USING NEUTRAL MODELS TO EVALUATE THE EFFECT OF TOPOGRAPHY ON LANDSCAPE PATTERNS OF FIRE SEVERITY: A CASE STUDY OF LASSEN VOLCANIC NATIONAL PARK

Open Access
Author:
Pierce, Andrew D
Graduate Program:
Geography
Degree:
Doctor of Philosophy
Document Type:
Dissertation
Date of Defense:
May 12, 2011
Committee Members:
  • Alan H Taylor, Dissertation Advisor
  • Alan H Taylor, Committee Chair
  • Erica A H Smithwick, Committee Member
  • Kenneth James Davis, Committee Member
  • Douglas Alan Miller, Committee Member
Keywords:
  • random forest
  • fire behavior modeling
  • vegetation dynamics
  • neutral model
  • landscape ecology
Abstract:
Fire is the most important disturbance process in forests of the American west. At the finest scales, its immediate effects on vegetation include killing of live vegetation and the consumption of dead organic debris. At the scale of one to many hectares, it drives variability in vegetation composition and structure. Heterogeneity in fire behavior and effects are not driven solely by the vegetation itself. Fire interacts with weather, climate and climate teleconnections, past human management of forested landscapes, and with topography. The effect of topography on heterogeneity in fire behavior and effects is not well knows, and some evidence is contradictory. I employ a neutral modeling approach generate an examine landscape scale patterns of fire intensity and fire effects. Broadly, I address three main research questions: 1) what is the distribution of surface and canopy fuels in LVNP, and how do they vary with topography? 2) what is the effect of topography on landscape level heterogeneity in neutral models of fireline intensity and fire type? 3) Is topography able to explain heterogeneity in the observed severity of historic and contemporary fires and is it able to explain heterogeneity in modeled fireline intensity? I accomplish this in stages. I use plot level data from 223 plots and 669 hemispherical photographs to assess the effect of topography—elevation, slope, and aspect—on the distribution of both surface and canopy fuels. Then I develop ten remotely sensed variables from Landast imagery and five topographic variables—elevation, slope, aspect, local topographic position, and landscape position—from the National Elevation Dataset. I use a Random Forest algorithm to model and then predictively map canopy fuels. In the second part, I model fire burning through homogenously distributed fuels at the 80th, 90th, and 97th percentile fuel moisture conditions and for a range of wind scenarios. This set of simulations forms the base of our neutral model expectations and is analyzed for the effect of topography. Finally, we map historic patches of high intensity fire effects from geo-referenced aerial photographs and extract severity data from remotely sensed images of recent fires. We compare the topographic conditions of these real high severity patches with information on high intensity fire from the neutral model approach to assess how important topography is in explaining the distribution of different severities. Our results show that slope angle is the most important variable for determining modeled fireline intensity, but that elevation and local topographic position are the most important variables for explaining the distribution of observed locations of high severity fire. These result support our conclusion that some patches of vegetation may be ‘fixed in space’ through the interaction of fire with topography and fuels.