A study on groundwater level calculation based on PCA-CIWOABP

In order to explore the relationship between groundwater levels and hydro-meteorological factors in Fengnan District, accurate estimation of groundwater levels in the area was undertaken. Real data on groundwater levels, water consumption, and rainfall from 2018 to 2021 in various townships within Fengnan District were selected. Utilizing the Principal Component Analysis method, the main influencing factors were extracted from the hydrological data of each township. Subsequently, a groundwater level calculation model was established using the CIWOABP(Cubic map - Intelligent weight adjustment - Whale Optimization Algorithm–Back Propagation) neural network in combination with these factors. The results indicate that: (1) Principal Component Analysis extracted a total of five principal components from various hydrological data in Fengnan District, namely, groundwater levels of monitoring wells #11 and #12, rainfall from rainfall station r1, and water consumption from Fengnan (FN) and Qianying (QY) towns. (2) The CIWOABP neural network was trained using 36 sets of actual measurement data and validated with 12 sets of simulated data. The mean absolute errors (MAE) for monitoring wells #11 and #12 were 0.19 and 0.23 respectively, and the mean squared errors (MSE) were 0.05 and 0.09 respectively. The model exhibited high computational accuracy and can be effectively employed to calculate actual groundwater levels. The research outcomes can provide theoretical and methodological insights for groundwater resource management in the North China Plain.