Agricultural and Forest Meteorology 182– 183 (2013) 342– 351
Contents lists available at ScienceDirect
Agricultural and Forest Meteorology
j our nal ho mep age : www.elsev ier .com/ locate /agr formet
uantifying the interactive impacts of global dimming and warmingn wheat yield and water use in China
iaoya Yanga,b,c,∗, Senthold Assengd, Mike Ting Fook Wongb, Qiang Yue, Jun Li c,nmin Liuc
Jiangsu Key Laboratory of Agricultural Meteorology, College of Applied Meteorology, Nanjing University of Information Science and Technology, Nanjing10044, ChinaCSIRO Land and Water, Private Bag 5, Wembley, WA 6913, AustraliaKey Laboratory of Water Cycle & Related Land Surface Processes, Institute of Geographic Sciences and Natural Resources Research, Beijing 100101, ChinaAgricultural & Biological Engineering Department, University of Florida, 221 Frazier Rogers Hall, P.O. Box 110570, Gainesville, FL 326 11-0570, USAPlant Functional Biology & Climate Change Cluster, University of Technology, Sydney, P.O. Box 123, Broadway, NSW 2007, Australia
r t i c l e i n f o
rticle history:eceived 30 November 2012eceived in revised form 14 March 2013ccepted 12 July 2013
eywords:olar radiationheat yield
vapotranspirationater use efficiency
PSIM-Nwheat
a b s t r a c t
Solar radiation has been declining across many parts of the world over the last 50 years as a consequenceof industrialization increasing atmospheric aerosols, known as ‘global dimming’. This study evaluatesthe impact of ‘global dimming’ and climate change on wheat yield and water use in China during thepast decades using the Agricultural Production Systems Simulator. Three regions, Beijing, Chengdu andUrumqi were selected to represent three different patterns of climate-light environments in China. Thedecline in solar radiation was in conjunction with a warming trend during the past decades. Solar radiationduring the wheat season declined by 20, 27 and 10% at Beijing, Chengdu and Urumqi, respectively, duringthe past four decades. Minimum temperature increased during the same period by 3.9, 1.5 and 2.3 ◦C,respectively. The reduction in solar radiation had no significant impact on simulated wheat yields in theBeijing region while simulated grain yields in the Chengdu region decreased by 32%. Variation of solar
radiation explained 74% of changes in grain yield at Chengdu. Simulated grain yields in the Urumqi regionincreased by 24% during the last decades due to increasing minimum temperature and rainfall. Simulatedevapotranspiration declined with the decline of solar radiation. Water use efficiency increased at Beijingand Urumqi, with no significant change at Chengdu. Declining solar radiation from high radiation levels
ield bntly.
had no effect on wheat yyields decreased significa
. Introduction
The amount of incident solar radiation governs the physiologicalnd biophysical processes of vegetation over land surface, such asanopy photosynthesis, and evapotranspiration (ET), as well as thenergy balance over the diurnal/seasonal time frames. The globalolar radiation reaching the surface decreased 1.3% per decade onverage from 1960 to 2000 (IPCC, 2007). A significant decline inolar radiation together with an increase in the diffuse fractionf light has been reported worldwide (Wild et al., 2005; Wild,009), especially across China since 1960 (Che et al., 2005, 2007;
aiser and Qian, 2002; Qian et al., 2006, 2007). The decline ofolar radiation in China was attributed to increased atmosphericerosols, which has been caused by a rapid economic growth. Solar
∗ Corresponding author at: Jiangsu Key Laboratory of Agricultural Meteorology,ollege of Applied Meteorology, Nanjing University of Information Science and Tech-ology, Nanjing 210044, China. Tel.: +86 25 5873 1539; fax: +86 25 5873 1539.
radiation is absorbed, or reflected back into space, by aerosols andother particulates.
Reductions in solar radiation and change in radiation com-position could have significant effects on plant growth, ET andwater use efficiency (WUE) (e.g., Choudhury, 2001; Rodriguez andSadras, 2007). Crop simulation models are useful tools to assessthe impacts of changing drivers on crop physiological and soilprocesses, diagnose crop growth, and predict crop yield (Assenget al., 2004; Lawes et al., 2009). Some researchers have showncrop yields would decrease with the declining solar radiation. Forexample, Chameides et al. (1999) found a close correlation betweendecrease in solar radiation and rice yield at Nanjing of China usingthe CERES model. Chen et al. (2009) found that simulated potentialyield and crop water demand were significantly reduced becauseof the declining trend in solar radiation in the North China Plain.These results were simulated under similar climate conditions over
eastern China, where solar radiation is at a medium level and theclimate is monsoonal.
Wheat is the dominant crop for food production in China. Itis grown across the Northeast, Northwest, North China Plain and
outhwest, with diverse climate factors limiting wheat production.n Northwest, solar radiation is high in the dry region, and mayot limit wheat yield, whereas it is lowest in the Southwest. Whenhe photosynthesis-light response curve is steep under low light,rop growth is sensitive to light change in low light condition, buts insensitive to light when light environment varies around lightaturation point (Yu et al., 2002). In addition, increasing temper-ture may have different impacts on wheat yield under differentemperature levels (Lobell and Ortiz-Monasterio, 2007).
The diverse effects of climate warming and dimming in theheat cultivation regions are complex and poorly understood
Stanhill and Cohen, 2001). Therefore, there is a need to synthe-ize this knowledge and extend it beyond experimental sites andesting years using agricultural system models, so that agriculturalesources can be used optimally under a changing climate acrossites and times, especially for determining the best response strate-ies for the future.
This study provides a mechanistic understanding of the interac-ive impacts of global dimming and climate warming under variableainfall on wheat production in China over the past decades. Thebjective of this study is to evaluate the effect of solar radiationecline on potential wheat yield, ET and WUE for some of the mainropping regions in China. The aims of this study are to (1) validatend calibrate a crop model, APSIM-Nwheat to simulate interac-ive impacts of changes in solar radiation and temperature; and2) to use the crop model to quantify wheat yields and water-usen response to changes in solar radiation and temperature over theast decades in China.
. Materials and methods
.1. Study sites and climate data
Representative climate stations were selected for three mainheat production regions of China. These included Beijing (39.8◦
Fig. 1. Monthly mean maximum and minimum temperature, solar radiation and precip
eorology 182– 183 (2013) 342– 351 343
N, 116.5◦ E) in the North China Plain with a temperate monsoonclimate, Chengdu (30.7◦ N, 104.0◦ E) in the Southwest with subtrop-ical climate, and Urumqi (43.8◦ N, 87.6◦ E) in the Northwest with anarid climate in the centre of the Eurasian continent. Solar radiationand its decline differed at each location. Solar radiation declined inthe order from the highest to lowest decline from Urumqi to Bei-jing, and to Chengdu. These three sites were selected due to theavailability of complete daily climate records from 1961 to 2009(climate data of Chengdu was from 1961 to 2003), including dailysolar radiation, diffuse radiation, maximum and minimum temper-ature and precipitation. Climate data were obtained from the ChinaMeteorological Administration.
Fig. 1 shows the mean values of maximum and minimum tem-perature, solar radiation and precipitation of each month at Urumqi(1961–2009), Beijing (1961–2009) and Chengdu (1961–2003). Themaximum and minimum temperatures of Chengdu were higherthan Beijing and Urumqi except the months from May to Septemberwhen temperatures converged, and for which Beijing’s temper-atures were higher than Urumqi’s. Reported solar radiations atUrumqi were the highest recorded for the May to September period.Precipitation at Beijing and Chengdu were higher than at Urumqiespecially from June to September.
2.2. APSIM-Nwheat model
The Agricultural Production Systems Simulator (APSIM)(Keating et al., 2003) for wheat (APSIM-Nwheat version 1.55s) isa crop simulation model, consisting of modules that incorporateaspects of soil water, nitrogen, crop residues, crop growth anddevelopment and their interactions within a wheat crop/soil sys-tem that is driven by daily weather data. Wheat crop (NWHEAT),
soil water (SOILWAT), soil N (SOILN) and residue (RESIDUE) aremost relevant to the simulation of wheat-based cropping systems.NWHEAT, SOILWAT and SOILN have evolved from the CERESmodels of Ritchie et al. (1985) and Jones and Kiniry (1986), and
itation at Urumiqi (1961–2009), Beijing (1961–2009) and Chengdu (1961–2003).
344 X. Yang et al. / Agricultural and Forest Me
F
tebauRtMe
bRisAreHrtdtsrctawdts
bc7tam
fcgtcun
Urumqi (1961–2009) respectively. Wheat cultivars were the same
he PERFECT model (Littleboy et al., 1992), as modified by Probertt al. (1995, 1998) and Keating et al. (2001). The main differencesetween the APSIM-Nwheat model and the CERES-Wheat modelre described by Keating et al. (2001). The model is availablender a license agreement from Agricultural Production Systemsesearch Unit (APSRU). APSIM-Nwheat has been extensivelyested against a range of field measurements in particular in the
editerranean climatic regions of Western Australia (Probertt al., 1995, 1998; Asseng et al., 1998a,b, 2001, 2004).
APSIM-Nwheat calculates potential daily biomass productionased on light interception and radiation-use efficiency (RUE). DailyUE is simulated using daily solar radiation (Fig. 2). RUE would
ncrease with the declining of solar radiation, especially whenolar radiation drops below 10 MJ m−2. The growth function inPSIM-Nwheat considers an increasing RUE with decreasing solaradiation, following the CERES Wheat model approach (Ritchiet al., 1985), based on wheat experiments by Spiertz and Van deaar (1978). Sub-optimal temperatures, water and N-deficit can
educe the potential growth. Potential water demand is a func-ion of transpiration efficiency (TE) modified by vapour pressureeficit (Monteith, 1988). Simulated plant water uptake is a func-ion of uptake demand, root length density distribution within theoil profile, and available soil water in different soil layers. Theate of rooting depth advance is a function of air temperature,rop water stress, and soil water content in the soil layer withhe deepest roots. Vertical soil water movement is simulated using
multi-layered soil model primarily using a cascading approach,ith movement both upward and downward also occurring byiffusive flow (Probert et al., 1998). Water (deficit) stress reducesillering, leaf area index (LAI) and photosynthesis, and enhancesenescence.
Grain yield is a function of grain number, grain filling and car-ohydrate remobilization. In the model, the potential amount ofarbohydrates available for remobilization to grain is defined as5% of biomass growth between 150 ◦C d before grain filling andhe commencement of grain filling. Crop phenology is a function ofccumulated degree-days, photoperiod and vernalization require-ents.In the model, critical and minimum crop N concentrations are a
unction of growth stage. Crop N demand is the difference betweenritical and actual N content and the N amount required for newrowth on a given day. Nitrogen uptake follows the approach ofhe CERES-Wheat model and is a function of potential N uptake
apacity of the root systems and crop N demand. The potential Nptake capacity is a function of root length density distribution, soilitrate and ammonium concentrations and soil water content.
teorology 182– 183 (2013) 342– 351
2.3. Experimental data for model evaluation
Observed data on wheat phenological stages, leaf area index(LAI), biomass, grain yield and evapotranspiration (ET) duringwheat growth seasons from three experimental stations were usedto calibrate and test the APSIM model. These three experimentalstations were Luancheng Agroecosystem Station (37.9◦ N, 114.7◦
E), Yanting Agroecosystem Station (31.3◦ N, 105.5◦ E), and Wulan-wusu Agrometeorology Experimental Station (44.3◦ N, 85.8◦ E)(Table 1). The soil characteristics of the three experimental sitesare shown in Table 2. The parameters of calibrated APSIM modelin Luancheng, Yanting and Wulanwusu were used for Beijing,Chengdu and Urumqi, respectively.
Sowing date, fertilization and irrigation date and amount(Table 3) were similar during three wheat growth seasons atLuancheng, Yanting and Wulanwusu respectively. The sowingdates of three sites were mainly determined by temperature. Wheatwas sown when average temperature declined around 10 ◦C. Fer-tilizer N was applied twice during wheat growth; at sowing andaround jointing. Irrigation was applied when soil water contentwas low; there was no irrigation at Yanting station.
Maximum and minimum temperature, solar radiation and rain-fall at the three experimental sites were recorded with an automaticmeteorological station. The automatic meteorological instrument ismade by Vaisala, and the model is Milos520. Wheat varieties sownwere Gaoyou 503 (Luancheng), 01-3570 (Yanting) and Kuidong 4(Wulanwusu). Ten plants of wheat were randomly selected fromthe sampling area for measuring LAI. Dry matter was determinedafter the harvested samples were oven-dried for 8–10 h. Actual ETwas measured daily with large-scale weighing lysimeters with a0.02 mm precision at Luancheng.
2.4. Model calibration
Simulations were initialized with soil profile measurements ofboth soil water and soil N at three sites five days before sow-ing. As APSIM-Nwheat does not include a function for snow cover,temperatures below 0 ◦C will damage the simulated leaf area. In sit-uations with prolonged periods of temperature below 0 ◦C this willlead to an unrealistic termination of the simulated crop. To avoidsuch a simulated termination in cold winters, temperatures below0 ◦C at Wulanwusu were replaced with 0 ◦C for the simulationstudy. Crop parameters for three experimental stations (Table 4)were established based on information from field experiments atLuancheng, Yanting and Wulanwusu. Two soil parameters whichaffect evapotranspiration of crop were u (first stage soil evaporationcoefficient) and cona (second stage soil evaporation coefficient),the values of 7 and 3.5, used for similar soils in other studies wereused.
The APSIM-Nwheat model was evaluated by comparing the sim-ulated results of all wheat seasons at three experimental stationswith measured data. The slope for a regression lines which had beenforced through the origin, the coefficient of determination (r2) forthe 1:1 regression line and the root mean square deviation (RMSD)were used to quantify the performance of the model.
2.5. Modelling the impacts of climate change on grain yield andwater use of wheat during past years
The APSIM-Nwheat model at Luancheng, Yanting and Wulan-wusu experimental stations was used to simulate wheat growthand water use of Beijing (1961–2009), Chengdu (1961–2003) and
as from the field experiments.Wheat was planted on the day of year 275 (Beijing), 300
(Chengdu) and 269 (Urumqi) according to the field experiments
X. Yang et al. / Agricultural and Forest Meteorology 182– 183 (2013) 342– 351 345
Table 1Experimental sites.
Location Latitude Longitude Soil type Yearly rainfalla
Wulanwusu 44.3 85.8 Sandy loam 236 179 2000–2003 Yang et al. (2011) Urumqi
a Average annual rainfall of four years.b Average rainfall of three wheat growth seasons.
Table 2Soil characteristics at Luancheng (Beijing), Yanting (Chengdu) and Wulanwusu (Urumqi), with lower limit for bulk density (BD), saturation (SAT), drained upper limit (DUL),and lower limit of plant available soil water (LL).
Table 3Average sowing date, fertilizer application date and amount, irrigation date and amount at Luancheng (Beijing), Yanting (Chengdu) and Wulanwusu (Urumqi) during threewheat growing seasons.
Location Sowing date Fertilize date Fertilize amount (kg ha−1 NO3 N) Irrigation date Irrigation amount (mm)
Luancheng 2 October 25 September 200 1 December 40(Beijing) 20 April 200 5 April 104
25 April 10210 May 10420 May 80
Yanting 27 October 25 October 200(Chengdu) 21 March 200
Wulanwusu 26 September 25 September 300 27 September 105(Urumqi) 12 April 300 16 October 90
10 November 13015 April 908 May 10525 May 10512 June 105
Table 4Wheat cultivar parameters (and represented locations in brackets) for cv Gaoyou 503 (Luancheng and Beijing), cv 01-3570 (Yanting and Chengdu) and cv Kuidong 4(Wulanwusu and Urumqi).
Parameter Gaoyou 503 01-3570 Kuidong 4
plv: Sensitivity to vernalization (1–5) 3.5 3.0 3.3pld: Sensitivity to photoperiod (1–5) 2.8 3.2 2.5p5: Thermal time (base 0 ◦C) from end-ear growth to maturity (◦C days) 560 680 650Grno: Coefficient of kernel number per stem weight at the ‘beginning of grain filling’ (kernel (g stem)−1). 20 25 27Fillrate: Maximum kernel growth rate (mg kernel−1 days−1) 2.0 2.5 2.5
−1
wWas
stmwt: Potential final dry weight of a single stem, excluding grain (g stem ).
phylo: Phyllochron interval (◦C days (leaf appearance)−1)
sla: specific leaf area
hich were representative of farmers practice in these regions.heat was harvested at physiological maturity. The soil char-
cteristics were used from the corresponding field experimentaltations which are representative of the regions. Wheat yield and
3 3 395 95 95
225 245 300
water use were simulated under the full irrigation condition byusing the automatic irrigation facility in the APSIM-Nwheat model.Stress factors such as nutrient deficit, disease and pests were notconsidered.
346 X. Yang et al. / Agricultural and Forest Meteorology 182– 183 (2013) 342– 351
Table 5Summary of the APSIM-Nwheat model performance at Luancheng (Beijing), Yanting (Chengdu) and Wulanwusu (Urumqi), China.
Model attribute Number of paired data points Observed range r2 a mb RMSDc
b Slop of linear regression (forced through origin).c Root mean squared deviation.d Days from sowing to maturity.
. Results
.1. Performance of APSIM-Nwheat model
The performance of the model compared with observations atuancheng, Yanting and Wulanwusu is summarized in Table 5.he simulated LAI, biomass and grain yield are compared withhe corresponding experimental data at Luancheng, Yanting and
ulanwusu (Fig. 3a–c). Simulated evapotranspiration (ET) isompared with the measured ET during the wheat growing sea-on at Luancheng (Fig. 3d). The coefficients of determination of:1 regression lines were 0.84 and 0.93, indicating a satisfac-ory performance of the model to simulate LAI and biomass.owever the model tended to over-predict biomass at Yant-
ng, and under-predict biomass at Wulanwusu near the endtage of wheat growth. The model could not reproduce well theiomass during grain filling stage in the different environments.he grain yield simulated with a coefficient of determination
ig. 3. Observed vs. simulated LAI, biomass and yield of wheat (a–c) at Luancheng (Beijinine is the 1:1 line. Comparison of observed and simulated data of cumulative ET (days) d
r2 = 0.37. The low coefficient of variation of observed grain yields(5.0–6.3 t ha−1) was one reason for the relatively low r2 val-ues. The RMSD of yield was 0.4 t ha−1 for an average yield of5.4 t ha−1.
Simulated cumulative ET during the wheat growing season cor-responded well with the daily measured values (Fig. 3d), althoughsome periods were overestimated between 5 December 1998 and2 April 1999.
3.2. Observed climate trends during the past years
Table 6 shows the average values and trends in solar radiation(Ra), maximum temperature (Tmax), minimum temperature (Tmin)and rainfall in wheat season and each year at Beijing (1961–2009),
Chengdu (1961–2003) and Urumqi (1961–2009). Solar radiationof Chengdu was less than Beijing and Urumqi. Solar radiationof three stations was significant decreasing by −0.061, −0.059and −0.026 MJ m−2 a−1 during wheat growing seasons at Beijing,
X. Yang et al. / Agricultural and Forest Meteorology 182– 183 (2013) 342– 351 347
Table 6Average observations and trends in solar radiation (Ra), maximum temperature (Tmax), minimum temperature (Tmin) and rainfall in wheat season and whole year at Beijing(1961–2009), Chengdu (1961–2003) and Urumqi (1961–2009). The wheat season of Beijing, Chengdu and Urumqi were from 1st October to 15 June of next year, 1st Novemberto 15 May next year, and 1st October to 30 June of next year respectively according to experimental data.
Beijing Chengdu Urumqi
Wheat season Annual Wheat season Annual Wheat season Annual
Ra (MJ m−2) Average 13.25 14.37 7.85 9.69 12.34 14.25Trend (a−1) −0.061** −0.070** −0.059** −0.068** −0.026** −0.034**
hengdu and Urumqi respectively. Trends of average solar radia-
ion and diffuse fraction (diffuse radiation/solar radiation) of eachheat growing season at three sites during past years are shown
n Fig. 4. Linear regression and determination coefficients of the
able 7egression analysis of observed solar radiation and diffuse light fraction (diffuse radia1961–2003) and Urumqi (1961–2009).
Solar radiation (MJ m−2) r2
Beijing y = −0.061x + 135 0.Chengdu y = −0.059x + 125 0.Urumqi y = −0.026x + 64.9 0.
radiation/solar radiation) of each wheat growing season at Beijing (1961–2009),f each linear regression is shown in Table 7.
regression analysis are shown in Table 7. Solar radiation during the
wheat season decreased by 18.9, 27.9 and 9% relative to the val-ues of 1961 at Beijing, Chengdu and Urumqi during the past years.Diffuse fraction trends of Chengdu significantly increased during
tion/solar radiation) of wheat growing seasons at Beijing (1961–2009), Chengdu
Diffuse fraction (%) r2
620 y = 0.059x − 65.1 0.055514 y = 0.179x − 269 0.396191 y = 0.095x − 136 0.083
348 X. Yang et al. / Agricultural and Forest Meteorology 182– 183 (2013) 342– 351
F eijingL .
pU
d1Twmi
jawe
3a
y(cwB
TRea
ig. 5. Simulated length of growing duration of wheat (a) and grain yields (b) at Bines show the linear trends, equation of each linear regression is shown in Table 8
ast years, while there was no significant change at Beijing andrumqi.
For Chengdu, Beijing and Urumqi the mean temperatureecreased. The average temperature of the wheat season was1.5, 7.8 and 3.0 ◦C at Chengdu, Beijing and Urumqi respectively.he maximum temperature at Beijing had significantly increased,ith no significant trend at Chengdu and Urumqi. The trends ofinimum temperature of the three stations were all significantly
ncreased, with Beijing showing the largest increase.The annual rainfall of Chengdu was the largest, followed by Bei-
ing and Urumqi. Rainfall during the wheat season was the largestt Urumqi, followed by Chengdu and Beijing. Rainfall during theheat season at Beijing and Urumqi had significantly increased,
specially at Urumqi.
.3. The impact of climate change on wheat growing durationnd grain yield
Simulated length of growing duration of wheat and grainield at Beijing (1961–2009), Chengdu (1961–2003) and Urumqi
1961–2009) are shown in Fig. 5. Linear model and determinationoefficients of the regression analysis are shown in Table 8. Theheat growing duration of Urumqi was the greatest followed byeijing and Chengdu with an average of 270, 254 and 203 days
able 8egression analysis of simulated wheat growth duration, grain yield, potentialvapotranspiration (ETpot), evapotranspiration (ET) and water use efficiency (WUE)t Beijing, Chengdu and Urumqi, respectively.
Linear model r2
Wheat growthduration (days)
Beijing y = −0.28**x + 802 0.536Chengdu y = −0.10x + 405 0.048Urumqi y = 0.02x + 229 0.002
Grain yield (t ha−1) Beijing y = −0.007x + 19.8 0.026Chengdu y = −0.050**x + 104 0.536Urumqi y = 0.032**x − 58.0 0.242
ETpot (mm) Beijing y = −2.88**x + 6294 0.654Chengdu y = −2.13**x + 4532 0.541Urumqi y = −0.57*x + 1705 0.098
ET (mm) Beijing y = −2.27**x + 4980 0.457Chengdu y = −2.08**x + 4386 0.591Urumqi y = −0.73**x + 1921 0.203
WUE(kg ha−1 mm−1)
Beijing y = 0.035*x − 58.0 0.115Chengdu y = −0.039x + 97.5 0.057Urumqi y = 0.090**x − 165.9 0.424
respectively. The wheat growing duration at Beijing was signifi-cantly shortened in the past 48 years by 0.28 days a−1 (Table 8) dueto a significant increase in maximum and minimum temperature of0.033 and 0.08 ◦C a−1 during the wheat growing seasons. The wheatgrowing duration of Chengdu and Urumqi had not changed signifi-cantly as there were only small increases in minimum temperatureby 0.034 and 0.047 ◦C a−1, respectively, during the wheat growingseason.
Simulated grain yields of Urumqi were greater than Beijing andChengdu, with average values of 6.3, 5.2 and 5.3 t ha−1 at Urumqi,Beijing and Chengdu respectively. The simulated grain yields atUrumqi were significantly increased by 0.032 t ha−1 a−1 during thepast decades. Simulated grain yields at Chengdu were significantlydecreased by −0.05 t ha−1 a−1. Simulated grain yields at Beijingshowed no significant trend.
3.4. Effect of climate change on potential evapotranspiration andwater use of wheat
Cumulative potential evapotranspiration (ETpot), evapotrans-piration (ET) and water use efficiency (WUE) simulated byAPSIM-Nwheat at Beijing (1961–2009), Chengdu (1961–2003)and Urumqi (1961–2009) are shown in Fig. 6. Linear regres-sion and determination coefficients of the regression analysis areshown in Table 8. ETpot at Chengdu was lower than Urumqiand Beijing, 312, 565 and 571 mm respectively. ETpot at theselocations had decreased significantly; the trends at Beijing andChengdu were −2.82 and −2.25 mm a−1 and greater than atUrumqi (−0.56 mm a−1), reflecting observed differences in solarradiation decline at Beijing (−0.061 MJ m−2 a−1) and Chengdu(−0.059 MJ m−2 a−1) and Urumqi (−0.026 MJ m−2 a−1).
Average simulated ET at Chengdu was 264 mm less than at Bei-jing (465 mm) and Urumqi (468 mm). The ET trends were similarto the trends of ETpot.
The average simulated WUE at Beijing, Chengdu and Urumqiwere 11.2, 19.9 and 13.5 kg ha−1 mm−1. WUE at Chengdu was thehighest as solar radiation was the least and the diffuse light fractionwas the greatest resulting in the highest RUE. WUE of Beijing andUrumqi was similar due to similar solar radiation and diffuse lightfraction at these locations.
WUE of Chengdu showed no significant change during the past42 years (Fig. 6), although grain yields and ET had declined by 32.4%and 28.2% respectively. WUE at Urumqi had significantly increased
by 0.09 kg ha−1 mm−1. Grain yields at Urumqi were increased by24.2% and ET had decreased by 6.2%. WUE at Beijing had increasedsignificantly during the past 48 years by 0.04 kg ha−1 mm−1 as EThad decreased by 18.3% during the same period.
X. Yang et al. / Agricultural and Forest Meteorology 182– 183 (2013) 342– 351 349
F ) (b) a( of ea
3w
gfarawpn
TC(
ig. 6. Simulated potential evapotranspiration (ETpot) (a), evapotranspiration (ET1961–2003) and Urumqi (�—) (1961–2009). Lines show the linear trend, equation
.5. Quantifying the impacts of solar radiation reduction onheat yield and ET
Table 9 shows the correlation studies between simulated wheatrowth duration, yield, ET, WUE and solar radiation (Ra), diffuseraction (DF), maximum temperature (Tmax), minimum temper-ture (Tmin) and rainfall at Beijing, Chengdu and Urumqi. Theelationship between Ra and other climate elements were differentt three sites. Ra was negatively related with DF at three sites. Ra
as negatively related with Tmax and Tmin at Beijing, while Ra wasositively related with Tmax at Chengdu and Urumqi. Rainfall wasegatively related with Ra and positively related with DF at Beijing
able 9orrelations between simulated wheat growth duration (Duration), yield, evapotranspirDF), maximum temperature (Tmax), minimum temperature (Tmin), rainfall in wheat seaso
and Urumqi, while rainfall had no relationship with other climatevariables at Chengdu.
The wheat growing duration was negatively correlated toTmax and Tmin at three study sites, and the absolute correlationcoefficients were greater than other climate variables. Wheat grow-ing duration was mainly determined by temperature.
Simulated grain yields at Beijing had no significant correlationwith any of the four climate drivers. Grain yields at Chengdu cor-related positively with Ra, and negatively with DF and Tmin. The
simulated grain yields at Urumqi correlated positively with DF,Tmin and rainfall. So the declining Ra and increasing DF had differ-ent effect on potential grain yield under different Ra, temperature
ation (ET) and water use efficiency (WUE) and solar radiation (Ra), diffuse fractionn at Beijing, Chengdu and Urumqi.
350 X. Yang et al. / Agricultural and Forest Meteorology 182– 183 (2013) 342– 351
Table 10Linear regression model results for simulated yield and ET at Beijing, Chengdu and Urumqi.
Dependent (y) Study sites Independent (x) Linear regression model r2
Yield (t ha−1) Beijing – – –Chengdu Solar radiation (MJ m−2) y = 0.709x − 0.25 0.736Urumqi Rainfall (mm) y = 0.006x + 5.15 0.312
ET (mm) Beijing Solar radiation (MJ m−2) y = 36.7x − 17.50 0.723iationiation
ajelt
wnaap
fatya
t7Ud9
4
do(rf2wCsttrs
m1otewyprbutds
Chengdu Solar radUrumqi Solar rad
nd rainfall levels. Declining Ra under the radiation level of Bei-ing (13.25 MJ m−2) and Urumqi (12.34 MJ m−2) had no significantffect on grain yield, while the declining Ra under the radiationevel of Chengdu (7.85 MJ m−2) reduced the grain yield even withhe increase of DF.
Simulated ET at all three locations was positively correlatedith Ra and negatively correlated with Tmin. WUE at Beijing wasegatively correlated with Ra and positively correlated with Tminnd rainfall. WUE of Chengdu was positively correlated with Rand negatively correlated with temperature. WUE of Urumqi wasositively correlated with DF, Tmin and rainfall.
The linear regression model results by using stepwise methodor simulated yield, ET and climate variables at Beijing, Chengdund Urumqi are shown in Table 10. The variation of solar radia-ion explained 73.6% of changes in grain yields at Chengdu. Grainield changes at Urumqi were mainly determined by rainfall, with
determination coefficient of 0.312.Changes in simulated ET were mainly determined by solar radia-
ion at all three locations. The variation of solar radiation explained2.3, 75.9 and 13.3% of changes of ET at Beijing, Chengdu andrumqi respectively. When solar radiation declined by 1 MJ m−2
uring the wheat growing seasons, ET decreased by 36.4, 28.3 and.7 mm at Beijing, Chengdu and Urumqi, respectively.
. Discussion
Significant declining trends of solar radiation were observeduring the past decades at Beijing, Chengdu and Urumqi as a resultf increased aerosol loading caused by anthropogenic activitiesChe et al., 2005; Liang and Xia, 2005). The declining trend in solaradiation at Beijing and Urumqi was −0.44% a−1 and −0.23% a−1
rom 1961 to 2009, at Chengdu is was −0.62% a−1 from 1961 to003. These trends are similar to reports by Liang and Xia (2005)ho showed trends of −0.52, −0.66 and −0.33% a−1 at Beijing,hengdu and Urumqi respectively for 1961–2000. All three stationshowed a warming trend mainly due to an increase in minimumemperature during the past years, which corresponds well withhe global warming trend from 1956 to 2005 (IPCC, 2007). Seasonalainfall at Beijing and Urumqi had significantly increased during theame period but had not changed at Chengdu.
Lobell and Field (2007) had shown that crop yields (wheat, rice,aize, barley, soybean, sorghum) decline globally by 0.6–8.9% per◦C rise in temperature. Challinor et al. (2003) showed that 50%f the variability in groundnut yield in India can be explained byhe variability in total seasonal rainfall from 1966 to 1995. Thempirical results of Zhang et al. (2010) indicated that rice yieldsere positively correlated to solar radiation, which determined
ield variation at 20 experimental stations in China during a studyeriods of 14–25 years. Swain et al. (2007) indicated a reduction inice grain yield by 12% with reductions in incident solar radiationy 30% by using historical weather data (1983–2002) in India. Sim-
lated results of our study at Beijing, Chengdu and Urumqi showedhat the effect of climate drivers on grain yield were different underifferent climate environments. Simulated grain yields at Beijinghowed no significant trend during the past 48 years. Chen et al.
(MJ m−2) y = 28.6x + 42.54 0.759 (MJ m−2) y = 9.7x + 355.5 0.133
(2009) reported that simulated potential yield of wheat was sig-nificantly decreased due to the declining trend in solar radiationat Beijing from 1961–2003 when using APSIM model version 5.3,which employs a constant for RUE and ignores diffused light as acomponent of light. Several studies have shown the importanceof light quality, especially diffuse radiation, for estimating RUE(Choudhury, 2001; Gu et al., 2002; Roderick et al., 2001; Rodriguezand Sadras, 2007). Our simulated results showed that grain yieldsat Chengdu decreased significantly by 32% during the past 42 years.This decrease was mainly due to the decline in solar radiation,which explained 74% of the change in grain yields. Grain yields atUrumqi increased by 24% during past 48 years mainly due to theincrease in rainfall.
The simulated results were modelled on a daily base by theAPSIM-Nwheat model. Solar radiation reduction could influencewheat growth on an hourly base in reality, and a shorter time-stepmight be necessary to further improve such simulations. However,this would require hourly measured solar radiation data which areoften not available.
In contrast to Lobell and Field (2007) our results showed that therelationship between wheat yield and minimum temperature wasnegative at Chengdu (mean wheat season minimum temperatureof 7.8 ◦C) and positive at Urumqi (mean wheat season minimumtemperature of −1.8 ◦C).
Some studies have evaluated the impact of climate trendon evapotranspiration (ET). For example, Roderick and Farquhar(2002) showed that the decrease in evaporation is consistent withdecreases in sunlight over the past 50 years. Xu et al. (2006) sug-gested that decreasing wind speed was responsible for the changesof reference evapotranspiration in the Yangtze River basin of China.Tao et al. (2008) reported that ET and irrigation water use woulddecrease due to a shorter growing season as a result of increasingtemperature in China. Supit et al. (2010) indicated that decliningwater requirement of wheat crops can be attributed to a lowerevaporative demand as a result of declining global solar radiation.Our simulated ET at Beijing, Chengdu and Urumqi was signifi-cantly decreased due to declining solar radiation and increasingminimum temperatures, while the impact of changing wind wasnot investigated due to a lack of wind speed measurements. 72,76 and 13% of the decline in ET at Beijing, Chengdu and Urumqi,respectively was due to the observed decline in solar radiation.Rodriguez and Sadras (2007) found that WUE during anthesis ofwheat increased from North to South (vapor pressure deficit andtemperature decreased and diffuse fraction increased from 0.41 to0.61) in eastern Australia. Tong et al. (2009) found that WUE wasgreater on cloudy days than on the sunny days in the North ChinaPlain due to an increased fraction of defuse light. This agrees withthe simulated increase in WUE during past 48 years at Beijing andUrumqi with decreasing solar radiation and increasing diffuse radi-ation. The WUE at Chengdu was the greatest among these threestations as Chengdu had the least solar radiation and the greatest
diffuse radiation.
Several studies indicated that global dimming would result incrop yields declining (Stanhill and Cohen, 2001; Swain et al., 2007),especially in China where atmospheric aerosols have increased
st Met
dtt7msfirtrti
5
wceciCsTtyrr
A
dP
R
A
A
A
A
C
C
C
C
C
C
G
I
X. Yang et al. / Agricultural and Fore
ue to industrialization (Zhang et al., 2010). Our studies revealedhat wheat yield significantly decreased only when solar radia-ion levels were already low (mean wheat season solar radiation of.8 MJ m−2). It is notable that the solar radiation dimming effect isore obvious around big cities where meteorological stations with
olar radiation observation are located. This study has just identi-ed a ‘hot-spot’ for a dimming effect on wheat yield, although theegional picture of the effect in the Southwest is still unclear dueo limited stations of solar radiation observations. The simulatedesults reflected climate change effects on potential crop produc-ion during past years, while observed yield trends also include anncrease due to improved variety and cultivation during past years.
. Conclusions
The impact of changes in solar radiation and temperature onheat yields is different depending on climate conditions. In dry
limates with high solar radiation, decreases in solar radiation,specially if accompanied by an increase in the diffuse radiationomponent, will either have less significant impact on productiv-ty (Beijing), or can even increase yields (Urumqi) (Stanhill andohen, 2001). In wet climates with low solar radiation, a decrease inolar radiation significantly reduces wheat productivity (Chengdu).he increase of minimum temperature resulted in wheat yieldso decrease at subtropical climate (Chengdu) and improved wheatield at a temperate continental climate (Urumqi). The simulatedesults indicated that crop production in low solar radiation envi-onments are particularly affected by global diming.
cknowledgements
This work was supported by the National Natural Science Foun-ation of China (No. 41171086) and the National Basic Researchrogramme of China (973 Programme) (No. 2010CB428404).
eferences
sseng, S., Fillery, I.R.P., Anderson, G.C., Dolling, P.J., Dunin, F.X., Keating, B.A., 1998a.Use of the APSIM wheat model to predict yield, drainage, and NO3− leaching fora deep sand. Aust. J. Agr. Res. 49, 363–377.
sseng, S., Keating, B., Fillery, I., Gregory, P., Bowden, J., Turner, N., Palta, J., Abrecht,D., 1998b. Performance of the APSIM-wheat model in Western Australia. FieldCrops Res. 57, 163–179.
sseng, S., Jamieson, P., Kimball, B., Pinter, P., Sayre, K., Bowden, J., Howden, S., 2004.Simulated wheat growth affected by rising temperature, increased water deficitand elevated atmospheric CO2. Field Crops Res. 85, 85–102.
sseng, S., Turner, N., Keating, B., 2001. Analysis of water-and nitrogen-use efficiencyof wheat in a Mediterranean climate. Plant Soil 233, 127–143.
hallinor, A., Slingo, J., Wheeler, T., Craufurd, P., Grimes, D., 2003. Toward a combinedseasonal weather and crop productivity forecasting system: determination ofthe working spatial scale. J. Appl. Meteorol. 42, 175–192.
hameides, W.L., Yu, H., Liu, S.C., Bergin, M., Zhou, X., Mearns, L., Wang, G., Kiang,C.S., Saylor, R.D., Luo, C., 1999. Case study of the effects of atmospheric aerosolsand regional haze on agriculture: an opportunity to enhance crop yields in Chinathrough emission controls? PNAS 96, 13626–13633.
he, H., Shi, G., Zhang, X., Zhao, J., Li, Y., 2007. Analysis of sky conditions using 40year records of solar radiation data in China. Theor. Appl. Climatol. 89, 83–94.
he, H.Z., Shi, G.Y., Zhang, X.Y., Arimoto, R., Zhao, J.Q., Xu, L., Wang, B., Chen, Z.H.,2005. Analysis of 40 years of solar radiation data from China, 1961–2000. Geo-phys. Res. Lett., 32.
hen, C., Wang, E., Yu, Q., Zhang, Y., 2009. Quantifying the effects of climate trends inthe past 43 years (1961–2003) on crop growth and water demand in the NorthChina Plain. Clim. Change 100, 559–578.
houdhury, B., 2001. Modeling radiation-and carbon-use efficiencies of maize,sorghum, and rice. Agric. For. Meteorol. 106, 317–330.
u, L., Baldocchi, D.D., Verma, S.B., Black, T.A., Vesala, T., Falge, E.M., Dowty, P.R.,2002. Advantages of diffuse radiation for terrestrial ecosystem productivity. J.Geophys. Res. 107 (D6), http://dx.doi.org/10.1029/2001JD00124.
PCC, 2007. In: Solomon, S., Qin, D., Manning, M., Chen, Z., Marquis, M., Averyt, K.B.,Tignor, M., Miller, H.L. (Eds.), Climate Change 2007: The Physical Science Basis:
eorology 182– 183 (2013) 342– 351 351
Working Group I Contribution to the Fourth Assessment Report of the IPCC.Cambridge University Press, Cambridge, p. 18.
Jones, C., Kiniry, J., 1986. CERES-Maize: A Stimulation Model of Maize Growth andDevelopment. Texas A&M Univ. Press, College Station, TX, pp. 194.
Kaiser, D., Qian, Y., 2002. Decreasing trends in sunshine duration over China for1954–1998: indication of increased haze pollution? Geophys. Res. Lett. 29, 2042.
Keating, B., Carberry, P., Hammer, G., Probert, M., Robertson, M., Holzworth, D., Huth,N., Hargreaves, J., Meinke, H., Hochman, Z., 2003. An overview of APSIM, a modeldesigned for farming systems simulation. Eur. J. Agron. 18, 267–288.
Keating, B.A., Meinke, H., Probert, M.E., Huth, N.I., Hills, I.G., 2001. NWheat: Doc-umentation and Performance of a Wheat Module for APSIM. Tropical Agr.Technical Memo.
Lawes, R.A., Oliver, Y.M., Robertson, M.J., 2009. Integrating the effects of climate andplant available soil water holding capacity on wheat yield. Field Crop. Res. 113,297–305.
Liang, F., Xia, X., 2005. Long-term trends in solar radiation and the associated climaticfactors over China for 1961–2000. Ann. Geophys.-Germany 23, 2425–2432.
Littleboy, M., Silburn, D., Freebairn, D., Woodruff, D., Hammer, G., Leslie, J., 1992.Impact of soil erosion on production in cropping systems. I. Development andvalidation of a simulation model. Soil Res. 30, 757–774.
Lobell, D.B., Field, C.B., 2007. Global scale climate-crop yield relationships and theimpacts of recent warming. Environ. Res. Lett. 2, 014002.
Lobell, D.B., Ortiz-Monasterio, J.I., 2007. Impacts of day versus night temperatureson spring wheat yields: a comparison of empirical and CERES model predictionsin three locations. Agron. J. 99, 469–477.
Monteith, J., 1988. Does transpiration limit the growth of vegetation or vice versa?J. Hydrol. 100, 57–68.
Probert, M., Dimes, J., Keating, B., Dalal, R., Strong, W., 1998. APSIM’s water andnitrogen modules and simulation of the dynamics of water and nitrogen in fallowsystems. Agr. Syst. 56, 1–28.
Probert, M.E., Keating, B.A., Thompson, J.P., Parton, W.J., 1995. Modelling water,nitrogen, and crop yield for a long-term fallow management experiment. Aust.J. Exp. Agric. 35, 941–950.
Qian, Y., Kaiser, D., Leung, L., Xu, M., 2006. More frequent cloud-free sky and less sur-face solar radiation in China from 1955 to 2000. Geophys. Res. Lett. 33, L01812.
Qian, Y., Wang, W., Leung, L., Kaiser, D., 2007. Variability of solar radiation undercloud-free skies in China: the role of aerosols. Geophys. Res. Lett. 34, L12804.
Ritchie, J., Godwin, D., Otter-Nacke, S., 1985. CERES-Wheat: A User-oriented WheatYield Model. Preliminary Documentation. AGRISTARS Publication No. YM-U3-04442-JSC-18892. Michigan State University, East Lansing, USA.
Roderick, M.L., Farquhar, G.D., 2002. The cause of decreased panevaporation over the past 50 years. Science 298, 1410–1411,http://dx.doi.org/10.1126/science.1075390-a.
Roderick, M., Farquhar, G., Berry, S., Noble, I., 2001. On the direct effect of clouds andatmospheric particles on the productivity and structure of vegetation. Oecologia129, 21–30.
Rodriguez, D., Sadras, V., 2007. The limit to wheat water-use efficiency in easternAustralia. I. Gradients in the radiation environment and atmospheric demand.Aust. J. Agric. Res. 58, 287–302.
Stanhill, G., Cohen, S., 2001. Global dimming: a review of the evidence for awidespread and significant reduction in global radiation with discussion of itsprobable causes and possible agricultural consequences. Agric. For. Meteorol.107, 255–278.
Spiertz, J.H.J., Van de Haar, H., 1978. Differences in grain growth, crop photosynthesisand distribution of assimilates between a semi-dwarf and a standard cultivar ofwinter wheat. Neth. J. Agr. Sci. 26, 233–249.
Supit, I., van Diepen, C., Boogaard, H., Ludwig, F., Baruth, B., 2010. Trend analysis ofthe water requirements, consumption and deficit of field crops in Europe. Agric.For. Meteorol. 150, 77–88.
Swain, D.K., Herath, S., Saha, S., Dash, R.N., 2007. CERES-Rice model: calibration, eval-uation and application for solar radiation stress assessment on rice production.J. Agrometeorol. 9, 138–148.
Tao, F., Hayashi, Y., Zhang, Z., Sakamoto, T., Yokozawa, M., 2008. Global warming,rice production, and water use in China: developing a probabilistic assessment.Agric. For. Meteorol. 148, 94–110.
Tong, X.J., Li, J., Yu, Q., Qin, Z., 2009. Ecosystem water use efficiency in an irrigatedcropland in the North China Plain. J. Hydrol. 374, 329–337.
Wild, M., 2009. Global dimming and brightening: a review. J. Geophys. Res. 114,D00D16, http://dx.doi.org/10.1029/2008JD011470.
Wild, M., Gilgen, H., Roesch, A., Ohmura, A., Long, C.N., Dutton, E.G., Forgan,B., Kallis, A., Russak, V., Tsvetkov, A., 2005. From dimming to brightening:decadal changes in solar radiation at earth’s surface. Science 308, 847–850,http://dx.doi.org/10.1126/science.1103215.
Xu, C., Gong, L., Jiang, T., Chen, D., Singh, V., 2006. Analysis of spatial distributionand temporal trend of reference evapotranspiration and pan evaporation inChangjiang (Yangtze River) catchment. J. Hydrol. 327, 81–93.
Yu, Q., Liu, Y.F., Liu, J.D., Wang, T.D., 2002. Simulation of leaf photosynthesis of winterwheat on Tibet plateau and in North China plain. Ecol. Model. 155, 205–216.
Zhang, T., Zhu, J., Wassmann, R., 2010. Responses of rice yields to recent climatechange in China: an empirical assessment based on long-term observations atdifferent spatial scales (1981–2005). Agric. For. Meteorol. 150, 1128–1137.
