The most useful finding in the federal sea level projections published in 2022 is not a number about the ocean. It is a statement about the projections themselves. NOAA’s application guide for the report records that its authors analyzed when the five scenarios separate from one another and “found that most of the divergence occurred after 2050. In other words, there is less uncertainty about the magnitude of SLR in the near term (i.e., before 2050), relative to the uncertainty in the more distant future.”
For a highway agency that changes what the science is good for. The next quarter century is the part the projections constrain tightly, and it happens to be the horizon on which capital programs, pavement rehabilitations and drainage retrofits are actually decided. The century-scale figures remain a planning input for the assets with the longest lives, which is a different question and belongs alongside how long bridges last.
What is exposed
FHWA’s estimate of the exposed network is unusually crisp for a figure of this kind. Hydraulic Engineering Circular 25, the agency’s coastal highway manual, focuses on roads “influenced by coastal tides and waves constantly, or occasionally during storms,” and states: “FHWA estimates that there are more than 60,000 miles of these ‘coastal highways’ in the US.” The manual’s authors put the same number in the peer-reviewed literature, as “over 60,000 miles (100,000 km) of coastal highways in the United States (US) that are occasionally exposed to coastal waves and water levels.”
The headline projection against that network, as NOAA states it, is that sea level along the US coastline is predicted to rise 10 to 12 inches over the next 30 years, and that flooding “is expected to occur, on average, more than 10 times as often as it does today.” The report was published in 2022, so that window closes near mid-century, inside the horizon where the scenarios still agree.
NOAA attaches limits to that projection which rarely survive requotation. The extreme water level analysis behind the flood-frequency claim excludes wave-driven water levels, which the application guide notes “can contribute 25-90% of extreme water levels” on exposed coastlines, so on a wave-swept shore the analysis is describing well under half of the water that arrives. The 1 percent annual chance levels in the same analysis “are not the same as those found in the Federal Emergency Management Agency’s (FEMA’s) regulatory products,” which rules them out as a substitute for a floodplain determination.
The flooding has a name problem, which is a data problem
HEC-25 devotes space to terminology because the terminology has shaped the record. The manual lists what the same phenomenon has been called: nuisance flooding, sunny-day flooding, high-tide flooding, chronic flooding, recurrent flooding, king-tide flooding. It settles on “increased flooding due to relative sea level rise” and explains the mechanism in plain terms. “While flooding happened before, the gradual RSLR over the past decades has brought the sea levels closer to the infrastructure levels. Areas that once flooded only every few years during major storms may be flooding several times per year.”
The failure mode on a paved street is frequently not the ocean arriving overland. It is salt water backing up through the storm drainage system at high tide, which the manual illustrates with photographed examples on Florida SH A1A at Hollywood Beach, on Dock Street in Annapolis and on an urban road in Honolulu. A drainage network sized to move rainfall outward becomes a pipe that admits tide inward, and no amount of pavement design addresses it. The inland version of the same regulatory problem is worked through in designing highways for flood resilience, and there is more on climate resilience across this section.
What inundation then does to the structure is slower and less visible. In its 2023 stocktake of pavement resilience, FHWA records that coastal flooding driven by sea level rise “tends to be slow moving and non-turbulent,” and that several coastal states “have also experienced such an increase in saturated soils that roads are sinking while sea level rises.” The review’s assessment of the remedy is blunt: “While there is not a particular design or material solution for this issue, frequent overlay is commonly used to maintain service in these locations.” A rising overlay schedule is a maintenance cost that never appears as a resilience project.
What raising a road actually buys
The clearest published case of the arithmetic sits in HEC-25’s chapter on relative sea level rise, drawn from an analysis of McNair Road in Annapolis, Maryland. The historic linear rate of rise at Annapolis is 3.61 millimeters per year, which is roughly a foot above today’s level by 2100. On that historic rate alone, the manual reports, “the roadway will be flooded about 80 times per year by 2060.” Under curves adding 1, 2 or 3 feet of rise this century above the historic trend, the count reaches 170, 300 or 430 floods per year by 2060.
Then the countermeasure. Raise the road one foot, and “the annual number of floods would reduce to just a few initially.” The manual’s phrase for the effect is that it would “turn back the flooding clock.” The same passage sets the expiry date on that effect. The elevated road “would flood as frequently as it is now by about 2065 to 2070,” under roughly 2 feet of rise this century.
One foot of elevation, on that road, under that scenario, buys roughly four decades and then the problem returns to its starting point. That is a defensible purchase. It is not a solution, and describing it as one misstates what the money did.
The social edge of the same work is visible in the manual’s Miami Beach example, where Sunset Harbour Drive was raised several feet and the local stormwater system re-engineered with pumps to move runoff up and out from behind the raised road. The retail businesses beside it remain at the original road elevation. Raising a street relocates the low point rather than removing it.
Five categories, not three
Public debate compresses coastal adaptation into elevate, armour or retreat. FHWA’s own taxonomy has five entries: manage and maintain, increase redundancy, protect, accommodate, relocate. Redundancy is the one most often missing from the shortlist, and the manual’s examples are concrete, including enhanced ferry service and a preference for “constructing or enhancing closely spaced roads versus one road.”
Under protection, the manual makes an observation about timing that indicts a good deal of past practice: “Many highway revetments were built in immediate response to a specific storm but the vulnerability prior to the storm had been increasing for years due to longer term processes of shoreline recession and sea level rise.” Structures also age against a moving target, since existing seawalls “may have to be modified to withstand higher wave conditions as sea level rises.” Soft and hybrid approaches carry evidence rather than aspiration: during Hurricane Sandy, a relic stone seawall buried beneath a nourished dune and beach “reduced wave forces by a factor of two, effectively protecting the upland infrastructure, including a coastal highway.”
Accommodation contains the manual’s single strongest claim, and it is a narrow one. “Increased elevation is the only proven adaptation option for coastal bridges subject to wave attack during extreme events.” The named replacements are the I-10 bridge over Escambia Bay near Pensacola, the I-10 bridge over Lake Pontchartrain near Slidell, and the US 90 bridges over Bay Saint Louis and Biloxi Bay. Accommodation also runs the other way, in an option that reads as counterintuitive until the damage mechanism is understood: “Lowering roadway profiles to allow overwash without pavement damage during extreme events.”
Relocation is already history in several places. Storm-related bluff erosion moved a portion of California 1 in April 2013. A coast-parallel road in Texas was abandoned after storm damage in 1989 because landward relocation was constrained by wetlands in a National Wildlife Refuge, and similar ecological constraints drove consideration of moving miles of NC 12 off its barrier island onto a bridge in the bay. Elsewhere, the manual notes, “relocation is constrained by private property.”
Retreat also happens without being chosen. In southern Dorchester County, Maryland, a significant portion of the roads flood on some high tides. HEC-25 puts the affected share at sometimes more than half, and records that residents “have adjusted to this reality though lifestyle changes (e.g. realizing that on some days they cannot drive to the city, buying vehicles with higher clearance, etc.).” No project was cancelled and no route was decommissioned. Service simply degraded until behavior changed around it, which is the outcome that asset-management triage produces by default rather than by decision, as rebuild prioritisation tends to show.
None of this becomes usable until someone converts a regional projection into a local elevation. Rise varies by coast, vertical land motion varies by county, and Dorchester County’s problem is subsidence as much as ocean volume. A national figure describes the sea. The number that decides whether a particular road floods is the difference between a tide gauge datum and the surveyed elevation of the pavement, and nobody in Washington can supply it.