Regional transitions in factors associated with groundwater recovery: Insights from fused monitoring-well and GRACE-FO observations coupled with interpretable deep learning in the North China Plain
Groundwater recovery influences regional ecohydrological conditions, with implications for groundwater-dependent ecosystems and sustainable water resources management. This study integrated monitoring-well observations, aquifer hydrogeological parameters, and GRACE-FO data to assess Groundwater Storage Anomalies(GWSA) across the North China Plain from 2018 to March 2025 at well, city, subplain, and regional scales. Dynamic fusion combined local information from monitoring wells with regional satellite constraints, resolving recovery hotspots while improving spatial continuity in sparsely monitored areas. Multi-source statistics and observations characterized potential influencing factors. A Temporal Convolutional Network(TCN) with Integrated Gradients(IG) identified changes in variables' predictive contributions before and after 2021 and their regional differences. Multiscale assessments consistently showed a widespread phased transition in GWSA around 2021, although the magnitude of recovery varied markedly among regions. Fusion-GWSA revealed that strong recovery was concentrated in the piedmont plain and generally weakened toward the east-central and coastal plains. In the medium-to-high water-receiving region(R1), the predictive contribution of reclaimed-water use was substantially greater than that in the low water-receiving region(R2), consistent with the continued growth of reclaimed-water use in cities such as Beijing and Tianjin. The predictive contribution of the South-to-North Water Diversion was greater in R1 than in R2, consistent with the physical basis of the regional delineation. Precipitation contributions increased by similar magnitudes in both regions, whereas the contribution of agricultural water use increased significantly only in R2. These regional differences call for ecologically and structurally safe groundwater-level management in areas such as R1, and attention should be given to coordinating irrigation demand with groundwater recharge in R2.
Authors
- Lin Zhu (ORCID: https://orcid.org/0000-0003-2725-6666)
- Beibei Chen (ORCID: https://orcid.org/0000-0001-8120-8476)
- Huili Gong (ORCID: https://orcid.org/0000-0001-8691-0264)
- Chaofan zhou
- Yongkang Wang
- Haigang Wang
Institutions
- China Institute of Geological Environmental Monitoring (CN)
- Capital Normal University (CN)
Publication Details
- Journal
- Ecological Indicators
- Published
- 2026-11-13
- DOI
- https://doi.org/10.1016/j.ecolind.2026.115842
- Primary Topic
- Geophysics and Gravity Measurements
- Type
- article
- Field-Weighted Citation Impact
- 0.00
Funders
- National Natural Science Foundation of China