← City rankings

Air events

Every named air event, with the evidence

822 causally classified PM2.5 events across the 100 largest US metros. Each is resolved to a root cause through retained evidence—official assessments, agency statements, news coverage, or NOAA records—or a clearly disclosed lower-confidence inference. The 425 remaining local episodes stay with their city records; they are not presented here as causal or regional events.

How to read this: local episodes come from each metro's monitor record. Nearby episodes may be retained internally as observation clusters for discovery, but proximity and timing do not make them one public event. Regional events appear only where retained evidence supports a shared source occurrence, transport progression, or mechanism. They have not been reviewed and signed off one by one, so boundaries, causes, and names can change as evidence accumulates, and provisional monitor data may still shift. Every cataloged event carries an evidence badge saying how far its cause is actually supported, and every cause links the evidence it rests on. Weak evidence remains visible as weak evidence; absence of shared evidence leaves the observation local.

Where the evidence is thin, that is often a gap in our reviewing rather than a gap in the record. The discovery totals below describe the frozen 1265-candidate baseline, before unresolved multi-metro parents were decomposed into place-first research units. News discovery surfaced 7,047 candidate articles; 7,047 have been assessed and 0 are still queued. Of the 1,265 candidate episodes, 1,093 have never been searched at all. Individual events say where they stand.

All air events, 1999 to 2026 Each bar is one year of evidence-backed events. Click a bar to filter to it, or a dot to open an event. Average PM2.5 across all covered metros, 15.7 µg/m³ in 1999 to 9.6 µg/m³ in 2026. Fixed baseline: it does not move when you filter.
Cause
Evidence
Severity
Filters
Rank by
Wildfire smoke Agency determination Jul 20 to Jul 21, 2026 provisional

2026 July 20–21 Boise City wildfire smoke episode →

57.7 µg/m³ peak ≈ 4.5 cigarettes per person 846k people affected

Wildfire smoke over Boise City, 2 days, peaking at 57.7 µg/m³ near Boise City.

Affected metros (1)
Evidence

NWS alert 1Agency/press 1

Uncertainties (1)
  • The supplied evidence does not identify a sufficiently specific shared fire or origin region for a broader merge.
Wildfire smoke Agency determination Jul 20, 2026 provisional

2026 July 20 St. Louis wildfire smoke episode →

51.5 µg/m³ peak ≈ 1.8 cigarettes per person 2.8M people affected

Wildfire smoke over St. Louis, 1 day, peaking at 51.5 µg/m³ near St. Louis.

Affected metros (1)
Evidence

NWS alert 1

Uncertainties (1)
  • The supplied evidence does not identify a sufficiently specific shared fire or origin region for a broader merge.
Wildfire smoke Documented Jul 17, 2026 provisional

2026-07-17 wildfire smoke episode affecting Spokane-Spokane Valley, WA →

130.6 µg/m³ peak ≈ 5.6 cigarettes per person 605k people affected

Wildfire smoke over Spokane, 1 day, peaking at 130.6 µg/m³ near Spokane.

Affected metros (1)
Evidence

Agency/press 1

Wildfire smoke Agency determination Jul 15 to Jul 20, 2026 provisional

July 15–20, 2026 Great Lakes and eastern U.S. wildfire-smoke episode →

Origin: Canada; Northern Minnesota

387.8 µg/m³ peak ≈ 13.0 cigarettes per person 86M people affected 14 city records set, 27 top-ten days

Wildfire smoke over 32 metros, 6 days, peaking at 387.8 µg/m³ near Toledo. 14 metros set an all-time record. Origin traced to Canada; Northern Minnesota.

Evidence

EPA AQS 1NWS alert 6Agency/press 6gdelt-doc-api 1

Uncertainties (1)
  • The supplied evidence does not identify or quantify every specific source contributing across all proposed members.
Dust Likely Jul 9, 2026 provisional

2026-07-09 Tucson, AZ dust episode →

30.5 µg/m³ peak ≈ 1.1 cigarettes per person 1.1M people affected 1 top-ten day

Dust over Tucson, 1 day, peaking at 30.5 µg/m³ near Tucson.

Affected metros (1)
Evidence

Reviewed source 1gdelt-doc-api 2

Uncertainties (1)
  • The classification is bounded to the retained evidence and does not apportion every source contributing to the daily PM2.5 value.
Fireworks Corroborated Jul 5, 2026 provisional

2026-07-05 fireworks episode affecting Washington-Arlington-Alexandria, DC-VA-MD-WV →

54.7 µg/m³ peak ≈ 2.0 cigarettes per person 6.4M people affected

Fireworks over Washington, 1 day, peaking at 54.7 µg/m³ near Washington.

Affected metros (1)
Evidence

Agency/press 2

Fireworks Corroborated Jul 5, 2026 provisional

2026-07-05 fireworks episode affecting Miami-Fort Lauderdale-West Palm Beach, FL →

51 µg/m³ peak ≈ 1.9 cigarettes per person 6.5M people affected 1 top-ten day

Fireworks over Miami, 1 day, peaking at 51 µg/m³ near Miami.

Affected metros (1)
Evidence

Agency/press 2

Fireworks Corroborated Jul 4 to Jul 5, 2026 provisional

2026-07-04 fireworks episode affecting Des Moines-West Des Moines, IA, Omaha-Council Bluffs, NE-IA, Kansas City, MO-KS →

72.1 µg/m³ peak ≈ 2.6 cigarettes per person 4.0M people affected 1 city record set, 2 top-ten days

Fireworks over 3 metros, 2 days, peaking at 72.1 µg/m³ near Omaha. 1 metro set an all-time record.

Evidence

Agency/press 2 4

Wildfire smoke Documented Jul 3, 2026 provisional

2026-07-03 wildfire smoke episode affecting Colorado Springs, CO →

25.8 µg/m³ peak ≈ 0.9 cigarettes per person 778k people affected

Wildfire smoke over Colorado Springs, 1 day, peaking at 25.8 µg/m³ near Colorado Springs.

Affected metros (1)
Evidence

Agency/press 2

Wildfire smoke Corroborated May 7, 2026 provisional

2026-05-07 wildfire smoke episode affecting Jacksonville, FL →

Named source South Canal Fire

64.9 µg/m³ peak ≈ 2.6 cigarettes per person 1.8M people affected 1 top-ten day

Wildfire smoke over Jacksonville, 1 day, peaking at 64.9 µg/m³ near Jacksonville.

Affected metros (1)
Evidence

Agency/press 1

Prescribed fire Likely May 6, 2026 provisional

2026-05-06 McAllen-Edinburg-Mission, TX prescribed-fire smoke episode →

44.1 µg/m³ peak ≈ 1.6 cigarettes per person 915k people affected

Prescribed fire over McAllen, 1 day, peaking at 44.1 µg/m³ near McAllen.

Affected metros (1)
Evidence

Reviewed source 2

Uncertainties (1)
  • The evidence supports transported cross-border burn smoke, but does not distinguish agricultural residue burning from other open burning or quantify its PM2.5 share.
Prescribed fire Likely Apr 27, 2026 provisional

2026-04-27 Austin-Round Rock, TX prescribed-fire smoke episode →

36.1 µg/m³ peak ≈ 1.2 cigarettes per person 2.6M people affected

Prescribed fire over Austin, 1 day, peaking at 36.1 µg/m³ near Austin.

Affected metros (1)
Evidence

Agency/press 1Reviewed source 1

Uncertainties (1)
  • The retained evidence is scoped to the places and dates stated; it is not a catalog promotion.
Dust Likely Mar 28, 2026

2026-03-28 Tucson, AZ dust episode →

30.3 µg/m³ peak ≈ 1.1 cigarettes per person 1.1M people affected 1 top-ten day

Dust over Tucson, 1 day, peaking at 30.3 µg/m³ near Tucson.

Affected metros (1)
Evidence

Reviewed source 2

Uncertainties (1)
  • Dust is confidently identified as the episode driver, but the evidence does not quantify its fraction of the certified daily PM2.5 value.
Prescribed fire EPA monitor flag Mar 26, 2026

2026-03-26 prescribed-fire smoke episode affecting Chattanooga, TN-GA →

43.8 µg/m³ peak ≈ 1.5 cigarettes per person 588k people affected

Prescribed fire over Chattanooga, 1 day, peaking at 43.8 µg/m³ near Chattanooga.

Affected metros (1)
Evidence

EPA AQS 1News 1

Mixed causes EPA monitor flag Mar 21, 2026

2026-03-21 mixed causes episode affecting Augusta-Richmond County, GA-SC →

50.2 µg/m³ peak ≈ 1.8 cigarettes per person 637k people affected 1 top-ten day

Mixed causes over Augusta, 1 day, peaking at 50.2 µg/m³ near Augusta.

Affected metros (1)
Evidence

EPA AQS 2

Uncertainties (1)
  • The supplied evidence does not quantify the relative contribution of each supported cause.
Prescribed fire EPA monitor flag Mar 20, 2026

2026-03-20 prescribed-fire smoke episode affecting Wichita, KS →

46.4 µg/m³ peak ≈ 1.7 cigarettes per person 661k people affected

Prescribed fire over Wichita, 1 day, peaking at 46.4 µg/m³ near Wichita.

Affected metros (1)
Evidence

EPA AQS 1Agency/press 2

Prescribed fire EPA monitor flag Mar 1, 2026

2026-03-01 prescribed-fire smoke episode affecting Jackson, MS →

35.8 µg/m³ peak ≈ 1.2 cigarettes per person 606k people affected

Prescribed fire over Jackson, 1 day, peaking at 35.8 µg/m³ near Jackson.

Affected metros (1)
Evidence

EPA AQS 1internet-archive-wayback-cdx 1

Other Documented Feb 19, 2026

2026-02-19 other transported pollution episode affecting McAllen-Edinburg-Mission, TX →

43.9 µg/m³ peak ≈ 1.6 cigarettes per person 915k people affected

Other over McAllen, 1 day, peaking at 43.9 µg/m³ near McAllen.

Affected metros (1)
Evidence

Agency/press 2

Prescribed fire Likely Feb 19, 2026

2026-02-19 Vegetation burning in Mexico episode →

Origin: Vegetation burning in Mexico

35.2 µg/m³ peak ≈ 1.2 cigarettes per person 2.8M people affected

Prescribed fire over San Antonio, 1 day, peaking at 35.2 µg/m³ near San Antonio. Origin traced to Vegetation burning in Mexico.

Affected metros (1)
Evidence

Reviewed source 3

Uncertainties (1)
  • The origin is assigned at the broad Mexico vegetation-burning level; individual burns and their fractional contribution are not known.
Mixed causes EPA monitor flag Feb 17, 2026

2026-02-17 mixed causes episode affecting Wichita, KS →

54.5 µg/m³ peak ≈ 2.1 cigarettes per person 661k people affected 1 top-ten day

Mixed causes over Wichita, 1 day, peaking at 54.5 µg/m³ near Wichita.

Affected metros (1)
Evidence

EPA AQS 1News 1gdelt-doc-api 2

Uncertainties (1)
  • The supplied evidence does not quantify the relative contribution of each supported cause.
Wildfire smoke Likely Feb 3, 2026

2026-02-03 Tiger Bay State Forest wildfires episode →

36.7 µg/m³ peak ≈ 1.3 cigarettes per person 740k people affected 1 top-ten day

Wildfire smoke over Deltona, 1 day, peaking at 36.7 µg/m³ near Deltona.

Affected metros (1)
Evidence

Reviewed source 1

Uncertainties (1)
  • No formal exceptional-event or transport-model attribution was found; the conclusion combines documented local smoke with the measured same-day PM2.5 event.
Wildfire smoke EPA monitor flag Jan 28, 2026

2026-01-28 wildfire smoke episode affecting Charleston-North Charleston, SC →

110.1 µg/m³ peak ≈ 4.6 cigarettes per person 870k people affected 1 city record set

Wildfire smoke over Charleston, 1 day, peaking at 110.1 µg/m³ near Charleston. 1 metro set an all-time record.

Affected metros (1)
Evidence

EPA AQS 1

Prescribed fire Documented Jan 19, 2026

2026-01-19 prescribed-fire smoke episode affecting Orlando-Kissimmee-Sanford, FL →

33.9 µg/m³ peak ≈ 1.2 cigarettes per person 2.9M people affected

Prescribed fire over Orlando, 1 day, peaking at 33.9 µg/m³ near Orlando.

Affected metros (1)
Evidence

News 1gdelt-doc-api 1

Mixed causes EPA monitor flag Jan 13, 2026

2026-01-13 mixed causes episode affecting Augusta-Richmond County, GA-SC →

48.5 µg/m³ peak ≈ 1.7 cigarettes per person 637k people affected 1 top-ten day

Mixed causes over Augusta, 1 day, peaking at 48.5 µg/m³ near Augusta.

Affected metros (1)
Evidence

EPA AQS 2

Uncertainties (1)
  • The supplied evidence does not quantify the relative contribution of each supported cause.

What happens to episodes without an established cause?

425 working episodes remain attached to their individual city records. They are real monitor observations, but IAQng does not promote them into this causal-event catalog until retained evidence supports the event boundary and cause. “Unknown” means unknown—not “no smoke,” “no dust,” or proof that simultaneous cities shared one event.

For discovery, nearby local episodes may still share an internal observation-cluster ID. That is a search convenience only. It is never displayed as evidence that the places experienced one regional event.

How this catalog is built

Episodes start as elevated-PM2.5 candidates in the monitor record (the same data behind the city rankings). Evidence is then gathered per candidate: agency and health-department statements, NOAA Storm Events records, and link-first news coverage. A resolution pass groups candidates into one named episode only when the retained evidence supports the same real-world source, and records its reasoning and uncertainties; everything else stays unattributed.

From monitor readings to a named event

What can merge episodes. Retained evidence that they share one real-world source occurrence: overlapping timing with a plausible progression, compatible origin regions, and corroborating official or news records. The resolution pass must cite the evidence it used, by ID.

What can never merge them. Severity alone, proximity alone, the season, article counts, or a shared cause label without connecting evidence. No PM2.5 threshold is treated as proof of any cause.

The three outcomes. Resolve into a named regional episode; abstain, leaving each local episode unattributed (unknown means unknown, not "no smoke"); or record a dispute and keep the candidates separate.

Evidence roles. Official assessments carry the most weight, then corroboration, hypothesis, and context. Counts are never a credibility score: one agency determination outweighs any number of headlines.

What changes a record. EPA certifying provisional days, new qualified evidence, or editorial review. No record here is final, and a correction is the fastest way to change one.

Grouping policy common_sense_root_cause_grouping.v2, resolved Jul 24, 2026 · working events pm25-unresolved-multimetro-local-episode-overlay.v1 (1444 records; frozen baseline 1265) · news evidence news-evidence-candidates.v1 (1786 items) · official matches pm25-official-source-matches.v2 (56 records). Attribution is automated grouping over retained evidence, not per-episode editorial sign-off. See editorial standards.