Dataset for 'Advancing Mediterranean Biodiversity Monitoring in South Africa through Machine Learning and Cost-Effective UAS Imagery'

This dataset contains high-resolution unmanned aerial system (UAS) multispectral imagery and the corresponding spectral and texture features extracted from burned sites within the Grootbos Private Nature Reserve, South Africa. The study area is located within the fynbos biome—a Mediterranean-type shrubland—known for its high biodiversity and fire-driven dynamics.

Spectral features derived from multispectral bands and spectral indices.

Texture features derived from gray-level co-occurrence matrix (GLCM) analyses.

These features were used to:

Classify vegetation by post-fire age using machine learning methods.

Predict alpha diversity (e.g., Shannon index) across the landscape.

The dataset supports the findings of the paper:“Advancing Mediterranean Biodiversity Monitoring in South Africa through Machine Learning and Cost-Effective UAS Imagery”

Authors: Manisha Das Chaity, Rob Chancia, Ramesh Bhatta, Jasper Slingsby, Glenn Moncrieff, and Jan van Aardt.