Simulation Of Evapotranspiration In The Loess Plateau Based On The Google Earth Engine Platform And The SEBAL Model

The ecological environment of the Loess Plateau is fragile and water resources are extremely scarce.Accurate estimation of the evapotranspiration(ET)in the Loess Plateau will help rationally manage and allocate water resources in the Loess Plateau.The methods of ET estimation based on remote sensing had been widely used recently,but evaluations of the methods are still needed in different areas in local ET estimation.The SEBAL(Surface Energy Balance Algorithm for Land)model,as a high-precision and effective method for ET estimation at large scales based on remote sensing,has been widely used in various regions in the world.However,this model also has some limitations,such as the complex processes of data collection and calculation,slow calculation process caused by the limited computing capacity of local computers,and the consequent large consumption of manpower and time.In order to estimate ET in the Loess Plateau and reduce restricts in SEBAL model application,this study acquired high-precision remote-sensing image data of Landsat satellite from the Google Earth Engine(GEE)and constructed SEBAL model on this platform.Then,we took Ansai and Changwu,two counties of Shaanxi Province in the Loess Plateau,as the study regions to verify the applicability of the SEBAL model that was drivem by satellite-remote-sensing data and the MOD 16 product that was a medium resolution imaging spectroradiometer ET product based on eddy covariance observation data in the Loess Plateau.Furthermore,we analyzed the sensitivity of the SEBAL model to the“hot”point's temperature,the energy balance components of inversion,the inversion results of main surface parameters,spatial distributions of daily and monthly ET,and daily ET and its dynamics at different land use conditions at Ansai and Changwu.Finally,the correlation between daily ET and surface temperature was analyzed through linear regression.Some main conclusions were drawn as follows.(1)In the Loess Plateau,the correlation between temperature and daily surface ET was relatively large,followed by sunshine hours.The correlation between relative humidity,wind speed and daily surface ET was relatively low.The meteorological factor of T_mean had the strongest correlation with the ET_a in Ansai(r=0.49),while T_max had the strongest correlation with the ET_a of Changwu(r=0.71).The correlation coefficients between ET_a and T_mean,T_min,T_max(0.68,0.59,and 0.71)at Changwu were all higher than those between ET_a and T_mean,T_min,T_max(0.49,0.47,and 0.46)at Ansai.The correlation between daily ET and temperature at Changwu was greater than that at Ansai.(2)The SEBAL model showed high sensitivity to“hot”spots both at Ansai and Changwu,but this high sensitivity had little influence on the ET estimated based on remote sensing.(3)The applicability of the MOD 16 model was lower than the SEBAL model.Compared with the eddy covariance observation data of the two sites,the RMSE(root mean square error)of the SEBAL model was 0.52 at Ansai and 0.37 at Changwu,respectively.The SEBAL model had high applicability in the Loess Plateau.(4)Daily ET of water bodies,arable land,and forest areas showed an overall increasing trend from April to July and an overall decreasing trend from August to November.There was a significant negative correlation between surface temperature and ET at Ansai(R²=0.94)and at Changwu(R²=0.95),respectively.(5)The results of this study showed that the combination of the SEBAL model and the GEE platform can quickly and accurately calculate ET in the Loess Plateau,which could facilitate the scientific management of water use in the Loess Plateau.