Microproduct data guideintermediate

Global Forest Watch Tree Cover Loss

Annual tree-cover loss statistics for building deforestation monitors and country comparison tools from satellite-derived forest change.

Start the guide Open official source

At a glance

Difficulty
Intermediate — some data preparation helps
Size
Tiny · ≤0.05 GB
Formats
JSON
Access
API
API key
Not required
Provider
World Resources Institute
Updates
Annual
Data terms
Creative Commons Attribution 4.0 International

From source to product signal

Test a product idea in four steps

Global Forest Watch exposes UMD/Hansen tree-cover loss through a SQL query API. Start with one year and the top countries by hectares lost. Loss is a stand-replacement disturbance, not a land-use class, and it includes fire, harvest, and conversion.

1

Check the setup

  • Python 3.10 or newer
  • A notebook environment such as Jupyter or Google Colab
  • An internet connection
2

Access the data

  1. 1.Open the GFW Data API documentation and locate the umd_tree_cover_loss dataset.
  2. 2.Query one loss year aggregated by ISO3 code with a small row limit.
  3. 3.Cite Hansen/UMD/Google/USGS/NASA and WRI when presenting results.
Open official source
3

Run the Python example

Install the packages, then run the notebook cell.

python -m pip install pandas requests

import pandas as pd
import requests

response = requests.get(
    "https://data-api.globalforestwatch.org/dataset/umd_tree_cover_loss/latest/query",
    params={
        "sql": (
            "SELECT iso, SUM(umd_tree_cover_loss__ha) AS loss_ha "
            "FROM data WHERE umd_tree_cover_loss__year = 2023 "
            "GROUP BY iso ORDER BY loss_ha DESC LIMIT 15"
        )
    },
    timeout=60,
)
response.raise_for_status()
loss = pd.DataFrame(response.json()["data"])
loss["retrieved_at_utc"] = pd.Timestamp.now(tz="UTC")
print(loss)
4

Test a useful signal

Rank countries by 2023 tree-cover loss

Test whether GFW country totals can power a bounded deforestation leaderboard.

  1. 01Confirm the query returned ISO3 codes and hectare totals for one loss year.
  2. 02Rank the bounded result and note countries with missing or zero values.
  3. 03Explain that tree-cover loss is not the same as deforestation of natural forest and includes harvest and fire.

Dataset details

World Resources Institute is a nonprofit source. Last verified 2026-08-18. Temporal coverage: 2001-present annual loss years.

Theme

Environment & Hazards

Domains

ClimateForestryEarth Observation

Data types

TabularGeospatialTime Series

Tasks

Deforestation MonitoringCountry ComparisonTrend Analysis

Geography

Global

Formats

JSON

Provider

World Resources Institute

Data terms

Creative Commons Attribution 4.0 International