Human- and lightning-caused wildland fire ignition clusters in British Columbia, Canada

Abstract

Wildland fire is a common occurrence in western Canada, with record-setting area burned recorded in British Columbia (BC) in the past decade. Here, we used the unsupervised machine learning algorithm HDBSCAN to identify high-density clusters of both human- and lightning-caused wildfire ignitions in BC using data from 2006 to 2020. We found that human-caused ignition clusters tended to occur around population centres, First Nations communities, roads and valleys, and were more common in the southern half of the province, which is more populated. Lightning-ignition clusters were generally fewer in number and larger in size than human-caused fires for most hyperparameter settings. There were significant differences (X2 = 1884.8, d.f. = 7, P-value <2.2 × 10−16) in fuels associated with lightning- versus human-caused ignition clusters, with human-ignition cluster fires being more often found within leafless aspen (D1) and ponderosas pine and Douglas fir (C7) fuel types. These high-density clusters highlight regions where the greatest densities of both lightning- and human-caused fires have occurred in the province, thereby identifying regions of potential interest to wildland fire managers, researchers and various communities and industries.