The effect of random and spatially explicit lightning and human-caused ignitions on simulated burn probabilities at small scales

James Lowery, The University of Montana

Abstract

Our overall goal was to assess the difference between annual wildfire burn probability (BP) generated from simulation models using random ignitions versus spatially explicit ignitions for wildland urban interface (WUI) areas and designated wilderness in the Rocky Mountain West, and infer whether one may be more useful than the other at small scales. We collected all available ignition data from logbooks, spreadsheets and the internet for a 1282 km2 study area within a 10 km buffer of the Rattlesnake National Recreation Area (RNRA) and Rattlesnake Wilderness (RW), Montana, which also included WUI. These ignitions were on small private landholdings, as well as on public land, for the years 2000-2010, and enabled the identification of specific wildfire ignition zones, that facilitated the parameterization of wildfire simulation models to more closely reflect where ignitions occur. The wildfire ignition model Randig was used to create BP maps for our focus area (the Rattlesnake Valley, the RNRA and RW) under eight ignition scenarios: two types of ignitions (human-caused and lightning-caused), two spatial patterns of ignition (random and spatially explicit), and two levels of wildfire ignition frequency (average and high). Spatially explicit ignition scenarios based on actual human-caused and lightning ignition locations generated higher BP in the WUI and Wilderness in our focus area relative to random ignition scenarios. Our results indicate that spatially explicit ignition data should be used whenever possible to estimate BP at small scales, to support placement of fuel treatments, deployment of wildfire suppression resources, and for WUI preparedness including prevention programs and homeowner education.