- 55 views
- 58 downloads
Modeling Wildfire Perimeter Formation for Strategic Containment Line Planning in the Boreal Forest Region of Alberta, Canada
-
- Author / Creator
- Mo, Siqi
-
The edge or perimeter of a wildland fire is an important characteristic that marks the extent of land area that has burned. It is not only an ecological boundary, but also an administrative way of documenting the direct impact of a wildfire. Weather has a well-established influence on fire cessation but changes rapidly and is difficult to predict over extended time periods. In contrast, more stable landscape factors such as land cover, road networks, water coverage, and topography are well-suited for informing strategic assessments over longer horizons. Previous studies have shown that the variations in these attributes influence where fires stop. However, these studies are mostly based on one or a few fire perimeter datasets, and the overall influences over a large spatial scale and a long temporal span (i.e., over a decade) are still unclear.
In this thesis, the hypothesis is that key landscape factors play an important role in fire cessation over large spatial and temporal scales. Over one hundred fire perimeters were selected from thousands of documented fires between 2008 and 2018 in the Boreal Forest Region of Alberta, covering various natural ecological sub-regions. Five categories of explanatory variables were chosen from multiple spatial datasets based on provincial statistical maps and remote sensing, to represent landscape factors, including water, topography, fuel, previous burns and human activity. Algorithms to automatically match case-control data sampling (i.e., inside and outside the fire perimeter) and perform data extraction were developed, including a custom Python toolbox and ArcGIS model, resulting in vastly improved processing time. A combined modeling framework connecting matched case-control conditional logistic regression and Random Forest classification was used to identify the influence of key landscape variables on fire cessation and to predict the probability of fire cessation at a given location.
Twenty key landscape factors were identified as having a clear influence on fire cessation, with an Area Under the Curve (AUC) of 0.7. Fire boundaries form in areas of aspen, grass, water, and lower topography. Mixedwood, with conifer contents up to 60%, also played a very strong role in fire cessation (P < 0.001). In contrast, lands where previous fires recently occurred showed strong fire-stopping behaviour, but the effect decreased with the passage of time to the point where land burned 70 years ago behaved equivalent to a hazardous fuel. Human activities also affected the location of wildfire perimeters. The closer a fire is to a roadway, the more likely it will stop. Four new fires were used to validate the model, and the highest AUC of the predictive model was 0.7.
This modelling framework could be used by wildfire management agencies to inform strategic planning. The tool supports decision-making by calculating the fire boundary forming probability for any area within the Alberta boreal zone, based on temporally stable landscape factors. It could potentially guide the allocation of firefighting resources, identify high and low-risk areas for proactive measures, and assist in more efficient fire line planning. -
- Graduation date
- Fall 2024
-
- Type of Item
- Thesis
-
- Degree
- Master of Science
-
- License
- This thesis is made available by the University of Alberta Library with permission of the copyright owner solely for non-commercial purposes. This thesis, or any portion thereof, may not otherwise be copied or reproduced without the written consent of the copyright owner, except to the extent permitted by Canadian copyright law.