Understanding where landslides are likely to occur is an important step in reducing disaster losses, especially where unstable terrain overlaps with residential development. A study published in Scientific Reports on 4 October 2026 applies machine learning to landslide susceptibility mapping in eThekwini, South Africa. By combining environmental data with an assessment of housing exposure, it provides spatial evidence that could inform local disaster risk planning.
In April 2022, prolonged heavy rainfall triggered destructive flooding and landslides in South Africa’s KwaZulu-Natal province. Homes, roads and bridges were damaged, while flooding in Durban also disrupted port operations. The disaster illustrates the importance of understanding how hazardous locations overlap with settlements and essential infrastructure. The researchers combined an April 2022 landslide inventory, Sentinel-2 imagery, elevation data and environmental layers at 30-metre resolution. Input variables captured terrain, vegetation, geology, land use and proximity to roads and rivers. Random forest and support vector machine models used 10,000 samples, equally divided between landslide and non-landslide locations, with an 80:20 training–testing split. The support vector machine used ten-fold cross-validation for parameter tuning. On the held-out test set, random forest achieved 99.45% accuracy and the support vector machine 98.63%. Land use, elevation, vegetation index and lithology were influential predictors. Both maps highlighted elevated susceptibility along the coast, although the support vector machine identified a broader susceptible area. Overlaying the support vector machine map with housing data showed that 1.33% of formal housing and 2.41% of informal housing fell within high-susceptibility zones. These percentages refer to each housing category. Potential planning uses include prioritising detailed site investigations and identifying settlements that require closer assessment. Applying these maps also requires information about the vulnerability of exposed communities and their capacity to cope with hazardous events. UNDRR guidance distinguishes spatial susceptibility from the timing and frequency of landslide triggers, making susceptibility mapping one component of a wider disaster risk assessment. The authors also recommend field validation and higher-resolution terrain data to strengthen future work.
Sources: Scientific Reports
Comment list ( 0 )
