Sarawak had a DOE station over API 100 on 41 of the 45 days in DOE’s archive (38 on fixed stations only), Johor and Negeri Sembilan on 35, in the season Indonesia logged its second-highest count of fire detections in 15 years.

Counts are satellite detections, and a detection is not a fire. Day counts are floors: DOE’s archive is missing all of 4 Sep and 67 other hours. The study follows the 2026 haze through three places: Malaysia’s air, Indonesia’s burning and the path the air took between them.

Built from public data: DOE APIMS hourly readings, NASA FIRMS satellite detections, Copernicus (CAMS, ERA5), WorldPop residents and NOAA CPC’s El Niño indices. NOAA ended S-NPP delivery at 13:00 UTC on 2 Nov 2026, after this study’s window; the window is unaffected, but this S-NPP-only series cannot continue. Correct as of the study date, 16 Oct 2026; never refreshed.

West Kalimantan Central Kalimantan South Sumatra Riau Papua Sarawak Johor Selangor

Detections, 2026Under 1,300 7 provinces1,300–2,699 6 provinces2,700–7,199 7 provinces7,200–13,999 6 provinces14,000 or more 8 provinces

Days with a station over API 1000–9 4 states10–19 2 states20–29 3 states30–39 6 states40 or more 1 state

Figure 1. Indonesia’s 2026 detections by province, and how many days each Malaysian state had a station over API 100. Provinces are shaded in fifths of their VIIRS S-NPP detections, 1 Jan – 15 Oct 2026; states by the days on which any of their stations read over 100. A detection is not a fire; S-NPP only; 2026 counts are lower bounds. State days are floors. Papua and West Papua sit in the inset at lower left; Singapore is drawn as a point. Pick a name above, or a shape, to zoom: the line under the map then gives that shape’s count (for a state, on all stations and on fixed stations only), as does hovering a shape. Whole region resets. Provinces: geoBoundaries 2017 set, 34 provinces, before the 2022 Papua split, simplified for this map by the study. Detections: NASA FIRMS, aggregated by the study. Map data from OpenStreetMap (https://www.openstreetmap.org/copyright), via geoBoundaries, available under the Open Database License (ODbL).

Every one of the 45 days in DOE’s archive had at least one station over API 100, fixed stations included; on 32 days 20 or more stations did (28 on fixed stations only).Sarawak Selangor Johor Negeri Sembilan

3 of 68 stations are mobile units, so every count is given both ways. Selangor moves most: 37 days counting its Johan Setia mobile unit, 29 on fixed stations only. Day counts are floors: the archive holds readings for 1,000 of its 1,091 hours (Where it landed).

All 68 stations65 fixed stations

Sarawak		41 38
Selangor		37 29
Johor		35 35
Negeri Sembilan		35 35
WP Kuala Lumpur		33 33
Melaka		31 31
WP Putrajaya		31 31
Pahang		26 26
Perak		24 24
Sabah		21 19
Terengganu		19 19
Pulau Pinang		12 12
Kedah		9 9
WP Labuan		8 8
Kelantan		4 4

Figure 2. Days with a station over API 100, by state, 1 Sep – 16 Oct 2026: all stations (ink, first number) and fixed stations only (grey, second).

On 16 September, 26.5 million of the 32.7 million residents linked to a station lived nearest one that read over API 100 (26.4 million counting fixed stations only).

People whose nearest DOE station, within 50 km, read over 100 that day. WorldPop 2020 residents, not citizens. The day-by-day count is on Where it landed.

All stationsFixed stations only

Johor		4.2 m 4.2 m
Selangor		4.2 m 4.1 m
WP Kuala Lumpur		4.0 m 4.0 m
Perak		2.7 m 2.7 m
Kedah		2.2 m 2.2 m
Pulau Pinang		1.9 m 1.9 m
Kelantan		1.4 m 1.4 m
Pahang		1.3 m 1.3 m
Terengganu		1.3 m 1.3 m
Negeri Sembilan		1.2 m 1.2 m
Melaka		1.0 m 1.0 m
WP Putrajaya		0.9 m 1.0 m
Sarawak		0.1 m 0

Figure 3. Residents nearest a station over API 100 on 16 Sep 2026, by state: all stations (ink, first number) and fixed stations only (grey, second); m is million. States with no station over 100 that day are left out.

Indonesia’s vegetation fires produced 518,186 S-NPP detections from 1 January to 15 October 2026, second in 15 years after 2015’s 648,333.Sumatra Kalimantan Papua
A detection is not a fire. S-NPP alone. 2026 is a lower bound: the satellite gaps hide about 4.0%, and the near-real-time filter puts Jul–Oct about 1% low. NOAA ended S-NPP delivery at 13:00 UTC on 2 Nov 2026, after this study’s window; the window is unaffected, but this S-NPP-only series cannot continue.
1. 2012
4. 648,3332015
15. 518,1862026
Figure 4. Indonesia’s VIIRS S-NPP vegetation-fire detections, 1 Jan – 15 Oct of each year, 2012–2026: 2026 in red, its dashed cap the detections the satellite gaps hide (an estimate); 2015 in grey. NASA FIRMS, aggregated by the study.
On the 26 haze days, back-trajectories say the air reaching Johor Bahru and Singapore had passed low over both Sumatran and Kalimantan detections, Sumatra more often.
Of the haze-day arrivals the ensemble agreed on, 52% at both receptors had passed low over Sumatran detections, 39% at Johor Bahru and 40% at Singapore over Kalimantan’s. No area reaches the 70% rule set before the run. The ensemble agrees on 39–40% of arrivals. Trajectories show where the air passed, not how much smoke it carried. Window: arrivals 1 Sep – 15 Oct UTC (How it travelled).

Johor Bahru 256 agreed arrivals
Sumatra 52% Kalimantan 39% Mixed under 1% No fire contact 9%
Singapore 250 agreed arrivals
Sumatra 52% Kalimantan 40% Mixed 1% No fire contact 7%
Figure 5. Haze-day arrivals the ensemble agreed on, by what the air passed low over: Sumatran detections (lime), Kalimantan’s (grey), a mix with no area holding 70% (pale), or no detection (white).
July–September 2026 ran at +1.69 on CPC’s Relative ONI, second since 1950 to 1997, and all 12 fire points had their driest August–September of the 12 seasons since 2015.
ERA5 rain starts in 2015, so 1997 compares on El Niño alone (Why 2026).

Limits

Seasons are not ranked on modelled smoke: Copernicus’ reanalysis puts 2015 and 2019 the other way round from its forecasts, for East Malaysia and for the Peninsula, so seasons are compared on detection counts only.
There is no fixed delay between burning in Indonesia and haze in Malaysia, and the study does not claim one.
A detection is a hot pixel seen from orbit, not a fire and not burned area.
Back-trajectories show where the air passed, not how much smoke it carried.
Papua and West Papua are counted in every Indonesia total; their smoke is not traced.
The study names no company or concession.
Nothing after 16 Oct 2026 is used, except NOAA’s end of S-NPP delivery.

Data, method and credits

Air quality: Department of Environment Malaysia (Jabatan Alam Sekitar), Air Pollutant Index Management System (APIMS), https://eqms.doe.gov.my/APIMS/main. “Data dan maklumat tertakluk di bawah Terma Penggunaan Data Terbuka Kerajaan Malaysia 1.0” (https://www.doe.gov.my/wp-content/uploads/2021/07/TermaPenggunaan.pdf).
Detections: “We acknowledge the use of data and/or imagery from NASA's Fire Information for Resource Management System (FIRMS) (https://earthdata.nasa.gov/firms), part of NASA's Land, Atmosphere Near real-time Capability for Earth observations (LANCE) (https://earthdata.nasa.gov/lance) and NASA's Earth Science Data and Information System (ESDIS).”
Residents: Bondarenko M., Kerr D., Sorichetta A., and Tatem, A.J. 2020, WorldPop, University of Southampton, UK. doi:10.5258/SOTON/WP00684 (CC BY 4.0). Summed to 1 km by the study.
Weather and modelled smoke: “Contains modified Copernicus Atmosphere Monitoring Service information 2026. Contains modified Copernicus Climate Change Service information 2026. Neither the European Commission nor ECMWF is responsible for any use that may be made of the Copernicus information or data it contains.”
El Niño: ONI and RONI: NOAA Climate Prediction Center (public domain).
Boundaries: geoBoundaries (Runfola et al. 2020, https://doi.org/10.1371/journal.pone.0231866), CC BY 4.0; provinces and states contain information from OpenStreetMap, under the ODbL.
Boundary files under the ODbL, free online at geoBoundaries’ commit-pinned URLs: github.com/wmgeolab/geoBoundaries/raw/9469f09/releaseData/gbOpen/IDN/ADM1/geoBoundaries-IDN-ADM1.geojson, github.com/wmgeolab/geoBoundaries/raw/9469f09/releaseData/gbOpen/MYS/ADM1/geoBoundaries-MYS-ADM1.geojson, github.com/wmgeolab/geoBoundaries/raw/9469f09/releaseData/gbOpen/TLS/ADM0/geoBoundaries-TLS-ADM0.geojson, github.com/wmgeolab/geoBoundaries/raw/9469f09/releaseData/gbOpen/PNG/ADM0/geoBoundaries-PNG-ADM0_simplified.geojson and github.com/wmgeolab/geoBoundaries/raw/9469f09/releaseData/gbOpen/THA/ADM0/geoBoundaries-THA-ADM0_simplified.geojson. We repaired and simplified them (tolerance 0.0005°) and used them to assign fire detections to provinces, states and countries.
No credited source or agency endorses this study, and no figure here is official.
Every licence, DOI and the method: Method and credits.