Deriving groundwater storage anomalies based on GRACE data and drought prediction using deep learning

The sustainability of water resources is significantly threatened by groundwater drought, primarily due to the increasing impacts of climate change. Therefore, understanding the variability of water resources and making accurate projections regarding their future availability is crucial. This study focuses on Shaanxi Province, which is divided into three distinct regions: Guanzhong in the center, Southern Shaanxi, and Northern Shaanxi. Each region has vastly different climates, which complicates drought monitoring. This study utilized data from the Global Land Data Assimilation System (GLDAS) and the Gravity Recovery and Climate Experiment (GRACE) satellites to analyze the changes in groundwater storage anomalies (GWSA) in Shaanxi Province from 2002 to 2021. We developed the Standardized Groundwater Index (SGI) to elucidate the spatial and temporal characteristics of groundwater drought and its changing trends. The Mann-Kendall (MK) trend test was employed to quantitatively assess groundwater drought conditions in three regions. Furthermore, to enhance drought monitoring and prevention capabilities, we applied four deep learning models: long short-term memory (LSTM), convolutional neural network–long short-term memory (CNN-LSTM), bidirectional long short-term memory (BiLSTM), and long short-term memory with attention mechanism (LSTM-Attention) to predict drought indices. Our findings indicate that while the drought trend in Southern Shaanxi is not as prominent, it is declining in Northern Shaanxi and Guanzhong. Overall, the average accuracy of the deep learning models in forecasting drought indices across all grid points in the three regions is above 84%, with the CNN-LSTM model demonstrating the best performance. This research highlights the significant potential of deep learning approaches for drought monitoring and forecasting, providing a new and effective strategy for water resource management and early warning systems. The implications for drought monitoring and prevention in Shaanxi Province and beyond are substantial, as these methods can enhance responses to drought occurrences and ensure the sustainable use of water resources.