Modelling deforestation, carbon stock changes, and identification of optimal forest restoration sites in a rapidly urbanising landscape

Abstract

Developing towns and cities worldwide face high deforestation rates, yet accurate information on its spatial extent, dynamics, and potential restoration sites remains limited. Forest ecosystems play a pivotal role in carbon sequestration; however, their degradation disrupts ecological functions, necessitating spatially explicit assessments of carbon stock changes over time. Understanding the spatiotemporal patterns of forest loss is essential for assessing carbon dynamics and guiding targeted restoration interventions across ecologically sensitive and socioeconomically vulnerable landscapes. This study aimed to assess deforestation dynamics, quantify forest cover, and identify potential restoration sites in Chinhoyi, Zimbabwe. Using Geographical Information Systems (GIS), Remote Sensing (RS), and the Random Forest (RF) machine learning algorithm, the study assessed forest cover loss and restoration suitability. Potential restoration sites were identified by evaluating factors such as slope, proximity to roads and settlements, and land use land cover (LULC) patterns using the Weighted Overlay Analysis (WOA) and Analytical Hierarchy Process (AHP). Findings revealed a 54.58 % net reduction in forest cover from 2014 to 2024, largely driven by agricultural expansion, urbanization, and land degradation. However, transition analysis also indicated localized regeneration, with 4.57 %– 15.17 % ha of forest gains observed in different intervals, highlighting natural regrowth and reforestation processes. Carbon stock analysis indicated significant losses, with 45,252 tons of carbon emissions exceeding regional averages over the decade. The study recommends prioritizing reforestation and forest restoration efforts in highly suitable areas to recover forest cover and mitigate the impacts of continued deforestation.