The chase is over. After 15 years of service, one of two NASA and German-run GRACE satellites is decaying into lower orbit, on course to burn up in the atmosphere in mid-March just as its companion satellite did at the end of 2017. The pair, sometimes referred to by scientists as “Tom” and “Jerry,” worked in tandem to measure changes in gravity on Earth. As floods and droughts impacted water quantities over large areas and humans depleted water supplies, the GRACE satellites recorded the change in the mass of bodies of water — producing data vital to understanding changes in water supply related to climate and human stresses.

A study published in the Proceedings of the National Academy of Sciences on Jan. 22 uses GRACE satellite data recorded between 2002 and 2014 to track water storage in 186 river basins around the world. Lead author Bridget Scanlon, a senior research scientist at the University of Texas at Austin, and her coauthors compared satellite observations to seven models commonly used by scientists investigating the water cycle, a topic of growing importance as the climate changes and population grows.

“Sometimes the models simulated the opposite trend” to what satellites observed, Scanlon said. “If the satellites go over an area that’s getting wetter or flooding, then the gravitational attraction would increase, and the leading satellite would speed up. The distance between the two satellites would then increase,” Scanlon said. “So these changes in distance between the satellites … provide an estimate of changes in water storage, from up in the atmosphere, to deep in the subsurface.”

Scanlon found that the models underestimated large changes in water storage, both with declines mostly caused by human-driven depletion, and rises related to precipitation and climate variability.

The models she investigated are split into two categories. Five of the seven were land models, components of global climate models, while the two others were expressly built to focus on human water use and predict scarcity in the future.

“Most of the land surface models only simulate soil moisture storage, not surface water or groundwater storage. All of the models may not have sufficient storage capacity in the soil profile to accommodate the changes in water storage that we get from wet and dry periods,” Scanlon said.

The GRACE data indicate there was a net increase in land water storage from 2002 to 2014, whereas all the models simulated a net decrease. From basin to basin, the GRACE observations vary: showing increasing storage in places like the Amazon and decreasing storage in the Ganges in India.

Many of the models the researchers looked at estimated the Ganges would gain water over the 12-year period, but GRACE observed a loss of 12 to 17 cubic kilometers per year. Scanlon said the decline was likely caused by human extraction, something many land models don’t capture.

“The results suggest that past studies that have used models to estimate global-scale water storage could have underestimated the effects of humans and climate variability. It means the potential effects of human water use and climate change in the future could be worse than we thought for some regions of the globe,” said Simon Gosling, a climate risk professor at the University of Nottingham who was not involved with the study.

One of Scanlon’s coauthors, Hannes Müller Schmied, a senior researcher at Goethe University Frankfurt, works on improving a global hydrological model that focuses on human alterations to the water cycle.

“The focus should now lay on the models themselves and Bridget provides the basis for it,” he said.