Estimation Of Regional Evapotranspiration Using Advanced SEBAL Model

As a very important part of the water circulation, Evapotranspiration (ET) plays a key role in the process of atmosphere, hydrosphere and biosphere.Estimating it correctly is very important for understanding large-scale energy and water balance. Among the several ET estimation methods, the remote sensing method is regarded as the only technology that can efficiently and economically provide regional and global coverage of actual water consumption.This study refines the original SEBAL algorithm, using the high resolution remote image Landsat TM, and estimates the day ET of Zuomaoxikong watershed in the source region of Yangtze river. Also fits the relation between some surface parameters, terrain factors and day ET ,and analyses the sensitivity of day ET to some parameters which either calculated by ground data or by experimental equations.In the original SEBAL model, the air transmittance was determined by a trial and error method using calculated albedo as an index. For this study, an elevation based transmittance equation was considered. The original SEBAL was limited to flat surfaces. A method of extending it to sloped surfaces, by using DEM information was developed.After validated the estimation results by some statistical information and Lysimeter measured ET, we consider that the developed SEBAL model can operate correctly in the study area and the ET results can imply the heat and vapor spatial pattern of this area accurately.The main conclusions of this dissertation are as follow:The average values of every portion of regional energy balance are: 673.6 W/m~2 for net radiation; 80.5 W/m~2 for soil heat flux; 167.6 W/m~2 for sensible heat flux; and 425.5 W/m~2 for the latent heat flux. The regional average value of Evapotranspiration Fraction reaches to 0.715, which is high enough to imply that the net radiation in the study area is mainly redistributed in latent heat flux at that certain time.Among those several key parameters, four of them have positive correlation with ET. They are surface emissivity, vegetation fraction, NDVI and surface roughness in sequence. Three parameters have negative correlations with ET, which are surface temperature, shortwave transmittance, and surface albedo.This study analyzed the sensitivity of ET to four parameters which are surface roughness, wind speed at 200 meters, surface emissivity and shortwave transmittance. For the change of the former two parameters (surface roughness and wind speed), ET almost has no sensitivity, which indicate that certain trial error of these two parameters will not affect the result of ET. But different phenomena appeared about the other two parameters, 3% change in emissivity and shortwave transmittance can cause 10% and 4.4% change in ET respectively. This means that we should be careful when we choose the equations for the latter two parameters.Based on our research, we should realize that an operational ET estimation model, which is physical-based and also resolve the scale-up problem, is very needed in the future.