Researchers used data from the GRACE/GRACE-FO platform to estimate that the region has lost a combined 34 cubic kilometers of groundwater since 2002.
By James Acker, NASA
The Colorado River is one of the most famous rivers in the United States, in part because of the Grand Canyon and other iconic landscapes it and its tributaries have created. The river is also notable because its water provides the American Southwest with large reservoirs, notably Lakes Mead and Powell. Millions of people in this desert region rely on the Colorado River for water, and it is also critical for local agriculture.
Increasing demands for the Colorado River’s water, combined with changing weather patterns in the region, have led to noticeably lower water levels in its major reservoirs. These declines are easy to see as “bathtub rings” on the shorelines of Lakes Mead and Powell. The marks of high water levels in the lake, compared to where the current reservoir levels stand, indicate that every drop of this vital resource is being used.
Using Satellite Observations to Monitor Aquifers
Flowing underneath the desert surface are important aquifers that also provide water for the people, industry, and agriculture of the region. The Colorado River and its reservoirs are not sufficient to meet the regional demand for water, so groundwater has been extracted from the aquifers, causing the aquifers to lose water at an increasing rate during the past 20 years. Unlike the reservoirs, this loss is not easily perceptible on the surface.
However, the loss in volume can be detected by satellites in space. By measuring the change in the force of gravity in the region caused by the decreased amount of water in the reservoirs and aquifers, researchers can estimate the volume of water removed.
In a recent study published in Geophysical Research Letters (Abdelmohsen et al. 2025), a research team used gravity data from NASA’s Gravity Recovery and Climate Experiment (GRACE) and GRACE Follow-On (GRACE-FO) to determine how much water has been withdrawn from Colorado River Basin aquifers since 2002. This study also used data from a second NASA source, the North American Land Data Assimilation System (NLDAS), to confirm that the estimates made with GRACE gravity data were accurate.
The results indicate that the reservoirs and the aquifers have lost a combined 52 cubic kilometers of water since 2002. The reduction of groundwater was 65% of that total, about 34 cubic kilometers. Such a loss is not sustainable, and the risk of running out of freshwater is potentially serious.
Gravity Data from GRACE and GRACE-FO
The GRACE missions are remarkable, both for the data they provide and the technology that is employed to acquire the data. The concept is simple: two trapezoid-shaped satellites orbit in tandem, with about 220 kilometers between them. Radio (microwave) signals determine the precise distance between the two satellites, with Global Positioning System (GPS) antennas indicating their orbital positions. As Earth’s gravitational field varies under the satellites as they orbit, the distance between the two satellites will vary as they are either pulled away from each other or pushed slightly toward each other. By taking repeated measurements over numerous orbits, researchers can precisely derive Earth’s gravity field from the small variations in the distance between the satellites.
There are two main reasons to measure Earth’s gravity field. One, a precise determination of the gravity field generates baseline data that can be used to measure the variation in the height of the surface of Earth. For example, such measurements can be used to determine how much a large earthquake has changed the height of Earth’s land surface, including the official altitude of mountain peaks.
Two, measuring the Earth’s gravity field allows researchers to measure variations of Earth’s gravity, which is the measurement used to assess the Colorado River Basin Aquifers. As the amount of water in a given region varies, Earth’s gravity in that region will change as the mass of water increases or decreases. This variation indicates changes in water storage on the surface of Earth, and underneath the surface as well. That’s where the “climate” part of the GRACE mission name comes from.
The figure below (from Humphrey, Rodell, and Eicker 2023) shows how this works.
The actual method used to calculate the changing gravity field is called “mascons,” which is a short name for “mass concentration.” Earth’s gravity field at a given time is actually a “solution” that unifies all the data that the GRACE mission collects. Mascons are a type of solution that indicates where the gravity is changing due to changes in mass at the surface. The mascons allow the calculation and depiction of gravity anomalies — where the gravity has increased or decreased in comparison to the average (mean) gravity field. The image below shows gravity anomalies for May 2025.
The May 2025 map of gravity anomalies has some notable features. The positive anomalies in the Congo and Amazon rainforest regions are due to the iconic “rainy season,” and this may also account for the positive anomaly in the Sahel region south of the Sahara Desert. In May 2025, rainfall was higher than normal in the Congo River basin, but approximately normal in the Sahel and the Amazon basin. This map also clearly shows the negative gravity anomaly in the Colorado River watershed in the U.S. desert Southwest. This persistent anomaly primarily results from the continued extraction of groundwater from the aquifers in this region, because slightly above average monthly rainfall amounts do not appreciably affect it.
Hydrologic Data from NLDAS
To check the accuracy of the measurements of water loss from the aquifers based on GRACE gravity data, the researchers utilized an independent data source, the North American Land Data Assimilation System (NLDAS).
NLDAS is a system of several land surface models (LSMs), which allow researchers to determine the values of hydrologic variables such as runoff, soil moisture, streamflow, and evaporation. LSMs use the input data to calculate the output variables. NLDAS models have different assumptions and different equations to calculate the variables, so the model results can be compared and assessed.
For the Colorado River watershed study, researchers used NLDAS data to check the results calculated with GRACE gravity data. According to Abdelmohsen et al. in their paper, “To assess the accuracy of GRACE data estimates in both the Upper Colorado River Basin and Lower Colorado River Basin, we conducted an independent water mass balance analysis using hydrological fluxes from three models (NOAH, VIC, and Mosaic) from NLDAS, including precipitation, evapotranspiration, and total runoff.
They also used the NLDAS data to calculate groundwater storage. The GRACE data indicated Total Water Storage, the combined amount of water in rivers, lakes, reservoirs, aquifers, soil moisture, and snowpack. Using the GRACE data in coordination with the NLDAS data thus provided a way to check the accuracy of the results, particularly the important estimate of how much water has been lost from the aquifers.
The results of this research are significant for resource managers in the region to consider, particularly for agricultural interests and communities, including Native American tribal lands.
References
Abdelmohsen, K., Famiglietti, J. S., Ao, Y. Z., Mohajer, B., and Chandanpurkar, H. A. (2025). Declining freshwater availability in the Colorado River basin threatens sustainability of its critical groundwater supplies. Geophysical Research Letters, 52, e2025GL115593. https://doi.org/10.1029/2025GL115593
Humphrey, V., Rodell, M., and Eicker, A. (2023) Using satellite-based terrestrial water storage data: a review. Surveys in Geophysics 44, 1489–1517. https://doi.org/10.1007/s10712-022-09754-9
Mitchell, K. E., Lohmann, D., Houser, P. R., Wood, E. F., Schaake, J. C., Robock, A., et al. (2004). The multi-institution North American Land Data Assimilation System (NLDAS): Utilizing multiple GCIP products and partners in a continental distributed hydrological modeling system. Journal of Geophysical Research, 109(D7), https://doi.org/10.1029/2003JD003823.
