Answers are grounded in this paper and cite their sources.
The reliability of gravity recovery and climate experiment (GRACE) and GRACE follow-on (GRACE-FO) terrestrial water storage (TWS) estimates remains uncertain in data-scarce regions, where limited ground observations, scale mismatch, and hydrological heterogeneity complicate validation. This study presents a quantitative metaanalysis of 234 peer-reviewed GRACE–GRACE-FO validation studies published from 2002 to 2024 to evaluate reported accuracy and identify the main controls on performance. Studies were classified by geographic region, basin area, climate zone, temporal scale, processing product, validation method, and reported accuracy metrics. Overall, GRACE showed useful basin-scale performance, with a pooled correlation coefficient of 0.74, pooled root mean square error of 5.8 cm water-equivalent height, pooled mean absolute error of 4.2 cm, pooled Nash–Sutcliffe efficiency of 0.51, and near-zero median bias. Accuracy was not uniform across settings. Higher agreement was generally reported in basins larger than approximately 100,000 km 2 and in regions with strong seasonal or long-term storage signals, whereas smaller, arid, mountainous, and cryosphere-influenced basins showed greater uncertainty because of leakage, weak signal amplitude, complex storage components, and limited reference data. Temporal aggregation improved agreement by reducing short-term noise, while differences among GRACE products mainly reflected processing-strategy variability rather than independent validation. Downscaling and machine-learning methods improved spatial detail, but they did not consistently guarantee higher physical accuracy without independent validation. These findings indicate that GRACE is most suitable for strategic basin-scale water assessment, groundwater depletion monitoring, drought evaluation, and long-term storage-change analysis in poorly instrumented regions. Future work should prioritize uncertainty-aware validation, transparent reporting of reference-data quality, improved component separation, and integration with next-generation gravity missions.
No takes yet. Share an insight, caveat, or question.
Hannan et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: