Research ArticleChanging Trends in Meteorological Elements andReference Evapotranspiration in a Mega City A Case Study inShenzhen City China
Haijun Liu Xian Zhang Liwei Zhang and Xuming Wang
College of Water Sciences Beijing Normal University Beijing 100875 China
Correspondence should be addressed to Haijun Liu shanxilhjbnueducn
Received 26 September 2014 Revised 3 January 2015 Accepted 5 January 2015
Academic Editor Andreas Matzarakis
Copyright copy 2015 Haijun Liu et al This is an open access article distributed under the Creative Commons Attribution Licensewhich permits unrestricted use distribution and reproduction in any medium provided the original work is properly cited
Shenzhen city was a farmland region before 1978 and it then developed to a mega city in China This type of change in citycan greatly affect the climatic conditions In this study the daily monthly and annual climatic variables and the reference cropevapotranspiration (ET
0) for Shenzhen from 1954 to 2012 were computed using the FAO Penman-Monteith equation (PM) and
these parameters were analyzed to study the temporal trends of ET0and meteorological factors The trends and the time points of
abrupt changes of ET0andmeteorological factors were tested usingMann-Kendall methods Results show that in the past 59 years
the annual ET0first decreased from 1954 to 1978 then increased from 1979 to 1990 and now varied slightly after 1990 The mean
air temperature rose gradually and the relative humidity decreased as a whole These trends finally resulted in an increasing trendin vapor pressure deficit (VPD) The wind speed showed a slightly decreasing trend Both the annual total sunshine duration andnet radiation showed trends of rapid decline ET
0change is sensitive to the hours of sunshine and VPDThe significant increase in
ET0after 1979 was mainly due to the increased air temperature and decreased relative humidity
1 Introduction
It has been confirmed that there has been a change in theglobal climate that is closely related to increases in the con-centrations of atmospheric greenhouse gases (CO
2 NO119909) [1]
These changes in climate are expected to causemajor changesin various climatic variables such as precipitation air tem-perature relative humidity and solar radiation [2] Accordingto the IPCC report [1] the air temperature at the earthrsquossurface level increased by 074∘C from 1906 to 2005 andthis rising trend of the air temperature is likely to continue inthe 21st century which will cause changes in the hydrologicalcycle by affecting precipitation and evaporation [3 4] Bates etal (2008) pointed out that over the last century precipitationhas primarily increased over land in high northern latitudesbut decreased from 10∘S to 30∘N since the 1970s and globallythe area of land classified as very dry has more than doubledsince the 1970s due to climate change [4]
The trends for changes in the city climate may differ fromthose for farmland and forest regions because of changes inthe energy balance caused by the surface albedo and compo-nents of air In cities the ground surface is primarily coveredby concrete and asphalt and the green ground surface andwater surface only account for a small portion of the totalarea For example it has been reported that the percentageof green ground surface and water surface of the total cityarea is approximately 207 for Beijing City (2012) 383 forShanghai City (2012) and 426 for Tianjin City (2012) Dueto the high albedo rate and low heat capacity of concrete andasphalt compared to those of green surfaces andwater bodiesthe air temperature is recorded as much higher in a city thanin farmland [1 4ndash7]The climate is also dryer in cities than infarmland due to the lower amount of water vapor caused bysmaller evaporative surface prevailing in cities [1 4 5]
The composition of the air may also change the energybalance in cities With city development a larger number of
Hindawi Publishing CorporationAdvances in MeteorologyVolume 2015 Article ID 324502 11 pageshttpdxdoiorg1011552015324502
2 Advances in Meteorology
Baoan district
Nanshan
district
districtdistrict
districtdistrict
FutianLuohu
Yantian
Longgang
20 10 0 20 40
(km)
Shenzhen authorityClimate station
NE
SW
Figure 1 Schematic diagram of the research region (Shenzhen) and the meteorological station
particles and particles with more complicated compositionsare emitted into the air In Beijing City it has been reportedthat vehicular emissions of HC CO and NO
119909were estimated
to reach 1333 times 104 10002 times 104 and 755 times 104 tonsrespectively in 2005 [8] and vehicle emissions in the urbanarea made up 75 of the total emissions in Beijing in 2002for vehicle-related pollutants [9] In recent years the ldquograyskyrdquo phenomenon caused by fine particles (particles less than25 120583m in aerodynamic diameter or PM
25) is an increasing
public concern caused by fossil fuels that play a decisive rolein the development of economic and urban growth in thecurrent period in China In Shanghai for example the annualaverage PM
25concentrations in 2005 reached 56120583gm3
which was much higher than the value proposed by theWorld Health Organization Air Quality Guidelines (WHOAQG) [10]This phenomenonwas also found in certain otherareas abroad Aziz and Bajwa (2008) showed that the majorsource of urban air pollution is the growing motor vehicularemission [11]They pointed out thatmotor vehicular emissionis a mixture of five key pollutants namely carbon monoxide(CO) hydrocarbons (HC) particulate matter (PM
10) nitro-
gen oxides (NO119909) and ozone (O
3)
Shenzhen city constitutes a special place in Chinese his-tory Before 1978 it was a small town occupied by farmlandand fishing areas with a population of approximately 314000permanent residents After that year it was chosen as thefirst Special Economic Zone in China after which the citydeveloped quickly nowadays it is one of the four largest citiesin China By the end of 2012 the city had a population of1055 million permanent residents approximately 35 timesthe number compared with 1978 and it had a gross domesticproduct (GDP) of 20937 million dollars approximately 6600times higher than that in 1978 From 1979 to 2012 thefarmland area decreased from 63553 to 8520 ha whereas theconstruction area increased from 29 to 5260 ha
We suppose that the urban development of Shenzhenmay have altered the heat balance of the ground surface
thereby influencing climatic variables as well as the referenceevapotranspiration changes
2 Data and Methods
21 Site Shenzhen city (22∘271015840ndash22∘521015840N 113∘461015840ndash114∘471015840E)lies in the southern region of Guangdong province in south-eastern China (Figure 1) The area administered by the Shen-zhen municipal government is 2020 km2 Shenzhen has atypical subtropical maritime climate with plenty of rain amild climate and numerous sunshine hoursThe yearlymeanprecipitation is 1914mm and the rainy season can rangebetween April and September The minimum precipitationis 912mm (1962) and the maximum is 2747mm (2001) Theannualmean air temperature is 223∘Cwith aminimumdailyair temperature of 02∘C (February 11 1957) and a maximumof 387∘C (July 10 1980)Themean yearly total sunshine hoursare 2120 hours
22 Meteorological Data All meteorological data were col-lected from a national climatic station (22∘331015840N 114∘061015840E182m above surface level) (Figure 1)Themeteorological datainclude the atmospheric pressure daily values of the precipi-tation the mean maximum andminimum air temperaturesmean relative humidity mean wind speed and sunshinehours from 1954 to 2012
23 Methods
231 Reference Crop Evapotranspiration The reference cropevapotranspiration (ET
0) was calculated using the FAO Pen-
man-Monteith method (hereafter denoted as PM) The PMmethod is [12]
is the reference crop evapotranspirationmmsdotdayminus1 119877
119899is the net radiation MJsdotmminus2sdotdayminus1 119866 is the
soil heat flux that can be neglected at daily intervals [12]MJsdotmminus2sdotdayminus1 120574 is the psychrometric constant kPasdot∘Cminus1 119880
2
is the wind speed measured at 2m above ground surfacemsdotsminus1 119890
119904and 119890
119886are the saturation and the actual vapor
pressure kPa and Δ is the slope of the saturation vaporpressure curve at the air temperature kPasdot∘Cminus1 The monthlyor yearly total ET
0s is the sum of daily ET0in an entire month
or yearThe net radiation was not directly measured at Shenzhen
stationTherefore the net radiation was calculated using datafor the daily sunshine hours and themaximumandminimumair temperatures following the method suggested by Allenet al (1998) [12]
The actual daily vapor pressure (119890119886) and vapor pressure
deficit (VPD) were based on the daily mean air temperatureand relative humidity and were calculated in the followingway
119890119886= RH times 119890
119900(119879mean)
VPD = (1 minus RH) times 119890119900 (119879mean) (2)
where RH is daily mean relative humidity 119879mean is thedaily mean temperature ∘C and 119890119900(119879mean) is saturated vaporpressure at 119879mean kPa 119890
119900(119879mean) is calculated as follows
119890119900(119879mean) = 06108 exp(
1727119879mean119879mean + 2373
) (3)
232 Mann-Kendall Test The Mann-Kendall test is one ofthemost widely used nonparametric tests to detect significanttrends in climatic variables and potential evapotranspirationin time series [13ndash15]
Mann-Kendall Test for Temporal-Trend Analysis The Mann-Kendall test is based on the statistic 119878 [16]
119878 =
119873minus1
sum
119894=1
119873
sum
119895=119894+1
sign (119909119895minus 119909119894) (4)
where 119909119894and 119909
119895are two generic sequential data values of the
variable119873 is the length of the data set and the sign (119883) takesthe following values
sign (119883) =
+1 if 119883 gt 0
0 if 119883 = 0
minus1 if 119883 lt 0
(5)
A positive 119878 in (4) represents a positive trend in the observeddata series and vice versa Under the null hypothesis ofno trend in the data 119867
0 the statistic 119878 is approximately
normally distributed with the mean 119864(119878) = 0 For data setswith more than ten values the variance associated with the
Mann-Kendall statistic 119878 (VAR(119878)) can be calculated afterconsidering the distribution as very close to normal
VAR (119878) = 1
18
[119873 (119873 minus 1) (2119873 + 5)
minus
119902
sum
119901=1
119905119901(119905119901minus 1) (2119905
119901+ 5)]
(6)
where 119902 is the number of tied groups and 119905119901is the number of
data values in the 119901th groupThe values of 119878 and VAR(119878) are used to compute the test
statistic 119885 as follows
119885 =
119878 minus 1
radicVAR (119878)if 119878 gt 0
0 if 119878 = 0119878 + 1
radicVAR (119878)if 119878 lt 0
(7)
The presence of a statistically significant trend is evaluatedusing the 119885 value A positive (negative) value of 119885 indicatesan upward (downward) trend The statistic 119885 has a normaldistribution To test for either an upward or downwardmono-tonic trend (a two-tailed test) at the 120572 level of significance1198670is rejected if the absolute value of119885 is greater than119885
1minus1205722
where 1198851minus1205722
is obtained from the standard normal cumula-tive distribution tables The tested significance level 120572 wasset to 001 in this study
Mann-Kendall Test for Mutation Point Analysis We supposethat a time series (119909
1 1199092 119909
119899) exists One order series119898
119894
is constructed to represent the sample accumulative numberof 119909119894gt 119909119895(1 le 119895 le 119894) 119889
119896is defined in the following way
119889119896=
119896
sum
1
119898119894
(2 le 119896 le 119899) (8)
The mean value and variance of 119889119896can be approximately
expressed as follows
119864 (119889119896) =
119896 (119896 minus 1)
4
(9)
var (119889119896) =
119899 (119899 minus 1) (2119899 + 5)
72
(2 le 119896 le 119899) (10)
Under the hypothesis that the time series is random andindependent the statistic is defined in the following way
UF119896=
119889119896minus 119864 (119889
119896)
var (119889119896)
(119896 = 1 2 119899) (11)
Given the significance level of 120572 |UF119896| gt UF
1205722means
that the series has an obvious change trend Time series 119909119894
is arranged in reverse order and is calculated with (9) whileensuring that
UB119896= minusUF
119896
119896 = 119899 + 1 minus 119896
(12)
4 Advances in Meteorology
By analyzing the statistical series UF119896and UB
119896 the
change trend of series 119909119894can be further analyzed and the
mutation time and region can be determined UF119896gt 0 indi-
cates that the series tends to increase and UF119896lt 0 indicates
that the series tends to decrease When the series exceed thecredibility line they exhibit an obvious increasing or decreas-ing trend If an intersection point exists between the curvesof UF
119896and UB
119896and falls between the credibility lines the
corresponding time of the intersection point is the startingmoment of mutation
In this study the statistical analysis software DPS (versionDPS 145 developed by Zhejiang University China) was usedto analyze the mutation point of the temporal trends of ET
0
and the meteorological variables [17] The credibility line isdrawn at the significance level of 120572 = 001
233 Sensitivity-Analysis Method Sensitivity analysis wasemployed to identify the climatic variables that most stronglyinfluence ET
0following themethod proposed byMoller et al
(2004) and Liu et al (2014) [15 18] In this study the temporaltrends of most climatic variables showed abrupt change in1978 and the climatic variables during 1954ndash1978 were sig-nificantly (119875 lt 001) higher or lower than those during 1992ndash2012 see Sections 311 to 316 in the following text Hencedata during these two periods were used to assess the climatechange impact on ET
0 The detailed processes of the sensitiv-
ity analysis are described as follows (1) themean values of theair temperature relative humidity wind speed and sunshinehours in each day of a year were calculated using climatic datafrom 1954 to 1978 and the corresponding daily and annualET0were calculated using these mean daily values and the
Penman-Monteith method The mean daily value for eachclimatic variablewas set as the reference climatic variable andthe calculated ET
0was set as the reference ET
0 (2)The sen-
sitivity of each climatic variable to ET0was analyzed by com-
paring the reference ET0and the ET
0calculated by changing
one variable with a rate of minus15 minus10 minus5 5 10 and 15 andkeeping the other variables identical to the reference climaticvariables Then a figure was drawn based on these data(Figure 10) (3) The mean values of the air temperaturerelative humidity wind speed and sunshine hours in each dayof a yearwere calculated using climatic data from 1992 to 2012The corresponding ET
0was calculated by setting one climatic
variable as those during 1992 and 2012 and the others as thereference climatic variable (in step (1)) and then calculatingET0and comparing it with the reference ET
0 The relative
change in ET0caused by each climatic change during 1992
to 2012 is marked in Figure 10 The most sensitive variable toa change in ET
0is determined by comparing the relative ET
0
changes caused by each variable
3 Results
31 Annual Distributions and Trends in theChanges of Climatic Variables
311 Sunshine Hours and Solar Radiation The annual sun-shine duration and annual total radiation during the period
1954 1962 1970 1978 1986 1994 2002 2010
Year
y = minus10077x + 21997
R2 = 0525
2700
2400
2100
1800
1500Tota
l sun
shin
e hou
rs (h
)
(a)
UF-
UB
UFUB
1954 1962 1970 1978 1986 1994 2002 2010
Year
minus8
minus4
4
0
(b)
Figure 2 Temporal change in the yearly sunshine duration (a) andthe trend results from the Mann-Kendall method (b)
1954 1962 1970 1978 1986 1994 2002 2010
Year
y = minus87059x + 21314
R2 = 0558
4600
4400
4200
4000
3800
3600
Tota
l sol
ar ra
diat
ion
(MJmiddotm
minus2)
(a)
UF-
UB
UFUB
1954 1962 1970 1978 1986 1994 2002 2010
Year
minus8
minus4
4
0
(b)
Figure 3 Temporal trend in the yearly total solar radiation (a) andthe trend results from the Mann-Kendall method (b)
from 1954 to 2012 are given in Figures 2 and 3 It can beseen in Figure 2 that the annual sunshine duration showed
Advances in Meteorology 5
y = minus06754x + 32529
R2 = 00008
3000
2500
2000
1500
1000
500Tota
l pre
cipi
tatio
n (m
m)
1954 1962 1970 1978 1986 1994 2002 2010
Year
(a)
UF-
UB
UFUB
1954 1962 1970 1978 1986 1994 2002 2010
Year
minus8
minus4
4
0
(b)
Figure 4 Temporal trend of the yearly total precipitation (a) andthe trend results from the Mann-Kendall method (b)
an obvious decline which decreased from 2651 to 1590 hourswith an average of 2014 hours (Figure 2(a))Through the two-tailed test (UF and UB lines) we found that the decreasingtrend of the sunshine hours is obvious for the period from1982 to 2012 and the increasing trend is obvious for the periodof 1954 to 1973 Similar to the trend for the sunshine durationthe solar radiation also showed an obvious declining trendwhich ranges from 3672 to 4520MJsdotmminus2 with an average of4050MJsdotmminus2 (Figure 3(a)) The two-tailed test showed thatthe solar radiation values from 1982 to 2012 were significantlylower than those from 1954 to 1973 The shifting mutationpoints of the sunshine duration and the solar radiation areall found in 1978 using the Mann-Kendall mutational test(119875 lt 001) (Figures 2 and 3(b))
312 Precipitation The annual total precipitation valuesduring the period from 1954 to 2012 are given in Figure 4 Itcan be seen in Figure 4 that the highest precipitation valuewas 2747mm in 2001 and the lowest was 912mm in 1963(Figure 4(a)) The annual mean was 1914mm during theperiod from 1954 to 2012 The precipitation varied greatlyover the years whereas the trend test showed that the tempo-ral trend of the annual precipitation is not significant and amutation point has not been tested (Figure 4(b))
313 Relative Humidity The daily mean relative humidityduring the period from 1954 to 2012 is given in Figure 5 Itcan be seen that the daily mean relative humidity increasedslightly in the first 20 years followed by an obvious decline
1954 1962 1970 1978 1986 1994 2002 2010
Year
y = minus01389x + 35181
R2 = 05331
84
80
76
72
68Mea
n re
lativ
e hum
idity
()
(a)
UF-
UB
UFUB
1954 1962 1970 1978 1986 1994 2002 2010
Year
minus8
minus4
4
0
(b)
Figure 5 Temporal trend of the mean relative humidity (a) and thetrend results from the Mann-Kendall method (b)
after 1978 the range is from 691 to 822 and the average is764 (Figure 5(a)) From the Mann-Kendall test (119875 lt 001)we found that the averaged relative humidity value during theperiod from 1992 to 2012 is significantly (119875 lt 001) lower thanthe period from 1954 to 1992
314 Air Temperature Themean air temperature during theperiod from 1954 to 2012 is shown in Figure 6 Accordingto Figure 6 the mean air temperature showed an obviouslyincreasing trend and the values range from 215 to 239∘Cwith an average value of 226∘C (Figure 6(a)) The statisticalresult showed that the mean air temperature over the periodfrom 1990 to 2012 is significantly (119875 lt 001) higher than thoseover the period from 1954 to 1983 (Figure 6(b))
315 Vapor Pressure Deficit Figure 7(a) shows the temporaltrend of the vapor pressure deficit (VPD) during the periodfrom 1954 to 2012 VPD ranges from 0449 to 0867 kPa andthe average is 0648 kPa There is no significant temporaltrend for VPD from 1954 to 1977 and after 1978 VPDincreased greatly It was found that the average VPD duringthe period from 1992 to 2012 is significantly (119875 lt 001) higherthan the values from 1954 to 1991 according to the Mann-Kendall mutational test
316 Wind Speed The annual mean wind speed variedgreatly during the period from 1954 to 2012 (Figure 8(a))Clearly decreasing trends were found from 1954 to 1977 andfrom 1987 to 2012 whereas increasing trend was found from
6 Advances in Meteorology
1954 1962 1970 1978 1986 1994 2002 2010
Year
y = 003x minus 36919
R2 = 06208
24
24
23
23
22
22
21
Mea
n te
mpe
ratu
re(∘
C)
(a)
UF-
UB
UFUB
1954 1962 1970 1978 1986 1994 2002 2010
Year
minus8
minus4
4
8
0
(b)
Figure 6 Temporal trend of themean temperature (a) and the trendresults from the Mann-Kendall method (b)
1954 1962 1970 1978 1986 1994 2002 2010
Year
y = 00055x minus 10261
R2 = 06721
100
040
020
060
080
Vapo
r pre
ssur
e defi
cit (
kPa)
(a)
UF-
UB
UFUB
1954 1962 1970 1978 1986 1994 2002 2010
Year
minus8
minus4
4
8
0
(b)
Figure 7 Temporal trend of the mean vapor pressure deficit (a) andthe trend results from the Mann-Kendall method (b)
1978 to 1986 It can be seen in Figure 8(a) that the maximumwind speed was 372ms in 1954 the minimum was 178ms
1954 1962 1970 1978 1986 1994 2002 2010
Year
y = minus00098x + 22013
R2 = 01214
40
30
20
10Win
d sp
eed
(mmiddotsminus
1)
(a)
UF-
UB
UFUB
1954 1962 1970 1978 1986 1994 2002 2010
Year
minus8
minus4
4
0
(b)
Figure 8 Temporal trend of the mean wind speed (a) and the trendresults from the Mann-Kendall method (b)
in 1977 and the annual mean was 260ms The statisticalresults show that there were no significant temporal trendsfrom 1954 to 2012
32 Reference Evapotranspiration (ET0) Theannual total ET0
varied from 946 to 1373mm with a total mean of 1187mmFigure 9(a) shows the change in the annual ET
0over the
past 59 years According to Figure 9(a) ET0firstly decreased
gradually from 1954 to 1978 then increased from 1978 to 1992and lastly varied slightly after 1992 From 1954 to 1978 theannual total ET
0decreased from 1285 to 946mmwith amean
value of 1110mm afterwards the annual total ET0increased
from 1118 to 1373mm with a mean value of 1284mm duringthe period from 1992 to 2012 The mean ET
0in the period
of 1992ndash2012 increased by 156 over those in the period of1954ndash1978 indicating a great increase in evaporation poten-tial The statistical results based on the Mann-Kendall test(119875 lt 001) (Figure 9(b)) showed that ET
0during the period
from 1972 to 1987 was significantly lower than that from 2001to 2012 which was significantly higher than those in otherperiods The shifting mutation point for ET
0is found at 1992
by the Mann-Kendall test (Figure 9(b))For analyzing the yearly ET
0distribution the total ET
0for
eachmonthwas calculated and the averagedmonth total ET0s
in period of 1954ndash1978 and 1992ndash2012were calculated respec-tively The monthly ET
0distributions in these two periods
and in all study period (1954ndash2012) are showed in Figure 10The highest monthly ET
0generally appears in July and
August and the lowest in January and FebruaryMonthly ET0
from May to November are generally higher than 100mm
Advances in Meteorology 7
Table 1 Mean values of each climatic variable in the periods of 1954ndash1978 and 1992ndash2012 and changes of each climatic variable to ET0variation
Air temperature Relative humidity Sunshine hours Wind speed∘C Hoursday ms
Mean values1954ndash1978 220 789 608 2681992ndash2012 233 732 504 252
Average ET0 over 1954ndash1978Average ET0 over 1992ndash2012
(a)
UF-
UB
UFUB
1954 1962 1970 1978 1986 1994 2002 2010
Year
minus8
minus4
4
8
0
(b)
Figure 9 Temporal trend of the yearly total ET0(a) and the trend
results from the Mann-Kendall method (b)
and the total ET0in this period accounts for approximately
13 the yearly total Monthly ET0s in the period of 1992ndash2012
are all higher by 5ndash30mm or 7ndash25 than those in the 1954ndash1978 period The most increases in monthly ET
0are found in
period from July to September and the lowest from January toApril with an increase of less than 7mmTherefore the greatincrease in ET
0in summer (generally from June to October)
makes main contribution to yearly ET0
33 Sensitivity Analysis of ET0 to the Change of ClimaticVariables Figures 2 to 8 show that most mutation pointsfor most of the climatic variables were found during theperiod from 1978 to 1992 and the mean values of the climaticvariables over the period from 1954 to 1978 were significantly
150
100
50
01 2 3 4 5 6 7 8 9 10 11 12
Months
Monthly mean from period of 1954ndash2012
Monthly mean from period of 1954ndash1978Monthly mean from period of 1992ndash2012
50
40
30
20
10
0Mon
thly
tota
ldie
renc
eET 0
(mm
)
Incr
ease
per
cent
age i
nET
0(
)
Increase in ET0 from 1954ndash1978 to 1992ndash2012Increase percentage in ET0
Figure 10 Yearly distribution of ET0averaged in the periods of
1954ndash1978 1992ndash2012 and 1954ndash2012 ET0increase amount and the
corresponding increase percentage in each month from periods of1954ndash1978 to 1992ndash2012 were showed
(119875 lt 001) higherlower than those during the period from1992 to 2012 Hence the mean values of each variable duringthe two periods from 1954 to 1978 and from 1992 to 2012 werecalculated and used to analyze their change effects on the ET
0
changes following themethod described in Section 233Thesummary of the mean values of each climate and their effectson ET
0were listed in Table 1 and Figure 11
It can be seen in Table 1 that the daily mean relativehumidity daily mean air temperature wind speed and sun-shine hours from the period of 1954ndash1978 to 1992ndash2012 wereminus72 59 minus60 and minus171 respectively which resulted inET0changes by 146 53 minus13 and minus32 respectively The
contribution of each climate variablersquos variation from 1954ndash1978 to 1992ndash2012 to ET
0change is shown in Figure 11 It
can be found that decrease in relative humidity accounted forapproximately 60 variation in ET
0 followed by temperature
increase with a contribution of 22 and sunshine hoursreduction of 13 Wind speed accounted for 6 variationin ET
0 Similarly in another mega city Beijing in China the
order of climate change to ET0variation frommain to weak is
air temperature relative humidity sunshine hours and windspeed [14]
8 Advances in Meteorology
Relative humidity
60Wind speed6
Temperature22
Sunshine hours13
Figure 11 The contribution of each climatic variable change to ET0variation
300
200
100
0
Dev
elopm
ent o
f She
nzhe
n ci
ty
Heavy industry output value (billion dollars)Resident population (10minus1 million)Electric energy production (102 MWh)
1954 1962 1970 1978 1986 1994 2002 2010
Year
Figure 12 Temporal trend of the development of Shenzhen city
4 Discussion
41 Change in the Sunshine Hours and Urban DevelopmentSunshine hours generally depend on the cloud cover man-made aerosols and certain air pollutants (including SO
2
NO119909 and PM) [19 20] Recent studies indicated that the
most probable cause for the depression of sunshine hours orsolar radiation is in the increased concentrations ofmanmadeaerosols and other air pollutants [1 19ndash22] In this studythe sunshine hours and amount of radiation showed cleardecreasing trends after 1970 and theywere significantly lowerafter 1980 during rapid urban development
Figure 12 shows the increasing trends of the residentpopulation heavy industry output value and electric energyproduction It is confirmed that the increasing trend in theenergy consumption corresponds to increased emission ofpolluted particles including CO CO
2 NO119909 SO119909 PM25 O3
andHCThese particlesmay result in an increase in the atmo-spheric aerosol concentration [23ndash27] which can directlyattenuate the surface solar radiation (SSR) by scattering andabsorbing solar radiation (direct effect) or can indirectlyattenuate SSR by their ability to act as cloud condensation
nuclei thereby increasing the cloud reflectivity and lifetime(first and second indirect effects) [24 28] A remarkabledecline in SSR between the 1950s and the 1980s was found inseveral studies that were performed at selected observationstations based on sites in Europe the Baltics the South PoleGermany and the former Soviet Union [21] Today a com-prehensive literature exists that confirms the declines of SSRduring this period in many places around the world [29]In Beijing and several other Chinese cities the decline ofsunshine hours and SSR has also been reported [15 30]
42 Air Temperature Change and Urban Development It hasbeen confirmed that there is a large air temperature differ-ence in urban and rural areas [4ndash7] These air temperaturedifferences result from the influence of the thermal emissivityproperties of urban surfaces and the three-dimensional con-figuration and heat capacity of erected structures onto the airtemperature patterns in an urban region [7 15 31]Most of theworldrsquos cities thus show higher air temperatures in the urbancore than in the surrounding rural areas [4 5 7] For examplein Beijing City China Kuang et al [32] measured that themean land surface temperature of urban impervious surfaceswas about 6ndash12∘C higher than that of the urban green spaceand that in built-up areaswas on average 3ndash6∘Chigher than inrural areas They showed the main reason is the higher ratioof sensible heat to net radiation (063) and lower ratio of thelatent heat to net radiation (019) on the urban impervioussurface as compared to the corresponding rates of 030 and063 in green space and cropland In this study there is noobvious temporal trend in the air temperature before 1978whereas after that time the temperature increased greatlyalthough it showed a slight declining trend in recent yearsThe large air temperature increase from 1978 to 2002 may bedue to the increasing area of construction and the decrease infarmland [32] Figure 13 shows the changes in the farmlandand construction areas It can be found that there was arapid increase in construction areas and an abrupt decreasein farmland areas after 1978 which led to a change in thethermal balance and then resulted in the increase in the airtemperature in Shenzhen city
Advances in Meteorology 9
1954 1962 1970 1978 1986 1994 2002 2010
Year
8000
6000
4000
2000
0Farm
land
and
cons
truc
tion
area
Farm land area (101 ha)Construction area (ha)
Figure 13 Temporal trend of the farmland area and the floor spaceof the buildings under construction
43 Changes in the Relative Humidity and the Vapor PressureDeficit due to Urban Development The vapor pressure deficitwas calculated by the air temperature and the relative humid-ity following themethod of Allen et al (1998) [12] It is clearlyshown in the calculation method that the vapor pressuredeficit (VPD) increases with air temperature and decreaseswith relative humidity In this study both relative humidityand mean air temperature varied slightly during the periodfrom 1954 to 1980 (Figures 5 and 6) which consequentlyresulted in a slight change in the vapor pressure deficit Afterthat period the mean air temperature increased significantlyand the relative humidity was decreased remarkably Thesetrends resulted in an increase in the vapor pressure deficit(Figure 7)
The relative humidity is defined as the ratio of the watervapor density (mass per unit volume) to the saturation watervapor density and it is also approximately the ratio of theactual to the saturation vapor pressure A greater evaporativearea may produce more water vapor for the atmosphere andthen increase the water vapor density as well as the relativehumidity at a given temperature In the city study area thefarmland area decreased whereas the construction areaincreased (Figure 13) These data indicate a decreasing trendin the evaporative area which may result in a decrease in thewater vapor density as well as the relative humidity [33] Sim-ilarly a decreasing trend of relative humidity and an increasein the vapor pressure deficit were observed in Beijing Datongin Shanxi province Zhang Jiakou in Hebei Province and BetDagan in Israel [15 34 35] In Beijing it was shown thatthe VPD increased with air temperature and decreased withrelative humidity from 1951 to 2010 [15] Cohen et al (2002)showed that themain factor responsible for the increased panevaporation was the growth in the aerodynamic componentof evaporation which was due to increases in both the airVPD and the wind speed at Bet Dagan from 1964 to 1997 [35]
The vapor pressure deficit represents a gradient acrosswhich water vapor is removed from the evapotranspiringsurface into the surrounding air [12] A greater vapor pressuredeficit generally causes a higher evaporative rate Hence the
increasing vapor pressure deficit in the Shenzhen area willresult in increasing plant evapotranspiration
44 ET0 Change and City Development In the current studyarea ET
0first decreased from 1950s to 1970s and then
increased greatly in the 1980s During the 1990s and 2000s itvaried slightly with amean value of 1287mm It was observedthat after the onset of urban development in 1978 the ET
0
value increased and became higher than this for the periodprior to the urban development Figures 2 5 6 and 8 showthat the mutation points for most climatic variables wereobserved near the onset year of urban development and sen-sitivity analysis shows that the higher ET
0during the period
of 1992ndash2012 is mainly attributed to the relative humiditydecrease and air temperature increaseHence it could be con-cluded that the quick development of Shenzhen city alteredthe climatic conditions and hence increased the local ET
0
In other large cities an increasing ET0trendwas also found in
recent decades For example in Beijing City the annual ET0
increased significantly from 1951 to 2010 and from the 1950sto 2000s it increased from 1039 to 1148mm [15] The annualpotential evapotranspiration (PET) displays a significantupward trend from 1970 to 2006 and the trend varied from 1to 4mm per year in the Pyrenees-Orientales and Audeadministrative departments respectively and the westernpart of the French Mediterranean area with an averageincrease in PET of between 34mm and 150mm in the last 36years [36]
The increasing trend in ET0that was observed in large
cities is different than this that was found in the farmlandFor example in the Haihe River basin in northern Chinadecreasing trends were observed in 26 stations while 16stations showed significant decreasing trends from 1950 to2007 with rates from minus20 to minus37mmyearminus1 [34] Similarly asignificant decreasing trend of ET
0with a rate of minus3mm per
yearwas found in the arid region of northwest China [37]Thedifference trend in ET
0between the large cities and the
farmland may be due to the variations in the energy balanceand the evaporative potential In the farmland areas morethan 60ndash80 of the net radiation is used for plant evapotran-spiration [33 38 39]This effect not only reduces the availableheat for heating the air environment but also increasesthe water vapor in the near atmosphere The latter effect mayincrease the relative humidity and reduce the vapor pressuredeficit and lastly it may reduce the reference evapotranspi-ration For urban conditions the decrease in green land willdecrease the energy consumption caused by crop evapotran-spiration increasing the available heat and decreasing thewater vapor which ultimately results in an increase in theevaporative potential
It should be noted in Figure 9 that the ET0values in 1997
and 2012 were much lower than those in the neighboringyears It is estimated that ET
0in 1997 and 2012 decreased
by minus104 and minus130 compared to the mean value for theperiod from 1990 to 2012 The sensitive analysis methoddescribed in Section 33 was used to determine the effects ofeach variable on the ET
0changes in 1997 and 2012 compared
to the mean value during the period of 1990ndash2012 The
10 Advances in Meteorology
results showed that the changes in the relative humidity airtemperature sunshine hours andwind speed in 1997 resultedin changes in ET
0by minus72 minus08 minus20 and minus05 respectively
and by minus126 minus08 minus06 and minus42 in 2012 respectively Itcould be concluded that the increase in the relative humidityis the main factor for ET
0reduction followed by the wind
speed air temperature and sunshine hours in the two yearsBased on the mean values of the climatic variables
averaged over the periods of 1954ndash1978 and 1992ndash2012 ET0
increased by 145 in the latter period For the sensitivityanalysis changes in the relative humidity air temperaturewind speed and sunshine hours during 1992ndash2012 caused thevariation of ET
0by 146 50 minus13 and minus38 respectively
compared to those for the period 1954ndash1978The total amountof change in ET
0was 156 based on the sensitivity analysis
This value is similar to the rate of increase of 145 by com-parison of the ET
0values between the two time periods Liu
et al (2014) [15] calculated the ET0change rates by directly
comparing the mean values and summing each ET0change
rate caused by climatic variables using the same sensitiveanalysis method the ET
0change rates were 107 and 105
respectively Liu et al (2009) [40] found that ET0inside the
screenhouse was reduced by 39 compared to that in theopen field By considering the effect of each climatic variablechange to ET
0using the sensitivity analysis the total ET
0
change rate sums to 44 which is similar to the value of39 Therefore it could be concluded that the sensitivity-analysis method used in this study is reliable and easy to useand hence it is recommended for the analysis of the effect ofclimate change on ET
0
5 Conclusions
(1) The development of Shenzhen city greatly affected thelocal climatic conditions Before the onset of urbandevelopment each climatic variable varied slightlywhereas afterward the air temperature increased sig-nificantly and the sunshine hours and relative humid-ity decreased significantly The mutation point formost climatic variables is observed at approximately1978 the onset year for urban development
(2) ET0first decreased from 1954 to 1978 and then
increased quickly and reached a maximal value of1373mm during the period from 1992 to 2012 Themean ET
0value for the period from 1954 to 1978 was
1110mm and increased to 1284mm during the periodfrom 1992 to 2012 indicating an increasing trend ofthe evaporative demand
(3) Sensitivity analysis showed that ET0is most sensitive
to relative humidity followed by air temperaturesunshine hours and wind speed
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This paper is partially supported by the National ScienceFoundation of China (Grant nos 51179005 51479004) Theauthors greatly acknowledge the comments from the editorand the two anonymous reviewers
References
[1] IPCC Climate Change 2014 Synthesis Report An Assessmentof Intergovernmental Panel on Climate Change IPCC GenevaSwitzerland 2014 httpipccchindexhtml
[2] J D Haskett Y A Pachepsky and B Acock ldquoEffect of climateand atmospheric change on soybean water stress a study ofIowardquo Ecological Modelling vol 135 no 2-3 pp 265ndash277 2000
[3] T G Huntington ldquoEvidence for intensification of the globalwater cycle review and synthesisrdquo Journal of Hydrology vol 319no 1ndash4 pp 83ndash95 2006
[4] B C Bates Z W Kundzewicz S Wu and J Palutikof ldquoClimatechange and waterrdquo Technical Paper of the IntergovernmentalPanel on Climate Change IPCC Secretariat Geneva Switzer-land 2008
[5] C M Philandras D A Metaxas and P T Nastos ldquoClimatevariability and urbanization in AthensrdquoTheoretical and AppliedClimatology vol 63 no 1-2 pp 65ndash72 1999
[6] R L Wilby ldquoPast and projected trends in Londonrsquos urban heatislandrdquoWeather vol 58 no 7 pp 251ndash260 2003
[7] N Schwarz U Schlink U Franck and K Groszligmann ldquoRela-tionship of land surface and air temperatures and its implica-tions for quantifying urban heat island indicatorsmdashan applica-tion for the city of Leipzig (Germany)rdquoEcological Indicators vol18 pp 693ndash704 2012
[8] H Wang L Fu X Lin Y Zhou and J C Chen ldquoA bottom-up methodology to estimate vehicle emissions for the Beijingurban areardquo Science of the Total Environment vol 407 no 6 pp1947ndash1953 2009
[9] DESE (Department of Environmental Science and Engineer-ingTsinghua University) Mobile Source Database EmissionInventory and Treatment Proposal for Beijing Tsinghua Univer-sity Beijing China 2005
[10] H Kan S J London G Chen et al ldquoDifferentiating the effectsof fine and coarse particles on daily mortality in ShanghaiChinardquo Environment International vol 33 no 3 pp 376ndash3842007
[11] A Aziz and I U Bajwa ldquoErroneous mass transit system andits tended relationship with motor vehicular air pollution (Anintegrated approach for reduction of urban air pollution inLahore)rdquo Environmental Monitoring and Assessment vol 137no 1ndash3 pp 25ndash33 2008
[12] R G Allen L S Perreira D Raes and M Smith Crop Evap-otranspiration Guidelines for Computing Crop Water Require-ments FAO Irrigation and Drainage Paper no 56 FAO RomeItaly 1998
[13] K H Hamed ldquoTrend detection in hydrologic data the Mann-Kendall trend test under the scaling hypothesisrdquo Journal ofHydrology vol 349 no 3-4 pp 350ndash363 2008
[14] L Q Liang L J Li andQ Liu ldquoTemporal variation of referenceevapotranspiration during 1961ndash2005 in the Taoer River basin ofNortheast Chinardquo Agricultural and Forest Meteorology vol 150no 2 pp 298ndash306 2010
Advances in Meteorology 11
[15] H Liu Y Li T Josef R H Zhang and G H HuangldquoQuantitative estimation of climate change effects on potentialevapotranspiration in Beijing during 1951ndash2010rdquo Journal ofGeographical Sciences vol 24 no 1 pp 93ndash112 2014
[16] M G Kendall and A StuartThe Advanced Theory of StatisticsGriffin London UK 1973
[17] Q-Y Tang and C-X Zhang ldquoData Processing System (DPS)software with experimental design statistical analysis and datamining developed for use in entomological researchrdquo InsectScience vol 20 no 2 pp 254ndash260 2013
[18] M Moller J Tanny Y Li and S Cohen ldquoMeasuring andpredicting evapotranspiration in an insect-proof screenhouserdquoAgricultural and Forest Meteorology vol 127 no 1-2 pp 35ndash512004
[19] G Kitsara G Papaioannou A Papathanasiou and A RetalisldquoDimmingbrightening in athens trends in sunshine durationcloud cover and reference evapotranspirationrdquoWater ResourcesManagement vol 27 no 6 pp 1623ndash1633 2013
[20] Y Wang Y Yang S Han Q X Wang and G H ZhangldquoSunshine dimming and brightening in Chinese cities (1955ndash2011) was driven by air pollution rather than cloudsrdquo ClimateResearch vol 56 no 1 pp 11ndash20 2013
[21] G Stanhill and S Cohen ldquoGlobal dimming a review of theevidence for a widespread and significant reduction in globalradiation with discussion of its probable causes and possibleagricultural consequencesrdquoAgricultural and ForestMeteorologyvol 107 no 4 pp 255ndash278 2001
[22] Q Liu and Z Yang ldquoQuantitative estimation of the impact ofclimate change on actual evapotranspiration in the Yellow RiverBasin Chinardquo Journal of Hydrology vol 395 no 3-4 pp 226ndash234 2010
[23] D I Stern ldquoReversal of the trend in global anthropogenic sulfuremissionsrdquoGlobal Environmental Change vol 16 no 2 pp 207ndash220 2006
[24] I Koren J V Martins L A Remer and H Afargan ldquoSmokeinvigoration versus inhibition of clouds over the AmazonrdquoScience vol 321 no 5891 pp 946ndash949 2008
[25] D Rosenfeld Y J Kaufman and I Koren ldquoSwitching cloudcover and dynamical regimes from open to closed Benard cellsin response to the suppression of precipitation by aerosolsrdquoAtmospheric Chemistry and Physics vol 6 no 9 pp 2503ndash25112006
[26] C Ruckstuhl R Philipona K Behrens et al ldquoAerosol and cloudeffects on solar brightening and the recent rapid warmingrdquoGeophysical Research Letters vol 35 no 12 Article ID L127082008
[27] DG Streets Y Fang CMian et al ldquoAnthropogenic andnaturalcontributions to regional trends in aerosol optical depth 1980ndash2006rdquo Journal of Geophysical Research Atmospheres vol 114 no10 Article ID D00D18 2009
[28] V Ramanathan P J Crutzen J T Kiehl and D RosenfeldldquoAtmospheremdashaerosols climate and the hydrological cyclerdquoScience vol 294 no 5549 pp 2119ndash2124 2001
[29] M Wild ldquoEnlightening global dimming and brighteningrdquoBulletin of the AmericanMeteorological Society vol 93 no 1 pp27ndash37 2012
[30] G D Liu Y Li H J Liu and J Xiao ldquoChanging trend of refer-ence crop evapotranspiration and its dominatedmeteorologicalvariables in Shanxi province in the past 55 yearsrdquo Journal ofIrrigation and Drainage vol 31 no 4 pp 26ndash30 2012
[31] C-S Rim ldquoThe effects of urbanization geographical and topo-graphical conditions on reference evapotranspirationrdquo ClimaticChange vol 97 no 3 pp 483ndash514 2009
[32] WKuang Y Liu YDou et al ldquoWhat are hot andwhat are not inan urban landscape quantifying and explaining the land surfacetemperature pattern in Beijing Chinardquo Landscape Ecology2014
[33] Z Qin Q Yu S Xu et al ldquoWater heat fluxes and water useefficiency measurement and modeling above a farmland in theNorth China Plainrdquo Science in China D Earth Sciences vol 48no 1 pp 207ndash217 2005
[34] B Tang L Tong S Z Kang and L Zhang ldquoImpacts ofclimate variability on reference evapotranspiration over 58 yearsin the Haihe river basin of north Chinardquo Agricultural WaterManagement vol 98 no 10 pp 1660ndash1670 2011
[35] S Cohen A Ianetz and G Stanhill ldquoEvaporative climatechanges at BetDagan Israel 1964ndash1998rdquoAgricultural and ForestMeteorology vol 111 no 2 pp 83ndash91 2002
[36] K Chaouche L Neppel C Dieulin et al ldquoAnalyses of precipita-tion temperature and evapotranspiration in a French Mediter-ranean region in the context of climate changerdquoComptes RendusGeoscience vol 342 no 3 pp 234ndash243 2010
[37] Z Huo X Dai S Feng S Kang and G Huang ldquoEffect of cli-mate change on reference evapotranspiration and aridity indexin arid region of Chinardquo Journal of Hydrology vol 492 pp 24ndash34 2013
[38] G Peng X Cai H Zhang A Li F Hu and M Y LeclercldquoHeat flux apportionment to heterogeneous surfaces using fluxfootprint analysisrdquo Advances in Atmospheric Sciences vol 25no 1 pp 107ndash116 2008
[39] Y Q Zhang Y J Shen C M Liu et al ldquoMeasurement andanalysis of water heat and CO
2flux from a farmland in the
North China plainrdquo Acta Geographica Sinica vol 57 no 3 pp333ndash342 2002 (Chinese)
[40] H-J Liu G-H Huang S Cohen and J Tanny ldquoChange in cropevapotranspiration and associated influencing factors underscreenhouse conditionsrdquo Chinese Journal of Eco-Agriculturevol 17 no 3 pp 484ndash488 2009 (Chinese)
Figure 1 Schematic diagram of the research region (Shenzhen) and the meteorological station
particles and particles with more complicated compositionsare emitted into the air In Beijing City it has been reportedthat vehicular emissions of HC CO and NO
119909were estimated
to reach 1333 times 104 10002 times 104 and 755 times 104 tonsrespectively in 2005 [8] and vehicle emissions in the urbanarea made up 75 of the total emissions in Beijing in 2002for vehicle-related pollutants [9] In recent years the ldquograyskyrdquo phenomenon caused by fine particles (particles less than25 120583m in aerodynamic diameter or PM
25) is an increasing
public concern caused by fossil fuels that play a decisive rolein the development of economic and urban growth in thecurrent period in China In Shanghai for example the annualaverage PM
25concentrations in 2005 reached 56120583gm3
which was much higher than the value proposed by theWorld Health Organization Air Quality Guidelines (WHOAQG) [10]This phenomenonwas also found in certain otherareas abroad Aziz and Bajwa (2008) showed that the majorsource of urban air pollution is the growing motor vehicularemission [11]They pointed out thatmotor vehicular emissionis a mixture of five key pollutants namely carbon monoxide(CO) hydrocarbons (HC) particulate matter (PM
10) nitro-
gen oxides (NO119909) and ozone (O
3)
Shenzhen city constitutes a special place in Chinese his-tory Before 1978 it was a small town occupied by farmlandand fishing areas with a population of approximately 314000permanent residents After that year it was chosen as thefirst Special Economic Zone in China after which the citydeveloped quickly nowadays it is one of the four largest citiesin China By the end of 2012 the city had a population of1055 million permanent residents approximately 35 timesthe number compared with 1978 and it had a gross domesticproduct (GDP) of 20937 million dollars approximately 6600times higher than that in 1978 From 1979 to 2012 thefarmland area decreased from 63553 to 8520 ha whereas theconstruction area increased from 29 to 5260 ha
We suppose that the urban development of Shenzhenmay have altered the heat balance of the ground surface
thereby influencing climatic variables as well as the referenceevapotranspiration changes
2 Data and Methods
21 Site Shenzhen city (22∘271015840ndash22∘521015840N 113∘461015840ndash114∘471015840E)lies in the southern region of Guangdong province in south-eastern China (Figure 1) The area administered by the Shen-zhen municipal government is 2020 km2 Shenzhen has atypical subtropical maritime climate with plenty of rain amild climate and numerous sunshine hoursThe yearlymeanprecipitation is 1914mm and the rainy season can rangebetween April and September The minimum precipitationis 912mm (1962) and the maximum is 2747mm (2001) Theannualmean air temperature is 223∘Cwith aminimumdailyair temperature of 02∘C (February 11 1957) and a maximumof 387∘C (July 10 1980)Themean yearly total sunshine hoursare 2120 hours
22 Meteorological Data All meteorological data were col-lected from a national climatic station (22∘331015840N 114∘061015840E182m above surface level) (Figure 1)Themeteorological datainclude the atmospheric pressure daily values of the precipi-tation the mean maximum andminimum air temperaturesmean relative humidity mean wind speed and sunshinehours from 1954 to 2012
23 Methods
231 Reference Crop Evapotranspiration The reference cropevapotranspiration (ET
0) was calculated using the FAO Pen-
man-Monteith method (hereafter denoted as PM) The PMmethod is [12]
is the reference crop evapotranspirationmmsdotdayminus1 119877
119899is the net radiation MJsdotmminus2sdotdayminus1 119866 is the
soil heat flux that can be neglected at daily intervals [12]MJsdotmminus2sdotdayminus1 120574 is the psychrometric constant kPasdot∘Cminus1 119880
2
is the wind speed measured at 2m above ground surfacemsdotsminus1 119890
119904and 119890
119886are the saturation and the actual vapor
pressure kPa and Δ is the slope of the saturation vaporpressure curve at the air temperature kPasdot∘Cminus1 The monthlyor yearly total ET
0s is the sum of daily ET0in an entire month
or yearThe net radiation was not directly measured at Shenzhen
stationTherefore the net radiation was calculated using datafor the daily sunshine hours and themaximumandminimumair temperatures following the method suggested by Allenet al (1998) [12]
The actual daily vapor pressure (119890119886) and vapor pressure
deficit (VPD) were based on the daily mean air temperatureand relative humidity and were calculated in the followingway
119890119886= RH times 119890
119900(119879mean)
VPD = (1 minus RH) times 119890119900 (119879mean) (2)
where RH is daily mean relative humidity 119879mean is thedaily mean temperature ∘C and 119890119900(119879mean) is saturated vaporpressure at 119879mean kPa 119890
119900(119879mean) is calculated as follows
119890119900(119879mean) = 06108 exp(
1727119879mean119879mean + 2373
) (3)
232 Mann-Kendall Test The Mann-Kendall test is one ofthemost widely used nonparametric tests to detect significanttrends in climatic variables and potential evapotranspirationin time series [13ndash15]
Mann-Kendall Test for Temporal-Trend Analysis The Mann-Kendall test is based on the statistic 119878 [16]
119878 =
119873minus1
sum
119894=1
119873
sum
119895=119894+1
sign (119909119895minus 119909119894) (4)
where 119909119894and 119909
119895are two generic sequential data values of the
variable119873 is the length of the data set and the sign (119883) takesthe following values
sign (119883) =
+1 if 119883 gt 0
0 if 119883 = 0
minus1 if 119883 lt 0
(5)
A positive 119878 in (4) represents a positive trend in the observeddata series and vice versa Under the null hypothesis ofno trend in the data 119867
0 the statistic 119878 is approximately
normally distributed with the mean 119864(119878) = 0 For data setswith more than ten values the variance associated with the
Mann-Kendall statistic 119878 (VAR(119878)) can be calculated afterconsidering the distribution as very close to normal
VAR (119878) = 1
18
[119873 (119873 minus 1) (2119873 + 5)
minus
119902
sum
119901=1
119905119901(119905119901minus 1) (2119905
119901+ 5)]
(6)
where 119902 is the number of tied groups and 119905119901is the number of
data values in the 119901th groupThe values of 119878 and VAR(119878) are used to compute the test
statistic 119885 as follows
119885 =
119878 minus 1
radicVAR (119878)if 119878 gt 0
0 if 119878 = 0119878 + 1
radicVAR (119878)if 119878 lt 0
(7)
The presence of a statistically significant trend is evaluatedusing the 119885 value A positive (negative) value of 119885 indicatesan upward (downward) trend The statistic 119885 has a normaldistribution To test for either an upward or downwardmono-tonic trend (a two-tailed test) at the 120572 level of significance1198670is rejected if the absolute value of119885 is greater than119885
1minus1205722
where 1198851minus1205722
is obtained from the standard normal cumula-tive distribution tables The tested significance level 120572 wasset to 001 in this study
Mann-Kendall Test for Mutation Point Analysis We supposethat a time series (119909
1 1199092 119909
119899) exists One order series119898
119894
is constructed to represent the sample accumulative numberof 119909119894gt 119909119895(1 le 119895 le 119894) 119889
119896is defined in the following way
119889119896=
119896
sum
1
119898119894
(2 le 119896 le 119899) (8)
The mean value and variance of 119889119896can be approximately
expressed as follows
119864 (119889119896) =
119896 (119896 minus 1)
4
(9)
var (119889119896) =
119899 (119899 minus 1) (2119899 + 5)
72
(2 le 119896 le 119899) (10)
Under the hypothesis that the time series is random andindependent the statistic is defined in the following way
UF119896=
119889119896minus 119864 (119889
119896)
var (119889119896)
(119896 = 1 2 119899) (11)
Given the significance level of 120572 |UF119896| gt UF
1205722means
that the series has an obvious change trend Time series 119909119894
is arranged in reverse order and is calculated with (9) whileensuring that
UB119896= minusUF
119896
119896 = 119899 + 1 minus 119896
(12)
4 Advances in Meteorology
By analyzing the statistical series UF119896and UB
119896 the
change trend of series 119909119894can be further analyzed and the
mutation time and region can be determined UF119896gt 0 indi-
cates that the series tends to increase and UF119896lt 0 indicates
that the series tends to decrease When the series exceed thecredibility line they exhibit an obvious increasing or decreas-ing trend If an intersection point exists between the curvesof UF
119896and UB
119896and falls between the credibility lines the
corresponding time of the intersection point is the startingmoment of mutation
In this study the statistical analysis software DPS (versionDPS 145 developed by Zhejiang University China) was usedto analyze the mutation point of the temporal trends of ET
0
and the meteorological variables [17] The credibility line isdrawn at the significance level of 120572 = 001
233 Sensitivity-Analysis Method Sensitivity analysis wasemployed to identify the climatic variables that most stronglyinfluence ET
0following themethod proposed byMoller et al
(2004) and Liu et al (2014) [15 18] In this study the temporaltrends of most climatic variables showed abrupt change in1978 and the climatic variables during 1954ndash1978 were sig-nificantly (119875 lt 001) higher or lower than those during 1992ndash2012 see Sections 311 to 316 in the following text Hencedata during these two periods were used to assess the climatechange impact on ET
0 The detailed processes of the sensitiv-
ity analysis are described as follows (1) themean values of theair temperature relative humidity wind speed and sunshinehours in each day of a year were calculated using climatic datafrom 1954 to 1978 and the corresponding daily and annualET0were calculated using these mean daily values and the
Penman-Monteith method The mean daily value for eachclimatic variablewas set as the reference climatic variable andthe calculated ET
0was set as the reference ET
0 (2)The sen-
sitivity of each climatic variable to ET0was analyzed by com-
paring the reference ET0and the ET
0calculated by changing
one variable with a rate of minus15 minus10 minus5 5 10 and 15 andkeeping the other variables identical to the reference climaticvariables Then a figure was drawn based on these data(Figure 10) (3) The mean values of the air temperaturerelative humidity wind speed and sunshine hours in each dayof a yearwere calculated using climatic data from 1992 to 2012The corresponding ET
0was calculated by setting one climatic
variable as those during 1992 and 2012 and the others as thereference climatic variable (in step (1)) and then calculatingET0and comparing it with the reference ET
0 The relative
change in ET0caused by each climatic change during 1992
to 2012 is marked in Figure 10 The most sensitive variable toa change in ET
0is determined by comparing the relative ET
0
changes caused by each variable
3 Results
31 Annual Distributions and Trends in theChanges of Climatic Variables
311 Sunshine Hours and Solar Radiation The annual sun-shine duration and annual total radiation during the period
1954 1962 1970 1978 1986 1994 2002 2010
Year
y = minus10077x + 21997
R2 = 0525
2700
2400
2100
1800
1500Tota
l sun
shin
e hou
rs (h
)
(a)
UF-
UB
UFUB
1954 1962 1970 1978 1986 1994 2002 2010
Year
minus8
minus4
4
0
(b)
Figure 2 Temporal change in the yearly sunshine duration (a) andthe trend results from the Mann-Kendall method (b)
1954 1962 1970 1978 1986 1994 2002 2010
Year
y = minus87059x + 21314
R2 = 0558
4600
4400
4200
4000
3800
3600
Tota
l sol
ar ra
diat
ion
(MJmiddotm
minus2)
(a)
UF-
UB
UFUB
1954 1962 1970 1978 1986 1994 2002 2010
Year
minus8
minus4
4
0
(b)
Figure 3 Temporal trend in the yearly total solar radiation (a) andthe trend results from the Mann-Kendall method (b)
from 1954 to 2012 are given in Figures 2 and 3 It can beseen in Figure 2 that the annual sunshine duration showed
Advances in Meteorology 5
y = minus06754x + 32529
R2 = 00008
3000
2500
2000
1500
1000
500Tota
l pre
cipi
tatio
n (m
m)
1954 1962 1970 1978 1986 1994 2002 2010
Year
(a)
UF-
UB
UFUB
1954 1962 1970 1978 1986 1994 2002 2010
Year
minus8
minus4
4
0
(b)
Figure 4 Temporal trend of the yearly total precipitation (a) andthe trend results from the Mann-Kendall method (b)
an obvious decline which decreased from 2651 to 1590 hourswith an average of 2014 hours (Figure 2(a))Through the two-tailed test (UF and UB lines) we found that the decreasingtrend of the sunshine hours is obvious for the period from1982 to 2012 and the increasing trend is obvious for the periodof 1954 to 1973 Similar to the trend for the sunshine durationthe solar radiation also showed an obvious declining trendwhich ranges from 3672 to 4520MJsdotmminus2 with an average of4050MJsdotmminus2 (Figure 3(a)) The two-tailed test showed thatthe solar radiation values from 1982 to 2012 were significantlylower than those from 1954 to 1973 The shifting mutationpoints of the sunshine duration and the solar radiation areall found in 1978 using the Mann-Kendall mutational test(119875 lt 001) (Figures 2 and 3(b))
312 Precipitation The annual total precipitation valuesduring the period from 1954 to 2012 are given in Figure 4 Itcan be seen in Figure 4 that the highest precipitation valuewas 2747mm in 2001 and the lowest was 912mm in 1963(Figure 4(a)) The annual mean was 1914mm during theperiod from 1954 to 2012 The precipitation varied greatlyover the years whereas the trend test showed that the tempo-ral trend of the annual precipitation is not significant and amutation point has not been tested (Figure 4(b))
313 Relative Humidity The daily mean relative humidityduring the period from 1954 to 2012 is given in Figure 5 Itcan be seen that the daily mean relative humidity increasedslightly in the first 20 years followed by an obvious decline
1954 1962 1970 1978 1986 1994 2002 2010
Year
y = minus01389x + 35181
R2 = 05331
84
80
76
72
68Mea
n re
lativ
e hum
idity
()
(a)
UF-
UB
UFUB
1954 1962 1970 1978 1986 1994 2002 2010
Year
minus8
minus4
4
0
(b)
Figure 5 Temporal trend of the mean relative humidity (a) and thetrend results from the Mann-Kendall method (b)
after 1978 the range is from 691 to 822 and the average is764 (Figure 5(a)) From the Mann-Kendall test (119875 lt 001)we found that the averaged relative humidity value during theperiod from 1992 to 2012 is significantly (119875 lt 001) lower thanthe period from 1954 to 1992
314 Air Temperature Themean air temperature during theperiod from 1954 to 2012 is shown in Figure 6 Accordingto Figure 6 the mean air temperature showed an obviouslyincreasing trend and the values range from 215 to 239∘Cwith an average value of 226∘C (Figure 6(a)) The statisticalresult showed that the mean air temperature over the periodfrom 1990 to 2012 is significantly (119875 lt 001) higher than thoseover the period from 1954 to 1983 (Figure 6(b))
315 Vapor Pressure Deficit Figure 7(a) shows the temporaltrend of the vapor pressure deficit (VPD) during the periodfrom 1954 to 2012 VPD ranges from 0449 to 0867 kPa andthe average is 0648 kPa There is no significant temporaltrend for VPD from 1954 to 1977 and after 1978 VPDincreased greatly It was found that the average VPD duringthe period from 1992 to 2012 is significantly (119875 lt 001) higherthan the values from 1954 to 1991 according to the Mann-Kendall mutational test
316 Wind Speed The annual mean wind speed variedgreatly during the period from 1954 to 2012 (Figure 8(a))Clearly decreasing trends were found from 1954 to 1977 andfrom 1987 to 2012 whereas increasing trend was found from
6 Advances in Meteorology
1954 1962 1970 1978 1986 1994 2002 2010
Year
y = 003x minus 36919
R2 = 06208
24
24
23
23
22
22
21
Mea
n te
mpe
ratu
re(∘
C)
(a)
UF-
UB
UFUB
1954 1962 1970 1978 1986 1994 2002 2010
Year
minus8
minus4
4
8
0
(b)
Figure 6 Temporal trend of themean temperature (a) and the trendresults from the Mann-Kendall method (b)
1954 1962 1970 1978 1986 1994 2002 2010
Year
y = 00055x minus 10261
R2 = 06721
100
040
020
060
080
Vapo
r pre
ssur
e defi
cit (
kPa)
(a)
UF-
UB
UFUB
1954 1962 1970 1978 1986 1994 2002 2010
Year
minus8
minus4
4
8
0
(b)
Figure 7 Temporal trend of the mean vapor pressure deficit (a) andthe trend results from the Mann-Kendall method (b)
1978 to 1986 It can be seen in Figure 8(a) that the maximumwind speed was 372ms in 1954 the minimum was 178ms
1954 1962 1970 1978 1986 1994 2002 2010
Year
y = minus00098x + 22013
R2 = 01214
40
30
20
10Win
d sp
eed
(mmiddotsminus
1)
(a)
UF-
UB
UFUB
1954 1962 1970 1978 1986 1994 2002 2010
Year
minus8
minus4
4
0
(b)
Figure 8 Temporal trend of the mean wind speed (a) and the trendresults from the Mann-Kendall method (b)
in 1977 and the annual mean was 260ms The statisticalresults show that there were no significant temporal trendsfrom 1954 to 2012
32 Reference Evapotranspiration (ET0) Theannual total ET0
varied from 946 to 1373mm with a total mean of 1187mmFigure 9(a) shows the change in the annual ET
0over the
past 59 years According to Figure 9(a) ET0firstly decreased
gradually from 1954 to 1978 then increased from 1978 to 1992and lastly varied slightly after 1992 From 1954 to 1978 theannual total ET
0decreased from 1285 to 946mmwith amean
value of 1110mm afterwards the annual total ET0increased
from 1118 to 1373mm with a mean value of 1284mm duringthe period from 1992 to 2012 The mean ET
0in the period
of 1992ndash2012 increased by 156 over those in the period of1954ndash1978 indicating a great increase in evaporation poten-tial The statistical results based on the Mann-Kendall test(119875 lt 001) (Figure 9(b)) showed that ET
0during the period
from 1972 to 1987 was significantly lower than that from 2001to 2012 which was significantly higher than those in otherperiods The shifting mutation point for ET
0is found at 1992
by the Mann-Kendall test (Figure 9(b))For analyzing the yearly ET
0distribution the total ET
0for
eachmonthwas calculated and the averagedmonth total ET0s
in period of 1954ndash1978 and 1992ndash2012were calculated respec-tively The monthly ET
0distributions in these two periods
and in all study period (1954ndash2012) are showed in Figure 10The highest monthly ET
0generally appears in July and
August and the lowest in January and FebruaryMonthly ET0
from May to November are generally higher than 100mm
Advances in Meteorology 7
Table 1 Mean values of each climatic variable in the periods of 1954ndash1978 and 1992ndash2012 and changes of each climatic variable to ET0variation
Air temperature Relative humidity Sunshine hours Wind speed∘C Hoursday ms
Mean values1954ndash1978 220 789 608 2681992ndash2012 233 732 504 252
Average ET0 over 1954ndash1978Average ET0 over 1992ndash2012
(a)
UF-
UB
UFUB
1954 1962 1970 1978 1986 1994 2002 2010
Year
minus8
minus4
4
8
0
(b)
Figure 9 Temporal trend of the yearly total ET0(a) and the trend
results from the Mann-Kendall method (b)
and the total ET0in this period accounts for approximately
13 the yearly total Monthly ET0s in the period of 1992ndash2012
are all higher by 5ndash30mm or 7ndash25 than those in the 1954ndash1978 period The most increases in monthly ET
0are found in
period from July to September and the lowest from January toApril with an increase of less than 7mmTherefore the greatincrease in ET
0in summer (generally from June to October)
makes main contribution to yearly ET0
33 Sensitivity Analysis of ET0 to the Change of ClimaticVariables Figures 2 to 8 show that most mutation pointsfor most of the climatic variables were found during theperiod from 1978 to 1992 and the mean values of the climaticvariables over the period from 1954 to 1978 were significantly
150
100
50
01 2 3 4 5 6 7 8 9 10 11 12
Months
Monthly mean from period of 1954ndash2012
Monthly mean from period of 1954ndash1978Monthly mean from period of 1992ndash2012
50
40
30
20
10
0Mon
thly
tota
ldie
renc
eET 0
(mm
)
Incr
ease
per
cent
age i
nET
0(
)
Increase in ET0 from 1954ndash1978 to 1992ndash2012Increase percentage in ET0
Figure 10 Yearly distribution of ET0averaged in the periods of
1954ndash1978 1992ndash2012 and 1954ndash2012 ET0increase amount and the
corresponding increase percentage in each month from periods of1954ndash1978 to 1992ndash2012 were showed
(119875 lt 001) higherlower than those during the period from1992 to 2012 Hence the mean values of each variable duringthe two periods from 1954 to 1978 and from 1992 to 2012 werecalculated and used to analyze their change effects on the ET
0
changes following themethod described in Section 233Thesummary of the mean values of each climate and their effectson ET
0were listed in Table 1 and Figure 11
It can be seen in Table 1 that the daily mean relativehumidity daily mean air temperature wind speed and sun-shine hours from the period of 1954ndash1978 to 1992ndash2012 wereminus72 59 minus60 and minus171 respectively which resulted inET0changes by 146 53 minus13 and minus32 respectively The
contribution of each climate variablersquos variation from 1954ndash1978 to 1992ndash2012 to ET
0change is shown in Figure 11 It
can be found that decrease in relative humidity accounted forapproximately 60 variation in ET
0 followed by temperature
increase with a contribution of 22 and sunshine hoursreduction of 13 Wind speed accounted for 6 variationin ET
0 Similarly in another mega city Beijing in China the
order of climate change to ET0variation frommain to weak is
air temperature relative humidity sunshine hours and windspeed [14]
8 Advances in Meteorology
Relative humidity
60Wind speed6
Temperature22
Sunshine hours13
Figure 11 The contribution of each climatic variable change to ET0variation
300
200
100
0
Dev
elopm
ent o
f She
nzhe
n ci
ty
Heavy industry output value (billion dollars)Resident population (10minus1 million)Electric energy production (102 MWh)
1954 1962 1970 1978 1986 1994 2002 2010
Year
Figure 12 Temporal trend of the development of Shenzhen city
4 Discussion
41 Change in the Sunshine Hours and Urban DevelopmentSunshine hours generally depend on the cloud cover man-made aerosols and certain air pollutants (including SO
2
NO119909 and PM) [19 20] Recent studies indicated that the
most probable cause for the depression of sunshine hours orsolar radiation is in the increased concentrations ofmanmadeaerosols and other air pollutants [1 19ndash22] In this studythe sunshine hours and amount of radiation showed cleardecreasing trends after 1970 and theywere significantly lowerafter 1980 during rapid urban development
Figure 12 shows the increasing trends of the residentpopulation heavy industry output value and electric energyproduction It is confirmed that the increasing trend in theenergy consumption corresponds to increased emission ofpolluted particles including CO CO
2 NO119909 SO119909 PM25 O3
andHCThese particlesmay result in an increase in the atmo-spheric aerosol concentration [23ndash27] which can directlyattenuate the surface solar radiation (SSR) by scattering andabsorbing solar radiation (direct effect) or can indirectlyattenuate SSR by their ability to act as cloud condensation
nuclei thereby increasing the cloud reflectivity and lifetime(first and second indirect effects) [24 28] A remarkabledecline in SSR between the 1950s and the 1980s was found inseveral studies that were performed at selected observationstations based on sites in Europe the Baltics the South PoleGermany and the former Soviet Union [21] Today a com-prehensive literature exists that confirms the declines of SSRduring this period in many places around the world [29]In Beijing and several other Chinese cities the decline ofsunshine hours and SSR has also been reported [15 30]
42 Air Temperature Change and Urban Development It hasbeen confirmed that there is a large air temperature differ-ence in urban and rural areas [4ndash7] These air temperaturedifferences result from the influence of the thermal emissivityproperties of urban surfaces and the three-dimensional con-figuration and heat capacity of erected structures onto the airtemperature patterns in an urban region [7 15 31]Most of theworldrsquos cities thus show higher air temperatures in the urbancore than in the surrounding rural areas [4 5 7] For examplein Beijing City China Kuang et al [32] measured that themean land surface temperature of urban impervious surfaceswas about 6ndash12∘C higher than that of the urban green spaceand that in built-up areaswas on average 3ndash6∘Chigher than inrural areas They showed the main reason is the higher ratioof sensible heat to net radiation (063) and lower ratio of thelatent heat to net radiation (019) on the urban impervioussurface as compared to the corresponding rates of 030 and063 in green space and cropland In this study there is noobvious temporal trend in the air temperature before 1978whereas after that time the temperature increased greatlyalthough it showed a slight declining trend in recent yearsThe large air temperature increase from 1978 to 2002 may bedue to the increasing area of construction and the decrease infarmland [32] Figure 13 shows the changes in the farmlandand construction areas It can be found that there was arapid increase in construction areas and an abrupt decreasein farmland areas after 1978 which led to a change in thethermal balance and then resulted in the increase in the airtemperature in Shenzhen city
Advances in Meteorology 9
1954 1962 1970 1978 1986 1994 2002 2010
Year
8000
6000
4000
2000
0Farm
land
and
cons
truc
tion
area
Farm land area (101 ha)Construction area (ha)
Figure 13 Temporal trend of the farmland area and the floor spaceof the buildings under construction
43 Changes in the Relative Humidity and the Vapor PressureDeficit due to Urban Development The vapor pressure deficitwas calculated by the air temperature and the relative humid-ity following themethod of Allen et al (1998) [12] It is clearlyshown in the calculation method that the vapor pressuredeficit (VPD) increases with air temperature and decreaseswith relative humidity In this study both relative humidityand mean air temperature varied slightly during the periodfrom 1954 to 1980 (Figures 5 and 6) which consequentlyresulted in a slight change in the vapor pressure deficit Afterthat period the mean air temperature increased significantlyand the relative humidity was decreased remarkably Thesetrends resulted in an increase in the vapor pressure deficit(Figure 7)
The relative humidity is defined as the ratio of the watervapor density (mass per unit volume) to the saturation watervapor density and it is also approximately the ratio of theactual to the saturation vapor pressure A greater evaporativearea may produce more water vapor for the atmosphere andthen increase the water vapor density as well as the relativehumidity at a given temperature In the city study area thefarmland area decreased whereas the construction areaincreased (Figure 13) These data indicate a decreasing trendin the evaporative area which may result in a decrease in thewater vapor density as well as the relative humidity [33] Sim-ilarly a decreasing trend of relative humidity and an increasein the vapor pressure deficit were observed in Beijing Datongin Shanxi province Zhang Jiakou in Hebei Province and BetDagan in Israel [15 34 35] In Beijing it was shown thatthe VPD increased with air temperature and decreased withrelative humidity from 1951 to 2010 [15] Cohen et al (2002)showed that themain factor responsible for the increased panevaporation was the growth in the aerodynamic componentof evaporation which was due to increases in both the airVPD and the wind speed at Bet Dagan from 1964 to 1997 [35]
The vapor pressure deficit represents a gradient acrosswhich water vapor is removed from the evapotranspiringsurface into the surrounding air [12] A greater vapor pressuredeficit generally causes a higher evaporative rate Hence the
increasing vapor pressure deficit in the Shenzhen area willresult in increasing plant evapotranspiration
44 ET0 Change and City Development In the current studyarea ET
0first decreased from 1950s to 1970s and then
increased greatly in the 1980s During the 1990s and 2000s itvaried slightly with amean value of 1287mm It was observedthat after the onset of urban development in 1978 the ET
0
value increased and became higher than this for the periodprior to the urban development Figures 2 5 6 and 8 showthat the mutation points for most climatic variables wereobserved near the onset year of urban development and sen-sitivity analysis shows that the higher ET
0during the period
of 1992ndash2012 is mainly attributed to the relative humiditydecrease and air temperature increaseHence it could be con-cluded that the quick development of Shenzhen city alteredthe climatic conditions and hence increased the local ET
0
In other large cities an increasing ET0trendwas also found in
recent decades For example in Beijing City the annual ET0
increased significantly from 1951 to 2010 and from the 1950sto 2000s it increased from 1039 to 1148mm [15] The annualpotential evapotranspiration (PET) displays a significantupward trend from 1970 to 2006 and the trend varied from 1to 4mm per year in the Pyrenees-Orientales and Audeadministrative departments respectively and the westernpart of the French Mediterranean area with an averageincrease in PET of between 34mm and 150mm in the last 36years [36]
The increasing trend in ET0that was observed in large
cities is different than this that was found in the farmlandFor example in the Haihe River basin in northern Chinadecreasing trends were observed in 26 stations while 16stations showed significant decreasing trends from 1950 to2007 with rates from minus20 to minus37mmyearminus1 [34] Similarly asignificant decreasing trend of ET
0with a rate of minus3mm per
yearwas found in the arid region of northwest China [37]Thedifference trend in ET
0between the large cities and the
farmland may be due to the variations in the energy balanceand the evaporative potential In the farmland areas morethan 60ndash80 of the net radiation is used for plant evapotran-spiration [33 38 39]This effect not only reduces the availableheat for heating the air environment but also increasesthe water vapor in the near atmosphere The latter effect mayincrease the relative humidity and reduce the vapor pressuredeficit and lastly it may reduce the reference evapotranspi-ration For urban conditions the decrease in green land willdecrease the energy consumption caused by crop evapotran-spiration increasing the available heat and decreasing thewater vapor which ultimately results in an increase in theevaporative potential
It should be noted in Figure 9 that the ET0values in 1997
and 2012 were much lower than those in the neighboringyears It is estimated that ET
0in 1997 and 2012 decreased
by minus104 and minus130 compared to the mean value for theperiod from 1990 to 2012 The sensitive analysis methoddescribed in Section 33 was used to determine the effects ofeach variable on the ET
0changes in 1997 and 2012 compared
to the mean value during the period of 1990ndash2012 The
10 Advances in Meteorology
results showed that the changes in the relative humidity airtemperature sunshine hours andwind speed in 1997 resultedin changes in ET
0by minus72 minus08 minus20 and minus05 respectively
and by minus126 minus08 minus06 and minus42 in 2012 respectively Itcould be concluded that the increase in the relative humidityis the main factor for ET
0reduction followed by the wind
speed air temperature and sunshine hours in the two yearsBased on the mean values of the climatic variables
averaged over the periods of 1954ndash1978 and 1992ndash2012 ET0
increased by 145 in the latter period For the sensitivityanalysis changes in the relative humidity air temperaturewind speed and sunshine hours during 1992ndash2012 caused thevariation of ET
0by 146 50 minus13 and minus38 respectively
compared to those for the period 1954ndash1978The total amountof change in ET
0was 156 based on the sensitivity analysis
This value is similar to the rate of increase of 145 by com-parison of the ET
0values between the two time periods Liu
et al (2014) [15] calculated the ET0change rates by directly
comparing the mean values and summing each ET0change
rate caused by climatic variables using the same sensitiveanalysis method the ET
0change rates were 107 and 105
respectively Liu et al (2009) [40] found that ET0inside the
screenhouse was reduced by 39 compared to that in theopen field By considering the effect of each climatic variablechange to ET
0using the sensitivity analysis the total ET
0
change rate sums to 44 which is similar to the value of39 Therefore it could be concluded that the sensitivity-analysis method used in this study is reliable and easy to useand hence it is recommended for the analysis of the effect ofclimate change on ET
0
5 Conclusions
(1) The development of Shenzhen city greatly affected thelocal climatic conditions Before the onset of urbandevelopment each climatic variable varied slightlywhereas afterward the air temperature increased sig-nificantly and the sunshine hours and relative humid-ity decreased significantly The mutation point formost climatic variables is observed at approximately1978 the onset year for urban development
(2) ET0first decreased from 1954 to 1978 and then
increased quickly and reached a maximal value of1373mm during the period from 1992 to 2012 Themean ET
0value for the period from 1954 to 1978 was
1110mm and increased to 1284mm during the periodfrom 1992 to 2012 indicating an increasing trend ofthe evaporative demand
(3) Sensitivity analysis showed that ET0is most sensitive
to relative humidity followed by air temperaturesunshine hours and wind speed
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This paper is partially supported by the National ScienceFoundation of China (Grant nos 51179005 51479004) Theauthors greatly acknowledge the comments from the editorand the two anonymous reviewers
References
[1] IPCC Climate Change 2014 Synthesis Report An Assessmentof Intergovernmental Panel on Climate Change IPCC GenevaSwitzerland 2014 httpipccchindexhtml
[2] J D Haskett Y A Pachepsky and B Acock ldquoEffect of climateand atmospheric change on soybean water stress a study ofIowardquo Ecological Modelling vol 135 no 2-3 pp 265ndash277 2000
[3] T G Huntington ldquoEvidence for intensification of the globalwater cycle review and synthesisrdquo Journal of Hydrology vol 319no 1ndash4 pp 83ndash95 2006
[4] B C Bates Z W Kundzewicz S Wu and J Palutikof ldquoClimatechange and waterrdquo Technical Paper of the IntergovernmentalPanel on Climate Change IPCC Secretariat Geneva Switzer-land 2008
[5] C M Philandras D A Metaxas and P T Nastos ldquoClimatevariability and urbanization in AthensrdquoTheoretical and AppliedClimatology vol 63 no 1-2 pp 65ndash72 1999
[6] R L Wilby ldquoPast and projected trends in Londonrsquos urban heatislandrdquoWeather vol 58 no 7 pp 251ndash260 2003
[7] N Schwarz U Schlink U Franck and K Groszligmann ldquoRela-tionship of land surface and air temperatures and its implica-tions for quantifying urban heat island indicatorsmdashan applica-tion for the city of Leipzig (Germany)rdquoEcological Indicators vol18 pp 693ndash704 2012
[8] H Wang L Fu X Lin Y Zhou and J C Chen ldquoA bottom-up methodology to estimate vehicle emissions for the Beijingurban areardquo Science of the Total Environment vol 407 no 6 pp1947ndash1953 2009
[9] DESE (Department of Environmental Science and Engineer-ingTsinghua University) Mobile Source Database EmissionInventory and Treatment Proposal for Beijing Tsinghua Univer-sity Beijing China 2005
[10] H Kan S J London G Chen et al ldquoDifferentiating the effectsof fine and coarse particles on daily mortality in ShanghaiChinardquo Environment International vol 33 no 3 pp 376ndash3842007
[11] A Aziz and I U Bajwa ldquoErroneous mass transit system andits tended relationship with motor vehicular air pollution (Anintegrated approach for reduction of urban air pollution inLahore)rdquo Environmental Monitoring and Assessment vol 137no 1ndash3 pp 25ndash33 2008
[12] R G Allen L S Perreira D Raes and M Smith Crop Evap-otranspiration Guidelines for Computing Crop Water Require-ments FAO Irrigation and Drainage Paper no 56 FAO RomeItaly 1998
[13] K H Hamed ldquoTrend detection in hydrologic data the Mann-Kendall trend test under the scaling hypothesisrdquo Journal ofHydrology vol 349 no 3-4 pp 350ndash363 2008
[14] L Q Liang L J Li andQ Liu ldquoTemporal variation of referenceevapotranspiration during 1961ndash2005 in the Taoer River basin ofNortheast Chinardquo Agricultural and Forest Meteorology vol 150no 2 pp 298ndash306 2010
Advances in Meteorology 11
[15] H Liu Y Li T Josef R H Zhang and G H HuangldquoQuantitative estimation of climate change effects on potentialevapotranspiration in Beijing during 1951ndash2010rdquo Journal ofGeographical Sciences vol 24 no 1 pp 93ndash112 2014
[16] M G Kendall and A StuartThe Advanced Theory of StatisticsGriffin London UK 1973
[17] Q-Y Tang and C-X Zhang ldquoData Processing System (DPS)software with experimental design statistical analysis and datamining developed for use in entomological researchrdquo InsectScience vol 20 no 2 pp 254ndash260 2013
[18] M Moller J Tanny Y Li and S Cohen ldquoMeasuring andpredicting evapotranspiration in an insect-proof screenhouserdquoAgricultural and Forest Meteorology vol 127 no 1-2 pp 35ndash512004
[19] G Kitsara G Papaioannou A Papathanasiou and A RetalisldquoDimmingbrightening in athens trends in sunshine durationcloud cover and reference evapotranspirationrdquoWater ResourcesManagement vol 27 no 6 pp 1623ndash1633 2013
[20] Y Wang Y Yang S Han Q X Wang and G H ZhangldquoSunshine dimming and brightening in Chinese cities (1955ndash2011) was driven by air pollution rather than cloudsrdquo ClimateResearch vol 56 no 1 pp 11ndash20 2013
[21] G Stanhill and S Cohen ldquoGlobal dimming a review of theevidence for a widespread and significant reduction in globalradiation with discussion of its probable causes and possibleagricultural consequencesrdquoAgricultural and ForestMeteorologyvol 107 no 4 pp 255ndash278 2001
[22] Q Liu and Z Yang ldquoQuantitative estimation of the impact ofclimate change on actual evapotranspiration in the Yellow RiverBasin Chinardquo Journal of Hydrology vol 395 no 3-4 pp 226ndash234 2010
[23] D I Stern ldquoReversal of the trend in global anthropogenic sulfuremissionsrdquoGlobal Environmental Change vol 16 no 2 pp 207ndash220 2006
[24] I Koren J V Martins L A Remer and H Afargan ldquoSmokeinvigoration versus inhibition of clouds over the AmazonrdquoScience vol 321 no 5891 pp 946ndash949 2008
[25] D Rosenfeld Y J Kaufman and I Koren ldquoSwitching cloudcover and dynamical regimes from open to closed Benard cellsin response to the suppression of precipitation by aerosolsrdquoAtmospheric Chemistry and Physics vol 6 no 9 pp 2503ndash25112006
[26] C Ruckstuhl R Philipona K Behrens et al ldquoAerosol and cloudeffects on solar brightening and the recent rapid warmingrdquoGeophysical Research Letters vol 35 no 12 Article ID L127082008
[27] DG Streets Y Fang CMian et al ldquoAnthropogenic andnaturalcontributions to regional trends in aerosol optical depth 1980ndash2006rdquo Journal of Geophysical Research Atmospheres vol 114 no10 Article ID D00D18 2009
[28] V Ramanathan P J Crutzen J T Kiehl and D RosenfeldldquoAtmospheremdashaerosols climate and the hydrological cyclerdquoScience vol 294 no 5549 pp 2119ndash2124 2001
[29] M Wild ldquoEnlightening global dimming and brighteningrdquoBulletin of the AmericanMeteorological Society vol 93 no 1 pp27ndash37 2012
[30] G D Liu Y Li H J Liu and J Xiao ldquoChanging trend of refer-ence crop evapotranspiration and its dominatedmeteorologicalvariables in Shanxi province in the past 55 yearsrdquo Journal ofIrrigation and Drainage vol 31 no 4 pp 26ndash30 2012
[31] C-S Rim ldquoThe effects of urbanization geographical and topo-graphical conditions on reference evapotranspirationrdquo ClimaticChange vol 97 no 3 pp 483ndash514 2009
[32] WKuang Y Liu YDou et al ldquoWhat are hot andwhat are not inan urban landscape quantifying and explaining the land surfacetemperature pattern in Beijing Chinardquo Landscape Ecology2014
[33] Z Qin Q Yu S Xu et al ldquoWater heat fluxes and water useefficiency measurement and modeling above a farmland in theNorth China Plainrdquo Science in China D Earth Sciences vol 48no 1 pp 207ndash217 2005
[34] B Tang L Tong S Z Kang and L Zhang ldquoImpacts ofclimate variability on reference evapotranspiration over 58 yearsin the Haihe river basin of north Chinardquo Agricultural WaterManagement vol 98 no 10 pp 1660ndash1670 2011
[35] S Cohen A Ianetz and G Stanhill ldquoEvaporative climatechanges at BetDagan Israel 1964ndash1998rdquoAgricultural and ForestMeteorology vol 111 no 2 pp 83ndash91 2002
[36] K Chaouche L Neppel C Dieulin et al ldquoAnalyses of precipita-tion temperature and evapotranspiration in a French Mediter-ranean region in the context of climate changerdquoComptes RendusGeoscience vol 342 no 3 pp 234ndash243 2010
[37] Z Huo X Dai S Feng S Kang and G Huang ldquoEffect of cli-mate change on reference evapotranspiration and aridity indexin arid region of Chinardquo Journal of Hydrology vol 492 pp 24ndash34 2013
[38] G Peng X Cai H Zhang A Li F Hu and M Y LeclercldquoHeat flux apportionment to heterogeneous surfaces using fluxfootprint analysisrdquo Advances in Atmospheric Sciences vol 25no 1 pp 107ndash116 2008
[39] Y Q Zhang Y J Shen C M Liu et al ldquoMeasurement andanalysis of water heat and CO
2flux from a farmland in the
North China plainrdquo Acta Geographica Sinica vol 57 no 3 pp333ndash342 2002 (Chinese)
[40] H-J Liu G-H Huang S Cohen and J Tanny ldquoChange in cropevapotranspiration and associated influencing factors underscreenhouse conditionsrdquo Chinese Journal of Eco-Agriculturevol 17 no 3 pp 484ndash488 2009 (Chinese)
is the reference crop evapotranspirationmmsdotdayminus1 119877
119899is the net radiation MJsdotmminus2sdotdayminus1 119866 is the
soil heat flux that can be neglected at daily intervals [12]MJsdotmminus2sdotdayminus1 120574 is the psychrometric constant kPasdot∘Cminus1 119880
2
is the wind speed measured at 2m above ground surfacemsdotsminus1 119890
119904and 119890
119886are the saturation and the actual vapor
pressure kPa and Δ is the slope of the saturation vaporpressure curve at the air temperature kPasdot∘Cminus1 The monthlyor yearly total ET
0s is the sum of daily ET0in an entire month
or yearThe net radiation was not directly measured at Shenzhen
stationTherefore the net radiation was calculated using datafor the daily sunshine hours and themaximumandminimumair temperatures following the method suggested by Allenet al (1998) [12]
The actual daily vapor pressure (119890119886) and vapor pressure
deficit (VPD) were based on the daily mean air temperatureand relative humidity and were calculated in the followingway
119890119886= RH times 119890
119900(119879mean)
VPD = (1 minus RH) times 119890119900 (119879mean) (2)
where RH is daily mean relative humidity 119879mean is thedaily mean temperature ∘C and 119890119900(119879mean) is saturated vaporpressure at 119879mean kPa 119890
119900(119879mean) is calculated as follows
119890119900(119879mean) = 06108 exp(
1727119879mean119879mean + 2373
) (3)
232 Mann-Kendall Test The Mann-Kendall test is one ofthemost widely used nonparametric tests to detect significanttrends in climatic variables and potential evapotranspirationin time series [13ndash15]
Mann-Kendall Test for Temporal-Trend Analysis The Mann-Kendall test is based on the statistic 119878 [16]
119878 =
119873minus1
sum
119894=1
119873
sum
119895=119894+1
sign (119909119895minus 119909119894) (4)
where 119909119894and 119909
119895are two generic sequential data values of the
variable119873 is the length of the data set and the sign (119883) takesthe following values
sign (119883) =
+1 if 119883 gt 0
0 if 119883 = 0
minus1 if 119883 lt 0
(5)
A positive 119878 in (4) represents a positive trend in the observeddata series and vice versa Under the null hypothesis ofno trend in the data 119867
0 the statistic 119878 is approximately
normally distributed with the mean 119864(119878) = 0 For data setswith more than ten values the variance associated with the
Mann-Kendall statistic 119878 (VAR(119878)) can be calculated afterconsidering the distribution as very close to normal
VAR (119878) = 1
18
[119873 (119873 minus 1) (2119873 + 5)
minus
119902
sum
119901=1
119905119901(119905119901minus 1) (2119905
119901+ 5)]
(6)
where 119902 is the number of tied groups and 119905119901is the number of
data values in the 119901th groupThe values of 119878 and VAR(119878) are used to compute the test
statistic 119885 as follows
119885 =
119878 minus 1
radicVAR (119878)if 119878 gt 0
0 if 119878 = 0119878 + 1
radicVAR (119878)if 119878 lt 0
(7)
The presence of a statistically significant trend is evaluatedusing the 119885 value A positive (negative) value of 119885 indicatesan upward (downward) trend The statistic 119885 has a normaldistribution To test for either an upward or downwardmono-tonic trend (a two-tailed test) at the 120572 level of significance1198670is rejected if the absolute value of119885 is greater than119885
1minus1205722
where 1198851minus1205722
is obtained from the standard normal cumula-tive distribution tables The tested significance level 120572 wasset to 001 in this study
Mann-Kendall Test for Mutation Point Analysis We supposethat a time series (119909
1 1199092 119909
119899) exists One order series119898
119894
is constructed to represent the sample accumulative numberof 119909119894gt 119909119895(1 le 119895 le 119894) 119889
119896is defined in the following way
119889119896=
119896
sum
1
119898119894
(2 le 119896 le 119899) (8)
The mean value and variance of 119889119896can be approximately
expressed as follows
119864 (119889119896) =
119896 (119896 minus 1)
4
(9)
var (119889119896) =
119899 (119899 minus 1) (2119899 + 5)
72
(2 le 119896 le 119899) (10)
Under the hypothesis that the time series is random andindependent the statistic is defined in the following way
UF119896=
119889119896minus 119864 (119889
119896)
var (119889119896)
(119896 = 1 2 119899) (11)
Given the significance level of 120572 |UF119896| gt UF
1205722means
that the series has an obvious change trend Time series 119909119894
is arranged in reverse order and is calculated with (9) whileensuring that
UB119896= minusUF
119896
119896 = 119899 + 1 minus 119896
(12)
4 Advances in Meteorology
By analyzing the statistical series UF119896and UB
119896 the
change trend of series 119909119894can be further analyzed and the
mutation time and region can be determined UF119896gt 0 indi-
cates that the series tends to increase and UF119896lt 0 indicates
that the series tends to decrease When the series exceed thecredibility line they exhibit an obvious increasing or decreas-ing trend If an intersection point exists between the curvesof UF
119896and UB
119896and falls between the credibility lines the
corresponding time of the intersection point is the startingmoment of mutation
In this study the statistical analysis software DPS (versionDPS 145 developed by Zhejiang University China) was usedto analyze the mutation point of the temporal trends of ET
0
and the meteorological variables [17] The credibility line isdrawn at the significance level of 120572 = 001
233 Sensitivity-Analysis Method Sensitivity analysis wasemployed to identify the climatic variables that most stronglyinfluence ET
0following themethod proposed byMoller et al
(2004) and Liu et al (2014) [15 18] In this study the temporaltrends of most climatic variables showed abrupt change in1978 and the climatic variables during 1954ndash1978 were sig-nificantly (119875 lt 001) higher or lower than those during 1992ndash2012 see Sections 311 to 316 in the following text Hencedata during these two periods were used to assess the climatechange impact on ET
0 The detailed processes of the sensitiv-
ity analysis are described as follows (1) themean values of theair temperature relative humidity wind speed and sunshinehours in each day of a year were calculated using climatic datafrom 1954 to 1978 and the corresponding daily and annualET0were calculated using these mean daily values and the
Penman-Monteith method The mean daily value for eachclimatic variablewas set as the reference climatic variable andthe calculated ET
0was set as the reference ET
0 (2)The sen-
sitivity of each climatic variable to ET0was analyzed by com-
paring the reference ET0and the ET
0calculated by changing
one variable with a rate of minus15 minus10 minus5 5 10 and 15 andkeeping the other variables identical to the reference climaticvariables Then a figure was drawn based on these data(Figure 10) (3) The mean values of the air temperaturerelative humidity wind speed and sunshine hours in each dayof a yearwere calculated using climatic data from 1992 to 2012The corresponding ET
0was calculated by setting one climatic
variable as those during 1992 and 2012 and the others as thereference climatic variable (in step (1)) and then calculatingET0and comparing it with the reference ET
0 The relative
change in ET0caused by each climatic change during 1992
to 2012 is marked in Figure 10 The most sensitive variable toa change in ET
0is determined by comparing the relative ET
0
changes caused by each variable
3 Results
31 Annual Distributions and Trends in theChanges of Climatic Variables
311 Sunshine Hours and Solar Radiation The annual sun-shine duration and annual total radiation during the period
1954 1962 1970 1978 1986 1994 2002 2010
Year
y = minus10077x + 21997
R2 = 0525
2700
2400
2100
1800
1500Tota
l sun
shin
e hou
rs (h
)
(a)
UF-
UB
UFUB
1954 1962 1970 1978 1986 1994 2002 2010
Year
minus8
minus4
4
0
(b)
Figure 2 Temporal change in the yearly sunshine duration (a) andthe trend results from the Mann-Kendall method (b)
1954 1962 1970 1978 1986 1994 2002 2010
Year
y = minus87059x + 21314
R2 = 0558
4600
4400
4200
4000
3800
3600
Tota
l sol
ar ra
diat
ion
(MJmiddotm
minus2)
(a)
UF-
UB
UFUB
1954 1962 1970 1978 1986 1994 2002 2010
Year
minus8
minus4
4
0
(b)
Figure 3 Temporal trend in the yearly total solar radiation (a) andthe trend results from the Mann-Kendall method (b)
from 1954 to 2012 are given in Figures 2 and 3 It can beseen in Figure 2 that the annual sunshine duration showed
Advances in Meteorology 5
y = minus06754x + 32529
R2 = 00008
3000
2500
2000
1500
1000
500Tota
l pre
cipi
tatio
n (m
m)
1954 1962 1970 1978 1986 1994 2002 2010
Year
(a)
UF-
UB
UFUB
1954 1962 1970 1978 1986 1994 2002 2010
Year
minus8
minus4
4
0
(b)
Figure 4 Temporal trend of the yearly total precipitation (a) andthe trend results from the Mann-Kendall method (b)
an obvious decline which decreased from 2651 to 1590 hourswith an average of 2014 hours (Figure 2(a))Through the two-tailed test (UF and UB lines) we found that the decreasingtrend of the sunshine hours is obvious for the period from1982 to 2012 and the increasing trend is obvious for the periodof 1954 to 1973 Similar to the trend for the sunshine durationthe solar radiation also showed an obvious declining trendwhich ranges from 3672 to 4520MJsdotmminus2 with an average of4050MJsdotmminus2 (Figure 3(a)) The two-tailed test showed thatthe solar radiation values from 1982 to 2012 were significantlylower than those from 1954 to 1973 The shifting mutationpoints of the sunshine duration and the solar radiation areall found in 1978 using the Mann-Kendall mutational test(119875 lt 001) (Figures 2 and 3(b))
312 Precipitation The annual total precipitation valuesduring the period from 1954 to 2012 are given in Figure 4 Itcan be seen in Figure 4 that the highest precipitation valuewas 2747mm in 2001 and the lowest was 912mm in 1963(Figure 4(a)) The annual mean was 1914mm during theperiod from 1954 to 2012 The precipitation varied greatlyover the years whereas the trend test showed that the tempo-ral trend of the annual precipitation is not significant and amutation point has not been tested (Figure 4(b))
313 Relative Humidity The daily mean relative humidityduring the period from 1954 to 2012 is given in Figure 5 Itcan be seen that the daily mean relative humidity increasedslightly in the first 20 years followed by an obvious decline
1954 1962 1970 1978 1986 1994 2002 2010
Year
y = minus01389x + 35181
R2 = 05331
84
80
76
72
68Mea
n re
lativ
e hum
idity
()
(a)
UF-
UB
UFUB
1954 1962 1970 1978 1986 1994 2002 2010
Year
minus8
minus4
4
0
(b)
Figure 5 Temporal trend of the mean relative humidity (a) and thetrend results from the Mann-Kendall method (b)
after 1978 the range is from 691 to 822 and the average is764 (Figure 5(a)) From the Mann-Kendall test (119875 lt 001)we found that the averaged relative humidity value during theperiod from 1992 to 2012 is significantly (119875 lt 001) lower thanthe period from 1954 to 1992
314 Air Temperature Themean air temperature during theperiod from 1954 to 2012 is shown in Figure 6 Accordingto Figure 6 the mean air temperature showed an obviouslyincreasing trend and the values range from 215 to 239∘Cwith an average value of 226∘C (Figure 6(a)) The statisticalresult showed that the mean air temperature over the periodfrom 1990 to 2012 is significantly (119875 lt 001) higher than thoseover the period from 1954 to 1983 (Figure 6(b))
315 Vapor Pressure Deficit Figure 7(a) shows the temporaltrend of the vapor pressure deficit (VPD) during the periodfrom 1954 to 2012 VPD ranges from 0449 to 0867 kPa andthe average is 0648 kPa There is no significant temporaltrend for VPD from 1954 to 1977 and after 1978 VPDincreased greatly It was found that the average VPD duringthe period from 1992 to 2012 is significantly (119875 lt 001) higherthan the values from 1954 to 1991 according to the Mann-Kendall mutational test
316 Wind Speed The annual mean wind speed variedgreatly during the period from 1954 to 2012 (Figure 8(a))Clearly decreasing trends were found from 1954 to 1977 andfrom 1987 to 2012 whereas increasing trend was found from
6 Advances in Meteorology
1954 1962 1970 1978 1986 1994 2002 2010
Year
y = 003x minus 36919
R2 = 06208
24
24
23
23
22
22
21
Mea
n te
mpe
ratu
re(∘
C)
(a)
UF-
UB
UFUB
1954 1962 1970 1978 1986 1994 2002 2010
Year
minus8
minus4
4
8
0
(b)
Figure 6 Temporal trend of themean temperature (a) and the trendresults from the Mann-Kendall method (b)
1954 1962 1970 1978 1986 1994 2002 2010
Year
y = 00055x minus 10261
R2 = 06721
100
040
020
060
080
Vapo
r pre
ssur
e defi
cit (
kPa)
(a)
UF-
UB
UFUB
1954 1962 1970 1978 1986 1994 2002 2010
Year
minus8
minus4
4
8
0
(b)
Figure 7 Temporal trend of the mean vapor pressure deficit (a) andthe trend results from the Mann-Kendall method (b)
1978 to 1986 It can be seen in Figure 8(a) that the maximumwind speed was 372ms in 1954 the minimum was 178ms
1954 1962 1970 1978 1986 1994 2002 2010
Year
y = minus00098x + 22013
R2 = 01214
40
30
20
10Win
d sp
eed
(mmiddotsminus
1)
(a)
UF-
UB
UFUB
1954 1962 1970 1978 1986 1994 2002 2010
Year
minus8
minus4
4
0
(b)
Figure 8 Temporal trend of the mean wind speed (a) and the trendresults from the Mann-Kendall method (b)
in 1977 and the annual mean was 260ms The statisticalresults show that there were no significant temporal trendsfrom 1954 to 2012
32 Reference Evapotranspiration (ET0) Theannual total ET0
varied from 946 to 1373mm with a total mean of 1187mmFigure 9(a) shows the change in the annual ET
0over the
past 59 years According to Figure 9(a) ET0firstly decreased
gradually from 1954 to 1978 then increased from 1978 to 1992and lastly varied slightly after 1992 From 1954 to 1978 theannual total ET
0decreased from 1285 to 946mmwith amean
value of 1110mm afterwards the annual total ET0increased
from 1118 to 1373mm with a mean value of 1284mm duringthe period from 1992 to 2012 The mean ET
0in the period
of 1992ndash2012 increased by 156 over those in the period of1954ndash1978 indicating a great increase in evaporation poten-tial The statistical results based on the Mann-Kendall test(119875 lt 001) (Figure 9(b)) showed that ET
0during the period
from 1972 to 1987 was significantly lower than that from 2001to 2012 which was significantly higher than those in otherperiods The shifting mutation point for ET
0is found at 1992
by the Mann-Kendall test (Figure 9(b))For analyzing the yearly ET
0distribution the total ET
0for
eachmonthwas calculated and the averagedmonth total ET0s
in period of 1954ndash1978 and 1992ndash2012were calculated respec-tively The monthly ET
0distributions in these two periods
and in all study period (1954ndash2012) are showed in Figure 10The highest monthly ET
0generally appears in July and
August and the lowest in January and FebruaryMonthly ET0
from May to November are generally higher than 100mm
Advances in Meteorology 7
Table 1 Mean values of each climatic variable in the periods of 1954ndash1978 and 1992ndash2012 and changes of each climatic variable to ET0variation
Air temperature Relative humidity Sunshine hours Wind speed∘C Hoursday ms
Mean values1954ndash1978 220 789 608 2681992ndash2012 233 732 504 252
Average ET0 over 1954ndash1978Average ET0 over 1992ndash2012
(a)
UF-
UB
UFUB
1954 1962 1970 1978 1986 1994 2002 2010
Year
minus8
minus4
4
8
0
(b)
Figure 9 Temporal trend of the yearly total ET0(a) and the trend
results from the Mann-Kendall method (b)
and the total ET0in this period accounts for approximately
13 the yearly total Monthly ET0s in the period of 1992ndash2012
are all higher by 5ndash30mm or 7ndash25 than those in the 1954ndash1978 period The most increases in monthly ET
0are found in
period from July to September and the lowest from January toApril with an increase of less than 7mmTherefore the greatincrease in ET
0in summer (generally from June to October)
makes main contribution to yearly ET0
33 Sensitivity Analysis of ET0 to the Change of ClimaticVariables Figures 2 to 8 show that most mutation pointsfor most of the climatic variables were found during theperiod from 1978 to 1992 and the mean values of the climaticvariables over the period from 1954 to 1978 were significantly
150
100
50
01 2 3 4 5 6 7 8 9 10 11 12
Months
Monthly mean from period of 1954ndash2012
Monthly mean from period of 1954ndash1978Monthly mean from period of 1992ndash2012
50
40
30
20
10
0Mon
thly
tota
ldie
renc
eET 0
(mm
)
Incr
ease
per
cent
age i
nET
0(
)
Increase in ET0 from 1954ndash1978 to 1992ndash2012Increase percentage in ET0
Figure 10 Yearly distribution of ET0averaged in the periods of
1954ndash1978 1992ndash2012 and 1954ndash2012 ET0increase amount and the
corresponding increase percentage in each month from periods of1954ndash1978 to 1992ndash2012 were showed
(119875 lt 001) higherlower than those during the period from1992 to 2012 Hence the mean values of each variable duringthe two periods from 1954 to 1978 and from 1992 to 2012 werecalculated and used to analyze their change effects on the ET
0
changes following themethod described in Section 233Thesummary of the mean values of each climate and their effectson ET
0were listed in Table 1 and Figure 11
It can be seen in Table 1 that the daily mean relativehumidity daily mean air temperature wind speed and sun-shine hours from the period of 1954ndash1978 to 1992ndash2012 wereminus72 59 minus60 and minus171 respectively which resulted inET0changes by 146 53 minus13 and minus32 respectively The
contribution of each climate variablersquos variation from 1954ndash1978 to 1992ndash2012 to ET
0change is shown in Figure 11 It
can be found that decrease in relative humidity accounted forapproximately 60 variation in ET
0 followed by temperature
increase with a contribution of 22 and sunshine hoursreduction of 13 Wind speed accounted for 6 variationin ET
0 Similarly in another mega city Beijing in China the
order of climate change to ET0variation frommain to weak is
air temperature relative humidity sunshine hours and windspeed [14]
8 Advances in Meteorology
Relative humidity
60Wind speed6
Temperature22
Sunshine hours13
Figure 11 The contribution of each climatic variable change to ET0variation
300
200
100
0
Dev
elopm
ent o
f She
nzhe
n ci
ty
Heavy industry output value (billion dollars)Resident population (10minus1 million)Electric energy production (102 MWh)
1954 1962 1970 1978 1986 1994 2002 2010
Year
Figure 12 Temporal trend of the development of Shenzhen city
4 Discussion
41 Change in the Sunshine Hours and Urban DevelopmentSunshine hours generally depend on the cloud cover man-made aerosols and certain air pollutants (including SO
2
NO119909 and PM) [19 20] Recent studies indicated that the
most probable cause for the depression of sunshine hours orsolar radiation is in the increased concentrations ofmanmadeaerosols and other air pollutants [1 19ndash22] In this studythe sunshine hours and amount of radiation showed cleardecreasing trends after 1970 and theywere significantly lowerafter 1980 during rapid urban development
Figure 12 shows the increasing trends of the residentpopulation heavy industry output value and electric energyproduction It is confirmed that the increasing trend in theenergy consumption corresponds to increased emission ofpolluted particles including CO CO
2 NO119909 SO119909 PM25 O3
andHCThese particlesmay result in an increase in the atmo-spheric aerosol concentration [23ndash27] which can directlyattenuate the surface solar radiation (SSR) by scattering andabsorbing solar radiation (direct effect) or can indirectlyattenuate SSR by their ability to act as cloud condensation
nuclei thereby increasing the cloud reflectivity and lifetime(first and second indirect effects) [24 28] A remarkabledecline in SSR between the 1950s and the 1980s was found inseveral studies that were performed at selected observationstations based on sites in Europe the Baltics the South PoleGermany and the former Soviet Union [21] Today a com-prehensive literature exists that confirms the declines of SSRduring this period in many places around the world [29]In Beijing and several other Chinese cities the decline ofsunshine hours and SSR has also been reported [15 30]
42 Air Temperature Change and Urban Development It hasbeen confirmed that there is a large air temperature differ-ence in urban and rural areas [4ndash7] These air temperaturedifferences result from the influence of the thermal emissivityproperties of urban surfaces and the three-dimensional con-figuration and heat capacity of erected structures onto the airtemperature patterns in an urban region [7 15 31]Most of theworldrsquos cities thus show higher air temperatures in the urbancore than in the surrounding rural areas [4 5 7] For examplein Beijing City China Kuang et al [32] measured that themean land surface temperature of urban impervious surfaceswas about 6ndash12∘C higher than that of the urban green spaceand that in built-up areaswas on average 3ndash6∘Chigher than inrural areas They showed the main reason is the higher ratioof sensible heat to net radiation (063) and lower ratio of thelatent heat to net radiation (019) on the urban impervioussurface as compared to the corresponding rates of 030 and063 in green space and cropland In this study there is noobvious temporal trend in the air temperature before 1978whereas after that time the temperature increased greatlyalthough it showed a slight declining trend in recent yearsThe large air temperature increase from 1978 to 2002 may bedue to the increasing area of construction and the decrease infarmland [32] Figure 13 shows the changes in the farmlandand construction areas It can be found that there was arapid increase in construction areas and an abrupt decreasein farmland areas after 1978 which led to a change in thethermal balance and then resulted in the increase in the airtemperature in Shenzhen city
Advances in Meteorology 9
1954 1962 1970 1978 1986 1994 2002 2010
Year
8000
6000
4000
2000
0Farm
land
and
cons
truc
tion
area
Farm land area (101 ha)Construction area (ha)
Figure 13 Temporal trend of the farmland area and the floor spaceof the buildings under construction
43 Changes in the Relative Humidity and the Vapor PressureDeficit due to Urban Development The vapor pressure deficitwas calculated by the air temperature and the relative humid-ity following themethod of Allen et al (1998) [12] It is clearlyshown in the calculation method that the vapor pressuredeficit (VPD) increases with air temperature and decreaseswith relative humidity In this study both relative humidityand mean air temperature varied slightly during the periodfrom 1954 to 1980 (Figures 5 and 6) which consequentlyresulted in a slight change in the vapor pressure deficit Afterthat period the mean air temperature increased significantlyand the relative humidity was decreased remarkably Thesetrends resulted in an increase in the vapor pressure deficit(Figure 7)
The relative humidity is defined as the ratio of the watervapor density (mass per unit volume) to the saturation watervapor density and it is also approximately the ratio of theactual to the saturation vapor pressure A greater evaporativearea may produce more water vapor for the atmosphere andthen increase the water vapor density as well as the relativehumidity at a given temperature In the city study area thefarmland area decreased whereas the construction areaincreased (Figure 13) These data indicate a decreasing trendin the evaporative area which may result in a decrease in thewater vapor density as well as the relative humidity [33] Sim-ilarly a decreasing trend of relative humidity and an increasein the vapor pressure deficit were observed in Beijing Datongin Shanxi province Zhang Jiakou in Hebei Province and BetDagan in Israel [15 34 35] In Beijing it was shown thatthe VPD increased with air temperature and decreased withrelative humidity from 1951 to 2010 [15] Cohen et al (2002)showed that themain factor responsible for the increased panevaporation was the growth in the aerodynamic componentof evaporation which was due to increases in both the airVPD and the wind speed at Bet Dagan from 1964 to 1997 [35]
The vapor pressure deficit represents a gradient acrosswhich water vapor is removed from the evapotranspiringsurface into the surrounding air [12] A greater vapor pressuredeficit generally causes a higher evaporative rate Hence the
increasing vapor pressure deficit in the Shenzhen area willresult in increasing plant evapotranspiration
44 ET0 Change and City Development In the current studyarea ET
0first decreased from 1950s to 1970s and then
increased greatly in the 1980s During the 1990s and 2000s itvaried slightly with amean value of 1287mm It was observedthat after the onset of urban development in 1978 the ET
0
value increased and became higher than this for the periodprior to the urban development Figures 2 5 6 and 8 showthat the mutation points for most climatic variables wereobserved near the onset year of urban development and sen-sitivity analysis shows that the higher ET
0during the period
of 1992ndash2012 is mainly attributed to the relative humiditydecrease and air temperature increaseHence it could be con-cluded that the quick development of Shenzhen city alteredthe climatic conditions and hence increased the local ET
0
In other large cities an increasing ET0trendwas also found in
recent decades For example in Beijing City the annual ET0
increased significantly from 1951 to 2010 and from the 1950sto 2000s it increased from 1039 to 1148mm [15] The annualpotential evapotranspiration (PET) displays a significantupward trend from 1970 to 2006 and the trend varied from 1to 4mm per year in the Pyrenees-Orientales and Audeadministrative departments respectively and the westernpart of the French Mediterranean area with an averageincrease in PET of between 34mm and 150mm in the last 36years [36]
The increasing trend in ET0that was observed in large
cities is different than this that was found in the farmlandFor example in the Haihe River basin in northern Chinadecreasing trends were observed in 26 stations while 16stations showed significant decreasing trends from 1950 to2007 with rates from minus20 to minus37mmyearminus1 [34] Similarly asignificant decreasing trend of ET
0with a rate of minus3mm per
yearwas found in the arid region of northwest China [37]Thedifference trend in ET
0between the large cities and the
farmland may be due to the variations in the energy balanceand the evaporative potential In the farmland areas morethan 60ndash80 of the net radiation is used for plant evapotran-spiration [33 38 39]This effect not only reduces the availableheat for heating the air environment but also increasesthe water vapor in the near atmosphere The latter effect mayincrease the relative humidity and reduce the vapor pressuredeficit and lastly it may reduce the reference evapotranspi-ration For urban conditions the decrease in green land willdecrease the energy consumption caused by crop evapotran-spiration increasing the available heat and decreasing thewater vapor which ultimately results in an increase in theevaporative potential
It should be noted in Figure 9 that the ET0values in 1997
and 2012 were much lower than those in the neighboringyears It is estimated that ET
0in 1997 and 2012 decreased
by minus104 and minus130 compared to the mean value for theperiod from 1990 to 2012 The sensitive analysis methoddescribed in Section 33 was used to determine the effects ofeach variable on the ET
0changes in 1997 and 2012 compared
to the mean value during the period of 1990ndash2012 The
10 Advances in Meteorology
results showed that the changes in the relative humidity airtemperature sunshine hours andwind speed in 1997 resultedin changes in ET
0by minus72 minus08 minus20 and minus05 respectively
and by minus126 minus08 minus06 and minus42 in 2012 respectively Itcould be concluded that the increase in the relative humidityis the main factor for ET
0reduction followed by the wind
speed air temperature and sunshine hours in the two yearsBased on the mean values of the climatic variables
averaged over the periods of 1954ndash1978 and 1992ndash2012 ET0
increased by 145 in the latter period For the sensitivityanalysis changes in the relative humidity air temperaturewind speed and sunshine hours during 1992ndash2012 caused thevariation of ET
0by 146 50 minus13 and minus38 respectively
compared to those for the period 1954ndash1978The total amountof change in ET
0was 156 based on the sensitivity analysis
This value is similar to the rate of increase of 145 by com-parison of the ET
0values between the two time periods Liu
et al (2014) [15] calculated the ET0change rates by directly
comparing the mean values and summing each ET0change
rate caused by climatic variables using the same sensitiveanalysis method the ET
0change rates were 107 and 105
respectively Liu et al (2009) [40] found that ET0inside the
screenhouse was reduced by 39 compared to that in theopen field By considering the effect of each climatic variablechange to ET
0using the sensitivity analysis the total ET
0
change rate sums to 44 which is similar to the value of39 Therefore it could be concluded that the sensitivity-analysis method used in this study is reliable and easy to useand hence it is recommended for the analysis of the effect ofclimate change on ET
0
5 Conclusions
(1) The development of Shenzhen city greatly affected thelocal climatic conditions Before the onset of urbandevelopment each climatic variable varied slightlywhereas afterward the air temperature increased sig-nificantly and the sunshine hours and relative humid-ity decreased significantly The mutation point formost climatic variables is observed at approximately1978 the onset year for urban development
(2) ET0first decreased from 1954 to 1978 and then
increased quickly and reached a maximal value of1373mm during the period from 1992 to 2012 Themean ET
0value for the period from 1954 to 1978 was
1110mm and increased to 1284mm during the periodfrom 1992 to 2012 indicating an increasing trend ofthe evaporative demand
(3) Sensitivity analysis showed that ET0is most sensitive
to relative humidity followed by air temperaturesunshine hours and wind speed
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This paper is partially supported by the National ScienceFoundation of China (Grant nos 51179005 51479004) Theauthors greatly acknowledge the comments from the editorand the two anonymous reviewers
References
[1] IPCC Climate Change 2014 Synthesis Report An Assessmentof Intergovernmental Panel on Climate Change IPCC GenevaSwitzerland 2014 httpipccchindexhtml
[2] J D Haskett Y A Pachepsky and B Acock ldquoEffect of climateand atmospheric change on soybean water stress a study ofIowardquo Ecological Modelling vol 135 no 2-3 pp 265ndash277 2000
[3] T G Huntington ldquoEvidence for intensification of the globalwater cycle review and synthesisrdquo Journal of Hydrology vol 319no 1ndash4 pp 83ndash95 2006
[4] B C Bates Z W Kundzewicz S Wu and J Palutikof ldquoClimatechange and waterrdquo Technical Paper of the IntergovernmentalPanel on Climate Change IPCC Secretariat Geneva Switzer-land 2008
[5] C M Philandras D A Metaxas and P T Nastos ldquoClimatevariability and urbanization in AthensrdquoTheoretical and AppliedClimatology vol 63 no 1-2 pp 65ndash72 1999
[6] R L Wilby ldquoPast and projected trends in Londonrsquos urban heatislandrdquoWeather vol 58 no 7 pp 251ndash260 2003
[7] N Schwarz U Schlink U Franck and K Groszligmann ldquoRela-tionship of land surface and air temperatures and its implica-tions for quantifying urban heat island indicatorsmdashan applica-tion for the city of Leipzig (Germany)rdquoEcological Indicators vol18 pp 693ndash704 2012
[8] H Wang L Fu X Lin Y Zhou and J C Chen ldquoA bottom-up methodology to estimate vehicle emissions for the Beijingurban areardquo Science of the Total Environment vol 407 no 6 pp1947ndash1953 2009
[9] DESE (Department of Environmental Science and Engineer-ingTsinghua University) Mobile Source Database EmissionInventory and Treatment Proposal for Beijing Tsinghua Univer-sity Beijing China 2005
[10] H Kan S J London G Chen et al ldquoDifferentiating the effectsof fine and coarse particles on daily mortality in ShanghaiChinardquo Environment International vol 33 no 3 pp 376ndash3842007
[11] A Aziz and I U Bajwa ldquoErroneous mass transit system andits tended relationship with motor vehicular air pollution (Anintegrated approach for reduction of urban air pollution inLahore)rdquo Environmental Monitoring and Assessment vol 137no 1ndash3 pp 25ndash33 2008
[12] R G Allen L S Perreira D Raes and M Smith Crop Evap-otranspiration Guidelines for Computing Crop Water Require-ments FAO Irrigation and Drainage Paper no 56 FAO RomeItaly 1998
[13] K H Hamed ldquoTrend detection in hydrologic data the Mann-Kendall trend test under the scaling hypothesisrdquo Journal ofHydrology vol 349 no 3-4 pp 350ndash363 2008
[14] L Q Liang L J Li andQ Liu ldquoTemporal variation of referenceevapotranspiration during 1961ndash2005 in the Taoer River basin ofNortheast Chinardquo Agricultural and Forest Meteorology vol 150no 2 pp 298ndash306 2010
Advances in Meteorology 11
[15] H Liu Y Li T Josef R H Zhang and G H HuangldquoQuantitative estimation of climate change effects on potentialevapotranspiration in Beijing during 1951ndash2010rdquo Journal ofGeographical Sciences vol 24 no 1 pp 93ndash112 2014
[16] M G Kendall and A StuartThe Advanced Theory of StatisticsGriffin London UK 1973
[17] Q-Y Tang and C-X Zhang ldquoData Processing System (DPS)software with experimental design statistical analysis and datamining developed for use in entomological researchrdquo InsectScience vol 20 no 2 pp 254ndash260 2013
[18] M Moller J Tanny Y Li and S Cohen ldquoMeasuring andpredicting evapotranspiration in an insect-proof screenhouserdquoAgricultural and Forest Meteorology vol 127 no 1-2 pp 35ndash512004
[19] G Kitsara G Papaioannou A Papathanasiou and A RetalisldquoDimmingbrightening in athens trends in sunshine durationcloud cover and reference evapotranspirationrdquoWater ResourcesManagement vol 27 no 6 pp 1623ndash1633 2013
[20] Y Wang Y Yang S Han Q X Wang and G H ZhangldquoSunshine dimming and brightening in Chinese cities (1955ndash2011) was driven by air pollution rather than cloudsrdquo ClimateResearch vol 56 no 1 pp 11ndash20 2013
[21] G Stanhill and S Cohen ldquoGlobal dimming a review of theevidence for a widespread and significant reduction in globalradiation with discussion of its probable causes and possibleagricultural consequencesrdquoAgricultural and ForestMeteorologyvol 107 no 4 pp 255ndash278 2001
[22] Q Liu and Z Yang ldquoQuantitative estimation of the impact ofclimate change on actual evapotranspiration in the Yellow RiverBasin Chinardquo Journal of Hydrology vol 395 no 3-4 pp 226ndash234 2010
[23] D I Stern ldquoReversal of the trend in global anthropogenic sulfuremissionsrdquoGlobal Environmental Change vol 16 no 2 pp 207ndash220 2006
[24] I Koren J V Martins L A Remer and H Afargan ldquoSmokeinvigoration versus inhibition of clouds over the AmazonrdquoScience vol 321 no 5891 pp 946ndash949 2008
[25] D Rosenfeld Y J Kaufman and I Koren ldquoSwitching cloudcover and dynamical regimes from open to closed Benard cellsin response to the suppression of precipitation by aerosolsrdquoAtmospheric Chemistry and Physics vol 6 no 9 pp 2503ndash25112006
[26] C Ruckstuhl R Philipona K Behrens et al ldquoAerosol and cloudeffects on solar brightening and the recent rapid warmingrdquoGeophysical Research Letters vol 35 no 12 Article ID L127082008
[27] DG Streets Y Fang CMian et al ldquoAnthropogenic andnaturalcontributions to regional trends in aerosol optical depth 1980ndash2006rdquo Journal of Geophysical Research Atmospheres vol 114 no10 Article ID D00D18 2009
[28] V Ramanathan P J Crutzen J T Kiehl and D RosenfeldldquoAtmospheremdashaerosols climate and the hydrological cyclerdquoScience vol 294 no 5549 pp 2119ndash2124 2001
[29] M Wild ldquoEnlightening global dimming and brighteningrdquoBulletin of the AmericanMeteorological Society vol 93 no 1 pp27ndash37 2012
[30] G D Liu Y Li H J Liu and J Xiao ldquoChanging trend of refer-ence crop evapotranspiration and its dominatedmeteorologicalvariables in Shanxi province in the past 55 yearsrdquo Journal ofIrrigation and Drainage vol 31 no 4 pp 26ndash30 2012
[31] C-S Rim ldquoThe effects of urbanization geographical and topo-graphical conditions on reference evapotranspirationrdquo ClimaticChange vol 97 no 3 pp 483ndash514 2009
[32] WKuang Y Liu YDou et al ldquoWhat are hot andwhat are not inan urban landscape quantifying and explaining the land surfacetemperature pattern in Beijing Chinardquo Landscape Ecology2014
[33] Z Qin Q Yu S Xu et al ldquoWater heat fluxes and water useefficiency measurement and modeling above a farmland in theNorth China Plainrdquo Science in China D Earth Sciences vol 48no 1 pp 207ndash217 2005
[34] B Tang L Tong S Z Kang and L Zhang ldquoImpacts ofclimate variability on reference evapotranspiration over 58 yearsin the Haihe river basin of north Chinardquo Agricultural WaterManagement vol 98 no 10 pp 1660ndash1670 2011
[35] S Cohen A Ianetz and G Stanhill ldquoEvaporative climatechanges at BetDagan Israel 1964ndash1998rdquoAgricultural and ForestMeteorology vol 111 no 2 pp 83ndash91 2002
[36] K Chaouche L Neppel C Dieulin et al ldquoAnalyses of precipita-tion temperature and evapotranspiration in a French Mediter-ranean region in the context of climate changerdquoComptes RendusGeoscience vol 342 no 3 pp 234ndash243 2010
[37] Z Huo X Dai S Feng S Kang and G Huang ldquoEffect of cli-mate change on reference evapotranspiration and aridity indexin arid region of Chinardquo Journal of Hydrology vol 492 pp 24ndash34 2013
[38] G Peng X Cai H Zhang A Li F Hu and M Y LeclercldquoHeat flux apportionment to heterogeneous surfaces using fluxfootprint analysisrdquo Advances in Atmospheric Sciences vol 25no 1 pp 107ndash116 2008
[39] Y Q Zhang Y J Shen C M Liu et al ldquoMeasurement andanalysis of water heat and CO
2flux from a farmland in the
North China plainrdquo Acta Geographica Sinica vol 57 no 3 pp333ndash342 2002 (Chinese)
[40] H-J Liu G-H Huang S Cohen and J Tanny ldquoChange in cropevapotranspiration and associated influencing factors underscreenhouse conditionsrdquo Chinese Journal of Eco-Agriculturevol 17 no 3 pp 484ndash488 2009 (Chinese)
By analyzing the statistical series UF119896and UB
119896 the
change trend of series 119909119894can be further analyzed and the
mutation time and region can be determined UF119896gt 0 indi-
cates that the series tends to increase and UF119896lt 0 indicates
that the series tends to decrease When the series exceed thecredibility line they exhibit an obvious increasing or decreas-ing trend If an intersection point exists between the curvesof UF
119896and UB
119896and falls between the credibility lines the
corresponding time of the intersection point is the startingmoment of mutation
In this study the statistical analysis software DPS (versionDPS 145 developed by Zhejiang University China) was usedto analyze the mutation point of the temporal trends of ET
0
and the meteorological variables [17] The credibility line isdrawn at the significance level of 120572 = 001
233 Sensitivity-Analysis Method Sensitivity analysis wasemployed to identify the climatic variables that most stronglyinfluence ET
0following themethod proposed byMoller et al
(2004) and Liu et al (2014) [15 18] In this study the temporaltrends of most climatic variables showed abrupt change in1978 and the climatic variables during 1954ndash1978 were sig-nificantly (119875 lt 001) higher or lower than those during 1992ndash2012 see Sections 311 to 316 in the following text Hencedata during these two periods were used to assess the climatechange impact on ET
0 The detailed processes of the sensitiv-
ity analysis are described as follows (1) themean values of theair temperature relative humidity wind speed and sunshinehours in each day of a year were calculated using climatic datafrom 1954 to 1978 and the corresponding daily and annualET0were calculated using these mean daily values and the
Penman-Monteith method The mean daily value for eachclimatic variablewas set as the reference climatic variable andthe calculated ET
0was set as the reference ET
0 (2)The sen-
sitivity of each climatic variable to ET0was analyzed by com-
paring the reference ET0and the ET
0calculated by changing
one variable with a rate of minus15 minus10 minus5 5 10 and 15 andkeeping the other variables identical to the reference climaticvariables Then a figure was drawn based on these data(Figure 10) (3) The mean values of the air temperaturerelative humidity wind speed and sunshine hours in each dayof a yearwere calculated using climatic data from 1992 to 2012The corresponding ET
0was calculated by setting one climatic
variable as those during 1992 and 2012 and the others as thereference climatic variable (in step (1)) and then calculatingET0and comparing it with the reference ET
0 The relative
change in ET0caused by each climatic change during 1992
to 2012 is marked in Figure 10 The most sensitive variable toa change in ET
0is determined by comparing the relative ET
0
changes caused by each variable
3 Results
31 Annual Distributions and Trends in theChanges of Climatic Variables
311 Sunshine Hours and Solar Radiation The annual sun-shine duration and annual total radiation during the period
1954 1962 1970 1978 1986 1994 2002 2010
Year
y = minus10077x + 21997
R2 = 0525
2700
2400
2100
1800
1500Tota
l sun
shin
e hou
rs (h
)
(a)
UF-
UB
UFUB
1954 1962 1970 1978 1986 1994 2002 2010
Year
minus8
minus4
4
0
(b)
Figure 2 Temporal change in the yearly sunshine duration (a) andthe trend results from the Mann-Kendall method (b)
1954 1962 1970 1978 1986 1994 2002 2010
Year
y = minus87059x + 21314
R2 = 0558
4600
4400
4200
4000
3800
3600
Tota
l sol
ar ra
diat
ion
(MJmiddotm
minus2)
(a)
UF-
UB
UFUB
1954 1962 1970 1978 1986 1994 2002 2010
Year
minus8
minus4
4
0
(b)
Figure 3 Temporal trend in the yearly total solar radiation (a) andthe trend results from the Mann-Kendall method (b)
from 1954 to 2012 are given in Figures 2 and 3 It can beseen in Figure 2 that the annual sunshine duration showed
Advances in Meteorology 5
y = minus06754x + 32529
R2 = 00008
3000
2500
2000
1500
1000
500Tota
l pre
cipi
tatio
n (m
m)
1954 1962 1970 1978 1986 1994 2002 2010
Year
(a)
UF-
UB
UFUB
1954 1962 1970 1978 1986 1994 2002 2010
Year
minus8
minus4
4
0
(b)
Figure 4 Temporal trend of the yearly total precipitation (a) andthe trend results from the Mann-Kendall method (b)
an obvious decline which decreased from 2651 to 1590 hourswith an average of 2014 hours (Figure 2(a))Through the two-tailed test (UF and UB lines) we found that the decreasingtrend of the sunshine hours is obvious for the period from1982 to 2012 and the increasing trend is obvious for the periodof 1954 to 1973 Similar to the trend for the sunshine durationthe solar radiation also showed an obvious declining trendwhich ranges from 3672 to 4520MJsdotmminus2 with an average of4050MJsdotmminus2 (Figure 3(a)) The two-tailed test showed thatthe solar radiation values from 1982 to 2012 were significantlylower than those from 1954 to 1973 The shifting mutationpoints of the sunshine duration and the solar radiation areall found in 1978 using the Mann-Kendall mutational test(119875 lt 001) (Figures 2 and 3(b))
312 Precipitation The annual total precipitation valuesduring the period from 1954 to 2012 are given in Figure 4 Itcan be seen in Figure 4 that the highest precipitation valuewas 2747mm in 2001 and the lowest was 912mm in 1963(Figure 4(a)) The annual mean was 1914mm during theperiod from 1954 to 2012 The precipitation varied greatlyover the years whereas the trend test showed that the tempo-ral trend of the annual precipitation is not significant and amutation point has not been tested (Figure 4(b))
313 Relative Humidity The daily mean relative humidityduring the period from 1954 to 2012 is given in Figure 5 Itcan be seen that the daily mean relative humidity increasedslightly in the first 20 years followed by an obvious decline
1954 1962 1970 1978 1986 1994 2002 2010
Year
y = minus01389x + 35181
R2 = 05331
84
80
76
72
68Mea
n re
lativ
e hum
idity
()
(a)
UF-
UB
UFUB
1954 1962 1970 1978 1986 1994 2002 2010
Year
minus8
minus4
4
0
(b)
Figure 5 Temporal trend of the mean relative humidity (a) and thetrend results from the Mann-Kendall method (b)
after 1978 the range is from 691 to 822 and the average is764 (Figure 5(a)) From the Mann-Kendall test (119875 lt 001)we found that the averaged relative humidity value during theperiod from 1992 to 2012 is significantly (119875 lt 001) lower thanthe period from 1954 to 1992
314 Air Temperature Themean air temperature during theperiod from 1954 to 2012 is shown in Figure 6 Accordingto Figure 6 the mean air temperature showed an obviouslyincreasing trend and the values range from 215 to 239∘Cwith an average value of 226∘C (Figure 6(a)) The statisticalresult showed that the mean air temperature over the periodfrom 1990 to 2012 is significantly (119875 lt 001) higher than thoseover the period from 1954 to 1983 (Figure 6(b))
315 Vapor Pressure Deficit Figure 7(a) shows the temporaltrend of the vapor pressure deficit (VPD) during the periodfrom 1954 to 2012 VPD ranges from 0449 to 0867 kPa andthe average is 0648 kPa There is no significant temporaltrend for VPD from 1954 to 1977 and after 1978 VPDincreased greatly It was found that the average VPD duringthe period from 1992 to 2012 is significantly (119875 lt 001) higherthan the values from 1954 to 1991 according to the Mann-Kendall mutational test
316 Wind Speed The annual mean wind speed variedgreatly during the period from 1954 to 2012 (Figure 8(a))Clearly decreasing trends were found from 1954 to 1977 andfrom 1987 to 2012 whereas increasing trend was found from
6 Advances in Meteorology
1954 1962 1970 1978 1986 1994 2002 2010
Year
y = 003x minus 36919
R2 = 06208
24
24
23
23
22
22
21
Mea
n te
mpe
ratu
re(∘
C)
(a)
UF-
UB
UFUB
1954 1962 1970 1978 1986 1994 2002 2010
Year
minus8
minus4
4
8
0
(b)
Figure 6 Temporal trend of themean temperature (a) and the trendresults from the Mann-Kendall method (b)
1954 1962 1970 1978 1986 1994 2002 2010
Year
y = 00055x minus 10261
R2 = 06721
100
040
020
060
080
Vapo
r pre
ssur
e defi
cit (
kPa)
(a)
UF-
UB
UFUB
1954 1962 1970 1978 1986 1994 2002 2010
Year
minus8
minus4
4
8
0
(b)
Figure 7 Temporal trend of the mean vapor pressure deficit (a) andthe trend results from the Mann-Kendall method (b)
1978 to 1986 It can be seen in Figure 8(a) that the maximumwind speed was 372ms in 1954 the minimum was 178ms
1954 1962 1970 1978 1986 1994 2002 2010
Year
y = minus00098x + 22013
R2 = 01214
40
30
20
10Win
d sp
eed
(mmiddotsminus
1)
(a)
UF-
UB
UFUB
1954 1962 1970 1978 1986 1994 2002 2010
Year
minus8
minus4
4
0
(b)
Figure 8 Temporal trend of the mean wind speed (a) and the trendresults from the Mann-Kendall method (b)
in 1977 and the annual mean was 260ms The statisticalresults show that there were no significant temporal trendsfrom 1954 to 2012
32 Reference Evapotranspiration (ET0) Theannual total ET0
varied from 946 to 1373mm with a total mean of 1187mmFigure 9(a) shows the change in the annual ET
0over the
past 59 years According to Figure 9(a) ET0firstly decreased
gradually from 1954 to 1978 then increased from 1978 to 1992and lastly varied slightly after 1992 From 1954 to 1978 theannual total ET
0decreased from 1285 to 946mmwith amean
value of 1110mm afterwards the annual total ET0increased
from 1118 to 1373mm with a mean value of 1284mm duringthe period from 1992 to 2012 The mean ET
0in the period
of 1992ndash2012 increased by 156 over those in the period of1954ndash1978 indicating a great increase in evaporation poten-tial The statistical results based on the Mann-Kendall test(119875 lt 001) (Figure 9(b)) showed that ET
0during the period
from 1972 to 1987 was significantly lower than that from 2001to 2012 which was significantly higher than those in otherperiods The shifting mutation point for ET
0is found at 1992
by the Mann-Kendall test (Figure 9(b))For analyzing the yearly ET
0distribution the total ET
0for
eachmonthwas calculated and the averagedmonth total ET0s
in period of 1954ndash1978 and 1992ndash2012were calculated respec-tively The monthly ET
0distributions in these two periods
and in all study period (1954ndash2012) are showed in Figure 10The highest monthly ET
0generally appears in July and
August and the lowest in January and FebruaryMonthly ET0
from May to November are generally higher than 100mm
Advances in Meteorology 7
Table 1 Mean values of each climatic variable in the periods of 1954ndash1978 and 1992ndash2012 and changes of each climatic variable to ET0variation
Air temperature Relative humidity Sunshine hours Wind speed∘C Hoursday ms
Mean values1954ndash1978 220 789 608 2681992ndash2012 233 732 504 252
Average ET0 over 1954ndash1978Average ET0 over 1992ndash2012
(a)
UF-
UB
UFUB
1954 1962 1970 1978 1986 1994 2002 2010
Year
minus8
minus4
4
8
0
(b)
Figure 9 Temporal trend of the yearly total ET0(a) and the trend
results from the Mann-Kendall method (b)
and the total ET0in this period accounts for approximately
13 the yearly total Monthly ET0s in the period of 1992ndash2012
are all higher by 5ndash30mm or 7ndash25 than those in the 1954ndash1978 period The most increases in monthly ET
0are found in
period from July to September and the lowest from January toApril with an increase of less than 7mmTherefore the greatincrease in ET
0in summer (generally from June to October)
makes main contribution to yearly ET0
33 Sensitivity Analysis of ET0 to the Change of ClimaticVariables Figures 2 to 8 show that most mutation pointsfor most of the climatic variables were found during theperiod from 1978 to 1992 and the mean values of the climaticvariables over the period from 1954 to 1978 were significantly
150
100
50
01 2 3 4 5 6 7 8 9 10 11 12
Months
Monthly mean from period of 1954ndash2012
Monthly mean from period of 1954ndash1978Monthly mean from period of 1992ndash2012
50
40
30
20
10
0Mon
thly
tota
ldie
renc
eET 0
(mm
)
Incr
ease
per
cent
age i
nET
0(
)
Increase in ET0 from 1954ndash1978 to 1992ndash2012Increase percentage in ET0
Figure 10 Yearly distribution of ET0averaged in the periods of
1954ndash1978 1992ndash2012 and 1954ndash2012 ET0increase amount and the
corresponding increase percentage in each month from periods of1954ndash1978 to 1992ndash2012 were showed
(119875 lt 001) higherlower than those during the period from1992 to 2012 Hence the mean values of each variable duringthe two periods from 1954 to 1978 and from 1992 to 2012 werecalculated and used to analyze their change effects on the ET
0
changes following themethod described in Section 233Thesummary of the mean values of each climate and their effectson ET
0were listed in Table 1 and Figure 11
It can be seen in Table 1 that the daily mean relativehumidity daily mean air temperature wind speed and sun-shine hours from the period of 1954ndash1978 to 1992ndash2012 wereminus72 59 minus60 and minus171 respectively which resulted inET0changes by 146 53 minus13 and minus32 respectively The
contribution of each climate variablersquos variation from 1954ndash1978 to 1992ndash2012 to ET
0change is shown in Figure 11 It
can be found that decrease in relative humidity accounted forapproximately 60 variation in ET
0 followed by temperature
increase with a contribution of 22 and sunshine hoursreduction of 13 Wind speed accounted for 6 variationin ET
0 Similarly in another mega city Beijing in China the
order of climate change to ET0variation frommain to weak is
air temperature relative humidity sunshine hours and windspeed [14]
8 Advances in Meteorology
Relative humidity
60Wind speed6
Temperature22
Sunshine hours13
Figure 11 The contribution of each climatic variable change to ET0variation
300
200
100
0
Dev
elopm
ent o
f She
nzhe
n ci
ty
Heavy industry output value (billion dollars)Resident population (10minus1 million)Electric energy production (102 MWh)
1954 1962 1970 1978 1986 1994 2002 2010
Year
Figure 12 Temporal trend of the development of Shenzhen city
4 Discussion
41 Change in the Sunshine Hours and Urban DevelopmentSunshine hours generally depend on the cloud cover man-made aerosols and certain air pollutants (including SO
2
NO119909 and PM) [19 20] Recent studies indicated that the
most probable cause for the depression of sunshine hours orsolar radiation is in the increased concentrations ofmanmadeaerosols and other air pollutants [1 19ndash22] In this studythe sunshine hours and amount of radiation showed cleardecreasing trends after 1970 and theywere significantly lowerafter 1980 during rapid urban development
Figure 12 shows the increasing trends of the residentpopulation heavy industry output value and electric energyproduction It is confirmed that the increasing trend in theenergy consumption corresponds to increased emission ofpolluted particles including CO CO
2 NO119909 SO119909 PM25 O3
andHCThese particlesmay result in an increase in the atmo-spheric aerosol concentration [23ndash27] which can directlyattenuate the surface solar radiation (SSR) by scattering andabsorbing solar radiation (direct effect) or can indirectlyattenuate SSR by their ability to act as cloud condensation
nuclei thereby increasing the cloud reflectivity and lifetime(first and second indirect effects) [24 28] A remarkabledecline in SSR between the 1950s and the 1980s was found inseveral studies that were performed at selected observationstations based on sites in Europe the Baltics the South PoleGermany and the former Soviet Union [21] Today a com-prehensive literature exists that confirms the declines of SSRduring this period in many places around the world [29]In Beijing and several other Chinese cities the decline ofsunshine hours and SSR has also been reported [15 30]
42 Air Temperature Change and Urban Development It hasbeen confirmed that there is a large air temperature differ-ence in urban and rural areas [4ndash7] These air temperaturedifferences result from the influence of the thermal emissivityproperties of urban surfaces and the three-dimensional con-figuration and heat capacity of erected structures onto the airtemperature patterns in an urban region [7 15 31]Most of theworldrsquos cities thus show higher air temperatures in the urbancore than in the surrounding rural areas [4 5 7] For examplein Beijing City China Kuang et al [32] measured that themean land surface temperature of urban impervious surfaceswas about 6ndash12∘C higher than that of the urban green spaceand that in built-up areaswas on average 3ndash6∘Chigher than inrural areas They showed the main reason is the higher ratioof sensible heat to net radiation (063) and lower ratio of thelatent heat to net radiation (019) on the urban impervioussurface as compared to the corresponding rates of 030 and063 in green space and cropland In this study there is noobvious temporal trend in the air temperature before 1978whereas after that time the temperature increased greatlyalthough it showed a slight declining trend in recent yearsThe large air temperature increase from 1978 to 2002 may bedue to the increasing area of construction and the decrease infarmland [32] Figure 13 shows the changes in the farmlandand construction areas It can be found that there was arapid increase in construction areas and an abrupt decreasein farmland areas after 1978 which led to a change in thethermal balance and then resulted in the increase in the airtemperature in Shenzhen city
Advances in Meteorology 9
1954 1962 1970 1978 1986 1994 2002 2010
Year
8000
6000
4000
2000
0Farm
land
and
cons
truc
tion
area
Farm land area (101 ha)Construction area (ha)
Figure 13 Temporal trend of the farmland area and the floor spaceof the buildings under construction
43 Changes in the Relative Humidity and the Vapor PressureDeficit due to Urban Development The vapor pressure deficitwas calculated by the air temperature and the relative humid-ity following themethod of Allen et al (1998) [12] It is clearlyshown in the calculation method that the vapor pressuredeficit (VPD) increases with air temperature and decreaseswith relative humidity In this study both relative humidityand mean air temperature varied slightly during the periodfrom 1954 to 1980 (Figures 5 and 6) which consequentlyresulted in a slight change in the vapor pressure deficit Afterthat period the mean air temperature increased significantlyand the relative humidity was decreased remarkably Thesetrends resulted in an increase in the vapor pressure deficit(Figure 7)
The relative humidity is defined as the ratio of the watervapor density (mass per unit volume) to the saturation watervapor density and it is also approximately the ratio of theactual to the saturation vapor pressure A greater evaporativearea may produce more water vapor for the atmosphere andthen increase the water vapor density as well as the relativehumidity at a given temperature In the city study area thefarmland area decreased whereas the construction areaincreased (Figure 13) These data indicate a decreasing trendin the evaporative area which may result in a decrease in thewater vapor density as well as the relative humidity [33] Sim-ilarly a decreasing trend of relative humidity and an increasein the vapor pressure deficit were observed in Beijing Datongin Shanxi province Zhang Jiakou in Hebei Province and BetDagan in Israel [15 34 35] In Beijing it was shown thatthe VPD increased with air temperature and decreased withrelative humidity from 1951 to 2010 [15] Cohen et al (2002)showed that themain factor responsible for the increased panevaporation was the growth in the aerodynamic componentof evaporation which was due to increases in both the airVPD and the wind speed at Bet Dagan from 1964 to 1997 [35]
The vapor pressure deficit represents a gradient acrosswhich water vapor is removed from the evapotranspiringsurface into the surrounding air [12] A greater vapor pressuredeficit generally causes a higher evaporative rate Hence the
increasing vapor pressure deficit in the Shenzhen area willresult in increasing plant evapotranspiration
44 ET0 Change and City Development In the current studyarea ET
0first decreased from 1950s to 1970s and then
increased greatly in the 1980s During the 1990s and 2000s itvaried slightly with amean value of 1287mm It was observedthat after the onset of urban development in 1978 the ET
0
value increased and became higher than this for the periodprior to the urban development Figures 2 5 6 and 8 showthat the mutation points for most climatic variables wereobserved near the onset year of urban development and sen-sitivity analysis shows that the higher ET
0during the period
of 1992ndash2012 is mainly attributed to the relative humiditydecrease and air temperature increaseHence it could be con-cluded that the quick development of Shenzhen city alteredthe climatic conditions and hence increased the local ET
0
In other large cities an increasing ET0trendwas also found in
recent decades For example in Beijing City the annual ET0
increased significantly from 1951 to 2010 and from the 1950sto 2000s it increased from 1039 to 1148mm [15] The annualpotential evapotranspiration (PET) displays a significantupward trend from 1970 to 2006 and the trend varied from 1to 4mm per year in the Pyrenees-Orientales and Audeadministrative departments respectively and the westernpart of the French Mediterranean area with an averageincrease in PET of between 34mm and 150mm in the last 36years [36]
The increasing trend in ET0that was observed in large
cities is different than this that was found in the farmlandFor example in the Haihe River basin in northern Chinadecreasing trends were observed in 26 stations while 16stations showed significant decreasing trends from 1950 to2007 with rates from minus20 to minus37mmyearminus1 [34] Similarly asignificant decreasing trend of ET
0with a rate of minus3mm per
yearwas found in the arid region of northwest China [37]Thedifference trend in ET
0between the large cities and the
farmland may be due to the variations in the energy balanceand the evaporative potential In the farmland areas morethan 60ndash80 of the net radiation is used for plant evapotran-spiration [33 38 39]This effect not only reduces the availableheat for heating the air environment but also increasesthe water vapor in the near atmosphere The latter effect mayincrease the relative humidity and reduce the vapor pressuredeficit and lastly it may reduce the reference evapotranspi-ration For urban conditions the decrease in green land willdecrease the energy consumption caused by crop evapotran-spiration increasing the available heat and decreasing thewater vapor which ultimately results in an increase in theevaporative potential
It should be noted in Figure 9 that the ET0values in 1997
and 2012 were much lower than those in the neighboringyears It is estimated that ET
0in 1997 and 2012 decreased
by minus104 and minus130 compared to the mean value for theperiod from 1990 to 2012 The sensitive analysis methoddescribed in Section 33 was used to determine the effects ofeach variable on the ET
0changes in 1997 and 2012 compared
to the mean value during the period of 1990ndash2012 The
10 Advances in Meteorology
results showed that the changes in the relative humidity airtemperature sunshine hours andwind speed in 1997 resultedin changes in ET
0by minus72 minus08 minus20 and minus05 respectively
and by minus126 minus08 minus06 and minus42 in 2012 respectively Itcould be concluded that the increase in the relative humidityis the main factor for ET
0reduction followed by the wind
speed air temperature and sunshine hours in the two yearsBased on the mean values of the climatic variables
averaged over the periods of 1954ndash1978 and 1992ndash2012 ET0
increased by 145 in the latter period For the sensitivityanalysis changes in the relative humidity air temperaturewind speed and sunshine hours during 1992ndash2012 caused thevariation of ET
0by 146 50 minus13 and minus38 respectively
compared to those for the period 1954ndash1978The total amountof change in ET
0was 156 based on the sensitivity analysis
This value is similar to the rate of increase of 145 by com-parison of the ET
0values between the two time periods Liu
et al (2014) [15] calculated the ET0change rates by directly
comparing the mean values and summing each ET0change
rate caused by climatic variables using the same sensitiveanalysis method the ET
0change rates were 107 and 105
respectively Liu et al (2009) [40] found that ET0inside the
screenhouse was reduced by 39 compared to that in theopen field By considering the effect of each climatic variablechange to ET
0using the sensitivity analysis the total ET
0
change rate sums to 44 which is similar to the value of39 Therefore it could be concluded that the sensitivity-analysis method used in this study is reliable and easy to useand hence it is recommended for the analysis of the effect ofclimate change on ET
0
5 Conclusions
(1) The development of Shenzhen city greatly affected thelocal climatic conditions Before the onset of urbandevelopment each climatic variable varied slightlywhereas afterward the air temperature increased sig-nificantly and the sunshine hours and relative humid-ity decreased significantly The mutation point formost climatic variables is observed at approximately1978 the onset year for urban development
(2) ET0first decreased from 1954 to 1978 and then
increased quickly and reached a maximal value of1373mm during the period from 1992 to 2012 Themean ET
0value for the period from 1954 to 1978 was
1110mm and increased to 1284mm during the periodfrom 1992 to 2012 indicating an increasing trend ofthe evaporative demand
(3) Sensitivity analysis showed that ET0is most sensitive
to relative humidity followed by air temperaturesunshine hours and wind speed
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This paper is partially supported by the National ScienceFoundation of China (Grant nos 51179005 51479004) Theauthors greatly acknowledge the comments from the editorand the two anonymous reviewers
References
[1] IPCC Climate Change 2014 Synthesis Report An Assessmentof Intergovernmental Panel on Climate Change IPCC GenevaSwitzerland 2014 httpipccchindexhtml
[2] J D Haskett Y A Pachepsky and B Acock ldquoEffect of climateand atmospheric change on soybean water stress a study ofIowardquo Ecological Modelling vol 135 no 2-3 pp 265ndash277 2000
[3] T G Huntington ldquoEvidence for intensification of the globalwater cycle review and synthesisrdquo Journal of Hydrology vol 319no 1ndash4 pp 83ndash95 2006
[4] B C Bates Z W Kundzewicz S Wu and J Palutikof ldquoClimatechange and waterrdquo Technical Paper of the IntergovernmentalPanel on Climate Change IPCC Secretariat Geneva Switzer-land 2008
[5] C M Philandras D A Metaxas and P T Nastos ldquoClimatevariability and urbanization in AthensrdquoTheoretical and AppliedClimatology vol 63 no 1-2 pp 65ndash72 1999
[6] R L Wilby ldquoPast and projected trends in Londonrsquos urban heatislandrdquoWeather vol 58 no 7 pp 251ndash260 2003
[7] N Schwarz U Schlink U Franck and K Groszligmann ldquoRela-tionship of land surface and air temperatures and its implica-tions for quantifying urban heat island indicatorsmdashan applica-tion for the city of Leipzig (Germany)rdquoEcological Indicators vol18 pp 693ndash704 2012
[8] H Wang L Fu X Lin Y Zhou and J C Chen ldquoA bottom-up methodology to estimate vehicle emissions for the Beijingurban areardquo Science of the Total Environment vol 407 no 6 pp1947ndash1953 2009
[9] DESE (Department of Environmental Science and Engineer-ingTsinghua University) Mobile Source Database EmissionInventory and Treatment Proposal for Beijing Tsinghua Univer-sity Beijing China 2005
[10] H Kan S J London G Chen et al ldquoDifferentiating the effectsof fine and coarse particles on daily mortality in ShanghaiChinardquo Environment International vol 33 no 3 pp 376ndash3842007
[11] A Aziz and I U Bajwa ldquoErroneous mass transit system andits tended relationship with motor vehicular air pollution (Anintegrated approach for reduction of urban air pollution inLahore)rdquo Environmental Monitoring and Assessment vol 137no 1ndash3 pp 25ndash33 2008
[12] R G Allen L S Perreira D Raes and M Smith Crop Evap-otranspiration Guidelines for Computing Crop Water Require-ments FAO Irrigation and Drainage Paper no 56 FAO RomeItaly 1998
[13] K H Hamed ldquoTrend detection in hydrologic data the Mann-Kendall trend test under the scaling hypothesisrdquo Journal ofHydrology vol 349 no 3-4 pp 350ndash363 2008
[14] L Q Liang L J Li andQ Liu ldquoTemporal variation of referenceevapotranspiration during 1961ndash2005 in the Taoer River basin ofNortheast Chinardquo Agricultural and Forest Meteorology vol 150no 2 pp 298ndash306 2010
Advances in Meteorology 11
[15] H Liu Y Li T Josef R H Zhang and G H HuangldquoQuantitative estimation of climate change effects on potentialevapotranspiration in Beijing during 1951ndash2010rdquo Journal ofGeographical Sciences vol 24 no 1 pp 93ndash112 2014
[16] M G Kendall and A StuartThe Advanced Theory of StatisticsGriffin London UK 1973
[17] Q-Y Tang and C-X Zhang ldquoData Processing System (DPS)software with experimental design statistical analysis and datamining developed for use in entomological researchrdquo InsectScience vol 20 no 2 pp 254ndash260 2013
[18] M Moller J Tanny Y Li and S Cohen ldquoMeasuring andpredicting evapotranspiration in an insect-proof screenhouserdquoAgricultural and Forest Meteorology vol 127 no 1-2 pp 35ndash512004
[19] G Kitsara G Papaioannou A Papathanasiou and A RetalisldquoDimmingbrightening in athens trends in sunshine durationcloud cover and reference evapotranspirationrdquoWater ResourcesManagement vol 27 no 6 pp 1623ndash1633 2013
[20] Y Wang Y Yang S Han Q X Wang and G H ZhangldquoSunshine dimming and brightening in Chinese cities (1955ndash2011) was driven by air pollution rather than cloudsrdquo ClimateResearch vol 56 no 1 pp 11ndash20 2013
[21] G Stanhill and S Cohen ldquoGlobal dimming a review of theevidence for a widespread and significant reduction in globalradiation with discussion of its probable causes and possibleagricultural consequencesrdquoAgricultural and ForestMeteorologyvol 107 no 4 pp 255ndash278 2001
[22] Q Liu and Z Yang ldquoQuantitative estimation of the impact ofclimate change on actual evapotranspiration in the Yellow RiverBasin Chinardquo Journal of Hydrology vol 395 no 3-4 pp 226ndash234 2010
[23] D I Stern ldquoReversal of the trend in global anthropogenic sulfuremissionsrdquoGlobal Environmental Change vol 16 no 2 pp 207ndash220 2006
[24] I Koren J V Martins L A Remer and H Afargan ldquoSmokeinvigoration versus inhibition of clouds over the AmazonrdquoScience vol 321 no 5891 pp 946ndash949 2008
[25] D Rosenfeld Y J Kaufman and I Koren ldquoSwitching cloudcover and dynamical regimes from open to closed Benard cellsin response to the suppression of precipitation by aerosolsrdquoAtmospheric Chemistry and Physics vol 6 no 9 pp 2503ndash25112006
[26] C Ruckstuhl R Philipona K Behrens et al ldquoAerosol and cloudeffects on solar brightening and the recent rapid warmingrdquoGeophysical Research Letters vol 35 no 12 Article ID L127082008
[27] DG Streets Y Fang CMian et al ldquoAnthropogenic andnaturalcontributions to regional trends in aerosol optical depth 1980ndash2006rdquo Journal of Geophysical Research Atmospheres vol 114 no10 Article ID D00D18 2009
[28] V Ramanathan P J Crutzen J T Kiehl and D RosenfeldldquoAtmospheremdashaerosols climate and the hydrological cyclerdquoScience vol 294 no 5549 pp 2119ndash2124 2001
[29] M Wild ldquoEnlightening global dimming and brighteningrdquoBulletin of the AmericanMeteorological Society vol 93 no 1 pp27ndash37 2012
[30] G D Liu Y Li H J Liu and J Xiao ldquoChanging trend of refer-ence crop evapotranspiration and its dominatedmeteorologicalvariables in Shanxi province in the past 55 yearsrdquo Journal ofIrrigation and Drainage vol 31 no 4 pp 26ndash30 2012
[31] C-S Rim ldquoThe effects of urbanization geographical and topo-graphical conditions on reference evapotranspirationrdquo ClimaticChange vol 97 no 3 pp 483ndash514 2009
[32] WKuang Y Liu YDou et al ldquoWhat are hot andwhat are not inan urban landscape quantifying and explaining the land surfacetemperature pattern in Beijing Chinardquo Landscape Ecology2014
[33] Z Qin Q Yu S Xu et al ldquoWater heat fluxes and water useefficiency measurement and modeling above a farmland in theNorth China Plainrdquo Science in China D Earth Sciences vol 48no 1 pp 207ndash217 2005
[34] B Tang L Tong S Z Kang and L Zhang ldquoImpacts ofclimate variability on reference evapotranspiration over 58 yearsin the Haihe river basin of north Chinardquo Agricultural WaterManagement vol 98 no 10 pp 1660ndash1670 2011
[35] S Cohen A Ianetz and G Stanhill ldquoEvaporative climatechanges at BetDagan Israel 1964ndash1998rdquoAgricultural and ForestMeteorology vol 111 no 2 pp 83ndash91 2002
[36] K Chaouche L Neppel C Dieulin et al ldquoAnalyses of precipita-tion temperature and evapotranspiration in a French Mediter-ranean region in the context of climate changerdquoComptes RendusGeoscience vol 342 no 3 pp 234ndash243 2010
[37] Z Huo X Dai S Feng S Kang and G Huang ldquoEffect of cli-mate change on reference evapotranspiration and aridity indexin arid region of Chinardquo Journal of Hydrology vol 492 pp 24ndash34 2013
[38] G Peng X Cai H Zhang A Li F Hu and M Y LeclercldquoHeat flux apportionment to heterogeneous surfaces using fluxfootprint analysisrdquo Advances in Atmospheric Sciences vol 25no 1 pp 107ndash116 2008
[39] Y Q Zhang Y J Shen C M Liu et al ldquoMeasurement andanalysis of water heat and CO
2flux from a farmland in the
North China plainrdquo Acta Geographica Sinica vol 57 no 3 pp333ndash342 2002 (Chinese)
[40] H-J Liu G-H Huang S Cohen and J Tanny ldquoChange in cropevapotranspiration and associated influencing factors underscreenhouse conditionsrdquo Chinese Journal of Eco-Agriculturevol 17 no 3 pp 484ndash488 2009 (Chinese)
Figure 4 Temporal trend of the yearly total precipitation (a) andthe trend results from the Mann-Kendall method (b)
an obvious decline which decreased from 2651 to 1590 hourswith an average of 2014 hours (Figure 2(a))Through the two-tailed test (UF and UB lines) we found that the decreasingtrend of the sunshine hours is obvious for the period from1982 to 2012 and the increasing trend is obvious for the periodof 1954 to 1973 Similar to the trend for the sunshine durationthe solar radiation also showed an obvious declining trendwhich ranges from 3672 to 4520MJsdotmminus2 with an average of4050MJsdotmminus2 (Figure 3(a)) The two-tailed test showed thatthe solar radiation values from 1982 to 2012 were significantlylower than those from 1954 to 1973 The shifting mutationpoints of the sunshine duration and the solar radiation areall found in 1978 using the Mann-Kendall mutational test(119875 lt 001) (Figures 2 and 3(b))
312 Precipitation The annual total precipitation valuesduring the period from 1954 to 2012 are given in Figure 4 Itcan be seen in Figure 4 that the highest precipitation valuewas 2747mm in 2001 and the lowest was 912mm in 1963(Figure 4(a)) The annual mean was 1914mm during theperiod from 1954 to 2012 The precipitation varied greatlyover the years whereas the trend test showed that the tempo-ral trend of the annual precipitation is not significant and amutation point has not been tested (Figure 4(b))
313 Relative Humidity The daily mean relative humidityduring the period from 1954 to 2012 is given in Figure 5 Itcan be seen that the daily mean relative humidity increasedslightly in the first 20 years followed by an obvious decline
1954 1962 1970 1978 1986 1994 2002 2010
Year
y = minus01389x + 35181
R2 = 05331
84
80
76
72
68Mea
n re
lativ
e hum
idity
()
(a)
UF-
UB
UFUB
1954 1962 1970 1978 1986 1994 2002 2010
Year
minus8
minus4
4
0
(b)
Figure 5 Temporal trend of the mean relative humidity (a) and thetrend results from the Mann-Kendall method (b)
after 1978 the range is from 691 to 822 and the average is764 (Figure 5(a)) From the Mann-Kendall test (119875 lt 001)we found that the averaged relative humidity value during theperiod from 1992 to 2012 is significantly (119875 lt 001) lower thanthe period from 1954 to 1992
314 Air Temperature Themean air temperature during theperiod from 1954 to 2012 is shown in Figure 6 Accordingto Figure 6 the mean air temperature showed an obviouslyincreasing trend and the values range from 215 to 239∘Cwith an average value of 226∘C (Figure 6(a)) The statisticalresult showed that the mean air temperature over the periodfrom 1990 to 2012 is significantly (119875 lt 001) higher than thoseover the period from 1954 to 1983 (Figure 6(b))
315 Vapor Pressure Deficit Figure 7(a) shows the temporaltrend of the vapor pressure deficit (VPD) during the periodfrom 1954 to 2012 VPD ranges from 0449 to 0867 kPa andthe average is 0648 kPa There is no significant temporaltrend for VPD from 1954 to 1977 and after 1978 VPDincreased greatly It was found that the average VPD duringthe period from 1992 to 2012 is significantly (119875 lt 001) higherthan the values from 1954 to 1991 according to the Mann-Kendall mutational test
316 Wind Speed The annual mean wind speed variedgreatly during the period from 1954 to 2012 (Figure 8(a))Clearly decreasing trends were found from 1954 to 1977 andfrom 1987 to 2012 whereas increasing trend was found from
6 Advances in Meteorology
1954 1962 1970 1978 1986 1994 2002 2010
Year
y = 003x minus 36919
R2 = 06208
24
24
23
23
22
22
21
Mea
n te
mpe
ratu
re(∘
C)
(a)
UF-
UB
UFUB
1954 1962 1970 1978 1986 1994 2002 2010
Year
minus8
minus4
4
8
0
(b)
Figure 6 Temporal trend of themean temperature (a) and the trendresults from the Mann-Kendall method (b)
1954 1962 1970 1978 1986 1994 2002 2010
Year
y = 00055x minus 10261
R2 = 06721
100
040
020
060
080
Vapo
r pre
ssur
e defi
cit (
kPa)
(a)
UF-
UB
UFUB
1954 1962 1970 1978 1986 1994 2002 2010
Year
minus8
minus4
4
8
0
(b)
Figure 7 Temporal trend of the mean vapor pressure deficit (a) andthe trend results from the Mann-Kendall method (b)
1978 to 1986 It can be seen in Figure 8(a) that the maximumwind speed was 372ms in 1954 the minimum was 178ms
1954 1962 1970 1978 1986 1994 2002 2010
Year
y = minus00098x + 22013
R2 = 01214
40
30
20
10Win
d sp
eed
(mmiddotsminus
1)
(a)
UF-
UB
UFUB
1954 1962 1970 1978 1986 1994 2002 2010
Year
minus8
minus4
4
0
(b)
Figure 8 Temporal trend of the mean wind speed (a) and the trendresults from the Mann-Kendall method (b)
in 1977 and the annual mean was 260ms The statisticalresults show that there were no significant temporal trendsfrom 1954 to 2012
32 Reference Evapotranspiration (ET0) Theannual total ET0
varied from 946 to 1373mm with a total mean of 1187mmFigure 9(a) shows the change in the annual ET
0over the
past 59 years According to Figure 9(a) ET0firstly decreased
gradually from 1954 to 1978 then increased from 1978 to 1992and lastly varied slightly after 1992 From 1954 to 1978 theannual total ET
0decreased from 1285 to 946mmwith amean
value of 1110mm afterwards the annual total ET0increased
from 1118 to 1373mm with a mean value of 1284mm duringthe period from 1992 to 2012 The mean ET
0in the period
of 1992ndash2012 increased by 156 over those in the period of1954ndash1978 indicating a great increase in evaporation poten-tial The statistical results based on the Mann-Kendall test(119875 lt 001) (Figure 9(b)) showed that ET
0during the period
from 1972 to 1987 was significantly lower than that from 2001to 2012 which was significantly higher than those in otherperiods The shifting mutation point for ET
0is found at 1992
by the Mann-Kendall test (Figure 9(b))For analyzing the yearly ET
0distribution the total ET
0for
eachmonthwas calculated and the averagedmonth total ET0s
in period of 1954ndash1978 and 1992ndash2012were calculated respec-tively The monthly ET
0distributions in these two periods
and in all study period (1954ndash2012) are showed in Figure 10The highest monthly ET
0generally appears in July and
August and the lowest in January and FebruaryMonthly ET0
from May to November are generally higher than 100mm
Advances in Meteorology 7
Table 1 Mean values of each climatic variable in the periods of 1954ndash1978 and 1992ndash2012 and changes of each climatic variable to ET0variation
Air temperature Relative humidity Sunshine hours Wind speed∘C Hoursday ms
Mean values1954ndash1978 220 789 608 2681992ndash2012 233 732 504 252
Average ET0 over 1954ndash1978Average ET0 over 1992ndash2012
(a)
UF-
UB
UFUB
1954 1962 1970 1978 1986 1994 2002 2010
Year
minus8
minus4
4
8
0
(b)
Figure 9 Temporal trend of the yearly total ET0(a) and the trend
results from the Mann-Kendall method (b)
and the total ET0in this period accounts for approximately
13 the yearly total Monthly ET0s in the period of 1992ndash2012
are all higher by 5ndash30mm or 7ndash25 than those in the 1954ndash1978 period The most increases in monthly ET
0are found in
period from July to September and the lowest from January toApril with an increase of less than 7mmTherefore the greatincrease in ET
0in summer (generally from June to October)
makes main contribution to yearly ET0
33 Sensitivity Analysis of ET0 to the Change of ClimaticVariables Figures 2 to 8 show that most mutation pointsfor most of the climatic variables were found during theperiod from 1978 to 1992 and the mean values of the climaticvariables over the period from 1954 to 1978 were significantly
150
100
50
01 2 3 4 5 6 7 8 9 10 11 12
Months
Monthly mean from period of 1954ndash2012
Monthly mean from period of 1954ndash1978Monthly mean from period of 1992ndash2012
50
40
30
20
10
0Mon
thly
tota
ldie
renc
eET 0
(mm
)
Incr
ease
per
cent
age i
nET
0(
)
Increase in ET0 from 1954ndash1978 to 1992ndash2012Increase percentage in ET0
Figure 10 Yearly distribution of ET0averaged in the periods of
1954ndash1978 1992ndash2012 and 1954ndash2012 ET0increase amount and the
corresponding increase percentage in each month from periods of1954ndash1978 to 1992ndash2012 were showed
(119875 lt 001) higherlower than those during the period from1992 to 2012 Hence the mean values of each variable duringthe two periods from 1954 to 1978 and from 1992 to 2012 werecalculated and used to analyze their change effects on the ET
0
changes following themethod described in Section 233Thesummary of the mean values of each climate and their effectson ET
0were listed in Table 1 and Figure 11
It can be seen in Table 1 that the daily mean relativehumidity daily mean air temperature wind speed and sun-shine hours from the period of 1954ndash1978 to 1992ndash2012 wereminus72 59 minus60 and minus171 respectively which resulted inET0changes by 146 53 minus13 and minus32 respectively The
contribution of each climate variablersquos variation from 1954ndash1978 to 1992ndash2012 to ET
0change is shown in Figure 11 It
can be found that decrease in relative humidity accounted forapproximately 60 variation in ET
0 followed by temperature
increase with a contribution of 22 and sunshine hoursreduction of 13 Wind speed accounted for 6 variationin ET
0 Similarly in another mega city Beijing in China the
order of climate change to ET0variation frommain to weak is
air temperature relative humidity sunshine hours and windspeed [14]
8 Advances in Meteorology
Relative humidity
60Wind speed6
Temperature22
Sunshine hours13
Figure 11 The contribution of each climatic variable change to ET0variation
300
200
100
0
Dev
elopm
ent o
f She
nzhe
n ci
ty
Heavy industry output value (billion dollars)Resident population (10minus1 million)Electric energy production (102 MWh)
1954 1962 1970 1978 1986 1994 2002 2010
Year
Figure 12 Temporal trend of the development of Shenzhen city
4 Discussion
41 Change in the Sunshine Hours and Urban DevelopmentSunshine hours generally depend on the cloud cover man-made aerosols and certain air pollutants (including SO
2
NO119909 and PM) [19 20] Recent studies indicated that the
most probable cause for the depression of sunshine hours orsolar radiation is in the increased concentrations ofmanmadeaerosols and other air pollutants [1 19ndash22] In this studythe sunshine hours and amount of radiation showed cleardecreasing trends after 1970 and theywere significantly lowerafter 1980 during rapid urban development
Figure 12 shows the increasing trends of the residentpopulation heavy industry output value and electric energyproduction It is confirmed that the increasing trend in theenergy consumption corresponds to increased emission ofpolluted particles including CO CO
2 NO119909 SO119909 PM25 O3
andHCThese particlesmay result in an increase in the atmo-spheric aerosol concentration [23ndash27] which can directlyattenuate the surface solar radiation (SSR) by scattering andabsorbing solar radiation (direct effect) or can indirectlyattenuate SSR by their ability to act as cloud condensation
nuclei thereby increasing the cloud reflectivity and lifetime(first and second indirect effects) [24 28] A remarkabledecline in SSR between the 1950s and the 1980s was found inseveral studies that were performed at selected observationstations based on sites in Europe the Baltics the South PoleGermany and the former Soviet Union [21] Today a com-prehensive literature exists that confirms the declines of SSRduring this period in many places around the world [29]In Beijing and several other Chinese cities the decline ofsunshine hours and SSR has also been reported [15 30]
42 Air Temperature Change and Urban Development It hasbeen confirmed that there is a large air temperature differ-ence in urban and rural areas [4ndash7] These air temperaturedifferences result from the influence of the thermal emissivityproperties of urban surfaces and the three-dimensional con-figuration and heat capacity of erected structures onto the airtemperature patterns in an urban region [7 15 31]Most of theworldrsquos cities thus show higher air temperatures in the urbancore than in the surrounding rural areas [4 5 7] For examplein Beijing City China Kuang et al [32] measured that themean land surface temperature of urban impervious surfaceswas about 6ndash12∘C higher than that of the urban green spaceand that in built-up areaswas on average 3ndash6∘Chigher than inrural areas They showed the main reason is the higher ratioof sensible heat to net radiation (063) and lower ratio of thelatent heat to net radiation (019) on the urban impervioussurface as compared to the corresponding rates of 030 and063 in green space and cropland In this study there is noobvious temporal trend in the air temperature before 1978whereas after that time the temperature increased greatlyalthough it showed a slight declining trend in recent yearsThe large air temperature increase from 1978 to 2002 may bedue to the increasing area of construction and the decrease infarmland [32] Figure 13 shows the changes in the farmlandand construction areas It can be found that there was arapid increase in construction areas and an abrupt decreasein farmland areas after 1978 which led to a change in thethermal balance and then resulted in the increase in the airtemperature in Shenzhen city
Advances in Meteorology 9
1954 1962 1970 1978 1986 1994 2002 2010
Year
8000
6000
4000
2000
0Farm
land
and
cons
truc
tion
area
Farm land area (101 ha)Construction area (ha)
Figure 13 Temporal trend of the farmland area and the floor spaceof the buildings under construction
43 Changes in the Relative Humidity and the Vapor PressureDeficit due to Urban Development The vapor pressure deficitwas calculated by the air temperature and the relative humid-ity following themethod of Allen et al (1998) [12] It is clearlyshown in the calculation method that the vapor pressuredeficit (VPD) increases with air temperature and decreaseswith relative humidity In this study both relative humidityand mean air temperature varied slightly during the periodfrom 1954 to 1980 (Figures 5 and 6) which consequentlyresulted in a slight change in the vapor pressure deficit Afterthat period the mean air temperature increased significantlyand the relative humidity was decreased remarkably Thesetrends resulted in an increase in the vapor pressure deficit(Figure 7)
The relative humidity is defined as the ratio of the watervapor density (mass per unit volume) to the saturation watervapor density and it is also approximately the ratio of theactual to the saturation vapor pressure A greater evaporativearea may produce more water vapor for the atmosphere andthen increase the water vapor density as well as the relativehumidity at a given temperature In the city study area thefarmland area decreased whereas the construction areaincreased (Figure 13) These data indicate a decreasing trendin the evaporative area which may result in a decrease in thewater vapor density as well as the relative humidity [33] Sim-ilarly a decreasing trend of relative humidity and an increasein the vapor pressure deficit were observed in Beijing Datongin Shanxi province Zhang Jiakou in Hebei Province and BetDagan in Israel [15 34 35] In Beijing it was shown thatthe VPD increased with air temperature and decreased withrelative humidity from 1951 to 2010 [15] Cohen et al (2002)showed that themain factor responsible for the increased panevaporation was the growth in the aerodynamic componentof evaporation which was due to increases in both the airVPD and the wind speed at Bet Dagan from 1964 to 1997 [35]
The vapor pressure deficit represents a gradient acrosswhich water vapor is removed from the evapotranspiringsurface into the surrounding air [12] A greater vapor pressuredeficit generally causes a higher evaporative rate Hence the
increasing vapor pressure deficit in the Shenzhen area willresult in increasing plant evapotranspiration
44 ET0 Change and City Development In the current studyarea ET
0first decreased from 1950s to 1970s and then
increased greatly in the 1980s During the 1990s and 2000s itvaried slightly with amean value of 1287mm It was observedthat after the onset of urban development in 1978 the ET
0
value increased and became higher than this for the periodprior to the urban development Figures 2 5 6 and 8 showthat the mutation points for most climatic variables wereobserved near the onset year of urban development and sen-sitivity analysis shows that the higher ET
0during the period
of 1992ndash2012 is mainly attributed to the relative humiditydecrease and air temperature increaseHence it could be con-cluded that the quick development of Shenzhen city alteredthe climatic conditions and hence increased the local ET
0
In other large cities an increasing ET0trendwas also found in
recent decades For example in Beijing City the annual ET0
increased significantly from 1951 to 2010 and from the 1950sto 2000s it increased from 1039 to 1148mm [15] The annualpotential evapotranspiration (PET) displays a significantupward trend from 1970 to 2006 and the trend varied from 1to 4mm per year in the Pyrenees-Orientales and Audeadministrative departments respectively and the westernpart of the French Mediterranean area with an averageincrease in PET of between 34mm and 150mm in the last 36years [36]
The increasing trend in ET0that was observed in large
cities is different than this that was found in the farmlandFor example in the Haihe River basin in northern Chinadecreasing trends were observed in 26 stations while 16stations showed significant decreasing trends from 1950 to2007 with rates from minus20 to minus37mmyearminus1 [34] Similarly asignificant decreasing trend of ET
0with a rate of minus3mm per
yearwas found in the arid region of northwest China [37]Thedifference trend in ET
0between the large cities and the
farmland may be due to the variations in the energy balanceand the evaporative potential In the farmland areas morethan 60ndash80 of the net radiation is used for plant evapotran-spiration [33 38 39]This effect not only reduces the availableheat for heating the air environment but also increasesthe water vapor in the near atmosphere The latter effect mayincrease the relative humidity and reduce the vapor pressuredeficit and lastly it may reduce the reference evapotranspi-ration For urban conditions the decrease in green land willdecrease the energy consumption caused by crop evapotran-spiration increasing the available heat and decreasing thewater vapor which ultimately results in an increase in theevaporative potential
It should be noted in Figure 9 that the ET0values in 1997
and 2012 were much lower than those in the neighboringyears It is estimated that ET
0in 1997 and 2012 decreased
by minus104 and minus130 compared to the mean value for theperiod from 1990 to 2012 The sensitive analysis methoddescribed in Section 33 was used to determine the effects ofeach variable on the ET
0changes in 1997 and 2012 compared
to the mean value during the period of 1990ndash2012 The
10 Advances in Meteorology
results showed that the changes in the relative humidity airtemperature sunshine hours andwind speed in 1997 resultedin changes in ET
0by minus72 minus08 minus20 and minus05 respectively
and by minus126 minus08 minus06 and minus42 in 2012 respectively Itcould be concluded that the increase in the relative humidityis the main factor for ET
0reduction followed by the wind
speed air temperature and sunshine hours in the two yearsBased on the mean values of the climatic variables
averaged over the periods of 1954ndash1978 and 1992ndash2012 ET0
increased by 145 in the latter period For the sensitivityanalysis changes in the relative humidity air temperaturewind speed and sunshine hours during 1992ndash2012 caused thevariation of ET
0by 146 50 minus13 and minus38 respectively
compared to those for the period 1954ndash1978The total amountof change in ET
0was 156 based on the sensitivity analysis
This value is similar to the rate of increase of 145 by com-parison of the ET
0values between the two time periods Liu
et al (2014) [15] calculated the ET0change rates by directly
comparing the mean values and summing each ET0change
rate caused by climatic variables using the same sensitiveanalysis method the ET
0change rates were 107 and 105
respectively Liu et al (2009) [40] found that ET0inside the
screenhouse was reduced by 39 compared to that in theopen field By considering the effect of each climatic variablechange to ET
0using the sensitivity analysis the total ET
0
change rate sums to 44 which is similar to the value of39 Therefore it could be concluded that the sensitivity-analysis method used in this study is reliable and easy to useand hence it is recommended for the analysis of the effect ofclimate change on ET
0
5 Conclusions
(1) The development of Shenzhen city greatly affected thelocal climatic conditions Before the onset of urbandevelopment each climatic variable varied slightlywhereas afterward the air temperature increased sig-nificantly and the sunshine hours and relative humid-ity decreased significantly The mutation point formost climatic variables is observed at approximately1978 the onset year for urban development
(2) ET0first decreased from 1954 to 1978 and then
increased quickly and reached a maximal value of1373mm during the period from 1992 to 2012 Themean ET
0value for the period from 1954 to 1978 was
1110mm and increased to 1284mm during the periodfrom 1992 to 2012 indicating an increasing trend ofthe evaporative demand
(3) Sensitivity analysis showed that ET0is most sensitive
to relative humidity followed by air temperaturesunshine hours and wind speed
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This paper is partially supported by the National ScienceFoundation of China (Grant nos 51179005 51479004) Theauthors greatly acknowledge the comments from the editorand the two anonymous reviewers
References
[1] IPCC Climate Change 2014 Synthesis Report An Assessmentof Intergovernmental Panel on Climate Change IPCC GenevaSwitzerland 2014 httpipccchindexhtml
[2] J D Haskett Y A Pachepsky and B Acock ldquoEffect of climateand atmospheric change on soybean water stress a study ofIowardquo Ecological Modelling vol 135 no 2-3 pp 265ndash277 2000
[3] T G Huntington ldquoEvidence for intensification of the globalwater cycle review and synthesisrdquo Journal of Hydrology vol 319no 1ndash4 pp 83ndash95 2006
[4] B C Bates Z W Kundzewicz S Wu and J Palutikof ldquoClimatechange and waterrdquo Technical Paper of the IntergovernmentalPanel on Climate Change IPCC Secretariat Geneva Switzer-land 2008
[5] C M Philandras D A Metaxas and P T Nastos ldquoClimatevariability and urbanization in AthensrdquoTheoretical and AppliedClimatology vol 63 no 1-2 pp 65ndash72 1999
[6] R L Wilby ldquoPast and projected trends in Londonrsquos urban heatislandrdquoWeather vol 58 no 7 pp 251ndash260 2003
[7] N Schwarz U Schlink U Franck and K Groszligmann ldquoRela-tionship of land surface and air temperatures and its implica-tions for quantifying urban heat island indicatorsmdashan applica-tion for the city of Leipzig (Germany)rdquoEcological Indicators vol18 pp 693ndash704 2012
[8] H Wang L Fu X Lin Y Zhou and J C Chen ldquoA bottom-up methodology to estimate vehicle emissions for the Beijingurban areardquo Science of the Total Environment vol 407 no 6 pp1947ndash1953 2009
[9] DESE (Department of Environmental Science and Engineer-ingTsinghua University) Mobile Source Database EmissionInventory and Treatment Proposal for Beijing Tsinghua Univer-sity Beijing China 2005
[10] H Kan S J London G Chen et al ldquoDifferentiating the effectsof fine and coarse particles on daily mortality in ShanghaiChinardquo Environment International vol 33 no 3 pp 376ndash3842007
[11] A Aziz and I U Bajwa ldquoErroneous mass transit system andits tended relationship with motor vehicular air pollution (Anintegrated approach for reduction of urban air pollution inLahore)rdquo Environmental Monitoring and Assessment vol 137no 1ndash3 pp 25ndash33 2008
[12] R G Allen L S Perreira D Raes and M Smith Crop Evap-otranspiration Guidelines for Computing Crop Water Require-ments FAO Irrigation and Drainage Paper no 56 FAO RomeItaly 1998
[13] K H Hamed ldquoTrend detection in hydrologic data the Mann-Kendall trend test under the scaling hypothesisrdquo Journal ofHydrology vol 349 no 3-4 pp 350ndash363 2008
[14] L Q Liang L J Li andQ Liu ldquoTemporal variation of referenceevapotranspiration during 1961ndash2005 in the Taoer River basin ofNortheast Chinardquo Agricultural and Forest Meteorology vol 150no 2 pp 298ndash306 2010
Advances in Meteorology 11
[15] H Liu Y Li T Josef R H Zhang and G H HuangldquoQuantitative estimation of climate change effects on potentialevapotranspiration in Beijing during 1951ndash2010rdquo Journal ofGeographical Sciences vol 24 no 1 pp 93ndash112 2014
[16] M G Kendall and A StuartThe Advanced Theory of StatisticsGriffin London UK 1973
[17] Q-Y Tang and C-X Zhang ldquoData Processing System (DPS)software with experimental design statistical analysis and datamining developed for use in entomological researchrdquo InsectScience vol 20 no 2 pp 254ndash260 2013
[18] M Moller J Tanny Y Li and S Cohen ldquoMeasuring andpredicting evapotranspiration in an insect-proof screenhouserdquoAgricultural and Forest Meteorology vol 127 no 1-2 pp 35ndash512004
[19] G Kitsara G Papaioannou A Papathanasiou and A RetalisldquoDimmingbrightening in athens trends in sunshine durationcloud cover and reference evapotranspirationrdquoWater ResourcesManagement vol 27 no 6 pp 1623ndash1633 2013
[20] Y Wang Y Yang S Han Q X Wang and G H ZhangldquoSunshine dimming and brightening in Chinese cities (1955ndash2011) was driven by air pollution rather than cloudsrdquo ClimateResearch vol 56 no 1 pp 11ndash20 2013
[21] G Stanhill and S Cohen ldquoGlobal dimming a review of theevidence for a widespread and significant reduction in globalradiation with discussion of its probable causes and possibleagricultural consequencesrdquoAgricultural and ForestMeteorologyvol 107 no 4 pp 255ndash278 2001
[22] Q Liu and Z Yang ldquoQuantitative estimation of the impact ofclimate change on actual evapotranspiration in the Yellow RiverBasin Chinardquo Journal of Hydrology vol 395 no 3-4 pp 226ndash234 2010
[23] D I Stern ldquoReversal of the trend in global anthropogenic sulfuremissionsrdquoGlobal Environmental Change vol 16 no 2 pp 207ndash220 2006
[24] I Koren J V Martins L A Remer and H Afargan ldquoSmokeinvigoration versus inhibition of clouds over the AmazonrdquoScience vol 321 no 5891 pp 946ndash949 2008
[25] D Rosenfeld Y J Kaufman and I Koren ldquoSwitching cloudcover and dynamical regimes from open to closed Benard cellsin response to the suppression of precipitation by aerosolsrdquoAtmospheric Chemistry and Physics vol 6 no 9 pp 2503ndash25112006
[26] C Ruckstuhl R Philipona K Behrens et al ldquoAerosol and cloudeffects on solar brightening and the recent rapid warmingrdquoGeophysical Research Letters vol 35 no 12 Article ID L127082008
[27] DG Streets Y Fang CMian et al ldquoAnthropogenic andnaturalcontributions to regional trends in aerosol optical depth 1980ndash2006rdquo Journal of Geophysical Research Atmospheres vol 114 no10 Article ID D00D18 2009
[28] V Ramanathan P J Crutzen J T Kiehl and D RosenfeldldquoAtmospheremdashaerosols climate and the hydrological cyclerdquoScience vol 294 no 5549 pp 2119ndash2124 2001
[29] M Wild ldquoEnlightening global dimming and brighteningrdquoBulletin of the AmericanMeteorological Society vol 93 no 1 pp27ndash37 2012
[30] G D Liu Y Li H J Liu and J Xiao ldquoChanging trend of refer-ence crop evapotranspiration and its dominatedmeteorologicalvariables in Shanxi province in the past 55 yearsrdquo Journal ofIrrigation and Drainage vol 31 no 4 pp 26ndash30 2012
[31] C-S Rim ldquoThe effects of urbanization geographical and topo-graphical conditions on reference evapotranspirationrdquo ClimaticChange vol 97 no 3 pp 483ndash514 2009
[32] WKuang Y Liu YDou et al ldquoWhat are hot andwhat are not inan urban landscape quantifying and explaining the land surfacetemperature pattern in Beijing Chinardquo Landscape Ecology2014
[33] Z Qin Q Yu S Xu et al ldquoWater heat fluxes and water useefficiency measurement and modeling above a farmland in theNorth China Plainrdquo Science in China D Earth Sciences vol 48no 1 pp 207ndash217 2005
[34] B Tang L Tong S Z Kang and L Zhang ldquoImpacts ofclimate variability on reference evapotranspiration over 58 yearsin the Haihe river basin of north Chinardquo Agricultural WaterManagement vol 98 no 10 pp 1660ndash1670 2011
[35] S Cohen A Ianetz and G Stanhill ldquoEvaporative climatechanges at BetDagan Israel 1964ndash1998rdquoAgricultural and ForestMeteorology vol 111 no 2 pp 83ndash91 2002
[36] K Chaouche L Neppel C Dieulin et al ldquoAnalyses of precipita-tion temperature and evapotranspiration in a French Mediter-ranean region in the context of climate changerdquoComptes RendusGeoscience vol 342 no 3 pp 234ndash243 2010
[37] Z Huo X Dai S Feng S Kang and G Huang ldquoEffect of cli-mate change on reference evapotranspiration and aridity indexin arid region of Chinardquo Journal of Hydrology vol 492 pp 24ndash34 2013
[38] G Peng X Cai H Zhang A Li F Hu and M Y LeclercldquoHeat flux apportionment to heterogeneous surfaces using fluxfootprint analysisrdquo Advances in Atmospheric Sciences vol 25no 1 pp 107ndash116 2008
[39] Y Q Zhang Y J Shen C M Liu et al ldquoMeasurement andanalysis of water heat and CO
2flux from a farmland in the
North China plainrdquo Acta Geographica Sinica vol 57 no 3 pp333ndash342 2002 (Chinese)
[40] H-J Liu G-H Huang S Cohen and J Tanny ldquoChange in cropevapotranspiration and associated influencing factors underscreenhouse conditionsrdquo Chinese Journal of Eco-Agriculturevol 17 no 3 pp 484ndash488 2009 (Chinese)
Figure 6 Temporal trend of themean temperature (a) and the trendresults from the Mann-Kendall method (b)
1954 1962 1970 1978 1986 1994 2002 2010
Year
y = 00055x minus 10261
R2 = 06721
100
040
020
060
080
Vapo
r pre
ssur
e defi
cit (
kPa)
(a)
UF-
UB
UFUB
1954 1962 1970 1978 1986 1994 2002 2010
Year
minus8
minus4
4
8
0
(b)
Figure 7 Temporal trend of the mean vapor pressure deficit (a) andthe trend results from the Mann-Kendall method (b)
1978 to 1986 It can be seen in Figure 8(a) that the maximumwind speed was 372ms in 1954 the minimum was 178ms
1954 1962 1970 1978 1986 1994 2002 2010
Year
y = minus00098x + 22013
R2 = 01214
40
30
20
10Win
d sp
eed
(mmiddotsminus
1)
(a)
UF-
UB
UFUB
1954 1962 1970 1978 1986 1994 2002 2010
Year
minus8
minus4
4
0
(b)
Figure 8 Temporal trend of the mean wind speed (a) and the trendresults from the Mann-Kendall method (b)
in 1977 and the annual mean was 260ms The statisticalresults show that there were no significant temporal trendsfrom 1954 to 2012
32 Reference Evapotranspiration (ET0) Theannual total ET0
varied from 946 to 1373mm with a total mean of 1187mmFigure 9(a) shows the change in the annual ET
0over the
past 59 years According to Figure 9(a) ET0firstly decreased
gradually from 1954 to 1978 then increased from 1978 to 1992and lastly varied slightly after 1992 From 1954 to 1978 theannual total ET
0decreased from 1285 to 946mmwith amean
value of 1110mm afterwards the annual total ET0increased
from 1118 to 1373mm with a mean value of 1284mm duringthe period from 1992 to 2012 The mean ET
0in the period
of 1992ndash2012 increased by 156 over those in the period of1954ndash1978 indicating a great increase in evaporation poten-tial The statistical results based on the Mann-Kendall test(119875 lt 001) (Figure 9(b)) showed that ET
0during the period
from 1972 to 1987 was significantly lower than that from 2001to 2012 which was significantly higher than those in otherperiods The shifting mutation point for ET
0is found at 1992
by the Mann-Kendall test (Figure 9(b))For analyzing the yearly ET
0distribution the total ET
0for
eachmonthwas calculated and the averagedmonth total ET0s
in period of 1954ndash1978 and 1992ndash2012were calculated respec-tively The monthly ET
0distributions in these two periods
and in all study period (1954ndash2012) are showed in Figure 10The highest monthly ET
0generally appears in July and
August and the lowest in January and FebruaryMonthly ET0
from May to November are generally higher than 100mm
Advances in Meteorology 7
Table 1 Mean values of each climatic variable in the periods of 1954ndash1978 and 1992ndash2012 and changes of each climatic variable to ET0variation
Air temperature Relative humidity Sunshine hours Wind speed∘C Hoursday ms
Mean values1954ndash1978 220 789 608 2681992ndash2012 233 732 504 252
Average ET0 over 1954ndash1978Average ET0 over 1992ndash2012
(a)
UF-
UB
UFUB
1954 1962 1970 1978 1986 1994 2002 2010
Year
minus8
minus4
4
8
0
(b)
Figure 9 Temporal trend of the yearly total ET0(a) and the trend
results from the Mann-Kendall method (b)
and the total ET0in this period accounts for approximately
13 the yearly total Monthly ET0s in the period of 1992ndash2012
are all higher by 5ndash30mm or 7ndash25 than those in the 1954ndash1978 period The most increases in monthly ET
0are found in
period from July to September and the lowest from January toApril with an increase of less than 7mmTherefore the greatincrease in ET
0in summer (generally from June to October)
makes main contribution to yearly ET0
33 Sensitivity Analysis of ET0 to the Change of ClimaticVariables Figures 2 to 8 show that most mutation pointsfor most of the climatic variables were found during theperiod from 1978 to 1992 and the mean values of the climaticvariables over the period from 1954 to 1978 were significantly
150
100
50
01 2 3 4 5 6 7 8 9 10 11 12
Months
Monthly mean from period of 1954ndash2012
Monthly mean from period of 1954ndash1978Monthly mean from period of 1992ndash2012
50
40
30
20
10
0Mon
thly
tota
ldie
renc
eET 0
(mm
)
Incr
ease
per
cent
age i
nET
0(
)
Increase in ET0 from 1954ndash1978 to 1992ndash2012Increase percentage in ET0
Figure 10 Yearly distribution of ET0averaged in the periods of
1954ndash1978 1992ndash2012 and 1954ndash2012 ET0increase amount and the
corresponding increase percentage in each month from periods of1954ndash1978 to 1992ndash2012 were showed
(119875 lt 001) higherlower than those during the period from1992 to 2012 Hence the mean values of each variable duringthe two periods from 1954 to 1978 and from 1992 to 2012 werecalculated and used to analyze their change effects on the ET
0
changes following themethod described in Section 233Thesummary of the mean values of each climate and their effectson ET
0were listed in Table 1 and Figure 11
It can be seen in Table 1 that the daily mean relativehumidity daily mean air temperature wind speed and sun-shine hours from the period of 1954ndash1978 to 1992ndash2012 wereminus72 59 minus60 and minus171 respectively which resulted inET0changes by 146 53 minus13 and minus32 respectively The
contribution of each climate variablersquos variation from 1954ndash1978 to 1992ndash2012 to ET
0change is shown in Figure 11 It
can be found that decrease in relative humidity accounted forapproximately 60 variation in ET
0 followed by temperature
increase with a contribution of 22 and sunshine hoursreduction of 13 Wind speed accounted for 6 variationin ET
0 Similarly in another mega city Beijing in China the
order of climate change to ET0variation frommain to weak is
air temperature relative humidity sunshine hours and windspeed [14]
8 Advances in Meteorology
Relative humidity
60Wind speed6
Temperature22
Sunshine hours13
Figure 11 The contribution of each climatic variable change to ET0variation
300
200
100
0
Dev
elopm
ent o
f She
nzhe
n ci
ty
Heavy industry output value (billion dollars)Resident population (10minus1 million)Electric energy production (102 MWh)
1954 1962 1970 1978 1986 1994 2002 2010
Year
Figure 12 Temporal trend of the development of Shenzhen city
4 Discussion
41 Change in the Sunshine Hours and Urban DevelopmentSunshine hours generally depend on the cloud cover man-made aerosols and certain air pollutants (including SO
2
NO119909 and PM) [19 20] Recent studies indicated that the
most probable cause for the depression of sunshine hours orsolar radiation is in the increased concentrations ofmanmadeaerosols and other air pollutants [1 19ndash22] In this studythe sunshine hours and amount of radiation showed cleardecreasing trends after 1970 and theywere significantly lowerafter 1980 during rapid urban development
Figure 12 shows the increasing trends of the residentpopulation heavy industry output value and electric energyproduction It is confirmed that the increasing trend in theenergy consumption corresponds to increased emission ofpolluted particles including CO CO
2 NO119909 SO119909 PM25 O3
andHCThese particlesmay result in an increase in the atmo-spheric aerosol concentration [23ndash27] which can directlyattenuate the surface solar radiation (SSR) by scattering andabsorbing solar radiation (direct effect) or can indirectlyattenuate SSR by their ability to act as cloud condensation
nuclei thereby increasing the cloud reflectivity and lifetime(first and second indirect effects) [24 28] A remarkabledecline in SSR between the 1950s and the 1980s was found inseveral studies that were performed at selected observationstations based on sites in Europe the Baltics the South PoleGermany and the former Soviet Union [21] Today a com-prehensive literature exists that confirms the declines of SSRduring this period in many places around the world [29]In Beijing and several other Chinese cities the decline ofsunshine hours and SSR has also been reported [15 30]
42 Air Temperature Change and Urban Development It hasbeen confirmed that there is a large air temperature differ-ence in urban and rural areas [4ndash7] These air temperaturedifferences result from the influence of the thermal emissivityproperties of urban surfaces and the three-dimensional con-figuration and heat capacity of erected structures onto the airtemperature patterns in an urban region [7 15 31]Most of theworldrsquos cities thus show higher air temperatures in the urbancore than in the surrounding rural areas [4 5 7] For examplein Beijing City China Kuang et al [32] measured that themean land surface temperature of urban impervious surfaceswas about 6ndash12∘C higher than that of the urban green spaceand that in built-up areaswas on average 3ndash6∘Chigher than inrural areas They showed the main reason is the higher ratioof sensible heat to net radiation (063) and lower ratio of thelatent heat to net radiation (019) on the urban impervioussurface as compared to the corresponding rates of 030 and063 in green space and cropland In this study there is noobvious temporal trend in the air temperature before 1978whereas after that time the temperature increased greatlyalthough it showed a slight declining trend in recent yearsThe large air temperature increase from 1978 to 2002 may bedue to the increasing area of construction and the decrease infarmland [32] Figure 13 shows the changes in the farmlandand construction areas It can be found that there was arapid increase in construction areas and an abrupt decreasein farmland areas after 1978 which led to a change in thethermal balance and then resulted in the increase in the airtemperature in Shenzhen city
Advances in Meteorology 9
1954 1962 1970 1978 1986 1994 2002 2010
Year
8000
6000
4000
2000
0Farm
land
and
cons
truc
tion
area
Farm land area (101 ha)Construction area (ha)
Figure 13 Temporal trend of the farmland area and the floor spaceof the buildings under construction
43 Changes in the Relative Humidity and the Vapor PressureDeficit due to Urban Development The vapor pressure deficitwas calculated by the air temperature and the relative humid-ity following themethod of Allen et al (1998) [12] It is clearlyshown in the calculation method that the vapor pressuredeficit (VPD) increases with air temperature and decreaseswith relative humidity In this study both relative humidityand mean air temperature varied slightly during the periodfrom 1954 to 1980 (Figures 5 and 6) which consequentlyresulted in a slight change in the vapor pressure deficit Afterthat period the mean air temperature increased significantlyand the relative humidity was decreased remarkably Thesetrends resulted in an increase in the vapor pressure deficit(Figure 7)
The relative humidity is defined as the ratio of the watervapor density (mass per unit volume) to the saturation watervapor density and it is also approximately the ratio of theactual to the saturation vapor pressure A greater evaporativearea may produce more water vapor for the atmosphere andthen increase the water vapor density as well as the relativehumidity at a given temperature In the city study area thefarmland area decreased whereas the construction areaincreased (Figure 13) These data indicate a decreasing trendin the evaporative area which may result in a decrease in thewater vapor density as well as the relative humidity [33] Sim-ilarly a decreasing trend of relative humidity and an increasein the vapor pressure deficit were observed in Beijing Datongin Shanxi province Zhang Jiakou in Hebei Province and BetDagan in Israel [15 34 35] In Beijing it was shown thatthe VPD increased with air temperature and decreased withrelative humidity from 1951 to 2010 [15] Cohen et al (2002)showed that themain factor responsible for the increased panevaporation was the growth in the aerodynamic componentof evaporation which was due to increases in both the airVPD and the wind speed at Bet Dagan from 1964 to 1997 [35]
The vapor pressure deficit represents a gradient acrosswhich water vapor is removed from the evapotranspiringsurface into the surrounding air [12] A greater vapor pressuredeficit generally causes a higher evaporative rate Hence the
increasing vapor pressure deficit in the Shenzhen area willresult in increasing plant evapotranspiration
44 ET0 Change and City Development In the current studyarea ET
0first decreased from 1950s to 1970s and then
increased greatly in the 1980s During the 1990s and 2000s itvaried slightly with amean value of 1287mm It was observedthat after the onset of urban development in 1978 the ET
0
value increased and became higher than this for the periodprior to the urban development Figures 2 5 6 and 8 showthat the mutation points for most climatic variables wereobserved near the onset year of urban development and sen-sitivity analysis shows that the higher ET
0during the period
of 1992ndash2012 is mainly attributed to the relative humiditydecrease and air temperature increaseHence it could be con-cluded that the quick development of Shenzhen city alteredthe climatic conditions and hence increased the local ET
0
In other large cities an increasing ET0trendwas also found in
recent decades For example in Beijing City the annual ET0
increased significantly from 1951 to 2010 and from the 1950sto 2000s it increased from 1039 to 1148mm [15] The annualpotential evapotranspiration (PET) displays a significantupward trend from 1970 to 2006 and the trend varied from 1to 4mm per year in the Pyrenees-Orientales and Audeadministrative departments respectively and the westernpart of the French Mediterranean area with an averageincrease in PET of between 34mm and 150mm in the last 36years [36]
The increasing trend in ET0that was observed in large
cities is different than this that was found in the farmlandFor example in the Haihe River basin in northern Chinadecreasing trends were observed in 26 stations while 16stations showed significant decreasing trends from 1950 to2007 with rates from minus20 to minus37mmyearminus1 [34] Similarly asignificant decreasing trend of ET
0with a rate of minus3mm per
yearwas found in the arid region of northwest China [37]Thedifference trend in ET
0between the large cities and the
farmland may be due to the variations in the energy balanceand the evaporative potential In the farmland areas morethan 60ndash80 of the net radiation is used for plant evapotran-spiration [33 38 39]This effect not only reduces the availableheat for heating the air environment but also increasesthe water vapor in the near atmosphere The latter effect mayincrease the relative humidity and reduce the vapor pressuredeficit and lastly it may reduce the reference evapotranspi-ration For urban conditions the decrease in green land willdecrease the energy consumption caused by crop evapotran-spiration increasing the available heat and decreasing thewater vapor which ultimately results in an increase in theevaporative potential
It should be noted in Figure 9 that the ET0values in 1997
and 2012 were much lower than those in the neighboringyears It is estimated that ET
0in 1997 and 2012 decreased
by minus104 and minus130 compared to the mean value for theperiod from 1990 to 2012 The sensitive analysis methoddescribed in Section 33 was used to determine the effects ofeach variable on the ET
0changes in 1997 and 2012 compared
to the mean value during the period of 1990ndash2012 The
10 Advances in Meteorology
results showed that the changes in the relative humidity airtemperature sunshine hours andwind speed in 1997 resultedin changes in ET
0by minus72 minus08 minus20 and minus05 respectively
and by minus126 minus08 minus06 and minus42 in 2012 respectively Itcould be concluded that the increase in the relative humidityis the main factor for ET
0reduction followed by the wind
speed air temperature and sunshine hours in the two yearsBased on the mean values of the climatic variables
averaged over the periods of 1954ndash1978 and 1992ndash2012 ET0
increased by 145 in the latter period For the sensitivityanalysis changes in the relative humidity air temperaturewind speed and sunshine hours during 1992ndash2012 caused thevariation of ET
0by 146 50 minus13 and minus38 respectively
compared to those for the period 1954ndash1978The total amountof change in ET
0was 156 based on the sensitivity analysis
This value is similar to the rate of increase of 145 by com-parison of the ET
0values between the two time periods Liu
et al (2014) [15] calculated the ET0change rates by directly
comparing the mean values and summing each ET0change
rate caused by climatic variables using the same sensitiveanalysis method the ET
0change rates were 107 and 105
respectively Liu et al (2009) [40] found that ET0inside the
screenhouse was reduced by 39 compared to that in theopen field By considering the effect of each climatic variablechange to ET
0using the sensitivity analysis the total ET
0
change rate sums to 44 which is similar to the value of39 Therefore it could be concluded that the sensitivity-analysis method used in this study is reliable and easy to useand hence it is recommended for the analysis of the effect ofclimate change on ET
0
5 Conclusions
(1) The development of Shenzhen city greatly affected thelocal climatic conditions Before the onset of urbandevelopment each climatic variable varied slightlywhereas afterward the air temperature increased sig-nificantly and the sunshine hours and relative humid-ity decreased significantly The mutation point formost climatic variables is observed at approximately1978 the onset year for urban development
(2) ET0first decreased from 1954 to 1978 and then
increased quickly and reached a maximal value of1373mm during the period from 1992 to 2012 Themean ET
0value for the period from 1954 to 1978 was
1110mm and increased to 1284mm during the periodfrom 1992 to 2012 indicating an increasing trend ofthe evaporative demand
(3) Sensitivity analysis showed that ET0is most sensitive
to relative humidity followed by air temperaturesunshine hours and wind speed
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This paper is partially supported by the National ScienceFoundation of China (Grant nos 51179005 51479004) Theauthors greatly acknowledge the comments from the editorand the two anonymous reviewers
References
[1] IPCC Climate Change 2014 Synthesis Report An Assessmentof Intergovernmental Panel on Climate Change IPCC GenevaSwitzerland 2014 httpipccchindexhtml
[2] J D Haskett Y A Pachepsky and B Acock ldquoEffect of climateand atmospheric change on soybean water stress a study ofIowardquo Ecological Modelling vol 135 no 2-3 pp 265ndash277 2000
[3] T G Huntington ldquoEvidence for intensification of the globalwater cycle review and synthesisrdquo Journal of Hydrology vol 319no 1ndash4 pp 83ndash95 2006
[4] B C Bates Z W Kundzewicz S Wu and J Palutikof ldquoClimatechange and waterrdquo Technical Paper of the IntergovernmentalPanel on Climate Change IPCC Secretariat Geneva Switzer-land 2008
[5] C M Philandras D A Metaxas and P T Nastos ldquoClimatevariability and urbanization in AthensrdquoTheoretical and AppliedClimatology vol 63 no 1-2 pp 65ndash72 1999
[6] R L Wilby ldquoPast and projected trends in Londonrsquos urban heatislandrdquoWeather vol 58 no 7 pp 251ndash260 2003
[7] N Schwarz U Schlink U Franck and K Groszligmann ldquoRela-tionship of land surface and air temperatures and its implica-tions for quantifying urban heat island indicatorsmdashan applica-tion for the city of Leipzig (Germany)rdquoEcological Indicators vol18 pp 693ndash704 2012
[8] H Wang L Fu X Lin Y Zhou and J C Chen ldquoA bottom-up methodology to estimate vehicle emissions for the Beijingurban areardquo Science of the Total Environment vol 407 no 6 pp1947ndash1953 2009
[9] DESE (Department of Environmental Science and Engineer-ingTsinghua University) Mobile Source Database EmissionInventory and Treatment Proposal for Beijing Tsinghua Univer-sity Beijing China 2005
[10] H Kan S J London G Chen et al ldquoDifferentiating the effectsof fine and coarse particles on daily mortality in ShanghaiChinardquo Environment International vol 33 no 3 pp 376ndash3842007
[11] A Aziz and I U Bajwa ldquoErroneous mass transit system andits tended relationship with motor vehicular air pollution (Anintegrated approach for reduction of urban air pollution inLahore)rdquo Environmental Monitoring and Assessment vol 137no 1ndash3 pp 25ndash33 2008
[12] R G Allen L S Perreira D Raes and M Smith Crop Evap-otranspiration Guidelines for Computing Crop Water Require-ments FAO Irrigation and Drainage Paper no 56 FAO RomeItaly 1998
[13] K H Hamed ldquoTrend detection in hydrologic data the Mann-Kendall trend test under the scaling hypothesisrdquo Journal ofHydrology vol 349 no 3-4 pp 350ndash363 2008
[14] L Q Liang L J Li andQ Liu ldquoTemporal variation of referenceevapotranspiration during 1961ndash2005 in the Taoer River basin ofNortheast Chinardquo Agricultural and Forest Meteorology vol 150no 2 pp 298ndash306 2010
Advances in Meteorology 11
[15] H Liu Y Li T Josef R H Zhang and G H HuangldquoQuantitative estimation of climate change effects on potentialevapotranspiration in Beijing during 1951ndash2010rdquo Journal ofGeographical Sciences vol 24 no 1 pp 93ndash112 2014
[16] M G Kendall and A StuartThe Advanced Theory of StatisticsGriffin London UK 1973
[17] Q-Y Tang and C-X Zhang ldquoData Processing System (DPS)software with experimental design statistical analysis and datamining developed for use in entomological researchrdquo InsectScience vol 20 no 2 pp 254ndash260 2013
[18] M Moller J Tanny Y Li and S Cohen ldquoMeasuring andpredicting evapotranspiration in an insect-proof screenhouserdquoAgricultural and Forest Meteorology vol 127 no 1-2 pp 35ndash512004
[19] G Kitsara G Papaioannou A Papathanasiou and A RetalisldquoDimmingbrightening in athens trends in sunshine durationcloud cover and reference evapotranspirationrdquoWater ResourcesManagement vol 27 no 6 pp 1623ndash1633 2013
[20] Y Wang Y Yang S Han Q X Wang and G H ZhangldquoSunshine dimming and brightening in Chinese cities (1955ndash2011) was driven by air pollution rather than cloudsrdquo ClimateResearch vol 56 no 1 pp 11ndash20 2013
[21] G Stanhill and S Cohen ldquoGlobal dimming a review of theevidence for a widespread and significant reduction in globalradiation with discussion of its probable causes and possibleagricultural consequencesrdquoAgricultural and ForestMeteorologyvol 107 no 4 pp 255ndash278 2001
[22] Q Liu and Z Yang ldquoQuantitative estimation of the impact ofclimate change on actual evapotranspiration in the Yellow RiverBasin Chinardquo Journal of Hydrology vol 395 no 3-4 pp 226ndash234 2010
[23] D I Stern ldquoReversal of the trend in global anthropogenic sulfuremissionsrdquoGlobal Environmental Change vol 16 no 2 pp 207ndash220 2006
[24] I Koren J V Martins L A Remer and H Afargan ldquoSmokeinvigoration versus inhibition of clouds over the AmazonrdquoScience vol 321 no 5891 pp 946ndash949 2008
[25] D Rosenfeld Y J Kaufman and I Koren ldquoSwitching cloudcover and dynamical regimes from open to closed Benard cellsin response to the suppression of precipitation by aerosolsrdquoAtmospheric Chemistry and Physics vol 6 no 9 pp 2503ndash25112006
[26] C Ruckstuhl R Philipona K Behrens et al ldquoAerosol and cloudeffects on solar brightening and the recent rapid warmingrdquoGeophysical Research Letters vol 35 no 12 Article ID L127082008
[27] DG Streets Y Fang CMian et al ldquoAnthropogenic andnaturalcontributions to regional trends in aerosol optical depth 1980ndash2006rdquo Journal of Geophysical Research Atmospheres vol 114 no10 Article ID D00D18 2009
[28] V Ramanathan P J Crutzen J T Kiehl and D RosenfeldldquoAtmospheremdashaerosols climate and the hydrological cyclerdquoScience vol 294 no 5549 pp 2119ndash2124 2001
[29] M Wild ldquoEnlightening global dimming and brighteningrdquoBulletin of the AmericanMeteorological Society vol 93 no 1 pp27ndash37 2012
[30] G D Liu Y Li H J Liu and J Xiao ldquoChanging trend of refer-ence crop evapotranspiration and its dominatedmeteorologicalvariables in Shanxi province in the past 55 yearsrdquo Journal ofIrrigation and Drainage vol 31 no 4 pp 26ndash30 2012
[31] C-S Rim ldquoThe effects of urbanization geographical and topo-graphical conditions on reference evapotranspirationrdquo ClimaticChange vol 97 no 3 pp 483ndash514 2009
[32] WKuang Y Liu YDou et al ldquoWhat are hot andwhat are not inan urban landscape quantifying and explaining the land surfacetemperature pattern in Beijing Chinardquo Landscape Ecology2014
[33] Z Qin Q Yu S Xu et al ldquoWater heat fluxes and water useefficiency measurement and modeling above a farmland in theNorth China Plainrdquo Science in China D Earth Sciences vol 48no 1 pp 207ndash217 2005
[34] B Tang L Tong S Z Kang and L Zhang ldquoImpacts ofclimate variability on reference evapotranspiration over 58 yearsin the Haihe river basin of north Chinardquo Agricultural WaterManagement vol 98 no 10 pp 1660ndash1670 2011
[35] S Cohen A Ianetz and G Stanhill ldquoEvaporative climatechanges at BetDagan Israel 1964ndash1998rdquoAgricultural and ForestMeteorology vol 111 no 2 pp 83ndash91 2002
[36] K Chaouche L Neppel C Dieulin et al ldquoAnalyses of precipita-tion temperature and evapotranspiration in a French Mediter-ranean region in the context of climate changerdquoComptes RendusGeoscience vol 342 no 3 pp 234ndash243 2010
[37] Z Huo X Dai S Feng S Kang and G Huang ldquoEffect of cli-mate change on reference evapotranspiration and aridity indexin arid region of Chinardquo Journal of Hydrology vol 492 pp 24ndash34 2013
[38] G Peng X Cai H Zhang A Li F Hu and M Y LeclercldquoHeat flux apportionment to heterogeneous surfaces using fluxfootprint analysisrdquo Advances in Atmospheric Sciences vol 25no 1 pp 107ndash116 2008
[39] Y Q Zhang Y J Shen C M Liu et al ldquoMeasurement andanalysis of water heat and CO
2flux from a farmland in the
North China plainrdquo Acta Geographica Sinica vol 57 no 3 pp333ndash342 2002 (Chinese)
[40] H-J Liu G-H Huang S Cohen and J Tanny ldquoChange in cropevapotranspiration and associated influencing factors underscreenhouse conditionsrdquo Chinese Journal of Eco-Agriculturevol 17 no 3 pp 484ndash488 2009 (Chinese)
Average ET0 over 1954ndash1978Average ET0 over 1992ndash2012
(a)
UF-
UB
UFUB
1954 1962 1970 1978 1986 1994 2002 2010
Year
minus8
minus4
4
8
0
(b)
Figure 9 Temporal trend of the yearly total ET0(a) and the trend
results from the Mann-Kendall method (b)
and the total ET0in this period accounts for approximately
13 the yearly total Monthly ET0s in the period of 1992ndash2012
are all higher by 5ndash30mm or 7ndash25 than those in the 1954ndash1978 period The most increases in monthly ET
0are found in
period from July to September and the lowest from January toApril with an increase of less than 7mmTherefore the greatincrease in ET
0in summer (generally from June to October)
makes main contribution to yearly ET0
33 Sensitivity Analysis of ET0 to the Change of ClimaticVariables Figures 2 to 8 show that most mutation pointsfor most of the climatic variables were found during theperiod from 1978 to 1992 and the mean values of the climaticvariables over the period from 1954 to 1978 were significantly
150
100
50
01 2 3 4 5 6 7 8 9 10 11 12
Months
Monthly mean from period of 1954ndash2012
Monthly mean from period of 1954ndash1978Monthly mean from period of 1992ndash2012
50
40
30
20
10
0Mon
thly
tota
ldie
renc
eET 0
(mm
)
Incr
ease
per
cent
age i
nET
0(
)
Increase in ET0 from 1954ndash1978 to 1992ndash2012Increase percentage in ET0
Figure 10 Yearly distribution of ET0averaged in the periods of
1954ndash1978 1992ndash2012 and 1954ndash2012 ET0increase amount and the
corresponding increase percentage in each month from periods of1954ndash1978 to 1992ndash2012 were showed
(119875 lt 001) higherlower than those during the period from1992 to 2012 Hence the mean values of each variable duringthe two periods from 1954 to 1978 and from 1992 to 2012 werecalculated and used to analyze their change effects on the ET
0
changes following themethod described in Section 233Thesummary of the mean values of each climate and their effectson ET
0were listed in Table 1 and Figure 11
It can be seen in Table 1 that the daily mean relativehumidity daily mean air temperature wind speed and sun-shine hours from the period of 1954ndash1978 to 1992ndash2012 wereminus72 59 minus60 and minus171 respectively which resulted inET0changes by 146 53 minus13 and minus32 respectively The
contribution of each climate variablersquos variation from 1954ndash1978 to 1992ndash2012 to ET
0change is shown in Figure 11 It
can be found that decrease in relative humidity accounted forapproximately 60 variation in ET
0 followed by temperature
increase with a contribution of 22 and sunshine hoursreduction of 13 Wind speed accounted for 6 variationin ET
0 Similarly in another mega city Beijing in China the
order of climate change to ET0variation frommain to weak is
air temperature relative humidity sunshine hours and windspeed [14]
8 Advances in Meteorology
Relative humidity
60Wind speed6
Temperature22
Sunshine hours13
Figure 11 The contribution of each climatic variable change to ET0variation
300
200
100
0
Dev
elopm
ent o
f She
nzhe
n ci
ty
Heavy industry output value (billion dollars)Resident population (10minus1 million)Electric energy production (102 MWh)
1954 1962 1970 1978 1986 1994 2002 2010
Year
Figure 12 Temporal trend of the development of Shenzhen city
4 Discussion
41 Change in the Sunshine Hours and Urban DevelopmentSunshine hours generally depend on the cloud cover man-made aerosols and certain air pollutants (including SO
2
NO119909 and PM) [19 20] Recent studies indicated that the
most probable cause for the depression of sunshine hours orsolar radiation is in the increased concentrations ofmanmadeaerosols and other air pollutants [1 19ndash22] In this studythe sunshine hours and amount of radiation showed cleardecreasing trends after 1970 and theywere significantly lowerafter 1980 during rapid urban development
Figure 12 shows the increasing trends of the residentpopulation heavy industry output value and electric energyproduction It is confirmed that the increasing trend in theenergy consumption corresponds to increased emission ofpolluted particles including CO CO
2 NO119909 SO119909 PM25 O3
andHCThese particlesmay result in an increase in the atmo-spheric aerosol concentration [23ndash27] which can directlyattenuate the surface solar radiation (SSR) by scattering andabsorbing solar radiation (direct effect) or can indirectlyattenuate SSR by their ability to act as cloud condensation
nuclei thereby increasing the cloud reflectivity and lifetime(first and second indirect effects) [24 28] A remarkabledecline in SSR between the 1950s and the 1980s was found inseveral studies that were performed at selected observationstations based on sites in Europe the Baltics the South PoleGermany and the former Soviet Union [21] Today a com-prehensive literature exists that confirms the declines of SSRduring this period in many places around the world [29]In Beijing and several other Chinese cities the decline ofsunshine hours and SSR has also been reported [15 30]
42 Air Temperature Change and Urban Development It hasbeen confirmed that there is a large air temperature differ-ence in urban and rural areas [4ndash7] These air temperaturedifferences result from the influence of the thermal emissivityproperties of urban surfaces and the three-dimensional con-figuration and heat capacity of erected structures onto the airtemperature patterns in an urban region [7 15 31]Most of theworldrsquos cities thus show higher air temperatures in the urbancore than in the surrounding rural areas [4 5 7] For examplein Beijing City China Kuang et al [32] measured that themean land surface temperature of urban impervious surfaceswas about 6ndash12∘C higher than that of the urban green spaceand that in built-up areaswas on average 3ndash6∘Chigher than inrural areas They showed the main reason is the higher ratioof sensible heat to net radiation (063) and lower ratio of thelatent heat to net radiation (019) on the urban impervioussurface as compared to the corresponding rates of 030 and063 in green space and cropland In this study there is noobvious temporal trend in the air temperature before 1978whereas after that time the temperature increased greatlyalthough it showed a slight declining trend in recent yearsThe large air temperature increase from 1978 to 2002 may bedue to the increasing area of construction and the decrease infarmland [32] Figure 13 shows the changes in the farmlandand construction areas It can be found that there was arapid increase in construction areas and an abrupt decreasein farmland areas after 1978 which led to a change in thethermal balance and then resulted in the increase in the airtemperature in Shenzhen city
Advances in Meteorology 9
1954 1962 1970 1978 1986 1994 2002 2010
Year
8000
6000
4000
2000
0Farm
land
and
cons
truc
tion
area
Farm land area (101 ha)Construction area (ha)
Figure 13 Temporal trend of the farmland area and the floor spaceof the buildings under construction
43 Changes in the Relative Humidity and the Vapor PressureDeficit due to Urban Development The vapor pressure deficitwas calculated by the air temperature and the relative humid-ity following themethod of Allen et al (1998) [12] It is clearlyshown in the calculation method that the vapor pressuredeficit (VPD) increases with air temperature and decreaseswith relative humidity In this study both relative humidityand mean air temperature varied slightly during the periodfrom 1954 to 1980 (Figures 5 and 6) which consequentlyresulted in a slight change in the vapor pressure deficit Afterthat period the mean air temperature increased significantlyand the relative humidity was decreased remarkably Thesetrends resulted in an increase in the vapor pressure deficit(Figure 7)
The relative humidity is defined as the ratio of the watervapor density (mass per unit volume) to the saturation watervapor density and it is also approximately the ratio of theactual to the saturation vapor pressure A greater evaporativearea may produce more water vapor for the atmosphere andthen increase the water vapor density as well as the relativehumidity at a given temperature In the city study area thefarmland area decreased whereas the construction areaincreased (Figure 13) These data indicate a decreasing trendin the evaporative area which may result in a decrease in thewater vapor density as well as the relative humidity [33] Sim-ilarly a decreasing trend of relative humidity and an increasein the vapor pressure deficit were observed in Beijing Datongin Shanxi province Zhang Jiakou in Hebei Province and BetDagan in Israel [15 34 35] In Beijing it was shown thatthe VPD increased with air temperature and decreased withrelative humidity from 1951 to 2010 [15] Cohen et al (2002)showed that themain factor responsible for the increased panevaporation was the growth in the aerodynamic componentof evaporation which was due to increases in both the airVPD and the wind speed at Bet Dagan from 1964 to 1997 [35]
The vapor pressure deficit represents a gradient acrosswhich water vapor is removed from the evapotranspiringsurface into the surrounding air [12] A greater vapor pressuredeficit generally causes a higher evaporative rate Hence the
increasing vapor pressure deficit in the Shenzhen area willresult in increasing plant evapotranspiration
44 ET0 Change and City Development In the current studyarea ET
0first decreased from 1950s to 1970s and then
increased greatly in the 1980s During the 1990s and 2000s itvaried slightly with amean value of 1287mm It was observedthat after the onset of urban development in 1978 the ET
0
value increased and became higher than this for the periodprior to the urban development Figures 2 5 6 and 8 showthat the mutation points for most climatic variables wereobserved near the onset year of urban development and sen-sitivity analysis shows that the higher ET
0during the period
of 1992ndash2012 is mainly attributed to the relative humiditydecrease and air temperature increaseHence it could be con-cluded that the quick development of Shenzhen city alteredthe climatic conditions and hence increased the local ET
0
In other large cities an increasing ET0trendwas also found in
recent decades For example in Beijing City the annual ET0
increased significantly from 1951 to 2010 and from the 1950sto 2000s it increased from 1039 to 1148mm [15] The annualpotential evapotranspiration (PET) displays a significantupward trend from 1970 to 2006 and the trend varied from 1to 4mm per year in the Pyrenees-Orientales and Audeadministrative departments respectively and the westernpart of the French Mediterranean area with an averageincrease in PET of between 34mm and 150mm in the last 36years [36]
The increasing trend in ET0that was observed in large
cities is different than this that was found in the farmlandFor example in the Haihe River basin in northern Chinadecreasing trends were observed in 26 stations while 16stations showed significant decreasing trends from 1950 to2007 with rates from minus20 to minus37mmyearminus1 [34] Similarly asignificant decreasing trend of ET
0with a rate of minus3mm per
yearwas found in the arid region of northwest China [37]Thedifference trend in ET
0between the large cities and the
farmland may be due to the variations in the energy balanceand the evaporative potential In the farmland areas morethan 60ndash80 of the net radiation is used for plant evapotran-spiration [33 38 39]This effect not only reduces the availableheat for heating the air environment but also increasesthe water vapor in the near atmosphere The latter effect mayincrease the relative humidity and reduce the vapor pressuredeficit and lastly it may reduce the reference evapotranspi-ration For urban conditions the decrease in green land willdecrease the energy consumption caused by crop evapotran-spiration increasing the available heat and decreasing thewater vapor which ultimately results in an increase in theevaporative potential
It should be noted in Figure 9 that the ET0values in 1997
and 2012 were much lower than those in the neighboringyears It is estimated that ET
0in 1997 and 2012 decreased
by minus104 and minus130 compared to the mean value for theperiod from 1990 to 2012 The sensitive analysis methoddescribed in Section 33 was used to determine the effects ofeach variable on the ET
0changes in 1997 and 2012 compared
to the mean value during the period of 1990ndash2012 The
10 Advances in Meteorology
results showed that the changes in the relative humidity airtemperature sunshine hours andwind speed in 1997 resultedin changes in ET
0by minus72 minus08 minus20 and minus05 respectively
and by minus126 minus08 minus06 and minus42 in 2012 respectively Itcould be concluded that the increase in the relative humidityis the main factor for ET
0reduction followed by the wind
speed air temperature and sunshine hours in the two yearsBased on the mean values of the climatic variables
averaged over the periods of 1954ndash1978 and 1992ndash2012 ET0
increased by 145 in the latter period For the sensitivityanalysis changes in the relative humidity air temperaturewind speed and sunshine hours during 1992ndash2012 caused thevariation of ET
0by 146 50 minus13 and minus38 respectively
compared to those for the period 1954ndash1978The total amountof change in ET
0was 156 based on the sensitivity analysis
This value is similar to the rate of increase of 145 by com-parison of the ET
0values between the two time periods Liu
et al (2014) [15] calculated the ET0change rates by directly
comparing the mean values and summing each ET0change
rate caused by climatic variables using the same sensitiveanalysis method the ET
0change rates were 107 and 105
respectively Liu et al (2009) [40] found that ET0inside the
screenhouse was reduced by 39 compared to that in theopen field By considering the effect of each climatic variablechange to ET
0using the sensitivity analysis the total ET
0
change rate sums to 44 which is similar to the value of39 Therefore it could be concluded that the sensitivity-analysis method used in this study is reliable and easy to useand hence it is recommended for the analysis of the effect ofclimate change on ET
0
5 Conclusions
(1) The development of Shenzhen city greatly affected thelocal climatic conditions Before the onset of urbandevelopment each climatic variable varied slightlywhereas afterward the air temperature increased sig-nificantly and the sunshine hours and relative humid-ity decreased significantly The mutation point formost climatic variables is observed at approximately1978 the onset year for urban development
(2) ET0first decreased from 1954 to 1978 and then
increased quickly and reached a maximal value of1373mm during the period from 1992 to 2012 Themean ET
0value for the period from 1954 to 1978 was
1110mm and increased to 1284mm during the periodfrom 1992 to 2012 indicating an increasing trend ofthe evaporative demand
(3) Sensitivity analysis showed that ET0is most sensitive
to relative humidity followed by air temperaturesunshine hours and wind speed
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This paper is partially supported by the National ScienceFoundation of China (Grant nos 51179005 51479004) Theauthors greatly acknowledge the comments from the editorand the two anonymous reviewers
References
[1] IPCC Climate Change 2014 Synthesis Report An Assessmentof Intergovernmental Panel on Climate Change IPCC GenevaSwitzerland 2014 httpipccchindexhtml
[2] J D Haskett Y A Pachepsky and B Acock ldquoEffect of climateand atmospheric change on soybean water stress a study ofIowardquo Ecological Modelling vol 135 no 2-3 pp 265ndash277 2000
[3] T G Huntington ldquoEvidence for intensification of the globalwater cycle review and synthesisrdquo Journal of Hydrology vol 319no 1ndash4 pp 83ndash95 2006
[4] B C Bates Z W Kundzewicz S Wu and J Palutikof ldquoClimatechange and waterrdquo Technical Paper of the IntergovernmentalPanel on Climate Change IPCC Secretariat Geneva Switzer-land 2008
[5] C M Philandras D A Metaxas and P T Nastos ldquoClimatevariability and urbanization in AthensrdquoTheoretical and AppliedClimatology vol 63 no 1-2 pp 65ndash72 1999
[6] R L Wilby ldquoPast and projected trends in Londonrsquos urban heatislandrdquoWeather vol 58 no 7 pp 251ndash260 2003
[7] N Schwarz U Schlink U Franck and K Groszligmann ldquoRela-tionship of land surface and air temperatures and its implica-tions for quantifying urban heat island indicatorsmdashan applica-tion for the city of Leipzig (Germany)rdquoEcological Indicators vol18 pp 693ndash704 2012
[8] H Wang L Fu X Lin Y Zhou and J C Chen ldquoA bottom-up methodology to estimate vehicle emissions for the Beijingurban areardquo Science of the Total Environment vol 407 no 6 pp1947ndash1953 2009
[9] DESE (Department of Environmental Science and Engineer-ingTsinghua University) Mobile Source Database EmissionInventory and Treatment Proposal for Beijing Tsinghua Univer-sity Beijing China 2005
[10] H Kan S J London G Chen et al ldquoDifferentiating the effectsof fine and coarse particles on daily mortality in ShanghaiChinardquo Environment International vol 33 no 3 pp 376ndash3842007
[11] A Aziz and I U Bajwa ldquoErroneous mass transit system andits tended relationship with motor vehicular air pollution (Anintegrated approach for reduction of urban air pollution inLahore)rdquo Environmental Monitoring and Assessment vol 137no 1ndash3 pp 25ndash33 2008
[12] R G Allen L S Perreira D Raes and M Smith Crop Evap-otranspiration Guidelines for Computing Crop Water Require-ments FAO Irrigation and Drainage Paper no 56 FAO RomeItaly 1998
[13] K H Hamed ldquoTrend detection in hydrologic data the Mann-Kendall trend test under the scaling hypothesisrdquo Journal ofHydrology vol 349 no 3-4 pp 350ndash363 2008
[14] L Q Liang L J Li andQ Liu ldquoTemporal variation of referenceevapotranspiration during 1961ndash2005 in the Taoer River basin ofNortheast Chinardquo Agricultural and Forest Meteorology vol 150no 2 pp 298ndash306 2010
Advances in Meteorology 11
[15] H Liu Y Li T Josef R H Zhang and G H HuangldquoQuantitative estimation of climate change effects on potentialevapotranspiration in Beijing during 1951ndash2010rdquo Journal ofGeographical Sciences vol 24 no 1 pp 93ndash112 2014
[16] M G Kendall and A StuartThe Advanced Theory of StatisticsGriffin London UK 1973
[17] Q-Y Tang and C-X Zhang ldquoData Processing System (DPS)software with experimental design statistical analysis and datamining developed for use in entomological researchrdquo InsectScience vol 20 no 2 pp 254ndash260 2013
[18] M Moller J Tanny Y Li and S Cohen ldquoMeasuring andpredicting evapotranspiration in an insect-proof screenhouserdquoAgricultural and Forest Meteorology vol 127 no 1-2 pp 35ndash512004
[19] G Kitsara G Papaioannou A Papathanasiou and A RetalisldquoDimmingbrightening in athens trends in sunshine durationcloud cover and reference evapotranspirationrdquoWater ResourcesManagement vol 27 no 6 pp 1623ndash1633 2013
[20] Y Wang Y Yang S Han Q X Wang and G H ZhangldquoSunshine dimming and brightening in Chinese cities (1955ndash2011) was driven by air pollution rather than cloudsrdquo ClimateResearch vol 56 no 1 pp 11ndash20 2013
[21] G Stanhill and S Cohen ldquoGlobal dimming a review of theevidence for a widespread and significant reduction in globalradiation with discussion of its probable causes and possibleagricultural consequencesrdquoAgricultural and ForestMeteorologyvol 107 no 4 pp 255ndash278 2001
[22] Q Liu and Z Yang ldquoQuantitative estimation of the impact ofclimate change on actual evapotranspiration in the Yellow RiverBasin Chinardquo Journal of Hydrology vol 395 no 3-4 pp 226ndash234 2010
[23] D I Stern ldquoReversal of the trend in global anthropogenic sulfuremissionsrdquoGlobal Environmental Change vol 16 no 2 pp 207ndash220 2006
[24] I Koren J V Martins L A Remer and H Afargan ldquoSmokeinvigoration versus inhibition of clouds over the AmazonrdquoScience vol 321 no 5891 pp 946ndash949 2008
[25] D Rosenfeld Y J Kaufman and I Koren ldquoSwitching cloudcover and dynamical regimes from open to closed Benard cellsin response to the suppression of precipitation by aerosolsrdquoAtmospheric Chemistry and Physics vol 6 no 9 pp 2503ndash25112006
[26] C Ruckstuhl R Philipona K Behrens et al ldquoAerosol and cloudeffects on solar brightening and the recent rapid warmingrdquoGeophysical Research Letters vol 35 no 12 Article ID L127082008
[27] DG Streets Y Fang CMian et al ldquoAnthropogenic andnaturalcontributions to regional trends in aerosol optical depth 1980ndash2006rdquo Journal of Geophysical Research Atmospheres vol 114 no10 Article ID D00D18 2009
[28] V Ramanathan P J Crutzen J T Kiehl and D RosenfeldldquoAtmospheremdashaerosols climate and the hydrological cyclerdquoScience vol 294 no 5549 pp 2119ndash2124 2001
[29] M Wild ldquoEnlightening global dimming and brighteningrdquoBulletin of the AmericanMeteorological Society vol 93 no 1 pp27ndash37 2012
[30] G D Liu Y Li H J Liu and J Xiao ldquoChanging trend of refer-ence crop evapotranspiration and its dominatedmeteorologicalvariables in Shanxi province in the past 55 yearsrdquo Journal ofIrrigation and Drainage vol 31 no 4 pp 26ndash30 2012
[31] C-S Rim ldquoThe effects of urbanization geographical and topo-graphical conditions on reference evapotranspirationrdquo ClimaticChange vol 97 no 3 pp 483ndash514 2009
[32] WKuang Y Liu YDou et al ldquoWhat are hot andwhat are not inan urban landscape quantifying and explaining the land surfacetemperature pattern in Beijing Chinardquo Landscape Ecology2014
[33] Z Qin Q Yu S Xu et al ldquoWater heat fluxes and water useefficiency measurement and modeling above a farmland in theNorth China Plainrdquo Science in China D Earth Sciences vol 48no 1 pp 207ndash217 2005
[34] B Tang L Tong S Z Kang and L Zhang ldquoImpacts ofclimate variability on reference evapotranspiration over 58 yearsin the Haihe river basin of north Chinardquo Agricultural WaterManagement vol 98 no 10 pp 1660ndash1670 2011
[35] S Cohen A Ianetz and G Stanhill ldquoEvaporative climatechanges at BetDagan Israel 1964ndash1998rdquoAgricultural and ForestMeteorology vol 111 no 2 pp 83ndash91 2002
[36] K Chaouche L Neppel C Dieulin et al ldquoAnalyses of precipita-tion temperature and evapotranspiration in a French Mediter-ranean region in the context of climate changerdquoComptes RendusGeoscience vol 342 no 3 pp 234ndash243 2010
[37] Z Huo X Dai S Feng S Kang and G Huang ldquoEffect of cli-mate change on reference evapotranspiration and aridity indexin arid region of Chinardquo Journal of Hydrology vol 492 pp 24ndash34 2013
[38] G Peng X Cai H Zhang A Li F Hu and M Y LeclercldquoHeat flux apportionment to heterogeneous surfaces using fluxfootprint analysisrdquo Advances in Atmospheric Sciences vol 25no 1 pp 107ndash116 2008
[39] Y Q Zhang Y J Shen C M Liu et al ldquoMeasurement andanalysis of water heat and CO
2flux from a farmland in the
North China plainrdquo Acta Geographica Sinica vol 57 no 3 pp333ndash342 2002 (Chinese)
[40] H-J Liu G-H Huang S Cohen and J Tanny ldquoChange in cropevapotranspiration and associated influencing factors underscreenhouse conditionsrdquo Chinese Journal of Eco-Agriculturevol 17 no 3 pp 484ndash488 2009 (Chinese)
Figure 11 The contribution of each climatic variable change to ET0variation
300
200
100
0
Dev
elopm
ent o
f She
nzhe
n ci
ty
Heavy industry output value (billion dollars)Resident population (10minus1 million)Electric energy production (102 MWh)
1954 1962 1970 1978 1986 1994 2002 2010
Year
Figure 12 Temporal trend of the development of Shenzhen city
4 Discussion
41 Change in the Sunshine Hours and Urban DevelopmentSunshine hours generally depend on the cloud cover man-made aerosols and certain air pollutants (including SO
2
NO119909 and PM) [19 20] Recent studies indicated that the
most probable cause for the depression of sunshine hours orsolar radiation is in the increased concentrations ofmanmadeaerosols and other air pollutants [1 19ndash22] In this studythe sunshine hours and amount of radiation showed cleardecreasing trends after 1970 and theywere significantly lowerafter 1980 during rapid urban development
Figure 12 shows the increasing trends of the residentpopulation heavy industry output value and electric energyproduction It is confirmed that the increasing trend in theenergy consumption corresponds to increased emission ofpolluted particles including CO CO
2 NO119909 SO119909 PM25 O3
andHCThese particlesmay result in an increase in the atmo-spheric aerosol concentration [23ndash27] which can directlyattenuate the surface solar radiation (SSR) by scattering andabsorbing solar radiation (direct effect) or can indirectlyattenuate SSR by their ability to act as cloud condensation
nuclei thereby increasing the cloud reflectivity and lifetime(first and second indirect effects) [24 28] A remarkabledecline in SSR between the 1950s and the 1980s was found inseveral studies that were performed at selected observationstations based on sites in Europe the Baltics the South PoleGermany and the former Soviet Union [21] Today a com-prehensive literature exists that confirms the declines of SSRduring this period in many places around the world [29]In Beijing and several other Chinese cities the decline ofsunshine hours and SSR has also been reported [15 30]
42 Air Temperature Change and Urban Development It hasbeen confirmed that there is a large air temperature differ-ence in urban and rural areas [4ndash7] These air temperaturedifferences result from the influence of the thermal emissivityproperties of urban surfaces and the three-dimensional con-figuration and heat capacity of erected structures onto the airtemperature patterns in an urban region [7 15 31]Most of theworldrsquos cities thus show higher air temperatures in the urbancore than in the surrounding rural areas [4 5 7] For examplein Beijing City China Kuang et al [32] measured that themean land surface temperature of urban impervious surfaceswas about 6ndash12∘C higher than that of the urban green spaceand that in built-up areaswas on average 3ndash6∘Chigher than inrural areas They showed the main reason is the higher ratioof sensible heat to net radiation (063) and lower ratio of thelatent heat to net radiation (019) on the urban impervioussurface as compared to the corresponding rates of 030 and063 in green space and cropland In this study there is noobvious temporal trend in the air temperature before 1978whereas after that time the temperature increased greatlyalthough it showed a slight declining trend in recent yearsThe large air temperature increase from 1978 to 2002 may bedue to the increasing area of construction and the decrease infarmland [32] Figure 13 shows the changes in the farmlandand construction areas It can be found that there was arapid increase in construction areas and an abrupt decreasein farmland areas after 1978 which led to a change in thethermal balance and then resulted in the increase in the airtemperature in Shenzhen city
Advances in Meteorology 9
1954 1962 1970 1978 1986 1994 2002 2010
Year
8000
6000
4000
2000
0Farm
land
and
cons
truc
tion
area
Farm land area (101 ha)Construction area (ha)
Figure 13 Temporal trend of the farmland area and the floor spaceof the buildings under construction
43 Changes in the Relative Humidity and the Vapor PressureDeficit due to Urban Development The vapor pressure deficitwas calculated by the air temperature and the relative humid-ity following themethod of Allen et al (1998) [12] It is clearlyshown in the calculation method that the vapor pressuredeficit (VPD) increases with air temperature and decreaseswith relative humidity In this study both relative humidityand mean air temperature varied slightly during the periodfrom 1954 to 1980 (Figures 5 and 6) which consequentlyresulted in a slight change in the vapor pressure deficit Afterthat period the mean air temperature increased significantlyand the relative humidity was decreased remarkably Thesetrends resulted in an increase in the vapor pressure deficit(Figure 7)
The relative humidity is defined as the ratio of the watervapor density (mass per unit volume) to the saturation watervapor density and it is also approximately the ratio of theactual to the saturation vapor pressure A greater evaporativearea may produce more water vapor for the atmosphere andthen increase the water vapor density as well as the relativehumidity at a given temperature In the city study area thefarmland area decreased whereas the construction areaincreased (Figure 13) These data indicate a decreasing trendin the evaporative area which may result in a decrease in thewater vapor density as well as the relative humidity [33] Sim-ilarly a decreasing trend of relative humidity and an increasein the vapor pressure deficit were observed in Beijing Datongin Shanxi province Zhang Jiakou in Hebei Province and BetDagan in Israel [15 34 35] In Beijing it was shown thatthe VPD increased with air temperature and decreased withrelative humidity from 1951 to 2010 [15] Cohen et al (2002)showed that themain factor responsible for the increased panevaporation was the growth in the aerodynamic componentof evaporation which was due to increases in both the airVPD and the wind speed at Bet Dagan from 1964 to 1997 [35]
The vapor pressure deficit represents a gradient acrosswhich water vapor is removed from the evapotranspiringsurface into the surrounding air [12] A greater vapor pressuredeficit generally causes a higher evaporative rate Hence the
increasing vapor pressure deficit in the Shenzhen area willresult in increasing plant evapotranspiration
44 ET0 Change and City Development In the current studyarea ET
0first decreased from 1950s to 1970s and then
increased greatly in the 1980s During the 1990s and 2000s itvaried slightly with amean value of 1287mm It was observedthat after the onset of urban development in 1978 the ET
0
value increased and became higher than this for the periodprior to the urban development Figures 2 5 6 and 8 showthat the mutation points for most climatic variables wereobserved near the onset year of urban development and sen-sitivity analysis shows that the higher ET
0during the period
of 1992ndash2012 is mainly attributed to the relative humiditydecrease and air temperature increaseHence it could be con-cluded that the quick development of Shenzhen city alteredthe climatic conditions and hence increased the local ET
0
In other large cities an increasing ET0trendwas also found in
recent decades For example in Beijing City the annual ET0
increased significantly from 1951 to 2010 and from the 1950sto 2000s it increased from 1039 to 1148mm [15] The annualpotential evapotranspiration (PET) displays a significantupward trend from 1970 to 2006 and the trend varied from 1to 4mm per year in the Pyrenees-Orientales and Audeadministrative departments respectively and the westernpart of the French Mediterranean area with an averageincrease in PET of between 34mm and 150mm in the last 36years [36]
The increasing trend in ET0that was observed in large
cities is different than this that was found in the farmlandFor example in the Haihe River basin in northern Chinadecreasing trends were observed in 26 stations while 16stations showed significant decreasing trends from 1950 to2007 with rates from minus20 to minus37mmyearminus1 [34] Similarly asignificant decreasing trend of ET
0with a rate of minus3mm per
yearwas found in the arid region of northwest China [37]Thedifference trend in ET
0between the large cities and the
farmland may be due to the variations in the energy balanceand the evaporative potential In the farmland areas morethan 60ndash80 of the net radiation is used for plant evapotran-spiration [33 38 39]This effect not only reduces the availableheat for heating the air environment but also increasesthe water vapor in the near atmosphere The latter effect mayincrease the relative humidity and reduce the vapor pressuredeficit and lastly it may reduce the reference evapotranspi-ration For urban conditions the decrease in green land willdecrease the energy consumption caused by crop evapotran-spiration increasing the available heat and decreasing thewater vapor which ultimately results in an increase in theevaporative potential
It should be noted in Figure 9 that the ET0values in 1997
and 2012 were much lower than those in the neighboringyears It is estimated that ET
0in 1997 and 2012 decreased
by minus104 and minus130 compared to the mean value for theperiod from 1990 to 2012 The sensitive analysis methoddescribed in Section 33 was used to determine the effects ofeach variable on the ET
0changes in 1997 and 2012 compared
to the mean value during the period of 1990ndash2012 The
10 Advances in Meteorology
results showed that the changes in the relative humidity airtemperature sunshine hours andwind speed in 1997 resultedin changes in ET
0by minus72 minus08 minus20 and minus05 respectively
and by minus126 minus08 minus06 and minus42 in 2012 respectively Itcould be concluded that the increase in the relative humidityis the main factor for ET
0reduction followed by the wind
speed air temperature and sunshine hours in the two yearsBased on the mean values of the climatic variables
averaged over the periods of 1954ndash1978 and 1992ndash2012 ET0
increased by 145 in the latter period For the sensitivityanalysis changes in the relative humidity air temperaturewind speed and sunshine hours during 1992ndash2012 caused thevariation of ET
0by 146 50 minus13 and minus38 respectively
compared to those for the period 1954ndash1978The total amountof change in ET
0was 156 based on the sensitivity analysis
This value is similar to the rate of increase of 145 by com-parison of the ET
0values between the two time periods Liu
et al (2014) [15] calculated the ET0change rates by directly
comparing the mean values and summing each ET0change
rate caused by climatic variables using the same sensitiveanalysis method the ET
0change rates were 107 and 105
respectively Liu et al (2009) [40] found that ET0inside the
screenhouse was reduced by 39 compared to that in theopen field By considering the effect of each climatic variablechange to ET
0using the sensitivity analysis the total ET
0
change rate sums to 44 which is similar to the value of39 Therefore it could be concluded that the sensitivity-analysis method used in this study is reliable and easy to useand hence it is recommended for the analysis of the effect ofclimate change on ET
0
5 Conclusions
(1) The development of Shenzhen city greatly affected thelocal climatic conditions Before the onset of urbandevelopment each climatic variable varied slightlywhereas afterward the air temperature increased sig-nificantly and the sunshine hours and relative humid-ity decreased significantly The mutation point formost climatic variables is observed at approximately1978 the onset year for urban development
(2) ET0first decreased from 1954 to 1978 and then
increased quickly and reached a maximal value of1373mm during the period from 1992 to 2012 Themean ET
0value for the period from 1954 to 1978 was
1110mm and increased to 1284mm during the periodfrom 1992 to 2012 indicating an increasing trend ofthe evaporative demand
(3) Sensitivity analysis showed that ET0is most sensitive
to relative humidity followed by air temperaturesunshine hours and wind speed
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This paper is partially supported by the National ScienceFoundation of China (Grant nos 51179005 51479004) Theauthors greatly acknowledge the comments from the editorand the two anonymous reviewers
References
[1] IPCC Climate Change 2014 Synthesis Report An Assessmentof Intergovernmental Panel on Climate Change IPCC GenevaSwitzerland 2014 httpipccchindexhtml
[2] J D Haskett Y A Pachepsky and B Acock ldquoEffect of climateand atmospheric change on soybean water stress a study ofIowardquo Ecological Modelling vol 135 no 2-3 pp 265ndash277 2000
[3] T G Huntington ldquoEvidence for intensification of the globalwater cycle review and synthesisrdquo Journal of Hydrology vol 319no 1ndash4 pp 83ndash95 2006
[4] B C Bates Z W Kundzewicz S Wu and J Palutikof ldquoClimatechange and waterrdquo Technical Paper of the IntergovernmentalPanel on Climate Change IPCC Secretariat Geneva Switzer-land 2008
[5] C M Philandras D A Metaxas and P T Nastos ldquoClimatevariability and urbanization in AthensrdquoTheoretical and AppliedClimatology vol 63 no 1-2 pp 65ndash72 1999
[6] R L Wilby ldquoPast and projected trends in Londonrsquos urban heatislandrdquoWeather vol 58 no 7 pp 251ndash260 2003
[7] N Schwarz U Schlink U Franck and K Groszligmann ldquoRela-tionship of land surface and air temperatures and its implica-tions for quantifying urban heat island indicatorsmdashan applica-tion for the city of Leipzig (Germany)rdquoEcological Indicators vol18 pp 693ndash704 2012
[8] H Wang L Fu X Lin Y Zhou and J C Chen ldquoA bottom-up methodology to estimate vehicle emissions for the Beijingurban areardquo Science of the Total Environment vol 407 no 6 pp1947ndash1953 2009
[9] DESE (Department of Environmental Science and Engineer-ingTsinghua University) Mobile Source Database EmissionInventory and Treatment Proposal for Beijing Tsinghua Univer-sity Beijing China 2005
[10] H Kan S J London G Chen et al ldquoDifferentiating the effectsof fine and coarse particles on daily mortality in ShanghaiChinardquo Environment International vol 33 no 3 pp 376ndash3842007
[11] A Aziz and I U Bajwa ldquoErroneous mass transit system andits tended relationship with motor vehicular air pollution (Anintegrated approach for reduction of urban air pollution inLahore)rdquo Environmental Monitoring and Assessment vol 137no 1ndash3 pp 25ndash33 2008
[12] R G Allen L S Perreira D Raes and M Smith Crop Evap-otranspiration Guidelines for Computing Crop Water Require-ments FAO Irrigation and Drainage Paper no 56 FAO RomeItaly 1998
[13] K H Hamed ldquoTrend detection in hydrologic data the Mann-Kendall trend test under the scaling hypothesisrdquo Journal ofHydrology vol 349 no 3-4 pp 350ndash363 2008
[14] L Q Liang L J Li andQ Liu ldquoTemporal variation of referenceevapotranspiration during 1961ndash2005 in the Taoer River basin ofNortheast Chinardquo Agricultural and Forest Meteorology vol 150no 2 pp 298ndash306 2010
Advances in Meteorology 11
[15] H Liu Y Li T Josef R H Zhang and G H HuangldquoQuantitative estimation of climate change effects on potentialevapotranspiration in Beijing during 1951ndash2010rdquo Journal ofGeographical Sciences vol 24 no 1 pp 93ndash112 2014
[16] M G Kendall and A StuartThe Advanced Theory of StatisticsGriffin London UK 1973
[17] Q-Y Tang and C-X Zhang ldquoData Processing System (DPS)software with experimental design statistical analysis and datamining developed for use in entomological researchrdquo InsectScience vol 20 no 2 pp 254ndash260 2013
[18] M Moller J Tanny Y Li and S Cohen ldquoMeasuring andpredicting evapotranspiration in an insect-proof screenhouserdquoAgricultural and Forest Meteorology vol 127 no 1-2 pp 35ndash512004
[19] G Kitsara G Papaioannou A Papathanasiou and A RetalisldquoDimmingbrightening in athens trends in sunshine durationcloud cover and reference evapotranspirationrdquoWater ResourcesManagement vol 27 no 6 pp 1623ndash1633 2013
[20] Y Wang Y Yang S Han Q X Wang and G H ZhangldquoSunshine dimming and brightening in Chinese cities (1955ndash2011) was driven by air pollution rather than cloudsrdquo ClimateResearch vol 56 no 1 pp 11ndash20 2013
[21] G Stanhill and S Cohen ldquoGlobal dimming a review of theevidence for a widespread and significant reduction in globalradiation with discussion of its probable causes and possibleagricultural consequencesrdquoAgricultural and ForestMeteorologyvol 107 no 4 pp 255ndash278 2001
[22] Q Liu and Z Yang ldquoQuantitative estimation of the impact ofclimate change on actual evapotranspiration in the Yellow RiverBasin Chinardquo Journal of Hydrology vol 395 no 3-4 pp 226ndash234 2010
[23] D I Stern ldquoReversal of the trend in global anthropogenic sulfuremissionsrdquoGlobal Environmental Change vol 16 no 2 pp 207ndash220 2006
[24] I Koren J V Martins L A Remer and H Afargan ldquoSmokeinvigoration versus inhibition of clouds over the AmazonrdquoScience vol 321 no 5891 pp 946ndash949 2008
[25] D Rosenfeld Y J Kaufman and I Koren ldquoSwitching cloudcover and dynamical regimes from open to closed Benard cellsin response to the suppression of precipitation by aerosolsrdquoAtmospheric Chemistry and Physics vol 6 no 9 pp 2503ndash25112006
[26] C Ruckstuhl R Philipona K Behrens et al ldquoAerosol and cloudeffects on solar brightening and the recent rapid warmingrdquoGeophysical Research Letters vol 35 no 12 Article ID L127082008
[27] DG Streets Y Fang CMian et al ldquoAnthropogenic andnaturalcontributions to regional trends in aerosol optical depth 1980ndash2006rdquo Journal of Geophysical Research Atmospheres vol 114 no10 Article ID D00D18 2009
[28] V Ramanathan P J Crutzen J T Kiehl and D RosenfeldldquoAtmospheremdashaerosols climate and the hydrological cyclerdquoScience vol 294 no 5549 pp 2119ndash2124 2001
[29] M Wild ldquoEnlightening global dimming and brighteningrdquoBulletin of the AmericanMeteorological Society vol 93 no 1 pp27ndash37 2012
[30] G D Liu Y Li H J Liu and J Xiao ldquoChanging trend of refer-ence crop evapotranspiration and its dominatedmeteorologicalvariables in Shanxi province in the past 55 yearsrdquo Journal ofIrrigation and Drainage vol 31 no 4 pp 26ndash30 2012
[31] C-S Rim ldquoThe effects of urbanization geographical and topo-graphical conditions on reference evapotranspirationrdquo ClimaticChange vol 97 no 3 pp 483ndash514 2009
[32] WKuang Y Liu YDou et al ldquoWhat are hot andwhat are not inan urban landscape quantifying and explaining the land surfacetemperature pattern in Beijing Chinardquo Landscape Ecology2014
[33] Z Qin Q Yu S Xu et al ldquoWater heat fluxes and water useefficiency measurement and modeling above a farmland in theNorth China Plainrdquo Science in China D Earth Sciences vol 48no 1 pp 207ndash217 2005
[34] B Tang L Tong S Z Kang and L Zhang ldquoImpacts ofclimate variability on reference evapotranspiration over 58 yearsin the Haihe river basin of north Chinardquo Agricultural WaterManagement vol 98 no 10 pp 1660ndash1670 2011
[35] S Cohen A Ianetz and G Stanhill ldquoEvaporative climatechanges at BetDagan Israel 1964ndash1998rdquoAgricultural and ForestMeteorology vol 111 no 2 pp 83ndash91 2002
[36] K Chaouche L Neppel C Dieulin et al ldquoAnalyses of precipita-tion temperature and evapotranspiration in a French Mediter-ranean region in the context of climate changerdquoComptes RendusGeoscience vol 342 no 3 pp 234ndash243 2010
[37] Z Huo X Dai S Feng S Kang and G Huang ldquoEffect of cli-mate change on reference evapotranspiration and aridity indexin arid region of Chinardquo Journal of Hydrology vol 492 pp 24ndash34 2013
[38] G Peng X Cai H Zhang A Li F Hu and M Y LeclercldquoHeat flux apportionment to heterogeneous surfaces using fluxfootprint analysisrdquo Advances in Atmospheric Sciences vol 25no 1 pp 107ndash116 2008
[39] Y Q Zhang Y J Shen C M Liu et al ldquoMeasurement andanalysis of water heat and CO
2flux from a farmland in the
North China plainrdquo Acta Geographica Sinica vol 57 no 3 pp333ndash342 2002 (Chinese)
[40] H-J Liu G-H Huang S Cohen and J Tanny ldquoChange in cropevapotranspiration and associated influencing factors underscreenhouse conditionsrdquo Chinese Journal of Eco-Agriculturevol 17 no 3 pp 484ndash488 2009 (Chinese)
Figure 13 Temporal trend of the farmland area and the floor spaceof the buildings under construction
43 Changes in the Relative Humidity and the Vapor PressureDeficit due to Urban Development The vapor pressure deficitwas calculated by the air temperature and the relative humid-ity following themethod of Allen et al (1998) [12] It is clearlyshown in the calculation method that the vapor pressuredeficit (VPD) increases with air temperature and decreaseswith relative humidity In this study both relative humidityand mean air temperature varied slightly during the periodfrom 1954 to 1980 (Figures 5 and 6) which consequentlyresulted in a slight change in the vapor pressure deficit Afterthat period the mean air temperature increased significantlyand the relative humidity was decreased remarkably Thesetrends resulted in an increase in the vapor pressure deficit(Figure 7)
The relative humidity is defined as the ratio of the watervapor density (mass per unit volume) to the saturation watervapor density and it is also approximately the ratio of theactual to the saturation vapor pressure A greater evaporativearea may produce more water vapor for the atmosphere andthen increase the water vapor density as well as the relativehumidity at a given temperature In the city study area thefarmland area decreased whereas the construction areaincreased (Figure 13) These data indicate a decreasing trendin the evaporative area which may result in a decrease in thewater vapor density as well as the relative humidity [33] Sim-ilarly a decreasing trend of relative humidity and an increasein the vapor pressure deficit were observed in Beijing Datongin Shanxi province Zhang Jiakou in Hebei Province and BetDagan in Israel [15 34 35] In Beijing it was shown thatthe VPD increased with air temperature and decreased withrelative humidity from 1951 to 2010 [15] Cohen et al (2002)showed that themain factor responsible for the increased panevaporation was the growth in the aerodynamic componentof evaporation which was due to increases in both the airVPD and the wind speed at Bet Dagan from 1964 to 1997 [35]
The vapor pressure deficit represents a gradient acrosswhich water vapor is removed from the evapotranspiringsurface into the surrounding air [12] A greater vapor pressuredeficit generally causes a higher evaporative rate Hence the
increasing vapor pressure deficit in the Shenzhen area willresult in increasing plant evapotranspiration
44 ET0 Change and City Development In the current studyarea ET
0first decreased from 1950s to 1970s and then
increased greatly in the 1980s During the 1990s and 2000s itvaried slightly with amean value of 1287mm It was observedthat after the onset of urban development in 1978 the ET
0
value increased and became higher than this for the periodprior to the urban development Figures 2 5 6 and 8 showthat the mutation points for most climatic variables wereobserved near the onset year of urban development and sen-sitivity analysis shows that the higher ET
0during the period
of 1992ndash2012 is mainly attributed to the relative humiditydecrease and air temperature increaseHence it could be con-cluded that the quick development of Shenzhen city alteredthe climatic conditions and hence increased the local ET
0
In other large cities an increasing ET0trendwas also found in
recent decades For example in Beijing City the annual ET0
increased significantly from 1951 to 2010 and from the 1950sto 2000s it increased from 1039 to 1148mm [15] The annualpotential evapotranspiration (PET) displays a significantupward trend from 1970 to 2006 and the trend varied from 1to 4mm per year in the Pyrenees-Orientales and Audeadministrative departments respectively and the westernpart of the French Mediterranean area with an averageincrease in PET of between 34mm and 150mm in the last 36years [36]
The increasing trend in ET0that was observed in large
cities is different than this that was found in the farmlandFor example in the Haihe River basin in northern Chinadecreasing trends were observed in 26 stations while 16stations showed significant decreasing trends from 1950 to2007 with rates from minus20 to minus37mmyearminus1 [34] Similarly asignificant decreasing trend of ET
0with a rate of minus3mm per
yearwas found in the arid region of northwest China [37]Thedifference trend in ET
0between the large cities and the
farmland may be due to the variations in the energy balanceand the evaporative potential In the farmland areas morethan 60ndash80 of the net radiation is used for plant evapotran-spiration [33 38 39]This effect not only reduces the availableheat for heating the air environment but also increasesthe water vapor in the near atmosphere The latter effect mayincrease the relative humidity and reduce the vapor pressuredeficit and lastly it may reduce the reference evapotranspi-ration For urban conditions the decrease in green land willdecrease the energy consumption caused by crop evapotran-spiration increasing the available heat and decreasing thewater vapor which ultimately results in an increase in theevaporative potential
It should be noted in Figure 9 that the ET0values in 1997
and 2012 were much lower than those in the neighboringyears It is estimated that ET
0in 1997 and 2012 decreased
by minus104 and minus130 compared to the mean value for theperiod from 1990 to 2012 The sensitive analysis methoddescribed in Section 33 was used to determine the effects ofeach variable on the ET
0changes in 1997 and 2012 compared
to the mean value during the period of 1990ndash2012 The
10 Advances in Meteorology
results showed that the changes in the relative humidity airtemperature sunshine hours andwind speed in 1997 resultedin changes in ET
0by minus72 minus08 minus20 and minus05 respectively
and by minus126 minus08 minus06 and minus42 in 2012 respectively Itcould be concluded that the increase in the relative humidityis the main factor for ET
0reduction followed by the wind
speed air temperature and sunshine hours in the two yearsBased on the mean values of the climatic variables
averaged over the periods of 1954ndash1978 and 1992ndash2012 ET0
increased by 145 in the latter period For the sensitivityanalysis changes in the relative humidity air temperaturewind speed and sunshine hours during 1992ndash2012 caused thevariation of ET
0by 146 50 minus13 and minus38 respectively
compared to those for the period 1954ndash1978The total amountof change in ET
0was 156 based on the sensitivity analysis
This value is similar to the rate of increase of 145 by com-parison of the ET
0values between the two time periods Liu
et al (2014) [15] calculated the ET0change rates by directly
comparing the mean values and summing each ET0change
rate caused by climatic variables using the same sensitiveanalysis method the ET
0change rates were 107 and 105
respectively Liu et al (2009) [40] found that ET0inside the
screenhouse was reduced by 39 compared to that in theopen field By considering the effect of each climatic variablechange to ET
0using the sensitivity analysis the total ET
0
change rate sums to 44 which is similar to the value of39 Therefore it could be concluded that the sensitivity-analysis method used in this study is reliable and easy to useand hence it is recommended for the analysis of the effect ofclimate change on ET
0
5 Conclusions
(1) The development of Shenzhen city greatly affected thelocal climatic conditions Before the onset of urbandevelopment each climatic variable varied slightlywhereas afterward the air temperature increased sig-nificantly and the sunshine hours and relative humid-ity decreased significantly The mutation point formost climatic variables is observed at approximately1978 the onset year for urban development
(2) ET0first decreased from 1954 to 1978 and then
increased quickly and reached a maximal value of1373mm during the period from 1992 to 2012 Themean ET
0value for the period from 1954 to 1978 was
1110mm and increased to 1284mm during the periodfrom 1992 to 2012 indicating an increasing trend ofthe evaporative demand
(3) Sensitivity analysis showed that ET0is most sensitive
to relative humidity followed by air temperaturesunshine hours and wind speed
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This paper is partially supported by the National ScienceFoundation of China (Grant nos 51179005 51479004) Theauthors greatly acknowledge the comments from the editorand the two anonymous reviewers
References
[1] IPCC Climate Change 2014 Synthesis Report An Assessmentof Intergovernmental Panel on Climate Change IPCC GenevaSwitzerland 2014 httpipccchindexhtml
[2] J D Haskett Y A Pachepsky and B Acock ldquoEffect of climateand atmospheric change on soybean water stress a study ofIowardquo Ecological Modelling vol 135 no 2-3 pp 265ndash277 2000
[3] T G Huntington ldquoEvidence for intensification of the globalwater cycle review and synthesisrdquo Journal of Hydrology vol 319no 1ndash4 pp 83ndash95 2006
[4] B C Bates Z W Kundzewicz S Wu and J Palutikof ldquoClimatechange and waterrdquo Technical Paper of the IntergovernmentalPanel on Climate Change IPCC Secretariat Geneva Switzer-land 2008
[5] C M Philandras D A Metaxas and P T Nastos ldquoClimatevariability and urbanization in AthensrdquoTheoretical and AppliedClimatology vol 63 no 1-2 pp 65ndash72 1999
[6] R L Wilby ldquoPast and projected trends in Londonrsquos urban heatislandrdquoWeather vol 58 no 7 pp 251ndash260 2003
[7] N Schwarz U Schlink U Franck and K Groszligmann ldquoRela-tionship of land surface and air temperatures and its implica-tions for quantifying urban heat island indicatorsmdashan applica-tion for the city of Leipzig (Germany)rdquoEcological Indicators vol18 pp 693ndash704 2012
[8] H Wang L Fu X Lin Y Zhou and J C Chen ldquoA bottom-up methodology to estimate vehicle emissions for the Beijingurban areardquo Science of the Total Environment vol 407 no 6 pp1947ndash1953 2009
[9] DESE (Department of Environmental Science and Engineer-ingTsinghua University) Mobile Source Database EmissionInventory and Treatment Proposal for Beijing Tsinghua Univer-sity Beijing China 2005
[10] H Kan S J London G Chen et al ldquoDifferentiating the effectsof fine and coarse particles on daily mortality in ShanghaiChinardquo Environment International vol 33 no 3 pp 376ndash3842007
[11] A Aziz and I U Bajwa ldquoErroneous mass transit system andits tended relationship with motor vehicular air pollution (Anintegrated approach for reduction of urban air pollution inLahore)rdquo Environmental Monitoring and Assessment vol 137no 1ndash3 pp 25ndash33 2008
[12] R G Allen L S Perreira D Raes and M Smith Crop Evap-otranspiration Guidelines for Computing Crop Water Require-ments FAO Irrigation and Drainage Paper no 56 FAO RomeItaly 1998
[13] K H Hamed ldquoTrend detection in hydrologic data the Mann-Kendall trend test under the scaling hypothesisrdquo Journal ofHydrology vol 349 no 3-4 pp 350ndash363 2008
[14] L Q Liang L J Li andQ Liu ldquoTemporal variation of referenceevapotranspiration during 1961ndash2005 in the Taoer River basin ofNortheast Chinardquo Agricultural and Forest Meteorology vol 150no 2 pp 298ndash306 2010
Advances in Meteorology 11
[15] H Liu Y Li T Josef R H Zhang and G H HuangldquoQuantitative estimation of climate change effects on potentialevapotranspiration in Beijing during 1951ndash2010rdquo Journal ofGeographical Sciences vol 24 no 1 pp 93ndash112 2014
[16] M G Kendall and A StuartThe Advanced Theory of StatisticsGriffin London UK 1973
[17] Q-Y Tang and C-X Zhang ldquoData Processing System (DPS)software with experimental design statistical analysis and datamining developed for use in entomological researchrdquo InsectScience vol 20 no 2 pp 254ndash260 2013
[18] M Moller J Tanny Y Li and S Cohen ldquoMeasuring andpredicting evapotranspiration in an insect-proof screenhouserdquoAgricultural and Forest Meteorology vol 127 no 1-2 pp 35ndash512004
[19] G Kitsara G Papaioannou A Papathanasiou and A RetalisldquoDimmingbrightening in athens trends in sunshine durationcloud cover and reference evapotranspirationrdquoWater ResourcesManagement vol 27 no 6 pp 1623ndash1633 2013
[20] Y Wang Y Yang S Han Q X Wang and G H ZhangldquoSunshine dimming and brightening in Chinese cities (1955ndash2011) was driven by air pollution rather than cloudsrdquo ClimateResearch vol 56 no 1 pp 11ndash20 2013
[21] G Stanhill and S Cohen ldquoGlobal dimming a review of theevidence for a widespread and significant reduction in globalradiation with discussion of its probable causes and possibleagricultural consequencesrdquoAgricultural and ForestMeteorologyvol 107 no 4 pp 255ndash278 2001
[22] Q Liu and Z Yang ldquoQuantitative estimation of the impact ofclimate change on actual evapotranspiration in the Yellow RiverBasin Chinardquo Journal of Hydrology vol 395 no 3-4 pp 226ndash234 2010
[23] D I Stern ldquoReversal of the trend in global anthropogenic sulfuremissionsrdquoGlobal Environmental Change vol 16 no 2 pp 207ndash220 2006
[24] I Koren J V Martins L A Remer and H Afargan ldquoSmokeinvigoration versus inhibition of clouds over the AmazonrdquoScience vol 321 no 5891 pp 946ndash949 2008
[25] D Rosenfeld Y J Kaufman and I Koren ldquoSwitching cloudcover and dynamical regimes from open to closed Benard cellsin response to the suppression of precipitation by aerosolsrdquoAtmospheric Chemistry and Physics vol 6 no 9 pp 2503ndash25112006
[26] C Ruckstuhl R Philipona K Behrens et al ldquoAerosol and cloudeffects on solar brightening and the recent rapid warmingrdquoGeophysical Research Letters vol 35 no 12 Article ID L127082008
[27] DG Streets Y Fang CMian et al ldquoAnthropogenic andnaturalcontributions to regional trends in aerosol optical depth 1980ndash2006rdquo Journal of Geophysical Research Atmospheres vol 114 no10 Article ID D00D18 2009
[28] V Ramanathan P J Crutzen J T Kiehl and D RosenfeldldquoAtmospheremdashaerosols climate and the hydrological cyclerdquoScience vol 294 no 5549 pp 2119ndash2124 2001
[29] M Wild ldquoEnlightening global dimming and brighteningrdquoBulletin of the AmericanMeteorological Society vol 93 no 1 pp27ndash37 2012
[30] G D Liu Y Li H J Liu and J Xiao ldquoChanging trend of refer-ence crop evapotranspiration and its dominatedmeteorologicalvariables in Shanxi province in the past 55 yearsrdquo Journal ofIrrigation and Drainage vol 31 no 4 pp 26ndash30 2012
[31] C-S Rim ldquoThe effects of urbanization geographical and topo-graphical conditions on reference evapotranspirationrdquo ClimaticChange vol 97 no 3 pp 483ndash514 2009
[32] WKuang Y Liu YDou et al ldquoWhat are hot andwhat are not inan urban landscape quantifying and explaining the land surfacetemperature pattern in Beijing Chinardquo Landscape Ecology2014
[33] Z Qin Q Yu S Xu et al ldquoWater heat fluxes and water useefficiency measurement and modeling above a farmland in theNorth China Plainrdquo Science in China D Earth Sciences vol 48no 1 pp 207ndash217 2005
[34] B Tang L Tong S Z Kang and L Zhang ldquoImpacts ofclimate variability on reference evapotranspiration over 58 yearsin the Haihe river basin of north Chinardquo Agricultural WaterManagement vol 98 no 10 pp 1660ndash1670 2011
[35] S Cohen A Ianetz and G Stanhill ldquoEvaporative climatechanges at BetDagan Israel 1964ndash1998rdquoAgricultural and ForestMeteorology vol 111 no 2 pp 83ndash91 2002
[36] K Chaouche L Neppel C Dieulin et al ldquoAnalyses of precipita-tion temperature and evapotranspiration in a French Mediter-ranean region in the context of climate changerdquoComptes RendusGeoscience vol 342 no 3 pp 234ndash243 2010
[37] Z Huo X Dai S Feng S Kang and G Huang ldquoEffect of cli-mate change on reference evapotranspiration and aridity indexin arid region of Chinardquo Journal of Hydrology vol 492 pp 24ndash34 2013
[38] G Peng X Cai H Zhang A Li F Hu and M Y LeclercldquoHeat flux apportionment to heterogeneous surfaces using fluxfootprint analysisrdquo Advances in Atmospheric Sciences vol 25no 1 pp 107ndash116 2008
[39] Y Q Zhang Y J Shen C M Liu et al ldquoMeasurement andanalysis of water heat and CO
2flux from a farmland in the
North China plainrdquo Acta Geographica Sinica vol 57 no 3 pp333ndash342 2002 (Chinese)
[40] H-J Liu G-H Huang S Cohen and J Tanny ldquoChange in cropevapotranspiration and associated influencing factors underscreenhouse conditionsrdquo Chinese Journal of Eco-Agriculturevol 17 no 3 pp 484ndash488 2009 (Chinese)
results showed that the changes in the relative humidity airtemperature sunshine hours andwind speed in 1997 resultedin changes in ET
0by minus72 minus08 minus20 and minus05 respectively
and by minus126 minus08 minus06 and minus42 in 2012 respectively Itcould be concluded that the increase in the relative humidityis the main factor for ET
0reduction followed by the wind
speed air temperature and sunshine hours in the two yearsBased on the mean values of the climatic variables
averaged over the periods of 1954ndash1978 and 1992ndash2012 ET0
increased by 145 in the latter period For the sensitivityanalysis changes in the relative humidity air temperaturewind speed and sunshine hours during 1992ndash2012 caused thevariation of ET
0by 146 50 minus13 and minus38 respectively
compared to those for the period 1954ndash1978The total amountof change in ET
0was 156 based on the sensitivity analysis
This value is similar to the rate of increase of 145 by com-parison of the ET
0values between the two time periods Liu
et al (2014) [15] calculated the ET0change rates by directly
comparing the mean values and summing each ET0change
rate caused by climatic variables using the same sensitiveanalysis method the ET
0change rates were 107 and 105
respectively Liu et al (2009) [40] found that ET0inside the
screenhouse was reduced by 39 compared to that in theopen field By considering the effect of each climatic variablechange to ET
0using the sensitivity analysis the total ET
0
change rate sums to 44 which is similar to the value of39 Therefore it could be concluded that the sensitivity-analysis method used in this study is reliable and easy to useand hence it is recommended for the analysis of the effect ofclimate change on ET
0
5 Conclusions
(1) The development of Shenzhen city greatly affected thelocal climatic conditions Before the onset of urbandevelopment each climatic variable varied slightlywhereas afterward the air temperature increased sig-nificantly and the sunshine hours and relative humid-ity decreased significantly The mutation point formost climatic variables is observed at approximately1978 the onset year for urban development
(2) ET0first decreased from 1954 to 1978 and then
increased quickly and reached a maximal value of1373mm during the period from 1992 to 2012 Themean ET
0value for the period from 1954 to 1978 was
1110mm and increased to 1284mm during the periodfrom 1992 to 2012 indicating an increasing trend ofthe evaporative demand
(3) Sensitivity analysis showed that ET0is most sensitive
to relative humidity followed by air temperaturesunshine hours and wind speed
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This paper is partially supported by the National ScienceFoundation of China (Grant nos 51179005 51479004) Theauthors greatly acknowledge the comments from the editorand the two anonymous reviewers
References
[1] IPCC Climate Change 2014 Synthesis Report An Assessmentof Intergovernmental Panel on Climate Change IPCC GenevaSwitzerland 2014 httpipccchindexhtml
[2] J D Haskett Y A Pachepsky and B Acock ldquoEffect of climateand atmospheric change on soybean water stress a study ofIowardquo Ecological Modelling vol 135 no 2-3 pp 265ndash277 2000
[3] T G Huntington ldquoEvidence for intensification of the globalwater cycle review and synthesisrdquo Journal of Hydrology vol 319no 1ndash4 pp 83ndash95 2006
[4] B C Bates Z W Kundzewicz S Wu and J Palutikof ldquoClimatechange and waterrdquo Technical Paper of the IntergovernmentalPanel on Climate Change IPCC Secretariat Geneva Switzer-land 2008
[5] C M Philandras D A Metaxas and P T Nastos ldquoClimatevariability and urbanization in AthensrdquoTheoretical and AppliedClimatology vol 63 no 1-2 pp 65ndash72 1999
[6] R L Wilby ldquoPast and projected trends in Londonrsquos urban heatislandrdquoWeather vol 58 no 7 pp 251ndash260 2003
[7] N Schwarz U Schlink U Franck and K Groszligmann ldquoRela-tionship of land surface and air temperatures and its implica-tions for quantifying urban heat island indicatorsmdashan applica-tion for the city of Leipzig (Germany)rdquoEcological Indicators vol18 pp 693ndash704 2012
[8] H Wang L Fu X Lin Y Zhou and J C Chen ldquoA bottom-up methodology to estimate vehicle emissions for the Beijingurban areardquo Science of the Total Environment vol 407 no 6 pp1947ndash1953 2009
[9] DESE (Department of Environmental Science and Engineer-ingTsinghua University) Mobile Source Database EmissionInventory and Treatment Proposal for Beijing Tsinghua Univer-sity Beijing China 2005
[10] H Kan S J London G Chen et al ldquoDifferentiating the effectsof fine and coarse particles on daily mortality in ShanghaiChinardquo Environment International vol 33 no 3 pp 376ndash3842007
[11] A Aziz and I U Bajwa ldquoErroneous mass transit system andits tended relationship with motor vehicular air pollution (Anintegrated approach for reduction of urban air pollution inLahore)rdquo Environmental Monitoring and Assessment vol 137no 1ndash3 pp 25ndash33 2008
[12] R G Allen L S Perreira D Raes and M Smith Crop Evap-otranspiration Guidelines for Computing Crop Water Require-ments FAO Irrigation and Drainage Paper no 56 FAO RomeItaly 1998
[13] K H Hamed ldquoTrend detection in hydrologic data the Mann-Kendall trend test under the scaling hypothesisrdquo Journal ofHydrology vol 349 no 3-4 pp 350ndash363 2008
[14] L Q Liang L J Li andQ Liu ldquoTemporal variation of referenceevapotranspiration during 1961ndash2005 in the Taoer River basin ofNortheast Chinardquo Agricultural and Forest Meteorology vol 150no 2 pp 298ndash306 2010
Advances in Meteorology 11
[15] H Liu Y Li T Josef R H Zhang and G H HuangldquoQuantitative estimation of climate change effects on potentialevapotranspiration in Beijing during 1951ndash2010rdquo Journal ofGeographical Sciences vol 24 no 1 pp 93ndash112 2014
[16] M G Kendall and A StuartThe Advanced Theory of StatisticsGriffin London UK 1973
[17] Q-Y Tang and C-X Zhang ldquoData Processing System (DPS)software with experimental design statistical analysis and datamining developed for use in entomological researchrdquo InsectScience vol 20 no 2 pp 254ndash260 2013
[18] M Moller J Tanny Y Li and S Cohen ldquoMeasuring andpredicting evapotranspiration in an insect-proof screenhouserdquoAgricultural and Forest Meteorology vol 127 no 1-2 pp 35ndash512004
[19] G Kitsara G Papaioannou A Papathanasiou and A RetalisldquoDimmingbrightening in athens trends in sunshine durationcloud cover and reference evapotranspirationrdquoWater ResourcesManagement vol 27 no 6 pp 1623ndash1633 2013
[20] Y Wang Y Yang S Han Q X Wang and G H ZhangldquoSunshine dimming and brightening in Chinese cities (1955ndash2011) was driven by air pollution rather than cloudsrdquo ClimateResearch vol 56 no 1 pp 11ndash20 2013
[21] G Stanhill and S Cohen ldquoGlobal dimming a review of theevidence for a widespread and significant reduction in globalradiation with discussion of its probable causes and possibleagricultural consequencesrdquoAgricultural and ForestMeteorologyvol 107 no 4 pp 255ndash278 2001
[22] Q Liu and Z Yang ldquoQuantitative estimation of the impact ofclimate change on actual evapotranspiration in the Yellow RiverBasin Chinardquo Journal of Hydrology vol 395 no 3-4 pp 226ndash234 2010
[23] D I Stern ldquoReversal of the trend in global anthropogenic sulfuremissionsrdquoGlobal Environmental Change vol 16 no 2 pp 207ndash220 2006
[24] I Koren J V Martins L A Remer and H Afargan ldquoSmokeinvigoration versus inhibition of clouds over the AmazonrdquoScience vol 321 no 5891 pp 946ndash949 2008
[25] D Rosenfeld Y J Kaufman and I Koren ldquoSwitching cloudcover and dynamical regimes from open to closed Benard cellsin response to the suppression of precipitation by aerosolsrdquoAtmospheric Chemistry and Physics vol 6 no 9 pp 2503ndash25112006
[26] C Ruckstuhl R Philipona K Behrens et al ldquoAerosol and cloudeffects on solar brightening and the recent rapid warmingrdquoGeophysical Research Letters vol 35 no 12 Article ID L127082008
[27] DG Streets Y Fang CMian et al ldquoAnthropogenic andnaturalcontributions to regional trends in aerosol optical depth 1980ndash2006rdquo Journal of Geophysical Research Atmospheres vol 114 no10 Article ID D00D18 2009
[28] V Ramanathan P J Crutzen J T Kiehl and D RosenfeldldquoAtmospheremdashaerosols climate and the hydrological cyclerdquoScience vol 294 no 5549 pp 2119ndash2124 2001
[29] M Wild ldquoEnlightening global dimming and brighteningrdquoBulletin of the AmericanMeteorological Society vol 93 no 1 pp27ndash37 2012
[30] G D Liu Y Li H J Liu and J Xiao ldquoChanging trend of refer-ence crop evapotranspiration and its dominatedmeteorologicalvariables in Shanxi province in the past 55 yearsrdquo Journal ofIrrigation and Drainage vol 31 no 4 pp 26ndash30 2012
[31] C-S Rim ldquoThe effects of urbanization geographical and topo-graphical conditions on reference evapotranspirationrdquo ClimaticChange vol 97 no 3 pp 483ndash514 2009
[32] WKuang Y Liu YDou et al ldquoWhat are hot andwhat are not inan urban landscape quantifying and explaining the land surfacetemperature pattern in Beijing Chinardquo Landscape Ecology2014
[33] Z Qin Q Yu S Xu et al ldquoWater heat fluxes and water useefficiency measurement and modeling above a farmland in theNorth China Plainrdquo Science in China D Earth Sciences vol 48no 1 pp 207ndash217 2005
[34] B Tang L Tong S Z Kang and L Zhang ldquoImpacts ofclimate variability on reference evapotranspiration over 58 yearsin the Haihe river basin of north Chinardquo Agricultural WaterManagement vol 98 no 10 pp 1660ndash1670 2011
[35] S Cohen A Ianetz and G Stanhill ldquoEvaporative climatechanges at BetDagan Israel 1964ndash1998rdquoAgricultural and ForestMeteorology vol 111 no 2 pp 83ndash91 2002
[36] K Chaouche L Neppel C Dieulin et al ldquoAnalyses of precipita-tion temperature and evapotranspiration in a French Mediter-ranean region in the context of climate changerdquoComptes RendusGeoscience vol 342 no 3 pp 234ndash243 2010
[37] Z Huo X Dai S Feng S Kang and G Huang ldquoEffect of cli-mate change on reference evapotranspiration and aridity indexin arid region of Chinardquo Journal of Hydrology vol 492 pp 24ndash34 2013
[38] G Peng X Cai H Zhang A Li F Hu and M Y LeclercldquoHeat flux apportionment to heterogeneous surfaces using fluxfootprint analysisrdquo Advances in Atmospheric Sciences vol 25no 1 pp 107ndash116 2008
[39] Y Q Zhang Y J Shen C M Liu et al ldquoMeasurement andanalysis of water heat and CO
2flux from a farmland in the
North China plainrdquo Acta Geographica Sinica vol 57 no 3 pp333ndash342 2002 (Chinese)
[40] H-J Liu G-H Huang S Cohen and J Tanny ldquoChange in cropevapotranspiration and associated influencing factors underscreenhouse conditionsrdquo Chinese Journal of Eco-Agriculturevol 17 no 3 pp 484ndash488 2009 (Chinese)
[15] H Liu Y Li T Josef R H Zhang and G H HuangldquoQuantitative estimation of climate change effects on potentialevapotranspiration in Beijing during 1951ndash2010rdquo Journal ofGeographical Sciences vol 24 no 1 pp 93ndash112 2014
[16] M G Kendall and A StuartThe Advanced Theory of StatisticsGriffin London UK 1973
[17] Q-Y Tang and C-X Zhang ldquoData Processing System (DPS)software with experimental design statistical analysis and datamining developed for use in entomological researchrdquo InsectScience vol 20 no 2 pp 254ndash260 2013
[18] M Moller J Tanny Y Li and S Cohen ldquoMeasuring andpredicting evapotranspiration in an insect-proof screenhouserdquoAgricultural and Forest Meteorology vol 127 no 1-2 pp 35ndash512004
[19] G Kitsara G Papaioannou A Papathanasiou and A RetalisldquoDimmingbrightening in athens trends in sunshine durationcloud cover and reference evapotranspirationrdquoWater ResourcesManagement vol 27 no 6 pp 1623ndash1633 2013
[20] Y Wang Y Yang S Han Q X Wang and G H ZhangldquoSunshine dimming and brightening in Chinese cities (1955ndash2011) was driven by air pollution rather than cloudsrdquo ClimateResearch vol 56 no 1 pp 11ndash20 2013
[21] G Stanhill and S Cohen ldquoGlobal dimming a review of theevidence for a widespread and significant reduction in globalradiation with discussion of its probable causes and possibleagricultural consequencesrdquoAgricultural and ForestMeteorologyvol 107 no 4 pp 255ndash278 2001
[22] Q Liu and Z Yang ldquoQuantitative estimation of the impact ofclimate change on actual evapotranspiration in the Yellow RiverBasin Chinardquo Journal of Hydrology vol 395 no 3-4 pp 226ndash234 2010
[23] D I Stern ldquoReversal of the trend in global anthropogenic sulfuremissionsrdquoGlobal Environmental Change vol 16 no 2 pp 207ndash220 2006
[24] I Koren J V Martins L A Remer and H Afargan ldquoSmokeinvigoration versus inhibition of clouds over the AmazonrdquoScience vol 321 no 5891 pp 946ndash949 2008
[25] D Rosenfeld Y J Kaufman and I Koren ldquoSwitching cloudcover and dynamical regimes from open to closed Benard cellsin response to the suppression of precipitation by aerosolsrdquoAtmospheric Chemistry and Physics vol 6 no 9 pp 2503ndash25112006
[26] C Ruckstuhl R Philipona K Behrens et al ldquoAerosol and cloudeffects on solar brightening and the recent rapid warmingrdquoGeophysical Research Letters vol 35 no 12 Article ID L127082008
[27] DG Streets Y Fang CMian et al ldquoAnthropogenic andnaturalcontributions to regional trends in aerosol optical depth 1980ndash2006rdquo Journal of Geophysical Research Atmospheres vol 114 no10 Article ID D00D18 2009
[28] V Ramanathan P J Crutzen J T Kiehl and D RosenfeldldquoAtmospheremdashaerosols climate and the hydrological cyclerdquoScience vol 294 no 5549 pp 2119ndash2124 2001
[29] M Wild ldquoEnlightening global dimming and brighteningrdquoBulletin of the AmericanMeteorological Society vol 93 no 1 pp27ndash37 2012
[30] G D Liu Y Li H J Liu and J Xiao ldquoChanging trend of refer-ence crop evapotranspiration and its dominatedmeteorologicalvariables in Shanxi province in the past 55 yearsrdquo Journal ofIrrigation and Drainage vol 31 no 4 pp 26ndash30 2012
[31] C-S Rim ldquoThe effects of urbanization geographical and topo-graphical conditions on reference evapotranspirationrdquo ClimaticChange vol 97 no 3 pp 483ndash514 2009
[32] WKuang Y Liu YDou et al ldquoWhat are hot andwhat are not inan urban landscape quantifying and explaining the land surfacetemperature pattern in Beijing Chinardquo Landscape Ecology2014
[33] Z Qin Q Yu S Xu et al ldquoWater heat fluxes and water useefficiency measurement and modeling above a farmland in theNorth China Plainrdquo Science in China D Earth Sciences vol 48no 1 pp 207ndash217 2005
[34] B Tang L Tong S Z Kang and L Zhang ldquoImpacts ofclimate variability on reference evapotranspiration over 58 yearsin the Haihe river basin of north Chinardquo Agricultural WaterManagement vol 98 no 10 pp 1660ndash1670 2011
[35] S Cohen A Ianetz and G Stanhill ldquoEvaporative climatechanges at BetDagan Israel 1964ndash1998rdquoAgricultural and ForestMeteorology vol 111 no 2 pp 83ndash91 2002
[36] K Chaouche L Neppel C Dieulin et al ldquoAnalyses of precipita-tion temperature and evapotranspiration in a French Mediter-ranean region in the context of climate changerdquoComptes RendusGeoscience vol 342 no 3 pp 234ndash243 2010
[37] Z Huo X Dai S Feng S Kang and G Huang ldquoEffect of cli-mate change on reference evapotranspiration and aridity indexin arid region of Chinardquo Journal of Hydrology vol 492 pp 24ndash34 2013
[38] G Peng X Cai H Zhang A Li F Hu and M Y LeclercldquoHeat flux apportionment to heterogeneous surfaces using fluxfootprint analysisrdquo Advances in Atmospheric Sciences vol 25no 1 pp 107ndash116 2008
[39] Y Q Zhang Y J Shen C M Liu et al ldquoMeasurement andanalysis of water heat and CO
2flux from a farmland in the
North China plainrdquo Acta Geographica Sinica vol 57 no 3 pp333ndash342 2002 (Chinese)
[40] H-J Liu G-H Huang S Cohen and J Tanny ldquoChange in cropevapotranspiration and associated influencing factors underscreenhouse conditionsrdquo Chinese Journal of Eco-Agriculturevol 17 no 3 pp 484ndash488 2009 (Chinese)
