Spatial and temporal variation patterns of reference

evapotranspiration across the Qinghai-Tibetan

Plateau during 1971–2004

Xueqin Zhang,1 Yu Ren,1,2 Zhi-Yong Yin,3 Zhenyao Lin,1 and Du Zheng1

Received 13 January 2009; revised 6 May 2009; accepted 4 June 2009; published 8 August 2009.

[1] Reference evapotranspiration (RET), an indicator of atmospheric evaporatingcapability over a hypothetical reference surface, was calculated using the Penman-Monteith method for 75 stations across the Qinghai-Tibetan Plateau between 1971 and2004. Generally, both annual and seasonal RET decreased for most part of the plateauduring the study period. Multivariate linear models were used to determine thecontributions of climate factors to RET change, including air temperature, air humidity,solar radiation, and wind speed. Spatial differences in the causes of RET change weredetected by K-means clustering analysis. It indicates that wind speed predominated thechanges of RET almost throughout the year, especially in the north of the study region,whereas radiation was the leading factor in the southeast, especially during thesummertime. Although the recent warming trend over the plateau would have increasedRET, the combined effect of the reduced wind speed and shortened sunshine durationnegated the effect of rising temperature and caused RET to decrease in general. Thesignificant decrease in surface wind speed corresponded to the decreasing trends ofupper-air zonal wind and the decline of pressure gradient, possibly as a result of therecent warming.

Citation: Zhang, X., Y. Ren, Z.-Y. Yin, Z. Lin, and D. Zheng (2009), Spatial and temporal variation patterns of reference

evapotranspiration across the Qinghai-Tibetan Plateau during 1971–2004, J. Geophys. Res., 114, D15105,

doi:10.1029/2009JD011753.

1. Introduction

[2] Warming-induced hydroclimatic changes, if fully re-alized, would likely vary greatly across any one continent[Ziegler et al., 2005], and climatic changes affect allelements of hydrologic cycle, such as precipitation, runoff,infiltration, groundwater flow, and evaportranspiration in avariety of ways [Kundzewicz and Somlyody, 1997]. Recentstudies on regional-scale hydroclimatic changes pay muchattention to the spatial and temporal patterns of evapotrans-piration, involved in the maintenance of forest and oasisecosystems, management of farmland and pasture irrigation,sustainable water supply for industrial and domesticdemands et al. [e.g., Yu et al., 2002; Jin et al., 2004;Francisco, 2005; Xu et al., 2006; Gao et al., 2006; Brunelet al., 2006; Calanca et al., 2006; Burns et al., 2007].[3] The Qinghai-Tibetan Plateau, famous as ‘‘the roof of

the world,’’ is rich in lakes, glaciers, and wetlands and is the

main source area of several major rivers in Asia [Shen andChen, 1996; Luosang, 2005]. The management and utiliza-tion of affluent water resources and hydropower on theplateau are not only important issues concerned with China,but also the neighboring countries located in the lowerreaches of the international rivers originated from theplateau [Liu and Qimeiduoji, 1999; Liu et al., 2006a].Significant warming trends were detected on the plateauduring the last several decades [e.g., Liu et al., 2006b]. Thepossible shift of water balance caused by climate changemight have impacted the water resources on the plateau.[4] As a result of monsoon climate, most precipitation

and evapotranspiration on the plateau occurs from June toSeptember. Evapotranspiration not only is a basic compo-nent of water balance, but also plays an important role in theenergy budget in the earth-atmospheric system. In thecentral and southeastern plateau, the latent heat flux dom-inates the surface energy balance and is a major energysource of the atmosphere during the summer monsoon [Li,2002; Xu et al., 2005]. Therefore the spatial and temporalpatterns of evapotranspiration may have a significant impacton the transport of water vapor and latent heat and furtherlocal hydroclimatic courses.[5] Actual evapotranspiration (AET) is influenced by

various factors of both local climate and land surfacecondition. Potential evapotranspiration (PET) was firstdefined as ‘‘the amount of water transpired in a given time

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 114, D15105, doi:10.1029/2009JD011753, 2009

1Institute of Geographical Science and Natural Resources Research,Chinese Academy of Science, Beijing, China.

2Graduate University of Chinese Academy of Science, Beijing, China.3Marine Science and Environmental Studies, University of San Diego,

San Diego, California, USA.

Copyright 2009 by the American Geophysical Union.0148-0227/09/2009JD011753

D15105 1 of 14

by a short green crop, completely shading the ground, ofuniform height and with adequate water status in the soilprofile’’ [Penman, 1948, 1956]. Similarly, the term ofmaximum possible evaporation or evaporation capacityusually refers to the amount of water evaporated fromsurfaces not limited by the supply of water [Gao et al.,1978]. FAO (Food and Agriculture Organization of theUnited Nations) accepted the definition of reference surfaceand recommended the concept of reference evapotranspira-tion (RET) instead of other ambiguous expressions such asPET [Allen et al., 1998]. The FAO Penman-Monteithmethod is recommended as the single standard method fordetermining RET, which represents the evaporating powerof the atmosphere at a specific location and time and doesnot consider the crop characteristics and soil factors [Allenet al., 1998].[6] As an observational and physical representation of

PET, pan evaporation has declined in many parts of theworld during the last several decades [e.g., Peterson et al.,1995; Chattopadhyay and Hulme, 1997; Golubev et al.,2001; Liu et al., 2004; Roderick and Farquhar, 2004;Tebakari et al., 2005]. Recent studies suggested that thedecline of pan evaporation, PET, or RET over the Qinghai-Tibetan Plateau was mainly attributable to the decrease ofsurface wind speed, regardless of the rising temperatures[Chen et al., 2006; Zhang et al., 2007]. Similar attributionshave been reported for some other parts of China [Chen etal., 2005; Xu et al., 2006], the northeast of India [Jhajhariaet al., 2009], the south of Canada [Burn and Hesch, 2007],and the most parts of Australia [Roderick et al., 2007]. Chenet al. [2006] indicated that changes in relative humidityplayed the supporting role affecting PET trends of theplateau during 1961–2000; while changes in sunshineduration played an insignificant role. Still, Zhang et al.[2007] suggested that the decrease of net total radiation wasthe secondary cause of the decreasing trend in RET of theplateau during 1966–2003. Since a climate factor may havea different impact on the RET process, its relative contri-bution to RET variation may also be different spatially. Therelative significance of the climate factors of RET variationneeds to be clearly quantified. Additionally, Chen et al.[2006] reckoned that the forcing factors of RET trendsacross the plateau were generally the same in all months,i.e., the decreasing wind speeds. However, there might bestill some seasonal variation of the relative significance ofthe climate factors. In our study, we focused on the spatialand seasonal differences in the contributing factors to theRET trends over the plateau, which has not been specifi-cally revealed in previous studies. Furthermore, we inves-tigated the causes why the surface wind speed haddecreased over the plateau, since it was identified as themain cause of the RET reduction.

2. Data and Methods

2.1. Meteorological Data

[7] Meteorological data are required for the FAO Penman-Monteith method. The discontinuities in the meteorologicaldata over the plateau are mostly caused by weather stationrelocations and missing observations. In our study, obser-vations with poor integrality before 1970s were discarded

and weather stations that moved more than 30’ in longitudeor latitude or 100 m in altitude were filtered out to reduceany potential impact of observation discontinuities on theanalyses. We abandoned five stations located in QinghaiProvince, i.e., Mangya (no. 51866), Menyuan (no. 52765),Xinghai (no. 52943), Henan (no. 56065), and Banma (no.56151), which were used in the studies by Chen et al.[2006] and Zhang et al. [2007].[8] The data from 75 stations in China’s National Meteo-

rological Observatory (NMO) network were used for theRET calculation (Figure 1, Table 1) in our study. Dailymaximum, minimum and mean air temperatures, meanrelative humidity, wind speed and sunshine duration wereprovided by the National Meteorological Information Cen-tre (NMIC) of China Meteorological Administration(CMA). The length of records for 74 stations is 1971–2004, but 1971–2000 for Chaka (no. 52842). As an averagefor the 74 stations, about 1.1% of the days were consideredmissing, if any of the weather observations was missing.Table 2 presents the number of months with missing data,i.e., month with all days missing, for all the 75 stations usedfor RET calculation during the period of 1971–2004.Besides, radiation observations from 16 stations (Figure 1,Table 1) were used to calibrate the Angstrom coefficients forestimating the solar radiation in the FAO Penman-Monteithmethod. Differing from the study by Zhang et al. [2007]which used 11 stations to calibrate the Angstrom coeffi-cients, 5 more stations located in Xinjiang Uygur Autono-mous Region and Gansu Province were used in our study toimprove the estimates of solar radiation over the north of theplateau.[9] In order to examine the changes in circulation patterns

and intensity as possible causes of climatic factors overthe plateau, we obtained the monthly 500 hPa zonal windspeed and geopotential height data from the NCEP/NCARReanalysis data set [Kalnay et al., 1996] provided by theNOAA/OAR/ESRL PSD (Boulder, Colorado, USA, http://www.cdc.noaa.gov/).

2.2. FAO Penman-Monteith Method

[10] The reference surface is defined as ‘‘a hypotheticalreference crop with an assumed crop height of 0.12 m, afixed surface resistance of 70 s m�1 and an albedo of 0.23,’’and the FAO Penman-Monteith method [Allen et al., 1998]to estimate RET (mm/d) is given as:

RET ¼0:408DðRn � GÞ þ g

900

T þ 273u2ðes � eaÞ

Dþ gð1þ 0:34u2Þ

where:

Rn net radiation at the crop surface [MJ m�2d�1]G soil heat flux density [MJ m�2 d�1]T mean daily air temperature at 2 m [�C]u2 wind speed at 2 m [m s�1]es saturation vapor pressure [kPa]ea actual vapor pressure [kPa]D slope of the saturation vapor pressure curve at air

temperature T [kPa �C�1]g psychrometric constant [kPa �C�1].

D15105 ZHANG ET AL.: TP REFERENCE EVAPOTRANSPIRATION PATTERNS

2 of 14

D15105

[11] To ensure the integrity of computation, the weathermeasurements should be made at 2 m (or converted to thatheight) above an extensive surface of green grass, shadingthe ground and not short of water [Allen et al., 1998]. InChina’s NMO network, air temperature and humidity aremeasured at 1.5 m and wind speed at 10–12 m aboveground. The measured daily minimum temperature, dailymaximum temperature, and daily mean specific humidity at1.5 m are respectively about 99%, 101%, and 101% of themeasurements at 2 m, estimated according to the fieldobservations during the Second Qinghai-Tibetan PlateauExperiment (TIPEX) from May to July 1998 [Liu et al.,2002]. In our study, the normal measurements of airtemperature and humidity at 1.5 m were used instead ofthe required measurements at 2 m, since the differences arerather small and could be ignored. Wind speed at 2 mheight was converted from the normal measurement at10–12 m based on the logarithmic wind speed profileequation given by the FAO Penman-Monteith method[Allen et al., 1998]:

u2 ¼ uz4:87

lnð67:8z� 5:42Þ

where:

z height of measurement above ground surface [m]uz measured wind speed at z m above ground surface

[m s�1].

2.2.1. Calibration of the Angstrom Coefficients[12] In the FAO Penman-Monteith method, the solar

radiation, if not measured, can be calculated using theAngstrom formula [Allen et al., 1998]:

Rs ¼ as þ bsn

N

� �Ra

where:

Rs solar or shortwave radiation [MJ m�2d�1]n actual duration of sunshine [hour]N maximum possible duration of sunshine or day-

light hours [hour]n/N relative sunshine duration [-]Ra extraterrestrial radiation [MJ m�2 d�1]As regression constant, representing the fraction of

extraterrestrial radiation reaching the earth onovercast days (n = 0)

as + bs the fraction of extraterrestrial radiation reachingthe earth on clear days (n = N).

[13] In our study, the observations of n and Rs from 16stations (Figure 1, Table 1) were used to calibrate theAngstrom coefficients as and bs. Ra and N were computedbased on date and latitude according to the equations givenby the FAO Penman-Monteith method [Allen et al., 1998].To ensure the continuity of Rs and n, the years with over 10missing days in any month were excluded (Table 3). Rs/Ra

and n/N were then computed and as and bs were derived

Figure 1. Locations of meteorological stations on and around the Tibetan Plateau. Black dots indicatestations used for calculating reference evapotranspiration, and open circles stand for stations withradiation observations. All the stations are labeled by their WMO numbers.

D15105 ZHANG ET AL.: TP REFERENCE EVAPOTRANSPIRATION PATTERNS

3 of 14

D15105

form the least squares fit of Rs/Ra on n/N by solving thelinear model.[14] The Angstrom coefficients of the 16 stations (Table 3)

are rather close to the values reported by Chen et al. [2004].Generally, the magnitudes of as + bs, i.e., the fraction ofextraterrestrial radiation reaching the earth on clear days, areabove 0.8 in higher altitude areas, but below 0.8 in loweraltitude areas. Overall, the Angstrom model is suitable fordaily global radiation estimation on the plateau as indicatedby high R2 values (Table 3). For stations with no observa-tion of solar radiation but sunshine duration, as and bs wereestimated by Kriging interpolation using Surfer 8.0 (GoldenSoftware, Golden, Colorado, USA).2.2.2. Calculation of the Soil Heat Flux[15] The soil heat flux (G) is the energy utilized in soil

heat exchange. Although G is small compared to Rn andmay often be ignored, the amount of energy gained or lostby the soil should be theoretically subtracted or added to Rn

when estimating evapotranspiration [Allen et al., 1998]. Inthe FAO Penman-Monteith Method, the monthly value of Gis given as:

Gmonth;i ¼ 0:07ðTmonth;iþ1 � Tmonth;i�1Þ

or, if Tmonth, i+1 is unknown:

Gmonth;i ¼ 0:14 Tmonth;i � Tmonth;i�1

� �

where:

Gmonth,i soil heat flux of month i [MJ m�2 day�1]Tmonth,i air temperature of month i [�C]

Tmonth, i�1 air temperature of the previous month [�C]Tmonth, i+1 air temperature of the next month [�C].

2.3. Statistical Methods

2.3.1. Long-term Means and Intraannual VariationPatterns of RET and Related Variables[16] The monthly means of air temperature (T, �C),

relative sunshine duration (S, %), relative humidity (RH,%), 2 m wind speed (U, m s�1) and sums of RET (mm)were calculated for spring (MAM), summer (JJA), fall(SON), and winter (DJF) for all 75 stations from 1971 to2004. The missing data for all 75 stations but Chaka(no. 52842) were filled using the means of the values fromthe other years.2.3.2. Regression Analysis[17] Linear regression was applied to examine the trends

of RET during the period of 1971–2004. Spatial distribu-tion pattern of RET was examined using stepwise regres-sions, where mean annual RET was regressed against

Table 1. Information for the National Meteorological Observatory

Stations Used in This Study

WMONumber Name

Latitude(�N)

Longitude(�E)

Altitude(m a.s.l.)

51709a Kashi 39.47 75.98 1291.351777a Ruoqiang 39.03 88.17 89051828a Hetian 37.13 79.93 1374.751855 Qiemo 38.15 85.55 1248.451931 Yutian 36.85 81.65 1423.352418a Dunhuang 40.15 94.68 113952533a Jiuquan 39.77 98.48 1477.252602 Lenghu 38.75 93.33 277052633 Tuole 38.8 98.42 336752645 Yeniugou 38.42 99.58 33202652 Zhangye 38.93 100.43 1482.752657 Qilian 38.18 100.25 2787.452679 Wuwei 37.92 102.67 1530.952707 Xiaozaohuo 36.8 93.68 276752713 Dachaidan 37.85 95.37 3173.252737 Delingha 37.37 97.37 2981.552754a Gangcha 37.33 100.13 3301.552787 Wuqiaoling 37.2 102.87 3045.152818a Geermu 36.42 94.9 2807.652825 Nuomuhong 36.43 96.42 2790.452836 Dulan 36.3 98.1 3191.152842 Chaka 36.78 99.08 3087.652856 Qiapuqia 36.27 100.62 283552866a Xining 36.72 101.75 2295.252868 Guide 36.03 101.43 2237.152876 Minhe 36.32 102.85 1813.952908 Wudaoliang 35.22 93.08 4612.252996 Huajialing 35.38 105 2450.655228a Shiquanhe 32.5 80.08 427855279 Bange 31.38 90.02 470055299a Naqu 31.48 92.07 450755472 Shenzha 30.95 88.63 467255578 Rikaze 29.25 88.88 383655591a Lhasa 29.67 91.13 3648.755598 Zedang 29.25 91.77 3551.755664 Dingri 28.63 87.08 430055680 Jiangzi 28.92 89.6 404055696 Longzi 28.42 92.47 386055773 Pali 27.73 89.08 430056004 Tuotuohe 34.22 92.43 4533.156018 Zaduo 32.9 95.3 4066.456021 Qumalai 34.13 95.78 417556029a Yushu 33.02 97.02 3681.256033 Maduo 34.92 98.22 4272.356034 Qingshuihe 33.8 97.13 4415.456038 Shiqu 32.98 98.1 420056046 Dari 33.75 99.65 3967.556067 Jiuzhi 33.43 101.48 3628.556079 Ruoergai 33.58 102.97 3439.656080 Hezuo 35 102.9 291056093 Minxian 34.43 104.02 231556106 Suoxian 31.88 93.78 4022.856116 Dingqing 31.42 95.6 3873.156125 Nangqian 32.2 96.48 3643.756137a Changdu 31.15 97.17 330656144 Dege 31.8 98.58 3201.256146a Ganzi 31.62 100 3393.556152 Seda 32.28 100.33 3893.956167 Daofu 30.98 101.12 2957.256172 Maerkang 31.9 102.23 2664.456173a Hongyuan 32.8 102.55 3491.656178 Xiaojin 31 102.35 2369.256178 Xiaojin 31 102.35 2369.256182 Songpan 32.65 103.57 2850.756202 Jiali 30.67 93.28 4488.856227 Bomi 29.87 95.77 275056247 Batang 30 99.1 2589.256257 Litang 30 100.27 3948.956287 Yaan 29.98 103 627.656312 Linzhi 29.67 94.33 2991.856357 Daocheng 29.05 100.3 3727.7

Table 1. (continued)

WMONumber Name

Latitude(�N)

Longitude(�E)

Altitude(m a.s.l.)

56374 Kangding 30.05 101.97 2615.756444 Deqin 28.48 98.92 331956462 Jiulong 29 101.5 2987.356543 Zhongdian 27.83 99.7 3276.156548 Weixi 27.17 99.28 2325.656043a Guoluo 34.47 100.25 3719

aStations with radiation observation.

D15105 ZHANG ET AL.: TP REFERENCE EVAPOTRANSPIRATION PATTERNS

4 of 14

D15105

latitude f, longitude l, and altitude z. To determine therelative contributions of the climatic variables, we alsoregressed annual and seasonal RET against T, S, U, andRH for each station. The significance level for a predictor tobe added into the model was set as 0.05, and the signifi-cance level for a predictor to be removed from the modelwas 0.1. The regression analysis was performed usingMatlab 7.0 (MathWorks, Natick, MA, USA).2.3.3. K-Means Clustering Analysis[18] K-means clustering analysis [Seber, 1984; Spath,

1985] was used to identify the difference in the annualRET series of 74 stations from 1971 to 2004. K-meansclustering can best be described as a partitioning methodthat divides the series into K mutually exclusive clusters.K-means treats each series as an object having a location inspace. It finds a partition in which objects within each clusterare as close to each other as possible, and as far from objectsin other clusters as possible. In our study, the ‘‘correlationdistance’’ which equals to one minus the sample correlationbetween series was selected from the five different distancemeasures provided by Matlab 7.0. Each cluster in thepartition is defined by its member objects and by itscentroid, or center, to which the sum of distances from allobjects in that cluster is minimized. K-means uses aniterative algorithm that minimizes the sum of distances fromeach object to its cluster centroid over all clusters. To avoidlocal minima as much as possible, we partition the 74 seriesinto K clusters (varying from 2 to 74) with ten replicatesperformed for each partitioning solution. Each of the tenreplicates begins from a different randomly selected set ofinitial centroids. For each clustering solution, there is acorresponding series of silhouette index, which ranges from+1.0, indicating points that are very distant from neighbor-ing clusters, through 0, indicating points that are notdistinctly in one cluster or another, to �1.0, indicatingpoints that are probably assigned to the wrong cluster.Practically, the average silhouette value sharply increasedwith the number of clusters when the 74 series werepartitioned into several decade clusters, but of little signif-icance. Therefore in our study, the correct number ofclusters should be smaller than ten while with a highaverage silhouette value and none (or few) minus silhouettevalues occurring.

3. Results and Discussion

3.1. Spatial Pattern of Mean Annual RET

[19] Spatial distribution of mean annual RET is shown inFigure 2. The regression equation was RET = 2342.028 �

12.055l � 0.062z, with R2 = 0.332. This indicates that themean annual RET generally decreases by 6 mm with rising100 m of altitude and by 12 mm moving one degree oflongitude eastward. Solar radiation increases with altitudebecause of thinner aerosphere, less air density, water vaporand aerosols; wind speed generally increases with altitudeas an observed fact across the plateau; relative humidity isnot clearly related to altitude because of the compleximpacts of topography and atmospheric circulation [Zhanget al., 1982; Guan et al., 1984; Dai, 1990]. The decrease ofRET with altitude is in accordance with the decrease of airtemperature with altitude. The decrease eastward of RET,which does not agree with the spatial distribution oftemperature, is determined by the spatial patterns of solarradiation, relative humidity, and wind speed, which gener-ally decrease from the northwest to the southeast of theplateau throughout the year [Dai, 1990]. The predictor fwas excluded from the regression model, implying that theinfluence of latitude on the distribution of RET is smallerthan and being replaced by that of altitude or longitude.Additionally, the low R2 value suggests that the spatialpattern of RET over the plateau is much determined by thecomplex influence of diverse terrain on the meteorologicalfactors across the Qinghai-Tibetan Plateau [Dai, 1990].

3.2. RET Trends During 1971–2004

[20] The trend slopes of annual and seasonal RET areshown in Figures 3 and 4. Significant decreasing trends

Table 3. Calibration of the Angstrom Coefficients for the 16

Stations With Radiation Observationa

WMONumber Name Length of Record d as bs R2

51709 Kashi 1958–2003 14729 0.248 0.485 0.69851777 Ruoqiang 1958–2003 15879 0.234 0.514 0.74251828 Hetian 1958–2003 15289 0.276 0.475 0.74452418 Dunhuan 1958–2003 15236 0.23 0.537 0.75952533 Jiuquan 1993–2003 3836 0.229 0.527 0.81152754 Gangcha 1993–2003 3909 0.193 0.669 0.83152818 Geermu 1958–2003 16122 0.262 0.563 0.8152866 Xining 1959–2003 15206 0.205 0.57 0.71755228 Shiquanhe 1972–2003 10428 0.134 0.662 0.38655299 Naqu 1961–2003 12942 0.208 0.561 0.51655591 Lhasa 1961–2003 13808 0.284 0.543 0.61656029 Yushu 1961–2003 13349 0.203 0.622 0.71556043 Guoluo 1993–2003 3831 0.256 0.579 0.81956137 Changdu 1961–2003 14353 0.203 0.635 0.66756146 Ganzi 1994–2003 3505 0.301 0.518 0.82556173 Hongyuan 1994–2003 3416 0.197 0.653 0.828aColumn d is the total number of days with available data. All values of

as and bs are statistically significant (p < 0.001).

Table 2. Months With Missing Data for the 75 Stations Used for RET Calculation During 1971–2004

WMONumber

Daily Maximum, Minimum,or Mean Temperature

Daily MeanRelative Humidity Sunshine Duration Wind Speed

52633 February 198052707 April –December 1974 April –December 1974 April –December 1974 April –December 197452818 December 1993 December 1993 December 1993 December 199352842 2001–2004 2001–2004 2001–2004 2001–200452876 February 1992 February 1992 February 1992 February 199255299 October–December 1994 October–December 1994 October–December 1994 October–December 199456038 June 197456257 January, June–August,

November 1971; November,December 1972

D15105 ZHANG ET AL.: TP REFERENCE EVAPOTRANSPIRATION PATTERNS

5 of 14

D15105

were detected in annual RET for 39 stations (i.e., 52% of allstations), while increasing trends were found for only6 stations (i.e., 8% of all stations). The remaining 40% ofthe stations presented no significant trends. Annual RET forstations with significant negative trends decreased by 20–

60 mm/decade, except for Qiemo (no. 51855), which had atrend slope of �141.9 mm/decade. For most of thesestations, significant negative trends were found in atleast two seasons. However, annual RET for stationswith significant positive trends increased by only about

Figure 2. Mean annual reference evapotranspiration for the period of 1971–2000.

Figure 3. Annual trend slopes of reference evapotranspiration during 1971–2004.

D15105 ZHANG ET AL.: TP REFERENCE EVAPOTRANSPIRATION PATTERNS

6 of 14

D15105

Figure 4. Seasonal trend slopes of reference evapotranspiration during 1971–2004.

Figure 5. Partition of stations according to the two-cluster solution by K-means analysis on the annualseries of reference evaportanspiration from 1971 to 2004. The size of the proportional symbol stands forthe inversed distance to the cluster centroid, which equals to one minus the correlation coefficientbetween the member series and the centroid.

D15105 ZHANG ET AL.: TP REFERENCE EVAPOTRANSPIRATION PATTERNS

7 of 14

D15105

20 mm/decade, and the increasing trends mainly occurred inautumn and summer.

3.3. Climate Factors for RET Change

[21] According to the K-means analysis on the annualRET series of the 74 stations from 1971 to 2004, twoclusters can be detected. The stations of Cluster 1 aremainly to the south of 33�N and stations of Cluster 2 tothe north (Figure 5). This pattern implies the possibledifferences in the relative contribution of climate factorsfor RET change, since RET is governed by solar radiation,air temperature, air humidity, and wind speed.[22] Stepwise regression was completed for each of the

75 stations, with the annual means of T, S, U and RH as thepredictors and the annual RET as the dependent variablefrom 1971 to 2004 (2000 for Chaka, Table 4). The regres-sion models are characterized by high R2 values, with anaverage of 0.915. For 58 stations, all four climate variableswere included in the regression models, which means thatair temperature, radiation, wind speed and air humidity allhad significant influences on RET variation at these sta-tions. As an exception, however, U was excluded for Pali(no. 55773) since its regression coefficient is negative,which is against the physical process of evaporation inwhich wind speed is positively correlated to ET, even thoughit is statistically significant (p < 0.1). RH did not enter themodels for three stations (no. 56227, 56093, and 56257),suggesting that humidity played a less important role thanthe other factors in humid areas. T entered all models butfor three stations (no. 51855, 55299, and 56543), and S wasexcluded for only two stations (no. 51777 and 55299). U didnot enter the models for twelve stations, mostly found inthe central plateau. Naqu (no. 55299) is distinguished fromall other stations, for only RH was included in its regressionmodel.[23] The absolute values of the standardized regression

coefficients for T, S, U, and RH are mapped as bar graphs inFigure 6 to reveal the spatial pattern of relative importance ofthe climate factors on RET variation. Apparently, the influ-ence of the same climate factor can vary spatially. T generallyplayed a more important role in the northeast of the plateau.S had a greater impact in the southeast than in the north. Theinfluence of U was much stronger to the north of 33�N thanthat to the south. RH mainly affected the variation of RET atseveral stations in the central plateau.[24] K-means clustering was also performed on the

absolute values of the standardized regression coefficientsof T, S, U, and RH, as a method to further determinethe leading factor of RET variation in different areas. The

Table 4. Stepwise Regression Analysis With Annual Means of T

(�C), S (%), U (m s�1), and RH (%) as Predictors and Annual RET

(mm) as Dependent Variable, Based on the Period of 1971–2004a

WMONumber

Intercept(mm)

Regression Coefficient for

R2T S U RH

51709 705.04 15.84 3.79 231.49 �6.19 0.9551777 796.71 34.68 0 287.72 �9.31 0.9151828 497.97 21.15 4.92 246.33 �4.87 0.9551855 513 0 7.54 223.75 �3.38 0.9951931 340.32 19.79 5.4 265.31 �2.9 0.9752418 504.94 28.14 4.73 215.04 �6.24 0.9452533 510.68 27.02 4.8 164.9 �5.5 0.9452602 508.54 32.95 5.55 142.64 �5.44 0.9552633 591.86 22.95 3.92 76.81 �2.69 0.8552645 698.11 29.86 4.84 39.77 �4.12 0.952652 674.24 31.3 5.24 135.01 �7.69 0.9652657 615.16 31.39 5.95 66.89 �4.97 0.8652679 427.37 19.79 6.69 156.87 �4.54 0.9752707 832.04 41.32 2.66 144.04 �8.38 0.9652713 593.48 27.66 4.94 117.65 �4.94 0.8452737 436.07 29.2 6.96 116.23 �4.12 0.9652754 792.52 37.31 4.8 45.67 �5.99 0.9152787 813.08 28 5.41 34.4 �8.23 0.9152818 554.98 34.97 5.76 163.53 �7.36 0.9752825 622.28 42.94 4.12 132.06 �5.43 0.9452836 743.92 33.12 4.14 106.22 �6.31 0.9552842 806.12 34.79 5.01 70.17 �7.62 0.9652856 1094.3 24.59 0 103.22 �6.72 0.9352866 539.66 16.3 4.9 97.41 �2.78 0.9552868 567.89 36.55 5.15 92.75 �5.23 0.8852876 503.23 29.93 4.6 122.38 �3.6 0.9552908 973.34 24.49 4.25 0 �6.61 0.9252996 906.08 26.71 6.17 0 �7.59 0.8255228 450.98 23.38 6.58 85.04 �2.66 0.8555279 842.05 34.68 7.14 0 �7.21 0.9455299 1439.8 0 0 0 �11.12 0.8655472 733.17 35.84 5.74 49.13 �5.78 0.9655578 675.6 30.92 5.6 106.07 �5.09 0.9655591 644.8 26.57 7.52 114.7 �6.98 0.9355598 805.64 35.67 5.6 118.12 �9.67 0.9955664 313.09 29.55 10.89 68.97 �5.06 0.9655680 522.22 34.89 7.3 103.37 �4.94 0.9755696 786.59 35.02 6.2 91.43 �7.98 0.9355773 912.59 19.07 5.78 �10.75 �6.06 0.956004 741.33 15.96 6.14 22.81 �6.3 0.8256018 709.81 21.87 5.13 45.86 �3.97 0.8756021 613.21 26.62 7.36 36.43 �4.79 0.8456029 381.92 24.94 8.22 84.88 �1.88 0.9256033 804.19 18.49 5.18 0 �4.76 0.8856034 698.02 22.47 5.25 0 �2.89 0.8756038 662.52 23.52 6.74 0 �3.42 0.956046 746.04 24.58 4.92 24.32 �3.65 0.9356067 766.77 16.67 6.09 0 �4.58 0.6556079 994.82 18.88 4.7 0 �7.08 0.7656080 507.03 22.98 6.07 50.66 �2.68 0.8556093 247.24 23.71 7.27 63.22 0 0.956106 684.64 25.31 5.47 63.54 �4.61 0.9356116 597.77 25.27 7.47 57.34 �4.74 0.9356125 584.56 24.49 7.11 76.52 �4.45 0.9356137 453.37 22.55 8.07 122.55 �3.06 0.9756144 606.03 29.9 6.72 92.84 �4.68 0.9856146 578.51 31.72 6.9 56.8 �3.66 0.9456152 752.16 27.07 5.66 26.55 �4.22 0.8756167 523.2 31.32 6.76 100.11 �3.67 0.9456172 468.28 21.56 9.62 77.16 �3.54 0.9356173 637.91 27.72 7.67 0 �4.15 0.8656178 432.16 33.11 8.85 176.47 �6.32 0.9856182 499.27 17.58 8.44 50.51 �2.81 0.856202 836.2 21.07 5.02 24.17 �5.57 0.8656227 213.49 21.61 9.93 80.09 0 0.9256247 375.84 25.75 6.96 226.87 �1.95 0.9756257 319.92 22.3 8.18 53.67 0 0.956287 858.21 13.36 10.21 61.48 �5.99 0.9356312 780.85 17.46 7.23 78.33 �6.37 0.91

Table 4. (continued)

WMONumber

Intercept(mm)

Regression Coefficient for

R2T S U RH

56357 697.96 19.95 8.19 76.17 �6.73 0.9556374 763.86 33.32 5.97 29.69 �5.96 0.9456444 959.17 28.34 7.35 0 �8.04 0.9456462 822.43 34.88 8.31 70.49 �9.19 0.9656543 959.68 0 8.33 0 �6.6 0.8756548 639.72 15.7 9.37 53.2 �4.25 0.93aAll the models are statistically significant (p < 0.01).

D15105 ZHANG ET AL.: TP REFERENCE EVAPOTRANSPIRATION PATTERNS

8 of 14

D15105

75 stations were partitioned into three clusters (Figure 7),with the leading factors of 1) wind speed, 2) humidity,and 3) radiation. The smaller increasing rates of daytimetemperatures than that of daily mean temperatures on theQinghai-Tibetan Plateau had limited the importance ofmean temperature trends for changes of PET rates [Chenet al., 2006]. The increase in mean temperature showed littlecorrelation with the declining trends in RET and panevaporation [Zhang et al., 2007]. Therefore there is nocluster with temperature as the leading factor. Generally,there is a contiguous area in the north of the study regiondominated by Cluster 1 stations, while the southeastern parthas mostly Cluster 3 stations. The Cluster 2 stations areintermingled with the other two clusters. This suggests thatthe changes in RET for most stations, especially those northof 33�N, were mainly attributable to the changes in surfacewind speed; while changes in sunshine duration weregenerally responsible for RET changes in the southeast ofthe plateau.[25] The relative importance of the climate factors on

RET may vary with seasons. Again, high R2 values indi-cated strong robustness of the models of the 75 stations(R2 = 0.97, 0.98, 0.91 and 0.95 for MAM, JJA, SON andDJF, respectively). Figure 8 indicates the seasonal variationof the relative significance of the climate factors for changesin RET across the plateau. In summer, radiation was thedominant factor for most stations to the south of 33�N,while wind speed for those to the north, similar to that of theannual RET (Figure 6). In spring and fall, the number ofstations strongly influenced by radiation was much smallerthan in summer. In fall, different from spring, humidity

became the dominant factor at several stations mainlylocated in the central plateau, e.g., Bange (no. 55279),Tuotuohe (no. 56004), and Maduo (no. 56033). In winter,the impact of radiation was much weaker than the otherfactors for most part of the plateau. Therefore wind speeddominated almost throughout the year for most stations tothe north of 33�N.

3.4. Main Reasons for RET Decrease

[26] The annual and seasonal trends of the climate factors,i.e., T, S, RH, and U, were examined for each station during1971–2004 to determine the reason for the detected RETtrends. In general, significant increases in T and significantdecreases in U were detected almost across the plateauthroughout the year; S significantly decreased to the southof 33�N in JJA, SON, and MAM; RH significantly increasedto the south of 33�N, but most of the increasing trends weredetected in DJF.[27] Five stations with significant RET trends were se-

lected to show the contributing factors and causes for theRET trends (Table 5). Zedang is located in the middlereaches of the Yalungzangbo River to the north of theHimalayas, and Maduo in the source area of the YellowRiver in central Qinghai Province. Dege and Hezuo areclosest to the centroids of Cluster 1 and Cluster 2, respec-tively, in the K-means analysis on annual RET change(Figure 5). Qiemo was selected because of its extremedecreasing trend of RET.[28] At Zedang, the impact of wind speed was the

strongest among the four climatic factors and, therefore thesignificant decrease of wind speed should be cause of

Figure 6. The absolute value of the standardized regression coefficients of T (annual meantemperature), S (annual mean relative sunshine duration), U (annual mean 2-m height wind speed),and RH (annual mean relative humidity), with annual reference evapotranspiration as the dependentvariable for the period of 1971–2004. Stepwise regression is used to derive the linear model.

D15105 ZHANG ET AL.: TP REFERENCE EVAPOTRANSPIRATION PATTERNS

9 of 14

D15105

Figure 7. Partition of stations according to the three-cluster solution, based on the K-means analysis onthe absolute value of the standardized regression coefficients of T (annual mean temperature), S (annualmean relative sunshine duration), U (annual mean 2-m height wind speed), and RH (annual mean relativehumidity), with annual reference evapotranspiration as the dependent variable during 1971–2004. Thesize of the proportional symbol stands for the inversed distance to the cluster centroid, which equals toone minus the correlation coefficient between the member series and the centroid.

Figure 8. Same as in Figure 6, but for the seasonal scales.

D15105 ZHANG ET AL.: TP REFERENCE EVAPOTRANSPIRATION PATTERNS

10 of 14

D15105

the decrease of annual RET (Table 5). At the seasonal scale,the impact of wind speed was essentially the same as that ofradiation in summer, but predominated in the other seasons.At Dege, the leading factor for annual RET change wasradiation (sunshine duration), followed by wind speed.Summer and fall were dominated by radiation; while windspeed had a slightly greater impact in spring and predomi-nated in winter. At Qiemo, the impact of wind speed wasmuch greater than that of the other factors throughout theyear. Therefore the observed significant decrease of RETwas mostly attributed to the significant decline of windspeed at both annual and seasonal scales, although thedecreasing sunshine duration and increasing relative humid-ity also played a role. At Maduo, the impact of humiditywas slightly stronger than that of temperature and radiationfor annual RET change, while wind speed had no significantimpact. Similar patterns were found at the seasonal scale. AtHezuo, radiation played the most important role in annualRET change. Summer and fall were predominated byradiation, while temperature and radiation contributed sim-ilarly in spring, and winter was predominated by windspeed.[29] The above analyses indicate that the decrease of

surface wind speed was mainly responsible for the RETreduction in the north of the plateau. To determine the causeof the decrease in surface wind speed over the plateau, wefirst examined the trends of the zonal wind speed at the500 hPa geopotential height surface for the warm months ofMarch through October when most RET occurs. The Spear-man’s rank correlation coefficients between the 500 hPazonal wind speed and the time variable YEAR were used asthe indicator for temporal trends during 1971–2004 [Mitchellet al., 1966]. For most part of the plateau, the March–October 500 hPa zonal wind speed decreased during thestudy period, with significant decreasing trends found inthe northeast of the region (Figure 9). This finding points tothe changes in circulation patterns and intensity as thepossible cause of the decline of surface wind speed. Toconfirm this, we analyzed the 500 hPa geopotential heightsat both 27.5�N and 40�N, with each averaged for thelongitudes 80�–102�E. Although the geopotential heightsat these two latitudes both increased during 1971–2004, theSouth-North pressure gradient declined with statistical sig-nificance (Figure 10), which led to the decreases in theupper-air zonal wind speed and the reduced strength ofcirculation over the plateau.[30] Our analyses also suggest that the reduced radiation

(sunshine duration) was the leading factor for RET decreaseat most stations in the southeastern plateau during summer-time. This conforms to the general decrease in pan evapo-

ration in the Northern Hemisphere associated with observedlarge and widespread decreases in sunlight due to increasingcloud coverage and aerosol concentration during the past50 years [Roderick and Farquhar, 2002]. Visibility in theclear sky reduced by the presence of aerosols has resulted innet global dimming over land from 1973 to 2007 [Wang etal., 2009]. Liu et al. [2004] suggested that the aerosol-caused decrease in solar irradiance (sunshine duration) wasmost likely the driving force for the reduced pan evapora-tion in China, which coincided with the study by Ren andGuo [2006]. Sunshine duration was also identified as themost influencing variable for pan evaporation changes inthe northeast of India in winter, monsoon and pre monsoonseasons [Jhajharia et al., 2009]. Du et al. [2007] indicatedthat the decrease of annual and summer sunshine duration inTibet during 1971–2005 was mainly related to the increaseof atmospheric water vapor pressure and precipitation. Upto now it is difficult to confirm the impact of atmosphericaerosol on sunshine duration because of the absence ofobservations on atmospheric aerosol in Tibet. However, arecent study revealed that a significant amount of brownclouds that consist of a mixture of light-absorbing and light-scattering aerosols is being generated in South Asia byfossil fuel and biomass burning [Ramanathan et al., 2007],which may be transported over the plateau by the summermonsoonal circulation pattern and cause the decreases insunshine duration in recent years.[31] In our study, using a corrected data set of climatic

variables, we confirmed the general pattern of decreasingtrends of RET across the Qinghai-Tibetan Plateau as iden-tified by Chen et al. [2006] and Zhang et al. [2007]. Still,we detected the spatial pattern of the RET variations acrossthe plateau, i.e., the difference between the north and southof 33�N (Figure 5), which implies some spatial variations ofthe controlling factors of the RET changes. We confirmedthat the combined effect of the reduced wind speed andshortened sunshine duration negated the effect of risingtemperature and caused the RET reduction [Zhang et al.,2007]. However, our analyses suggest that there were bothspatial and seasonal differences of the contributing factorsto the RET trends across the plateau. It is concluded thatwind speed predominated the changes of RET almostthroughout the year, especially in the north of the studyregion; while radiation was the leading factor in the south-east, especially during the summertime. Moreover, wefound that the South-North pressure gradient in the upperair decreased significantly over the plateau, which was themain cause of the reduced strength of circulation and thereduced surface wind speed.

Table 5. Trend Slope and Absolute Value of the Standardized Regression Coefficient of the Climatic Factors for Five Selected Stations

During 1971–2004a

Number Name RET (mm/10a)

Trend Slope/Absolute Value of the Standardized Regression Coefficient

T S U RH

55598 Zedang �57.6** 0.365**/19.03 �1.5**/11.92 �0.464**/62.86 0.8/23.2656033 Maduo 18.9** 0.425**/15.34 1.2**/12.89 �0.107**/0 �1.9**/18.6756144 Dege �58.1** 0.097/12.46 �5.1**/38.75 �0.189**/21.52 2.1**/14.5256080 Hezuo 12.3** 0.407**/12.96 1*/16.38 �0.112**/7.7 �0.8*/5.4951855 Qiemo �141.9** 0.407**/0 �2.9**/35.73 �0.512**/116.44 1.6**/10.43aA star (*) indicates significance at p < 0.05 and two stars (**) at p < 0.01 through the t test. Trend slope of T, S, U, and RH are expressed in �C/10a,

%/10a, m s�1/10a, and %/10a, respectively.

D15105 ZHANG ET AL.: TP REFERENCE EVAPOTRANSPIRATION PATTERNS

11 of 14

D15105

3.5. Possible Change in AET Related With DecreasingRET

[32] Ultimately, AET is the hydrologic flux of interest,while pan evaporation, PET, or RET matters insofar as itcan offer a useful clue to the direction of the change in AET

[Ohmura and Wild, 2002]. However, decreasing pan evap-oration, PET, or RET cannot simply indicate decreasingAET. On the contrary, AET and pan evaporation, PET, orRET can be inversely related in certain cases. Actually, thedecline in pan evaporation has been reported to be associ-

Figure 10. Change of the South-North gradient of geopotential heights for March–October on the500 hPa surface over the Tibetan Plateau. The gradient is indicated by DGH, i.e., the difference ingeopotential heights between the latitudes of 27.5�N and 40�N, averaged for 80�–102.5�E. The lineartrend line of DGH is statistically significant at p < 0.01.

Figure 9. The Spearman’s rank correlation coefficients between March and October mean 500 hPazonal wind speed and the time variable YEAR during 1971–2004. The thin solid contours indicatepositive values; while the dotted contours and the shaded area indicate negative values. The thick solidcontours stand for statistical significance levels.

D15105 ZHANG ET AL.: TP REFERENCE EVAPOTRANSPIRATION PATTERNS

12 of 14

D15105

ated with an incline in actual evaporation in some climates[e.g., Lawrimore and Peterson, 2000; Golubev et al., 2001],which supports the proposal by Brutsaert and Parlange[1998] that decreasing pan evaporation actually provides astrong indication of increasing terrestrial evaporation inmany situations. Theoretically, PET and AET for largehomogeneous surfaces with minimal advection of heatand moisture fall in a complementary relationship, i.e., theBouchet-Morton complementary relationship [Bouchet,1963; Morton, 1975]. Ramırez et al. [2005] provided directobservational evidence of the complementary relationship inregional evapotranspiration hypothesized by Bouchet in1963. This complementary relationship has been examinedand adopted to estimate regional AET [e.g., Sugita et al.,2001; Hobbins et al., 2001; Xu and Singh, 2005; Ozdoganet al., 2006].[33] As for the Qinghai-Tibetan Plateau, regional water

budget-derived AET generally presented insignificantincreasing trends [Zhang et al., 2007], which coincidedwith the detected increasing trend in vapor pressure, as wellas temperature and precipitation [Wu et al., 2007; Du et al.,2007]. AET and PET exhibited some complementarybehavior on the plateau, which, however, did not fall inBouthet’s hypothesis [Zhang et al., 2007]. To explain thetrend in water budget-derived AET and the relationshipbetween pan evaporation and AET, Hobbins et al. [2004]emphasized that both two driving components, i.e., theradiative energy budget (Qn) and the vapor transfer budget(EA), must be considered together. These two budgets wereseparately addressed by Szilagyi et al. [2001], Milly andDunne [2001], and Roderick and Farquhar [2002]. Accord-ing to our analyses, the reduced sunshine duration thatcaused Qn to decrease was mainly responsible for thedecreasing trend of RET in the southeast of the plateau,though the decreasing wind speed and increasing relativehumidity that caused EA to decrease also contributed todecreasing RET. The possible increasing trend in AET canbe explained in the context of the complementary relation-ship, i.e., AET = lWET � PET, where WET is wetenvironment evapotranspiration, and l is a constant greaterthan one but less than two according to Zhang et al. [2007].WET decreases with declining Qn, PET decreases withdeclining Qn and EA, but AET may increase when thedecreasing rate of PET exceeds that of lWET.

4. Conclusion

[34] In this study we examined the spatial and temporalpatterns of reference evapotranspiration (RET) calculatedusing the FAO recommended Penman-Monteith methodover the plateau and its vicinity during 1971–2004. Gen-erally, both annual and seasonal RET decreased for mostpart of the plateau during the study period. At the annualand seasonal scales, we identified the relative contributionof wind speed, sunshine duration, temperature, and relativehumidity to variations of RET. The combined influence ofwind speed, radiation (sunshine duration), and humidity wasgreater than that of temperature on the change of atmo-spheric evaporating power, while the impacts of the climatefactors of RET varied spatially and seasonally. Wind speedpredominated the change of RET almost throughout theyear, especially in the north of the study region; while

sunshine duration was the leading factor in the southeast,especially during the summertime. Although the recentwarming trend over the plateau could have increased RETthrough changes in temperature, the combined effect ofthe reduced wind speed and shortened sunshine durationnegated the effect of the rising temperature and causedRET to decrease in general across the plateau. The signif-icant decline of surface wind speed over most part of theplateau was in accordance with the detected decreasingtrends of zonal wind speed on the 500 hPa surface. Wefound that the geopotential height generally increased overthe plateau for the months March–October, but the 500 hPapressure gradient decreased statistically over the studyregion. The reduced pressure gradient between 27.5�Nand 40�N, possibly as the result of the recent warming, isidentified as the main cause of the reduced surface windspeed.

[35] Acknowledgments. This study was in part supported by theNational Basic Research Program of China (2005CB422006), the ChineseAcademy of Sciences Knowledge Innovation Program (KZCX2-YW-310),the National Natural Science Foundation of China (40871044, 40625002),NASA (NNG05GB85G/EOS/03-0063-0069), and University of San Diego(FRG 07-08, 08-09). We thank the Climate Data Center of NationalMeteorological Information Center of China Meteorological Administrationfor providing surface observations, the NOAA/OAR/ESRL PSD (Boulder,Colorado) for providing the NCEP/NCAR reanalysis data, and the anon-ymous reviewers for their valuable comments and constructive suggestions.

ReferencesAllen, R. G., L. S. Pereira, D. Raes, and M. Smith (1998), Crop Evapo-transpiration — Guidelines for Computing Crop Water Requirements —FAO Irrigation and Drainage Paper 56, Food and Agriculture Organiza-tion of the United Nations, Rome.

Bouchet, R. J. (1963), Evapotranspiration reelle et potentielle, significationclimatique. IAHS Publ. 62, pp. 134–142, Int. Assoc. Sci. Hydrol.,Berkeley, Calif.

Brunel, J. P., J. Ihab, A. M. Droubi, and S. Samaan (2006), Energy budgetand actual evapotranspiration of an arid oasis ecosystem: Palmyra (Syria),Agric. Water Manage., 84, 213–220, doi:10.1016/j.agwat.2006.02.005.

Brutsaert, W., and M. Parlange (1998), Hydrologic cycle explains the eva-poration paradox, Nature, 396, 30.

Burn, D. H., and N. M. Hesch (2007), Trends in evaporation for theCanadian Prairies, J. Hydrol., 336, 61–73.

Burns, D. A., J. Klaus, and M. R. McHale (2007), Recent climate trendsand implications for water resources in the Catskill Mountain region,New York, USA, J. Hydrol., 336, 155 –170, doi:10.1016/j.jhydrol.2006.12.019.

Calanca, P., A. Roesch, K. Jasper, and M. Wild (2006), Global warming andthe summertime evapotranspiration regime of the Alpine region, Clim.Change, 79, 65–78, doi:10.1007/s10584-006-93-9.

Chattopadhyay, N., and M. Hulme (1997), Evaporation and potential eva-potranspiration in India under conditions of recent and future climatechange, Agric. For. Meteorol., 87, 55–73.

Chen, R.-S., E.-S. Kang, J.-P. Yang, S.-H. Lu, and W.-Z. Zhao (2004),Validation of five global radiation modes with measured daily data in China,Energy Convers. Manage., 45, 1759–1769, doi:10.1016/j.enconman.2003.09.019.

Chen, D.-L., G. Gao, C.-Y. Xu, J. Guo, and G.-Y. Ren (2005), Comparisonof the Thornthwaite method and pan data with the standard Penman-Monteith estimates of reference evapotranspiration in China, Clim.Res., 28, 123–132.

Chen, S.-B., Y.-F. Liu, and A. Thomas (2006), Climatic change on theTibetan Plateau: Potential evapotranspiration trends from 1961–2000,Clim. Change, 76, 291–319.

Dai, J.-X. (1990), Climate of the Tibetan Plateau, pp. 23–214, ChinaMeteorological Press, Beijing.

Du, J., B. Duo, J. Hu, J. Liao, and M.-J. Zhou (2007), Climatic change ofsunshine duration and its influencing factors over Tibet during the last35 years, Acta Geogr. Sin., 62(5), 492–500.

Francisco, J. M. (2005), Variability of reference evapotranspiration andwater demands. Association to ENSO in the Maipo river basin, Chile,Global Planet. Change, 47, 212 – 220, doi:10.1016/j.gloplacha.2004.10.013.

D15105 ZHANG ET AL.: TP REFERENCE EVAPOTRANSPIRATION PATTERNS

13 of 14

D15105

Gao, G.-D., Y.-R. Lu, and H.-J. Li (1978), The calculation and distributionof the maximum possible evaporation in China, Acta Geogr. Sin., 33(2),102–111.

Gao, G., D.-L. Chen, G.-Y. Ren, Y. Chen, and Y.-M. Liao (2006), Trend ofpotential evapotranspiration over China during 1956 to 2000, Geogr.Res., 25(3), 378–387.

Golubev, V. S., J. H. Lawrimore, P. Y. Groisman, N. A. Speranskaya, S. A.Zhuravin, M. J. Menne, T. C. Peterson, and R. W. Malone (2001), Eva-poration changes over the contiguous United States and the formerUSSR: A reassesssment, Geophys. Res. Lett., 28(13), 2665–2668.

Guan, Z.-H., C.-Y. Chen, Y.-X. Ou, Y.-Q. Fan, Y.-S. Zhang, Z.-M. Chen,S.-H. Bao, Y.-T. Zu, X.-W. He, and M.-T. Zhang (1984), Rivers andLakes of Xizang, pp. 35–40, Science Press, Beijing.

Hobbins, M. T., J. A. Ramırez, and T. C. Brown (2001), The complementaryrelationship in estimation of regional evapotranspiration: An enhancedadvection-aridity model, Water Resour. Res., 37(5), 1389–1403.

Hobbins, M. T., J. A. Ramırez, and T. C. Brown (2004), Trends in panevaporation and actual evaporation across the conterminous U.S.: Para-doxical or complementary?, Geophys. Res. Lett., 31, L13503,doi:10.1029/2004GL019846.

Jhajharia, D., S. K. Shrivastava, D. Sarkar, and S. Sarkar (2009), Temporalcharacteristics of pan evaporation trends under the humid conditions ofnortheast India, Agric. For. Meteorol., 149, 763–777, doi:10.1016/j.agrformet.2008.10.024.

Jin, L.-Y., J. Li, X. Wang, and F.-H. Chen (2004), The temporal and spatialdistribution of surface dry-wet conditions over northwestern China inrecent 50 years, Acta Geogr. Sin., 59(6), 847–854.

Kalnay, E., et al. (1996), The NCEP/NCAR 40-year reanalysis project, Bull.Am. Meteorol. Soc., 77(3), 437–471, doi:10.1175/1520-0477(1996)077<0437:TNYRP>2.0.CO;2.

Kundzewicz, Z. W., and L. Somlyody (1997), Climatic change impact onwater resources in a systems perspective, Water Resour. Manage., 11,407–435.

Lawrimore, J. H., and T. C. Peterson (2000), Pan Evaporation Trends inDry and Humid Regions of the United States, J. Hydrometeorol., 1, 543–546.

Li, G.-P. (2002), Dynamic Meteorology of the Tibetan Plateau, pp. 89–98,China Meteorological Press, Beijing.

Liu, T.-C., and Qimeiduoji (1999), Characteristics, development and utili-zation prospects of water resources of the international river area inQinghai-Tibet Plateau, Acta Geogr. Sin., 54(Suppl.), 11–20.

Liu, H.-Z., H.-S. Zhang, L.-G. Bian, J.-Y. Chen, M.-Y. Zhou, X.-D. Xu,S.-M. Li, and Y.-J. Zhao (2002), Characteristics of micrometeorology inthe surface layer in the Tibetan Plateau, Adv. Atmos. Sci., 19(1), 73–88.

Liu, B., M. Xu, M. Henderson, and W. Gong (2004), A spatial analysis ofpan evaporation trends in China, 1955–2000, J. Geophys. Res., 109,D15102, doi:10.1029/2004JD004511.

Liu, H., L.-H. Geng, H.-P. Zhong, and Y. Gu (2006a), On expediting theexploitation of water resource of international rivers in China, YangtzeRiver, 37(7), 32–33.

Liu, X., Z.-Y. Yin, X. Shao, and N. Qin (2006b), Temporal trends andvariability of daily maximum and minimum, extreme temperature events,and growing season length over the eastern and central Tibetan Plateauduring 1961 – 2003, J. Geophys. Res., 111, D19109, doi:10.1029/2005JD006915.

Luosang, L. (2005), Utilization and protection of water resources in theQinghai-Tibet Plateau, Resources Science, 27(2), 23–27.

Milly, P. C. D., and K. A. Dunne (2001), Trends in evaporation and surfacecooling in the Mississippi River basin, Geophys. Res. Lett., 28(7), 1219–1222.

Mitchell, J. M., B. Dzerdzeevskii, H. Flohn, W. L. Hofmeyr, H. H. Lamb,K. N. Rao, and C. C. Wallen (1966), Climate Change, Technical Note 79,79 pp., World Meteorological Organization, Geneva, Switzerland.

Morton, F. I. (1975), Estimating evaporation and transpiration from clima-tological observations, J. Appl. Meteorol., 14, 488–497.

Ohmura, A., and M. Wild (2002), Is the hydrological cycle accelerating?,Science, 298, 1345–1346.

Ozdogan, M., G. D. Salvucci, and B. T. Anderson (2006), Examination ofthe Bouchet-Morton complementary relationship using a mesoscale cli-mate model and observations under a progressive irrigation scenario,J. Hydrometeorol., 7(2), 235–251.

Penman, H. L. (1948), Natural evaporation from open water, bare soil andgrass, Proc. R. Soc. London , 193, 120–145.

Penman, H. L. (1956), Evaporation: An introductory survey, Neth. J. Agric.Sci., 4, 9–29.

Peterson, T. C., V. S. Golubev, and P. Y. Groisman (1995), Evaporationlosing its strength, Nature, 377, 687–688.

Ramanathan, V., M. V. Ramana, G. Roberts, D. Kim, C. Corrigan,C. Chung, and D. Winker (2007), Warming trends in Asia amplified bybrown cloud solar absorption, Nature, 448, 575–578.

Ramırez, J. A., M. T. Hobbins, and T. C. Brown (2005), Observationalevidence of the complementary relationship in regional evaporation lendsstrong support for Bouchet’s hypothesis, Geophys. Res. Lett., 32,L15401, doi:10.1029/2005GL023549.

Ren, G.-Y., and J. Guo (2006), Change in pan evaporation and the influen-tial factors over China: 1956–2000, J. Nat. Resour., 21(1), 31–44.

Roderick, M. L., and G. D. Farquhar (2002), The cause of decreased panevaporation over the past 50 years, Science, 298, 1410 – 1411,doi:10.1126/science.1075390.

Roderick, M. L., and G. D. Farquhar (2004), Changes in Australian panevaporation from 1970 to 2002, Int. J. Climatol., 24, 1077–1090.

Roderick, M. L., L. D. Rotstayn, G. D. Farquhar, and M. T. Hobbins(2007), On the attribution of changing pan evaporation, Geophys. Res.Lett., 34, L17403, doi:10.1029/2007GL031166.

Seber, G. A. F. (1984), Multivariate Observations, pp. 380–382, Wiley,New York.

Shen, D.-J., and C.-Y. Chen (1996), Water resources of the Qinghai-XizangPlateau and its exploitation, J. Nat. Resour., 11(1), 8–14.

Spath, H. (1985), Cluster Dissection and Analysis: Theory, FORTRANPrograms, Examples, translated by J. Goldschmidt, 226 pp., HalstedPress, New York.

Sugita, M., J. Usui, I. Tamagawa, and I. Kaihotsu (2001), Complementaryrelationship with a convective boundary layer model to estimate regionalevaporation, Water Resour. Res., 37(2), 353–365.

Szilagyi, J., G. G. Katul, and M. B. Parlange (2001), Evapotranspirationintensifies over the conterminous United States, J. Water Resour. Plann.Manage, 127(6), 354–362.

Tebakari, T., J. Yoshitani, and C. Suvanpimol (2005), Time-space trendanalysis in pan evaporation over Kingdom of Thailand, J. Hydrol.Eng., 10(3), 205–215.

Wang, K.-C., R. E. Dickinson, and S.-L. Liang (2009), Clear sky visibilityhas decreased over land globally from 1973 to 2007, Science, 323,1468–1470, doi:10.1126/science.1167549.

Wu, S.-H., Y.-H. Yin, D. Zheng, and Q.-Y. Yang (2007), Climatic trendsover the Tibetan Plateau during 1971–2000, J. Geogr. Sci., 17(2), 141–147, doi:10.1007/s11442-007-0141-7.

Xu, C.-Y., and V. P. Singh (2005), Evaluation of three complementaryrelationship evapotranspiration models by water balance approach toestimate actual regional evapotranspiration in different climatic regions,J. Hydrol., 308, 105–121, doi:10.1016/j.jhydrol.2004.10.024.

Xu, J.-Q., S. Haginoya, K. Masuda, and R. Suzuki (2005), Heat and waterbalance estimates over the Tibetan Plateau in 1997–1998, J. Meteorol.Soc. Jpn., 83(4), 577–593.

Xu, C.-Y., L.-B. Gong, T. Jiang, D.-L. Chen, and V. P. Singh (2006),Analysis of spatial distribution and temporal trend of reference evapo-transpiration and pan evaporation in Changjiang (Yangtze River) catch-ment, J. Hydrol., 327, 81–93, doi:10.1016/j.jhydrol.2005.11.029.

Yu, P., T. Yang, and C. Chou (2002), Effects of climate change on evapo-transpiration from paddy fields in southern Taiwan, Clim. Change, 54,165–179.

Zhang, R.-Z., D. Zheng, and Q.-Y. Yang (1982), Physical Geography ofXizang (Tibet), pp. 32–54, Science Press, Beijing.

Zhang, Y., C. Liu, Y. Tang, and Y. Yang (2007), Trends in pan evaporationand reference and actual evapotranspiration across the Tibetan Plateau,J. Geophys. Res., 112, D12110, doi:10.1029/2006JD008161.

Ziegler, A. D., E. P. Maurer, J. Sheffield, B. Nijssen, E. F. Wood, and D. P.Lettenmaier (2005), Detection time for plausible changes in annual pre-cipitation, evapotranspiration, and streamflow in three Mississippi Riversub-basins, Clim. Change, 72, 17–36, doi:10.1007/s10584-005-579-4.

�����������������������Z. Lin, Y. Ren, X. Zhang, and D. Zheng, Institute of Geographical

Science and Natural Resources Research, Chinese Academy of Science,A11 Datun Road, Anwai, Chaoyang District, Beijing 100101, China.(renyu06@mails.gucas.ac.cn; zhangxq@igsnrr.ac.cn; zhengd@igsnrr.ac.cn)Z.-Y. Yin, Marine Science and Environmental Studies, University of San

Diego, 5998 Alcala Park, San Diego, CA 92110, USA. (zyin@sandiego.edu)

D15105 ZHANG ET AL.: TP REFERENCE EVAPOTRANSPIRATION PATTERNS

14 of 14

D15105