U3b · September 2026 · Users and data stewards

Sources, coverage and data lifecycle

Understand where evidence comes from, what it means and how it changes.

U4 source catalog

The source families below come from the released definitions fixture, not a live query to the publishers. Publication dates and observation periods differ. Lookout serves an accepted, pinned runtime; it does not silently replace it whenever an upstream site changes.

U.S. Census Bureau

acs_tables. U.S. Census Bureau, 2020-2024 American Community Survey 5-Year Estimates, Detailed Tables B01001, B01003, B08201, B19013, B23025, B25003, B25004, B25070, B25091; table-based Summary File, released January 29, 2026. Source: U.S. Census Bureau, 2020-2024 American Community Survey 5-Year Estimates. This product uses Census Bureau data but is not endorsed or certified by the Census Bureau.

  • Values are 2020–2024 five-year survey estimates with 90% margins of error, not single-year counts.
  • Shares and their margins are calculated by Lookout from the published Census estimates; the margins are approximate.
  • A value Census could not compute is shown as not available, never as a number.
  • A low-reliability value has a margin of error of 30% or more of the value; it stays in the view and is marked "wide range".
  • Thurston County values may be cited under the source terms. Values for neighbouring counties are for exploring only: they cannot be cited or exported.
  • These values are not in the cited annex yet.

Centers for Disease Control and Prevention (CDC)

cdc_places. Centers for Disease Control and Prevention. PLACES: Local Data for Better Health, County Data and Census Tract Data (GIS Friendly Format), 2025 release. https://data.cdc.gov (datasets swc5-untb, i46a-9kgh, yjkw-uj5s), accessed 2026-09-24. Source: CDC PLACES, 2025 release (model-based estimates). Use of this data does not imply endorsement by CDC, ATSDR, HHS or the United States Government. This material is otherwise available on the CDC website at no charge: https://www.cdc.gov/places/.

  • PLACES values are model-based small-area estimates for adults, not direct survey measurements or counts.
  • Tract values are crude prevalence with CDC's 95% confidence interval. County age-adjusted values are for comparing counties and are not comparable with tract values.
  • Some measures describe a narrower group, such as adults aged 65 and over or women aged 50 to 74; each measure names its group.
  • Most measures use 2023 survey data; teeth lost, dental visits, mammography, colorectal screening and short sleep use 2022.
  • Thurston County values may be cited under the source terms. Values for neighbouring counties are for exploring only: they cannot be cited or exported.
  • These values are not in the cited annex yet.

Federal Emergency Management Agency (FEMA)

fema_nri. Federal Emergency Management Agency (FEMA), National Risk Index Dataset: National Risk Index Census Tracts and National Risk Index Counties - v1.20 (December 2025). Retrieved from https://services.arcgis.com/XG15cJAlne2vxtgt/arcgis/rest/services/National_Risk_Index_Census_Tracts/FeatureServer/0 and https://services.arcgis.com/XG15cJAlne2vxtgt/arcgis/rest/services/National_Risk_Index_Counties/FeatureServer/0 on 2026-09-24T16:53:07Z. This product uses the Federal Emergency Management Agency's National Risk Index dataset API or downloadable datasets but is not endorsed by FEMA. The Federal Government or FEMA cannot vouch for the data or analyses derived from these data after the data have been retrieved from the Agency's website(s).

  • NRI scores are national percentile ranks among all U.S. tracts (or counties), and ratings are FEMA's relative classes. They are not predictions of loss or damage in any year.
  • Expected annual loss values are modelled average yearly dollar losses (December 2024 dollars) for broad nationwide comparison, not a substitute for a local risk assessment.
  • A tract FEMA rates "Not Applicable", "No Rating" or "No Expected Annual Losses" has no score here and is shown with FEMA's own words, never as 0.
  • Social vulnerability in NRI v1.20 comes from the Census Bureau Community Resilience Estimates, not from the CDC Social Vulnerability Index.
  • Community resilience is FEMA's county score, ranked among U.S. counties, so it is shown as county context and never as a tract map.
  • Tracts are 2020 Census tracts from FEMA's 2021 boundary files, drawn on the 2024 tract boundaries.
  • Thurston County values may be cited under the source terms. Values for neighbouring counties are for exploring only: they cannot be cited or exported.
  • These values are not in the cited annex yet.

Geography and coverage

U4 maps 314 tracts across these five counties. The values use 2020 tract definitions displayed on 2024 tract boundaries; FEMA’s source geography uses its 2021 boundary files. Two ACS-only water tracts are listed as not mapped. Do not join by row position across different geography releases.

National county context has a different coverage surface. The overlap UI is Washington-only in U4; that restriction does not expand M1 coverage beyond its five counties.

Geography and coverage
CountyFIPSMapped tractsU4 use
Grays Harbor5302720Explore only
Lewis5304121Explore only
Mason5304519Explore only
Pierce53053193Explore only
Thurston5306761Explore and cite under source terms; new annex integration pending

Units, estimates and comparisons

  • PLACES reports model-based adult prevalence with 95% intervals. Read each measure’s universe; some apply only to particular ages or groups. Crude prevalence is not interchangeable with age-adjusted county prevalence.
  • ACS values are five-year survey estimates. Derived shares and margins are calculated by Lookout from Census tables. They are not a single-year census count.
  • FEMA NRI scores are national relative ranks, and rating classes are the publisher’s classes. Expected annual loss uses modeled annual dollars; neither is a forecast of an event at a particular property.
  • County values come from publisher county rows, not the sum of tract values. Community resilience is county-only, ranked among U.S. counties.
  • Thurston values marked registry-authoritative stay on the existing admitted record. In the M1 county matrix, reason RA points to that authoritative path rather than supplying a duplicate value.

Missing values and reliability

Always read the reason code with the value. A null can mean publisher missingness, source withdrawal, an unsupported universe or a separately authoritative county record. It is not automatically zero. The dictionary lists every published reason, rating, reliability and annotation code.

annotation codes

annotation codes
CodeMeaningValue / method behavior
CControlled estimate: Census publishes no sampling margin of errorSee publisher annotation
MThe margin of error could not be computed: too few sample observationsSee publisher annotation
XThe estimate could not be computed: too few sample observationsSee publisher annotation

rating codes

rating codes
CodeMeaningValue / method behavior
ABNot in FEMA's data for this tractSee measure and method context
IDInsufficient DataSee measure and method context
NANot ApplicableSee measure and method context
NENo Expected Annual LossesSee measure and method context
NRNo RatingSee measure and method context
RHRelatively HighSee measure and method context
RLRelatively LowSee measure and method context
RMRelatively ModerateSee measure and method context
VHVery HighSee measure and method context
VLVery LowSee measure and method context

reason codes

reason codes
CodeMeaningValue / method behavior
ABNot in the publisher's data for this tractmissing
CControlled estimate: Census publishes no sampling margin of errorvalue_counts
IDInsufficient Data (FEMA): FEMA could not rate this tractmissing
MThe margin of error could not be computed: too few sample observations (Census)value_counts
NANot Applicable (FEMA): the hazard is not geographically possible hereassessed_not_met
NENo Expected Annual Losses (FEMA)assessed_not_met
NPNot published by CDC for this tractmissing
NRNo Rating (FEMA): no expected annual loss from this hazardassessed_not_met
RAThurston's cited county value is served by its existing recordmissing
XThe estimate could not be computed: too few sample observations (Census)missing
ZDNo households or people in the denominator, so the share is not definedmissing
source_unavailableNot available to you for this countymissing

reliability codes

reliability codes
CodeMeaningValue / method behavior
HHigh reliabilitySee measure and method context
LLow reliability: wide rangeSee measure and method context
MModerate reliabilitySee measure and method context
NNo published rangeSee measure and method context
PPublished 95% intervalSee measure and method context

Facilities and record context

Seven layers are available in the U4 facility catalog: schools, hospitals, bridges, HIFLD hospitals, fire/EMS stations, ambulance services and police stations. The four new D8 layers carry incomplete-inventory disclosures and record-specific access/status notes.

CMS hospital listings are linked to HIFLD facilities and have no independently verified map point in this runtime. Use linked facility identifiers rather than counting a listing and its mapped campus as two facilities. A linked listing can refer to more than one campus.

An owner may see additional bridge fields; member projections omit them. Facility location does not prove operational status, capacity, public access, road access or emergency readiness. Consult the source record and local authority for operational decisions.

NOAA storm records, HRSA shortage designations and Washington DOH detail records retain their own periods, limitations and access rules. A shortage designation is not a count of available appointments; a storm point is not a complete hazard footprint.

Refresh and versioning

  1. Identify a new publisher release and compare schema, geography, field meaning, availability and source terms with the accepted release.
  2. Run the reviewed data ingestion and acceptance workflow in the data repository. Retain attribution, source periods, permission decisions and field allow-lists; compare missingness and geography joins.
  3. Build a new immutable data package with its manifest, hashes and acceptance receipts. Do not edit the deployed package to make a new release look like the old one.
  4. Run provider/contract and UI checks, review visible changes, and deploy through a new approved release. Record changed source vintages and any comparison implications in the changelog.
  5. Retain prior pins needed for artifact traceability and rollback. Do not claim continuous, nightly or real-time refresh: no such schedule is established for the pinned U4 runtime.

Withdrawal, retention and reuse

A source withdrawal or measure revocation can make previously visible data unavailable. The tract provider re-reads configuration through the directory mount; reviewed tools record a receipt chain before replacing configuration. Linked tenure/cost-burden measures are withheld together because their values can reconstruct one another.

Revoking M1 does not revoke related U3a county evidence. The operator must assess those rights separately. Do not restore old configuration merely to make a chart render.

Explore-and-cite is subject to source terms and does not itself enable annex export. Neighbor values are explore-only. Do not use a share link or browser capture to circumvent a source restriction.

The application’s artifact holds and retention controls do not establish an organization’s retention policy by themselves. Follow the approved retention decision for exports, raw data, backups and support records; do not invent a duration from this documentation.

Clinic practice locations

The clinic release uses the HRSA site download (https://data.hrsa.gov/topics/health-centers/), the CMS NPPES Version 2 dissemination files (https://download.cms.gov/nppes/NPI_Files.html), current CMS organization responses (https://npiregistry.cms.hhs.gov/api-page), and the Census Geocoder (https://geocoding.geo.census.gov/geocoder/Geocoding_Services_API.html). The September 2026 baseline is updated through the September 14–20 weekly file. Only active Type 2 organizations reporting a 261Q Clinic/Center taxonomy are selected; primary and secondary Washington practice locations are considered. Mailing addresses and individual practitioners are excluded.

Geocoding uses the publicly reported street, city, state and ZIP. Admission requires matching normalized street and ZIP, a Census county assignment, and containment in the retained county boundary. Nonmatches, conflicting results and ambiguous matches are excluded without inferring county from city or ZIP. Census coordinates interpolate address ranges. HRSA permanent-site coordinates must fall inside their reported county; mobile or suppressed-address sites remain list-only.

Specialty labels are retained from individual public CMS provider responses, not a bundled NUCC codebook. They are provider-reported classifications, not independently verified services. Distinct provider/practice identifiers remain separate even when names and addresses coincide. Published CMS aliases are searchable. Nine reviewed provider-directory locations supplement missing sites or labels; their specialties remain separate from CMS taxonomy descriptions. Counts describe registry practice records, not buildings.

The accuracy release contains 1,221 source records, 1,219 mapped points and two list-only sites: Grays Harbor 62, Lewis 86, Mason 51, Pierce 721 and Thurston 301. The increase mainly reflects identities previously combined by name/address, plus directory additions; it does not mean newly opened clinics. DaVita Olympia, Lacey and Tumwater and four Indigo locations are searchable. Six explicitly described urgent-care directory sites supplement the registry classification. These totals do not establish complete coverage.

Enable Clinics and specialty clinics (OpenStreetMap) alongside the CMS/HRSA layer to compare them. Registry markers are pink; OSM markers are blue. Each source has independent name/address and specialty filters. The comparison table counts records and mapped points in the current county scope and filters; the Cross-source matching panel suggests links and shows name/address/distance evidence, ambiguity and missing or differing specialties. Suggested links do not merge records or confirm identity. OSM counts refer to mapped objects, so a node and an area may describe the same clinic.

The earlier regional OSM snapshot contained 627 objects across five counties: Grays Harbor 38, Lewis 38, Mason 20, Pierce 452 and Thurston 79. That snapshot had a July 24, 2026 source timestamp and was downloaded September 27, 2026. The separate Lewis and Pierce workspaces were accepted September 28 with 39 and 459 OSM objects respectively from September 28 snapshots; do not use the older regional totals as current counts for those workspaces. Opening this layer never waits for an Overpass query. The browser reuses county records for toggles and filters during the current authenticated explorer session. Reloading the page fetches the local snapshot again. A weekly validated refresh attempts to update the snapshot. Incomplete, stale or regressing acquisitions retain the previous accepted layer and its source date, with an operational warning.

OSM includes community-tagged clinics, doctors, dentists, dialysis and selected outpatient specialties. A name containing Urgent Care does not imply that OSM supplies an urgent-care specialty tag. Use name search and Specialty not supplied to investigate missing tags. Details preserve OSM object links, brand/operator aliases and source tags. Locations, opening hours, services and completeness are not independently verified.

OSM-derived records are separately downloadable from the layer controls for the current county scope and filters as GeoJSON Lines, with attribution to OpenStreetMap contributors and ODbL 1.0 licensing. Source and license information: https://www.openstreetmap.org/copyright. Neither clinic layer contributes to annex exports. The broader OSM Places overlay is separate and still uses its existing viewport query cache.

Implementation references for this manual

Repository paths reviewed for this U3b documentation. Optional code is not a claim of live availability.

  • tests/lookout/product/fixtures/u4/definitions.json
  • docs/implementation-plans/execution/evidence/U4/api-contract.md
  • server/evidence/tractMeasuresProvider.mjs
  • scripts/u4/tract-measures-revocation.mjs
  • tests/lookout/product/fixtures/u4/matrix-5x5-worst.json