Thirty-four billion dollars buys you the largest coastal protection project in Army Corps of Engineers history. It buys you the Coastal Texas Protection Project, the system of gates, barriers, and dunes anchored across Galveston Bay that everybody calls the Ike Dike, featuring the largest barrier gate system in the world. Two components entered active design in 2025. The full magnitude of the thing, as one project official put it, is only now becoming real.
It may be realer than anyone involved has accounted for. Two independent studies published in the past fourteen months have established that the number the project is being designed to, the baseline from which all its sea level projections launch, is wrong. The sea is higher than the models say, and the ground is sinking faster than standard measurements capture. The $34 billion project sits between those two surfaces, engineered to a gap that doesn't exist.
Where the water should be versus where it is
I spent five years on cargo ships, and one thing you learn early is the difference between the chart and the ocean. The chart is a theory. The ocean is wet.
Geoid models are charts. They're mathematical surfaces derived from Earth's gravity and rotation, describing where the sea surface should be according to physics. Tide gauges are the ocean. They measure where the water actually sits, accounting for winds, currents, temperature, salinity, all the local conditions that physics alone can't predict.
In March 2026, Karin Seeger and Philip Minderhoud published a study in Nature examining how coastal flood risk assessments handle this distinction. More than 99 percent handled it badly. Ninety percent relied on the chart instead of the ocean. The discrepancy, globally averaged, runs between 9.4 and 10.6 inches. The actual sea surface, measured at the coast, is higher than the geoid predicts.
Jonathan Bamber, a glaciologist at the University of Bristol, observed that this single measurement error exceeds total global sea level rise since 1900.
"The error in the ruler is bigger than the thing being measured."
The Gulf Coast offset likely differs from the global mean. Minderhoud has noted that discrepancies run "relatively small" for Northern Europe and the US East Coast, larger in the tropics and Southeast Asia. The Gulf of Mexico falls somewhere between, and the specific regional figure has not been publicly reported. But the direction is not in question. Corrected calculations show 37 percent more area and 68 percent more people falling below sea level after one meter of rise than previous assessments suggested.
Meanwhile, underneath
One year before Seeger's paper, Leonard Ohenhen and colleagues at Columbia's Lamont-Doherty Earth Observatory published findings in Nature Cities identifying Houston as the fastest-sinking major city in the United States. Forty-two percent of Houston's land area is subsiding faster than 5 millimeters per year. Near Fulshear, in the booming western suburbs, the ground is dropping at 1.3 inches per year. More than a foot per decade.
Groundwater extraction drives roughly 80 percent of it. The Harris-Galveston Subsidence District has spent fifty years shifting users to surface water, and in the oldest regulated areas, they've nearly stopped the sinking. But development keeps pushing outward. The fastest-growing suburbs are the fastest-sinking ones. Across the region, groundwater withdrawals have caused about 3,200 square miles to drop more than a foot. Some areas have fallen twelve to thirteen feet.
Aquifer compaction is permanent. You can stop pumping and the sinking slows, but the ground that already dropped is gone the way a squeezed sponge is gone.
Two rulers, one coast
At the Galveston Pier 21 tide gauge, recording since 1904, sea level has risen roughly two feet in a century. The relative rise rate, 6.51 millimeters per year, is nearly four times the global mean. A USGS-supported study found that subsidence contributed as much as 85 percent of that rise.
Stack those corrections. Seeger says the sea surface is higher than models assumed. Ohenhen says the land beneath it is dropping faster than anyone priced in. Independent findings from independent teams using independent methods, and they compound beneath the same coastline. The effective margin of protection for anything built between those two surfaces shrinks from both sides at once.
The Coastal Texas Project's 2021 feasibility study, the document on which the entire $34 billion authorization rests, computed its recommended plan using the Corps' intermediate sea level rise scenario. Corps guidance requires these projections to extrapolate from tide gauge baselines. If the baseline is wrong, the projection is wrong. The error widens over the project's 50-year design life as accelerating sea level rise compounds on top of a starting point that was already too low, while the land beneath continues to sink at rates the feasibility study may have underestimated.
Tina Petersen, executive director of the Harris County Flood Control District, said in February:
"When we have access to better, more accurate information, we can have better decision making on behalf of our community."
She was discussing FEMA's new draft flood maps for Harris County, the first comprehensive revision in nearly twenty years. Her district has at least $3.5 billion in bond-funded projects going to bid over the next three to five years, with $400 million in construction contracts expected in the next twelve months alone. Every one of those projects is engineered to design standards derived from existing flood maps and elevation data.
The new FEMA maps account for increased rainfall and development. They do not incorporate sea level rise projections. They do not incorporate subsidence corrections. They do not reflect either study. During Hurricane Harvey, 68 percent of flooded structures in Harris County were outside the mapped 100-year floodplain. Those buildings weren't in a flood zone because the maps said they weren't. The maps were wrong then. They are being updated now to fix some of what was wrong. The baseline beneath them remains uncorrected.
Better information has arrived. The chain from peer-reviewed paper to revised engineering specification runs through federal guidance updates, state adoption, local permitting. That chain takes years. The Ike Dike's design window is open now. Petersen's capital projects are going to bid now. So the concrete goes in on the old number, because the concrete always goes in on the old number. The bureaucracy will catch up, or it won't, and the pour will have set either way.
The sea is where it is. The ground is where it is. The number in the box is the only thing anyone can argue about, and it was wrong before the first pour.
Things to follow up on...
- The subsidence keeps spreading: A century-spanning review of Houston-area land subsidence documents the sinking migrating from coastal to inland areas, tracking urban expansion into northern Fort Bend and southern Montgomery counties where regulation arrived decades after the pumping started.
- AMOC and the Gulf: A Nature Communications study published May 3 finds that a weakening Atlantic overturning circulation would intensify atmospheric rivers hitting the US West Coast, reshaping precipitation patterns across the continent in ways current infrastructure planning doesn't anticipate.
- Hazardous sites in the path: A companion analysis to the Ohenhen subsidence work identifies 5,500 US facilities at risk of a 100-year flood by 2100 under high emissions, including coastal power plants and petrochemical infrastructure concentrated in Louisiana, Florida, New Jersey, and Texas.
- The preprint pushback: A not-yet-peer-reviewed response to the Seeger baseline study argues that uncertainties in coastal impact modeling aren't resolved by a single correction, contending that multiple biases remain even after the geoid offset is addressed.