Comparison of Five Models to Scale Daily Evapotranspiration from One-Time-of-Day Measurements

Calculation of regional, spatially distributed evapotranspiration (ET) is possible using remotely sensed surface temperatures from sensors aboard air or space platforms. These platforms provide instantaneous data at frequencies of days to weeks, so that instantaneous latent heat flux can be computed from energy balance algorithms. However, instantaneous latent heat flux must be converted to ET and then scaled to daily (24 h) totals for most practical applications. We compared five scaling models where a single measurement of 0.5 h ET was used to estimate the daily total during clear days. Each model takes advantage of the clear day, quasi-sinusoidal nature of daytime ET and other daytime parameters including solar radiation, available energy, or reference ET. The surfaces were fully irrigated alfalfa, partially irrigated cotton, dryland grain sorghum, and bare soil (tilled fallow sorghum). Actual ET was measured by precision weighing lysimeters. Model agreement was evaluated on the basis the modified index of agreement (D) and the modified coefficient of efficiency ( epsilon ), in addition to standard statistical parameters. For cropped surfaces, the models based on grass reference ET resulted in the best agreement between observed and predicted daily ET totals. For bare soil, the model based on available energy (i.e., evaporative fraction) resulted in the best agreement. Relative error between observed and predicted daily ET increased as daily ET decreased. Observed and predicted daily ET agreed well for the transpiring crops (RMSE of 0.33 to 0.46 mm d super(-1) for mean daily ET of 3.9 to 5.8 mm d super(-1)) but poorly for bare soil (RMSE of 0.47 mm d super(-1) for mean daily ET of 1.4 mm d super(-1)).