Two NWS Forecast Offices (Phoenix and Tucson) Filed 74% of All US Heat Deaths 2020-2025, and the Same Heat Warning Standard Produces 100x Different Death-Per-Event Ratios Across WFOs

USE CASE. NWS Heat program coordinators at NWS Headquarters and the NWS Eastern/Central/Southern/Western Regions, CDC NCEH environmental epidemiology teams, OSHA outdoor worker heat hazard alert allocation, and state climatologists tracking heat mortality all rely on per-WFO heat-event filing data. The conventional CONUS-aggregate heat death narrative (about 700-1,500 heat deaths per year nationally per CDC NCHS) doesn't reveal which forecast offices are doing the filing or how the filing relates to the warning issuance. RESULT 1 (the concentration). Top 5 NWS WFOs by 2020-2025 Excessive Heat / Heat direct deaths filed in the Storm Events Database: PSR (Phoenix) 1,135 deaths, TWC (Tucson) 357, VEF (Las Vegas) 173, PQR (Portland OR) 155, EWX (Austin/San Antonio) 53. PSR + TWC = 1,492 deaths = 73.8% of the national 2020-2025 total of 2,022 direct heat deaths. Adding VEF brings the desert Mojave/Sonoran cluster to 1,665 = 82.3% of national. The Pacific Northwest's PQR shows 155 deaths driven entirely by the late June 2021 heat dome event. After the desert SW + 2021 PNW event, no other WFO is in triple digits. RESULT 2 (the calibration heterogeneity). Per-event death ratios vary by more than 100x across WFOs filing similar heat-event volumes. WFOs that file high event counts with near-zero deaths: SHV (Shreveport LA) 2,686 events / 7 deaths (0.003 per event), MEG (Memphis) 3,832 events / 3 deaths (0.0008 per event), PAH (Paducah KY) 1,555 / 2 (0.001), TSA (Tulsa OK) 643 / 3 (0.005), CRP (Corpus Christi) 897 / 17 (0.019), HGX (Houston) 194 / 11 (0.057). WFOs filing fewer events with high death rates: PSR (Phoenix) 1,527 / 1,135 (0.743 deaths per event), TWC (Tucson) 477 / 357 (0.749), PQR (Portland OR) 74 / 155 (2.094 — the 2021 heat dome inflates this). The ratio between PSR's 0.743 deaths/event and MEG's 0.0008 is 929x — meaning the same NWS Excessive Heat Warning is being filed under operationally completely different verification practices in the two regions. STRUCTURAL READING. NWS Directive 10-515 (the Heat Warning Service directive) defines criteria for issuing Excessive Heat Warnings (typically 105°F+ heat index for at least 2 hours, or 110°F+ in the desert SW) but leaves the post-event Storm Events filing to the local WFO. The data show three clean operational categories: (a) desert SW WFOs (PSR, TWC, VEF, EWX) that file relatively few events and verify each one against confirmed deaths from the local medical examiner — the PSR 1,135 number is consistent with Maricopa County Department of Public Health's published 2020-2024 heat death counts of ~400-645/year with a 2023 peak around 645; (b) the Pacific Northwest one-time 2021 heat dome cluster (PQR 155), which represents an extraordinary single-event mass casualty rather than ongoing climatology; (c) Mid-South / Southern Plains / Midwest WFOs (MEG, SHV, PAH, ALY, BRO) that file thousands of heat events per year as a routine operational record but rarely log fatalities. These regions actually have substantial heat mortality (CDC tracks Mississippi, Tennessee, Louisiana as having high heat death rates per capita), but the deaths are reported by county medical examiners to CDC NCHS rather than by NWS WFOs into Storm Events. The 74% concentration of Storm Events heat deaths in PSR + TWC is therefore not a 'desert SW has all the heat deaths' finding — it's a 'desert SW has the only WFOs that systematically file heat deaths into the federal Storm Events catalog' finding. NWS heat program harmonization should standardize WFO-level filing practice so the federal record reflects actual heat mortality rather than the desert SW WFO operational habit. CAVEATS. (1) NWS Storm Events fatality counts are not the canonical US heat mortality count; CDC NCHS death certificates are. The CDC numbers are typically 2-3x higher than the Storm Events numbers. (2) The 'direct' vs 'indirect' death distinction depends on the local WFO's coding practice. (3) The PSR 1,135 figure tracks Maricopa County's own count well; the Memphis MEG 3 figure is dramatically below the Shelby County Health Department's reported heat deaths for the same period.

When a heat wave kills people, the National Weather Service local forecast office files each death into the federal Storm Events database. That database is the headline source for 'how many people did heat kill in the US' coverage. I downloaded the database for 2020-2025 and added up direct heat deaths by NWS forecast office. Two offices dominate. Phoenix (PSR) filed 1,135 direct heat deaths over the period, and Tucson (TWC) filed 357. Together those two desert Southwest offices account for 1,492 of the 2,022 heat deaths in the database — about 74% of all US heat deaths according to NOAA's federal record. Add Las Vegas (VEF, 173 deaths) and the desert Southwest cluster reaches 82% of the national total. Outside the desert Southwest, the only forecast office in triple digits is Portland Oregon (PQR, 155 deaths) — and ALL of those came from the late June 2021 'heat dome' that broke records across the Pacific Northwest. After Phoenix, Tucson, Las Vegas, and the one-time Portland event, no other US forecast office has more than 53 heat deaths in five years. Does that mean only the desert Southwest has heat deaths? No — and that's the structural finding. CDC and state health departments record substantial heat mortality across the Southeast (Mississippi, Tennessee, Louisiana, Texas Gulf Coast). When you compare event counts vs deaths by WFO, the picture is wild. Memphis (MEG) filed 3,832 heat events into the federal database with exactly 3 confirmed deaths. Shreveport (SHV) filed 2,686 events with 7 deaths. Albany NY (ALY) filed 715 events with 0 deaths. Brownsville TX (BRO) filed 512 events with 0 deaths. By contrast, Phoenix filed 1,527 events and got 1,135 deaths — a death-per-event ratio of 0.74. Memphis's ratio is 0.0008. The same NWS Excessive Heat Warning directive is being applied with completely different verification practices, with a death-per-event ratio differing by more than 900x between the two regions. The takeaway: the Storm Events 'heat deaths' rankings reflect WFO filing habits, not the actual geographic distribution of heat mortality. The desert Southwest WFOs file each death into the federal record because their county medical examiners are aggressive about coding heat-related fatalities and the WFOs cross-reference. The Mid-South and Southeast WFOs file heat warnings into the database but rarely loop back to file the deaths that those events caused, because in those states heat-related deaths are reported through the CDC NCHS death certificate stream rather than through the WFO. Why this matters: federal heat program coordinators trying to allocate resources or harmonize WFO practice should know the federal Storm Events 'heat deaths' field is operationally inconsistent across WFOs. The right way to measure US heat mortality is the CDC NCHS death certificate count, not the Storm Events filing — and the two should be reconciled before policy decisions get made on the basis of either.

PSR Phoenix 1,135 deaths (1,527 events) · TWC Tucson 357 (477) · VEF Las Vegas 173 (619) · PQR Portland OR 155 (74) · EWX Austin/SA 53 (144) · FWD Dallas/FW 30 (1,806) · CRP Corpus Christi 17 (897) · MTR San Francisco Bay 16 (70) · HGX Houston 11 (194) · PDT Pendleton OR 8 (96)

Highest: PQR Portland 2.09 (2021 heat dome) · PSR Phoenix 0.74 · TWC Tucson 0.75 · EWX Austin 0.37 · VEF Las Vegas 0.28 · MTR Bay Area 0.23 · Lowest: ALY Albany NY 0.00 (715 events / 0 deaths) · MEG Memphis 0.0008 (3,832 / 3) · SHV Shreveport 0.003 (2,686 / 7) · PAH Paducah KY 0.001 (1,555 / 2) · BRO Brownsville 0.00 (512 / 0) · ratio between PSR and MEG: ~900x

26,114 Excessive Heat + Heat events filed by 84 NWS WFOs · 2,022 direct deaths · 1,005 indirect deaths · system death/event ratio 0.08 · PSR + TWC concentration 73.8% · PSR + TWC + VEF 82.3% · PNW PQR contributes 7.7% (essentially all from 2021 heat dome) · Mid-South + Southeast WFOs contribute <1% of deaths despite ~30% of events