Mapping Changes in the Peatland of Ngamakala, Republic of Congo

Abstract

Peatlands, a type of wetland, act as a natural carbon store, which, when left undisturbed, prevent, this carbon from further warming the global climate. However, these peatlands, in the Republic of Congo are subject to numerous anthropogenic pressures. Improvements have been made using Sentinel-2 multispectral images between 2017 and 2023 in the Ngamakala peatland forest. We applied a Machine Learning (ML) model using the Random Forest (RF) algorithm to map changes over the period studied. The methodology involved preprocessing Sentinel-2B images, creating training samples, designing the ML model, and then predicting (classifying) and validating the results. The overall accuracies of the classifications range from 91% to 96%. The time series classifications show large changes in land cover type through time. This is owing to anthropogenic activities that are threatening to the Ngamakala peatland forest. We recommend that the authorities take action to protect this site, which is almost 25,000 years old.