You can see a shrunken Colorado River or a parched Danube riverbed from the bank. Groundwater is harder to see, because it vanishes underground. Gizmodo’s report on a new study puts it this way: the crisis beneath our feet is far less conspicuous, though just as severe.
The study, by NASA Goddard scientists Mary Michael O’Neill, Matthew Rodell and Bryant Loomis, appeared in PNAS. It covers global terrestrial water storage from April 2002 to November 2025. Terrestrial water storage means all the water held on and under land. These are the numbers that matter.
40 hotspots, out of 94
The team found 94 regions with shifts in water storage above background levels. In 40 of them, human land or water use was the main influence rather than climate variability. Those 40 are the study’s focus.
3,122 Gt lost, 2,432 Gt gained
Across these regions, cumulative losses of 3,122 gigatons substantially exceeded gains of 2,432 gigatons. One gigaton of water is roughly one cubic kilometre. The total loss is about 750 cubic miles, more than Lake Victoria holds. The gains came from things like rain-fed agriculture, surface-water irrigation, reservoirs and deforestation.
33% larger trends
Compared with the traditional monthly approach, the new method gave regional trend magnitudes that were on average 33% larger. The authors conclude that groundwater pumping is hitting harder than earlier work suggested. Live Science and Gizmodo cite a 45% figure for groundwater-related declines specifically. I could not find that figure in the abstract.
58,000 square miles
Earlier methods could not reliably detect changes in areas smaller than about 58,000 square miles (150,000 square kilometres). The new method fits the raw satellite-to-satellite ranging data directly. According to Earth.com, the resulting map picks out changes five to ten times finer.
Every continent but one
Human land and water use was significantly linked to strong trends on every non-polar continent except Australia. Widespread losses showed up in the Middle East and northern Africa, northern India, the western United States and southeastern Brazil.
The caveats
The method shows long-term trends but cannot track how changes unfold year to year. Monitored wells are too sparse to check the map against the ground worldwide. The paper gives no population total for the affected areas.
Taken together, the figures show that the satellite picture has sharpened, and that sharpening has made the drain look bigger, not smaller. The worst of it sits where farms pump hardest. O’Neill told Earth.com: “The regions of greatest concern are those already recognized as water-stressed, where we now observe losses even greater than previously estimated.”
