FEMA NFIP Repetitive Loss & Flood Insurance Market Screener

Give it a US coordinate. It resolves that point to the NFIP community whose published FEMA boundary contains it, then returns that community's repetitive-loss property profile, its Community Rating System class and premium discounts, its NFIP participation status, and the county's flood-insurance market penetration - four FEMA OpenFEMA datasets joined in one call.

Nobody publishes these four together. FEMA's repetitive-loss list was historically FOIA-restricted; it went public as an API in February 2024, the community boundary layer in February 2024, and the county penetration rates in November 2024.

Why there is no radius search, and why that is the honest answer

The obvious build is "repetitive-loss properties within N miles". That product cannot be built from this data, and anyone selling it is selling a fabricated number.

FEMA redacts the coordinates. Its own data dictionary says so:

"Approximate latitude of the structure redacted to 1 decimal place. The precision has been lessened to ensure individual privacy. This redaction as well as the inherent error in geocoding addresses may result in a point location that exists in an incorrect community, county or state."

Measured on the live file 2026-08-19: of Colorado's 113 records, 84 carry one decimal place, 23 carry zero decimal places, and 6 have no coordinate at all. On a 4,000-row Louisiana page it is 2,909 / 1,068 / 23. Zero decimals of latitude is roughly +/- 55 km.

So this actor answers at the scope the data actually supports: point -> NFIP community -> community and county rollup. Every row carries multiple_loss_coordinates_redacted, radius_screening_supported: false, and the decimal precision measured on that run.

Who it is for

• Phase I ESA and environmental due-diligence firms adding a flood-loss and flood-insurance-market section to a site report.
• Commercial real-estate acquisition and underwriting teams screening a portfolio for communities with concentrated repetitive-loss exposure and thin insurance take-up.
• P&C and flood underwriters / MGAs pricing against the CRS discount schedule and the county SFHA penetration rate.
• Floodplain managers and mitigation consultants benchmarking their community's repetitive-loss count, mitigation rate and CRS class against the county.

Example input

{
  "assets": [
    { "lat": 39.7392, "lon": -104.9903, "label": "Denver City Hall, CO" },
    { "lat": 39.7047, "lon": -105.0814, "label": "Lakewood, CO (city inside Jefferson County)" },
    { "lat": 29.9511, "lon": -90.0715, "label": "New Orleans, LA" }
  ],
  "includeCountyCommunity": true,
  "includeTribalEtjLayer": true,
  "includeLossProperties": false
}

What comes back

Three row types, distinguished by record_type:

A site inside an incorporated place usually returns two communities: the municipality (which governs, because FEMA defines a community as "the lowest level of government with land use authority") and the county community covering unincorporated areas. Both are emitted, with is_governing_community and governing_community_basis saying which is which and why. Measured example: Lakewood CO resolves to CID 085075 (Lakewood, City of) as the governing community and CID 080087 (Jefferson County *) as the broader one - two different CRS classes and two different loss rollups.

What this actor refuses to do

• It will not publish a distance to a repetitive-loss property. See above.
• It will not report "not participating in the NFIP" for a community whose Status Book record it never read. Unchecked is null; checked-and-negative is false.
• It will not derive a CRS premium discount from the class rating. FEMA's own file contradicts the textbook rule: Colorado carries class-10 communities published with a 10% and a 15% SFHA discount alongside class-10 communities with none. Discounts are transcribed verbatim, with the raw FEMA text beside the parsed number.
• It will not recompute or clamp a penetration rate. FEMA publishes 593 counties with zero SFHA residential structures alongside a non-zero SFHA policy count, and 27 counties with an SFHA penetration rate above 100%. Those figures ship as published, with penetration_data_quality_flags naming the inconsistency.
• It will not coerce a corrupted census block group. On a live Louisiana page 96.1% of block groups arrive as Excel scientific notation ("2.2055E+11") and have irrecoverably lost seven of twelve GEOID digits. Those are emitted as null with an explicit reason.
• It will not produce a screen level while any contributing FEMA dataset was unavailable. It reads INCONCLUSIVE instead.
• If it cannot answer at all, it fails the run and bills nothing.

Live drift gate

Nine probes run before a single billable row is produced, each with its own error boundary and a three-valued outcome (pass / fail / unavailable):

A load-bearing probe that cannot complete fails the run. A corroborating probe that cannot complete lets the run proceed and says so (

drift_gate_status: verified_degraded

Use as an MCP tool

This actor is callable by AI agents over Apify's MCP server.

{
  "mcpServers": {
    "apify": {
      "command": "npx",
      "args": ["-y", "@apify/actors-mcp-server", "--actors", "malonestar/fema-repetitive-loss-flood-market-screener"],
      "env": { "APIFY_TOKEN": "<your Apify token>" }
    }
  }
}

Worked agent prompt:

For 1600 Broadway, Denver CO and 900 Camp St, New Orleans LA, tell me which NFIP community each sits in, how many repetitive-loss properties that community has, what its CRS class and SFHA premium discount are, and what share of the county's SFHA homes carry flood insurance.

Billing is unchanged when called as a tool, and a run that cannot answer fails without billing.

Sources

All keyless. Counts measured 2026-08-19.

Output fields (196 declared)

Every row: site, sources and provenance

NFIP community identity and boundary

NFIP Community Status Book: participation, FIRM dates and CRS

Repetitive-loss rollup for the community

County flood-insurance market penetration

Screen verdict

Per-property rows (opt-in)

FAQ

Can it tell me how many repetitive-loss properties are within a mile of my site? No, and neither can anything else built on this file. FEMA redacts the coordinates to one decimal place, and 20-27% of records carry zero decimals. See the section above.

What is the difference between NFIP RL and FMA RL? They are four independent flags, not a hierarchy. nfipRl / nfipSrl are insurance-side designations driven by claim history; fmaRl / fmaSrl are grant-side designations that drive Flood Mitigation Assistance eligibility. A property can carry any combination, and the live file contains FMA-only severe designations - so counting the NFIP flag alone understates severity. All four are counted separately.

Why do I get two rows for one address? Because the address is inside a municipality that is an NFIP community AND inside a county that is a separate NFIP community. Both have their own CRS class and their own repetitive-loss rollup. Set includeCountyCommunity: false to receive only the governing one.

Why is res_penetration_rate_sfha null for my county? Because FEMA published it as null, usually alongside a totalResStructuresSfha of zero. It is not recomputed from the underlying counts, because dividing by FEMA's zero would invent a rate. The counts are on the row so you can see exactly what FEMA published.

Does a CRS class of 10 mean no discount? Usually, but not always in FEMA's own file. Colorado carries class-10 communities published with a 10% and a 15% SFHA discount. The discount is read from the file, never derived from the class.

What happens if FEMA takes one of these datasets offline? The run fails with a message naming the dataset, and nothing is billed. FEMA publishes machine-readable deprecation dates on its catalog endpoint and this actor reads them, so an announced end-of-life shows up in openfema_deprecation_notice on every row before anything breaks.

Related actors

• nfip-flood-loss-risk-screener - paid NFIP claim history by location.
• fema-nri-county-risk-profile - FEMA National Risk Index perils by county and tract.
• epa-contaminated-site-screener - Phase I ESA contamination screen for the same site.
• site-due-diligence-bundle - 20-layer site screen in one call.

Pricing

Pay per result. One row = one billable result. A run that fails the drift gate, or that cannot resolve any site, emits nothing and bills nothing.