Evaluation Framework of Landsat 8–Based Actual Evapotranspiration Estimates in Data-Sparse Catchment

Abstract

Remote sensing has revolutionized the assessment of evapotranspiration by continuous monitoring of the variable at a global scale. However, it is difficult to accurately estimate actual evapotranspiration (AET) in areas with high spatial heterogeneity along with very little ancillary data available. The presence of spatial heterogeneity is observed over India and being a country, whose economy largely depends on agriculture, estimation of AET is crucial for efficient water management. This study proposed a new framework to evaluate AET at a fine spatial scale suitable for such heterogeneous and data-sparse environments, with spatial and temporal validation using global datasets. Landsat 8 data were used to estimate AET for eight cloud-free days of 2014 over the Malaprabha River Basin in India using the modified surface energy balance algorithm for land (M-SEBAL) and two-source energy balance (TSEB) models. These two AET estimates were compared with Moderate resolution Imaging Spectroradiometer (MODIS) and Global Land Evaporation Amsterdam Model (GLEAM) AET for spatial and temporal validation respectively. M-SEBAL outperformed TSEB in capturing the magnitude and spatial variability of AET (e.g., spatial correlation between M-SEBAL and MODIS AET was 0.56 and 0.45 for day of year (DOYs) 040 and 136, respectively, whereas for the same DOYs the correlation between TSEB and MODIS AET was 0.16 and 0.36). The results demonstrate the challenge in AET estimation at a fine spatial resolution and highlight the importance of choosing a suitable algorithm.