Application of spatial statistics for climate-sensitive natural hazard management

This is the authors' abstract. We don't add key points for this paper.

Recent climate records show a marked intensification of seasonal climate dynamics over the past three decades, with projections suggesting continued intensification through 2100. The resulting variability in high-intensity rainfall or drought events combined with changes in vegetation is a challenge for the maintenance and safety of public infrastructure such as roads and railways. This study integrates hazard event databases with climate model outputs to identify the climate indices spatially associated with clusters of slope hazards along the Austrian motorway network, using global and local Moran's I (LISA) on two complementary spatial supports. The analyses associate rockfall and debris-flow clusters most strongly with the annual number of frost days, and earth-and-rock-slide clusters with daily rainfall intensity. Against the RCP8.5 change signal the bivariate statistic reverses sign on both supports: the documented hazard clusters coincide with the cells of strongest projected change, placing formerly clustered sections in a high–low regime. The analytical results serve as diagnostic indicators of spatial co-exposure rather than causal triggers, highlighting zones where documented hazards and the strongest projected climate change coincide along the road network. For these areas, a screening methodology incorporating remote sensing and spatial analysis extends monitoring beyond individual assets. The findings underscore the value of climate-informed asset-management strategies under evolving environmental conditions.