Due to increasing requirements on water resources and a lower recharge rate, the farming seasons are a vital season for the management of groundwater and surface water resource management. This condition necessitates the use of combined water distribution to meet the full water requirements. Analysis of existing surface water resources and related restrictions, this research suggested an algorithm for aquifer stabilization and fulfilling optimum water requirements. To manage the optimum withdrawals and the subsequent drop, this technique first employed the MODFLOW model for simulating the water levels. Next, an improved feed-forward neural network (IFFNN) was combined with an optimization method to create a machine learning (ML) framework. During the last phase, the findings of the optimized connectives approach as well as the relevant fields technologies to determine using improved bald eagle search with least square SVM(IBES-LSSVM) method that predicted the level of water deficit for every period, especially during farming seasons. This approach is based on an improved bald eagle search (IBES) optimization technique for finding the best settings for a least-squares support vector machine (LSSVM). The findings revealed that between 2005 and 2020, the year with the biggest water deficit was 2018 when only roughly 64 percent of water need was satisfied by groundwater (69 percent) and surface water (64 percent) (33 percent). The water depth may have risen by around 0.7 m during the study period if the optimum model had been used. The outcome of this research will help the management forecast future water shortages and make smarter water strategic choices.
Yan Jixuan, Li Guang, Qi Guangping, Yao Xiangdong, Song Miao
Agriculture, Groundwater, Machine learning, Neural networks, Optimization, Surface water, Water management, Water supply