Key points are not available for this paper at this time.
Urbanization, deforestation, and forest degradation significantly contribute to atmospheric carbon emissions and heightened climate change risks. Reforestation, a sustainable long-term land use strategy, offers mitigation by sequestering carbon dioxide. To assess reforestation efficacy within urban contexts, continuous carbon stock evaluation in reforested areas is essential for informed management and monitoring. Remote sensing techniques have gained traction in landscape analysis, driven by improved spatial-spectral data characteristics and novel indices. Notably, the Planetscope multispectral imagery, characterized by enhanced spatial and spectral attributes, has potential in enhancing carbon stock estimation. This study examines Planetscope's spectral, derived spectral features and terrain variables' effectiveness in estimating reforested urban landscape carbon stock. Employing extreme gradient boosting algorithm in Buffelsdraai, South Africa, the study's results are compared with an artificial neural network model to test the robustness of the model. Encouragingly, Planetscope spectral data accurately estimated reforested carbon stock with high R2 (0.78 and 0.81) and low RMSE (27.33 and 29.75 t. ha-1) from calibration and validation datasets. Notably, the green normalized vegetation index (GNDVI), red-edge normalized difference vegetation index (NDVIRED), and red-edge simple ratio index (SRRED) are optimal predictors. These findings underscore the value of Planetscope spectral data and extreme gradient boosting for precise carbon stock predictions in reforested urban environments. This study's insights are pivotal for designing effective reforestation ecosystem management and monitoring strategies, with implications for larger-scale carbon sequestration projects and resilient urban landscapes.
Matiza et al. (Fri,) studied this question.