Yes, a filter at the helmet is a sensible tool for the next five fire seasons. The usual advice is to stay indoors, but in rural Central Kalimantan the indoor to outdoor PM2.5 ratio was 1.01, so the house gave almost no cover (Graham et al., 2024). A rider has even less.
The smoke is easing this week. The fire years are not going away.
It is Sunday 1 November, and Singapore's air has eased from last month's peak.
The 24-hour PSI read 68 to 86 across the five regions at 8 am today, the Moderate band, down from the 284 µg/m³ one-hour PM2.5 spike in the west on 14 October (NEA, 2026; Reuters, 2026). The west still shows a 24-hour PM2.5 of 43 µg/m³, which is about 2.9 times WHO's 24-hour guideline of 15 (NEA, 2026; WHO, 2021).
So the headline is calming down. This post is about what the research says for the years after the headline.
Key takeaways
- In rural Central Kalimantan the median indoor to outdoor PM2.5 ratio during fire smoke was 1.01, so staying in gave almost no protection (Graham et al., 2024).
- Peatland fire smoke kills on a large scale: about 33,100 adults and 2,900 infants a year on average, and more than 87,000 modelled deaths nationwide in 2019 (Hein et al., 2022; Grosvenor et al., 2024).
- Even finished peat restoration would have cut 2015 PM2.5 emissions by 24%, which leaves 76% of them in the air (Kiely et al., 2021).
- Fire peaks in 2019 and 2023 lined up with El Niño, and forecasters expect El Niño effects to last until the first quarter of 2027 (Gaveau et al., 2026; Reuters, 2026).
The research behind this
Peer-reviewed measurements this post is built on. Each card says what was measured, what was found, and what it can't tell you.
Smoke exposure in Central Kalimantan, 2023
- What they measured
- 13 low-cost PM2.5 sensors, indoors and outdoors, across urban and rural sites in Central Kalimantan from 1 September to 31 October 2023, plus a regional air quality model and a time-use questionnaire.
- What they found
- Mean outdoor PM2.5 was 136 µg/m³, of which fires added 90. The median indoor to outdoor ratio in rural areas was 1.01, and an estimated 1.62 million people were exposed to unhealthy or worse air.
- Can't tell you: two months of one El Niño onset, low-cost sensors, and the exposure total is modelled rather than counted.
The 2019 Palangka Raya fires
- What they measured
- A low-cost sensor network around Palangka Raya through the 2019 fire season, used to check the Copernicus model and then estimate health impacts.
- What they found
- Over 1,200 excess deaths in the Palangka Raya region, over 3,200 across Central Kalimantan and more than 87,000 nationwide in 2019 from fire-induced PM2.5.
- Can't tell you: the deaths are modelled estimates with uncertainty ranges, and they cover one year.
Health impacts of Indonesian peatland fires
- What they measured
- Satellite-derived PM2.5 linked to peatland fires in Sumatra and Kalimantan, combined with epidemiological response functions and the age and disease profile of the exposed population.
- What they found
- On average around 33,100 adults and 2,900 infants die prematurely each year from peatland fire PM2.5. The adult range runs from 19,900 to 64,800 deaths depending on the response function.
- Can't tell you: the biggest uncertainty is the concentration response function, and the estimate is an annual average, not a forecast.
Costs of fires and the benefit of restoring peat
- What they measured
- Fire emissions and land cover data for 2004 to 2015, then a what-if for Indonesia's pledge to restore 2.49 million hectares of degraded peatland.
- What they found
- Had restoration already been finished, the 2015 burned area would have fallen by 6% and PM2.5 emissions by 24%, preventing about 12,000 premature deaths.
- Can't tell you: it is a model of one past year, not a promise about the next.
Monthly burned area in Indonesia, 2019 to 2024
- What they measured
- Monthly Sentinel-2 burned-area maps for all of Indonesia from January 2019 to December 2024, compared with El Niño and Indian Ocean Dipole indices.
- What they found
- Fires burned 5.62 million hectares in total. Burning accelerated around July and peaked in September to October in 2019 and 2023, when the El Niño index was at or above +0.5 °C.
- Can't tell you: it ends in 2024, so it does not measure the 2026 season.
What the studies actually measured
Start with the sensors, because a sensor on a wall does not lie about the room it is in. Graham's team put 13 low-cost PM2.5 monitors indoors and outdoors in Central Kalimantan for the first two months of the 2023 dry season (Graham et al., 2024).
Outdoors, the mean was 136 µg/m³, and fires accounted for 90 of that (Graham et al., 2024). Indoors, in rural homes, the median ratio to outdoors was 1.01 (Graham et al., 2024).
Grosvenor's group did the same job around Palangka Raya in the 2019 season. Their sensors showed the Copernicus model performed fairly well in extreme smoke, which matters because the health numbers lean on it (Grosvenor et al., 2024).
Hein and Kiely then zoom out. Hein turned satellite PM2.5 into annual deaths across Sumatra and Kalimantan (Hein et al., 2022). Kiely asked what restoring peat would have changed in 2015 (Kiely et al., 2021).
Gaveau supplies the calendar. His monthly burned-area maps show when the fire seasons actually peak, and what drives them (Gaveau et al., 2026).
Ride with filtered air
Filters haze PM2.5 on every ride
Easi Breezi clips onto any helmet and feeds you H11 HEPA-filtered air. Bike-powered, no batteries, no charging.
The stay-indoors advice assumes a house that works
Official guidance in smoke events is to stay indoors with windows shut. That advice rests on an idea borrowed from wildfire research elsewhere: a sealed building keeps a large share of the particles out.
Graham tested that idea in Indonesia and the ratio came back at 1.01 in rural homes. The authors note this is considerably higher than ratios reported in wildfires in places such as the USA, so local housing offers little protection (Graham et al., 2024).
The median PM2.5 ratio in rural Central Kalimantan homes during fire smoke. A value of 1 means indoors is as smoky as outdoors.
Source: Graham et al., 2024A rider cannot even use that weak cover. The commute, the delivery round and the school run all happen in the plume.
The BBC's reporting from Palangkaraya shows the same bind from the human side: the official advice is to stay in, and one mother says that is simply not possible (BBC, 2026).
The numbers behind the headlines are large and recurring
Reuters reported 283,270 respiratory illness cases in seven Indonesian provinces as of 13 October, more than double the 113,336 recorded on 9 September (Reuters, 2026). Those are counts of people who got ill, not a model.
The deaths come from models. Hein estimates around 33,100 adult and 2,900 infant premature deaths in an average year, with the adult figure ranging from 19,900 to 64,800 depending on the response function (Hein et al., 2022).
Grosvenor puts 2019 alone at more than 87,000 deaths nationwide (Grosvenor et al., 2024).
Those two figures are not in conflict. One is an average across all years, and the other is a single bad one.
Restoration helps, but it does not erase the smoke
Indonesia has pledged to restore 2.49 million hectares of degraded peatland at an estimated cost of US$3.2 to 7 billion (Kiely et al., 2021). Kiely's what-if says finishing that job before 2015 would have cut PM2.5 emissions by 24% and prevented about 12,000 deaths (Kiely et al., 2021).
That is a big win. It is also a partial one.
By my own arithmetic, a 24% cut leaves 76% of the 2015 PM2.5 emissions in the air. That is a derived figure, not one the paper states.
Plan for the smoke that restoration does not remove. Treat any fire-season week as filter weather, not only the days the headline says Unhealthy.
Our earlier post on restoration versus drought goes deeper on that trade-off.
The fire calendar is tied to El Niño, and 2026 is not over
Gaveau's burned-area maps show fire activity accelerating around July and peaking in September to October in 2019 and 2023, with El Niño at or above +0.5 °C. In the neutral and negative years from 2020 to 2022, burning was minimal (Gaveau et al., 2026).
The same paper also finds that burns in 2019 and 2023 stayed lower than during earlier events, which suggests prevention efforts are limiting forest fires (Gaveau et al., 2026). That is genuine good news, and it is worth stating.
2026 looks like a different scale of season. The BBC cites Nusantara Atlas for about 1.9 million hectares burned across Indonesia since the start of the year, with one million hectares in September alone (BBC, 2026).
Reuters says the effects of El Niño are expected to last until the first quarter of 2027, and that the rains forecast for October are now delayed to November (Reuters, 2026).
What five more seasons look like, on the evidence we have
No source we found gives a number for fire exposure in 2031, so we will not invent one.
What the evidence supports is a pattern. Bad years follow warm El Niño phases (Gaveau et al., 2026). Drained and deforested peatland burns more readily in those years (Graham et al., 2024).
Smoke from the fires travels as far as Singapore, Malaysia and the Philippines (BBC, 2026). And the health cost of a bad year is measured in tens of thousands of deaths (Grosvenor et al., 2024; Hein et al., 2022).
Restoration may shrink the bad years. The El Niño cycle will keep supplying dry ones. Plan for the overlap.
Where a helmet air filter fits, and where it does not
If the house does not shelter you, and the road certainly does not, the only place left to act is the air at your own mouth and nose.
Easi Breezi clips onto your helmet and passes the air you breathe through H11 HEPA media that stops at least 95% of PM2.5. It filters at your intake, which is the one point of the journey you control. See the clip-on helmet air purifier, and check conditions on the live haze page.
It is a particle filter, so it does not remove carbon monoxide or other gases, and a filter does nothing about heat or fatigue on a long ride.
The bigger picture is on our helmet air filtration guide.
Frequently asked questions
Does a motorcycle helmet air filter work against haze?
Easi Breezi's H11 HEPA media is rated to stop at least 95% of PM2.5, the fine-particle class the studies above measure. It does not remove gases such as carbon monoxide.
Is staying indoors enough during haze?
Not always. In rural Central Kalimantan homes the median indoor to outdoor PM2.5 ratio was 1.01 (Graham et al., 2024). The authors conclude that local housing offers little protection.
Will the haze come back next year?
It depends on El Niño and on how dry the peat is. Fire peaks in 2019 and 2023 matched El Niño years, and minimal burning marked 2020 to 2022 (Gaveau et al., 2026). Expect risk to rise from July if a warm phase returns.
How bad is haze compared with WHO limits?
Central Kalimantan's two-month mean in 2023 was 136 µg/m³ (Graham et al., 2024), against WHO's 24-hour PM2.5 guideline of 15 (WHO, 2021). Singapore's west read 43 µg/m³ over 24 hours today (NEA, 2026).
- Do not trust stay-indoors advice alone. Rural homes in Central Kalimantan matched outdoor PM2.5 (Graham et al., 2024).
- Check a local reading before each long ride. Singapore's west read 43 µg/m³ today while the other regions read lower (NEA, 2026).
- Watch the El Niño signal from June. Burning accelerated around July in the bad years (Gaveau et al., 2026).
- Fit your filter in the quiet months. The Easi Breezi clip-on air purifier filters PM2.5 at your intake.
Sources
- Graham, A.M., Spracklen, D.V., McQuaid, J.B., Smith, T.E.L., Nurrahmawati, H., Ayona, D. and colleagues (2024). “Updated smoke exposure estimate for Indonesian peatland fires using a network of low-cost PM2.5 sensors and a regional air quality model.” GeoHealth. doi.org/10.1029/2024gh001125
- Grosvenor, M.J., Ardiyani, V., Wooster, M.J., Gillott, S., Green, D.C., Lestari, P., Suri, W. (2024). “Catastrophic impact of extreme 2019 Indonesian peatland fires on urban air quality and health.” Communications Earth & Environment. doi.org/10.1038/s43247-024-01813-w
- Hein, L., Spadaro, J.V., Ostro, B., Hammer, M., Sumarga, E., Salmayenti, R., Boer, R., Tata, H. and colleagues (2022). “The health impacts of Indonesian peatland fires.” Environmental Health. doi.org/10.1186/s12940-022-00872-w
- Kiely, L., Spracklen, D.V., Arnold, S.R., Papargyropoulou, E., Conibear, L., Wiedinmyer, C. and colleagues (2021). “Assessing costs of Indonesian fires and the benefits of restoring peatland.” Nature Communications. doi.org/10.1038/s41467-021-27353-x
- Gaveau, D.L.A., Descals, A., Salim, M.A. (2026). “Nationwide monthly burned area monitoring in Indonesia using Sentinel-2.” PLOS ONE. doi.org/10.1371/journal.pone.0331831
- World Health Organization (2021). “WHO global air quality guidelines: particulate matter (PM2.5 and PM10), ozone, nitrogen dioxide, sulfur dioxide and carbon monoxide.” who.int
- Reuters (2026). “Explainer: Why is the haze so bad in Southeast Asia this year?” 15 October. reuters.com
- BBC News (2026). “Borneo burning: Life on the frontline of South East Asia's toxic haze.” bbc.com
- National Environment Agency Singapore (2026). Real-time PSI readings, 1 November 2026, 8 am. api-open.data.gov.sg
Written by Ash, mechanical engineer and founder of Easi Breezi, building an active HEPA filtration system for motorcycle helmets (Patent Pending). Based between Hong Kong and Bali, riding daily in the traffic this blog writes about.