Prediction of farmland water environment after reconstruction of water-saving in the large-scale irrigation district based on the ANN technology

In order to investigate the change of regional farmland water environment and its possible negative effect caused by reconstruction of water-saving in the large-scale irrigation district, the Hetao irrigation district of Inner Mongolia in the Yellow River valley for a study case was selected. Through the monitoring data of regional water and salt movement and analysis of data of many years’ hydrology and groundwater in the three different scale research areas, the ANN technology (BP model and RBF model) was applied for prediction of farmland groundwater hydrology and soil water and salt movement in the three different scale research areas. Then, the movement trend of groundwater hydrology and soil water and salt contents were predicted in the Hetao Irrigation District in future different water-saving level years. Furthermore, the results of ANN model were compared with those of finite element method and hydrological budget method. As a result, the predicted results were rational and accordant. At the same time, the ANN method showed more advantage comparing with traditional method. This indicates that the ANN technology expresses prominent advantage in the prediction for the complicated system and can obtain better results. So, the ANN is suitable for regional trend estimation in the future when the influence factors are changed. Furthermore, this paper brings forward a moderate environment safety threshold about water-saving of irrigation district in the arid and soil salinization irrigation areas. This can provide instructions for macro-management and decision of reconstruction of water-saving project in the large-scale irrigation district.