Measuring Indonesia’s 2026 fire season while it burns — where the fire is, what it is costing, and who is carrying the bill.
The 2026 season is being recorded as it happens, from satellite detections through health costs to the modeled hit to output, and published in short intervals rather than held for a year-end report.
Fire in Indonesia is a recurring economic event with a predictable trigger. A strong El Niño dries peat that was drained for cultivation, and land that is cleared with fire in a dry year does not stop at the boundary of the plot. The damage then travels: haze closes schools and airports, respiratory admissions rise across the region, harvests fall, and the fiscal cost of response and reconstruction competes with planned spending.
What is usually missing during a season is the number. Hotspot feeds report detections without attribution; health effects are estimated long after the smoke clears; and the economic cost is settled in retrospect, when the decisions it should have informed have already been taken. This initiative closes that gap by publishing the measurement continuously — the count, where it falls, what it costs, and what the modeling says happens next.
Four instruments, each answering a different question, and each published with its method attached.
Detection. NASA FIRMS hotspots from VIIRS and MODIS, filtered to high confidence, counted daily and by province, with volcanic thermal anomalies removed. Each point is counted once even where designations overlap.
Exposure. The same detections overlaid on the protected area estate from the WDPA extract, so the share of fire burning inside the conservation estate — and which specific sites carry it — is visible rather than inferred. Density is computed against the protected area itself, not the province.
Health and fiscal cost. The population exposed to elevated particulate concentrations, the resulting respiratory burden, and the public cost of treating it. The August 2026 fact sheet put the health burden at Rp120 trillion and prompted the call for systemic rather than seasonal response.
Economy-wide impact. ENSOID26-V2, a calibrated system-dynamics model of how a super El Niño propagates through the Indonesian economy — the finding being that the indirect wave through supply chains, prices and public budgets costs more than the burned area itself. ENSONINAID26 extends it to the La Nina that frequently follows, and finds that damage in the dry year raises damage in the wet year.
The 2026 season escalated sharply in the second half of August and has not let up. SiPongi recorded 1,224,060 hotspots from 1 January to 17 October, and fire emissions reached 673 Mt CO2e by 15 October, two thirds of it from burning peat and 74 percent of 2015 at the same date.
The cost is now visible well beyond the burning provinces. Smoke has reached 17 provinces, including Jakarta, Banten and West Java; 2,458 domestic flights were disrupted between 11 August and 17 October; and respiratory infections in seven provinces passed 300,000. On 19 October high-confidence hotspots in Central Kalimantan jumped from 157 to 937, and fire has reached Mount Lawu and Mount Sumbing on Java. BMKG expects hot, dry weather to continue to 26 October, and the El Niño is near its peak.
The pattern the season exposed is not the one policy is built around.
In the protected-area analysis, between 13 August and 5 September, satellites recorded 31,161 high-confidence hotspots nationally. Of those, 2,067 fell inside a protected area — seven in every hundred — so most of the season’s fire is burning outside the conservation estate. But inside the estate, ten sites carry 70.4 percent of everything recorded, and legal category tracks none of it: the strictest IUCN class holds more points across more sites than the national parks do.
Two provinces hold 62 percent of the hotspots recorded inside protected areas. Papua has the volume, spread across an enormous estate; Kalimantan Tengah has the density, four times as many points per unit of protected land. The five Kalimantan provinces together hold more than half the national figure.
That combination — concentration in a small number of sites, and density rather than count as the operative measure — is what makes targeted intervention possible. Protection on paper and protection on the ground are separate things, and the difference is now measurable while the season is still running.
A calibrated model of how an El Niño event travels through the Indonesian economy: from Pacific sea-surface temperature anomalies to rainfall, harvests, fire and haze, and on to food prices, health costs, labor productivity and the public budget.
It is the engine behind Fire on the Horizon. Every assumption and transmission channel is exposed as an adjustable parameter, so the run can be re-tested rather than taken on trust.
A causal-loop stress test that runs both phases of the cycle — El Niño drought and La Nina rainfall surplus, including multi-year events and cyclone activity — through direct damage, the indirect second wave, and the macroeconomic reckoning.
It adds the policy side: food-import buffers, fire suppression and peat condition, fiscal relief, monetary tightening, and adaptation portfolios weighed against annualized avoided losses.
Built on the CST26 engine, it runs an El Niño year followed by a La Nina year on a dated quarterly calendar from 2026 to 2029, anchored to the observed Oceanic Niño Index, and links the dry chain of drought, fire and haze to the wet chain of extreme rain, cyclones, floods and landslides through nine cross-phase channels.
It is the engine behind Floods After Fires, and adds an adaptation layer that weighs early warning, peatland rewetting and other measures against the damage they avoid.
The official national record: hotspot counts by province and satellite, area burned by year and district, and daily situation maps. It is the number every other figure is measured against, and the source of the 202,004 hectares reported for January to July 2026.
Active fire detections from the VIIRS and MODIS instruments, updated within three hours of satellite overpass. Detections carry a confidence value and can be filtered by date and drawn over protected area and concession boundaries — the layer behind most independent hotspot counts.
Full-resolution daily satellite imagery with smoke and aerosol layers, so a haze plume can be traced from its source fires across the Strait of Malaka and the South China Sea. Aerosol optical depth and the aerosol index show the drift that reaches Sarawak, Singapore and Manila.
Published image analyses of individual fire episodes, including the peat fires of the 2026 season. Each piece pairs the imagery with an explanation of what the sensor is seeing — useful where a raw hotspot map needs interpretation.
Daily fire radiative power converted into carbon and particulate emission estimates, with forecast smoke transport. The source of the comparison between the 2026 season and 2015 by tonnes of carbon released.
Independent mapping of deforestation and burn scars across Indonesia, attributed to concessions and company groups. Burn scars are mapped from post-fire imagery rather than from heat detections, so the area figures are typically higher and more reliable than hotspot-derived estimates.
Fire alerts layered over tree cover loss, peatland extent, protected areas and concession boundaries in one map, with downloadable statistics by administrative unit. The straightforward tool for asking whether a given fire sits inside a license or a conservation area.
Fire inside 29 of 39 Verra carbon projects released an estimated 16.7 Mt CO2e in 2026, 45 percent of a year of their expected credits, against buffers that cover about a quarter.
Download →
The 2026 El Niño near its peak: a record weekly Niño 3.4 of 3.3°C, fires that may run into November, and a 13 percent CPC chance of La Nina by mid-2027.
Download →
Hotspots, burned area and fire emissions since January: 1,224,060 hotspots to October 17 and 673 Mt CO2e to October 15, two thirds from peat and 74 percent of 2015 at the same date.
Download →
Indonesia’s conservation estate in the 2026 season: 2,067 high-confidence hotspots inside the estate, ten sites carrying 70 percent of them.
Download →
Indonesia, the health price of the 2026 fires.
Download →
Indonesia, January 1 to August 26.
Download →
Kalimantan, August 13 to 22.
Download →
The 2026 fire season, week of August 13 to 20.
Download →
The modeled impacts of the 2026 super El Niño and the probable super La Nina that follows on the Indonesian economy, checked against the events of February to mid-October 2026.
Read →
The modeled impacts of the 2026 super El Niño on the Indonesian economy — why the indirect wave costs more than the scorched fields. Includes the ENSOID26-V2 system dynamics model.
Read →
Increasing resilience through nature-based recovery in Aceh, North Sumatera, and West Sumatera.
Read →Press coverage of the fire findings, including the Rp120 trillion health burden and the call for systemic response.
Where the peat, fire management and ecosystem service work sits, and the projects it runs.
The land-use pressure behind the clearing: food security, biodiesel feedstock and the FOLU Net Sink target on one map.