Nationwide Susceptibility Mapping of Landslides in Kenya Using the Fuzzy Analytic Hierarchy Process Model

Abstract

Landslide susceptibility mapping (LSM) is a cost-effective tool for landslide hazard mitigation. To date, no nationwide landslide susceptibility maps have been produced for the entire Kenyan territory. Hence, this work aimed to develop a landslide susceptibility map at the national level in Kenya using the fuzzy analytic hierarchy process method. First, a hierarchical evaluation index system containing 10 landslide contributing factors and their subclasses was established to produce a susceptibility map. Then, the weights of these indexes were determined through pairwise comparisons, in which triangular fuzzy numbers (TFNs) were employed to scale the relative importance based on the opinions of experts. Ultimately, these weights were merged in a hierarchical order to obtain the final landslide susceptibility map. The entire Kenyan territory was divided into five susceptibility levels. Areas with very low susceptibility covered 5.53% of the Kenyan territory, areas with low susceptibility covered 20.58%, areas with the moderate susceptibility covered 29.29%, areas with high susceptibility covered 29.16%, and areas with extremely high susceptibility covered 15.44% of Kenya. The resulting map was validated using an inventory of 425 historical landslides in Kenya. The results indicated that the TFN-AHP model showed a significantly improved performance (AUC = 0.86) compared with the conventional AHP (AUC = 0.72) in LSM for the study area. In total, 31.53% and 29.88% of known landslides occurred within the “extremely high” and “high” susceptibility zones, respectively. Only 8.24% and 1.65% of known landslides fell within the “low” and “very low” susceptibility zones, respectively. The map obtained as a result of this study is beneficial to inform planning and land resource management in Kenya.

Identifying Landslides Prone-Areas Using GIS-based Fuzzy Analytical Hierarchy Process Model in Ziz Upper Watershed (Morocco)

“…Some research has shown that for large areas with low availability of historical landslide inventories, the MCDA methods are suitable due to their significant advantages over statistical and physical methods (Glade et al, 2012 ;Barella et al, 2018;. For large areas, a review of previous studies has shown that AHP and Fuzzy-AHP are one of the best approaches for landslide susceptibility mapping (Zhou et al, 2020).…”

“…Heuristic analysis, Inventory analysis and Analytical Hierarchy Processing (AHP) are the most important qualitative LSM models. The quantitatively developed methods showed that the most widely used are frequency ratios (FR), logistic regressions (LR), weights of evidence (Wof E), artificial neural networks (ANN) and support vector machines (SVM).Whereas, SHALSTAB, SINMAP and TRIGRS are the most important physically based LSM models, developed based on the integration of slope stability models and groundwater flow models to calculate the safety factor per each slope unit (Zhou et al, 2020).Looking at all these models, it was found through comparative studies that the optimal choice of one of the LSM methods depends largely on nature and availability of data in the study area.…”

Waterlogging Risk Assessment: An Undervalued Disaster Risk in Coastal Urban Community of Chattogram, Bangladesh

“…AHP was chosen because it can deal with complex decision-making and be useful for checking the consistency of the evaluation measures, as suggested by the experts. Thus, it is widely used for vulnerability and risk mapping (i.e., [27,35,[37][38][39]).…”

Landslide Disasters’ Causal/Trigger Factors as Understood by the Indigenous People in Murang’a County, Kenya

“…Like many areas of the world, particularly in the tropical developing countries, Kenya is at risk of landslides and their associated effects [11] which leads to hundreds billions of Dollars loss [12]. Landslide disasters are not new phenomena in Kenya [13][14][15] yet a dearth of information about the same has been cited in the country [13,14,16]. Murang'a County presents a unique case due to its geographical location within the Aberdare Ranges, which runs across other counties such as Nyeri, Nyandarua and Kiambu, but with Murang'a landslides being recurrent in the recent past [15] and deadly [9].…”