When ground water saturates a river basin, the risk for flooding goes up. So does the strength of Earth’s gravity in that region, ever so slightly, because of the extra mass of the underground water. By using tiny variations in gravity detected from space, researchers report online today in Nature Geoscience thatthey can identify basins that are primed for flooding if additional rains come—sometimes with several months’ warning. As a test case, the scientists looked at the gravity signals leading up to catastrophic floods in 2011 on the Missouri River (pictured above). They used data from NASA’s Gravity Recovery and Climate Experiment (GRACE), a pair of orbiting satellites that get tugged around the Earth faster in places where gravity is slightly stronger. Using the gravity signal—with its sensitivity to groundwater stores—improved model forecasts and predicted a high discharge 5 months before the 2011 flood. The gravity data was more important for forecasts than measures of snowmelt and soil wetness were. The approach is not yet ready to be rolled into flood forecasts. One problem is that the gravity signal is less important in places where rainfall alone dominates—such as monsoon-driven floods in India and Pakistan. Another problem is that GRACE, with its fuzzy resolution, can see groundwater anomalies only over large river basins, 200,000 square kilometers or more. And third, it takes 3 months or more to get data down from GRACE and process it, which erases much of the warning time.