GIS-BASED CATASTROPHE MODELLING OF GEOPHYSICAL AND REMOTE SENSING DATA FOR LANDSLIDE SUSCEPTIBILITY MAPPING IN OKEMESI-EKITI, SOUTHWESTERN NIGERIA

Abstract

The need to develop a viable decision support system tool to mitigate the effect of frequent landslides in the steep slopes of Okemesi-Ekiti and Environ was addressed in this study. Relevant datasets such as surface data (remote sensing, climatic and ancillary data) and subsurface data (VES and aeromagnetic data) were acquired. The data were processed by application of relevant geophysical (Oasis Montaj and winResist) and remote sensing software. Results obtained from geospatially processed surface data revealed that geomorphology, geology and climatic factors play important roles in the landslides occurrences in the area. Presence of magnetic anomalies from subsurface geologic structures (mainly faults and fractures) and subsurface lithological layers consisting of topsoil, sandy-clay, fractured basement and fresh basement at varying depths were delineated from the processed magnetic and VES data. Based on the results obtained from the interpreted surface and subsurface data, six (6) Landslide Causative Factors (LCFs) namely; slope, drainage density, rainfall, lithology, lineament density, and transverse resistance were derived and constructed into spatial database. Exploring the efficacy of Catastrophe Theory-Based data mining model and Weight Linear Algorithm (WLA), the produced LCFs thematic maps in the GIS environment were used to establish the Landslide Susceptibility Mapping Index (LSMI). The results of the applied LSMI gives index estimate in the range of 25.26 to 62.58 which were geospatially modelled to produce the landslide susceptibility map (LSM) of the study area. Similarly, the output of the applied Analytical Hierarchy Process (AHP) data mining model in this study also produced landslide susceptibility map for the area. The resultant landslide susceptibility maps (LSMs) produced in the context of Catastrophe and Analytical Hierarchy Process (AHP) for comparison classified the study area into five zones of very low, low, moderate, high, and very high landslide susceptibilities. The maps were validated via application of Ground Control Points (GCPs), Landslide Density and Area Under the Curve (AUC) techniques, using existing landslides in the study area. Qualitatively, the GCPs validation result shows that the produced LSMs’ predicted result matched with the existing field conditions; while the landslide density validation result proves that areas with high landslide densities correlates with the predicted high susceptibilities zones. The AUC validation technique quantitatively established 61 % and 57 % prediction accuracy for the CT and AHP-based approaches respectively. Findings from this work has established that the landslides in the area is a function of multiple causative factors driving by structural and geomorphological influence. In accordance with the obtained results, the efficacy of Catastrophe Theory (CT) data mining model in Landslide mapping has been established. Thus, the developed CT-based LSM model is a viable decision support system tool that can be explored for cost effective geo-hazard management in Okemesi-Ekiti and areas with similar geology.