Correlation Between Geotechnical Indexes and Landslide Occurrence in Southwestern Cyprus Using GIS and Machine Learning

Abstract Landslides are significantly influenced by the properties of the geological materials. As such, effective landslide susceptibility and hazard assessment necessitates the use of carefully selected and well-organized spatial data on geology and ground characteristics. The present study explores the spatial correlation between landslides and geotechnical indexes pertinent to the problem of slope stability. For this purpose, a geodatabase containing geotechnical properties was created for the southwestern part of the island of Cyprus, an area noted for its frequent ground instability issues and the availability of a comprehensive landslide database. Then, statistical correlations are established between the recorded landslides (active and inactive) in the region and key geotechnical, geological and geomorphological factors. The analysis was initially performed using the Frequency Ratio method, followed by two advanced machine learning techniques, namely Random Forest and Shapley Additive Explanations. The analysis results reveal that for weak argillaceous geomaterials, the clay content and plasticity index constitute high importance variables, followed by the geomorphological factors such as the slope angle. In rocky formations with clear rock mass structure, as main factors emerge the Geological Strength Index and the uniaxial compressive strength. The strong correlation identified between geotechnical factors and the spatial distribution of landslides underscores the potential benefit of integrating these variables in landslide susceptibility and hazard analysis methodologies. This adjusted emphasis provides clearer insights into the relationship between geological properties and landslide occurrences, which is crucial for developing more accurate predictive models and effective mitigation strategies.