Scientists Say Something Strange Is Happening Across the Great Lakes

In a first-of-its-kind survey, a team of scientists from more than a dozen US and Canadian institutions set out across the Great Lakes this month. They traveled on snowmobiles, ATVs, and airboats, all to collect valuable data. Look down on the middle of North America and you see five enormous lakes carved out by ice, connected across an international border, and draining through a single narrow exit to the Atlantic. For generations, a system this massive was assumed to change slowly. But in 2026, that assumption started getting harder to defend. So how can a freshwater system this enormous begin changing this quickly? And what happens when those changes start affecting one another? The basin that drains through one door. Start with a fact that surprises most people who live on these shores. Lake Michigan and Lake Huron are not two lakes in any physical sense. They are joined at the Straits of Mackinac by a channel deep enough and wide enough that water levels equalize between them. Engineers treat them as one body and publish one elevation number for both. A heavy snowmelt over Wisconsin registers on a dock in Georgian Bay, Ontario. That connectivity is the organizing principle of everything that follows. Heat, nutrients, sediment, pollutants, and living organisms all travel the same gradient toward the same outlet. There are no isolation compartments in a system with one door. Something that starts in a tributary in Ohio ends up in the water column of a lake that borders four states and a Canadian province. It is worth understanding how the numbers you hear are actually produced because they are easy to misread. Official lake levels are still water surface elevations averaged across an entire lake and reported as monthly averages, not daily readings. The lake surface you actually see is never that flat. Wind pushes water toward one shore and pulls it off another, sometimes by more than a foot in a strong event. So when the Army Corps reports that a lake sits three inches above where it stood a year earlier, that is a basin-wide accounting figure, not a promise about your beach. Those figures still drive real decisions. Freighters set cargo weight against expected channel depth. Marinas dredge or don't. Municipal intakes, shoreline erosion projections, and hydropower schedules all key off the same monthly bulletins. The connectivity has a nutrient dimension, too, and it is severely lopsided. The Maumee River, draining farmland in Ohio and Indiana into the shallow western end of Lake Erie, delivers somewhere between 5 and 10 percent of the water entering that lake. It delivers roughly half of the phosphorus. That is 5 to 10 times its share of the load, arriving through one channel into the shallowest, warmest basin in the chain. For most of the last century, this arrangement moved slowly enough that planners could work from long-term averages and be roughly right. One recent season broke that assumption in both directions inside six months. A winter that ran low. A spring that ran over. Through the winter of 2026, the forecast read like a story about scarcity. The Army Corps of Engineers projected Michigan and Huron running 11 to 12 inches below their long-term average straight through August. Lake St. Clair, the shallow connector between Huron and Erie, was expected to sit 5 to 7 inches below normal. Shipping operators were already trimming cargo, because every inch of missing draft is tonnage you cannot load. The cause was not a shortage of rain. It was evaporation. The preceding autumn had been dry, and hot summers had loaded the lakes with stored heat. Warm water under cold, dry air loses volume fast. Then, the balance flipped. Superior's watershed took in about 155 percent of its normal precipitation in February. A blizzard in March dumped more than four feet of snow across parts of Michigan's Upper Peninsula, and April precipitation came in well above average across the basin. Snowpack is just stored precipitation waiting on a melt curve, which is why a single storm can show up as a lake-wide elevation change weeks later. Michigan and Huron rose 10 inches between March and April, reaching 578.58 feet, three inches higher than the same point the year before. Superior climbed three inches to 601.12 feet, four inches above its prior year mark. In its updated outlook, the Corps cited record-high water supplies across the basin. But the lakes did not all move the same way. And this is where a simple narrative falls apart. Lake Erie also rose 10 inches, up to 571.65 feet, and still finished four inches below where it had been a year earlier. Projections had it staying one to three inches lower than the previous year through August. Same basin, same storms, opposite year-over-year direction. That divergence comes down to volume and residence time. Erie is shallow and flushes quickly, so it responds sharply to short-term supply and sheds water fast through the Niagara. Superior is deep and slow, holding its balance across seasons rather than weeks. Net basin supply is the accounting behind all of it. Precipitation plus runoff minus evaporation minus outflow. The winter deficit was an evaporation story. The spring recovery was a snow and rainfall story. Both happened inside the same 12-month stretch. The Corps can measure that swing to the hundredth of a foot. What it cannot tell you is which direction the trend runs over the next several decades. And that raises a harder question about the physics itself. A body of water this large should not be able to move this fast. Why a lake this big should not move this fast? Start with the thing most people get backwards. Evaporation on the Great Lakes does not peak in July. It peaks in late autumn and early winter, when cold, dry continental air slides over water still holding summer heat. The steeper the gap between water and air temperature, the faster the lake gives up moisture. That is why lake-effect snowbands form downwind of Superior and Michigan in November and December. Those squalls are the lakes losing volume in real time. Ice is the break. A frozen lid seals the surface and cuts the exchange between water and air almost completely. Take the lid away, and the lake keeps bleeding moisture all winter. That relationship builds a feedback loop that manufactures instability. A cold winter with heavy ice suppresses evaporation and leaves the lake colder going into spring, which can favor more ice the following year. A warm winter with little ice lets the water lose heat and volume, but also warms earlier the next season, favoring less ice again. Researchers with the Great Lakes Integrated Sciences and Assessments Program have documented that handoff between high-ice and low-ice winters. It is a system built to swing. The record shows the swing did not arrive gradually. Ice cover dropped in steps. Michigan, Erie, and Ontario shifted abruptly in the mid-1980s. Superior and Huron shifted in the winter of 1998, alongside one of the strongest El Nino events on record. And the ice season on those lakes has run from a lower baseline since. If DiscoverTube's mix of geography, hard measurement, and current science is your kind of thing, subscribing helps more of it reach you. Temperature records tell a matching story, though getting them required work. Reliable satellite measurements of lake surface temperature only begin in the mid-1990s, which is not long enough to identify a regime change. So researchers at the University of Michigan and the Cooperative Institute for Great Lakes Research reconstructed lake temperatures back to the 1940s using a three-dimensional hydrodynamic model, then searched that record for extreme heat events. On most of these lakes, extreme temperature events now appear more than twice as often as before that late-90s shift. Superior shows the largest increase in heatwave intensity of the five. Here is the comparison that matters. Lake levels historically drifted across decades, which is why planners could work from long averages. Michigan and Huron moved 10 inches in a single month. On a surface that behaves as one connected body, that is close to the fastest change water of that mass can physically make. The lakes are not just absorbing weather, either. Modeling at Argonne National Laboratory found that roughly 1.4 degrees Celsius of surface warming raised near-surface air temperature, increased evaporation, suppressed large thunderstorms upstream, and increased smaller, more frequent storms downwind. Most climate models do not resolve that coupling, treating 6 quadrillion gallons of lake water as a passive boundary rather than an active partner. Now, the honest limits. Yi Hong, a research scientist with the Cooperative Institute for Great Lakes Research, has pointed out that the recent decline in levels is not unusually steep on its own. What stands out is the trajectory toward increasing variability. Whether levels trend lower or higher across the century is genuinely unsettled. One regional climate assessment projects mid-century increases, with Superior up around 7.5 inches and Michigan and Huron up around 17. That is a model estimate, and it cuts against the assumption that these lakes are drying out. Recent winters have included both record low ice and seasons near or above the long-term average. This is not a one-way collapse. It is a widening range, and a widening range is harder to plan against than a steady decline. Because shipping companies, water utilities, and shoreline engineers buy forecast reliability, not averages. Physics is only half of the instability. The other half is biological. It moves upstream, and it reproduces. 80 miles and a stop work order. In August, crews from the Illinois Department of Natural Resources working the Marseilles Pool of the Illinois River pulled several hundred silver carp out of the water. The fish were four to six inches long. The location sits roughly 80 miles from Lake Michigan, the closest this species has ever been documented to the open basin. The size is the evidence. Fish that small are almost certainly local recruits, not strays that swam a long way upstream. Illinois DNR noted that summer flood pulses and water temperatures created conditions favorable to spawning. Biologists now fear the fish reproduced upriver of the Marseilles Pool, which would put spawning closer still to the lake. That is a stated concern based on the size and number of juveniles, not a confirmed finding. Silver carp are filter feeders. They strip plankton directly from the water column, competing with the base of the food web native fish depend on. Where they are established in the Illinois River, they can dominate total fish biomass. The engineered answer sits downstream at Brandon Road Lock and Dam near Joliet, a choke point on the Des Plaines River designed to concentrate underwater deterrents and barriers on the artificial connection between the Mississippi Basin and Lake Michigan. Phase 1 was funded with about $274 million federal and $114 million state against a total project cost near $1.15 billion. Industry groups value the fishery and boating economy it protects in the billions. Weeks before those carp were removed, the Army Corps of Engineers issued a stop work order at Brandon Road. The project went under administrative review following an earlier review that had already produced delays. The Corps has not publicly given a reason and has not offered a timeline. That is the entire public explanation available. Juvenile carp advancing on one side, construction halted on the other, and no announced restart. The basin has been rewritten by an invader before. The last one arrived through a canal engineers dug for shipping. The food web nobody voted for. Sea lampreys reached the upper lakes through improved canal navigation and fastened onto lake trout. Harvest there fell from roughly 15 million pounds a year to about 300,000, around two percent of what it had been. Control efforts have since cut lamprey numbers by about 90 percent in most areas. Out of more than 180 non-native species in the basin, it remains the only one controlled basin-wide. The current disruption is smaller and harder to fight. Quagga mussels average about an inch, filter enormous volumes of water, and their spread coincided with major whitefish decline across large parts of Michigan and Huron. A 90-year population analysis found whitefish growth rates responding negatively both to mussel establishment and to a later period of low productivity. Two causes, not one. The same pattern shows up at the bottom of the food web. Sediment cores from Great Bear Lake, Great Slave Lake, and Lake Hazen, all Arctic Canadian lakes rather than part of the Laurentian system, show a long-established diatom community almost entirely replaced. Queen's University biologist John Small and colleagues found larger species giving way to smaller ones, driven by surface warming and weaker wind mixing. Small's point is blunt. Size and latitude no longer buffer a lake. Similar shifts toward smaller, less nutritious diatoms have been recorded in Lake Michigan, though invasive mussels and nutrients complicate that picture. Not every change begins in the sediment. On the shallowest lake in the chain, the change is visible from orbit and it is forecast months in advance. Half the phosphorus from one river. Every summer, the shallow western end of Lake Erie turns green. Not everywhere, not evenly, but reliably enough that federal scientists now put a number on it before the season starts. The forecast comes from NOAA's National Centers for Coastal Ocean Science, built with Heidelberg University, the U.S. Geological Survey, the University of Michigan, and North Carolina State. This year, they projected a bloom severity of 3.5, with a likely range of 3 to 4.5. That is moderate. It runs higher than the previous summer's mild 2.4, and it lands close to what the lake produced two seasons back. The severity index measures bloom biomass over the peak 30 days. Anything from 3 to 5 is moderate. Above 5 is severe. Above 7 is what researchers call particularly severe. A moderate bloom is not a quiet one. At its peak, it can spread across roughly a third of the western basin. The forecast tightens as spring goes on. An early projection spanned 2 to 5.5, a range so wide it barely constrains anything. By the time discharge data from the MAUMI came in, the window had narrowed to a .5 wide. Nathan Manning, a research scientist at Heidelberg, tied the shape of the season to a single, large April storm. That is the same April rainfall that pushed lake levels up 10 inches. This is the payoff of the lopsided nutrient math from earlier. The MAUMI delivers 5 to 10 percent of Erie's water, and about half its phosphorus. Spring discharge sets the load. The load sets the biomass. Wind and temperature decide where it piles up. That is why a forecast issued in June can describe a bloom that has not formed yet. The measurement behind it is not new. Heidelberg's National Center for Water Quality Research runs tributary sampling stations across the MAUMI watershed, a river chemistry record now stretching back roughly half a century, with more than 20 stations feeding it. Without that data set, the forecast would be guesswork. One caution matters. Bloom size is not a reliable indicator of toxicity. A large bloom can carry modest toxin levels. A moderate one can turn dangerous in a patch of concentrated scum. This summer, monitoring found toxins in the western basin below the recreational threshold, though concentrations inside surface scums can run far higher than open water readings. There is a less obvious risk, too. Municipal drinking water intakes sit below the surface, not at it. When cells sink out of a scum layer, they can carry toxins down toward the intake depth. The most visible day on the beach is not necessarily the most dangerous day at the treatment plant. Toledo learned that when bloom toxins left hundreds of thousands of people unable to drink their tap water. And that event still sets the reference point for what a bad season means, which raises the question worth putting to you directly. If DiscoverTube covered one of these long-running stories in full, would you rather see the nutrient fight in the MAUMI watershed or the engineering fight over invasive species at the Chicago Waterway? The policy side is where the frustration lives. In 2016, the United States and Canada agreed to cut phosphorus loading from the MAUMI and other western basin tributaries by 40 percent from an earlier baseline. Chris Winslow of Ohio Sea Grant has said plainly that the target has not been reached. A decade on, the forecast still gets issued, and the bloom still arrives. All of this depends on one thing this film has not examined yet. The level bulletins, the severity index, the CARP surveys, the temperature reconstructions. Every number here exists because someone is paid to go out and collect it, watching the watchers. The forecasting architecture is genuinely bi-national. Environment and Climate Change Canada and the Army Corps' Detroit District jointly produce the six-month water level outlooks under a shared coordinating committee published in both countries. One elevation number for a lake split by an international border, agreed on by two governments. Underneath sit NOAA's Great Lakes Environmental Research Laboratory and the Cooperative Institute for Great Lakes Research in Ann Arbor, running the buoys, the hydrodynamic models, and the bloom projections that get updated through the season rather than once. That network has a structural blind spot, and it is the worst possible one. Buoys are pulled out before winter because ice and storms destroy them. So the season with the least observation is exactly the season when evaporation peaks and the ice feedback described earlier does its work. Researchers have flagged that gap themselves. The mechanism driving the instability is the mechanism watched least closely. That network has also been thinning. Over the past two years, GLERL lost roughly a third of a 52-person workforce, and funding constraints delayed buoy deployment. That is an earlier development, not a new one. What is current is the budget request moving through this year's federal process. It proposes reducing NOAA's funding by about $1.3 billion, roughly a third of current levels, and would zero out programs Great Lakes researchers describe as the foundation for weather observation, water quality, maritime safety, and recreation. A proposal is not an appropriation. Congress decides. But the proposal is real, and it is on the table. The irony is arithmetic. Variability makes long records more valuable, not less. You cannot detect a widening range from a short one. The harder the system swings, the more essential an unbroken record becomes. So here is where the story actually stands. The lakes are not behaving mysteriously. Evaporation timing. The ice feedback loop. Phosphorus loading. Invasive pressure. These are individually well understood. What researchers cannot yet say is which direction the long-term trend runs. They can measure the swing's widening while the destination stays unsettled. Three things are worth watching. Whether construction resumes at Brandon Road. And how far upriver juvenile carp turn up next season. Whether the phosphorus target finally moves. And whether the instruments stay in the water long enough to record what happens. One basin. One outlet. 40 million people drinking from it. If you want more work like this, subscribe. And tell DiscoverTube in the comments which one you would track first. The carp front, the bloom index, or the water levels.