Maps · Geospatial

Indonesia’s forest, 2001–2025: where it went, and what came next.

Twenty-five years of tree-cover loss across the archipelago. Three islands carry the story; 2025 changes the tempo.

By Caleb Tutty Published Jul 5, 2026 12 min read Data · Hansen/UMD GFC v1.11, MapBiomas Indonesia, KLHK, WDPA

Fig. 01 · Tree-cover loss, 2001–2025 (Hansen/UMD GFC v1.13)

Loading…

Each polygon is one of Indonesia’s 34 provinces (GADM 4.1). Darker shades carry more cumulative loss. Author analysis · Hansen GFC v1.13

33Mha

Tree-cover loss across Indonesia, 2001–2025

91%

Concentrated on Sumatra, Kalimantan, Papua

2.4Mha

Single worst year, 2016 — El Niño fires

0.85×

2024–25 loss vs. the 25-year average

Between 2001 and 2025, Indonesia lost 33 million hectares of tree cover — an area larger than Italy.

Tree-cover loss is not the same thing as deforestation. The Hansen/UMD product, which we use here, registers any patch of trees taller than five metres that disappeared in a given year. That includes plantation harvest, replant rotations, and salvage logging after fire — alongside the irreversible conversion of natural forest that the word “deforestation” actually means.

Indonesia, 2001

38 provinces · geometry only

One polygon, one province

Thirty-eight provinces stretched across nearly five thousand kilometres of equator. North is up; the Java Sea sits roughly in the middle.

Twenty-five years of loss

Cumulative tree-cover loss, 2001–2025. Darker means more hectares gone. The hot colour band runs along the Sumatran spine, fans across Kalimantan, and brightens unevenly on Papua. Java barely registers — there isn’t much forest left to lose.

Three islands carry the story

Sumatra, Kalimantan and Papua together account for 91% of Indonesian tree-cover loss in the satellite record — Sumatra alone is 45%, Kalimantan another 40%. Each frontier has its own timeline: Sumatra peaked early and is now mostly cleared; Kalimantan ramped through the plantation decade; Papua is the youngest frontier with the most natural forest still standing.

Inside the lines

Coloured here by the share of each province’s loss that fell inside protected areas: IUCN-categorised parks and reserves plus the broader WDPA polygon set. Teal-heavy provinces are losing forest from places that are supposed to be off-limits.

What replaced the forest

Where each hectare of natural forest went, between 2001 and 2024, per MapBiomas Indonesia. Oil palm and bare ground absorb most of it; agriculture and plantation forest take the rest. (MapBiomas also records 4.4 Mha of the reverse direction — regrowth from bare to forest — which is a separate story this Sankey doesn’t show, and a big part of why the headline numbers from different sources never quite agree.)

2024–25 in context

Back to the map, coloured by just 2024 and 2025 — the last two years of the record. The top three provinces by recent loss carry a darker outline. The aggregate signal is actually 15% below the 25-year mean: the El Niño fire spikes of 2012 and 2016 dominate the long arc. But Auriga Nusantara’s natural-forest tracker reports a 66% increase in 2025 over 2024 — the conversion frontier and the harvest-and-replant churn have decoupled. The next iteration of this page will split them.

Tree-cover loss is not the same thing as deforestation.

The map below trades the province-aggregated story for the raw 30-metre pixels, every year of loss painted at its native resolution. Pale yellow is the start of the record (2001); hot magenta is the most recent year (2025). Pan to Sumatra’s eastern lowlands or central Kalimantan to see the plantation frontier in close-up.

Fig. 02 · Tree-cover loss, every pixel, 2001–2025 (Hansen/UMD GFC v1.13)

Loss year

2001 2025

Tiles: Global Forest Watch · Base: CARTO Positron · Pan and zoom

Why the numbers disagree

If you read three different reports of Indonesian forest loss in the same year, you will get three different numbers — sometimes by a factor of two. That isn’t carelessness. It’s that each source draws the line in a different place.

Source	What it measures	Why it differs
Hansen / UMD GFC	Tree cover taller than 5 m that disappeared, anywhere	Includes plantation harvest and tree-crop turnover
MapBiomas Indonesia	Annual land-cover class, pixel by pixel	Lets you distinguish natural forest from oil palm at source
KLHK (Ministry of Forestry)	Official “net” deforestation, with replanting subtracted	Defines forest by legal land-use category, not satellite signal
Auriga Nusantara	NGO estimate using GFW alerts + ground checks	Often higher than KLHK; published earlier

The page above is built on Hansen, MapBiomas for the “what replaced it” flows in step 6, and WDPA + Wetlands International for the protected-area and peat overlays. KLHK and Auriga numbers are reproduced in the table because they are the ones quoted in the press and the parliament.

Notebooks and data

Everything on this page is built from public, mostly auth-free data. The notebooks and the JSON shapes the components read are in github.com/tuttinator/indonesia-deforestation.

Notebook	Source	What it does
`00_setup_and_geometries.py`	GADM 4.1 + Natural Earth	Province + district polygons, simplified TopoJSON
`01_gfw_treecover_loss.py`	Hansen GFC v1.11 via GEE	Annual loss zonal-stats per province + top-200 district
`02_overlays_pa_peat.py`	WDPA + Wetlands International peat	Loss inside protected areas; loss on peat
`03_mapbiomas_transitions.py`	MapBiomas Indonesia Coll. 2	Forest → sink class transitions 2001–2023
`04_export_blog_json.py`	parquet → JSON	Schema-validated outputs synced into `static/data/deforestation/`

Heavy downloads cache to data/raw/. The full pipeline is uv sync && make notebooks && make sync-blog.