GRACE Satellite Data Reveals Groundwater Anomalies in Ghana

A recent study featured on arXiv (2608.10233) examines groundwater storage anomalies in Ghana using GRACE satellite data spanning from 2004 to 2024, an area noted for its scarcity of in-situ observations. Employing an ensemble-based Isolation Forest framework, the research integrates statistical analysis with unsupervised machine learning to identify anomalies, standardizing groundwater anomalies via Z-scores. The results reveal notable temporal fluctuations, with consistent groundwater deficits occurring from 2004 to 2009, followed by a rise in positive anomalies post-2018. The analysis identified 12 anomalous months, including 5 deficit and 7 surplus events, with deficits showing stronger anomalies. Additionally, spatial analysis indicates that deficits are more prevalent in northern Ghana, while surpluses are more pronounced in the south. The study's machine learning outcomes were compared to statistical thresholds, highlighting the framework's utility. This research holds importance for managing water resources in Ghana, where long-term data on groundwater variability is lacking.