# Building floor elevation coverage — 2026-10-20

**No: we do not know every building's floor height above ground.** The study contains 15,840 OSM building polygons. Public City elevation certificates provide a useful, small surveyed-record subset; neither the footprint nor roof elevation establishes the floor or lowest opening. No UI, hydraulic inputs, or simulation data were changed by this audit.

## Fresh official records and coverage

Fresh retrieval from [City PublicSiteGIS layer 87](https://nbgis.newportbeachca.gov/arcgis/rest/services/PublicSiteGIS/MapServer/87) and its [all-record query](https://nbgis.newportbeachca.gov/arcgis/rest/services/PublicSiteGIS/MapServer/87/query?f=geojson&where=1%3D1&outFields=*&outSR=4326) returned **343 records**. Raw data: `floor-certificates-fresh-20261020.geojson`; schema: `floor-certificates-layer-20261020.json`. The cached display file has 342 geometries because the additional record (UNIQUE_TAG 1369, 127 BAY FRONT S) has **null geometry, floor, date, and datum**, rather than a newly usable measurement. Its location cannot be independently checked from that geometry.

| Measure | Result |
|---|---:|
| All study building footprints | 15,840 |
| City certificate records | 343 |
| Records with numeric top-of-bottom-floor value | 337 |
| Numeric, unambiguous datum, dated records | 332 |
| NAVD88 / NGVD29 / unknown or malformed datum records | 151 / 186 / 6 |
| Geometries / distinct parcel geometries | 342 / 295 |
| Conservative unique building address-and-parcel matches | 160 (1.01% of 15,840) |
| NAVD88 subset of matched buildings | 77 (0.49% of 15,840) |
| Certificate date range | June 5, 1986 – July 24, 2024 |

These are **parcel/address polygons**, not surveyed building outlines or floor-measurement points. Unit records can share a parcel, geometry, and floor value. Record totals are not unique-building coverage. The fresh [City elevation-certificate portal](https://nbgis.newportbeachca.gov/gispub/NewportBeachElevationCerts/default.aspx) provides a broader address inventory of 1,803 rows: 343 marked Y, 1,458 marked N, two blank. It exposes the same certificate count and no newer certificate date than the GIS subset. This is an address/unit inventory, not a complete footprint inventory; do not divide its counts by the study building count as a coverage estimate.

## Matching method and limits

We spatially queried the City's parcel polygons against actual study footprints, requiring the footprint's interior representative point to be covered by the parcel; then required an exact normalized street number/address match between the published City address and the footprint's OSM address. Normalization only capitalizes, removes punctuation and standardizes explicit street/direction words. No nearest neighbor, guessed address, parcel-center point, or proximity match was used. Among all 343 records, 337 parcels contain one footprint representative point, one contains two, and five contain none (including the null geometry). Only 179 records have one address-consistent spatial candidate; after requiring numeric floor, known datum, certificate date, and deduplicating building IDs, 160 unique buildings remain. No conflicting floor/datum pairs occur among those matched duplicates.

The other 164 records have no exact address-consistent spatial match under this conservative rule; many could be recovered through original document, parcel, directional-address, or multiunit review. A parcel's sole footprint is still only a candidate when the published address disagrees. Even the 160 matches remain **record-to-footprint candidates, not original-certificate review or field validation**. Changes since a certificate date, basement/enclosure diagrams, unit-specific floors, and a building's lowest opening require checking original certificate and permit/as-built evidence. Do not label this 1.01% as independently verified current interior flood thresholds.

## Floor elevation versus height above ground

`Top_of_Bottom_Flr` is an absolute elevation in the indicated vertical datum, interpreted as feet from the City/FEMA certificate context; it is not inherently a height above surrounding ground or the occupied living floor. NGVD29 values cannot be directly subtracted from NAVD88 terrain. The malformed `NAVD_88_X` value `7.986` is excluded rather than silently interpreted as a datum flag.

For 77 NAVD88 matched buildings, the JSON includes illustrative native 0.5 m bare-earth samples from existing local 2023 OPR TIFFs at each footprint representative point, transformed to EPSG:6340. The computation is `published_floor_ft × 0.3048 − bare_earth_NAVD88_m`. The full sample coordinates, source TIFF, source floor value, date, ground value, and difference are retained in `building-floor-elevation-coverage.json`.

These samples are **not surveyed adjacent grade**. Bare-earth terrain beneath a roof can be interpolated; it does not resolve slab steps or thresholds. For example, 310 38TH ST has published floor 8.4 ft NAVD88 and sampled ground 2.432 m, an apparent difference 0.128 m (~5.1 in), whereas 308 38TH ST yields 1.288 m. Some differences are negative. Those discrepancies demonstrate why these differences must not automatically become building ingress heights or accuracy claims. Prefer the same certificate's lowest/highest adjacent grade and relevant opening/floor diagram, then verify current conditions.

## Additional public data checked

The [City elevation-certificate information page](https://newportbeachca.gov/government/departments/community-development/building-division/floodplain-management-information/elevation-certificates) explicitly provides copies **if the City has a certificate**, with online search options. The portal's Document links target the City's Laserfiche CDD Docs collection with document type Elevation Certificates and a specific street number/name. The [City permit reference guide](https://apps.newportbeachca.gov/pwizard/iframe.aspx?id=3&step=114) includes elevation-certificate and slab/floor inspection requirements. These are promising recovery paths for existing records, not evidence of complete building coverage. The [public records page](https://www.newportbeachca.gov/government/departments/public-records) describes further record access; this audit sent no record requests or messages.

The current [Orange County building-footprint layer](https://www.ocgis.com/arcpub/rest/services/Map_Layers/Building_Footprints/FeatureServer/0) exposes `Height`, `ABS_HT`, geometry, and address fields, but no finished-floor or lowest-opening fields. Schema retained in `oc-building-footprints-layer-20261020.json`. These building-height attributes do not establish floor elevation. Existing OSM has 9,218 `height` tags, 1,131 `building:levels`, nine `ele`, zero `min_height`, and zero `building:min_level`. Roof height and floor count do not establish first floor elevation. LiDAR/DSM roof-minus-ground measurements similarly do not provide a surveyed interior floor or door opening, and cannot establish complete floor coverage.

## Insurance-company records and public availability

The user's insurance-data lead is useful: a homeowner or insurer may possess an elevation certificate not already in the municipal index. But that does **not** make insurer files a public complete building inventory. [FEMA's current NFIP data FAQ](https://agents.floodsmart.gov/flood-maps-and-data/faqs-about-nfip-data) describes anonymized public policy/claims data, coordinate truncation to 0.1 degree (approximately six miles), and elevation fields suitable for descriptive aggregation. Such records cannot be defensibly joined to individual study footprints. Repeated policy transactions also are not distinct buildings.

[FEMA's Risk Rating 2.0 FAQ](https://agents.floodsmart.gov/sites/default/files/media/document/2026-01/fema-nfip-risk-rating-2.0-FAQs.pdf), question 10, states certificates are no longer required to purchase coverage under that rating system; owners can optionally submit one. FEMA-sourced rating elevations must not be treated as a public survey for every building. The [current FloodSmart certificate page](https://www.floodsmart.gov/get-insured/elevation-certificates) directs owners first to local floodplain managers and otherwise to a surveyor, engineer, or architect. Therefore, widespread insurance coverage would not imply widespread certified floors.

The next useful feature can distinguish surveyed/document-backed floors, inferred ground comparisons, and unknown floors. Before changing any water physics or predicting indoor entry, retrieve the underlying certificate for matched buildings, read its building diagram and grade/opening details, resolve units/datums, and preserve observation dates and uncertainty. There is no established all-building finished-floor dataset in the public sources examined here.

## Reproduction

`audit_building_floor_coverage.py` produces the retained JSON and portal summary using existing study footprints and downloaded snapshots; it only reads local source TIFFs and writes research artifacts. Full portal HTML is retained as `newport-elevation-certs-portal-20261020.html`, parsed inventory as `floor-certificates-portal-summary-20261020.json`. All matching counts use the same 15,840 study footprints; citywide portal row counts are reported separately.

## Addendum: USACE NSI offers broader *estimated* foundation heights

A live [USACE NSI API](https://www.hec.usace.army.mil/confluence/nsi/technicalreferences/2026/api-reference-guide) query around Balboa Island (center -117.8908, 33.6073, ±0.001°) returned **145 structure points**. Of these, 139 points fall inside 129 distinct study footprints; six fall outside them. Multiple points can represent one OSM footprint. This neighborhood result establishes local availability, **not coverage of all 15,840 buildings**. No nearest-neighbor joins were used.

Sample NSI `fd_id=44163700` at (-117.891057, 33.607332) reports crawl foundation (`C`), `found_ht=2` feet, `ground_elv=7.616307` feet and `grnd_elv_m=2.3214502` meters, with Bing footprint provenance. The sample includes five recurring foundation-height values: 0.25, 0.5, 0.75, 2 and 5.25 ft. Stored raw response: `nsi-balboa-sample-20261020.geojson`; request URL and spatial match audit: `nsi-balboa-sample-request-20261020.json` and `nsi-balboa-sample-audit-20261020.json`.

[USACE's 2026 technical documentation](https://www.hec.usace.army.mil/confluence/nsi/technicalreferences/2026/technical-documentation) describes a modeled national inventory, with foundation type sourced from parcel data where available or statistical/regional assignment, and foundation heights assigned by foundation type. It gives illustrative defaults of 0.75 ft for RES1 slabs and 2 ft for basements. Ground elevations derive from the 10 m USGS NED source. Its warning explicitly covers errors in individual locations and foundation heights. `found_ht` is a relative height in feet, not an absolute NAVD88/NGVD29 floor survey. The returned point attributes do not supply a record-level vertical datum or observation date validating these values. Do not add NSI ground elevation to foundation height and label the result a surveyed NAVD88 floor.

NSI is a credible source for a separately labeled **estimated foundation-height** feature, subject to spatial matching and local review. It does not establish lowest openings, split-level floors, basements, occupied floor diagrams, or current building-specific ingress thresholds. It must remain distinct from certificate-backed elevations; a 2026 release is not a 2026 site survey. No NSI data were added to the app or hydraulic model.

## Addendum: FEMA LOMA/LOMR records

[FEMA's official MSC training](https://emilms.fema.gov/is_0273/groups/173.html) confirms that Letters of Map Change are publicly discoverable under MSC **Search All Products**. LOMA/LOMR-F submissions may use elevation certificates as supporting evidence, as described in the [FEMA elevation-certificate instructions](https://www.fema.gov/pdf/library/elvcert.pdf). These are a further case-specific lead for building elevations and adjacent grade. Public determination-letter availability does not establish that every supporting certificate is publicly attached or that every study building has a case. This audit did not retrieve local determination cases or count additional verified building floors from them.