LINEAR MULTI-REGRESION MODEL FOR ESTIMATING ACTUAL DAILY EVAPOTRANSPIRATION BASED ON S-SEBI METHOD OVER CAU RIVER BASIN OF VIETNAM USING LANDSAT-7 IMAGES

Calculating the actual evapotranspiration (ETa) is an important task in monitoring water balance in river basins, serving irrigation management in agriculture and in weather forecasting. In fact, it is difficult to measure ETa directly and most ETa applications are calculated using empirical-theoretical models. This paper proposes a linear multi-regression model to calculate ETa based on the relationship between ETa extracted by the Simple Surface Energy Balance Index (S-SEBI) with the surface biological-physical parameters, such as albedo, normal difference vegetation index (NDVI), surface temperature (Ts), and net surface solar radiation (Rnd) that were extracted from the Landsat-7 imagery dated November 23, 2001. The proposed model was evaluated based on ETa field observations on meteorological stations on November 4th, 2000 with Root Mean Square Error (RMSE) of 1.585 mm/day and Mean Absolute Error (RAE) of 35.44%. After eliminating the Mean Bias Error (MBE) the model achieves accuracy with RMSE = 0.702 mm/day and MAE = 17% (model accuracy is 83%).