In May 2019, weeks of severe weather and record-breaking rainfall swamped Keystone reservoir with nearly six times its volume, amounting to more than twice the inflow during Tulsa’s 1986 flood. The U.S. Army Corps of Engineers released 2.8 times as much water from Keystone Dam as in 1986,1 loading Tulsa’s levees longer than any period in their history.
David Williams, chief hydrologist with the Corps’ Tulsa District, said the 2019 flood was one rainfall away from releases of up to 315,000 cubic feet per second (cfs) from Keystone Dam: an unprecedented volume likely to have breached the Tulsa-West Tulsa Levee System.2 Although the reservoir reached a new pool of record, releases peaked at 275,000 cfs.3
“The reason why 2019 was not also the release of record was that we did everything we could to not make a bigger release,” Williams said. “At those types of release levels, every increase has an effect. And we had a lot of concern about Tulsa-West Tulsa. Those levees. Additional loading on those levees. We had concern about residential areas downstream. … Town and Country, and the neighborhoods adjacent to Town and Country, over there along Highway 51—obviously, they flooded. They flooded at around 200,000 cfs.”
To buy time, the Corps obtained emergency authorization to let Keystone reservoir continue to rise—a deviation that would have extended to a Water Surface Elevation of 758 feet. The pool peaked at a record 757.2 feet in the last week of May 2019, surpassing 1986 by 2.4 inches.
As floodwaters rise against a sand levee, hydrostatic forces drive water inside it. If they’re working, which Tulsa’s are not, relief wells manage the pressure by moving the seepage to toe drains, where pumps return the water to the river.4 When these components fail or cannot keep up, water pushes through to the “dry” side of the levee, sometimes creating volcano-like formations called sand boils.
Ranging in size from streaming pinholes to billowing plumes a few yards in diameter, sand boils can rapidly destabilize the ground beneath them. If they’re leaching material from inside a levee, they require immediate attention. The most common treatment, a ring of sandbags, allows a pool to form over the top, applying back pressure to slow the flow of water. After you’ve addressed one, it’s on to the next: Without intervention, sand boils can liquefy a levee.
Working alongside the Corps, Levee Commissioner Todd Kilpatrick summoned civilian volunteers, the Oklahoma National Guard, Tulsa County Sheriff’s Department and police and firefighters from Sand Springs and Tulsa to try to preempt a worst-case scenario.
Volunteers filled sandbags as dozens of levee walkers, guards on ATVs, and drones scoured the landscape around the clock. Backup pumps pulled water from long-defunct relief wells and crumbling toe drains. Black Hawk and Chinook helicopters circled with 2,500-pound sandbags as the National Guard detonated C-4, blowing the anchors of transmission lines that spanned the river to forestall them being ripped from the levees.
“We had no idea exactly where those weak points were gonna be,” Kilpatrick said. “But we definitely knew that there were gonna be some.”
To emergency workers’ horror, sand boils surfaced far beyond the levees, deep in neighborhoods.
“So we’re looking at 10, 20 feet away thinking it would pop up there,” he said. “But no, it pops up 300 yards away, and the next thing you know, you’ve got a series of sand boils. … It’s just unimaginable the places that you would have the water show up.”
As the potential for sand boils in crawl spaces and backyards grew, Kilpatrick urged residents to call in and report them. Within days, his voicemail was full.
“I couldn’t keep up with the phone calls,” he said. “We’re talking a 10-foot wall of water sitting behind people’s houses.”
The levees held, but an estimated 500-600 structures flooded along the Arkansas River and local streams.5 FEMA’s 100-year floodplain, where flood insurance is required for federally backed loans, is based on an outdated estimate of 205,000 cfs. Against the recommendations of its own hazard mitigation advisory board and the Arkansas River Corridor Master Plan, the City of Tulsa continues to regulate development in the Arkansas River floodplain using 205,000 cfs plus one foot.
Jon Phillips, the National Flood Insurance Program (NFIP) coordinator for Oklahoma, said many residents affected by May 2019 flooding didn’t know they needed flood insurance because they lived outside the regulatory floodplain.
“But what they were told was, you are not required to have flood insurance,” Phillips said. “But because you’re so close to the 100-year floodplain, yes, you probably need flood insurance. … I think it comes down to the message of the community.”
Joe Kralicek, executive director of the Tulsa Area Emergency Management Agency, cautioned that inundation areas behind accredited levees also are not mapped as part of the FEMA floodplain. (Tulsa’s failing levees remain accredited despite their condition.)
“That means that people have an unrealistic assessment of the risk,” Kralicek said. “Human beings are, by their very nature, horrible—we are awful—at understanding and assessing risk. We don’t do it. We don’t do it properly. And so [the federal government is] telling these folks behind the levee systems, ‘Oh, you’re not in a floodplain.’ They don’t get the nuance.”
The 2019 flood also thrashed the new Gathering Place shoreline, reclaiming much of a habitat restoration project that was required to mitigate the park’s impact on the ecosystem. About 8.9 acres of Gathering Place were built in the Arkansas River floodway, with a shoreline designed for flows of just 205,000 cfs.6 The habitat restoration area, also known as the “Wilds,” was built without any meaningful level of flood protection.7
Josh Miller, program officer with George Kaiser Family Foundation, said in an email that Gathering Place sustained more than $7.2 million in damages during the 2019 flood: more than three times the total in the rest of River Parks. (The park occupies one mile of River Parks’ 26-mile system of paved trails.) FEMA reimbursements to Gathering Place covered less than $4 million.8 The rest came out of Gathering Place funds.
The habitat restoration area alone cost more than $4 million to rebuild. It is likely to be swept away again and again if Gathering Place does not reckon with the potential for major flooding. Miller said the area was rebuilt after the 2019 flood with “much more rock armoring” to protect the soil and plants—“up to at least a river flow of 40,000 cfs.” (Flows during Tulsa’s 1986 and 2019 Arkansas River floods peaked at 307,000 cfs and 277,000 cfs, respectively.)9
In a 2019 memo, engineering consultant and floodplain manager Bill Smith reflected on major damage to Gathering Place’s “Lakeview Lawn,” built in the floodway above Zink Dam.
“My opinion was and still is that this should never have been constructed as I predicted this type of damage,” Smith wrote.
“It will rain again, It will flood again!!” he added beneath an image of the damage.
Nationally recognized floodplain manager Ron Flanagan said building portions of the park in the river is an extreme example of development and political interests coming before common-sense hazard mitigation in Tulsa.
“Why in the world does the Gathering Place need to fill in the Arkansas River?” Flanagan said. “It’s just insane. … Why do you think, other than the power and the influence that George Kaiser has, that they would do this for creating an open-space play area? It just doesn’t make any sense at all.”
I asked Miller: Why build in the Arkansas River floodway?
In an email, Miller said GKFF sought to “build a true riverfront park, one that gave all Tulsans an opportunity to interact and experience the Arkansas River in a way the previous River Parks trails did not.” He said the areas built in the floodway “were created to allow citizens to safely experience the river’s edge,” and Gathering Place followed permitting requirements to construct them.10
Flanagan said 2019 should have been a wake-up call.
“If they’d released another foot, it would have been the biggest disaster on the Arkansas River floodplain that we’ve ever known,” he said. “All of these things that they’re building along the Arkansas River would have flooded. The levees would have been overtopped.”
When the City of Tulsa broke ground on the new Zink Dam in 2020, Mayor G.T. Bynum claimed it will make Tulsa “safer from a flood control standpoint … because you’ll have gates that can all lay down flat on the riverbed.” The statement contradicts basic design details and data about the dam.
First, so-called full-height gates make up just half of the dam’s length. The rest consists of large concrete piers; a massive concrete weir (a fixed-height barrier); and crest gates, which are three feet high and mounted at least seven feet above the riverbed on large expanses of concrete.
Additionally, the full-height gates lower onto a base that stands as much as five feet above the riverbed. The base consists of concrete built on top of the concrete base of the original Zink Dam, which remains in place across the length of the new dam. Between the dam and Gathering Place, an engineered island and whitewater flume also encroach on the Arkansas River floodway.
Lastly, hydraulic data provided by River Parks Authority estimates that the new dam and flume will have nearly the same net impact on flooding as the original Zink Dam and pedestrian bridge. (The estimated net hydraulic impact of the new bridge is negligible.)
After a public presentation April 25, 2023, Bynum again suggested that the new Zink Dam would have a meaningful positive impact on flood mitigation.
“The explanation we were given on the dams that are being built is that they would—especially the one at Zink—actually improve the flood capacity,” Bynum said.
“Have you looked at that water data—the hydraulic data?” I asked. “Because it actually shows that there’s no improvement.”
“That is not what we were told at the time,” Bynum said.
Bill Smith, who has been involved with Arkansas River projects since the 1970s, said in a 2022 phone interview that lowering Tulsa’s risk in a flood “was never the intent” of rebuilding Zink Dam.
David Williams, chief hydrologist with the Corps’ Tulsa District, said the 2019 flood was one rainfall away from releases of up to 315,000 cubic feet per second (cfs) from Keystone Dam: an unprecedented volume likely to have breached the Tulsa-West Tulsa Levee System.
How much time is left to reckon with Tulsa’s flood risks? Between 2015 and 2019, Houston experienced multiple 500- or 1,000-year flood events, including Hurricane Harvey: the wettest storm of its kind on record in the U.S.11
The Intergovernmental Panel on Climate Change (IPCC) warned in 2021 that without “immediate, rapid and large-scale reductions” in emissions, we will lose the chance to prevent worst-case climate extremes. The IPCC reports that heavy precipitation and flooding events are likely to continue to intensify and occur more frequently with rising temperatures.12
In 2019, the Corps reported that the average temperature in the Southern Great Plains could increase by 5.1 degrees Fahrenheit by 2050, or 8.4 degrees Fahrenheit by 2100 (compared to averages from 1976-2000), if we effectively reduce emissions. If emissions continue at the current rate, the region could see as many as 60 days per year over 100 degrees Fahrenheit by 2100.13
After the 2019 flood, the Corps recalculated Tulsa’s 100- and 500-year Arkansas River flood probabilities. The 100-year estimate rose to 270,000 cfs, signifying a 1% chance of near-2019 flooding occurring in any year.14 The estimate’s uncertainty range, or confidence interval, is further illuminating: The Corps reported 95 percent confidence that a so-called 100-year event would fall between 174,900 cfs and 417,000 cfs—a flood likely to destroy the levee system.15
In the next article, I’ll unpack the levee study to address questions including: How did the Corps justify the decision not to raise the height of the levees or repair Levee C? How did Gathering Place factor into the outcome of the levee study? How much flooding can the levees withstand?
If you’re new to Watershed, start at the beginning:
Watershed is read in 28 states and 6 countries.
If you have the means, help us share essential reporting with everyone who needs access:
Subscribe for $7/month or $64/year.
Make a sustaining contribution at a higher level of your choice (choose Critical Mass and enter any amount above $64).
Purchase a gift or group subscription.
All free and paid subscribers get the latest updates delivered to your inbox, plus access to the full archive at mollybullock.substack.com. You can modify your subscription or cancel any time by logging in to your account.
Reporters Committee for Freedom of the Press provides pro bono legal counsel for this series. You can support their work here.
This comparison reflects the total volume of water released during the period of levee loading (150,000 cfs or greater, 24-hour average) for each event: 10 consecutive days in 2019, and 4 consecutive days in 1986.
The Corps estimates that flows of 330,000 cfs would breach Levee B, filling the area behind Levees A and B to flood depths as high as 15 feet. (A breach of one floods the areas behind both.) However, the levees are likely to breach earlier and in different ways than projected due to their condition. This occurred at Levee A in 1986—a flood with peak flows of 307,000 cfs through Tulsa.
If you live behind the levees, it is also critical to understand that the above estimate does not account for stream flooding along tiebacks, where the levees tie into tributaries. According to Tulsa’s 2019 All Hazard Mitigation Plan, the tiebacks provide a so-called 25 to 30 year level of protection from stream flooding. In other words, flooding is far more likely to occur in these areas.
I will describe the Corps’ loading and overtopping estimates for the Tulsa-West Tulsa Levee System in greater detail in a future article.
Peak releases during the 2019 flood (275,000 cfs) occurred between 8 a.m. May 27, 2019, and 4 p.m. May 29, 2019.
All relief wells and toe drains were inoperable as of January 2024.
These numbers appeared in news reports at the time, citing the Tulsa Area Emergency Management Agency. Executive Director Joe Kralicek did not fulfill requests to verify their accuracy prior to publication.
Bill Smith, an engineering consultant and floodplain manager involved with Arkansas River projects since the 1970s, said “certain places” along the shoreline at Gathering Place were armored above 205,000 cfs but that “every place was kind of different.” For more information about development in the Arkansas River floodplain, go back to “Still Fighting It,” “Channel Markers” and “The Mentality.”
Josh Miller, program officer with George Kaiser Family Foundation, said in an email that the Corps did not require rock armoring for the initial habitat restoration project. He said Gathering Place built the islands with “very little rock armoring” and plantings that did not have time to take root before the flood wiped them out less than a year after the park opened.
FEMA reimburses 75% of the cost of approved repair projects. Although the habitat restoration project qualified, much of the other damage to Gathering Place did not because it involved landscaping, an ineligible expense exceeding $2 million.
For additional context about the 1986 flood and releases from Keystone Dam, go back to “Still Fighting It,” the third article in this series.
Miller said Gathering Place met permitting requirements for the encroachment by excavating and surveying compensatory storage in the channel. The park must maintain the storage over time. Miller said Gathering Place would survey the storage for sedimentation before the project opens, re-survey after major flood events and/or every five years, and report results to the Corps and City of Tulsa.
In 2018, the Houston area expanded its regulatory floodplain to a 500-year standard and increased requirements for construction in the floodplain.
The Intergovernmental Panel on Climate Change (IPCC) has been reporting on the escalating risks of anthropogenic climate change for decades. A longtime campaign by industry and elected officials including former Sen. Jim Inhofe has sowed public doubt about climate change; however, the IPCC represents the climatological findings and consensus of the vast majority of the scientific community.
FEMA’s 100-year floodplain does not reflect this update.
The confidence interval is inclusive of its lower and upper bounds. For essential information about the status of the Tulsa-West Tulsa Levee System, go back to “The Known Unknowns.”