← Air events

Prescribed fire Likely

2003-05-08 to 2003-05-11 Dallas-Fort Worth-Arlington, TX, Houston-The Woodlands-Sugar Land, TX, and 1 other metros prescribed-fire smoke episode

May 8, 2003 to May 11, 2003

Daily PM2.5 peaked at 56.7 µg/m³ near Dallas across 3 covered metros home to 17 million people. 2 metros placed a top-ten day in their record.

Counting the extra PM2.5 above each metro's typical day, this event handed the average affected resident ≈ 3.3 cigarettes. Across everyone living in the affected metros, that is the collective equivalent of ≈ 56 million cigarettes. An awareness illustration of outdoor exposure at 22 µg/m³ per cigarette (Berkeley Earth), counting only the event's ranked days; not a medical dose.

Why we think this happened

The basis for “Prescribed fire”

Manual research connected 3 source events to one prescribed-fire smoke episode using the deterministic grouping rule.

Likely Applicable reporting or a supported inference makes this the best current causal explanation.

NASA Earth Observatory corroborates

Fires in Mexico and Central America

NASA imagery shows smoke from agricultural and other fires in Mexico and Central America spreading across the Gulf into the southern United States during the event.

See every retained source and limitation ↓

Sequence: Dallas (May 8) → McAllen (May 9) → Houston (May 11)

Evidence path

From source to city

A cause is more convincing when the record connects three things: something happened, the air mass moved toward the region, and local instruments responded. Here is how much of that chain is documented for this event.

Open the evidence ledger ↓
  1. 1

    Source or mechanism

    Prescribed fire

    NASA imagery shows smoke from agricultural and other fires in Mexico and Central America spreading across the Gulf into the southern United States during the event.

    View supporting source ↗
    Documented
  2. 2

    Air-mass movement

    Transport into the affected region is documented

    NASA imagery shows smoke from agricultural and other fires in Mexico and Central America spreading across the Gulf into the southern United States during the event.

    View transport source ↗
    Documented
  3. 3

    Local response

    56.7 µg/m³ near Dallas

    EPA monitors recorded the event across 3 metros. The peak daily concentration is roughly 2.6 cigarette-equivalents at the outdoor daily rate.

    Inspect the pollutant and weather signals ↓
    Documented

Event autopsy

What the instruments saw

Before, during, and after the episode: nearby pollutant monitors and surface-weather stations show what changed. This context tests the “Prescribed fire” explanation; it does not assign the cause by itself.

Observed context · not attribution

Dallas-Fort Worth-Arlington, TX

Monitor comparison window 2003-05-05 to 2003-05-13 · PM2.5 target: Fort Worth Northwest

6 pollutant signals
Fine-particle pulse 41.9 µg/m³

52% above nearby days

Air movement 11.3 kt

62% higher than surrounding days

PM10 64 µg/m³

No adjacent-day comparison

Pollution fingerprint

Metro-wide median of reporting monitors
PM2.5 41.9 µg/m³

52% higher than surrounding days

Before27.5 µg/m³ Event41.9 µg/m³ After6.80 µg/m³
11 monitors · 24 HOUR
PM10 64 µg/m³

No adjacent-day comparison

Before Event64 µg/m³ After
7 monitors · 24 HOUR
Carbon monoxide 0.592 ppm

21% higher than surrounding days

Before0.504 ppm Event0.592 ppm After0.362 ppm
3 monitors · 1 HOUR
Nitrogen dioxide 5.08 ppb

21% lower than surrounding days

Before6.46 ppb Event5.08 ppb After7.92 ppb
11 monitors · 1 HOUR
Sulfur dioxide 0.283 ppb

239% higher than surrounding days

Before0.083 ppb Event0.283 ppb After0.083 ppb
4 monitors · 1 HOUR
Ozone 0.045 ppm

About the same as surrounding days

Before0.038 ppm Event0.045 ppm After0.035 ppm
19 monitors · 8-HR RUN AVG BEGIN HOUR

Atmospheric conditions

Nearby EPA weather observations
Humidity 69.7 °F

About the same as surrounding days

Pressure 987 Millibars

About the same as surrounding days

Temperature 80.8 °F

About the same as surrounding days

Wind 11.3 kt

62% higher than surrounding days

Source: U.S. EPA AirData daily observations and, where available, nearby ASOS/AWOS surface observations. Values are descriptive medians over the retained monitor set; missing measurements remain missing. Transport and source evidence remain separate layers because this instrument pattern does not assign a cause by itself.

Measured daily PM2.5 at EPA monitors, as cigarette-equivalents
EPA monitors sit in metro areas, so the colour clusters there. Land between and beyond them is unmeasured, not clean: this map shows where IAQng has readings, not the extent of the smoke.

Episode peak

-- cigarettes

The smoke map's scale, replayed over this event: the surface interpolates between covered metros' daily records, framed to the affected region. Hover for a reading, click a metro to pin it to the card. Days a monitor did not report draw nothing.

Views from the selected metro

Camera views near

Browse the camera map →

Select a metro on the map to see nearby cameras. Where an operator archive reaches the event, drag the divider to compare two moments, adjust either date and time, or play the available event-day frames. Camera imagery is visual context, not an air-quality measurement or proof of smoke.

Affected metros

MetroPeak PM2.5Cigarettes, whole eventPeak dayAll-time standing
Dallas-Fort Worth-Arlington, TX 56.7 µg/m³ 6.7 May 10, 2003 #3 of its top days
McAllen-Edinburg-Mission, TX 54 µg/m³ 2.5 May 9, 2003 #9 of its top days
Houston-The Woodlands-Sugar Land, TX 43.9 µg/m³ 2.0 May 11, 2003 #17 of its top days

"Cigarettes, whole event" totals a metro's ranked event days at 22 µg/m³ per cigarette: what a person outdoors there breathed across the event, as the illustration. "All-time standing" is where this episode's days sit in each metro's ranked record (worst-area measure, since 1999). Full local context is on each metro's Air Records page.

Public evidence library

What supports this account

Each card states what a source supports—and what it does not. Official records, research, reporting, community observations, and transparent inferences remain visibly different kinds of evidence.

4 retained sources

Official record

Smoke from Yucatan Fires Reaches Gulf Coast States

NASA Earth Observatory

NASA's May 11 satellite analysis shows smoke from intense Central American burning entrained over the Gulf Coast states.

What it supports

A retained source supports the attributed cause for the stated place and dates.

Limit

The satellite narrative is regional.

Official record

Fires in Mexico and Central America

NASA Earth Observatory

NASA imagery shows smoke from agricultural and other fires in Mexico and Central America spreading across the Gulf into the southern United States during the event.

What it supports

A retained source supports the attributed cause for the stated place and dates.

Official record

Fires in Mexico and Central America

NASA Earth Observatory

NASA reports smoke from scores of fires across southern Mexico and Central America spreading north over the Gulf and southern United States on the event date.

What it supports

A retained source supports the attributed cause for the stated place and dates.

News report

New alert issued for smoke from Mexico, Central America

Houston Chronicle via MyPlainview

Contemporaneous Texas reporting says the state's smoky haze peaked over the weekend and came from agricultural fires in Mexico and Central America.

What it supports

A retained source supports the attributed cause for the stated place and dates.

Limit

The report does not identify individual fires.

What remains uncertain

Status and provenance

Likely Applicable reporting or a supported inference makes this the best current causal explanation.

This episode was grouped from the monitor record by the catalog pipeline and given its cause from the evidence listed above. It has not been individually reviewed and signed off, so its boundaries, cause, and name may change as evidence accumulates. Grouping policy common_sense_root_cause_grouping.v2. The cause label reflects the assembled evidence above, not an IAQng detection formula. If it is wrong, tell us.