<![CDATA[This study aims to produce a landslide susceptibility map for Temanggung Regency, Central Java, using a logistic regression approach optimized through a stepwise Akaike Information Criterion (AIC) procedure. This method was selected to address the limitations of subjective scoring and weighting approaches commonly used in previous studies. The stepwise AIC was applied to an initial model incorporating topographic, hydrological, and anthropogenic variables, resulting in a gradual reduction of the AIC value from 1267.33 to 1258.46. The optimal model with the lowest AIC consists of elevation, slope aspect (cos), land use, distance to roads, Stream Power Index (SPI), and Topographic Wetness Index (TWI), representing the most statistically efficient combination of controlling factors. The dataset comprises 586 landslide events recorded between 2020 and 2024, divided into training (80%) and testing (20%) datasets, with an equal number of randomly selected non-landslide points. All variables show Variance Inflation Factor (VIF) values below 5, indicating no significant multicollinearity. The model demonstrates good predictive performance, with an Area Under the Curve (AUC) value of 0.703. Spatial analysis reveals that moderate and high susceptibility classes dominate the study area, covering more than 86% of the total area.]]>
Copyrights © 2026
