Applicability of R statistics in analyzing landslides spatial patterns in Northern Turkey

Statistical analysis of rainfall-triggered landslides inventory patterns is a key for landslide hazard and risk prediction analysis of susceptible areas. The main objective of the study is to test if the landslides locations are spatially auto correlated, that could either be clustered (spatial attraction), dispersed or randomly distributed (spatially independent). Two categories of spatial distance functions were applied, first using, first-order distance analysis using Quadrat Counts function and kernel density analysis. The second category used second order distance analysis includes Diggle's empty space F-function and nearest neighbor distance G-function, and also, more sophisticated Ripley's K-function, which evaluates the distribution of all neighbor distances within the space taking into consideration the edge correction effect. Based on the generated curves by the G, F and K functions, we observed that landslides locations clearly tend to be clustered in certain areas rather than randomly distributed. Eventually, Moran's I autocorrelation function used to find where the highest amount of landslides are clustered using four conditioning factors (Elevation, Slope, Land-cover, and Geology). This study tests the landslides distribution pattern in landslide prone area of Trabzon city, northern turkey. The current study aims to facilitate the integration of spatial data and the coding in R environment through using the R extensive research tools and libraries.