The River That Was Restored

The water knew where to go. We had to stop insisting otherwise.

For a century, in many countries, a good river was a straight river: navigable, drainable, bordered by real estate that did not intend to be a marsh. The U.S. Army Corps of Engineers, Rijkswaterstaat, and their cousins elsewhere were not villains in a children’s book. They were answering a brief that cities and farms had written in flood marks and drowned cattle. Straighten the channel. Armor the bank. Cut the oxbow. Move the water out. The hidden costs arrived as they usually do—somewhere else, or later. Faster peaks. Emptier biology. A flood that had been a season becoming a disaster because the floodplain had been rezoned as a neighborhood. Sediment that used to feed a delta trapped behind a wall of concrete.

Restoration, in this setting, is not nature winning a lawsuit. It is people deciding that a concrete idea had gone too far, and then hiring the same kinds of machines that did the first job to do a second one. Earth is moved. Canals are backfilled. Dikes are set back. Dams are taken apart in a planned sequence so that a century of silt does not become a single catastrophe. Birds return to a margin. Fish find a nursery. A city discovers that a park can be a sponge. The work looks like construction because it is construction: an older pattern rebuilt with invoices, models, and a ribbon that someone will cut.

This is good news that still requires a wrench. It does not ask you to feel that everything is fine. It asks you to notice that a straightened mistake is not necessarily permanent—and that the next flood may still exceed the design.

The cost of a straight line

A meandering river on a floodplain is slow on purpose. Water leaves the channel, spreads, drops sediment, recharges groundwater, and returns. The biology of that delay is not decorative. Shallow, warm, vegetated water grows invertebrates; invertebrates feed fish and birds; the floodplain is a nursery that a trapezoidal canal cannot imitate. When engineers confine a river between levees and cut its length, they increase slope and speed. The same volume arrives downstream sooner and higher. Neighbors who were promised protection inherit a peak. The channel, deprived of its store of wood and gravel, becomes a flume.

Channelization was often locally rational. Kissimmee ranchers wanted dry pasture. Dutch polder farmers wanted dry feet. Olympic Peninsula mills wanted power. Seoul wanted a road. Each project had a benefit that could be photographed in the first decade. The deficit accumulated in a currency that did not appear on the original ledger: wading birds that stopped coming, salmon that stopped passing, a downtown that baked over a buried creek, a flood stage that arrived like a wall.

Reconnecting a floodplain does not automatically reverse the ledger. A given project’s flood benefit depends on the whole basin. Upstream land use can undo a beautiful reach. Climate-shifted extremes may exceed the design hydrograph that justified the earthmoving. Restoration ecology is now a profession with failures as well as successes, which is how you know it is a profession. The following cases are not a tour of miracles. They are a tour of construction sites that admitted the river was still a river.

Kissimmee: filling the canal that replaced the river

The Kissimmee River once wandered about 103 miles from the lakes south of Orlando to Lake Okeechobee, across a floodplain up to a couple of miles wide. Between the 1960s and 1971 the Corps, answering a flood-control brief, replaced most of that wander with the C-38 canal: a 56-mile ditch, roughly 30 feet deep and 300 feet wide, with water-control structures that turned a pulsing wetland into a staircase of pools. The floodplain dried. Wading birds collapsed. Game fish lost the oxygenated, invertebrate-rich edges they had used. Nutrients that the marsh had processed moved more directly toward Okeechobee. The hydrologic and ecological failure was obvious enough, and soon enough, that the same agencies began planning to undo their own work.

Congress authorized the Kissimmee River Restoration Project in the Water Resources Development Act of 1992. Construction began in 1999. The Corps and the South Florida Water Management District acquired on the order of 100,000 acres. They backfilled about 22 miles of C-38, recarved missing meanders, and took out structures. As of July 2022 the Corps reported that all 22 miles of canal backfill were complete, restoring continuous flow to 44 miles of historic channel. The design restores more than 40 square miles of river-floodplain ecosystem. Costs have run at or above a billion dollars—more, in real terms, than the original straightening. That ratio is not an argument against the restoration. It is a statement about how expensive it is to buy back a floodplain after you have sold it.

The biology moved faster than the most cautious models. The Kissimmee River Restoration Evaluation Program tracks fish, vegetation, and birds against restoration expectations rather than against nostalgia. Mean monthly wading-bird abundance in restored portions during the 2021–2022 dry season was about 100 birds per square kilometre; the three-year running average rose to 54, against an expectation of 30.6. White ibis, great egrets, snowy egrets, and little blue herons increased. Ring-necked ducks, American avocets, and black-necked stilts—absent in pre-construction surveys—returned. Sportfish shifted: largemouth bass and sunfishes made up a larger share of the community in restored sections than in remaining canal, consistent with more oxygen and the return of mayfly and caddisfly larvae. Invasives arrived too, because water that is newly connected is newly available. The evaluation program is now in the business of living with that.

The project is not a time machine. Forty-four restored miles are not 103. The Headwaters Revitalization Schedule—the operating rules for Lakes Kissimmee, Hatchineha, and Cypress that are supposed to reestablish pre-channelization flow pulses—is still being implemented in increments. Flood-risk management remains a project purpose; the restoration is not a vote to drown the new houses. Hurricane Ian in 2022 was used locally as a demonstration that a rehydrated floodplain can slow water. One storm is not a frequency analysis. The honest claim is narrower and still large: a federal-state construction project put bends back in a river, birds and fish answered, and the answering was measured.

Room for the River: moving the dike, not only raising it

The Netherlands had a different argument with water, and then changed it. After high discharges on the Rhine in 1993 and 1995—the 1995 event prompting the evacuation of some 250,000 people, even though the dikes held—the country could have done what it had always done: build the walls higher. Higher walls increase the damage if they fail. They also continue a centuries-long confiscation of floodplain. The program that followed, Ruimte voor de Rivier, Room for the River, ran roughly from 2006 to 2018, with a budget on the order of €2.3 billion and some 30 to 39 measures, depending on how one counts, along the Rhine, Waal, Nederrijn, Lek, and IJssel.

The aim was dual and explicit: raise the design discharge of the Rhine branches from about 15,000 to 16,000 cubic metres per second, and improve “spatial quality”—a Dutch phrase covering recreation, landscape, and the look of a place that is no longer only a fortress. The toolbox was horizontal rather than only vertical: set dikes back, lower floodplains, cut side channels, lower groynes, remove obstacles, depolder selected polders so that they flood on purpose. At Nijmegen, the dike at Lent was moved about 350 metres inland and a 3.5-kilometre flood-relief channel was dug, making an island and a river park and taking on the order of 30 centimetres off extreme water levels at that bottleneck. At Noordwaard, a polder completed in 1970 was opened: the river levee lowered over kilometres so that high water could spread across roughly 2,000 hectares of former enclosed farmland, with houses and farms raised on terps. In 2016 the new threshold overtopped in a surge, and the polder did the job it had been rebuilt to do.

This was not the abandonment of Dutch dikes. It was a change in geometry. Land that earlier generations had drained became, again, part of the flood system, with compensation and argument. Some farmers left. Some stayed on mounds. Evaluations have called the program a rare large infrastructure effort that finished near time and near budget. It has not made the Rhine climate-proof. Design discharges are already being revisited as upstream extremes grow. Room for the River is a completed program and an incomplete answer, which is the usual condition of honest floodworks. What it demonstrates is that a rich, crowded delta can buy width as well as height, and that the purchase can be a park as well as a calculation.

Elwha: taking the dams apart and living with the silt

Two dams on the Elwha River, on Washington’s Olympic Peninsula, blocked salmon from more than 70 miles of mainstem and tributary habitat, most of it inside what became Olympic National Park. Elwha Dam went up in 1913, Glines Canyon Dam in 1927. They produced power. They also trapped on the order of 21 million cubic metres of sediment—USGS has described it as enough to fill nearly two million dump trucks—and starved the coast. The delta eroded. The estuary simplified. Five species of Pacific salmon and steelhead were confined to a short, simplified reach below the lower dam.

The Elwha River Ecosystem and Fisheries Restoration Act of 1992 set the unusual goal of full restoration of the river and its native anadromous fisheries. Removal began in September 2011. Elwha Dam was down by 2012; Glines Canyon, larger and holding more sediment, came down more slowly and was finished in 2014. The river then did what a river does with a century of stored gravel and fines. Turbidity went up. Channels filled and swung. Something like two-thirds of the impounded sediment moved downstream within a few years; most of that reached the Strait of Juan de Fuca. The delta expanded by tens of hectares. New beaches appeared. The old estuary was partly buried and then rebuilt farther seaward. Nearshore forage fish gained substrate. The Lower Elwha Klallam Tribe, whose creation story and fishery had been interrupted by the first dam, was a partner in the monitoring, not a spectator.

Salmon did not wait for a conference. Adults were found above the former dam sites within months. Chinook returns, combining hatchery and natural origin, increased during and after removal; researchers reporting in 2024 estimated average naturally spawning Chinook before, during, and after dam removal at about 1,400, 1,900, and 3,500 fish, with large uncertainty. Natural-origin subyearling migrants jumped in 2019 and 2020, once the worst of the sediment pulse had passed, into the hundreds of thousands. Coho production rose enough that the Tribe held a ceremonial and subsistence fishery in 2023, the first since removal. Steelhead have been seen upstream; few of them hatchery-marked. Hatchery programs still run. The Elwha Act’s goal is viable wild populations, not a permanent truck from a hatchery. Ten years is not a salmon generation’s last word. Summer temperatures, remaining barriers, and a still-evolving estuary are open problems. The dams are nevertheless gone, the sediment is on the move, and the fish are using miles they had not used in a century. That is a restoration, not a metaphor.

Cheonggyecheon: a stream that is also a pump

Seoul’s Cheonggyecheon is the case that restoration tourism likes, and the case that ecological restorationists argue about. A seasonal stream through the old city was covered, then buried under a road, then overtopped by an elevated expressway. By the 1990s the highway was aging, the air was bad, and downtown was losing people. Mayor Lee Myung-bak campaigned on tearing the road down. Between 2003 and 2005 the city demolished about six kilometres of expressway and opened a 5.84-kilometre lined channel at a final cost of more than ₩386 billion, then about US$281 million, over 27 months. Vendors were moved. Water was not left to the monsoon. The channel is fed by pumping: recent city figures on the order of 40,000 tonnes a day of treated Han River water plus several thousand tonnes of groundwater, tens of thousands of tonnes daily to keep a visible flow. Annual maintenance runs around ₩2.9 billion.

What returned is real and also designed. Counts of plants, fish, birds, and invertebrates rose sharply from the near-sterile baseline of a covered road. A 2022 Seoul Institute survey reported hundreds of species in the corridor. Summer temperatures along the water have been measured several degrees cooler than nearby streets. Nitrogen dioxide and coarse particles fell. Bus and subway use ticked up. Traffic did not produce the predicted permanent gridlock; the project is cited as a live example of Braess’s paradox. People walk there in great numbers. That is not nothing. A downtown that had been a sewer of cars became a place a person would choose.

Korean environmental groups and some hydrologists have said, from the beginning, that this is not a restored stream. The original Cheonggyecheon ran dry outside the rains. The present one is an urban canal with floodwalls, a recreation brief, and a water bill. Algae blooms have been a recurring complaint. Fish kills after heavy rain have been recorded over years. Historical reconstruction of the old creek’s cultural sites was partial. The Cheonggyecheon Museum’s own labels concede an unfinished, engineered object. Treat it honestly and it still belongs in this essay—as evidence that a city can unbuild a highway and buy a cool corridor, not as evidence that a watershed was reborn. Confusing the two is how restoration becomes a brand. Flying River will not do that. A pumped stream can be good urbanism and weak ecology at the same time. Both assessments can be documented. Only one of them should be used to sell the next river.

Failures, unfinished reaches, and a climate that was not in the original model

Not every reconnection holds. Replanted banks wash out. Side channels fill with the wrong sediment. A dam removal that is under-studied can send a pulse of contaminants downstream. Projects sold on flood reduction can disappoint if the bottleneck was never where the park was built. Some restorations are landscaping with a grant number. The literature is full of sites that look wet in the consultant’s year-three photo and dry in year ten.

Even the successes in this article are incomplete on purpose. Kissimmee’s headwaters schedule is still being phased. Room for the River’s 16,000 cubic metres per second is a number already under review. The Elwha’s hatchery is still a scaffold. Cheonggyecheon’s pumps are still running. Upstream of every pretty reach is a land-use map that restoration staff do not control: a subdivision, a feedlot, a warming tributary, an abandoned mine. Follow the connection far enough and a restored river is a local correction in a basin that may still be wrong.

Climate is the term that most threatens to make a completed project look naïve. Design storms are moving. Rain that used to be a one-in-a-hundred-year event arrives in clusters. Sea-level rise changes the backwater of tidal rivers. A floodplain reconnection sized for the twentieth century’s hydrograph can still be overtopped in a way the neighborhood was not told to expect. That is not an argument for walls only. Walls fail too, and they fail worse. It is an argument for saying, in the same paragraph as the bird counts, that restoration is a bet on a range of futures, not a restoration of a past climate.

Construction in the service of an older pattern

If flying rivers are moisture in the sky, this is their grounded cousin: water allowed to be a system rather than a pipe. The same editorial question applies. What is happening here that most people cannot easily see? The floodplain. The sediment trapped for a hundred years. The pumped water that makes a postcard. The operating schedule of a lake that decides whether a restored channel actually pulses. What happens if we follow the connection far enough? We might build as if the river were still a river—and we might keep the invoices in view, so that no one confuses a construction project with a miracle.

The water knew where to go. It still does. Our contribution, when we make one, is to stop insisting on the straight line, to buy back the width we sold, to take down the wall that has finished its useful work, and to tell the truth about the stream that is really a pump. That is not despair and it is not innocence. It is a set of completed jobs, a set of remaining jobs, and a flood that will test both.

Sources

• U.S. Army Corps of Engineers, Jacksonville District. Kissimmee River Restoration project pages and Congressional fact sheets (C-38 backfill of ~22 miles complete as of July 2022; 44 miles of historic channel; ~40 square miles of floodplain; land acquisition ~100,000 acres).
• South Florida Water Management District. South Florida Environmental Report, Chapter 9, Kissimmee River Restoration Evaluation Program (wading-bird densities versus restoration expectation of 30.6 birds/km²; fish community shifts).
• National Geographic. “The Kissimmee River Has Been Brought Back to Life—and Wildlife Is Thriving.” 2023 (construction complete; >$1 billion; 44 miles / 40 square miles).
• Rijke, Jeroen, et al., and Rijkswaterstaat program documents. Room for the River: ~€2.3 billion; design discharge 15,000 to 16,000 m³/s; ~30–39 measures, 2006–2018.
• Klijn, Frans, et al. Papers on the 1993 and 1995 Rhine high waters, the 1995 evacuations, and the policy shift from dike raising to conveyance.
• Dutch Water Sector / Rijkswaterstaat. Nijmegen–Lent dike relocation and side channel; Noordwaard depoldering and 2016 overtopping.
• U.S. Congress. Elwha River Ecosystem and Fisheries Restoration Act, Pub. L. 102-495 (1992).
• National Park Service, Olympic National Park. Elwha restoration research pages: dam-removal timeline 2011–2014; >70 miles of habitat; sediment and delta response.
• U.S. Geological Survey. Fact Sheet 2018-3025: ~21 million m³ of sediment; ~two-thirds released; partnership with the Lower Elwha Klallam Tribe.
• Duda, Jeffrey J., et al., and Pess, George R., et al. “Initial Responses of Chinook Salmon and Steelhead to Removal of Two Dams on the Elwha River.” Frontiers in Ecology and Evolution (2024): adult and subyearling estimates before, during, and after removal.
• Lower Elwha Klallam Tribe and co-managers. Elwha salmon recovery plan update, December 2024 (2023 ceremonial coho fishery; remaining habitat goals).
• Seoul Metropolitan Government and project histories: Cheonggyecheon restoration, 2003–2005; 5.84 km; cost >₩386 billion.
• The Guardian. “Everyone Thought It Would Cause Gridlock: The Highway That Seoul Turned into a Stream.” 17 January 2025 (pumped Han River water and groundwater volumes; annual maintenance; temperature and NO₂ reports; engineered-stream critique).
• Cho, Myung-Rae, and related urban-planning literature; Korean environmental-group critiques of ecological authenticity; reports of algal blooms and post-storm fish kills in the lined channel.
• Ritchie, Andrew C., et al. Elwha sediment budget after dam removal. Scientific Reports / related USGS papers (2018).