How the USPR Score and reports are built
The USPR Score ranks each ZIP area's expected yearly building losses from natural hazards against every other US ZIP area. Everything else on a report is a separate, sourced fact, kept out of the score on purpose.
The USPR Score
The USPR Score is a number from 0 to 100. A ZIP area that scores 70 has higher modelled natural-hazard losses, per dollar of building value, than 70% of the 33,559 US ZIP areas we score. It is a ranking, not a probability: 70 does not mean a 70% chance of anything.
The loss measure underneath is FEMA's expected annual building loss: the average yearly damage to buildings that FEMA's National Risk Index models for 17 hazards, from inland and coastal flooding, hurricanes and tornadoes to earthquakes, wildfire and winter weather. We divide it by the value of the buildings exposed, so every report also gives the score as a plain figure, such as "$132 a year per $100,000 of building value", the median for US ZIP areas.
We chose expected losses because they let different hazards be added in dollars without anyone choosing weights. Earlier drafts of this project weighted flood, toxic sites, wildfire, sea level and noise by hand; that would have rated a San Francisco ZIP low despite its earthquake exposure, and the weights could not be defended. FEMA's model already prices each hazard in the same unit.
How it is calculated
- Start from census tracts. The National Risk Index (December 2025 version) publishes expected annual building loss and building value for 85,154 census tracts.
- Allocate tracts to ZIP areas. ZIP code tabulation areas (ZCTAs, the Census Bureau's ZIP areas) do not follow tract lines. We split each tract between the ZIP areas it overlaps by its share of 2020 Census housing units, using the Census block-to-ZCTA relationship file, and sum the pieces. Tracts with no housing are split by land area. Connecticut's tracts are matched to their 2022 planning-region codes.
- Compute the loss rate. Expected annual building loss divided by building value, for all hazards together and for each hazard.
- Rank nationally. The score is the percentile rank of the loss rate among the 33,559 ZIP areas with at least one housing unit and a building value. Each hazard row on a report is ranked the same way among ZIP areas where that hazard has a modelled loss.
The largest median hazard loss rates across US ZIP areas are inland flooding ($84 per $100,000), tornado ($5 per $100,000), earthquake ($3 per $100,000), tsunami ($1 per $100,000).
Bands and percentiles
Scores fall into five equal bands, always relative to other US ZIP areas:
| Band | Score | Loss rate per $100,000 a year |
|---|---|---|
| Very low | 0–19 | under $90 |
| Low | 20–39 | $90–$117 |
| Moderate | 40–59 | $117–$149 |
| High | 60–79 | $149–$211 |
| Very high | 80–100 | over $211 |
Differences of a few points between neighbouring ZIP areas are within the precision of the underlying model and should not be read as meaningful. A report is marked lower confidence when the ZIP area has fewer than 50 homes or buildings, or when less than 80% of its housing falls in tracts with complete data. ZIP areas with almost no housing, such as a military base or a post-office-box ZIP, get no score or page.
What the score leaves out
The National Risk Index also publishes social vulnerability and community resilience measures and combines them with losses in its own risk score. We use neither. They describe the people living in an area, not the hazard to a building, and a property score that moves with an area's demographics would be both misleading and inappropriate.
Environmental sites, sea level rise and aviation noise are reported beside the score, not inside it. None of them has an expected-loss estimate that can be added to hazard losses honestly. Agricultural losses, injuries and deaths, which the Index also models, are excluded because the score is about buildings.
Flood zone counts
We place every building centroid in FEMA/ORNL's USA Structures inventory (June 2025 release for Florida) on FEMA's National Flood Hazard Layer, using a national snapshot of the effective maps compiled in July 2026, and count buildings by zone. Where flood polygons overlap, a building takes the higher-hazard zone. Land shares are measured on the same polygons in an equal-area projection.
As a check, our count for Miami-Dade gives 38.4% of residential buildings in the Special Flood Hazard Area; FEMA's own county table gives 40.5% of residential structures (199,250 of 492,181). The difference comes from the two building inventories. Counties with only paper maps, or none, are reported as not mapped rather than as zero.
Insurance and claims
Policy figures come from OpenFEMA's redacted NFIP policies dataset (version 3). We count policies in force on August 31, 2026 by the ZIP code FEMA records for the property, sum insured units, and take the median annual policy cost for single-family houses (occupancy code 11) and for single units in multi-unit buildings (code 16). The cost is FEMA's policyCost: premium plus reserve fund assessment, federal policy fee and HFIAA surcharge. Medians by zone are shown only with at least 10 policies.
Claims come from OpenFEMA's redacted NFIP claims dataset (version 3), 1978 to September 2026, again by reported ZIP code. Paid amounts are net building, contents and increased cost of compliance payments, in dollars of the year paid, not adjusted for inflation.
Environmental sites
Distances run from the ZIP area's boundary to each site's point: 0 means inside the ZIP. National Priorities List sites come from EPA's NPL layer with status (final, proposed, deleted), within 5 miles. Toxics Release Inventory facilities are the 2024 reporters within 3 miles, with on-site releases summed across chemicals in pounds. Brownfields, other Superfund-program sites and large-quantity hazardous waste generators come from EPA's Facility Registry Service, excluding points geocoded only to a ZIP centre. A listing records an EPA program interest; it is not a finding of contamination at a parcel.
Wildfire, sea level and noise
Wildfire: the Forest Service's Wildfire Hazard Potential 2023 ZIP summary gives the share of land in each hazard class. It describes land, so ZIPs with large natural areas can show high shares with few homes exposed; the hazard table's wildfire row, from FEMA, measures losses to buildings.
Sea level rise: we test each building centroid against NOAA's Sea Level Rise Viewer extent rasters for 1, 2 and 3 feet above mean higher high water. The extents show ocean-connected inundation only, not depth. Projections are the 2022 Interagency Sea Level Rise Technical Report scenarios at the nearest tide gauge, relative to 2000.
Aviation noise: we sample the 2020 aviation layer of the US Department of Transportation's National Transportation Noise Map at each building centroid, from its published map tiles (about 34 m cells over a 30 m model grid). Values are modelled 24-hour equivalent sound levels; the publisher cautions against using them for single locations.
Summaries and checklists
The three summary bullets and the due diligence checklist on each report are generated from fixed sentence rules filled with the ZIP area's own figures. A checklist item appears only when the data triggers it: the Elevation Certificate item when buildings stand in the Special Flood Hazard Area, the Phase I item when an active NPL site lies within a mile or a brownfield or Superfund-program site lies inside the ZIP, and so on. No report text is written freehand or by an AI model.
Limits
- Reports describe ZIP areas. A single parcel can sit in a different flood zone, at a different elevation or further from a site than the area figures suggest.
- The National Risk Index is a model with its own periods of record and assumptions; it does not include levees, dams or mitigation except as they appear in its source data, and it does not project future climate.
- Flood maps change. The reports use maps in effect when the data were compiled; Miami-Dade's new maps take effect on February 19, 2027.
- ZIP codes and ZIP code tabulation areas differ: post-office-box and single-organisation ZIP codes have no area of their own, and NFIP records filed under them are not shown separately.
- Nothing on this site is an official flood zone determination, an appraisal, an insurance quote, legal advice or an environmental site assessment.
Updates and corrections
We rebuild the dataset when a source publishes a new release: the National Risk Index (about yearly, each December), the NFIP policy and claims files (monthly; we refresh at least quarterly), EPA's NPL layer and TRI (as released) and FEMA's flood maps (as counties' new maps take effect). Each report shows the date its data were compiled. To report an error, write to the address on the contact page; we check every report against the source and correct confirmed errors in the next rebuild.
Area reports currently cover 78 ZIP codes in Miami-Dade County, Florida; further counties are added as their data are checked.