Seasonal Assessment, with Letter

The hard drive had water damage along one edge, a brown tideline across the casing that looked like a topographic contour. Elin found it in a banker’s box wedged behind the furnace, between a folder of insurance documents and a Ziploc bag of Lake Superior agates her mother had collected over the years — smooth, banded stones the size of thumbs, reds and grays and one translucent quartz she remembered holding up to a window as a child. She plugged the drive into her tablet’s adapter without expecting anything. Most of the directories were corrupt — nested folders of gibberish names, file sizes listed as zero, the digital equivalent of water-damaged paper. One folder opened clean: SEASONAL_ASSESSMENT_2026_DRAFT_3.

Outside, the pipeline was down for the second day. A pressure fault at Relay Station 14, somewhere between the pumping complex and the western distribution junction, and the maintenance crews couldn’t get to it because the roads north of Two Harbors were sheeted in ice and the Transfer Authority wouldn’t authorize a helicopter for anything below a Category 2 breach. So Elin was in her prefab outside Silver Bay with nothing to do and too much quiet. The prefab was company housing — Transfer Authority standard issue, four rooms and a utility closet, identical to every other unit in the Palisade Head development. She had lived there three years without putting anything on the walls. She opened the file.

Water samples were collected at seventeen established monitoring stations across the St. Louis River watershed during the period of June 1-30, 2026. Sampling protocols followed MPCA standard methodology (MPCA-SOP-WQ-2024, Rev. 3). Flow measurements were obtained at USGS gauge 04024000 (Scanlon, MN) and correlated with stage height observations at six tributary confluences.

June discharge at Scanlon averaged 3,340 cfs, within historical norms for the period (30-year mean: 3,520 cfs). However, disaggregated daily readings show a decline gradient steeper than any recorded in the June record. Peak flow occurred June 3 at 4,100 cfs; by June 28, the gauge read 2,780 cfs. This represents a within-month decline of 32.2%, compared to the 30-year June average decline of 11.4%.

Dissolved oxygen concentrations in the lower estuary ranged from 5.8 to 7.1 mg/L, with the lowest readings at Station 12 (Spirit Island) during the afternoon thermal maximum. Secchi depth at the harbor transect averaged 2.4 meters, reduced from the 2023 baseline of 3.1 meters.

The lake level at Duluth harbor on June 30 was 182.67 meters (IGLD85), continuing the drawdown trend documented in the 2024 and 2025 seasonal reports.

She skips the methodology tables. Her mother was meticulous about methodology tables — three pages of sampling equipment, calibration dates, chain-of-custody logs — and Elin has seen enough of them in her own work, different context, same format, the bureaucratic architecture of measurement that is the same whether you are documenting a watershed or a pipeline. Numbers in columns. Dates. Equipment serial numbers. The pipeline logistics system she manages has its own version — tracks water by the cubic meter from intake to distribution, flow rate, pressure, temperature, chloramine residual, turbidity post-treatment — and none of these numbers describe water. They describe a product.

She scrolls to July.

July discharge at Scanlon averaged 1,840 cfs.

I wrote that sentence and then sat with it for eleven minutes before writing the next one, because the number is wrong. Not wrong in the sense of erroneous — I confirmed it with Bill Kessler at the Cloquet field office, and he confirmed it with the datalogger, and the datalogger has no reason to lie. Wrong in the sense that a number like that should not exist in July on the St. Louis River. The 30-year July mean is 2,680 cfs. The previous record low was 2,110, set in 2012. 1,840 is not a fluctuation. It is not an artifact of a single dry week. It is a tributary system dropping below its capacity to sustain the cold-water assemblage — the brook trout, the sculpin, the invertebrate communities that depend on dissolved oxygen concentrations above 6 mg/L and water temperatures below 18C. At 1,840 cfs and the water temperatures we recorded in the third week of July, the modeling suggests DO levels at the western estuary confluence dropped to 4.2 mg/L.

This is the number I have been watching for. I did not want to find it.

When I first surveyed this reach in 1995, the St. Louis carried spring runoff that turned the banks into a country of mud and standing water all the way to Fond du Lac. The willows along the floodplain went knee-deep. Herons stood in the shallows between Scanlon and the Thomson Dam. My first field season, I walked the riparian corridor in boots that leaked and I was twenty-seven years old and the river was so full it seemed impossible that it could be going anywhere but where it had always gone, which was toward the lake, which was toward the horizon, which was toward something I did not yet have a name for but understood to be permanent.

You were three. You were staying with your father in Eagan that summer, and I called every night from the motel in Cloquet, and you told me about the neighbor’s cat and I told you about the river, and neither of us was listening to the other, which was training for everything that came after.

Elin puts the tablet on the couch cushion beside her. You were three. She stares at the far wall of the prefab — insulated panel, the seam where the factory didn’t quite align the sections, a hairline crack she’s been meaning to seal since October. She does not remember being three. She remembers being seven or eight and standing on a dock somewhere on the North Shore while her mother crouched at the waterline with a thermometer and a clipboard, and she remembers the cold, and she remembers asking about it, and she remembers her mother saying something about glaciers. She does not remember the words. She remembers the feeling of being answered by someone whose attention was already elsewhere — the way her mother’s eyes tracked the water while her mouth formed an explanation, the divided focus that Elin learned to read before she could read clocks.

Ice clicks against the windows. The furnace cycles on with the low hum she has learned to sleep through — the heat pump pulling calories from the air outside and pushing them through the ductwork, energy transferred from where it is to where it is wanted.

She picks the tablet back up.

August field observations are presented here in summary form. Full data tables are appended (Appendix C, Tables 8-14). Conditions during the August sampling period were characterized by sustained high air temperatures (mean daily max: 29.4C, 6.2C above the 30-year August average), below-normal precipitation (42 mm recorded, representing 38% of the August mean), and the development of a cyanobacterial bloom in the inner harbor that was first observed August 9 and persisted through the end of the sampling period.

I will describe the bloom as I found it on August 14, because the data tables do not contain the smell.

The water in the inner harbor at 0640 was the color of pea soup left out overnight. Not green exactly — a green-brown, with a surface film that broke and reformed around the bow of the sampling skiff like skin on heated milk. The air temperature was already 24C. The smell was organic decay — not the clean rot of autumn leaves in a creek but something thicker, a sweetness underneath the sulfur that catches in the soft palate and stays there through lunch. Decomposing cyanobacteria release volatile organic compounds — geosmin, 2-methylisoborneol, dimethyl sulfide — and when the bloom is this dense and the water this warm, the compounds volatilize so aggressively that you taste them before you smell them. You carry them home in your hair.

I collected twelve samples at one-meter depth intervals from the skiff. My hands were green to the wrists within four minutes. The nitrile gloves had split — I’d been reusing them because the supply budget was cut in March, a small humiliation I mention only because it is representative of a larger pattern, which is that the people who decide how much money to spend on monitoring a watershed do not have green hands. The Secchi disk disappeared at 0.6 meters — I could have been lowering it into paint. At Station 7, the DO meter read 3.1 mg/L at two meters depth. Below that, effectively anoxic. The fish kills hadn’t started yet but they would come in the next week — first the sculpin floating belly-up along the breakwall, then the perch, then the algal mat thickening over them until the harbor looks less like water and more like a wound trying to scab.

I rinsed the sample jars with distilled water three times and the glass was still filmed green. My forearms itched for two days. The lab results came back September 3: microcystin concentrations at Stations 5, 7, and 9 exceeded the EPA recreational advisory threshold by a factor of four.

This is what August smells like now. It did not used to smell like this.

Elin knows this harbor. She manages the intake manifold schedules for the western transfer line, and the harbor is where Pipeline 3 drew its source water before the intake was relocated twelve miles up the shore in 2041. The relocation was her first real assignment at the Transfer Authority — she had been hired six months earlier, a logistics coordinator fresh from the distribution hub in Ashland, and the relocation project was the kind of work she was good at: routing, scheduling, coordinating the pipe-laying crews and the environmental compliance paperwork. The paperwork cited water quality degradation. She remembers the phrase but not the numbers behind it. Her mother’s numbers, possibly. Her mother’s exact numbers, stripped of their source, treated, delivered to a use their collector never intended.

The extraction corridor proposal is on her tablet too, in a different folder, the one marked GLTA-EC-2048-017. Seven hundred pages of environmental assessment, flow projections, economic modeling. She has read it twice. The corridor would draw an additional 340 million gallons per day from the Superior basin, routed through the existing pipeline infrastructure to the Fox River diversion junction and from there south to the Illinois distribution network. The Illinois allocation has been oversubscribed since 2044 — the aquifers downstate collapsed faster than anyone projected, not gradually but in a staircase of sudden drops that everyone calls unprecedented because the alternative is to call them predictable.

Legal authority for the corridor derives from the 2039 amendment to the Great Lakes Compact — the emergency provisions that everyone knew were permanent the day the governor signed them. Similar diversions are already operational: the Yangtze-to-Yellow-River, the Irtysh transfers in western China. The Great Lakes are late to this. Water goes where the money sends it. This is not ideology. It is plumbing.

Elin pulls up the Transfer Authority’s current water quality dashboard on a second window and runs her mother’s August numbers against it. The comparison takes four seconds. The harbor DO in August 2048 was 2.1 mg/L — lower than her mother’s 3.1 by a full milligram per liter, a decline that proceeded at almost exactly the rate her mother’s modeling predicted. Elin is good at this: the side-by-side, the trend confirmation, the conversion of one set of numbers into another set of numbers that support or complicate a decision.

She opens her mother’s file again. September.

September precipitation at the Two Harbors station totaled 31 mm, 41% below the monthly mean. The St. Louis River flow gauge at Scanlon recorded a monthly average of 1,620 cfs. I will note for the record that in thirty-one years of field work on this watershed, I have never measured a September average below 2,000 cfs. The previous low was 2,240, recorded in September 2021.

Lake level at Duluth harbor: 182.41 meters. Down 26 centimeters since June. The rate is not alarming if you only look at the rate. Twenty-six centimeters in four months. But the starting point was already below the five-year mean, and the five-year mean is below the thirty-year mean, and the thirty-year mean is below the centennial estimate, and each of these numbers lives inside the next one the way smaller boxes live inside larger ones, and the smallest box is the one I am standing in, on the harbor breakwall on September 12, looking at the rocks below the waterline — the ones that are supposed to be below the waterline, the ones that hold the breakwall’s structural base — now exposed to the air and drying in the sun. The zebra mussels on them were dead and chalky. It looked like the lake was pulling its skirts back.

I deployed a linear array of six geophones along the lakebed at the Lester River mouth on September 18, in accordance with the nearshore sediment assessment protocol requested by the Army Corps (reference: USACE-LRB-2026-044). The purpose of the deployment was to measure shear-wave velocity in the lakebed substrate, which provides data on sediment compaction, erosion susceptibility, and structural integrity of the nearshore zone.

The method relies on the detection of Scholte waves — acoustic interface waves that propagate along the boundary between the water column and the underlying sediment. Scholte waves were first described by J.G. Scholte in 1947. They are characterized by their concentration at the fluid-solid interface: the wave energy is maximal at the contact between water and earth and decays exponentially in both directions, upward into the water column and downward into the substrate. They exist because of the boundary. Without it — without the meeting of two different media — the wave does not propagate. The energy has nowhere to live.

The measurements from the Lester River mouth indicate a shear-wave velocity of 88 m/s in the upper sediment layer, consistent with loosely consolidated glaciolacustrine deposits. Erosion modeling based on this value suggests the shoreline at this location will recede approximately 2.3 meters over the next decade under current wave-energy conditions. Under projected conditions — reduced ice cover, increased storm frequency, lower lake levels — the recession rate doubles.

I was kneeling in shallow water hammering the geophones into the lakebed substrate when a kayaker went past, close enough to wave. He didn’t ask what I was doing. People on the North Shore are accustomed to solitary figures doing inexplicable things at the water’s edge. I waved back and then I knelt there for a while longer, my knees on the sand, the water at 11C, the geophones recording the vibrations of trucks on the highway and the Scholte waves propagating along the interface beneath me, and I thought about boundaries. About what lives at the contact between one thing and another.

I think about you at the boundary of things. Between the job I chose and the life I should have been paying attention to. Between the water and the ground.

Elin reads this passage twice. The first time quickly, the way she reads everything — looking for the data, the shear-wave velocity, the erosion rate, numbers she can cross-reference against the current shoreline surveys in the extraction corridor assessment. Her mother’s 2.3-meter-per-decade projection: she can check that against what actually happened. Twenty-two years of shoreline data, somewhere in the Transfer Authority’s archives. She makes a mental note to pull the files when the system comes back online.

The second time she reads it slowly. Something in the prose has changed — a pressure drop, a shift in the texture that is not yet weather but is no longer just air.

I think about you at the boundary of things.

She puts the tablet down. She walks to the kitchen and fills a glass from the tap. Pipeline 2, the domestic supply line — treatment plant at Beaver Bay, forty miles of buried conduit, the Palisade Head development. The water is cold and tastes faintly of chloramine. She drinks it standing at the counter with one hand on the glass and the other on the edge of the sink.

Her mother has been dead for nine years. Pancreatic cancer, fast, the diagnosis in April and the funeral in August, and between those months a series of phone calls — weather, work, the dog — except shorter now and with silences that neither of them knew how to fill. Elin had driven up to Two Harbors twice. Both times they sat in the kitchen and drank coffee and talked about the pipeline construction tearing up the road outside Beaver Bay. Her mother had looked out the window at the lake and said, “The level’s down again this year,” and Elin had said, “I know,” and they finished their coffee.

She goes back to the couch. Outside the prefab, the ice is still falling. The pipeline is still down. The extraction corridor assessment is still in its folder.

She opens October.

October 2. The Lester River at its mouth. Water temperature: 8.2C. Air temperature: 4.6C. Wind from the northeast at 22 km/h. The river is running dark with sediment — there was a heavy rain three days ago, the first significant precipitation since mid-August, and the watershed is flushing everything it held during the dry months. Turbidity at the river mouth: 47 NTU. I can’t see the rocks at the bottom, and I’ve been wading this stretch since before the rocks were laid down. Not literally, but it feels that way — there are days in this work when your body is older than your memory of the landscape and your memory of the landscape is older than the landscape itself, because the thing you remember has already become the thing that used to be.

I collected the final samples of the season here: six surface grabs, three at one-meter depth, two sediment cores from the river channel. The cores were heavy and dark — anoxic sediment, sulfur-smelling, the bottom of the river giving up its buried chemistry like an old house opened to the air. I labeled the jars and packed them into the cooler and sat on the rocks and ate an apple and watched the osprey that nests above the culvert on the upstream side make its last passes of the season, low over the water, hunting with a focus that I have spent thirty-one years trying to match and have never quite achieved, because the osprey does not need to write down what it sees.

The wind shifted northeast and brought the lake smell — cold water and iron and the faint mineral sharpness of exposed basalt. My boots were wet through. My field notebook had a water stain on the page where I wrote the turbidity reading, and I left it there because the stain was data — the river reaching up to mark the document that was measuring it. The subject gets into the record. The record is never clean.

These samples will complete the October data set. The seasonal assessment is nearly finished. This document will be submitted to the Minnesota Pollution Control Agency and the Environmental Protection Agency Region 5 office, with copies to the Great Lakes Commission and the Army Corps of Engineers. It will contain data showing that the St. Louis River watershed experienced its lowest recorded flows in 2026, that Lake Superior continued its multi-year drawdown, that the nearshore zone at multiple monitoring stations showed accelerated erosion, and that the cyanobacterial bloom in the Duluth harbor was the most extensive and longest-lasting in the monitoring record.

The document will be received. It will be filed. It may be cited in a future environmental impact statement, the way my 2019 report was cited in the Phase II assessment for the pipeline corridor — my data appearing in someone else’s argument, supporting a conclusion I would not have drawn, the numbers separated from the hands that collected them and put to work building the thing the numbers warned against.

I am fifty-eight years old. After this season, the field position is eliminated — budget reallocation, the polite word for a decision made in St. Paul by people who have not stood in the Duluth harbor at dawn and tasted the algae on the back of their teeth. I will be reassigned to the data management division. I will sit at a desk and organize other people’s field observations into databases. Not the worst afterlife for a field scientist. Not the one I would have chosen.

But I have been trying to write this paragraph for three weeks, and what I want to say is not in the data. The data is in Appendix C. What I want to say is:

You used to ask me why the water was so cold and I told you it was because it came from glaciers, which was not true, but it was the kind of lie that contains a truth, and I wish I had more of those to give you. The water is cold because it is deep and because it is old and because the lake turns over twice a year, the whole volume of it rolling in the dark so that what was on the bottom comes to the top and what was warm becomes cold, and this has been happening since the last ice age and will continue after both of us are finished with our respective tasks on its shore. I told you about glaciers because the truth — thermal stratification, convective mixing, the physics of density as a function of temperature — was not what a seven-year-old needed. You needed the glaciers. You needed the ancient ice. You needed the cold to come from somewhere beautiful and distant, and I gave you that, and I think it was the best thing I ever gave you, better than the facts I spent my career collecting, because the facts have turned out to be a record of loss and the glaciers, though they were not true, were at least a way of saying: the world is larger than what is happening to it.

I am sorry I was in the field when you were growing up. I am sorry that the attention I gave to this watershed was the attention I should have been giving to you, and I am sorry that I cannot tell you, even now, sitting on these rocks with the October wind coming off the lake and the sediment plume spreading at the river mouth like a bruise, I cannot tell you which loss is larger. Whether the river’s decline or my absence from your life is the thing I should have spent these years trying to stop. I am ashamed that I cannot tell. A better person would know. A worse scientist, maybe, but a better person.

This paragraph is not part of the seasonal assessment. I’m going to leave it in.

Elin is in her car. She does not remember deciding to leave the prefab, putting on her coat, starting the engine, but the heater is running and the windshield wipers are knocking ice off the glass and she is driving south on 61, the only road, the road her mother drove for thirty-one years to reach the monitoring stations, the tributaries, the watershed that Elin now manages from the other side — not documenting the loss but administering it, routing it through pipe.

The extraction corridor form is on her tablet on the passenger seat. She has not signed it. She has not decided not to sign it. These are different conditions, and she is good at distinguishing between them, because that is what her work requires: not decisions but the precise management of undecided things, keeping the options open, the schedules current, the pressure balanced across the distribution network until someone above her in the Authority decides what happens next.

The Lester River appears on her left — she almost misses it because the signage is gone, replaced two years ago by the pipeline access markers, the blue-and-white Transfer Authority placards that mark every waterway within the extraction zone. She pulls off the road into a gravel turnout that wasn’t here in her mother’s time. It was built for the survey crews.

The river mouth is narrower than the photographs. She can see this without measuring — the banks wider, the channel compressed, the rocks her mother described wading over now exposed and dry, lichened in a way that means years, not days. The lake is out there, still enormous, still cold, but the shoreline has pulled back and the exposed shelf rock runs thirty meters or more before the waterline begins. The lake level last month was 180.12 meters. Her mother measured 182.67 in June of that year. The math is simple. The math was always simple. That was never the problem.

She stands where the river meets the lake. The wind is from the northeast, steady, carrying the smell of cold water and something else — iron, maybe, or the mineral absence that cold air has when there is nothing between you and the horizon but lake. She does not have a thermometer but the air feels like single digits, and the water pushing past her boots — she is wearing the wrong shoes, office shoes, and they are soaked through — the water is cold enough that her ankles ache within a minute.

There is no osprey nest above the culvert. She looks for it because her mother described it, and the culvert itself is different — larger, a corrugated steel pipe three times the diameter of anything a stream this size would need, because it was upsized during the pipeline construction to accommodate a feeder line that was never installed. The osprey are gone. She does not know whether this means they are elsewhere or they are gone. The distinction matters to ornithologists. To Elin it is the same absence.

The rocks at the river mouth are the same rocks her mother sat on to eat an apple and write about failure. Her mother would have corrected her — rocks change, just slower than you think, and the evidence is in the shear-wave data, the erosion models, the retreating shoreline that has moved these rocks from waterline to dry land in two decades. Her mother would have had the numbers. Elin has the rocks.

The light is doing something on the water — late afternoon, November, the sun at an angle that turns the surface of the lake into a field of moving silver, each wave catching and releasing the light independently so that the whole visible expanse seems to be signaling, though to whom and about what she could not say. She stands there long enough for the light to change. The silver goes amber. The amber goes gray. The river keeps running, thinner than it was, carrying less than it carried, entering the lake at a place where her mother stood twenty-two years ago with sample jars and a seasonal assessment that was turning into something else, something her mother could not stop writing and did not submit.

She walks back to her car. Her feet are numb. The tablet is on the passenger seat, the extraction corridor form still open, the signature line still blank. She picks it up. She puts it down. She picks it up.

She has not finished reading.

October 14. Lester River mouth. Final entry.

The lake this afternoon is the color of itself and nothing else. There is no simile for Superior in October — it is not steel, not slate, not any of the gray words people reach for when they describe Great Lakes water in autumn. It is the color of a body of cold fresh water ten thousand years old under a sky that is about to snow, and if I have learned anything in three decades of field work, it is that the thing itself is always more precise than the comparison.

The osprey are gone south. The sediment plume has cleared since the last rain. The water at the river mouth is so transparent that I can see the bottom at two meters, the glaciolacustrine sand and the dark shapes of the rocks, and it occurs to me that this clarity is not health but absence — the river is carrying so little suspended material because the river is carrying so little water, and the beauty of the clear water is the beauty of a system running out.

I do not know who will read this. The agency will file it. You may never find it. But I am writing it here, in the space where the assessment ends and something else begins, because the alternative is to describe only what is measurable, and the measurable world, though it is the world I have given my life to, is not the whole of what I have seen.

The light on the lake at this hour. The rocks at the river mouth, water-dark on one side and dry on the other. The cold running over my hands while I rinsed the last sample jars of my career, and the cold felt like something I would miss, which is a strange thing to miss but there it is. The lake, the river, the shore, the sky. I wanted you to know about them and I am telling you now, too late and in the wrong document, because I did not know another way.