About the Oklahoma Climate Resilience Atlas

Why this exists

Oklahoma's metros plan for tornadoes, floods, and heat with data scattered across a dozen federal portals. This atlas puts the region's hazard, vulnerability, and demographic record on one map, so a planner, an emergency manager, or a resident can answer a concrete question in minutes: where does water go, where have tornadoes actually tracked, who gets hottest, what have storms cost, and who has the least capacity to cope.

Click any census tract and the atlas returns one combined readout: social vulnerability, the demographic flags that matter in an emergency, and the tract's hazard exposure. That cross-layer view, not any single dataset, is the point.

What each layer is for

Flooding

FEMA flood zones. Check any parcel or corridor against the regulatory floodplain before you zone it, buy it, or build on it.

Projected inland flood exposure. Compare how often each tract will see two-inch rain days by mid-century, and size culverts and detention where the increase is steepest.

Extreme heat

Observed urban heat. Find the Oklahoma City blocks that ran hottest on an August afternoon, and target tree canopy and cooling centers block by block.

Projected extreme heat days. Overlay projected days at or above 95 °F on the SVI to find neighborhoods that are both hot and vulnerable, and put cooling investment there first.

Impervious surface. Spot the tracts where pavement drives both heat and runoff, and write redevelopment standards that pay down both risks at once.

Severe weather

Tornado tracks. See every rated tornado path since 1950 against the neighborhoods that exist today, and site shelters and sirens against the record rather than the reputation.

Hail and damaging wind. Hail is the largest driver of property insurance claims in this market and leaves no trace on any other layer; read the report density before pricing a roof or a deductible.

Earthquakes. Oklahoma’s shaking is induced, it peaked in the mid-2010s, and it has fallen sharply since disposal was restricted — the decline matters to an underwriter as much as the peak.

Historical storm events. Filter 77 years of tornado, hail, wind, and flood reports by type and decade to brief responders on what actually hits each corridor.

Storm damage intensity. Rank jurisdictions by inflation-adjusted storm losses to argue mitigation budgets with evidence instead of anecdotes.

People and vulnerability

Social Vulnerability Index. Layer any hazard over the SVI to see who has the least capacity to prepare and recover, then aim outreach there first.

Demographic exposure. Switch the lens to seniors, young children, poverty, or car-free households, and plan evacuation and shelter capacity around the people who actually live there.

Loss and insurance

Expected annual loss. Put a dollar figure on what hazards are expected to cost a tract each year, and compare that against what mitigation would cost.

Flood insurance claims paid. See where the NFIP has actually written cheques, which is not the same map as the regulatory floodplain — claims regularly land outside the mapped high-risk zone.

Property and market

House price change. See which tracts have moved most over five years before setting an adjustment or arguing a trend.

Home values, rents and price-to-rent. Screen ZIPs on the ratio an investor actually underwrites, rather than on price alone.

Sales activity and construction. Read days on market, months of supply and permitted units together to tell a tightening market from a loosening one.

Job density. Separate places people work from places people sleep, which is what drives commute patterns, retail siting and multifamily demand.

Data sources and credits

Every layer comes from a federal public-domain source or an attribution-only open source, snapshotted into this atlas's own files. Nothing here hot-links a government server at view time.

The market layers (Zillow, Redfin) are free to use with attribution but are commercial market indicators, not public records. They support market analysis; they are not appraisals and not comparable sales.

Dataset	Publisher	Used for	License / credit
National Flood Hazard Layer (snapshot)	FEMA	Flood zone layer, tract SFHA shares	U.S. Government work. Credit: FEMA.
Severe weather GIS: tornado tracks, 1950–2025	NOAA Storm Prediction Center	Tornado track layer	U.S. Government work. Credit: NOAA SPC.
CMRA Screening Data (tract summaries)	Interagency (resilience.climate.gov / USGCRP)	Projected heat days, 2-inch rain days	U.S. Government work. Credit: CMRA.
Urban Heat Watch, Oklahoma City, August 12 2023	NOAA NIHHIS + CAPA Strategies (2023 cohort)	Observed heat traverse points	Public campaign data. Credit: NIHHIS / CAPA Strategies. Oklahoma City is the metro with a campaign; the traverse covers the OKC core south into Cleveland County, and Tulsa has no campaign to date, so the layer is blank there.
National Risk Index (census tracts)	FEMA	Expected annual loss in dollars, by hazard	U.S. Government work. Credit: FEMA. Modelled screening estimates, not a property-level appraisal.
NFIP redacted claims	FEMA (OpenFEMA)	Flood insurance claims paid, by tract	U.S. Government work. Credit: FEMA. Claim locations are redacted to the tract; coordinates in the source are rounded too coarsely to map.
Severe weather GIS: hail and wind reports, 1955–2025	NOAA Storm Prediction Center	Hail and damaging wind layers	U.S. Government work. Credit: NOAA SPC. Observer reports, so density reflects reporting as well as occurrence.
Earthquake catalogue	USGS	Earthquake epicentre layer	U.S. Government work. Credit: USGS.
Developmental house price index (census tracts)	FHFA	Five-year price change layer	U.S. Government work. Credit: FHFA. A repeat-sales index, not a valuation; thin-sales tracts are noisy.
ZHVI and ZORI, ZIP level	Zillow Research	Home value, rent and price-to-rent layer	Free for use with attribution. Credit: Zillow. A smoothed market index over the homes Zillow observes, not a complete sales record.
County market tracker	Redfin Data Center	Median sale price, days on market, supply	Free for use with attribution. Credit: Redfin. MLS-derived county aggregates — market analysis, not comparable sales.
Building Permits Survey	U.S. Census Bureau	Permitted units and construction value	Public domain. Credit: U.S. Census Bureau.
LEHD LODES 2023	U.S. Census Bureau	Job density and employment centres	Public domain. Credit: U.S. Census Bureau. Covers jobs under state unemployment insurance, so most self-employment is absent.
Storm Events Database, 1950–2026	NOAA NCEI	Event points, damage choropleth, loss model	U.S. Government work. Credit: NOAA NCEI.
Social Vulnerability Index 2022	CDC / ATSDR	SVI layer, readout, equity weighting	Public domain. Credit: CDC/ATSDR.
TIGER/Line cartographic boundaries 2024	U.S. Census Bureau	Tracts, county, place outlines	Public domain. Credit: U.S. Census Bureau.
ACS 5-year estimates 2018–2022	U.S. Census Bureau	Demographic layer, exposure model	Public domain. Credit: U.S. Census Bureau.
NLCD 2021 impervious surface	USGS / MRLC	Impervious layer	Public domain. Credit: USGS/MRLC.
GlobalMLBuildingFootprints (2026-07-24 release)	Microsoft (Bing Maps)	Building footprints layer, 2D and 3D views, ML height estimates	CDLA Permissive 2.0. Credit: Microsoft GlobalMLBuildingFootprints. Heights are model estimates from imagery, not survey data.
Protomaps basemap (build 2026-07-29)	Protomaps, from OpenStreetMap	Basemap ("Streets")	Open style; data ODbL. © OpenStreetMap contributors.
World Imagery tile service (view-time)	Esri	Optional "Satellite" basemap	Tiles © Esri — Source: Esri, Maxar, Earthstar Geographics, and the GIS User Community. Streamed at view time, not snapshotted; the only layer that needs the network.
CPI-U (CPIAUCSL)	U.S. BLS via FRED	Inflation adjustment of storm damages	Public data. Credit: BLS / FRED.

Excluded by license: First Street / Risk Factor (proprietary), Esri-hosted basemap data beyond the optional view-time satellite tiles, and Google Maps data.

Methodology & assumptions (mitigation analytics)

The analytics panel and the companion workbook (analytics/loss_model.xlsx) price mitigation with public, citable methods. Every number they show is illustrative, built for ranking places and hazards, not for appraising a specific project.

Annualized loss. NCEI Storm Events damages (property plus crops), each year converted to current dollars with annual-average CPI-U (FRED CPIAUCSL), divided by the 77-year record.
Discount rate: 7%. The OMB Circular A-94 (1992) rate, which FEMA reinstated for Hazard Mitigation Assistance benefit-cost analysis effective April 8, 2025. The workbook exposes it as an input.
Project life: 30 years, a common FEMA BCA default for building projects. Also an input.
Benefit-cost ratios. NIBS, Natural Hazard Mitigation Saves (2019 report, Table 1, federal-grants row): riverine flood 7:1, wind 5:1, earthquake 3:1, wildland-urban fire 3:1, 6:1 across all hazards. The 2019 study remains the institute's latest full report as of this build (July 2026). No headline ratio is hardcoded from memory; the workbook reads these values from a labeled table with the citation attached.
Exposure. Property exposure multiplies each tract's housing units by its median home value by its share inside the special flood hazard area. Vulnerable population counts residents of tracts at or above the 75th SVI percentile that also touch the SFHA or fall in the top quartile of projected heat days.

Storm Events undercounts small losses and geocodes damage to the reporting jurisdiction, so treat jurisdiction totals as floors. Commission a project-level FEMA BCA before committing funds.

Geographic scope

The atlas covers four Oklahoma counties: Oklahoma, Cleveland, and Canadian in the Oklahoma City metro, plus Tulsa County. That takes in Oklahoma City, Norman, Moore, Edmond, Midwest City, Yukon, Mustang, El Reno, Tulsa, Broken Arrow, Jenks, and Owasso, along with the smaller incorporated cities and towns inside those counties. The footprint is deliberately non-contiguous — the OKC metro and Tulsa sit about 90 miles apart — so the map opens zoomed out far enough to show both, and the gap between them holds no data. Oklahoma City's limits reach slightly into Pottawatomie County; that sliver falls outside the county footprint and is clipped off. The pipeline clips to a single list of FIPS codes in config.py; edit that list and re-run to cover more of Oklahoma.