The water that doesn't refill
There's a line on this site's water-stress ranking that should stop you: a handful of countries withdraw more than 100% of the water nature renews for them each year. We've said before that this happens two ways — you desalinate the sea, or you mine an aquifer that doesn't refill. This is the note about the second one, and about how we now know which countries are doing it.
What's fossil water?
Some groundwater is a savings account that fills back up: rain seeps down, the aquifer recharges, and as long as you don't withdraw faster than it refills, it lasts forever. But some groundwater is different. It seeped into the rock thousands of years ago, under climates that no longer exist, and nothing is topping it up now. Pump that and you're not borrowing — you're spending a balance that will never be replaced. It's called fossil water, and like the other kind of fossil fuel, there's a fixed amount and then there's none.
The cruel geography is that the biggest fossil-water reserves sit under the driest land — the Sahara, the Arabian Peninsula — precisely where there's no rain to recharge anything and every drop has to come from underground. So that's exactly where it's being mined hardest.
How do we know?
You can't see underground, and almost nobody has good meter data on what's pumped. So for a long time the honest answer was that we didn't really know. Then came GRACE — a pair of NASA satellites that measure Earth's gravity precisely enough to weigh the water beneath a region from orbit. When an aquifer loses mass, the satellites feel the dip. For the first time, groundwater loss could be measured from space rather than guessed from patchy wells.
The landmark result: of the planet's 37 largest aquifer systems, studied with GRACE over 2003–2013, 21 are being depleted faster than they refill, and 8 are "overstressed" — heavy withdrawals against almost no natural replenishment. Here are the worst, with one column borrowed from this site: the live water-stress ratio of the country sitting on top.
| Aquifer system | Where | GRACE class | Recharge | Country stress ratio |
|---|---|---|---|---|
| Arabian Aquifer System · #1 most stressed | Saudi Arabia, Yemen | Overstressed | Almost none — fossil water | 974% (Saudi Arabia) |
| Indus Basin · #2 most stressed | Pakistan, north-west India | Overstressed | Far below withdrawals | 110% (Pakistan) |
| Murzuk–Djado Basin · #3 most stressed | Libya, Niger, Algeria | Overstressed | Almost none — fossil water | 817% (Libya) |
| Northwest Sahara Aquifer System | Algeria, Tunisia, Libya | Overstressed | Almost none — fossil water | 145% (Algeria) |
| California Central Valley | United States | Highly stressed | Some, but mined hard in drought | 28% (United States) |
Aquifer classification: Richey et al., Water Resources Research (2015), NASA GRACE 2003–2013. Stress ratios: FAO AQUASTAT, latest year, as elsewhere on this site.
So how long until it's gone?
Here's the part that should worry you most: nobody knows. The same researchers tried to pair their loss rates with estimates of how much water is actually left in each aquifer, and found the estimates almost useless. For the overstressed Northwest Sahara system, the projected time-to-depletion ranged from 10 years to 21,000 years depending on whose volume figure you believed. Saudi Arabia, sitting on the most overstressed aquifer on Earth, is meanwhile running a water-stress ratio of about 974% — drawing many times what the desert renews. You can't manage a resource when your estimate of how much is left spans four orders of magnitude. We are flying blind and pumping hard.
Is it getting worse?
Yes, and faster. A 2025 study using the full GRACE record found that drying land is now spreading across the Northern Hemisphere, and that groundwater depletion accounts for about 68% of the freshwater the continents have lost since 2002 — more than melting glaciers. The land is now handing more water to rising seas than the ice caps are. The aquifers aren't just local accounts being overdrawn; together they've become a planet-scale leak.
Why it belongs in the footnotes here, not the rankings
We've put no aquifer number on the main spine of this site, and that's deliberate. GRACE measures basins, not countries; the volumes left are genuinely unknown; and a tidy national ranking would imply a precision the science doesn't have. So this stays a note, sourced and caveated. But read it next to the stress ratio and the wealth-and-water story and the picture sharpens: when you see a country drawing more than 100% of its renewable water, ask the next question. Is it buying the difference from the sea — or digging it out of the past? One of those you can keep doing. The other ends.