The Robust Analysis Of Remote Sensing-based Precipitation Products And Evapotranspiration Model

Precipitation and evapotranspiration(ET)are key components of the hydrological cycle.Traditional observation techniques of precipitation and ET are limited to point scale.Remote sensing technique is able to obtain important information of surface parameters over space and time continuously and dynamically,providing a means for obtaining spatial-distributed precipitation and evapotranspiration(ET)on large scales.Due to the indirect measurement or the empirical assumptions and parameterizations in modeling,the accuracy of areal precipitation and ET results based on remote sensing always attracts extensive attention from the researchers.The potential uncertainty or error impedes the application of remote sensing-based precipitation and ET on related fields.Aiming at soloving this problem,this study mainly focused on the evaluation and calibration of the satellite rainfall products,and the robustness of satellite-based ET models,in order to derive more reliable precipitation and ET data on large scales.The mechanism and potential problems of the existing satellite rainfall evaluation methods were summarized firstly,based on which the consistency rule for precipitation and the integrated evaluation method for satellite rainfall were proposed,realizing the evaluation of satellite rainfall within grids without gauges for the first time.Secondly,the relation between rainfall and related factors such as location and topography information was deeply explored through data mining technique,and the ensemble calibrating method for satellite rainfall based on multi-factor data mining technique was put forwarded.The calibration provided an effective way to fix the problem that the exact relation between actual rainfall and its influencing factors is still unclear so far.Thirdly,the exciting resistance schemes for ET modeling were summarized based on sufficient literiture review.The potential schemes being suitable for each reference dry and wet limit were listed out,and their influence on the extreme limits calculation and the final flux estimation from ET model were analyzed.Finally,the more ideal algorithms for reference dry and wet limit were explored through three different ways in order to avoid some of the empirical parameterizations and unreasonable assumptions in current algorithm.These three methods were varying the dry and wet limit of the trapezoid framework,varying the vegetation height assumption for reference vegetated wet limit,and adopting two source resistance schemes,respectively.All the new algorithms for reference dry and wet limits were applied on remote sensing ET models.The effectivenesses of the new algorithms on the calculation of extreme limits,and the estimation of flux both on point and areal scales were analyzed and discussed detailedly.A series of methods including literature review,data mining,model optimizing,reference dry and wet limit algorithm,remote sensing-based ET model,were comprehensively applied in this study,which was benefical to the development of the evaluation/calibration of remote sensing-based rainfall,and the theorrtical derivation of extreme limits in remote sensing-based ET models.Some original concepts,such as consistency rule for precipitation and ideal reference dry and wet limit,were proposed for the first time and applied to improving the current methods,enriching the research area of remote sensing-based precipitation and ET.This study provides significant mechanism and technique spports for the obtaining of reliable precipitation and ET data with resonable temporal and spatial resolution based on remote sensing,which is invaluable contribution to hydrological cycle-related scientific research and water resources-related practical applications.