Statistical And Gis Approaches To Landslide Susceptibility Assessment And Mapping In Mila Basin (ne Algeria)

Résumé: The severe landslides affecting Mila Basin (located in the North-East region of Algeria) are serious threats not only to the environment and the local populations but also inflecting economic burdens to local authorities by the non-ending reconditioning and restoration projects. In addition, these landslides affect the current landscape evolution and the geodynamics of the basin. Therefore, predicting and delineating landslides are crucial tasks to reduce their associated damages. However, landslide risk prevention requires prone areas delineation using an assessment that can integrate into GIS environments and considering the spatial and temporal space component of the basin. This should theoretically provide probabilities for both the spatial and temporal components of this hazard in the form of susceptibility toward landsliding. That being said, no systematic and accountable landslide susceptibility models or even susceptibility maps are available for the basin yet, despite the tremendous losses. In an attempt to fill this gap, an advanced statistical-based modeling approach (i.e. Machine Learning) was used to provide state-of-the-art models capable of providing the highest landslide prediction capabilities. The main research workflow was rather simplistic as it focuses essentially on elaborating predictive models using some advanced techniques that can be integrated successfully in GIS environments in order to develop customized models for the basin. A partial focus was given to mapping and zoning areas toward landsliding. The obtained results highlight the overall benefits of using advanced machine learning methods for landslide susceptibility assessment, as the implemented models exhibit reasonably good predictive performance (AUC>0.85, Acc> 78% and kappa>0.56). The generated landslide susceptibility maps were proven to be useful as a technical framework for spatial prediction to develop countermeasures and regulatory policies by decision-makers to minimize the damages introduced by either existing or future landslides.

Publié dans la revue: