Atlas · Rwanda / 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

The rules behind recommendations

Each recommendation comes from a stated rule with evidence numbers, a calculation note and a priority.

All methodologies Ask the analyst about this
Advanced 1 min read recommendationsrulesintelligencepriority

Atlas generates recommendations for every country with a mangrove map. Each is produced by a transparent rule.

Examples of rules

  • Net change. Net loss of ≥5% since 1996 is high; any loss is medium; gain is low.
  • Fragmentation. Many tiny patches with most area in a few large stands suggests focusing on the large stands and checking the small-patch tail.
  • Pressure. More than 30% of mangrove within 10 km of a major development is high (Lamu).
  • Validation. Low agreement between field plots and the map is high.
  • Governance. Datasets with no stated licence raise a licence recommendation.

Each recommendation stores a category, priority, confidence, evidence (the numbers that triggered it), the calculation and a suggested action. They are regenerated whenever the data changes.

Honest limits. The rules encode judgement about what is worth flagging. They are transparent, but not exhaustive.

Limits. Rules are tuned for mangrove screening, not for legal or financial decisions.

Related reading

Confidence labels: calculated, illustrative, assumed

Every number is tagged with how it was produced so you know how far to trust it.

How recommendations and briefs are generated

A deterministic rules engine, not a language model.

More in Methodologies

Still have a question?

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

Ask Atlas
Guest preview · 169s 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