Estimating large area wheat evapotranspiration from remote sensing data

Abstract

Presents an approach for evapotranspiration (ET) estimation using remote sensing data for wheat in a large irrigation district in northwest Mexico, during the whole growing season, based on the relationship between vegetation index and crop coefficient (Kc). The ratio of measured ET to potential ET (the so called crop coefficient, Kc) using four different models (Shuttleworth, Penman, Priestley-Taylor and Makkink) was compared to two vegetation indices (Normalized Difference Vegetation Index, NDVI, and Soil Adjusted Vegetation Index, SAVI). Continuous measurements were made of surface fluxes and other meteorological variables from January through April, following almost the entire vegetative cycle of the plant using a station equipped with standard meteorological instruments and an eddy-correlation system (Campbell Scientific International), placed in the northern part of an 80 ha field, since the prevailing winds were from the southwest. Weekly measurements of surface reflectance were made using a handheld radiometer with 5 spectral bands (Cropscan MSR5). Several Landsat 7, MODIS and vegetation images were acquired. A negative exponential model with three parameters was fitted to the vegetation index measurements. The fitted model was later used to estimate the crop coefficient as a function of the vegetation index for each day of the vegetative cycle. Actual evapotranspiration was then computed as the product of the estimated crop coefficients and potential evapotranspiration. RMSE were on the order of 30 W m/sup -2/ day/sup -1/. Crop coefficient and actual evapotranspiration were estimated for the whole irrigation district using the Landsat images to demonstrate the applicability to large areas.