How tree-cover loss is measured

The forest counter is fast. The science underneath is annual, pixel by pixel.

The question this tool answers

This dashboard asks a simple question: at the latest annual-average pace, how quickly is the world losing tree cover? The answer is shown as football fields because hectares are abstract. The underlying number is not a live sensor. It comes from annual satellite analysis by the University of Maryland's Global Land Analysis and Discovery lab.

The Hansen/UMD record

The dataset most people know through Global Forest Watch began with work led by Matthew Hansen at the University of Maryland, with long-time collaborators including Peter Potapov. The team used Landsat imagery to map global forest cover, gain, and loss at about 30-metre resolution. The widely used annual series starts in 2001, the period when Landsat 7's global observations made consistent wall-to-wall monitoring possible.

A pixel is marked as tree-cover loss when the satellite record shows a stand-replacement disturbance: canopy that was there, then was not. That can be a clear-cut, fire, plantation harvest, storm damage, mining, flooding, or conversion to agriculture. It is a tree-cover-loss dataset, not a moral label and not always “deforestation” in the narrow policy sense.

Global Forest Watch

World Resources Institute turned the UMD data into Global Forest Watch, the public platform used by journalists, researchers, companies, and governments. GFW also hosts alert products such as GLAD and RADD that can flag recent forest disturbance. This dashboard does not use those alert APIs. It downloads the annual country series mirrored by Our World in Data and presents a fast, readable front door.

Why “real-time” needs a caveat

The page refreshes daily, but the annual Hansen layer changes once a year. The ticking counter uses the latest annual global total divided into days, minutes, and seconds. It is useful because it makes scale tangible during a reading session. It should not be read as a minute-by-minute tally from satellites passing overhead.

Drivers are global systems

The driver bar uses Curtis et al. 2018 and WRI synthesis work to name broad causes of tree-cover loss: commodity agriculture, forestry, shifting agriculture, wildfire, and urbanization. The better shorthand is geographic and supply-chain specific: cattle pasture and soy expansion in the Amazon, palm oil plantations in Southeast Asia, boreal megafires and logging in Canada and Russia.

Country-of-cutting is not country-of-blame. Deforestation drivers often include multinational commodity chains, overseas consumption, finance, infrastructure, weak enforcement, and climate-amplified fires. The country leaderboard is a measurement lens, not an accusation index.

FAO numbers differ on purpose

The table includes a FAO net forest-area-change column for context. FAO Forest Resources Assessment data are country-reported and use land-use definitions across multi-year intervals. Hansen/UMD is satellite-observed gross tree-cover loss. A plantation harvest can appear as tree-cover loss in Hansen while a country reports stable forest area if trees are replanted. Both views are useful; they answer different questions.

Credits

This dashboard depends on the University of Maryland GLAD team, the Landsat programme, WRI Global Forest Watch, Our World in Data, FAO forest reporting, and the researchers who translate pixels into public evidence. It is a public-facing summary, not a replacement for GFW analysis tools or local land-rights investigations.

← Back to the dashboard