A Simple Algorithm to Identify Irrigated Croplands by Remote Sensing

− The identification of irrigated cropland is essential for crop monitoring, yield estimation and water management assessment in drylands. The standard approach is to use supervised classification on multispectral bands, vegetation indices or the Principal Components or, more recently, the combination of optical images with radar data. More complex methods, such as time-series analysis, sub-pixel calculation method and decision-tree based supervised classification have been proposed to differentiate the irrigated areas and identify the irrigation system. As an alternative approach to identify irrigated land, this paper introduces a simple and easily implemented algorithm, based on the logical operation and thresholding of a combination of thermal temperature (Ts) and vegetation indices (e.g. NDVI). This approach is illustrated through a case study in northern Syria using Landsat TM images. The results show a good consistence with the field observations (99%).