A Simple Statistical Model for Estimating Evapotranspiration of Tallgrass Prairies in the Central United States Using MODIS Data

Abstract:

Accurate estimation of evapotranspiration (ET) across space and time is of great significance for quantifying global water cycle and improving land and water resources management. The Moderate Resolution Imaging Spectroradiometer (MODIS)-derived enhanced vegetation index (EVI) and ground-based climatic variables were integrated with eddy covariance tower-based ET (ET_EC) at six AmeriFlux tallgrass prairie sites in the Central United States to develop a simple and robust statistical model for predicting tallgrass prairie ET. The EVI tracked the seasonal variation of ET_EC well at both individual (R² > 0.70) and across six (R² = 0.76) sites, suggesting that the EVI could be used to estimate ET. The inclusion of photosynthetically active radiation (PAR) measured at tower sites further improved the ET-EVI relationship (R² = 0.86). Based on this result, we aggregated ET_EC, EVI, and PAR (MJ m^-2 d^-1) data from four sites (15 site-years) to develop a statistical model (ET = 0.11 R_g + 5.49 EVI – 1.43, adj. R² = 0.86, P < 0.0001) for predicting ET (mm d^-1) at 8-day intervals. This predictive model was evaluated against additional two years of ET_EC data from one of the four model development sites and two independent sites. The predicted ET (ET_EVI+PAR) captured the seasonal patterns and magnitudes of ET_EC, and correlated well with ET_EC, with R² of 0.87-0.96 and RMSE of 0.35-0.49 mm d^-1, and it was significantly improved compared to the standard MODIS ET product and ET estimates from two remote sensing ET models (Coupling Photosynthesis and Transpiration Model and Vegetation Transpiration Model). Overall, the results from this study demonstrated that tallgrass prairie ET can be accurately predicted using a multiple regression model that uses EVI and PAR which can be readily derived from remote sensing data.