Mechanism Of Reference Dry And Wet Limits In Remote Sensing Evapotranspiration Model: Study And Application

Estimating land surface evapotranspiration (ET) using remote sensing is recently aresearch focus in the study of global climate change and water cycle. Based on thereview of international remotely sensed ET models and their applications, it was foundthat ET estimation seemed to be highly un-robust due to the spatial ambiguity issueexisting in common remotely sensed ET models. The concepts of reference dry and wetlimits were proposed in this study to solve the spatial ambiguity issue and, thus, toimprove ET estimation. It is hoped that the present study could give its contributions tothe study of water cycle in regional scale.The study proposed the concepts and the assumptions of reference dry and wetlimits to solve ambiguity issue. The implicit solution of the radiometric temperature ofreference dry and wet limits was abtained according to some important assumptions.Furthermore, the explicit solution was derived through Taylor expansion to make thesolution easier to be understood. This study incorporated numerical simulations andobservations in field scale to verify the assumptions of available energy allocation andthe characteristics of some key parameters of reference dry and wet limits.Since the boundaries of conventional Ts/VI models were hard to be determined,the study redetermined the boundaries of Ts/VI feature space based on the results ofreference dry and wet limits. Then two inkinds of interpolation methods were used toestimate the actual evapotranspiration. The in-situ data from two flux stations were usedto test the reliabilities of the re-determined Ts/VI feature space. The results indicatedthat the ET models based on the re-determined Ts/VI feature space were acceptablecompared with other researchers’ studies.To reduce the uncertainties of selecting dry and wet pixels in SEBAL, REDRAWwas proposed by using the new trapezoid framework determined by reference dry andwet limits. The important assumption that there was a linearity relationship betweennear-surface temperature gradient and radiometric temperature was applied to the rangesof different vegetation indexes. Thus REDRAW could consider the impact of vegetationindex on the partition of available energy more reasonable. The validation suggestedthat REDRAW had better performance than SEBAL in relatively arid and humid areas.Considering that conventional methods in downscaling surface temperature data might fail in the condition of low vegetation coverage, a trapezoid method was putforward to avoid this problem. The experiments of downscaling surface temperature intwo areas indicated the validity of the trapezoid method in different conditions ofvegetation coverage.Finally, REDRAW was applied in the Hekou-Longmen section of Yellow RiverBasin using10years MODIS data. Observed ET from water balance analysis in basinscale showed that REDRAW could provide relatively better ET estimations thanSEBAL. The analysis of the response to the precipitation events also indicated thatREDRAW could better describe the effects of the precipitation on the actualevapotranspiration.