Agglomeration and scanning methods detect edge effected reductions in fire frequency in simulated boreal forest landscapes

Keywords: fire frequency, forest fire, landscape, models
Session Type: Virtual Paper Abstract
Day: Saturday
Session Start / End Time: 2/26/2022 08:00 AM (Eastern Time (US & Canada)) - 2/26/2022 09:20 AM (Eastern Time (US & Canada))
Room: Virtual 12

Abstract

In real forested landscapes, fires ignited outside the permeable boundary of a study area may spread into that area, but in simulated forest landscapes fires are not ignited outside of it and, thus, fewer fires occur near its rigid boundary. Therefore, the rigid-boundary edge effect of a simulated forest fire landscape greatly reduces burn probability and fire frequency along the outer portions. The width of the edge effected region varies with the organism and process of interest, ranging from meters to kilometers. Thus, identification of the edge effected region is required to accurately estimate the fire frequency from a simulated forest landscape. We developed an agglomeration method and a scanning method to detect and remove the edge effected region and applied them to a boreal forest landscape simulated by LANDIS-II model. The detected edge effected region ranged in mean width from 5.9 to 17.3 km and occupied 4.9 to 21.3% of the simulated landscapes. The maximum width of the edge effected region covaried with wind predominance, indicating it is not possible to prescribe a standard buffer width for all simulation studies. The two detection methods differ in their optimality, with the best results provided by a consensus of the two methods. The result suggests that future landscape forest simulation studies should, to ensure their results near the rigid boundary are not misrepresentative, simulate an appropriately enlarged study area and then employ edge detection methods to remove the edge effected region.