This story was originally published by High Country News.

Floyd Dominy, the commissioner of the federal Bureau of Reclamation in the 1960s, was largely responsible for the construction of Glen Canyon Dam on the Colorado River. In 1963, when the dam was completed, he could not have foreseen the climate situation we find ourselves in today, with declining snowpack, record-high temperatures and alarmingly low water levels in Lake Powell, year after year. But he and his engineers could have, and should have, foreseen that the way they designed the dam would leave little room to maneuver should a water-supply crisis ever impact the river and its watershed.

Indeed, a state of crisis has been building on the Colorado for decades, even as the parties that claim its water argue over how to divide its rapidly diminishing flows. Lately, things have entered a new and perilous phase. Last Nov. 11 was a long-awaited deadline: Either the states involved — California, Arizona, Nevada, Utah, New Mexico, Colorado and Wyoming — would have to agree on a new management plan, or else the federal government would impose its own, something none of the parties would welcome. Meanwhile, the 30 tribes that also hold claims to the river have historically been and continue to be excluded from these negotiations.

That deadline came and went, and instead of acting, the government punted, this time to Feb. 14. Nobody was surprised: Unmet deadlines and empty ultimatums have been business as usual on the river for years. Decades of falling reservoir levels and clear warnings from scientists about global warming and drought have prompted much hand-wringing and some temporary conservation measures, but little in the way of permanent change in how water is used in the Colorado River Basin. This month’s deadline also came and went.

For decades, the seven Basin states have used more water than the river delivers by drawing their entitlements from surpluses banked in reservoirs during the wet 1980s and ’90s, chiefly in Lake Mead and Lake Powell. Never mind that those entitlements were based on an over-estimate of river flows in 1922, when the Colorado River Compact was established, rendering the “paper” water of the entitlements essentially a fiction, not to mention a source of continual conflict. That savings account has now been drained: Mead and Powell are each below 30 percent full, and the trend is steadily downward. Global warming has only accelerated the decline: So far this century, the river’s flow has fallen 20 percent from its long-term annual averages, and scientists forecast more of the same as the climate continues to heat up.

Meanwhile, the physical infrastructure that enables Colorado River water management is on the verge of its own real and potentially catastrophic crisis — and yet Reclamation has barely acknowledged this, with the exception of an oblique reference in an unposted technical memorandum from 2024. The falling reservoir levels reveal another, deeper set of problems inside Glen Canyon Dam, which holds back the Colorado and Lake Powell. The 710-foot-tall dam was designed for a Goldilocks world in which water levels would never be too high or too low, despite the well-known fact that the Colorado is by far the most variable river in North America, prone to prodigious floods and extended droughts. But the Bureau, bursting with Cold War confidence — or hubris — chose to downplay the threat. In the record-breaking El Niño winter of 1983, the Bureau almost lost the dam to overtopping, due to both its mismanagement and its design, because the dam lacks sufficient spillway capacity for big floods. Only sheets of plywood installed across its top and cooler temperatures that slowed the melting of that year’s snowpack saved Glen Canyon Dam.

Today, the dam is threatened not by too much water but too little. In March 2023, the water level of Lake Powell dropped to within 30 feet of the minimum required for power generation, known as “minimum power pool.” At 3,490 feet above sea level, minimum power pool is 20 feet above the generators’ actual intakes, or penstocks, but the dam’s eight turbines must be shut down at minimum power pool to avoid cavitation — when air is sucked down like a whirlpool into the penstocks, forming explosive bubbles which can cause massive failure inside the dam.

Even more worrisome is what would happen next. At minimum power pool, the penstocks would have to be closed, and the only remaining way to pass water through the dam is the river outlet works, or ROWs: two intakes in the rear face of the dam leading to four 96-inch-diameter steel pipes with a combined maximum discharge capacity of 15,000 cubic feet per second. However, the ROWs, also known as bypass tubes, have a serious design flaw: They are unsafe to use for extended intervals, and start to erode when the reservoir is low.

In 2023, when the ROWs were used to conduct a high-flow release into the Grand Canyon at low-reservoir levels, there was, in fact, damaging cavitation, and the Bureau has warned that there would likely be more in the event of their extended use. In practice, safe releases downstream may only be a fraction of their claimed capacity — and if the tubes begin to experience cavitation, flows may need to be cut off entirely. Such a scenario would compromise the dam’s legal downstream delivery requirements, or, to put it bluntly, its ability to deliver enough water to the 25 million people downstream who rely on it — as well as the billions of dollars’ worth of agriculture involved. This means that Lake Powell — and with it, the entire Colorado River system — is perilously close to operational failure.

If reservoir levels drop to the ROWs’ elevation of 3,370 feet above sea level, Lake Powell would reach “dead pool,” where water would pass through the dam only when the river’s flow exceeded the amount of water lost to evaporation from the reservoir. No other intakes nor spillways exist below the ROWs. There is no “drain plug.” Yet there is more dam — 240 feet more before the bottom of the reservoir, effectively the old riverbed. This not-insignificant impoundment — about 1.7 million acre-feet of water — would be trapped, stagnant and heating in the sun, prone to algal blooms and deadly anoxia. The lake would rise and fall wildly, as much as 100 feet in a season, because of the martini-glass shape of Lake Powell’s vertical cross section.

Insufficient or no flows through Glen Canyon Dam would be a disaster of unprecedented magnitude, affecting vast population centers and some of the biggest economies in the world, not to mention ecosystems that depend on the river all the way to the Gulf of California in Mexico. The Lower Basin states of California, Arizona and Nevada warned as much in a recent letter to Interior Secretary Doug Burgum, saying that Reclamation’s failure to mention the dam’s plumbing problems in its current environmental impact statement for post-2026 operations is against federal law. The letter reads: “Addressing the infrastructure limitations may be the one long-term measure that would best achieve operation and management improvements to the Glen Canyon Dam.”

To date, however, the Bureau has made no formal response.

One thing is clear: Glen Canyon Dam will need to be modified to meet its legal and operational requirements. In the process, the health of the ecosystems in Glen Canyon, above the dam, and in Grand Canyon, below it, must be considered. The best way to avoid operational failure and the economic and ecological disasters that would follow is to re-engineer the dam to allow the river to run through it or around it at river level, transporting its natural sediment load into the Grand Canyon.

As it happens, Floyd Dominy himself provided us with a simple and elegant plan for how to do it. In 1997, the former commissioner sketched on a cocktail napkin how new bypass tunnels could be drilled through the soft sandstone around the dam and outfitted with waterproof valves to control the flow of water and sediment. What it prescribes is treating the patient — the Colorado River, now on life support — with open-heart surgery, a full bypass. Dominy’s napkin, which he signed and gave to my colleague Richard Ingebretsen, the founder of Glen Canyon Institute, is effectively a blueprint for a healthier future for the Colorado River and the people and ecosystems that depend on it.

But the window for action to avoid dead pool is dauntingly narrow and closing fast, especially given the time that would likely be required for the government to study, design and implement a fix. The Trump administration’s gutting of federal agency expertise and capacity adds yet more urgency to the issue. The feds and the basin states need to look beyond the water wars and start building a lasting, sustainable future on the Colorado River.

Wade Graham is a historian and writer based in Los Angeles. He is the author of books on landscape, urbanism and environmental history. Since 1999, he has been a trustee of the Glen Canyon Institute.