A new method to distinguish between irrigated dry land and rain-fed dry land using multi-temporal MODIS and ancillary data: An application example in China

Abstract

Accurate information on the extent of irrigated dry land and rain-fed dry land is fundamental to many aspects of human existence and agricultural water resource sustainable development. This paper proposes a new method for distinguishing between irrigated dry land and rain-fed dry land. It chooses three kinds of classification features which are Fourier components feature, fractal texture feature and terrain feature and then uses classification algorithm called GA optimized BP-ANN. Results show that the method this paper proposed is feasible and the accuracy of this result is much higher than others such as USGS's and IWMI's.