On July 1, 2021, the heat dome that had been sitting over the Pacific Northwest broke. The National Weather Service let its excessive-heat warnings expire. The forecasts changed and the coverage moved on.
The Columbia River took longer. At Bonneville Dam, the five-day average water temperature on July 7 was 68.0°F, against a recent five-year average of 65.6°F for that date. Two weeks after the air cooled, Yakama Nation biologists surveying adult sockeye near Priest Rapids found most of the fish they examined covered in skin lesions — a condition U.S. Fish and Wildlife Service specialists attributed to warm water. Sockeye do best at or below 60°F. The river was running in the low 70s.
Oregon's fisheries agency was still reporting above-average temperatures at Bonneville into mid-August. Six weeks after the weather app said the heat wave was over, the water hadn't finished with it.
A study published in Nature Geoscience in July 2026 attaches numbers to something fisheries biologists and water-treatment operators have understood for a long time without being able to cite it. Zhou et al. reconstructed daily water temperatures and dissolved oxygen levels across 796 river basins in the United States and Central Europe, 1981 through 2019. Where an atmospheric heat wave and a river heat wave overlapped, the river's outlasted the air's in 99.6 percent of cases. The average overshoot was 2.67 days. In the extreme cases, more than 12.
Air is thin and gives up heat fast. Water is dense, takes days to absorb heat, and releases it on a schedule set by flow volume, depth, shade, groundwater inputs, and whatever is happening upstream — none of which respond to a high-pressure ridge moving east. When atmospheric heat drove the river event rather than the river heating on its own, water temperatures rose an additional 16 percent and dissolved oxygen dropped a further 2.9 percent.
Dissolved oxygen is the part that does the killing. Oxygen dissolves less readily in warm water, and fish, aquatic insects, and the microbes that process nutrients all need it at concentrations that hot water physically struggles to hold. EPA's freshwater criteria set a 7-day mean of 9.5 milligrams per liter for cold-water species during early life stages, easing to a 1-day minimum of 3.0 mg/L for warm-water species later in life. States can adopt their own defensible standards. The direction doesn't change: warmer water, less oxygen, less life.
So the alert expires, the air cools, and the river keeps running hot with the oxygen still down. What happens then depends on whose clock you're reading.
The biologist's clock
Oregon's district fisheries biologists don't wait for the Weather Service to tell them a river is dangerous. They work from gauge temperatures and from what the fish are doing. The Fish and Wildlife Director holds emergency authority to regulate the taking of wildlife during a crisis, and biologists make the calls reach by reach: what the thermometer says, whether salmon and steelhead are stacking up in the cold water at tributary mouths, how the fish look.
In August 2019 the agency prohibited angling within 200 feet of tributary mouths along a stretch of the Umpqua. The gauge at Elkton read 79°F. District fisheries biologist Greg Huchko put the threshold at 68°F — the point past which salmon and steelhead start struggling. The restriction held from August 1 through September 30.
During the 2021 heat dome, ODFW imposed afternoon fishing restrictions across several Willamette basin rivers beginning July 1, the day the principal atmospheric heat wave ended and the day after the excessive-heat warnings came down. Then came the slower work of taking them off. On July 16 the agency lifted restrictions on two specific reaches — below Big Cliff Dam on the North Santiam and below Foster Dam on the South Santiam — because those dams release cold water drawn from deep in their reservoirs, which kept the stretches downstream survivable. Other reaches stayed closed. Their daytime temperatures hadn't come down.
Washington's framework is built the same way: the WDFW director can close or shorten a season by emergency rule when biological data or unforeseen climate conditions put a population in jeopardy.
Huchko made the Umpqua call. Assistant district fish biologist Alex Farrand explained the partial Santiam reopening. These are people who have spent years watching rivers hold heat after the air has let go of it, learning which reaches cool first, which tributary mouths draw stressed fish, which dam releases can buy a stretch of water a few degrees. The discretion exists in statute because somebody has to be making judgment calls in the days after the atmosphere clears. Those calls reach the public through press releases and regulation notices, not through the phone.
The operator's clock
Inside a treatment plant drawing from a warm river or reservoir, the problems don't end when the air does.
Warm water speeds up the formation of disinfection byproducts — compounds like trihalomethanes that appear when chlorine reacts with organic material in the source water. Federal limits cap total trihalomethanes in finished water at 0.080 mg/L. Warm water also causes chlorine to dissipate faster, which makes it harder to keep a detectable amount of disinfectant in the water all the way through miles of distribution pipe. Both things happen at once, pulling the operator in opposite directions: more byproducts forming, less disinfectant staying put.
The Portland Water Bureau adds chlorine and ammonia to form chloramine, a longer-lasting disinfectant, and reports increasing chlorine slightly in warmer months. Seattle uses ozonation to handle the seasonal taste-and-odor compounds that come with algal blooms driven by sunlight and temperature. Both adjustments are seasonal and specific to the plant. No universal source-water temperature obliges every operator to act.
When dissolved oxygen falls in a reservoir, the chemistry moves again. Low-oxygen conditions at the bottom let iron and manganese leach out of the sediments and into the water the plant is drawing, which arrives as color, taste, and filtration trouble. A study of one Nova Scotia plant measured raw-water dissolved oxygen as low as 1.4 mg/L while the reservoir was stratified — layered, with the deep water cut off from the surface — and iron and manganese concentrations spiking to roughly 2,300 micrograms per liter.
The operator's workday resets when the source water cools, which, per Zhou et al., is days after the forecast says otherwise.
The outfitter's clock (if there is one)
On most rivers, nobody tells the outfitter the water is too warm to float.
Steamboat Springs, Colorado, is the exception people cite. The city can suspend commercial outfitting on the Yampa when flow drops below 85 cubic feet per second, when water temperature stays above 75°F for two consecutive days, or when average dissolved oxygen falls below 6 mg/L. Temperature triggered the closure in July 2025. In June 2026, low flow did. Commercial operators have to stop. Private tubers and boaters get voluntary restrictions — which means the person floating the same water on their own inner tube, with no guide reading the gauge for them, is under no rule at all.
Reopening isn't automatic either. The city consults Colorado Parks and Wildlife, weighs current readings against the forecast, and factors in how much stress the fish have already absorbed. Backdoor Sports owner Mike Welch said the city's parameter updates help the company plan. Fly-fishing operations moved trips to other water during closures, adding drive time and concentrating pressure on different stretches.
Steamboat Springs has a closure protocol because someone there decided the Yampa needed one. Elsewhere the go/no-go call rests on the forecast, the booking calendar, and maybe how the water looks — none of which register what the river is doing thermally.
A few hundred miles away, another outfitter checks the same clear forecast and loads the boats. That outfitter may have a USGS flow gauge upstream, if the stretch has one. Flow tells you whether the river is navigable. It doesn't tell you the water is 74°F with oxygen sliding past what trout can tolerate. The nearest real-time temperature reading may be on a different river.
What the gauges can and can't see
USGS runs more than 12,000 continuous surface-water monitoring sites nationally. Most of them measure stage — how high the water is — or streamflow. The subset reading water temperature in real time is much smaller: roughly 2,000 stream sites in the agency's water-quality network, and not all of those measure temperature, and not all the temperature records transmit live.
In September 2025, USGS and partners installed 14 new gauges in remote spring Chinook spawning reaches of the Clackamas, Santiam, Molalla, McKenzie, and Middle Fork Willamette. Some transmit in real time. Others write to an internal chip that staff collect on site visits. A gauge logging temperature every 15 minutes to a chip in the woods is producing science, but nothing a fisheries biologist or a treatment operator can act on while the river is hot.
Zhou et al. reconstructed their temperature and oxygen records with deep learning models trained on irregular observations — a method that worked better for temperature than for dissolved oxygen. The study could see the lag because it built a tool capable of seeing it after the fact. In real time, on most rivers, the lag doesn't show up anywhere.
The signal that reaches the public
The Weather Service warning is atmospheric, and it expires when atmospheric conditions improve. That expiration is the loudest thing most people hear about a heat wave — the notification, the headline, the general sense that it's over.
Everything downstream runs on something else. The fisheries biologist watches gauge temperature and fish behavior. The treatment operator watches source-water chemistry. The Steamboat Springs closure turns on a threshold that requires two consecutive days above 75°F, so it lags the atmosphere by design. On most rivers the recreation decision turns on nothing measured at all.
These people also don't have the same authority or the same pressures. The biologist can restrict fishing but absorbs public frustration when the restriction outlives the weather story. The operator adjusts chemistry on a schedule nobody outside the utility sees. The Yampa outfitter works under a municipal protocol that exists because one city wrote one.
There's real uncertainty in how much the study's numbers translate. Zhou et al. defined their heat waves relative to each basin's own historical range — hot for that river, not hot for a fish or a filter. Whether 2.67 days understates or overstates the functional lag, the time until a particular species or treatment process is no longer under strain, depends on the river and the organism, and the paper doesn't settle it. The average describes a pattern across hundreds of basins. Any single river's behavior depends on its depth, flow, shading, groundwater, and how it's managed upstream. The Columbia in 2021 stayed hot for weeks, with heat compounded by drought, thin snowpack, and dam operations. A shallow, spring-fed creek might shed it in an afternoon.
What holds is the asymmetry in who knows. The professionals working in the days after the all-clear have thermometers, fish, decades of pattern recognition, and in some cases the authority to close the water. The family loading the car on Saturday morning has a notification that the heat wave ended.
-
Compound heatwaves are tripling: The same Nature Geoscience study found that atmospheric-riverine compound heatwaves increased from roughly 0.76 to 2.3 events per year between 1981 and 2019, and under a high-emissions scenario, 98.5 percent of river heatwaves would coincide with atmospheric heatwaves by 2100.
-
Ecological recovery runs on multiple clocks: A 2026 synthesis of the 2021 western heat dome found that fire activity increased 395 percent in the week after the event, while shoreline species like rockweed and mussels showed only partial recovery through 2024, shaped by later predation and salinity conditions.
-
Heatwaves leave a fire tail: A June Science Advances study found that heatwaves and the five days following them accounted for only 12–15 percent of fire-season days but 42 percent of western U.S. burned area during 2001–2024, with forest fire activity peaking about seven days after heatwave onset.
-
July 2026 broke the record on warm nights: NOAA data reported by AP show that the contiguous U.S. set its hottest-month record in July 2026, driven more by warm minimum temperatures than by daytime maxima — the same overnight pattern that keeps rivers from cooling.