A comparative assessment of SWAT-model-based evapotranspiration against regional-scale estimates

We attempted to assess the actual evapotranspiration (ET_a) rates estimated by the Soil and Water Assessment Tool (SWAT) model in an arid region by comparing other regional-based ET_a estimates: the Operational Simplified Surface Energy Balance (SSEBop) products, the MODerate Resolution Imaging Spectroradiometer (MODIS) Global Evapotranspiration product (MOD16), and the Generalized Complementary Relationship (GCR) model. The results by the SWAT model clearly showed that ET_a rates in the SWAT model are limited by soil water availability. From the comparison of the annual ET_a estimates, the GCR model among those ET_a estimation methods simulated slightly higher values than the water-balance derived ET_a rates (i.e., the difference between annual precipitation and annual stream flow rates). The annual ET_a rates of the SSEBop products showed the closest slope of the regression line to 1.0, while the highest R² (coefficient of determination) value was found from the annual ET_a estimates by the SWAT model. The monthly ET_a by the GCR model had the highest linear relationship with those of the SSEBop products. However, the SWAT model did not have a strong linear relationship with the monthly ET_a of the SSEBop products unlike the comparison against the MOD16 products. This study recommends a further study on more reliable ET_a estimates in arid or semi-arid regions by the SWAT model.