Modeling wildfire smoke air quality impact trendline

How wildfire smoke exposure has changed across the United States

This model estimates how many U.S. residents encountered wildfire-smoke PM₂.₅ high enough that the air would be unhealthy for sensitive groups from smoke alone—crossing the AQI 101 threshold. Canadian smoke is included whenever it reached U.S. counties.

Comparable model: 2006–2023 · preliminary smoke and fire updates through Jul 18, 2026

47%of Americans breathed unhealthy smoke in 2023155M residents had at least one qualifying day
1 → 40widespread smoke days, then vs now10M+ residents at once · 2006–15 vs 2016–23
23 → 1,646counties touched at least once per yearmore than half of 3,108 counties qualified in 2023
1 in 4Americans on the single worst dayJun 29, 2023 · 88.5M residents
9.3× higher recent burden.The last five comparable years averaged more than nine times the population-days of the first five.
01

The modeled smoke trend

Comparable nationwide estimates begin in 2006; current conditions remain a separate preliminary series

Averaged across the country, the typical American now lives through about three-quarters of a day of unhealthy wildfire smoke per year—roughly one day every 16 months, up from about one every 12 years in the late 2000s.

Annual exposure and five-year trend

Person-days of unhealthy smoke air · fixed 2020 population · 2006–2023

1.88Bperson-days since 2006
Annual estimate5-year average trend
491M1.5 days/person327M1.0 days/person164M0.5 days/person020062010201420182023
thin gray: individual years blue: trailing five-year average. Hover either series for exact values and annual context. The right scale expresses the same burden as days per fixed resident.

The worst individual smoke days

Largest modeled daily reach · comparable series, 2006–2023

RankDatePeople reachedShare of fixed population
1Jun 29, 202388.5M27.0% · more than 1 in 4
2Jun 7, 202372.6M22.2% · more than 1 in 5
3Jun 28, 202370.8M21.6% · more than 1 in 5
4Jun 8, 202356.0M17.1%
5Jun 30, 202354.2M16.6%
6Jun 6, 202347.3M14.4%
7Jun 27, 202345.4M13.9%
8Sep 14, 202045.0M13.7%
9Sep 12, 202044.7M13.7%
10Sep 13, 202041.8M12.8%
The worst day of 2006–2010 reached 11.0M people. 36 days since 2016 exceeded it; the longest continuous stretch lasted 8 days in September 2020.

U.S. wildfire-smoke evidence: years with available data

Thirty non-contiguous years · one month per square · grayscale represents evidence or exposure intensity

Documented reportsModeled estimatesRecent screen
195019881998200020012002200320042005200620072008200920102011201220132014201520162017201820192020202120222023202420252026
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
documented strength1 · 2 · 3+modeled smoke days0 · 1–3 · 4–7 · 8–14 · 15+recent screennot yet observed
The columns intentionally skip years without accepted documentary evidence or national monthly estimates. This removes empty decades while keeping documentary reports, modeled estimates and the recent screen visibly separated.
Historical source-review coverage, 1950–2005one square per year · months systematically reviewed
1950196019701980199020002005
months reviewed01–34–89–1112
Systematic month-by-month source screening begins in 1996. Earlier accepted episodes mark documented cases, not complete annual review. Unreviewed years remain unknown rather than zero.
Annual source-covered fire records, 1950–2026within-country count quintiles · exact counts on hover
195019701990200620202026
U.S.
Canada
lowest fifthhighest fifth · calculated separately by countryno national catalog
This is an archive record-density view, not one uniform fire census. Canada uses NFDB from 1950; U.S. coverage begins with MTBS large fires in 1984, broadens to FPA FOD in 1992–2020, then uses a dated InFORM snapshot. Hover any year for the exact count, largest available incidents and smoke evidence.

When and where the burden concentrated

Monthly share of person-days, then each year’s leading state shaded by burden

2006–2015
JFMAMJJASOND
2016–2023
JFMAMJJASOND
Top state
WA06CA07CA08OR09MA10FL11WA12NV13NV14OR15GA16WA17CA18OR19CA20CA21WA22NY23
Lower leading-state burdenHigher
June now carries the largest share of the recent-era burden; 85% of 2023’s person-days occurred in June. New York led in 2023, while five northeastern and midwestern states carried 52% of the national burden.

What the available evidence says about 2026

The published wildfire-attribution series above ends in 2023, so it is not used to rank 2026. This section compares every year from 2006–2026 under one separate screen: high PM₂.₅ at ground monitors on days NOAA mapped smoke overhead, always from January 1 through July 18.

2026 through Jul 18, 2026
Same calendar window across all 21 yearsJan 1–Jul 18 · monitor + satellite screen only
2ndburden rankof 21 years2ndpeak-reach rankof 21 years4days reaching 10M+3rd by annual count
Indicative population-days2026 · 214.8M
060708091011121314151617181920212223242526
Peak daily reach2026 · 58.9M
060708091011121314151617181920212223242526
other yearshighest year2026
2026 ranks 2nd for cumulative burden and 2nd for peak daily reach. 2023 remains higher on both measures. Values are comparable within this screen only, not with the Stanford modeled series.

2026 is already an outlier under the same-window screen. Its 214.8M indicative population-days exceed 2025 by 19%; its 58.9M-resident peak reached about 18% of the fixed contiguous population. 2023 remains the leader.

The Washington Post reported → D.C. reached AQI 175 between 4 and 5 a.m. on July 18—the 30th consecutive hour in unhealthy or very unhealthy territory. This episode is context, not an input to the annual model.

02

Historical burned-area trendlines

Complete annual Canadian burned area and mapped U.S. large-fire area, with current 2026 totals kept as separate markers

Canada and United States burned area

Annual acres · complete historical series plus Jul 18, 2026 YTD markers

Canada · NBACU.S. · MTBS large fires2026 YTD
40M20M01972198019902000201020202026
Canada: NBAC adjusted annual area, 1972–2025. United States: MTBS mapped large wildfires, 1984–2024. The 2026 markers use current operational totals through Jul 18 and are deliberately not connected to either complete-year line.
03

North American fire seasons now burn more land

Canada’s recent decade averaged 2.0× its 1970s baseline; U.S. mapped large fires averaged 3.4× their 1980s baseline

More burned land increases the amount of smoke available for transport. The exposure result still varies sharply with wind direction: a Canadian fire season can produce a large U.S. burden even when U.S. acreage is comparatively low.

Where reported burned area concentrates2.47M georeferenced records · U.S. + Canada · 1992–2020

45.1% of reported burned acres fall in the darkest 10% of occupied grid cells.

Reported burned acres per cell
≤3.5K3.5K–45.6K45.6K–461.7K461.7K–1.77M1.77M–3.48M>3.48M
Six distribution-aware bands preserve contrast from the lowest cells through the highest-decile cells; hover any cell for exact acres and incidents. The map combines U.S. FPA FOD and Canadian NFDB records over their shared 1992–2020 period. Counts are not compared across countries because reporting coverage differs.
2.0×Canada recent decade vs 1972–1981
3.4×U.S. recent decade vs 1984–1993
7.20MCanada · Jan 1–Jul 18, 2026
3.85MU.S. · Jan 1–Jul 18, 2026

2023 broke the exposure record without a large U.S. fire year. U.S. mapped burned area was 69% below its preceding-decade average, while Canada recorded 6.1× its average. Documented June 2023 episodes traced the smoke to Canadian fires; this is documented co-occurrence, not model-derived plume attribution.

A Washington Post graphic shared by Ben Noll → shows Canada’s largest 1972–2025 fire seasons clustering among warmer May–September years. That relationship explains rising fire potential; the smoke-exposure model above measures who was actually reached.

04

How the numbers are calculated

The same threshold and fixed population are applied to every comparable year

The headline metric answers one question: How many people encountered a high wildfire-smoke increment, and for how many days?

1

Begin with one wildfire-smoke estimate per county-day

Input
Stanford ECHO v2 beta daily county predictions, 2006–2023.
Calculation
The Stanford model combines ground monitors, NOAA satellite smoke plumes, weather, fire and land variables to estimate the PM₂.₅ attributable to wildfire smoke—separate from other pollution.
Output
Daily county-derived output → with date, qualifying population and national exposure totals.
2

Apply the same high-smoke rule to every county-day

Input
Modeled wildfire-attributable PM₂.₅ and county GEOID.
Calculation
Mark a county-day as high smoke when smoke PM₂.₅ ≥ 35.5 µg/m³. That concentration is the lower PM₂.₅ breakpoint for AQI 101; the calculation applies it to the wildfire-only increment, not total-air AQI.
Output
A binary high-smoke flag for each county and date. The threshold does not change between years.
3

Attach a fixed population to qualifying counties

Input
July 1, 2020 Census county population joined by GEOID.
Calculation
A qualifying county contributes its full fixed population for that day; a non-qualifying county contributes zero. Holding population constant keeps population growth and migration from manufacturing a trend.
Output
county population exposed = high-smoke flag × fixed 2020 population.
4

Aggregate people across dates, then summarize each year

Input
All qualifying county populations for every date in a calendar year.
Calculation
Sum county populations by date, then sum those daily totals across the year for person-days. Record the yearly maximum daily exposure and count a widespread smoke day when at least 10 million residents qualify at once.
Output
Annual metric output → with person-days, widespread days, annual reach and peak population exposed.
  1. Person-days of unhealthy smoke air. The formal unit remains population-days: ten people exposed for one day and one person exposed for ten days both equal ten person-days.
  2. Days per person. Annual person-days divided by the fixed 327,345,959-resident population. It is a national average burden, not proof that every resident experienced the average.
  3. Share reached at least once. Fixed residents of any county that crossed the threshold on one or more days, divided by the same fixed population. Residence is not daily movement.
  4. Widespread smoke day. Counties containing at least 10 million fixed residents crossed the threshold at once.
  5. Severe smoke day. A sensitivity threshold: modeled wildfire-smoke PM₂.₅ reached at least 50 µg/m³ in counties containing 10 million or more residents.
  6. Benchmark and streaks. The benchmark is the worst modeled daily reach in 2006–2010. A streak counts adjacent calendar days that each meet the widespread-day rule; missing dates break a streak.
  7. Threshold, not a literal wildfire AQI. Total-air AQI includes every pollution source. This project applies the same concentration breakpoint to the modeled wildfire-only increment.
  8. Comparable series begins in 2006. Earlier decades lack the same national PM₂.₅ monitoring and daily satellite smoke record. Documented historical episodes are shown as context and are never backfilled as modeled exposure.
05

Sources, papers and downloads

Original publisher records, cited research and reproducible outputs

Every original input, exact publisher file or feed, research paper, and project-generated table is linked below. The machine-readable source manifest → preserves retrieval dates, checksums, licenses and comparability notes; the repository → contains the complete processing and validation code.

Original data sources

Use “Original source” for the publisher’s documentation and “Direct data” for the exact file, API, archive, or feed used by the project.

  1. U.S. Census Bureau. 2020 Census DHC P1 Total Population by ZCTA. 2020; 33,774 ZCTAs in the DHC response.
    Original source →Direct data ↓
    2020 fixed population denominator
  2. U.S. Census Bureau. 2020 Gazetteer: ZIP Code Tabulation Areas. 2020 ZCTAs for the 50 states, DC, and Puerto Rico.
    Original source →Direct data ↓
    ZCTA representative points and land area
  3. U.S. Census Bureau. 2020 ZCTA to County Relationship File. 2020 national ZCTA-to-county land intersections.
    Original source →Direct data ↓
    Assign ZCTA to the state containing its largest land-area share
  4. U.S. Census Bureau. 2020 Census Apportionment Results. Resident population by state and decade through the 2020 Census.
    Original source →Direct data ↓
    Independent 2020 CONUS-plus-DC resident-population coverage denominator
  5. Colorado State University Libraries. Spatially interpolated PM2.5 concentrations for the US for 2021. CONUS, daily 2021, 15 km grid.
    Original source →Direct data ↓
    Independent-model validation example using the released 2021 median-background grid; not an input to the Stanford annual series
  6. Dryad / Colorado State University. Spatially interpolated non-smoke and smoke PM2.5 concentrations for the US: 2024 mean-background grid. CONUS, daily 2024, 15 km grid; standalone file in the 2006-2024 Dryad v5 release.
    Original source →Direct data ↓
    Pinned independent-model 2024 processor target using the publisher-recommended mean background. No 2024 output is checked into this release because automated retrieval was blocked; it is never appended to Stanford as the same estimator
  7. Dryad / Colorado State University. Spatially interpolated non-smoke and smoke PM2.5 concentrations for the US: 2006-2023 mean-background grids. CONUS, daily 2006-2023, 15 km grid; 18 annual netCDF members in one 5.71 GB archive.
    Original source →Direct data ↓
    Pinned optional all-year independent-model input. The streaming processor supports it, but the release workstation intentionally does not download this multi-GB archive when storage is constrained
  8. Stanford Environmental Change and Human Outcomes Lab. Wildfire Smoke PM2.5 Predictions v2.0 beta. U.S. counties, daily 2006-2023.
    Original source →Direct data ↓
    Controlling 2006-2023 annual website series and validation reference
  9. U.S. Census Bureau. Annual Resident Population Estimates for Counties, 2010 to 2020. July 1, 2020 county estimates for the United States.
    Original source →Direct data ↓
    Fixed July 1, 2020 county population weights for Stanford aggregation
  10. Harvard Dataverse / Stanford ECHO. Daily local-level estimates of ambient wildfire smoke PM2.5 for the contiguous US. U.S. counties, daily 2006-2020 (version 1, not the website's v2 beta).
    Original source →Direct data ↓
    Archived prior-version reference; not used to calculate the 2006-2023 dashboard series
  11. U.S. Environmental Protection Agency. AQS Daily Data, PM2.5 parameters 88101 and 88502. CONUS and DC regulatory monitor-day observations, 2006-2026; 2026 lags the operational AirNow supplement.
    Original source →Direct data ↓
    Official total-PM2.5 source for the consistent 2006-2026 AQS/HMS screen. Event-included summaries are retained; daily bulk files do not expose specific RF/RT/RM fire qualifier codes
  12. U.S. Environmental Protection Agency. AQS Qualifier Codes. Current and legacy AQS qualifier definitions, including RF Canadian fire, RT U.S. wildfire, and RM prescribed fire.
    Original source →Direct data ↓
    Defines fire qualifier semantics. The codes are not present in the bulk daily summaries and are therefore documented but not imputed into the monitor screen
  13. U.S. Environmental Protection Agency AirNow. AirNow Daily Data File: PM2.5 24-hour site averages. U.S. daily PM2.5-24hr monitoring sites, 2026-06-01 through 2026-07-18; preliminary operational supplement.
    Original source →Daily-file pattern →
    Preliminary daily PM2.5 supplement after the lagged AQS 2026 snapshot; combined with HMS only in the explicitly non-comparable operational proxy
  14. NOAA/NESDIS. Hazard Mapping System Smoke Polygons. Daily satellite-analyst smoke plume polygons, 2005-present; annual bundles processed for 2006-2026.
    Original source →Direct data ↓
    Smoke-presence screen for every 2006-2026 monitor-day; includes U.S. and Canadian smoke reaching the U.S. but does not identify plume origin
  15. NOAA National Centers for Environmental Information. Integrated Surface Database Global Hourly present-weather observations. Global hourly station observations, 1901-present; North American coverage increases sharply in the 1940s and 1970s.
    Original source →Direct data ↓
    Candidate pre-2006 observation-context index using explicit smoke present-weather codes; not PM2.5 exposure and not comparable across changing station coverage
  16. USGS EROS and USDA Forest Service. MTBS Fire Occurrence Points. Large U.S. fires, 1984-2026 current archive; 2025-2026 incomplete/provisional.
    Original source →Direct data ↓
    Source-covered normalized U.S. large-wildfire records, annual counts and acres, and non-attributive tooltip context; not smoke exposure
  17. USDA Forest Service Research Data Archive. Spatial wildfire occurrence data for the United States, 1992-2020 (FPA FOD sixth edition). United States, 2.3 million deduplicated final-fire reports, 1992-2020.
    Original source →Direct data ↓
    Best available comprehensive U.S. all-size wildfire occurrence backbone; complements rather than replaces MTBS mapped large-fire area
  18. National Interagency Fire Center. InFORM Fire Occurrence Data Records. United States public wildfire occurrence records, 2021-2026 snapshot; 565,764 retrieved records include approved and unapproved records and ongoing historical incorporation.
    Original source →Direct data ↓
    Dated post-2020 U.S. incident catalog lane. Live-service changes during retrieval are recorded in its receipt; counts are not official annual statistics and are not appended to FPA as a trend
  19. U.S. Environmental Protection Agency. AQI Breakpoints. Current PM2.5 AQI breakpoints; table updated July 14, 2026.
    Original source →
    Defines 35.5 micrograms per cubic meter as the lower PM2.5 concentration breakpoint for AQI 101 and supplies the official category scale
  20. Natural Resources Canada, Canadian Forest Service. National Burned Area Composite annual summary. Canada, complete annual adjusted burned area, 1972-2025; revision 20260513.
    Original source →Direct data ↓
    Canadian burned-area context; fire counts intentionally omitted because NBAC does not supply a homogeneous count series
  21. Natural Resources Canada, Canadian Forest Service. Canadian National Fire Database large-fire point records. Canadian reported large-fire point records, 1946-2025 in this release; completeness varies by agency and year.
    Original source →Direct data ↓
    Source-covered normalized Canadian large-fire records, annual rollup and non-attributive tooltip context; never causal attribution for U.S. smoke exposure
  22. Natural Resources Canada, Canadian Forest Service. Canadian National Fire Database all reported fire point records. Canada, provincial and territorial reported fire records 1950-2025 in the current archive; agency-year completeness varies and early coverage is partial.
    Original source →Direct data ↓
    Best available comprehensive Canadian reported-fire occurrence backbone; agency-year coverage flags are required before trend analysis
  23. National Interagency Fire Center / National Interagency Coordination Center. Incident Management Situation Report. United States, 2026 year-to-date through 2026-07-18.
    Original source →Direct data ↓
    Current 2026 U.S. fire count and acres-burned context; preliminary and not spliced into MTBS history
  24. Natural Resources Canada. Reported Fire Statistics by Response Type. Canada, 2026 year-to-date through 2026-07-18.
    Original source →Direct data ↓
    Current 2026 Canadian fire count and area-burned context; preliminary and not spliced into NBAC history
  25. Natural Resources Canada. Canadian Wildland Fire Information System. Canadian active fires, fire weather, and smoke products.
    Original source →
    Canadian fire context and future attribution validation
  26. Smoke Exposure project. Documented U.S. wildfire-smoke events, 1950-2005. 19 accepted documentary U.S. wildfire-smoke episodes from 1950 through 2005; intentionally non-exhaustive.
    Original source →Direct data ↓
    Controlling pre-2006 documentary event table; never a complete incidence census or comparable PM2.5 series
  27. Smoke Exposure project. Observation evidence for documented U.S. wildfire-smoke events. 27 government-source observation records supporting the accepted 1950-2005 event catalog.
    Original source →Direct data ↓
    Traceable evidence behind each interpreted historical event
  28. Smoke Exposure project. Source citations for documented U.S. wildfire-smoke events. 27 authoritative government or published source records used by the historical evidence catalog.
    Original source →Direct data ↓
    Original-source URLs and authority metadata for the documentary event and observation tables
  29. NOAA National Centers for Environmental Information. NOAA Storm Events annual detail files. Annual detail files 1950-2005; Wildfire is present as an event type only from 1996 in these files.
    Original source →Direct data ↓
    Deterministic wildfire-incident candidate and narrative smoke-keyword screen; changing event types and narrative practices prevent a complete smoke series

Research and methodology papers

  1. Daily Local-Level Estimates of Ambient Wildfire Smoke PM2.5 for the Contiguous US. Childs, M. L., et al. Environmental Science & Technology 56, 13607-13621 (2022).
    Paper →PDF ↓peer reviewed
    Published method for the Stanford v1 daily 10 km and local-area wildfire-smoke PM2.5 estimates.
  2. Growing wildfire-derived PM2.5 across the contiguous U.S. and implications for air quality regulation. Childs, M. L., et al. EarthArXiv preprint, revised August 18, 2025.
    Paper →PDF ↓preprint; not peer reviewed
    Method paper identified by Stanford for the v2.0 beta dataset that controls this report's 2006-2023 estimates.
  3. Contribution of Wildland-Fire Smoke to US PM2.5 and Its Influence on Recent Trends. O'Dell, K., Ford, B., Fischer, E. V. & Pierce, J. R. Environmental Science & Technology 53, 1797-1804 (2019).
    Paper →peer reviewed
    Published monitor-plus-satellite interpolation method underlying the independent O'Dell/Ford smoke grids.
  4. The contribution of wildfire to PM2.5 trends in the USA. Burke, M., et al. Nature 622, 761-766 (2023).
    Paper →peer reviewed
    National 2000-2022 research context for wildfire smoke's effect on U.S. PM2.5 trends.
  5. Summertime Trends in U.S. Population Exposure to Fine Particulate Air Pollution (PM2.5) from Wildfire Smoke. Volckens, J. Environmental Science & Technology Letters 11, 1182-1186 (2024).
    Paper →PDF ↓peer reviewed; open access
    Independent 2000-2023 monitor and satellite-plume analysis of population exposure and high-PM2.5 smoke events.
  6. Long-range PM2.5 pollution and health impacts from the 2023 Canadian wildfires. Zhang, Q., et al. Nature (2025).
    Paper →PDF ↓peer reviewed; open access
    Transboundary attribution and U.S. population-exposure context for Canadian wildfire smoke in 2023.
  7. Health and Regulatory Impacts of PM2.5 From Wildland Fires for 2019-2024 in the US. Jaffe, D. A., et al. GeoHealth 10 (2026).
    Paper →PDF ↓peer reviewed; open access
    Latest published national monitor and satellite-smoke analysis extending research context through 2024.
  8. The Great Smoke Pall - September 24-30, 1950. U.S. Weather Bureau. Weatherwise 3, 129-130 (1950).
    Paper →published historical report
    Primary published documentation for the 1950 transboundary episode; contextual, not comparable PM2.5 exposure data.

Project data downloads

  1. Stanford annual smoke, 2006–2023 ↓Annual comparable exposure metrics used in the headline trend.
  2. Stanford daily smoke, 2006–2023 ↓County-day aggregation behind widespread days and person-days.
  3. Monitor + satellite smoke check, 2024–2026 ↓Recent high-PM₂.₅ observations where NOAA also mapped smoke overhead.
  4. Monitor + satellite annual screen, 2006–2026 ↓A conceptually consistent annual AQS/HMS screen; 2026 adds preliminary AirNow and remains non-comparable with Stanford.
  5. Monitor + satellite daily screen, 2006–2026 ↓All 7,504 daily screen rows behind the annual and same-window summaries.
  6. Monitor + satellite same-window screen ↓Every year compared from Jan 1 through July 18, with a fixed-site coverage sensitivity.
  7. Monitor-network coverage audit ↓Monitor counts, population footprint, HMS coverage, and balanced-panel size by year.
  8. Monitor + satellite processing audit ↓Threshold, event-row policy, source cutoffs, balanced-panel membership, and comparability boundary.
  9. Canada annual burned area, 1972–2025 ↓NBAC adjusted complete-year series.
  10. U.S. mapped large-fire area, 1984–2024 ↓MTBS complete historical extract.
  11. Current fire activity, 2026 ↓Preliminary NIFC and NRCan CWFIF year-to-date totals.
  12. Historical large-fire incidents ↓Every normalized source-covered U.S. and Canadian large-fire record; coverage varies by product, agency and year.
  13. Historical large-fire annual rollup ↓Year-country counts and acres reconciled to the normalized incident table.
  14. All-fire annual source rollup ↓Country-specific annual coverage from U.S. FPA FOD and the full Canadian NFDB point archive.
  15. Monthly source-covered fire records, 1950–2026 ↓Month-country counts and recorded acres from one designated archive per era; blanks mean no national catalog.
  16. Annual incident catalog, 1950–2026 ↓Annual reconciliation of the monthly source-covered incident catalog.
  17. Incident catalog source lanes ↓Exact archive, scope and coverage boundary used for every country-year.
  18. North American fire-density grid ↓The exact 30×24 spatial aggregate rendered in the reported-fire density map.
  19. Annual fire context ↓Largest named or identified large-fire records by year and country for chart context.
  20. Modeled smoke burden by state ↓State-year exposure metrics reconstructed from the Stanford county estimates.
  21. Historical smoke events, 1950–2005 ↓One reviewed U.S. smoke-impact episode with evidence strength, attribution, regions, and source keys.
  22. Historical smoke observations ↓One source-specific observation supporting each interpreted event.
  23. Historical smoke source index ↓The original government and published sources behind the documentary record.
  24. Historical monthly search ledger ↓Every month from 1950–2005 labeled documented, source-screened without a case, or not systematically reviewed.
  25. NOAA wildfire candidate index ↓NOAA Storm Events Wildfire records from 1996–2005; candidate context, not a smoke census.
  26. Source manifest ↓Retrieval dates, checksums, roles and comparability notes for the project inputs.

Historical and interpretive context

These records document episodes, reporting, or health analogies. They do not control the comparable annual exposure calculation.

  1. USA Today. We charted every wildfire since 1984. There's been a drastic increase. Published July 2026; MTBS-based historical fire context.
    Original source →
    Inspiration and journalistic context; MTBS controls reproduced values
  2. U.S. Weather Bureau / Weatherwise. The Great Smoke Pall - September 24-30, 1950. Documented September 1950 transboundary smoke episode.
    Original source →
    Pre-instrumental historical episode context only
  3. NASA Earth Observatory. Smoke from Canadian Fires Blankets Eastern U.S.. July 2002 Quebec wildfire-smoke episode.
    Original source →
    Pre-2006 documented U.S. air-quality episode context only
  4. Ben Noll. What's with all the smoke the last few years?. Published July 18, 2026.
    Original source →
    Event/context lead; underlying Canadian fire and temperature datasets must control any reused values
  5. The Washington Post. Thick wildfire smoke lingers over Mid-Atlantic early Saturday. Published July 18, 2026.
    Original source →
    Documented 2026 episode and lead to operational NOAA/EPA products
  6. Robert Rohde. CWFIS active fires and estimated smoke for July 18. Published July 18, 2026.
    Original source →
    Event-attribution lead for Ontario/northwestern Canadian fires; CWFIS controls measurements
  7. Berkeley Earth. Air Pollution and Cigarette Equivalence. Published 2015; rough PM2.5 health-impact analogy.
    Original source →
    Hover-card context only; supplies the approximate 22 micrograms per cubic meter for one day per cigarette-equivalent rule
  8. IQAir. 2025 World Air Quality Report. Calendar 2025 city-level annual PM2.5; released March 24, 2026.
    Original source →
    Hover-card context only; supplies Loni, India annual average PM2.5 of 112.5 micrograms per cubic meter
  9. Energy Policy Institute at the University of Chicago. Air Quality Life Index Methodology. Long-term PM2.5 exposure and life-expectancy methodology.
    Original source →
    Hover-card context only; supplies the 0.98 life-years per sustained additional 10 micrograms per cubic meter relationship