The Man Who Built It Right and Watched It Drown

The man on the other side of this conversation never existed. Or rather, he existed as dozens of men, none of whom left behind a memoir. When the Interagency Performance Evaluation Task Force spent four years forensically reconstructing the original design logic of New Orleans' hurricane protection system, they were essentially interviewing a ghost: reverse-engineering what the 1966 engineer knew, what tools he had, what assumptions governed his pencil. What follows is a conversation with that ghost. We're calling him Clement "Clem" Thibodaux, a composite Army Corps civil engineer whose answers are grounded in the IPET's nine-volume forensic report, the ASCE External Review Panel findings, and the original design documents they excavated. His accent is invented. His math is not.

You grew up in Terrebonne Parish. What was the marsh like when you were young?

Clem: You couldn't see the end of it. Cypress so thick you'd lose the sun by ten in the morning. My père trapped muskrat out past Cocodrie. You'd take the pirogue through channels that weren't on any map, and the land just kept going. Wasn't land, exactly. Wasn't water. Something in between that moved under your feet. You learned to read it or you sank.

Then you went to work for the Corps.

Clem: And turned that landscape into a friction coefficient. (half-smile) I know how that sounds. But after Betsy put eight feet of water in the Lower Ninth, Congress wasn't interested in poetry about cypress trees. They wanted elevations. They wanted a number on a wall that said: the water stops here. And we had a method for getting that number. The Standard Project Hurricane, the SPH.¹ Best tool available. We'd been developing it with the Weather Bureau since '59.

Walk me through what the SPH actually was.

Clem: You take every significant hurricane in the historical record for the region, analyze the meteorological parameters: central pressure, wind field, forward speed. You build a composite. A steady-state storm representing the most severe conditions "reasonably characteristic" of the area.² For Lake Pontchartrain, that gave us surface winds of a hundred miles an hour. After Betsy, we revised the wind speeds upward, kept everything else. The recurrence interval worked out to roughly once every two hundred years.³

"Reasonably characteristic." That phrase is doing a lot of work.

Clem: More than we knew. The idea was you don't design for the absolute worst hurricane that could theoretically exist. You design for the worst one that's reasonable to expect. Which means somebody has to decide what's reasonable. And the way they decided was... they threw out the outliers.⁴

They removed the most extreme storms from the dataset.

Clem: From what I understand now, yes. Storms that fell too far outside the statistical norm got excluded. At the time, in '65, '66, I didn't know that. I don't think most of us did. We trusted the methodology. It came from the Weather Bureau. It was the standard. You use the standard. That's what a standard is.

Then Camille hit in 1969.

Clem: Category 5. Winds over 175 miles an hour. Storm surge of 24 feet in Pass Christian. That was not in our dataset. That was the kind of storm we'd defined as unreasonable.

Let's talk about the ground beneath the levees. You designed to specific elevations, 9.3 to 13.5 feet depending on location.⁵ How confident were you?

Clem: Extremely. We surveyed to NGVD29 benchmarks, which was the national standard. We had freeboard allowances built in, extra height above the calculated surge level. Margin of safety. I signed off on those numbers without hesitation.

The IPET found that by 2005, some of your levees had lost most of that freeboard to subsidence and datum errors.⁶

Clem: I've read that. The benchmarks we used were already moving. Southern Louisiana, the ground doesn't sit still. And the levees themselves, the sheer weight of them on that soil, they sank fastest right after construction.⁷ My elevations were becoming wrong while the concrete was still curing. I didn't have satellite radar. I didn't have GPS. I had a surveyor's rod and a benchmark that was sinking along with everything else. You can't measure subsidence from a surface that's subsiding. That's not a failure of technique. It's a problem of physics.

There's something I want to ask about the wetlands. The Chalmette-area levee design explicitly incorporated the existing marsh as a storm-surge buffer.⁸

Clem: Of course it did. You'd be a fool not to. Miles of cypress swamp between the Gulf and the city. That's a horizontal levee. Slows the surge, breaks the wave energy, reduces the load hitting your structure. We factored that in. It was sound engineering. It was good engineering.

The Mississippi River-Gulf Outlet was completed the same year your project was authorized. Same agency. Same Corps of Engineers.

Clem: (sets down coffee cup, doesn't pick it back up)

The channel introduced saltwater into those freshwater cypress swamps. The north bank eroded at 35 feet per year. A hundred acres of marsh lost annually.⁹ The buffer you designed against was being destroyed by your own agency.

Clem: I know.

Did you know in 1966?

Clem: I knew the channel was there. I knew it was... look, we were the Corps. We built things. The navigation division built MRGO, the flood protection division built levees. Different offices, different authorizations, different Congressional mandates. Nobody sat in a room and said, "This channel is going to kill the swamp that this levee needs." That room didn't exist. I'm not sure anyone would have known what to do in it if it had.

NOAA revised the SPH criteria in 1979 to reflect significantly more intense storms. The Lake Pontchartrain project was never updated to match.¹⁰

Clem: The new criteria called for lower central pressure, stronger storm. But the project was already under construction. You don't tear out what you've built and start over. There's no mechanism for that. There's no line item in the budget that says "go back and redo the math." Congress authorized a project. We were building that project. Slowly, underfunded, but building it.

The project was supposed to take thirteen years. It wasn't finished when Katrina hit, forty years later.¹¹

Clem: If you'd told me in 1966 that the thing wouldn't be done in 2005, I'd have called you a liar. Thirteen years was already ambitious. But forty? With declining appropriations, design changes forced by lawsuits? My original barrier plan called for gates at the Rigolets, at Chef Menteur Pass. Stop the surge before it enters the lake. Benefit-cost ratio of almost nineteen to one.¹² Killed by a lawsuit in '76. So we went to the high-level plan, which was the backup. We built the backup. Over four decades. And didn't finish that, either.

The IPET's final conclusion was that most of your structures "performed as designed."¹³ The design itself was inadequate.

The post-Katrina system, $15 billion, built to a 100-year standard, is also sinking. A 2025 study found some floodwalls losing elevation faster than sea levels are rising.¹⁴

Clem: Same soil?

Same soil.

The 2025 Tulane study found subsidence rates of up to 47 millimeters per year beneath sections of the post-Katrina flood protection system, structures built on the same deltaic geology that swallowed Clem Thibodaux's freeboard allowances a generation earlier. The $15 billion system is, in places, sinking faster than the ocean is climbing to meet it. The engineers who designed it had better tools, better data, better models. They also had the same ground.

Footnotes

1. The Standard Project Hurricane methodology was developed collaboratively by the Army Corps and the U.S. Weather Bureau beginning in 1959. See NOAA Technical Report NWS 23: https://www.weather.gov/media/owp/oh/hdsc/docs/TR23.pdf ↩

2. IPET Final Report, Vol. I, U.S. Army Corps of Engineers (2009): https://biotech.law.lsu.edu/katrina/ipet/ipet.html ↩

3. GAO-06-244T, "Army Corps of Engineers: History of the Lake Pontchartrain and Vicinity Hurricane Protection Project" (2005): https://www.gao.gov/assets/a112581.html ↩

4. ASCE Hurricane Katrina External Review Panel, The New Orleans Hurricane Protection System: What Went Wrong and Why (2007), noting Berkeley researchers' finding that "excluding outlier data is not appropriate in the context of dealing with extreme hazards": https://biotech.law.lsu.edu/katrina/reports/erpreport.pdf ↩

5. DOI Water Resources, Lake Pontchartrain Project History: https://www.doi.gov/ocl/water-resources ↩

6. IPET Volume II, Geodetic Vertical and Water Level Datums (2006): https://biotech.law.lsu.edu/katrina/ipet/FINAL%20Vol%20II%20Geodetic%20Vertical%20and%20Water%20Level%20Datums%20-%20maintext.pdf ↩

7. Dixon, T.H., et al., "Subsidence and Flooding in New Orleans," Nature 441:587-588 (2006): https://earthobservatory.nasa.gov/images/6623/subsidence-in-new-orleans ↩

8. Rogers, J.D., et al., "How a Navigation Channel Contributed to Most of the Flooding of New Orleans During Hurricane Katrina," Public Organization Review (2009): https://link.springer.com/article/10.1007/s11115-009-0093-8 ↩

9. Louisiana Coastal Area MRGO Study, U.S. Army Corps of Engineers: https://lacoast.gov/new/Data/Reports/ITS/MRGO.pdf ↩

10. NOAA Technical Report NWS 23 (1979) revised SPH criteria; the Lake Pontchartrain project continued using original 1965 criteria through Katrina. See Link, L.E., "Water Resources Policy and Practice Issues Exposed by Katrina," Journal of Contemporary Water Research & Education (2009): https://onlinelibrary.wiley.com/doi/full/10.1111/j.1936-704X.2009.00030.x ↩

11. GAO-05-1050T (2005): https://www.gao.gov/assets/a112281.html ↩

12. CRS Report RL33188, "Protecting New Orleans: From Hurricane Barriers to Floodwalls" (2006): https://www.everycrsreport.com/reports/RL33188.html ↩

13. IPET Final Report (2009), finding that "with the exception of four I-wall design failures, all of the major breaches were caused by overtopping and subsequent erosion": https://biotech.law.lsu.edu/katrina/ipet/ipet.html ↩

14. Fiaschi, S. et al., "Vertical land motion in Greater New Orleans," Science Advances (2025): https://www.science.org/doi/10.1126/sciadv.adt5046 ↩