2026 Volume 46 Issue 1 Pages 22-39
For better estimation of hazardous areas affected by landslide occurrences, this paper has proposed a Causal- and Trigger-factor Influence map (CTI map) using geospatial information for landslide hazardous areas assessment. The CTI map can be produced by the proposed CTI algorithm based on the Structural Equation Modeling (SEM). The training data set (i.e., endogenous variable) of past landslide occurrences induced by the Niigata Chuetsu earthquake in Nagaoka area are prepared. Causal factors (i.e., topography, surface geology, soil, slope, relief, elevation, aspect, drainage, and vegetation) are used as “observed variables” on geospatial information, while trigger factor caused landslides are regarded as “unobserved latent variable.” A path diagram in which the single-headed paths are drawn from the two exogeneous variables (i.e., causal and trigger factors) to endogenous variable (i.e., training data) is considered. Through the goodness-of-fit test on the path diagrams, the recall rate, the precision rate and the F1 score curves with respect to the training data set, the proposed path diagram can be accepted. Based on the structural and measurement equations in SEM, the causal and trigger factor influence maps are newly produced, respectively. As a final outcome, the differences between those maps are delineated on the CTI map which is effective in comparative evaluation of the influence between causal and trigger factors for landslide hazard assessment, supporting the installation of various measurement devices, as well as identifying candidate categories for subdivision within causal factor maps.
