A review of current best practices and future directions in assimilating GRACE/-FO terrestrial water storage data into numerical models

Core Idea

Assimilating GRACE/-FO terrestrial water storage (TWS) data into hydrological and land surface models improves reanalyses, capturing climate-driven variability, human impacts, and extremes (droughts/floods).

Why It Matters

GRACE/-FO data assimilation is more than a technical advance — it directly supports decision-making in water management and climate adaptation. By improving estimates of groundwater depletion, soil moisture variability, and snowpack dynamics, assimilation provides actionable insights for agriculture, urban water supply, and drought preparedness. Enhanced flood forecasting and drought attribution strengthen early warning systems, reducing risks to communities and economies. Multi-sensor frameworks further expand applications, enabling integrated monitoring of ecosystems and supporting climate model evaluation. Beyond hydrology, GRACE/-FO DA contributes to geophysical research by correcting GNSS deformation signals and refining Earth system models. In practice, these advances help governments, water managers, and researchers make better-informed decisions, ensuring resilience against extremes and sustainable use of freshwater resources.

How It Works

GRACE and GRACE‑FO satellites measure monthly changes in Earth’s gravity field, which reflect total terrestrial water storage (TWS) across all compartments — groundwater, soil moisture, snow, and surface water. Hydrological and land surface models, by contrast, simulate these compartments daily or sub‑daily on fine grids. Data assimilation bridges the two: the coarse satellite observations are merged with the high‑resolution model outputs, updating each storage component and related fluxes (like evapotranspiration, river discharge, and infiltration) in a physically consistent way.

Figure 1: The GRACE and GRACE Follow-On (GRACE/-FO) data assimilation (DA) concept: A numerical model simulates daily or even subdaily individual water storage components and fluxes on a user-defined grid. In contrast, the GRACE/-FO satellites observe the total variation in monthly TWS aggregated over large footprints and over all storage compartments. During DA, the model’s individual storage compartments are updated towards the GRACE/-FO observations. These updates influence water fluxes – illustrated by the arrows in the figure – as well as other related model variables such as soil temperature, energy fluxes, and plant growth.

Key Advances

Downscaling & disaggregation: Coarse GRACE/-FO signals mapped to finer model grids.
Groundwater focus: Major improvements in depletion detection (China, Iran, Australia, California).
Snow & floods: Better snowpack, flood forecasting, drought attribution.
Multi-sensor DA: Combining GRACE/-FO with soil moisture, streamflow, LAI, flood extent → more consistent improvements.
Applications beyond hydrology: GNSS deformation correction, CMIP6 evaluation, machine learning hybrid approaches.

Figure 2: (a)General concept for assimilating monthly GRACE/-FO-derived TWSA into GHMs and LSMs along with most relevant Ensemble Kalman Filter (EnKF) equations showing two options for applying the assimilation increments: δx_i¹…δx_i³⁰ After computing the increments, the model is either (A) rewound and re-run over the month with the increments distributed across all days, or (B) updated by applying the full monthly increment ;δx_i^m at the end of the month. Please note that the equations provided refer to the EnKF and are expressed for each ensemble member i. The Kalman Gain matrix determines the update weights based on the state and observation error covariance matrices. and X^'f and Y^'f are matrices of forecast state anomalies and forecast state observation-space anomalies, respectively. Each column represents the deviation of one ensemble member from the ensemble mean. For further details on the equations for other DA algorithms.

Figure 3: Statistics of common settings in GRACE/-FO DA experiments, including the (a) hydrological model used, (b) continent of study, (c) GRACE/-FO observation product (analysis approach), and (d) observation error model (Status as of October 2024).

Best Practices

Across reviewed studies, GRACE/-FO DA has demonstrated robust improvements in groundwater estimation and drought monitoring, while impacts on streamflow remain region-dependent. Multi-sensor approaches provide the most consistent performance across variables. The ensemble Kalman filter (EnKF) remains the most widely adopted approach, often enhanced with localization techniques to prevent spurious correlations and improve computational efficiency. Successful frameworks explicitly account for correlated observation errors and apply bias correction to both model outputs and GRACE/-FO observations, ensuring that assimilation increments reflect true hydrological signals rather than systematic offsets. Signal separation and multi-source assimilation are also emphasized, allowing unmodeled processes—such as anthropogenic water use or surface deformation—to be treated appropriately before integration. Collectively, these practices yield physically consistent updates across groundwater, soil moisture, snow, and related fluxes, strengthening the reliability of water cycle reanalyses and forecasts.

Challenges

Despite these advances, several persistent challenges limit the full potential of GRACE/-FO data assimilation. The coarse spatial and temporal resolution of GRACE/-FO observations—typically monthly and several hundred kilometers—creates mismatches with high-resolution hydrological models that operate at daily or sub-daily scales. Assimilation increments are not always allocated to the correct storage compartments, leading to potential misrepresentation of groundwater or soil moisture changes. Moreover, many models still lack explicit representation of human activities such as irrigation, pumping, and reservoir operations, which can distort assimilation results. The diversity of DA algorithms and observation products has also led to a lack of consensus within the community regarding optimal configurations. These issues highlight the need for standardized validation protocols and improved coupling between models and observations to ensure consistent and interpretable outcomes.

Figure 4: Spatial maps of TWSA from the CLSM model only (open loop), GRACE DA into CLSM and GRACE for a US western region centered around Colorado in April 2006 (panels a, b and c); TWSA time series during 2003–2007 for locations A and B (panels d and e). CLSM refers to the NASA Catchment LSM. Further details on CLSM and this GRACE DA simulation can be found in Li et al. (2019).

Findings

The synthesis of approximately sixty GRACE/-FO data assimilation studies reveals clear evidence of the technique’s transformative impact on hydrological modeling and water cycle reanalyses. Assimilation has been shown to significantly improve improves the representation of groundwater dynamics, soil moisture variability, and snow processes, leading to more accurate estimates of total water storage and related fluxes. In regions such as Northern India, China, and Iran, GRACE/-FO DA has successfully captured irrigation-driven depletion trends that conventional models often miss. Multi-sensor assimilation frameworks further enhance consistency across variables, demonstrating that integrating GRACE/-FO with soil moisture, streamflow, and vegetation data yields the most robust improvements. The review also highlights that GRACE/-FO DA strengthens drought and flood prediction capabilities, supports attribution of hydrological extremes to specific storage components, and provides valuable constraints for evaluating climate model performance. Overall, the findings confirm that GRACE/-FO assimilation bridges the gap between coarse satellite observations and fine-scale hydrological processes, enabling physically consistent, data-driven insights into terrestrial water storage variability worldwide.

Future Directions

The next generation of GRACE-like missions and modeling frameworks promises to overcome many of these limitations. Research is moving toward low-latency data products that enable near-real-time assimilation for operational water management and disaster response. Enhanced multi-sensor integration—combining GRACE/-FO with soil moisture, streamflow, vegetation, and surface elevation data—will provide more comprehensive constraints on water storage dynamics. Hybrid approaches that merge data assimilation with machine learning are gaining traction, offering improved downscaling and predictive skill. Furthermore, coupled subsurface models with three-dimensional groundwater flow are emerging as powerful tools to represent complex hydrological interactions. Together, these innovations point toward a future of more accurate, consistent, and actionable water cycle reanalyses capable of supporting global resource management and climate resilience.

Key Takeaways

Improves groundwater and drought monitoring
Enables physically consistent water cycle estimates
Integrates multiple satellite observations
Supports climate adaptation and water management

Please Find the full publication here: https://doi.org/10.5194/hess-30-985-2026

RECENTLY PUBLISHED

Weight-Supported Random Forest Downscaled GRACE(-FO) Data

This study investigates groundwater depletion in the Lower Indus Basin (LIB), where intensive irrigation has led to unsustainable groundwater extraction. To address the coarse spatial resolution of satellite-derived water storage data, a Weight-Supported Random Forest (WSRF) model was developed to downscale data to a finer 0.1° resolution using precipitation, evapotranspiration, vegetation index, and soil moisture as predictors. The results showed high agreement with groundwater well observations (R² > 0.85) and revealed detailed spatial patterns of groundwater depletion from 2003–2023, highlighting the combined influence of climate variability and irrigation practices.