On a summer morning in western Kansas, a centre-pivot irrigation system makes its slow mechanical circuit across a quarter-section of corn. It moves like the minute hand of a clock \u2014 imperceptible in real time, undeniable over hours. Tens of thousands of such systems run across the High Plains, and from altitude they carve the flat land into green circles that look, on satellite imagery, like a civilization’s signature. The water pulling through those pipes fell as rain during the Pleistocene, when glaciers still covered the northern half of the continent. It soaked into the ground over tens of thousands of years. It is not coming back.<\/p>\n\n\n\n
Beneath the Great Plains, from the Texas Panhandle north through Kansas, Nebraska, Colorado, Oklahoma, Wyoming, South Dakota, and New Mexico, lies the Ogallala Aquifer: roughly 174,000 square miles of saturated sandstone and gravel, holding water that accumulated over millions of years. It is not a lake. There are no underground rivers. The water exists the way water exists in a sponge \u2014 dispersed through porous rock, held in place by geology, extractable by drilling down and pumping up. This is an aquifer in the technical sense, and the Ogallala is the largest in North America by volume.<\/p>\n\n\n\n
What makes it categorically different from the reservoirs people imagine when they think about water supply is the recharge rate. The Colorado River is depleted and can be refilled \u2014 theoretically \u2014 by upstream precipitation. A reservoir behind a dam empties and can fill again with the next wet year. The Ogallala does not work this way. Recharge rates across most of the aquifer range from less than one millimetre per year in the Texas Panhandle to roughly six inches per year in parts of south-central Kansas \u2014 and those rates measure natural precipitation percolating down through rock that may take decades to reach the water table. Annual withdrawals for irrigation run orders of magnitude above what the aquifer receives back. The USGS has been measuring this since the 1930s, and the numbers have never, not once, suggested a balance.<\/p>\n\n\n\n
Geologists call it a fossil aquifer. The water in it is a relic of a wetter climate, a different world. The Pleistocene snowmelt that fed it is not returning. The climate that deposited it is gone. What remains is a finite stock of ancient water, and the distinction between a stock and a flow is not semantic \u2014 it is the entire argument. A flow can be managed. A stock can only be spent.<\/p>\n\n\n\n
According to USGS Scientific Investigations Report 2023-5143 \u2014 the most recent comprehensive assessment of the High Plains Aquifer, published in February 2024 \u2014 total recoverable water in storage as of 2019 stood at approximately 2.91 billion acre-feet. It represents a decline of 286.4 million acre-feet from predevelopment levels \u2014 roughly 9% of the original volume already gone. The number reads like a reassurance. The south doesn’t agree. The average water-level decline across the entire aquifer since predevelopment is 16.5 feet, but these aggregate figures conceal the actual geography of the crisis, and that geography is what matters. Nebraska, which sits above the aquifer’s thick northern reaches \u2014 saturated thicknesses that can exceed 1,000 feet in west-central areas \u2014 holds roughly two-thirds of the remaining volume and has, in several regulatory districts, achieved rough stabilization. The southern aquifer, under Kansas and Texas, has not. The 16.5-foot average is the average of a place that’s losing relatively little and places that are losing everything.<\/p>\n\n\n\n
If pumping stopped today across the most depleted zones, scientists estimate it would take 500 to 1,300 years for those sections to refill by natural recharge. Full recovery of the aquifer as a whole: roughly 6,000 years.<\/p>\n\n\n\n
There is no technology that changes this. Water recycling, efficiency gains, drip irrigation \u2014 these slow the rate of depletion. They do not reverse it.<\/p>\n\n\n\n
The Ogallala and the High Plains aquifer system<\/strong>\n\nThe USGS formally classifies the Ogallala as part of the High Plains Aquifer (HPA) system, which includes hydraulically connected formations across the eight-state region. \"The Ogallala\" is the name in common use \u2014 in journalism, in policy discussions, in most scientific literature \u2014 and refers specifically to the Ogallala Formation, the dominant geologic unit within the HPA. The distinction rarely matters for the argument; the article uses \"Ogallala\" as the conventional shorthand throughout. Researchers working with USGS data, including the SIR 2023-5143 report cited here, typically report figures for the High Plains Aquifer system as a whole, which substantially overlaps with the Ogallala Formation.<\/code><\/pre>\n\n\n\nThe machine that runs on it<\/h3>\n\n\n\n
A pump goes down a borehole sunk into saturated rock. A motor turns. Water comes up, is distributed across a pipe arm that rotates around a central pivot, and falls onto crops that could not otherwise survive in this climate at these yields. That is the entire mechanism. More than 200,000 wells draw from it across the High Plains, and together those pumps underwrite a food production system that, for most of the people who depend on its outputs, is invisible by design.<\/p>\n\n\n\n
The region above the Ogallala produces approximately 20% of the nation’s wheat, corn, cotton, and cattle \u2014 agricultural output valued at roughly $35 billion annually. Ninety-four percent of everything pulled from the aquifer goes to agriculture, according to USDA data. The Ogallala provides 30% of all groundwater used for irrigation in the United States, which makes it the largest single source of agricultural irrigation water in the country, by a considerable margin. In the most intensively irrigated parts of western Kansas and the Texas Panhandle, rainfall alone cannot support the yields that make these farming operations economically viable. The crops grown here are, without qualification, dependent on Ogallala water. Not supplemented by it. Dependent on it.<\/p>\n\n\n\n
This system was built in a remarkably short period. Through the early twentieth century, the Plains were largely dryland-farmed \u2014 wheat, primarily, with yields that tracked rainfall. The 1930s Dust Bowl was, in part, the consequence of overextending dryland cultivation onto marginal land. Then came cheap centrifugal pump technology in the 1940s and 50s, the spread of rural electrification, and federal price supports that rewarded high-yield irrigated production. By the 1970s and 80s, the High Plains had been transformed. The circles appeared on the landscape. The yields climbed. The water tables started to fall, and the monitoring data that documented the fall was published by the same federal government that was subsidising the expansion.<\/p>\n\n\n\n
The machine was built to run on a permanent supply. It turns out to be a battery. A very large battery, but one that discharges in one direction only.<\/p>\n\n\n\n
The numbers that don’t lie<\/h3>\n\n\n\n
The aggregate numbers understate the crisis because they average together places that are losing relatively little and places that are losing everything. The Ogallala is not declining everywhere at the same pace, and the question of when irrigated farming ends is not the same question in Nebraska as it is in southwest Kansas. In the regions where depletion is most advanced, that question has already shifted from “if” to “when” \u2014 and the calendar is measured in decades, not centuries.<\/p>\n\n\n\n
Start in Texas.<\/p>\n\n\n\n
The Texas Panhandle sits above the thinnest and least rechargeable part of the aquifer. Water levels have declined substantially since predevelopment \u2014 in the most severely depleted parts, declines exceeding 100 to 200 feet have been recorded, with some areas in the southern High Plains approaching or having crossed economic exhaustion already. The High Plains Underground Water Conservation District No. 1, which manages much of the Texas Panhandle water, has reported average annual declines in recent measurement cycles. Texas retains the rule of capture at the state level, meaning landowners have a near-unlimited legal right to pump whatever lies beneath their property, regardless of the impact on shared supply. Regulation, where it exists at all, is local and voluntary.<\/p>\n\n\n\n
Move north to Kansas. The Kansas Geological Survey’s annual water-level reports track the aquifer’s decline at monitoring wells across the state, and the 2024 data is not encouraging: southwest Kansas fell 1.36 feet between January 2024 and January 2025 \u2014 against a long-term average of 1.67 feet annually, a bad year even by regional standards. Northwest Kansas dropped 1.16 feet in 2024, after a year in which water levels had actually risen slightly. The declines are not stabilizing. They are not following any trend toward equilibrium.<\/p>\n\n\n\n
The structural picture for Kansas was established clearly in a landmark 2013 PNAS study by Steward and colleagues at Kansas State University. Their analysis of the High Plains Aquifer in Kansas found that 30% of the aquifer’s total groundwater volume had already been pumped as of 2010, and that under existing trends, another 39% would be depleted over the following 50 years \u2014 with natural recharge supplying only 15% of current annual withdrawals. Some parts of western Kansas, the study found, had less than 25 years of economically viable irrigation remaining at then-current pumping rates. That study was published thirteen years ago. The pumping has continued.<\/p>\n\n\n\n
Sanderson and colleagues, writing in The Conversation in 2020 (and citing the Steward PNAS data), put the Kansas situation plainly: the effective pumping threshold \u2014 the point at which water tables fall too low to sustain economical irrigation \u2014 has been reached across a significant portion of the Kansas aquifer’s footprint. Steward et al.’s own figures confirm that approximately 30% of the Kansas aquifer’s groundwater volume had been extracted by 2010; the geographic extent of what the researchers termed “Day Zero” conditions has expanded since then.<\/p>\n\n\n\n
Nebraska is different, and the difference is instructive. The state holds roughly two-thirds of the aquifer’s remaining volume. Its Natural Resources Districts have had authority to restrict pumping since the 1970s, and several districts have achieved what amounts to rough stabilization \u2014 not recovery, but a slowing of decline that the Texas and southern Kansas data cannot claim. The contrast is not accidental. It reflects a combination of physical conditions \u2014 a thicker, more generously recharged northern formation \u2014 and governance choices that were made earlier and enforced more consistently. The point is not that the Ogallala is fine because Nebraska is managing. The point is that “the Ogallala is being destroyed” and “the Ogallala is being destroyed everywhere equally” are not the same claim. The southern High Plains is not saved by Nebraska’s relative stability. It is instead a demonstration that the destruction there was a choice, compounded over decades.<\/p>\n\n\n\n
The aggregate figures, from USGS SIR 2023-5143: 286.4 million acre-feet of total depletion from predevelopment to 2019. To render this abstract number concrete \u2014 annual extraction from the Ogallala has historically run at a rate that multiple sources, drawing on USGS volumetric estimates, have compared to many times the annual flow of the Colorado River. The Colorado River, whose own overallocation has triggered years of crisis-level negotiations, is itself running dry. The Ogallala loses that much water every single year, and loses it permanently.<\/p>\n\n\n\n
Economic exhaustion is not the same as empty<\/strong>\n\nMost coverage of the Ogallala depletion frames the crisis as the point at which the water \"runs out\" \u2014 as if an aquifer were a swimming pool that eventually hits the bottom. In practice, the threshold that ends irrigation agriculture arrives much earlier. When the saturated thickness of the aquifer falls below roughly 30 feet, the cost of pumping per acre-foot of water extracted rises above what the crops can return in revenue. The wells don't go dry; the economics do. This distinction matters because it means the crisis arrives before the aquifer is technically empty \u2014 but it also means that \"there's still water down there\" is not a reassurance. There is still oil in mature oil fields too. What matters is whether pulling it out makes economic sense, and the math on the southern Ogallala is increasingly delivering the same answer.<\/code><\/pre>\n\n\n\nThe system that wouldn’t stop<\/h3>\n\n\n\n
The USGS has been publishing water-level measurements for the High Plains Aquifer since the 1930s. The declines were documented in federal reports through the 1950s, the 60s, the 70s. Congress knew. State governments knew. Agricultural economists published studies. The farmers pumping the water knew \u2014 they could read a depth gauge. And still the pumping accelerated through the expansion of centre-pivot technology in the 1970s and 80s, still it continued at scale through the 1990s and 2000s, still it is continuing now. This is not a story about ignorance.<\/p>\n\n\n\n
Three overlapping mechanisms explain it.<\/p>\n\n\n\n
The first is the commons structure. Groundwater in most High Plains states is governed by either the rule of capture or prior appropriation doctrine \u2014 in practical terms, whoever pumps it first owns it. Texas retains the rule of capture at the state level: a landowner pumping water from beneath their property has an essentially unlimited legal right to do so, regardless of how that pumping depletes the shared supply beneath a neighbor’s land. The individual incentive to conserve water that a neighbor can legally extract is approximately zero. This is not a market failure in the usual sense; it is a market design that was never designed to manage a common exhaustible resource, producing the result that economic theory would predict and that the data has confirmed for fifty years. Sanderson and Hughes documented the production-treadmill dynamics driving this in their 2019 Social Problems paper \u2014 the logic that compels individual farmers to pump more as their neighbors pump more, not because they are reckless but because the institutional structure leaves them no rational alternative.<\/p>\n\n\n\n
The second mechanism is federal policy. Sloggy and colleagues, publishing in the Journal of Environmental Economics and Management in 2025, used field-level panel data and spatial discontinuities in crop insurance prices at county borders to establish a direct causal relationship between federal crop insurance subsidies and groundwater extraction. Their finding: a 1% increase in irrigated crop insurance prices leads to a decrease in total groundwater use of approximately 0.5%, irrigated acreage by 0.27%, and irrigation intensity per acre by 0.28%. The federal government currently subsidises approximately 62% of producers’ crop insurance premiums, according to GAO data. The arithmetic is not complicated. By suppressing insurance prices below their market cost, federal policy actively stimulates groundwater extraction.<\/p>\n\n\n\n
And it gets worse. The Revenue Protection insurance product used most widely in the region calculates payouts based on a farmer’s ten-year average yield history, which means farmers in zones where water tables are falling are partially insured against the consequences of irrigation-dependent yields they can no longer fully sustain. The same federal government that funds conservation programs to protect the aquifer also runs insurance programs that reward the extraction those conservation programs exist to prevent.<\/p>\n\n\n\n
Sanderson, Griggs, and Miller-Klugesherz made this argument accessibly in a 2020 Conversation essay \u2014 “Farmers are depleting the Ogallala Aquifer because the government pays them to do it” \u2014 and the Sloggy 2025 paper provides the peer-reviewed empirical confirmation.<\/p>\n\n\n\n
The third mechanism is the price of water itself, which is zero. Pump costs reflect electricity and equipment. They do not reflect the depletion value of a nonrenewable resource. If crude oil carried no extraction cost, no depletion royalty, no reflection of its scarcity \u2014 if you could simply sink a pipe and pump as much as you could sell \u2014 oil fields would be emptied faster than they already are. The Ogallala is being depleted at the pace it is being depleted partly because the water carries none of its own cost. Individual farmers are behaving rationally within the incentive structure they face. The pathology is structural, not individual. That distinction matters for policy, though it doesn’t change the outcome for the aquifer.<\/p>\n\n\n\n
What happens after<\/h3>\n\n\n\n
In Greeley County, Kansas \u2014 one of the westernmost counties in the state, sitting above some of the thinnest remaining saturated thickness \u2014 farmers have been converting to dryland production for years. Not because they want to. Because the water table has dropped below the economic threshold. Wheat and grain sorghum, the dryland crops that can survive on High Plains rainfall, yield dramatically less per acre than irrigated corn or soybeans. Land values in these areas reflect the loss: irrigated farmland commands prices built on irrigated yields, and when the water goes, those prices and the debt structures built on them do not automatically adjust. A forced transition from irrigated to dryland farming is not a soft landing. It is a balance-sheet event.<\/p>\n\n\n\n
The towns follow the farms. Greeley County had 1,534 residents in 2000 and 1,284 by 2020 \u2014 a 16% decline in two decades, making it the least populous county in Kansas. The trajectory is not stabilizing. Small agricultural communities across the western High Plains built their populations and tax bases on the expansion of irrigation agriculture in the postwar decades. As the water-dependent economic base contracts, that process accelerates. Schools close. Main streets empty. This is not a future risk; it is already happening, and the pace is determined largely by how fast the water table falls.<\/p>\n\n\n\n
The High Plains Aquifer region underlies the states that produce approximately 64% of US wheat, with the southern Plains states \u2014 Kansas, Texas, Oklahoma, Colorado \u2014 accounting for roughly 31% of total US wheat production, according to USDA-NASS data. The Ogallala itself provides 30% of all groundwater used for irrigation in the United States. Irrigated production from this region is not interchangeable with irrigated production elsewhere; the scale required to substitute for High Plains output does not exist in the US agricultural system, and developing it would require decades and water supplies that are also under stress. A structural reduction in High Plains irrigated capacity is not a marginal adjustment to the national food supply. It removes a pillar.<\/p>\n\n\n\n
The global exposure is harder to quantify precisely, but the direction is clear. The United States is the world’s largest corn exporter and a significant wheat exporter. Global buyers in the Middle East, North Africa, and parts of Asia have built import dependency on US grain volumes predicated on current production capacity. What makes the Ogallala scenario different from a drought year \u2014 different in a way that commodity markets are not currently pricing \u2014 is that a drought ends. Depleted irrigated acreage does not recover. A temporary supply reduction raises prices until the next harvest. A permanent baseline reduction changes the underlying supply curve. Buyers who have been sourcing US wheat and corn on the assumption of continued volumes from the High Plains are exposed to a structural shift, not a weather event.<\/p>\n\n\n\n
The Saudi precedent<\/strong>\n\nSaudi Arabia provides the most complete example of where the Ogallala's trajectory leads. Beginning in the 1970s, the kingdom used massive subsidies and deep well technology to pump its own fossil aquifer \u2014 the Saq Aquifer, a nonrenewable formation in the northwestern part of the country \u2014 and achieved wheat self-sufficiency by the 1990s. It was briefly the world's sixth-largest wheat exporter. Then, as the water tables collapsed and extraction costs spiralled, the government began phasing out the program. By 2016, Saudi Arabia had terminated its domestic wheat production entirely and become a net importer. The transformation took roughly forty years from peak production to complete withdrawal. The aquifer, once effectively empty of accessible water, offered nothing to reverse the decision. The US is not Saudi Arabia \u2014 the political and market structures are different, the aquifer is vastly larger, the adjustment will be slower and messier rather than state-directed. But the sequence \u2014 a fossil aquifer exploited to support an agricultural model, then depleted to the point where the model collapses \u2014 is not a hypothetical. It is a completed historical example, playing out now on a larger scale.<\/code><\/pre>\n\n\n\nThe response to this situation, at the scale the depletion requires, has been nothing. Federal conservation programs \u2014 the USDA’s Environmental Quality Incentives Program, the Conservation Reserve Program \u2014 exist and fund some reduction in irrigated acreage. Kansas has Local Enhanced Management Areas that allow producers within a groundwater management district to agree, voluntarily, to reduce pumping. Nebraska’s Natural Resources Districts have achieved meaningful results in some areas. These are not nothing. But they are nowhere near commensurate with the scale of what is being lost, and they are competing against the same federal insurance architecture that Sloggy et al. demonstrated is actively stimulating extraction. The USGS keeps publishing the measurement data. The water tables keep falling. The pivot systems keep turning.<\/p>\n\n\n\n
Out in the field in western Kansas, the centre-pivot system completes another circuit. It will complete another tomorrow. The farmer operating it is not a villain \u2014 the system was built, over decades of deliberate policy, to make this the rational choice. The Ogallala does not run a countdown clock visible from above. It simply gets shallower, year by year, depth measurement by depth measurement, until one day the pump pulls sediment and the calculation changes. The monitoring data for 2024 has already been published. The numbers went the wrong way again.<\/p>\n\n\n\n
Nobody with the power to change the fundamental structure of what is happening has made a serious attempt to do so.<\/p>\n\n\n\n