| This article has been：Browse 939Times Download 2798Times |
scan it! |
|
|
| DOI：10.13522/j.cnki.ggps.2022123 |
|
| Water Allocations in Irrigation Districts Determined by Random Forest Combined with SHAP Method |
|
SU Nan, ZHANG Shaohui, BAI Meijian, ZHANG Baozhong
|
|
1. China Institute of Water Resources and Hydropower Research, Beijing 100038, China;
2. State Key Laboratory of Simulation and Regulation of Water Cycle in River Basin, Beijing 100038, China
|
| Abstract: |
| 【Objective】Decades of management of various irrigation districts have accumulated valuable experiences, but how to make most of these experiences has not yet been appreciated. The purpose of this paper is to fill this gap, proposing a method to harness these experiences to improve water allocation calculation in irrigation districts.【Method】The analysis is based on data measured over a three-year period from an irrigation area at the Waxi branch in the Pishihang irrigation district. We consider spatial variation in variables such as temperature, rainfall and soil moisture in integrating the interpretation method of the random forest model with the SHAP model. Nonlinear quantitative relationship between the target water allocation and each variable was constructed using limited data samples.【Result】The model obtains the important scores and changes in each variable in a long time series for different typical years. In addition to the combination of characteristic variables suitable for actual water dispatching, the indexes of characteristic variables that dispatchers mainly refer to different dispatching scenarios can be analyzed and obtained. Combined with the analysis of SHAP value, the positive and negative directions of the response of water dispatching target flow to each characteristic variable is obtained.【Conclusion】The method proposed has realized quantitative representation of historical experience of water dispatching in irrigated areas and provided guidelines for rational prediction of different water dispatching flows in irrigation districts. |
| Key words: Pishihang irrigation district; water dispatching; random forest model; SHAP method; characteristic variable; nonlinear characterization |
|
