Estimation of Crop Coefficients Using Landsat-8 Remote Sensing Image at Field Scale for Maize Crop

A widely used method is to estimate crop water requirements using reference evapotranspiration and crop coefficient. The crop coefficients can be estimated using a relationship between satellite-derived vegetation index and crop coefficient values for efficient and timely agricultural water management strategies. In the present decade several remotesensing based vegetation indices are applied to simulate crop coefficients but almost all are based on linear relationship. The relationship between the NDVI and Kc was established by using both a traditional regression method and ANN model. Given the complex meteorological and biophysical phenomena related with crop coefficients and satellite-derived vegetation index, a linear relationship between these two variables is insufficient to extract the non-linearity and non-stationarity between them. Therefore in this study widely applied feed forward back propagation Artificial neural networks (FFBP-ANN), a soft computing techniques for mapping complex input and output relationship, was applied. Performance of FFBP-ANN for mapping crop coefficients with NDVI was also compared with traditional regression method. It was found that FFBP-ANN can be applied to accurately estimate crop coefficient values using the remote sensing derived NDVI values. This advancement in calculating crop coefficient using free satellite images is a significant step forward in the development of agricultural irrigation demand models. As a result, this research paves the way for near-real-time irrigation decision-making systems.