Land Cover Map of Northern Madagascar

This map focuses on northern Madagascar using Google Earth Engine & Sentinel-2 imagery for land cover classification. The main aims of this work are to produce a current land cover map of northern Madagascar with a high spatial resolution and, at the same time, to make the generation scripts freely accessible to all interested parties, enabling easy updates, and improvement. Four classes were defined: forest, soil, non-forest vegetation, water. Spectral indices & covariates enhance classification. Random Forest algorithm employed with sample-based evaluation. The 10 meters land cover classification achieved an overall accuracy of 96% and a precision of approximately 93%. While broadly aligning with field observations (1994-2023), the model lacks the precision to distinguish various forest types due to insufficient field data. Further improvements could involve statistical optimization, exploration of RADAR/LIDAR data, hyperspectral imaging, and unsupervised classification.

Organizational unit

enviroSPACE

Type

Dataset

DOI

10.26037/yareta:2wxp7yonczaepoewhmvucn33my

License

Creative Commons Attribution 4.0 International

Keywords

land cover, madagascar, sentinel-2, random forest, forest, vegetation