Tropical Peatland Classification Using Multi-Sensor Sentinel Imagery and Random Forest Algorithm in Greater Amanzule, Ghana

Abstract

Tropical peatlands such as Ghana's Greater Amanzule peatland are important ecosystems due to the magnitude of their greenhouse gas emissions under human and climatic pressures. Accurate measurement of their occurrence and extent is required to facilitate sustainable management. A key challenge however is the high cloud coverage in the tropics that limits optical remote sensing data acquisition. We combined optical, radar and elevation data to optimise Land Use and Land Cover (LULC) classification for the Greater Amanzule tropical peatland. Sentinel-1, Sentinel-2 and SRTM data were acquired, and appropriate features were selected and integrated to develop a machine learning LULC classification using a Random Forest classifier. A total of six LULC classifications were made. Results showed that the best overall accuracy (OA) was found for the integrated Sentinel-1, Sentinel-2 and SRTM features (S1+S2+DEM), significantly outperforming all the other classifications with an OA of 94%, followed by the integrated Sentinel-1 and Sentinel-2 (S1+S2) (92%). Sentinel-1 only (S1) had the worse OA of 70%. The integration of more features systematically increased the classification accuracy. We estimated Ghana's Greater Amanzule peatland at 60,187ha. Our proposed methodological framework contributes a robust workflow for accurate and detailed landscape-scale monitoring of tropical peatlands, while our findings and research outputs provide timely information critical for the sustainable management of the Greater Amanzule peatland.