In the fall of 1922, Aldo Leopold and his brother Carl paddled a canoe into the delta of the Colorado River and found a wilderness of green lagoons, a hundred channels braiding through two million acres of cottonwood and willow, egrets standing in the shallows, deer in the cane, and somewhere beyond the firelight the jaguar they never saw and never stopped feeling. Leopold wrote about that trip a quarter century later and confessed that he never went back. He preferred the delta as memory held it, because he understood what was coming for it. He was right to stay away. In November of that same year, in a lodge outside Santa Fe, commissioners from seven states signed the Colorado River Compact and divided the river on paper. The canoe and the signature belong in the same paragraph because one of them canceled the other.

A century later the delta Leopold paddled is a cracked plain. For most of the past sixty years the Colorado has died in the sand south of Yuma, miles short of the Gulf of California, a river with a mouth and no arrival. The Cocopah, whose name means the river people, fished that estuary for centuries; when the water stopped coming the fishery collapsed, and the people named for the river outlived the river’s ending.
Landlocked is the correct word for a river severed from its sea, and the tense matters, because this was never weather. No river loses its ocean by accident. The severing was performed upstream, by name-able people signing datable documents, and forty million of us now live inside their signatures.
And the landlocking is happening a second time, higher up, inside the river’s own plumbing.
On the first of this month Lake Mead fell to 1,038.82 feet above sea level, the lowest surface elevation since the reservoir first filled behind Hoover Dam; it sits this week at about 26 percent of capacity.
Lake Powell broke its own record in late August at 3,518.79 feet, roughly 22 percent full, after a winter that delivered 36 percent of average inflow, and this year the Bureau of Reclamation cut Glen Canyon Dam’s annual release by 1.48 million acre-feet to protect the upstream pool, a decision the Bureau itself acknowledged would accelerate Mead’s decline.
The numbers below are worse. Beneath 3,490 feet, Powell’s turbines begin to starve. Beneath 3,370 feet, dead pool, water can no longer pass Glen Canyon Dam by gravity at all. Mead’s dead pool is 895 feet. A reservoir at dead pool still holds millions of acre-feet; it simply cannot give one of them downstream. Water, present and stranded, sealed behind concrete we poured ourselves. That is a river landlocked twice, once at its mouth and once inside the machines built to improve it.
How did this happen? It began with the arithmetic of 1922, because every cruelty since is that arithmetic coming due. The commissioners at Bishop’s Lodge, Herbert Hoover in the chair, split fifteen million acre-feet a year between the upper and lower basins, and a 1944 treaty later promised Mexico a million and a half more. Behind them lay gauge records from the wettest run of years the basin had seen in centuries.
The government’s own hydrologist, E.C. LaRue of the Geological Survey, spent that decade warning that the river could not supply every project being promised. He was correct, and it changed nothing, because there were farms to sell, a Boulder Canyon bill to pass, an empire to irrigate. Tree-ring science later put the river’s long-term average closer to fourteen million acre-feet. Since 2000 the flow has averaged nearer twelve and a half, about a fifth below the last century’s mean, and the best climate physics finds the river losing on the order of nine percent more for every degree Celsius of warming.
So the founding document of Western water obligates sixteen and a half million acre-feet, more or less, from a river that produces twelve. We have spent a hundred years spending the difference, first out of the delta, then out of the reservoirs, and the reservoirs are nearly gone. Then came the machinery for spending it. Hoover Dam rose in 1935 and Lake Mead behind it. Glen Canyon closed its gates in 1963 to bank Lake Powell as the Upper Basin’s insurance against its delivery obligations, and in the first years of filling, barely any water passed below at all.
The delta’s allocation, in all that paper, was zero acre-feet, a number nobody bothered to write down because nobody believed the sea was owed anything. By the late 1960s the river had stopped arriving at the Gulf in any regular way.
Two great reservoirs now lie open on the desert like pans, and together with the canals they lose well over a tenth of the system to evaporation and seepage every year, several entire Nevada allocations boiled into the sky, off an artificial lake we named, with a straight face, for the one man who told us all of this would happen. I will come back to him.
Where has all that water gone? That question finally has a full and audited answer, and it is the most damning document in the basin. In 2024 a team of twelve researchers led by Brian Richter published, in Communications Earth and Environment, the first complete accounting of every drop the Colorado moves, from the headwaters through the Gila to the dry line in Mexico, covering 2000 through 2019.
Total consumption averaged about 19.3 million acre-feet a year against a river producing far less, which is how the basin overdrew its account by ten percent a year on average, running a deficit in sixteen of twenty-one years while the reservoirs quietly covered the checks. Irrigated agriculture consumes 52 percent of everything the basin uses and 74 percent of the water people directly divert. Within that, one category, cattle-feed crops, alfalfa and grass hay, takes 46 percent of all direct human use, close to two-thirds of the agricultural share, a third of the entire river. Every city, suburb, factory, casino, and lawn from Denver to Tijuana combined draws about 18 percent.
Reservoir evaporation claims another tenth. Set those shares side by side and the sentence writes itself: the largest single consumer of the Colorado River is a cow. A portion of the hay never even meets an American cow. Compressed and containerized, it crosses the Pacific to feed dairy herds abroad, the basin’s blood exported in bales while the states that bled it hold press conferences about shower heads.
The Law of the River then decides who suffers in which order, and the order is the cruelty. Western water runs on priority, first in time, first in right. The Imperial Irrigation District’s claims date to 1901, and its entitlement runs to some 2.6 million acre-feet a year, nearly nine times what the entire state of Nevada receives.
The Central Arizona Project, the canal that carries the river to Phoenix and Tucson, agreed in 1968 to stand last in line among the big users as the price of getting built at all, so more than five million city dwellers rank behind flood-irrigated forage when the cuts arrive. And beneath everyone, holding the oldest rights and the fewest pipes, stand the thirty federally recognized tribal nations of the basin, whose reserved rights amount on paper to about a quarter of the river’s flow. Nobody invited them to Bishop’s Lodge in 1922.
A century later, about a third of the homes on the Navajo Nation still haul their water in barrels while the river of their maps runs past toward the hayfields. Seniority was supposed to mean fairness through time. It has come to mean that thirst flows downhill toward the people who caused the least of this.
Which brings us to the present administration of the disaster, and I use the word administration deliberately, because governance would be too generous. The rules that have run the reservoirs since 2007 were titled interim guidelines; they have now been interim for nineteen years, and they expire this December alongside the 2019 drought plans and the current implementing agreement with Mexico.
The seven states were given years to negotiate a successor and a final deadline this past winter. They produced competing proposals and no agreement. The upper and lower basins could not agree on whose water the shortage is, so in January the federal government published a draft plan with no preferred alternative, an environmental document shaped like a shrug, and then finished the job alone: a final impact statement on July 31, a Record of Decision on August 21 adopting what it calls a Decision Framework, and a first set of operating guidelines covering only 2027 and 2028, under which the Lower Basin absorbs the reductions it had proposed for itself and the Upper Basin faces no mandatory cut at all. Years three through ten remain a promissory note.
Lawyers across the basin are already pronouncing the phrase original jurisdiction, which is how water people say Supreme Court. After a century of the Law of the River, this is the estate: two record-empty reservoirs, no agreement among the heirs, and a framework its own signers praise as “a level of stability,” which is what you call a tourniquet when you cannot say cure.
Is there a real way out? Yes, and it is dull, and its dullness is the reason we keep refusing it. The way out is subtraction. Richter’s team ran the number: stabilizing Mead and Powell requires cutting basin consumption 22 to 29 percent, between 2.4 and 3.4 million acre-feet a year, permanently. No pipeline. No miracle membrane. Arithmetic.
The only open question is where the subtraction lands, and the ledger answers it, because nobody conserves their way out of a cow pasture by shortening showers.
Weigh the options honestly. Retiring and fallowing cattle-feed acreage is effective because that is where the water is: trim the forage irrigation by a quarter and you recover more water than every city on the river consumes, and it must be paid for like the permanent transition it is, at prices that let a farm family and a farm county exit whole, because the growers did what the law invited them to do and the bill belongs to all of us.
Converting the fixed acre-foot entitlements of 1922 into percentages of actual flow is effective because it ends the paper river forever and makes the law tell the truth about the sky; every state fears it because reopening the compact shrinks somebody’s number in public, though no formal amendment is even required; the shortage agreements already in force prove the states can bind themselves to conditional numbers whenever they are cornered.
Charging reservoir evaporation to the users who incur it is effective because it is honest bookkeeping of a tenth of the system that currently belongs to no one; California resists because California would pay the most.
Completing and funding the tribal water settlements is effective because it is justice and flexibility in the same instrument, wet water for nations that hold senior paper, and leasable supply that can move where the need is.
Against these stand the fantasies, and they deserve the name. A pipeline from the Mississippi is not effective because it fails on energy, on cost, on treaty, on Congress, and on the fifteen or twenty years of construction it would demand while the reservoirs are scheduled to fail within this decade. Towing icebergs is a joke drought tells about itself. Desalination is not effective as rescue because its plausible scale serves a coastal city at the margin, useful for trading paper, useless for growing hay.
My recommendation, for whatever a writer’s recommendation weighs against a hundred years of concrete: cap the river’s obligations at the river’s actual production, pay agriculture’s exit generously and forever, write evaporation and climate into the Law of the River in numbers, finish the tribal settlements, and grant the delta a standing water right, so that for the first time since 1922 the mouth of the river appears in its own ledger.
Restoration, by the way, is proven. We proved it ourselves, then stopped. In the spring of 2014, under an agreement called Minute 319, the United States and Mexico opened the gates at Morelos Dam and released a pulse of about 105,000 acre-feet, less than one percent of the river’s former annual flow, across eight weeks. The river found the sea in mid-May. In San Luis Río Colorado, people who had lived whole lives beside a dry channel walked into moving water; children who had never seen the river played in it while grandparents stood on the banks and wept.
Cottonwood and willow seed germinated on the wet sand within weeks. Small binational restoration reaches have been kept green since then on trickles and volunteers. The river remembered its way to the ocean after half a century. It was only ever waiting for us to get out of the way.
Now open the other door and look at what the schedule says if we do nothing. Reclamation’s own projections put Powell below 3,510 feet during the coming water year, into what the agency calls the Low Elevation Infrastructure Protection Range, and the pessimistic runs put it beneath reliable power generation as soon as this fall. At that point the dam that electrifies about five million customers across seven states stops paying its way, replacement power gets bought on the open market, and the first swollen bills land on the rural cooperatives and tribal utilities least able to carry them.
Below power pool, Glen Canyon can pass water only through four river-outlet tubes eight feet across, plumbing installed for the construction era, never engineered to carry a basin’s entire legal obligation year after year, its capacity shrinking as the lake above it falls and its linings already showing damage in testing.
Mead’s endgame is blunter. At 895 feet, Hoover Dam becomes a wall with a lake behind it, and by plain physics nothing flows to anyone below.
Las Vegas, to its credit, believed the projections a decade ago and spent well over a billion dollars on a third intake and a deep pumping station that can drink the lake nearly to the mud, so the city the rest of the country mocks for its fountains will be the one still pouring while the canals of the pious go dry. Arizona’s canal would shut down in the order the contracts dictate. California’s seniority would keep the Imperial and Yuma fields green longest, and there sits the grim December punchline at the bottom of the system: the last certain water out of a dying river will be growing winter lettuce, since those two valleys supply on the order of ninety percent of the continent’s cold-season leafy greens, with export hay ripening beside it.
And the people, since the people are the point and the people are the plural the paper always forgets. First the farmworkers of Yuma and Imperial and the Mexicali Valley, hundreds of thousands whose winter wages are irrigation made visible, gone when the fields go. Then the wells, because when surface water fails the West mines its groundwater, and mined groundwater fails from the bottom of the economy upward, domestic wells before corporate ones, trailer parks before dairies, the land itself sinking and cracking canals as the aquifers compact, as central Arizona’s fissured earth already testifies. The grocery store comes next, in February, everywhere, because a Wednesday salad in New Jersey is Colorado River water wearing a disguise.
After that, the small towns that exist because a headgate exists, hollowing the way farm towns hollow, school by school. And last the quiet migrations nobody will call climate migration, because the movers will call it retirement, or opportunity, or family. Running underneath it all is the sorting that shames us.
The Cocopah lost their river first, by decades. The tribal nations with the oldest rights and the newest pipes are told to wait behind cities junior to them in law. Mexicali’s farmers stand at the end of a river that ends before it reaches them. The arrangement pays its seniority holders handsomely to take less of what the sky no longer sends, while the people who took least from this river are billed first for its death. That is the cruelty as designed, operating as designed.
So name the responsible, because this condition was constructed, and construction has contractors. The commissioners of 1922, warned by their own hydrologist and signing anyway, because the empire could not wait for honest data. A Bureau of Reclamation a century deep in building magnificent plumbing for a river that does not exist, still filing the shortfall under weather. Growth boosters of Phoenix and Las Vegas and the Front Range, selling desert lots against a paper river and naming the subdivisions for whatever they bulldozed.
Hay barons and their bankers, shipping a landlocked river out of the Port of Long Beach in bales. Politicians of both parties who renewed an interim arrangement for nineteen years because the true number would have cost one election, any election, theirs. And the state negotiators of these past three years, who watched the deadlines pass like passengers watching a fuel gauge, each certain some other hand would land the plane, until the federal government seized the yoke in August and produced a two-year patch.
None of them wanted a dead delta or a stranded reservoir. All of them accepted both, year after year, as the reasonable price of never being the one to say the number out loud. Intent is their alibi. Acceptance is the crime.
In October 1893, at the International Irrigation Congress in Los Angeles, a one-armed geologist stood up before the boosters of the whole irrigation gospel and ruined the banquet.
John Wesley Powell had run the Grand Canyon and measured the arid lands for two decades, and he told the hall: “I tell you, gentlemen, you are piling up a heritage of conflict and litigation over water rights, for there is not sufficient water to supply the land.” They hissed and booed him off the program, and he was dead within nine years, and we answered his warning in the most American way available: we built the second-largest reservoir in the country on the river he measured and named the evaporation after him.
Every acre-foot that boils off Lake Powell this summer is his sentence, still finishing. South of Yuma there is a line in the sand where a river used to become an ocean.
Twice a day the tide still comes up from the Gulf looking for its river, on schedule, faithful as ever, and finds sand. Rivers keep their appointments. We are the ones who stopped showing up.
Leave a ReplyCancel reply