Landscape character assessment of Nigeria: A machine-assisted learning approach

Landscape Character Assessment (LCA) is an approach used to understand the unique characteristics of places, typically transcending conventional politico-administrative boundaries. While LCA is well-established in Europe and other parts of the world, its application in Africa, particularly in rapidly urbanising countries, remains underexplored. This study focuses on the LCA of Nigeria, using open-access geospatial datasets and a machine-assisted learning approach. We adopted and modified the LANMAP typology, incorporating climate, parent material, and elevation as physiographic elements and land cover as a proxy for cultural elements at a broad scale (1 km resolution). These elements were analysed using K-means clustering, Gap statistics, Silhouette coefficients, and landscape structure metrics to examine the derived landscape units. Essentially, 24 physiographic and 131 landscape character typologies (LCTs) were identified. Assessment of internal coherence between the derived typologies and input elements, along with qualitative evaluation (including stakeholders' feedback), yielded agreement exceeding 74% in both cases. Also, the class-level structure metrics computed for the LCTs show varying levels of spatial heterogeneity in the country’s landscape. In addition, our study produces a flexible and FAIRly-Open LCA framework, including an interactive dashboard to share our results and obtain user feedback for (future) improvements. Thus, this study can serve as a potential basis for landscape ecological assessments, planning, protection, resource management, and for broad strategic policy development in Nigeria and beyond - particularly where LCA is non-existent.