Texas coastal flood days and hourly persistence

Two days with a long sampled sequence

NOAA counted two high-tide-flood days at its Rockport gauge during the May 2025 to April 2026 meteorological year. HomeRiskIQ’s analysis of the gauge’s verified hourly heights found 15 readings above the current NOS minor-flood threshold. All 15 occurred consecutively, from October 5 at 17:00 UTC through October 6 at 07:00 UTC, a 14-hour span between observations. [1–3]

Boston’s gauge had six official flood days over that same period, but only six hourly readings above its own minor-flood threshold. Each occurred on a different date. The Texas gauge therefore had fewer flood days but a longer sequence of above-threshold hourly observations. This is an illustrative comparison between selected gauges, not a national ranking or a statement that Rockport properties face more risk than Boston properties.

The readings describe water at gauges. They do not establish the duration of flooding on a street, in a house or across an entire city. Fifteen hourly observations also do not equal 15 continuous hours of flooding: the first and last observations are 14 hours apart.

The October episode across three Texas gauges

The strongest Texas sequences in this selected set clustered on October 5 and 6, 2025. Eagle Point, in Galveston Bay, had a run of 21 above-threshold hourly readings spanning 20 hours. Galveston Pier 21 had nine, spanning eight hours. Rockport’s 15-reading run began later in the same window. These are station-specific observations, not interchangeable measures of conditions throughout the bays or surrounding neighborhoods.

Contemporaneous Texas Division of Emergency Management situation briefs preserved NWS coastal-flood advisories for this episode. An October 6 brief reproduced the Corpus Christi office’s October 5 advisory covering bays as well as Gulf-facing Middle Texas beaches through Monday evening; the October 5 statewide brief also listed Houston/Galveston coastal advisories. Those records establish that coastal flooding was forecast and advisories were issued, not that every warned location flooded. [4–5]

What the annual count leaves out

A flood-day total counts dates meeting a threshold. It does not distinguish a single above-threshold hourly observation from many consecutive observations, or show whether those observations straddle midnight. At Rockport, the sampled sequence crossed the UTC date boundary, contributing to two dates in the annual total. Reporting both frequency and timing gives readers a more precise description of the record.

HomeRiskIQ previously examined this distinction in its Annual Flood Day Totals study. This analysis is a geographically focused extension: it adds three Texas gauges, a dated episode timeline, an illustrative Boston contrast and explicit quality and threshold checks. It does not claim to discover flood duration as a new concept or to reveal an unreported coastal-flood episode. [6]

The numbers depend on the threshold

At the nominal thresholds, the longest runs were 15 readings at Rockport, 21 at Eagle Point, nine at Galveston Pier 21 and one at Boston. Moving each threshold down or up by one centimeter changes the exact lengths. Rockport’s result ranges from 10 to 17 readings; Eagle Point’s ranges from 18 to 38. The Rockport–Boston contrast remains in this diagnostic, while Eagle Point’s precise run length is especially threshold-sensitive.

Values are longest consecutive zero-flag hourly runs, in readings. The one-centimeter perturbation is a sensitivity check, not an instrument-error estimate, confidence interval or historical threshold revision.

A closer check at Rockport

The denser six-minute record contains 141 observations between the endpoints of Rockport’s hourly run. All exceed the nominal threshold, but one carries a flat-tolerance flag. NOAA marks it verified; verified does not mean unflagged. Excluding that observation breaks the all-zero-flag six-minute sequence. This check supports a sustained pattern of elevated gauge readings but does not establish an entirely unflagged continuous record, much less continuous street flooding. [1–2]

The practical takeaway is limited but useful: pair annual flood-day counts with the timing and concentration of gauge observations. Neither metric alone measures property damage, access disruption, future annual risk or a long-term trend.

Methods and reproducibility

Window and units. Verified hourly_height observations cover every UTC hour from May 1, 2025 at 00:00 through April 30, 2026 at 23:00. All six validated gauges have the exact 8,760-point grid, without duplicates, missing timestamps or nonfinite heights. Heights and thresholds are in meters relative to station datum, STND. STND differs among stations; comparisons use each gauge’s own threshold, not raw cross-station elevations. NOAA’s inundation application converts its metadata NOS thresholds by subtracting MHHW, supporting the station-datum interpretation. [1–3,7]

Definitions. An exceedance is a height strictly greater than the retrieved NOS minor threshold. Reading totals are reported both with all values and with zero-flag values. A clean run requires zero flags and exactly one-hour adjacency; a missing or flagged point breaks it. Official flood-day counts are retained separately from the custom strict-threshold screen. For this six-gauge shortlist, custom UTC day counts match the saved official daily and annual products.

Validation. The reproduction script parses JSON numeric tokens directly as Decimal. It checks station identity, the complete timestamp set, finite values, daily-date coverage, each daily maximum and maximum hour, minor/moderate/major flags and annual flag sums. All 2,190 station-days pass. Every selected above-threshold hourly value is unflagged, but below-threshold flagged or inferred records exist; the complete series must not be called wholly clean or directly measured.

Selection. The discovery screen attempted 22 purposively chosen stations, initially 12 and then 10 additional Texas/Florida stations after a Texas contrast appeared. Nineteen succeeded. Freeport and Port Isabel lacked a successful threshold response; Naples returned no hourly data. Those failures are missing, not zero. The six-station validation set was selected after exploration. Boston is an illustrative contrast, not the least-persistent or safest city. This is neither a national sample nor a complete Texas survey.

Threshold boundaries. Toke Point and Grand Isle, outside the validation shortlist, differ between the strict arithmetic screen and official totals at observations equal to rounded metadata thresholds. Their screen values are not substituted for official totals. Thresholds are a current retrieved snapshot, not proven historical thresholds at every date. Exact equality is excluded from the custom hourly metric.

Source record and reuse limits

All source data were retrieved October 24, 2026. The archived provenance records full request URLs and download-completion times reconstructed from file modification times; these are not independently logged HTTP response timestamps. The evidence archive freezes 48 raw NOAA files with SHA256 hashes and sizes. The offline script verifies those bytes and hourly request parameters before analyzing them, preventing accidental reuse of a different year, product or datum.

Run reproduce.py with Python 3 to recreate the publication table and sensitivity results from the frozen evidence. The numerical reproduction uses only the standard library. charts.py uses matplotlib; build_document.py uses python-docx. The archive includes editable SVG graphics, PNG versions, a machine-readable table and the full selected-gauge results. A package manifest fingerprints code, data, methods and outputs.

The sampled heights cannot identify flooded homes, establish a damage total, prove a meteorological cause or attribute conditions to climate change. No property addresses are inferred. The warning context is corroborative, not an observed-impact dataset. Reuse of a local warning as proof of specific property flooding would exceed this evidence.

Sources

[1] NOAA CO-OPS Data API and response fields. Hourly and six-minute water levels. https://api.tidesandcurrents.noaa.gov/api/prod/ and https://api.tidesandcurrents.noaa.gov/api/prod/responseHelp.html

[2] NOAA station threshold metadata. Example Rockport request: https://api.tidesandcurrents.noaa.gov/mdapi/prod/webapi/stations/8774770/floodlevels.json?units=metric

[3] NOAA high-tide-flood daily and meteorological-year annual products. https://api.tidesandcurrents.noaa.gov/dpapi/prod/webapi/htf/htf_met_year_annual.json?met_year=2025 and https://tidesandcurrents.noaa.gov/high-tide-flooding/annual-outlook

[4] Texas Division of Emergency Management State of Texas Situation Awareness Brief, October 6, 2025. https://olympus.soc.texas.gov/services/Sitrep/GetFile.aspx?id=15656

[5] Texas Division of Emergency Management State of Texas Situation Awareness Brief, October 5, 2025. https://olympus.soc.texas.gov/services/Sitrep/GetFile.aspx?id=15653

[6] HomeRiskIQ, Annual Flood Day Totals. https://www.homeriskiq.com/research/annual-flood-day-totals

[7] NOAA inundation application source, getHTFThreshold function. https://tidesandcurrents.noaa.gov/inundationdb/assets/js/inundation.js