| Irrigation areas play an important role in national economy and the development of society in our country. However, as a result of unenlightened management means and methods, the water use efficiencies are universally low, and the waste of water is serious in irrigation areas. Consequently, to enhance the management of irrigation areas is imperative under the situation. To estimate crop water requirements is one important part of agricultural irrigation management. In order to carry through scientific and efficient irrigation management, crop water requirements must be calculated and forecasted accurately. This paper will study on prediction model of crop water requirements and apply it to the irrigation management information system of Shulehe irrigation area.There are a variety of methods to calculate crop water requirements,and the most widely used one is using the reference crop water requirements & crop coefficient to calculate actual crop water requirements. This study takes vernal wheat (one of primary crops in Shulehe irrigation area) for example, calculates reference crop water requirements using Penman-Monteith method and computes crop coefficients of different growth stages in the way of single crop coefficient method. And then, actual crop water requirements can be reckoned .This study uses actual crop water requirements which have already been calculated as target outputs, weather factors as inputs, to establish the BP neural network model of crop water requirements forecast. Different models are created by choosing diverse weather factors combination,number of hidden neurons and network training function. And the forecasting precisions of these models are compared to choose the best one. The optimized model is applied to predict the crop water requirements of Shulehe irrigation area to provide guidance for crop irrigation in the irrigation area.This study also has reference value to the crop water requirements prediction of other arid and semi-arid regions.
|
PDF Full Text Request