How sensitive is SEBAL to changes in input variables, domain size and satellite sensor?

摘要: [1] Estimation of evapotranspiration (ET) over large heterogeneous areas using numerous satellite-based algorithms is increasing; however, further analysis uncertainties limited. The objective this study was to evaluate impacts varying input variables, size the modeling domain, and spatial resolution satellite sensors on sensible heat flux (H) estimates from Surface Energy Balance Algorithm for Land (SEBAL). First, sensitivity SEBAL conducted by its variables Moderate Resolution Imaging Spectroradiometer (MODIS) data 29 cloud-free days in 2007 covering Baiyangdian watershed North China. Domain dependence H quantified estimating subwatersheds different sizes entire MODIS 4 May 2007. Landsat Thematic Mapper (TM) based are compared effect sensors. Results indicate that most sensitive temperatures hot cold pixels available energy pixel. domain show mean absolute percentage difference (MAPD) root square deviation (RMSD) between up 53.9% 75.7 W m−2, respectively. Although areally averaged TM similar, MODIS-based an RMSD 52.3 m−2 a bias 26.5 relative TM-based counterparts. Unlike other models, standard high images can be smaller than low images. Furthermore, upscaling scheme (aggregating variables) generally consistent with those output output) same sensor, given similar differences resolution. resulting ET therefore vary differing extreme selected operator, size, This provides insights into various factors should considered when applying estimate helps correctly interpret outputs.