The use of high‐resolution image time series for crop classification and evapotranspiration estimate over an irrigated area in central Morocco

Abstract

A time series of eight high‐resolution Landsat TM images, ranging over the crop season, has been acquired over an irrigated area in central Morocco. From this time series, a Normalized Difference Vegetation Index (NDVI) profile was generated for each pixel. In order to get significant profiles, the images were radiometrically corrected, first, using invariant objects located on the scene, based on visual observation of the images, and second, using the reflectance of these objects, estimated from a previously corrected image. In the following step, these NDVI profiles were used to identify four main crop types—bare soil, annual crops, trees on bare soil and trees on annual understory—using a decision tree algorithm. The resulting land cover map and the associated NDVI profiles were then used for an evapotranspiration estimate over the whole area, using the Food and Agriculture Organization (FAO) model. Daily outputs of the Moroccan meteorological model Aire Limitée Adaptation Dynamique développement International (ALADIN) were used to generate reference evapotranspiration (ET0) maps and K c estimates were determined using the NDVI profiles.