Water storage trends derived from the GRACE/-FO global gravity-based groundwater product (G3P)

The Global Gravity-based Groundwater Product (G3P) reflects observations of global groundwater storage (GWS) variations derived from GRACE/-FO satellite gravimetry. It is calculated from terrestrial water storage (TWS) anomalies by subtracting aggregated and filtered contributions from root-zone soil moisture, glaciers, surface water storage, and snow water equivalent. As such, G3P provides a crucial observational constraint for assessing global groundwater depletion, recharge, and long-term water storage changes related to climate variability and human activities. A central challenge in the analysis of GRACE/-FO-derived water storage time series is the reliable separation of long-term trends, arising from anthropogenic forcing and climate change, from stochastic signals attributable to natural climate variability (“climate noise”) and observational system instabilities. To address this, we introduce a trend analysis framework that uses calibrated parametric time series models to jointly represent trends, seasonal variability, and temporally correlated stochastic processes. By explicitly accounting for short- and long-range memory in the water storage time series, this approach requires minimal assumptions about the underlying physical processes and provides a robust basis for separating long-term trends from stochastic variability using statistical significance testing.

We first validate the framework for TWS by comparing detected trends with previously reported GRACE-based results and by providing consistent and reliable estimates of trend magnitudes and uncertainties. We then apply the framework to derive and analyse trends in GWS variations. Our results show that groundwater storage decrease is the dominant contributor to negative TWS trends in many regions, with Asia (specifically the Middle East, Northern India, Northern China, and South-east Asia) experiencing a decline of about −43 km³ yr⁻¹. At the same time, we reveal previously unobserved trends, including increasing groundwater levels in large parts of Africa (+34 km³ yr⁻¹) and declining trends attributed to droughts in regions such as Southern Africa, Asia, and Eastern Europe. The resulting global budget indicates significant GWS losses of −22 km³ yr⁻¹ and TWS losses of −154 km³ yr⁻¹ (excluding Antarctica and Greenland). Beyond the regional patterns, this study demonstrates how accounting for stochastic memory fundamentally affects trend significance and uncertainty estimates in GRACE/-FO time-variable gravity-field observations.

The proposed framework is scalable and transferable to other Essential Climate Variables, contributing to a more reliable detection of subtle long-term changes in Earth system mass variations.