Static maps, dynamic threats: re-evaluating U.S. wildfire risk with spatiotemporal clustering

Abstract Wildfire (WF) risk in the United States is escalating, yet conventional risk assessment methods often treat this risk as static. This study provides a national-scale comparison of purely spatial and spatiotemporal clustering methods to identify WF hotspots across the contiguous United States (CONUS) from 2014 to 2024. Using a comprehensive dataset of WF occurrences and spatial and spatiotemporal scan statistics, we analyze how the inclusion of a temporal dimension alters risk identification. The results demonstrate a fundamental divergence: purely spatial analysis identifies persistent, chronic hotspots, often in the Southeastern U.S., with smaller at-risk populations. In contrast, spatiotemporal analysis reveals transient, acute hotspots, frequently shifting risk to the Western U.S. and Southern Plains, and encompassing at-risk populations an order of magnitude larger. The distinction between chronic and acute risk necessitates a dual-strategy approach to fire management, combining long-term mitigation with agile, short-term operational responses. This research provides a new framework for understanding and managing the dynamic nature of contemporary WF risk on a national scale.