Estimation Of Evapotranspiration Based On SWH Model In The Loess Plateau In Northern Shaanxi

Evapotranspiration(ET)is an important process of the water cycle and energy balance in the ecosystem,which determines the water and heat transfer in the Soil-Plant-Atmosphere system and is closely related to ecosystem productivity.Accurate quantification of regional daily evapotranspiration is of great significance for the study of regional water cycle and use of vegetation in different growth stages and for the guidance of agricultural water management.Nowadays,the combination of hydrological models and remote sensing technology is considered to be an advanced means for estimating regional ET and hydrological processes.However,due to the limitations of cloud cover,seasonal snow cover,sensors and other factors,remote sensing products are difficult to meet the requirements of high temporal resolution and high spatial resolution at the same time,which limits the simulation accuracy and temporal resolution of ET simulated by remote sensing evapotranspiration model.In this study,SWH-m TSF and SWH-METRIC daily evapotranspiration models were constructed based on SWH model with source data,including MODIS remote sensing data,meteorological data and eddy covariance measurements.The models established improved the quantization accuracy and temporal resolution of ET,and simulated ET with a temporal resolution of 1 day and a spatial resolution of 1 km.Meanwhile,the temporal and spatial evolution characteristics of ET in the Loess Plateau in Northern Shaanxi(LPNS)were analyzed in this paper.The main conclusions of the study are as follows:(1)SWH dual-source evapotranspiration model can accurately simulate the ET with a temporal resolution of 8 days in the LPNS.In general,the trend of ET within the year simulated by the SWH model was consistent with the eddy covariance measurements at Changwu station,and the simulation performance was good.The determination coefficient,root mean square error,mean absolute error and fitting index were 0.69,0.76 mm·d^-1,0.53mm·d^-1 and 0.87 respectively,indicating that SWH model was suitable for simulating ET in the Loess Plateau in Northern Shaanxi.(2)SWH-m TSF and SWH-METRIC daily evapotranspiration models were constructed based on the SWH model with meteorological raster data,MODIS data products and eddy covariance measurements as source data.Both models have strong theoretical basis.SWH-m TSF model focused on the high spatial and temporal resolution of the model's input parameters.The improved spatio-temporal filtering algorithm(m TSF)was used to reconstruct the continuous daily LAI/FPAR in the study area,and then the SWH model was used to simulate the continuous daily ET.The SWH-METRIC model,combining the SWH model with METRIC model combined with the climate and environment level of the study area,simulated continuous daily ET in the LPNS.(3)Based on the site measured data verification,the daily ET simulation performance and temporal-spatial resolution of SWH-METRIC model were significantly better than those of SWH-m TSF and SWH models.Compared with the eddy covariance measurements of Changwu station and SWH model,the SWH-m TSF model has a poor simulation performance on daily ET,with the determination coefficient of 0.44,the root mean square error and the mean absolute error increased to 0.86 mm·d^-1 and 0.62 mm·d^-1,and the fitting index was 0.80.SWH-METRIC model had the best simulation performance,the determination coefficient is up to 0.70,and the root mean square error and mean absolute error are reduced to 0.67 mm·d^-1and 0.48 mm·d^-1.Meanwhile,the research showed that the SWH-METRIC model after coupling METRIC model did not cause error accumulation,but improved the simulation accuracy,which was suitable for the simulation of continuous daily ET in the Loess Plateau in Northern Shaanxi.(4)SWH-METRIC model was applied to the ET simulation in the LPNS with a spatial resolution of 1 km and a temporal resolution of 1 day,revealing the spatial-temporal evolution characteristics of annual ET in the study area.The annual ET in the LPNS showed a fluctuating decreasing trend from 2002 to 2009.From the perspective of spatial distribution,the high evapotranspiration areas were mainly concentrated in the Luochuan County and the central part of the study area,while the areas of low evapotranspiration are mainly distributed in the surrounding areas of Luochuan County,including Huanglong County,Huangling County and Fu County.In terms of the temporal evolution of annual ET,only the western boundary of the study area showed an increasing trend,and the other regions showed a decreasing trend,among which the central region showed a significant decreasing trend(p<0.05),accounting for about 38%of the area of the Loess Plateau in northern Shaanxi.