Application of a combined model based on wavelet analysis for predicting crop water requirement

Wavelet analysis was used to reconstruct the time series of the crop water requirement into different scales in order to reduce the randomicity, and then the reconstructed time series was predicted using grey and time series prediction method to increase the prediction accuracy of agricultural water requirement (non-stationary time series).The space of prediction and application of non-stationary time series were expanded through the combined model of wavelet analysis, gray and time series prediction methods.The crop water requirement in Erdos was validate by the method, and the results showed that the prediction accuracy was high and relative error less than 3%（2009）. It can provide a new method for prediction of agricultural water requirement and has great significance to Erdos for rational use of water resources, planning and management, promoting social and economic sustainable development.