KM-GBF Global Review
From October 19–30, 2026, delegates from nearly 200 countries gathered in Yerevan, Armenia, for COP17 of the Convention on Biological Diversity (CBD). The convened Parties will review collective progress in the implementation of the Kunming-Montreal Global Biodiversity Framework (KM-GBF) a joint agreement adopted in 2022 that sets a global target to halt and reverse biodiversity loss by 2030. The primary basis for this review is the Seventh National Report (NR7), submitted by each country to the CBD Secretariat prior to the COP. Indonesia has also submitted its NR7. However, the true report card lies not in that document, but in the reality on the ground. Since early April 2026, BMKG had warned that this year's dry season would be drier and longer than usual, with the likelihood of El Niño emerging in the second half of the year. A drier dry season translates to a heightened risk of forest and land fires (karhutla). That warning proved accurate. The government had roughly four months to prepare, yet the spike in fires still materialized at the height of the dry season. This was not a sudden disaster, but the result of neglect.
Indicative Burned Area (AIT) data compiled by Madani indicates that approximately 2.5 million hectares of forest and land were burned between January and September 2026. The vast majority of those fires occurred within the last three months. In July, AIT recorded 101 thousand hectares. In August, that figure climbed to 788 thousand hectares. In September, it reached 1.56 million hectares. Within two months, the monthly indicative burned area surged fifteenfold.
Source: MADANI Berkelanjutan, January–September 2026 AIT Update
That total figure is important, but it is not the crux of this piece. What matters more is its location. When Madani overlays the AIT onto policy maps, the emerging pattern is troubling. Fires do not choose empty land. Instead, fires burn places pledged by the government for climate action, places legally protected, and places that serve as home to Indonesia's biodiversity.
Nearly 44% of the AIT (1.11 million hectares) falls within the FOLU Net Sink Operational Plan area. Almost a third (800 thousand hectares) lies within the moratorium area on new permits. More than half a million hectares are situated in natural forest cover. And 157 thousand hectares are within Key Biodiversity Areas locations identified by the global scientific community as crucial to the survival of the world's biodiversity.
This article attempts to examine the available data more closely and pose a fundamental question that Indonesia should answer in Yerevan.
Reading the AIT Modeling Figures
In any writing on forest and land fires, it is necessary to emphasize that the AIT is not an official fire extent figure. AIT is a model developed by Madani since 2019 as an early observation system. The model processes satellite imagery and updates estimated indicative burned areas up to three days before official reports are released via SiPongi (D-3). Its purpose is to provide an early snapshot, enabling the government, companies, and the public to take action before official numbers are published.
Because it is indicative in nature, the AIT needs to be tested. Madani compared the provincial distribution of AIT with the Ministry of Forestry's SiPongi data for the January–July 2026 period. The correlation coefficient between the two reached 90.02%, with an R² of 81.04%. Madani also cross-referenced model outputs with Sentinel-2 imagery to confirm the presence of smoke plumes and burn scars at locations flagged by the AIT.
The high correlation between AIT and SiPongi demonstrates that fires occurred in the same provinces with similar patterns. Correlation does not prove the magnitudes are identical. For January–July, SiPongi recorded 202 thousand hectares, whereas AIT recorded 229 thousand hectares. Through July, this discrepancy appears reasonable. The issue arises in subsequent months, where SiPongi's burned area data stops at July. As of this writing, AIT lacks official comparative data from SiPongi because available figures only extend through July even though a very substantial surge took place in August and September.
This means that of the approximately 2.35 million hectares identified by AIT in August–September, the figures are purely modeled estimates. Official figures released later may be significantly smaller. It is also possible that the official figures will closely match AIT, and their non-release to date may stem from various reasons, potentially political. Whatever the case, as an environmental organization, Madani has a moral obligation to disclose its analysis transparently.
Therefore, the analysis in this piece does not rely on total numbers. Instead, it relies on proportions and locations. The question is: of all the areas flagged by AIT, what percentage falls within specific designated areas? A proportion-based analysis is more resilient to correction than aggregate figures. If the model is wrong in reading one land type more often than another, the proportions will shift as well. Even so, field validation remains essential.
One further technical note: the spatial categories in this piece overlap. A single hectare can simultaneously lie within peatland, a moratorium area, and an oil palm concession. The total AIT remains a discrete figure of around 2.5 million hectares because each hectare is counted only once. However, one cannot aggregate percentages across categories, as the same hectare may appear in multiple categories at once.
Indonesia's Climate Commitments Burned Away
Indonesia has made the forestry and other land use (FOLU) sector the primary anchor of its climate commitments. Through Indonesia's FOLU Net Sink 2030, the government targets this sector to sequester more carbon than it emits, aiming for a net emission target of minus 140 million metric tons of CO₂e by 2030. To achieve this, the government mapped out an Operational Plan (RO) areas where the government has committed to preventing deforestation, restoring peatlands, rehabilitating forests, and preserving carbon stocks.
AIT data indicates that 1.11 million hectares, or 43.77% of the total AIT between January and September 2026, falls within the FOLU Net Sink RO area. The surge mirrors the national trajectory, rising from 19.6 thousand hectares in July to 312 thousand hectares in August, and reaching 740 thousand hectares in September.
When the government designates areas for the RO based on their carbon value, and when 44% of the burned area falls within them, the fire does not merely consume vegetation it also consumes the projected achievement of the climate targets themselves.
Peatlands exacerbate the problem. AIT recorded 995 thousand hectares within peatland ecosystem functions, or 39.19% of the total. Of that amount, 555 thousand hectares are located in protected peatland ecosystems (fungsi lindung), areas that peatland protection regulations mandate to be kept wet and intact. In September alone, AIT in protected peatlands reached 395 thousand hectares, climbing from 9.5 thousand hectares in July. Burned peat releases carbon that has been stored for thousands of years. Peat fires can also smolder underground for weeks, far beyond the reach of firefighters.
Then there are the rehabilitation areas. The General Plan for Forest and Land Rehabilitation (RURHL) maps lands the state intends to restore. AIT shows that 349 thousand hectares of RURHL areas were indicatively burned, with 256 thousand hectares occurring in September. This presents an inexplicable paradox: land slated for rehabilitation burns before restoration can take place, or burns once again after rehabilitation has commenced. AIT data cannot differentiate between the two, but both outcomes are equally detrimental.
Given this data, the question for the government is no longer whether the FOLU Net Sink targets can be achieved. Rather, the question is: have the FOLU Net Sink carbon sequestration calculations accounted for fire risks of this magnitude? If not, the government's targets rest upon figures detached from realities on the ground.
The Last Ecosystems Consumed by Fire
Two KM-GBF targets are most directly relevant to this data. Target 2 calls for 30% of degraded ecosystems to be under effective restoration by 2030. Target 3 calls for 30% of the world's terrestrial and inland water areas to be effectively conserved by 2030, especially areas of particular importance for biodiversity. The keyword is “effective.” Areas recorded as protected on paper, yet burning every dry season, do not live up to that word.
AIT data reveals fires occurring across multiple layers of high-value ecosystems simultaneously:
Source: MADANI Berkelanjutan, January–September 2026 AIT Update
Because these layers of protection status overlap, the total burned area across these zones cannot be obtained simply by summing the five figures above. A direct summation yields 1.74 million hectares, but that figure double-counts the same areas for instance, natural forests on protected peatlands that also fall within a KBA. The actual extent lies somewhere between 555 thousand hectares (the largest single layer, protected peatlands) and 1.74 million hectares. The exact figure requires an inter-layer spatial overlay analysis.
This situation illustrates that this year's fires have not only scorched land, but also eroded the very foundation of Indonesia's biodiversity. The five layers above represent the ecosystems most difficult to replace. Peatlands, natural forests, and conservation areas serve as home to numerous species found nowhere else. When fire encroaches, these habitats shrink, become fragmented into small patches, and lose their capacity to sustain wildlife and plant populations.
This condition also threatens the achievement of the KM-GBF. Target 1 calls for bringing the loss of areas of high biodiversity importance close to zero by 2030; fires in natural forests and KBAs move in the exact opposite direction. Target 2 and Target 3 demand effective restoration and protection, yet AIT data reveals fires encroaching directly into rehabilitation areas and protected zones.
Indonesia has translated these targets into the Indonesian Biodiversity Strategy and Action Plan (IBSAP) 2025–2045. If fires of this magnitude recur, the IBSAP targets will be exceedingly difficult to achieve not because the plan itself is flawed, but because the targeted ecosystems are steadily shrinking.
Nor does this destruction end once the fires are extinguished. Burned land becomes more susceptible to repeat burning and is more prone to land-use conversion. Without rigorous oversight, areas burned this year could become permanently lost in the years to come.
KBAs are not merely green patches on a map. KBAs are locations identified by the global conservation community through shared scientific standards because they determine the survival of biodiversity globally. Fire in a KBA requires no further interpretation: fire there directly threatens the very reason the site was listed in the first place.
At first glance, a 6.17% share may appear small. But that proportion measures the share of KBAs within the total AIT, not the extent of damage to the KBAs themselves. A more accurate metric is the percentage of Indonesia's total KBA area that has burned.
What is more alarming is the velocity. The AIT in confirmed KBAs recorded 11.1 thousand hectares in July, 35 thousand hectares in August, and 96 thousand hectares in September. Within two months, the figure surged nearly ninefold. If this pattern of national surges continues or recurs next year, KBAs will not remain safe simply because their current share appears small today.
Natural Forests: Damage That Does Not Heal in a Single Season
Natural forest is the layer people understand most intuitively. In January, the AIT in natural forest cover was only 2.8 thousand hectares. In August, it reached 108 thousand hectares. In September, 377 thousand hectares.
Indonesia's tropical rainforests did not evolve alongside fire. Unlike savannas, which are adapted to periodic burning, rainforest trees are generally thin-barked. A single fire can kill many large canopy trees and alter forest structure for decades. Forests that have previously burned are also drier and far more susceptible to reburning. It is this feedback cycle that makes recurrent fires so destructive: each successive round cuts deeper into what remains.
That is why figures for natural forests must not be read as “land that will naturally regenerate.” For biodiversity, the 505 thousand hectares of natural forest indicatively burned from January through September represents habitat that is shrinking, fragmented, and fundamentally transformed.
Distribution of Fires Through September
Check out and download the infographic here
Our analysis shows that the largest AIT extents from January through September were in Central Kalimantan (658 thousand hectares), West Kalimantan (463 thousand hectares), and South Papua (296 thousand hectares). At the regency level, Ketapang led with 269 thousand hectares, followed by Merauke (234 thousand hectares) and Pulang Pisau (149 thousand hectares). These areas harbor Kalimantan's peat swamp forests as well as the wetlands and savannas of southern Papua ecosystems with no equivalent anywhere else. The single most heavily burned KBA during this period was in the Kapuas coastal swamps, with burned area reaching 35 thousand hectares, while the most severely affected conservation areas were National Parks across several provinces, where AIT reached 140 thousand hectares.
This distribution demonstrates an unusual spatial scale, to the extent that even Papua was not spared from this season's fires. This represents an active disruption not merely a potential threat to ecosystems that will directly impact the state of Indonesia's biodiversity. Under these circumstances, how will Indonesia uphold its 30x30 target?
Who Holds the Land That Burned
Nearly half of the AIT, specifically 48.12% or 1.22 million hectares, is located within permit and concession areas. The breakdown is as follows:
Permit Type | Jan–Sep 2026 AIT | Share of Concession AIT | July – September |
Oil palm plantations | 385 thousand ha | 31.48% | 10 thousand → 206 thousand ha |
PBPH (forest utilization) | 300 thousand ha | 24.58% | 8.4 thousand → 200 thousand ha |
Overlapping permits | 283 thousand ha | 23.17% | 5.6 thousand → 162 thousand ha |
Oil and gas | 172 thousand ha | 14.11% | 7 thousand → 108 thousand ha |
Mineral and coal mining | 81 thousand ha | 6.65% | 2.4 thousand → 52 thousand ha |
This pattern aligns with state forest area data. Of the 1.75 million hectares of AIT within state forest areas, 1.22 million hectares are situated in production forests areas where management has largely been delegated by the state to permit holders.
Indonesian regulations place the obligation for fire prevention and control squarely on permit holders within their concession areas. Therefore, when fires occur within concession areas, the question is not merely “who started them.” It must also answer “who failed to prevent and suppress them.” The AIT analysis is not aimed at scapegoating companies that experience fires within their concessions, as fire ignitions can stem from various sources. Its primary purpose is to urge the government to rigorously investigate and examine what failed to be done that allowed the 2026 fires to have such a profound impact on communities.
The figure for overlapping permits warrants special attention. Across 283 thousand hectares, more than one permit is active on the same land. When fire strikes, who is obligated to extinguish it? Who bears the sanctions? Overlapping permits obscure accountability, and this ambiguity creates room for all parties to deflect blame.
The State Is “Absent” in Fires Outside Concessions
Spanning 1.32 million hectares, or 51.88% of fires identified by AIT, fires occurred outside permit areas. On these lands, management responsibility including for fires that break out rests entirely with the state and cannot be shifted to other parties.
Fires outside concessions are not ownerless fires. These lands comprise several categories: conservation areas managed directly by National Park Offices and Natural Resources Conservation Centers (BKSDA) under the Ministry of Forestry; unpermitted protection forests and production forests under the purview of Forest Management Units (KPH) and provincial forestry agencies; and other utilization areas (APL) without business permits, which fall under the authority of regional governments. Across every level, there is a state institution with a clear mandate.
AIT data demonstrates that this mandate has not been fulfilled. Conservation forests, managed directly by the state, recorded 306 thousand hectares of AIT, with 231 thousand hectares occurring in September alone. A sharp rise was already visible in July, yet the AIT extent continued to climb through September. This is where the term “absent” applies: the state is present on paper as the manager, but absent when the fires strike.
When fires break out in concessions, the public can hold corporations accountable. When fires break out outside concessions, there is no other entity to point to but state institutions themselves. Therefore, data transparency becomes a prerequisite for accountability. Without official SiPongi data for August and September, the public cannot assess the performance of the institutions mandated to prevent and extinguish fires on state land.
Beyond the various permit layers and protected zones outlined earlier, there is one more layer that requires close scrutiny: food and energy reserve areas. AIT in these areas reached 446.5 thousand hectares (17.58%). A summation of the monthly tables indicates that approximately 178 thousand hectares are located in unpermitted corridors, 168 thousand hectares in PBPH corridors, and 100 thousand hectares in PIAPS corridors. The PIAPS corridor consists of lands allocated by the state for social forestry; thus, fires there directly impact the interests of communities living around forests.
AIT data cannot determine why food and energy reserve areas burned. Fires could stem from severe drought, land-clearing practices, or both. However, when large-scale fires occur on lands being prepared by the state for major national strategic projects, ground-level verification becomes an absolute imperative.
What Indonesia Must Do
AIT data does not provide all the answers, but it offers ample justification to demand six key actions:
Honest biodiversity reporting. Every progress report on KM-GBF Target 2 and Target 3 whether presented at COP forums or in subsequent national reports must account for fires occurring within conservation areas, natural forests, protected peatlands, and KBAs. The extent of protected areas on paper means nothing if those areas are consumed by fire.
Prompt and transparent official data. The Ministry of Forestry must immediately update SiPongi data for August and September, accompanied by spatial distribution maps. The public has a right to compare official figures with independent estimates.
Audits of burned concessions. The government must audit oil palm, PBPH, oil and gas, and mining permit holders with the largest AIT footprints and publicly disclose the findings. Concession areas with overlapping permits require definitive clarity regarding who bears legal responsibility.
Recalculate FOLU Net Sink and RURHL. The government must incorporate fire risk into carbon sequestration calculations and establish fire prevention as a core prerequisite across Operational Plan (RO) and rehabilitation zones.
Enforce comprehensive peatland ecosystem protection. Strengthen safeguards across all peatlands, both conservation and cultivation zones, and immediately prioritize burned peat areas for restoration under the Peatland and Mangrove Management Agency.
Utilize early warning systems. The fire surge was already evident in July. Data like AIT and daily hotspots exist to trigger early action, not to be read after the fires have already died out.
Conclusion
Armenia chose the theme “Taking Action for Nature” for COP17. That theme demands concrete action, not policy documents. Indonesia already possesses an array of sound frameworks: IBSAP, FOLU Net Sink targets, permit moratoria, peat protection regulations, rehabilitation plans, and national biodiversity strategies. The 2026 AIT data serves as a filter testing whether those documents actually work on the ground.
That test is far from over, and the final figures may shift once official data is released. But one thing is already clear: this year's fires did not halt at the boundaries of protected areas, nor did they stop at the perimeter of climate pledges and did not stop at the boundaries of KBAs. Warnings had already arrived as early as April. The deliberations in Yerevan, and every report thereafter, without a chapter on fire, represent an incomplete report card.
Sources
MADANI Berkelanjutan, Indicative Burned Area (AIT) Update January–September 2026.
Convention on Biological Diversity, COP17, Yerevan, October 19–30, 2026.
BMKG, Development of the Dry Season and El Niño Predictions for 2026: Drier and Longer Dry Season, April 5, 2026 (https://www.bmkg.go.id/berita/perkembangan-musim-kemarau-dan-prediksi-el-nino-tahun-2026-kemarau-lebih-kering-dan-panjang-bmkg-tekankan-pentingnya-antisipasi).
