Machine learning model improves prediction of groundwater depth in China's Ningxia region

Groundwater is one of the most important sources of drinking water for Ningxia, a region located in the arid and semi-arid regions of China, but little research has been done on the application of machine learning models to predict groundwater quality in this region.

Professor Sun Bo of Nanjing University of Information Science and Technology and his colleagues conducted a study on groundwater level forecasting in Ningxia Hui Autonomous Region and found that two hybrid machine learning models, namely Multi-head Attention Convolutional Neural Network Long Short-term Memory (MH-CNN-LSTM) and Multi-head Attention Convolutional Neural Network Gated Recurrent Unit (MH-CNN-GRU), have great potential for groundwater level forecasting in Ningxia Hui Autonomous Region. The results of this research were recently published in the journal “Chinese Academy of Sciences” of the Chinese Academy of Sciences. Atmospheric and Oceanic Sciences Letters.

In this study, we select groundwater-related factors such as precipitation, combine two hybrid deep learning models, CNN-LSTM and CNN-GRU, with multi-head attention, and compare them with a traditional statistical model, the multiple linear regression model.

Furthermore, we use the dung beetle optimization algorithm (DBO) to further enhance the predictive ability of the hybrid deep learning model by optimizing the parameters. We improve the DBO using tent maps, adaptive T-distribution and spiral search strategies, and compare the prediction results of the model using the improved DBO and the original DBO.

The predictive performance is better than that of the traditional multiple linear regression model. Moreover, the DBO algorithm can further improve the predictive accuracy of the model. Compared with the original DBO, the performance of the model using the improved DBO is improved.

Precipitation in Ningxia is mainly concentrated in summer, and groundwater in this region increases significantly in summer compared with the other three seasons. In the future, the research team will focus on summer groundwater in Ningxia and study the related physical mechanisms. They will then further explore whether adding factors related to these physical mechanisms can significantly improve the prediction results.