Forest Fires

• overview
• subprojects
• Fire History Analysis
• Fire Management - National Resource Planning Under a Changing Climate
• Fire growth modelling and fire/weather data analysis and visualization
• Fire history analysis
• Fire-Weather Science
• Identification of Smoke Plumes from Satellite Images
• Lightning storm tracking using a data sharpening-based clustering algorithm
• Models for predicting fire occurrence
• Remote Sensing Analysis of Forest Fire Smoke for Investigating Health Effects
• Spatio-Temporal Models Examining Fire Frequency and Severity from Panel Data
• software
• publications

Fire growth modelling and fire/weather data analysis and visualization

Researchers: Braun, Kulperger, Stanford, Wotton, Tymstra
Students: Davies, Garcia, Han, Sun, Woolford, Zhou

The Western Ontario team has been working in two main areas: fire growth modelling and fire/weather data analysis and visualization. A grid-based stochastic fire growth model and its simulator have been developed. Mean rates of spread can be calibrated using the Canadian Fire Behaviour Prediction (FBP) system. An important feature of this new model is its ability to simulate jump-fires across fire breaks.

Technical difficulties of the popular PROMETHEUS fire growth simulator, such as aberrant behaviour on the simulated fire perimeter, have now been mostly addressed. Inclusion of stochasticity in the PROMETHEUS model is under development using a bootstrap resampling approach to inference; this will allow probability risk contour maps to be produced. More details on the research related to PROMETHEUS can be found under Project Area 3. A model to provide initial estimates of the likelihood of fire perimeter containment for new fires has also been developed; initial testing has been performed based on certain fuel types for locations in Ontario.

Katherine Davies (PhD through this project) is working on maximum likelihood estimation for a two-dimensional renewal point process which was proposed by Gail Ivanoff at our 2005 workshop.

Lutong Zhou (MSc) developes data visualization tools and nonparametric tolerance interval estimation. She used this methodology to determine bounds on upper percentiles of annual total area burned by wildfire in Ontario. This information is useful to fire managers to aid in strategic planning.

Jessica Sun (MSc) studied fire risk in Ontario due to human causes using generalized additive model methodology.

Ou Feng (MSc, OMNR) applied linear models to the midyear wildfire suppression budget problem (for predicting year-end budget totals) for the Ontario Ministry of Natural Resources.

Cory Ahn (NSERC USRA, undergraduate) continued the study of the problem of adding randomness to PROMETHEUS by smoothing and resampling. She constructed some as-yet unvalidated probability maps.

Kimberley Bridge (NSERC and CFS, undergraduate) did some preliminary modelling of the flush date for aspen. This is an important event, since it signals the end of a short spring fire season in some Boreal regions.

Yu Han (NSERC, undergraduate) assisted in the study of parameter estimation for a stochastic fire spread model.

Qiang Fu (NSERC, undergraduate) assisted in the study of parameter estimation for a stochastic fire spread model.

Kristy Alexander (NSERC, undergraduate) considered a fire risk analysis for Northeastern Ontario using generalized additive models.

Yiwen Diao (NSERC, undergraduate) considered a similar fire risk analysis for Northwestern Ontario using generalized additive models.

Coop placements: Kim Bridge worked at the University of Toronto Firelab under the supervision of CFS researcher Mike Wotton; Ou Feng made several visits to Sault Ste. Marie to work with researchers at the Ontario Ministry of Natural Resources center; Jessica Sun had frequent discussions with Firelab researchers, including Mike Wotton of CFS.