Author(s) :

Jessy Houston Tchoffo Tsoala, Michel Gustave Opelele, Yann Edwin Essi Avoto, Cedrick Belmich Nkoua Ondon, Raphael Muamba Tshimanga, Papy Claude Bolaluembe, Bonaventure Nyami Lele, Eric Landu Lutete, Jean Ngabinzeke Semeki.

Volume/Issue :

Abstract :

The The commune of Campo (South Cameroon), which hosts part of Campo Ma’an National Park, faces growing agricultural pressure. This study analyses land use and land cover (LULC) dynamics in 2000, 2015 and 2025, then simulates their evolution under a business-as-usual scenario for 2035 and 2045. Six-class classification (Random Forest, Google Earth Engine, Landsat) achieved overall accuracies of 95.3–98.2% (Kappa 0.69–0.81). Dense forest declined by about 40,181 ha between 2000 and 2025. Comparison with the independent JRC-TMF product shows that the degradation mapped over 2000–2015 mainly results from classification errors (96.7% of the pixels concerned remain undisturbed forest), but confirms the agricultural expansion of 2015–2025 (83.5% of the pixels concerned deforested or degraded), which coincides with clearing in the Camvert oil palm concession (since 2019–2020). Simulation relied on a model coupling cellular automata and artificial neural networks (CA-ANN). Validation on 2025 yields a Kappa of 0.34 and a Figure of Merit of 0.19: although less accurate than a null persistence model (Kappa = 0.40), the model locates part of the changes (AUC of 0.82–0.92 for transitions from dense forest). Under the business-as-usual scenario, dense forest would decrease from 84.2% of the modelled area in 2025 to 82.1% in 2045, and agricultural area would increase by about 23%. These results inform sustainable land management in Campo, although the fine spatial location of simulated changes should be interpreted with caution.

No. of Downloads :

0