Reliable and up-to-date digital soil data is crucial for achieving Sustainable Development Goal 13 (Climate Action) by enabling improved monitoring of soil carbon and land degradation, thereby supporting climate-smart agriculture and ensuring stable crop yields in sub-Saharan Africa. This study focuses on the spatial mapping of soil organic carbon (SOC) comparing predictive models that integrate Landsat 8 variables and DEM derivatives within a Random Forest framework. Three models were evaluated: Model A, which incorporates only Landsat 8 derivatives; Model B, based solely on DEM variables; and Model C, which integrates both Landsat 8 and DEM datasets. The results indicate that Model A achieved an RMSE of 0.15 (%) and an R 2 of 0.67, while Model B achieved an RMSE of 0.19 (%) and an R 2 of 0.54. Model C (the combined model) achieved the highest explanatory power with an R 2 of 0.69. The findings highlight the significant influence of DEM-derived variables, such as profile and plan curvature, on SOC distribution, emphasizing the role of terrain attributes in SOC mapping. This study demonstrates the potential of RF modeling for SOC prediction, reinforcing the importance of integrating spectral and topographic variables to enhance accuracy. To achieve sustainable farming and resilient crop production in sub-Saharan Africa, accurate digital soil mapping is essential. These datasets empower climate action by tracking soil health and carbon sequestration, providing the necessary evidence base for effective land management strategies. • DEM variables strongly influence SOC prediction, with models using terrain attributes achieving the highest accuracy. • Random Forest models perform well, especially when integrating both spectral and topographic data. • Digital SOC mapping is essential for climate-smart agriculture, soil conservation, and SDG 13 implementation. • Reliable SOC data supports policy and land management, enabling targeted interventions for soil resilience and carbon sequestration in sub-Saharan Africa.
Yaqub et al. (Sun,) studied this question.