Atlas · Kenya / Methodologies
With data in Atlas
Angolafull · 18 datasets Beninfull · 14 datasets Cameroonfull · 47 datasets Comorosfull · 16 datasets Côte d'Ivoirefull · 14 datasets Democratic Republic of the Congofull · 44 datasets Djiboutifull · 15 datasets Egyptfull · 18 datasets Equatorial Guineafull · 13 datasets Eritreafull · 15 datasets Gabonfull · 28 datasets Ghanafull · 16 datasets Guineafull · 15 datasets Guinea-Bissaufull · 14 datasets Kenyafull · 344 datasets Liberiafull · 13 datasets Madagascarfull · 18 datasets Mauritaniafull · 16 datasets Mauritiusfull · 9 datasets Mozambiquefull · 35 datasets Nigeriafull · 16 datasets Republic of the Congofull · 41 datasets Senegalfull · 15 datasets Seychellesfull · 9 datasets Sierra Leonefull · 14 datasets Somaliafull · 13 datasets South Africafull · 26 datasets Sudanfull · 16 datasets São Tomé and Principefull · 14 datasets Tanzaniafull · 170 datasets The Gambiafull · 14 datasets Togofull · 15 datasets Algeriareference · 5 datasets Botswanareference · 32 datasets Burkina Fasoreference · 2 datasets Burundireference · 2 datasets Central African Republicreference · 2 datasets Chadreference · 2 datasets Ethiopiareference · 2 datasets Kingdom of eSwatinireference · 1 datasets Lesothoreference · 2 datasets Libyareference · 3 datasets Malawireference · 2 datasets Malireference · 2 datasets Moroccoreference · 6 datasets Namibiareference · 56 datasets Nigerreference · 1 datasets Republic of Cabo Verdereference · 1 datasets Rwandareference · 7 datasets South Sudanreference · 1 datasets Tunisiareference · 3 datasets Ugandareference · 29 datasets Western Saharareference · 1 datasets Zambiareference · 54 datasets Zimbabwereference · 31 datasets
Compare all countries
METHODOLOGIES

How Global Mangrove Watch maps are made

Satellite pixels are classified as mangrove, cleaned, and compared across years to find change.

All methodologies Ask the analyst about this
Advanced 2 min read Global Mangrove Watchremote sensingmethodaccuracy

Atlas does not make these maps; it stores, measures and explains them. Understanding how they are made tells you how far to trust them.

Steps (simplified)

  1. Baseline extent. A reference map of where mangrove is, built by classifying satellite imagery with training samples and a mask of areas where mangrove can occur (coastal, tropical, low-lying).
  2. Change detection. Each year's imagery is compared with the baseline; pixels that persistently change are flagged as loss or gain.
  3. Time series. Yearly maps are produced by applying the changes to the baseline.
  4. Vectorisation. Mangrove pixels are joined into polygons; v4.1 adds a country label to each polygon.
  5. Validation. The authors check the maps against independent reference samples and publish accuracy results in the scientific papers. Read the paper before citing precision.

What it means for you. Errors are most likely: at forest edges; in narrow creeks; for small patches; where mangrove is mixed with other vegetation; and in cloudy regions for optical years.

In Atlas every polygon is stored with its source id, its area is recalculated on the ellipsoid, and the original archive file is kept with a checksum.

Limits. Atlas does not repeat the authors' accuracy assessment; it reports their maps as published.

Related reading

Global Mangrove Watch: version 3 and version 4 explained

Version 3 covers 1996 and 2007–2020 from radar; version 4.1 extends the series to 2025 using a 10 m baseline.

Why Atlas totals are 2.3% higher than the official figures

Atlas measures polygons on the ellipsoid; the authors publish a "corrected area". The gap is a constant 2.3% in every country.

How satellites see mangroves

Radar sees through cloud and optical sensors see colour; maps combine both.

More in Methodologies

Still have a question?

The AI analyst answers from these articles and from the data.

Ask Atlas
Guest preview · 170s left Explore until the timer ends. Then you’ll be asked to log in or create an account (about two minutes) to keep going.
Log in Create an account
Ask Atlas