Comparing two approaches for landslide susceptibility mapping - a case study in Freetown, Sierra Leone

The city of Freetown in Sierra Leone (Africa) is located on a coastal peninsula characterized by the regular occurrence of floods and landslides that have directly affected the population. The most relevant event corresponds to a landslide and a mudflow (as a cascade effect of the landslide) during a heavy rainy day that occurred in the Regent neighborhood. This disaster occurred in August of 2017 and resulted in the death of 1141 people mainly due to the mudflow.
Considering the high impact of disasters in the urban and rural areas of Freetown and the increase in the degree of exposure generated by urban growth, it is necessary to implement susceptibility studies as the first approach to risk management. For the evaluation of susceptibility in the area of interest (Freetown peninsula), the Regent landslide of 2017 was taken as a reference, seeking to determine the conditioning and detonating factors that favor the occurrence of landslides in the entire area. This extrapolation of factors is possible since the evaluation is local, and there are no substantial differences at geological, geomorphological, structural and climatic levels.
This study applies two methods for landslide susceptibility assessment: i) Landslide Statistical Index (LSI) and ii) Analytical Hierarchy Process (AHP). LSI strongly depends on the quality and completeness of landslide catalogs. AHP does not depend directly on landslide catalogs but is mainly based on the degree of experience of the evaluator with respect to the behavior of the phenomena in the evaluated area. The work with AHP involved the participation of different professionals with different degrees of knowledge of the study area. The quality in the allocation of weights was evaluated according to the Consistency Ratio (CR). The results reflect the rich variety of opinions regarding the conceptualization of landslide phenomena. Additionally, they help to identify future challenges related to the conceptualization of the hazard itself, given that the limit between what is considered as conditioning and triggering factors does not correspond to a consensus.