Estimation Of Evapotranspiration Under All-sky Condition Over The Tibetan Plateau Using Satellite Observations

The Tibetan Plateau(TP),with unique dynamic and thermal effects,has a great influence on the weather and climate in China,Asia and even the globe.Therefore,it is important to study and quantitatively understand the energy and water cycle over the TP.Evapotranspiration(ET)(the sum of surface evaporation and vegetation transpiration)is not only a crucial part of the water cycle in the land-atmosphere system but also a major component of the surface energy balance.Furthermore,ET is constrained by many controlling factors and consequently varies with the environment,making it one of the most uncertain parts of the land surface process.Compared with discrete ground-based observation methods,satellite remote sensing enables spatially continuous estimation of land-atmosphere system at a large-scale.However,the observations of land surface key parameters from the optical sensors are significantly influenced by clouds,and the applicability of the estimated ET within the study area still needs to be validated.Although some studies have been carried out,there is still a lack of systematic understanding of ET over the TP because of the harsh environment and the difficulty of observation.Up to now,the existing ET products are only representative for clear-sky conditions while their accuracies are insufficient for practical applications.In this study,the satellite observations from optical and microwave sensors were combined to estimate ET under all-sky conditions over the TP in this study.Firstly,ET estimation schemes were established by using the multi-layer air temperature,relative humidity,net radiation flux,wind speed,precipitation,and soil moisture observations from the Coordinated Enhanced Observing Period(CEOP)Asia-Australia Monsoon Project(CAMP)over the Tibetan Plateau(CAMP/Tibet).These schemes can provide more ground-based information for validation of satellite remote sensing ET.In this study,we also found that the dominant factors influencing the ET over the northern TP are energy-related parameters(net radiation and air temperature).Secondly,based on the HANTS algorithm and the narrowband to broadband conversion formula,time series of normalized difference vegetation index(NDVI)and land surface emissivity were reconstructed to remove the cloud effect.Improved broadband albedo,radiation flux and reconstructed NDVI were coupled into two different models,and their ET estimations were compared with each other.The results showed that the TESEBS model performed better over the TP.Thirdly,land surface temperature(LST)and downwelling radiation are two vital input parameters of ET models and their detections are highly influenced by clouds.Thus,the corresponding retrieval algorithms need to be developed.Considering the apparent albedo as an integrated factor to describe the effect of clouds on radiation,an improved Heliosat algorithm was proposed to determine all-sky surface downwelling radiation fluxes based on polar-orbiting satellite.Additionally,the Penetration Depth(PD)was found to be an ideal integrated factor to describe the attenuation of soil heat transfer,and a PD-based microwave LST retrieval model was developed accordingly.Finally,all-sky ET over the TP was estimated and its spatio-temporal distribution characteristics were analyzed.Based on above results,a remote sensing-based ET model was established.The model includes four modules:cloud removal of low-frequency land surface parameters,retrieval of the land surface temperature,estimation of the downwelling radiation fluxes,and the all-sky ET estimation.Combining multi-source satellite data(MODIS,VGT,OMI,CERES,AMSR-E)and meteorological forcing data,the cloud effects were removed.On this basis,the spatio-temporal characteristics of the all-sky ET were further analyzed.The results show that the estimated ET was in good agreement with the actual ET with correlation coefficient and mean bias of 0.79 and 0.06 mm/h.The temporal variation of ET over the TP showed a"single-peak" shape with winter-low and summer-high pattern while the spatial distribution showed a northwest-low and southeast-high pattern.Additionally,there was a low-value and a high-value center,located in the Tarim Basin and the water vapor transport belt,respectively.