Journal of Geophysical Research: Atmospheres

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

Authors

  • Di Long,

    1. Department of Biological and Agricultural Engineering, Texas A&M University, College Station, Texas, USA
    Search for more papers by this author
  • Vijay P. Singh,

    1. Department of Biological and Agricultural Engineering, Texas A&M University, College Station, Texas, USA
    2. Department of Civil and Environmental Engineering, Texas A&M University, College Station, Texas, USA
    Search for more papers by this author
  • Zhao-Liang Li

    1. LSIIT, UdS, CNRS, Illkirch, France
    2. State Key Laboratory of Resources and Environment Information System, Institute of Geographic Sciences and Natural Resources Research, Chinese Academy of Sciences, Beijing, China
    Search for more papers by this author

Abstract

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

Ancillary