Quantifying Forest Cover Change at District and Forest Reserve Levels in Southern and Eastern Sierra Leone Using Landsat Data

Abstract

Aim: In Sierra Leone, the paucity of data on forest cover distribution and degradation limits the design and implementation of forest protection measures and the evaluation of the effectiveness of such measures. In view of the above, this study presents the first comparative analysis of district-level and forest reserve-level forest cover change in Sierra Leone, using a case study of the southern and eastern regions of the country. Study Design: A case-study research design and a remote-sensing approach using optical satellite imagery were adopted to quantify forest cover change at district and reserve levels over a 40-year interval. Place and Duration of Study: The study mapped and quantified forest cover change in Moyamba and Bo (Southern Region) and Kenema and Kailahun (Eastern Region) districts in 1986 and 2026. For each district, a forest reserve was identified and changes in its forest cover over the four decades were mapped and quantified. Methodology: Landsat imagery was acquired in 1986 (Landsat-5) and 2026 (Landsat-8). A simplified land-use land-cover (LULC) classification scheme was adopted, comprising six classes: forest, farm bush, grassland, mangrove, water body and others, to which the Support Vector Machine (SVM) classifier was applied. Post-classification change detection was used to quantify LULC changes, and the corresponding district- and forest reserve-level statistics were summarised in bar charts. Accuracies were assessed using confusion matrices based on Google Earth imagery and ground-truth data for the 1986 and 2026 LULC maps, respectively. Results: The LULC classification achieved overall accuracies of 85.9% for the 1986 map and 96.6% for the 2026 map, and the higher accuracy recorded in the 2026 map is attributable to the improvement in radiometric resolution in Landsat-8 relative to Landsat-5. Forest declined across all districts, and a similar trend was recorded at the forest reserve level, except for Kangari in Bo, which recorded a slight increase of 1.4%. The decline in forest occurred in tandem with increases in either farm bush or grassland, or both. The overall decline in forest cover at the district level was greater (-35.2%) than that at the forest reserve level (-12.2%). Moyamba District, which is located along the Atlantic coast recorded a substantial decline (-28.6%) in mangrove. Conclusion: The lesser forest loss in forest reserves suggests a possible effect of protection measures. However, as unprotected forests decline, protected forests and mangroves would face increasing threats. Thus, there is a need to understand the dynamics of community forests, protected forests, and forest reserves. Moreover, studies are required of transboundary forest reserves, such as the Gola Forest, where cross-border differences in approaches and priorities have varying implications for forest management.