Landslides remain a persistent hazard in Japan’s mountainous regions, where complex topography intersects with residential vulnerability. This study investigates how micro-topographical cues shape residents’ perception of landslide risk by integrating high-resolution Digital Elevation Model data with a household survey (n = 204) in Ehime Prefecture. Initial exploratory analysis indicated that, while official hazard designations and topographical features were associated with risk perception, significant discrepancies remained between official risk and subjective evaluation. To examine these relationships further, we conducted an Ordered Logistic Regression analysis, controlling for socio-demographic factors. The results revealed that sex and age are dominant predictors, with women exhibiting higher risk perception and older residents showing a potential “normalization bias”. Regarding environmental cues, upslope housing density exhibited a significant non-linear effect, following an inverted U-shaped trajectory that peaks at a density of two houses ( B = −0.568, p = .034). This reflects the Weber-Fechner Law of psychophysics, where sensitivity is highest at initial stimulus levels before reaching cognitive saturation. Furthermore, a significant interaction effect ( B = 0.024, p = .019) indicated that while housing density can provide a sense of “safety in numbers” in low-relief areas, physical cues of steepness override this social reassurance in high-elevation terrain. These findings suggested that risk communication should move beyond abstract mapping to address intuitive spatial heuristics and socio-demographic biases.
MARUI et al. (Fri,) studied this question.