Predicting Fire Ignition in Eastern Oregon using Maximum Entropy Modeling

Authors: Michael W Palace*, University of New Hampshire, Joel Hartter, University of Colorado, Angela Boag, University of Colorado, Mark J Ducey, University of New Hampshire, Lawerence C Hamilton, University of New Hampshire, Forrest R Stevens, University of Louisville, Nils D Christoffersen, Wallowa Resources, Paul T Oester, Oregon State University
Topics: Spatial Analysis & Modeling, Coupled Human and Natural Systems, Mountain Environments
Keywords: Fire, Eastern Oregon, Spatial Modeling, Maximum Entropy
Session Type: Paper
Day: 4/12/2018
Start / End Time: 10:00 AM / 11:40 AM
Room: Endymion, Sheraton, 8th Floor
Presentation File: No File Uploaded

Wildfire regimes are generally determined by interactions between climate, vegetation and topography, and increasingly human activities. In western North America, theory suggests fire regimes in low elevation dry forests, where fire season climatic conditions are generally conducive to burning. Understanding causal factors for ignition is important to aid in identifying at-risk forests, their characteristics, and the proximity of high-risk locales to housing and development is important in developing fire wise practices and forest policy on both public and private forests. We use maximum entropy modelling to develop fire ignition probability maps and demonstrate its utility in the Blue Mountains of Eastern Oregon. We used fire ignition data from the US Forest Service to model nine ignition sources: arson, campfire, children, debris burning, equipment, lightning, smoking, railroads, and miscellaneous. Lightning accounted for the majority of fires ignited, whereas human ignitions had different predicted distributions and driving factors. We used geospatial layers such as solar radiation, percent tree cover, vegetation type, topography (elevation, slope, aspect, and ruggedness), temperature, precipitation, and topographic wetness index in our modeling effort. The influence of these variables differed based on fire ignition cause, but all models included strong contributions from vegetation type, temperature, and precipitation. Our geospatial estimates of ignition probability have the potential to aid in the placement of equipment and personnel to combat wildfires. Work examining the interannual variability and seasonal dynamics using geospatial modeling can also aid in annual fire forecasting efforts.

Abstract Information

This abstract is already part of a session. View the session here.

To access contact information login