Assessment of AHP AND Fuzzy Logic Landslide Susceptibility Mapping Models. A Case Study of Bududa District.
MetadataShow full item record
Globally, landslides have been noticed to rise (Neema et al., 2018). Over the years, they have been triggered by a number of factors which include rainfall and human activities among others. The occurrence of landslides leads to loss of lives and massive damages which leads to a fall back in the economy of the cities where they occur. Landslides susceptibility mapping (LSM) is an answer to grasping and modelling upcoming natural threats so that their effects are minimized. (Feizizadeh and Blaschke, 2013). This process involves deriving a relationship between landslide occurrence and causal factors which is done using an appropriate model. Various models have been developed to assign weights and combine the landslide causal factors such that landslide susceptibility maps are generated. This process can however be very uncertain and if care is not taken, the derived landslide susceptibility maps may misinform the GIS experts who use these models and in turn, this may misinform the decision makers who use the maps in planning for mitigation of the landslides. This study therefore was focused on performing an assessment and compare the two most widely used models of landslide susceptibility mapping which are AHP and fuzzy logic in order to guide the GIS experts on which is a more reliable model for use in the process of landslides susceptibility mapping. The main objective of this study was to evaluate the performance of AHP and Fuzzy logic landslide susceptibility mapping models in Bududa and the specific objectives included, to determine landslide susceptible areas using AHP and fuzzy logic and to assess the reliability of landslide susceptible areas generated by AHP and fuzzy logic. To achieve these objectives, landslide inventory data, a DEM, LULC map, rainfall map and geological map were used and with the help of GIS, this data was analyzed in order to generate a relationship between landslide occurrence and the causal factors which were slope, aspect, rainfall, elevation, LULC, soil type, sink distance to stream, using each model. Landslide susceptibility maps were generated using AHP and fuzzy logic. Thereafter, data validation was performed using the ROC method and during this process, prediction curves were generated from which the area under curve was used for assessment where the AUC for Fuzzy logic (97.47%) was better than that for AHP (62.53%) for Bududa. The generated susceptibility maps can be used for the purposes of land use planning and hazard mitigation.