Causes of accelerated High-Tide Flooding in the U.S. since 1950

High-tide flooding (HTF) along U.S. coastlines has risen rapidly in recent decades, increasingly disrupting transportation, businesses, and property. Although individual events are often minor, their cumulative impacts can exceed those of less frequent extremes, motivating improved understanding and projections to support adaptation and mitigation planning. HTF arises from the interaction of tides, storm surges, and anomalies in relative mean sea level (RMSL). Prior work linked decadal HTF variability to tidal modulations, lunar and solar declination cycles, and interannual RMSL variability, with anthropogenic alterations within estuaries also modifying tides and HTF. However, the long-term increases in HTF have been broadly attributed to RMSL rise reducing the gap to local flood thresholds. Global mean sea level has risen since 1900 due to steric expansion and barystatic mass inputs from ice melt, with acceleration since the 1960s. Regional departures from the global mean reflect gravity, rotation, and deformation (GRD) fingerprints of mass changes, sterodynamic variations driven by circulation, winds, and pressure, and vertical land motion (VLM). Along U.S. coasts, these processes have produced RMSL rise of about 0.9–7.0 mm/yr since 1950. Yet, a clear local attribution of HTF changes to these contributors has been lacking. This study quantifies the contributions of RMSL and its components (GRD, SDSL, VLM) to HTF changes across U.S. coasts and selected Pacific islands since 1950, and examines their temporal variability from seasonal to decadal scales.

The study synthesizes prior findings that: (1) HTF variability and trends are influenced by astronomical tidal modulations and long-period nodal and declination cycles; (2) anthropogenic modifications within bays and estuaries can alter tidal amplitudes and contribute to HTF; (3) rising RMSL is the dominant driver of long-term HTF increases by elevating the baseline relative to flood thresholds; (4) global mean sea-level rise results from steric expansion and barystatic mass additions, with acceleration in recent decades; (5) regional sea-level patterns deviate from the global mean due to GRD fingerprints, sterodynamic processes (circulation, winds, pressure), and VLM; and (6) prior reports document widespread recent HTF increases along U.S. coasts and provide impact-based thresholds (NOS, NWS). The paper builds on this by explicitly decomposing observed HTF changes into RMSL components at local scales.

Data and sites: Hourly verified water levels from 41 NOAA tide gauges around the contiguous U.S. (CONUS) and six Pacific island sites were analyzed. Records with at least 90% hourly data coverage in both comparison periods (1950–1968 and 2002–2020) were retained. Primary flood thresholds were NOAA National Ocean Service (NOS) nominal thresholds tied to local impacts; National Weather Service (NWS) thresholds were used in sensitivity analyses where available. All levels are referenced to a control period (1950–1968). HTF definition and RMSL removal: RMSL time series were derived using a 30-day cutoff LOESS smoothing (MATLAB smooth, LOESS) to retain seasonal-to-interannual variability and long-term rise. RMSL anomalies were referenced to the control period mean. Control water levels were constructed by subtracting the monthly RMSL from hourly verified levels, representing tide and surge contributions under 1950–1968 baseline conditions. HTF days were counted from hourly exceedances of the local threshold. HTF attributed to RMSL was computed as the difference between HTF days from raw hourly levels and from control water levels. Attribution of RMSL components: RMSL was decomposed into three contributors: (i) GRD fingerprints (barystatic mass change fingerprints, 100-member annual ensemble from published estimates), spline-interpolated to hourly values under the assumption of minor intra-annual variability far from freshwater sources; (ii) VLM, estimated as the difference between local tide-gauge RMSL and a nearby hybrid sea-level reconstruction (monthly resolution), following prior work, yielding VLM time series that can be nonlinear at some sites (e.g., Texas) and consistent with independent geodetic estimates; and (iii) sterodynamic sea level (SDSL), inferred as the residual after removing GRD and VLM from RMSL. HTF decomposition method: For each day with one or more hourly threshold exceedances, the fractional contributions of each RMSL component to the exceedance were computed and averaged across hours. Annual HTF days attributed to each component were obtained by summing daily fractional contributions. This ensures the sum of component-attributed HTF equals HTF due to RMSL and component water levels sum to total RMSL. Uncertainty quantification: At each site, 10,000 Monte Carlo realizations were generated accounting for uncertainties in the individual components (e.g., GRD ensemble), and annual HTF attributions were reported as ensemble means with one-sigma standard errors. Regional and temporal analyses: Spatial patterns were summarized by comparing 19-year periods (1950–1968 vs 2002–2020). Temporal evolution was examined using decadal sums of HTF days and decadal RMSL trends for regional “virtual stations” representing the CONUS West Coast, Gulf of Mexico and South Atlantic Bight, Mid-Atlantic Bight, and Gulf of Maine. Sensitivity to threshold choice (NOS vs NWS) was evaluated, and relationships between RMSL rise and HTF changes were explored relative to local threshold placement within water-level distributions. Data availability: Processed datasets and code are archived at https://doi.org/10.5281/zenodo.7677796.

• RMSL rise is the dominant long-term driver of HTF increases: RMSL accounts for over 84% of the HTF increases at 33 of 41 sites when comparing 1950–1968 with 2002–2020. Sites with smaller changes showed minimal or no increase. - Spatial patterns: Large increases in annual HTF days were observed along the U.S. East Coast, especially in the Mid-Atlantic Bight and Gulf of Maine. Examples include increases of about 8 days/year in Atlantic City and Boston, and 13 days/year in Eastport. Increases along the Southeast and Gulf coasts range roughly from 1.3 (St. Petersburg) to 6.5 (Galveston Pier) days/year; along the West Coast from 0.3 (Port San Luis) to 3.6 (San Diego) days/year. - RMSL vs HTF correlation: Spatial patterns of RMSL rise only moderately correlate with HTF increases (r ≈ 0.57). Differences arise from local flooding thresholds and water-level distribution characteristics, which control sensitivity to baseline shifts. - Component attribution varies regionally: • Northeast and Mid-Atlantic (north of Cape Hatteras): VLM dominates increases in HTF days in many locations (e.g., 46–57% between Virginia and New York, 2.4–3.6 days/year; up to 48–63% around Long Island Sound though magnitudes are <1 day/year), reflecting GIA-driven subsidence gradients. • Gulf of Maine: Despite relatively smaller GRD-induced RMSL rise, GRD contributes substantially (23–32%; 1.6–3.9 days/year) due to thresholds situated in sensitive parts of the water-level distribution and reduced VLM contributions; overall HTF increases are among the largest nationally. • Estuarine/tidal river sites (e.g., Baltimore, Washington, D.C., Providence): SDSL is the leading contributor (36–56%; 1.5–2.6 days/year), likely reflecting influences from river discharge, tides, and local wind forcing. • South Atlantic Bight and Gulf of Mexico: SDSL often dominates (≈50–80% in the South Atlantic Bight; 1.5–2.0 days/year). In the Gulf of Mexico, SDSL is generally leading, with exceptions at Galveston (Pier 21, Pleasure Pier) where VLM (~47%) and SDSL (~45%) contributions are comparable. • West Coast: SDSL leads but HTF increases remain smaller (0.3–3.6 days/year) given historically suppressed RMSL rise (≈1–2 mm/yr) due to Pacific decadal variability; VLM contributions vary, with decreases where uplift occurs. • Pacific islands: Few or no HTF days under NOS thresholds, but more (up to 28 additional days) under NWS thresholds; process percentage contributions are similar across thresholds, with SDSL generally dominant and VLM highly variable (−21% to 27%). - Temporal evolution and accelerations: The Gulf of Mexico experienced a ≈220% increase in HTF frequency over the last decade, linked to recent SDSL acceleration that shifted water-level distributions. In the Mid-Atlantic Bight and Gulf of Maine, decadal SDSL trends exceeded 10 cm/decade in the early 2000s, yielding 45–100% increases in HTF relative to previous decades; despite moderating RMSL trends in the 2010s, HTF increases accelerated due to threshold sensitivity. - Case study (Sewell’s Point, Norfolk, VA): Of 280 observed HTF days since 1950–1968, only 36 would have occurred without RMSL changes; maximum annual HTF decreased from 14 to 4 in the no-RMSL scenario. Local VLM is ~3.4 mm/yr; GRD accelerates over time; SDSL exhibits variability and trend. The nonlinear relation between threshold exceedance counts and RMSL implies linear VLM trends do not translate linearly to HTF-day increases.

The study addresses the attribution of rising HTF frequency to specific physical processes affecting RMSL. Findings confirm that RMSL rise explains most long-term HTF increases, but also reveal that local sensitivity depends strongly on flood-threshold placement within the water-level distribution and on variance characteristics, explaining only moderate spatial correspondence between RMSL rise and HTF increases. Regionally, VLM dominates in the northeastern U.S. due largely to ongoing GIA-related subsidence, implying persistent HTF increases irrespective of climate mitigation. Elsewhere and on seasonal-to-decadal time scales, SDSL variations—driven by ocean circulation, wind, and pressure—play the leading role, making HTF highly sensitive to internal climate variability (e.g., Pacific Decadal Oscillation) and anthropogenic forcing. GRD fingerprints have recently grown in importance but typically contribute less than VLM and SDSL. The marked recent acceleration of SDSL in the Gulf of Mexico illustrates how decadal dynamical changes can rapidly amplify HTF. These insights enhance physical understanding needed for both short-term forecasting and long-term coastal risk planning, and underscore that adaptation strategies must consider both persistent (VLM, GRD) and variable (SDSL) drivers and local threshold sensitivities.

This work provides a comprehensive, process-based attribution of HTF increases along U.S. and selected Pacific island coasts since 1950. RMSL rise accounts for the vast majority of long-term HTF increases, with VLM dominating in the northeast and SDSL elsewhere and at shorter time scales; GRD contributions are increasing but remain secondary. Spatial disparities in HTF growth arise from local threshold positioning within water-level distributions and regional hydrographic dynamics. Practically, communities in regions with strong VLM (e.g., northeast) will face continued HTF increases even under aggressive climate mitigation, while regions where SDSL dominates (e.g., Gulf of Mexico, South Atlantic Bight, parts of the West Coast) may experience pronounced decadal variability and accelerations driven by ocean dynamics. Future research should further resolve oceanographic mechanisms behind SDSL accelerations, quantify and monitor nonlinear VLM, refine threshold selection and local hydrodynamic influences (tides, river discharge, winds), and improve integrated forecasts and projections of HTF to inform adaptation.

• Flood-threshold dependence: Results depend on local thresholds (NOS vs NWS). While percentage contributions were robust across thresholds in most regions, absolute HTF counts vary, and NOS thresholds are not available everywhere. - Control-period sensitivity: HTF counts are sensitive to the choice of control period, though the conclusion that RMSL drives most increases is robust. - Component estimation assumptions: GRD fingerprints were interpolated from annual to hourly assuming minimal intra-annual variability; SDSL was inferred as a residual; a classical sea-level budget closure was not the aim. - VLM estimation: Derived from differences between tide gauges and a hybrid reconstruction; while consistent with independent estimates, uncertainties remain and some sites exhibit nonlinear VLM. - Data coverage: Some sites (e.g., Rockport, Galveston Pleasure Pier) have data gaps; the analysis assumes available data are representative of each 19-year period. - Spatial representativeness: "Virtual stations" summarize regional behavior and may mask site-specific dynamics. - Moderate RMSL–HTF correlation: Local water-level variance and threshold placement limit the predictability of HTF increases from RMSL trends alone.