11
Spatially Explicit Burn Probability Spatially Explicit Burn Probability across A Landscape in Extreme Fire across A Landscape in Extreme Fire
Weather YearWeather Year
Wenbin Cui, David L. MartellWenbin Cui, David L. Martell
Faculty of Forestry,Faculty of Forestry, University of TorontoUniversity of Toronto
22
Outline Outline
Burn Probability (BP) ModelBurn Probability (BP) Model
BP Model ApplicationBP Model Application Predict BP in an Extreme Fire Weather YearPredict BP in an Extreme Fire Weather Year
DiscussionsDiscussions Possible Applications, Limitations & Future Possible Applications, Limitations & Future
ResearchResearch
33
Burn Probability of Next Fire SeasonBurn Probability of Next Fire Season
44
Burn Probability Burn Probability CalculationCalculation
Forest Forest Burn ProbabilityBurn Probability of next fire of next fire season at location(i,j)season at location(i,j)
BPBPxyxy: Burn probability at location (x,y): Burn probability at location (x,y) N: number of years(iterations)N: number of years(iterations) NNxy: xy: number of times of having been burned at number of times of having been burned at location(i,j)location(i,j)
N
NBP xy
Nxy Lim
55
Main Factors affecting Main Factors affecting BPBP
Fuel Type
Fire Occurrence
Topography
(elevation, slopes & slope aspects)
Weather
Level of Protection
Fire
Spr
ead
66
Burn Probability ModelBurn Probability Model
Fuel Type
Fire Occurrence
Level of Protection
Fire Spread
SPATIAL
Daily Weather
Burn Probability
77
Fuel Type Fuel Type ClassificationClassification
The Canadian Forest Fire Behavior Prediction (FBP) System is used
88
Burn Probability ModelBurn Probability Model
Fuel Type
Fire Occurrence
Level of Protection
Fire Spread
Weather
Burn Probability
Fire Occurrence
99
Fire OccurrenceFire Occurrence That total number of fires will occur each That total number of fires will occur each
year follows a Poisson distribution with an year follows a Poisson distribution with an average number equal to historical average average number equal to historical average number of fires in this landscape.number of fires in this landscape.
The conditions that cause past ignition The conditions that cause past ignition pattern will continue in the next fire seasonpattern will continue in the next fire season
affected by the fuel at the location and the affected by the fuel at the location and the weather condition at the time of ignition.weather condition at the time of ignition.
Ignition Patterns differ by causeIgnition Patterns differ by cause
1010
Fire ignition patterns Fire ignition patterns (density)(density)
People-caused and Lightning-People-caused and Lightning-caused density mapscaused density maps
1111
Burn Probability ModelBurn Probability Model
Fuel Type
Fire Occurrence
Level of Protection
Fire Spread
Weather
Burn ProbabilityLevel of Protection
1212
Level of ProtectionLevel of Protection
Percent of forest fires controlled at initial Percent of forest fires controlled at initial attack (IA)attack (IA)
If a fire is controlled, it only “burns” one If a fire is controlled, it only “burns” one cell of the landscape. Otherwise it cell of the landscape. Otherwise it escapes IA and we used a fire growth escapes IA and we used a fire growth model to “spread” it.model to “spread” it.
Escape Index: (EI)Escape Index: (EI) HFI is the Head Fire Intensity (kW/m)HFI is the Head Fire Intensity (kW/m) RT is response time (hours)RT is response time (hours)
RTHFIEI
1313
Spatially Different Spatially Different Response Time to Forest Response Time to Forest
FiresFires
10 hours2 hours
3 hours
4 hours
1414
Burn Probability ModelBurn Probability Model
Fuel Type
Fire Occurrence
Level of Protection
Weather
Burn Probability
Fire Spread
1515
Fire SpreadFire Spread
The escaped fires are simulated by The escaped fires are simulated by using using WildfireWildfire program. program.
Wildfire is a fire growth model that Wildfire is a fire growth model that incorporates GIS data, FBP System incorporates GIS data, FBP System calculations and weather data to calculations and weather data to estimate patterns of hourly fire estimate patterns of hourly fire perimeters. (Todd 1999)perimeters. (Todd 1999)
1616
Burn Probability ModelBurn Probability Model
Fuel Type
Fire Occurrence
Level of Protection
Fire Spread
Weather
Burn Probability
Weather
1717
WeatherWeather
Daily historical weather dataDaily historical weather data Each record includes:Each record includes:
Temperature, Relative Humidity, Temperature, Relative Humidity, Wind speed, wind direction, rain Wind speed, wind direction, rain fall, FFMC, DMC, DC, BUI, ISIfall, FFMC, DMC, DC, BUI, ISI
Data from more than one Data from more than one station can be used. station can be used.
1818
Output of the ModelOutput of the Model
BP Maps
Fire Information
Burn FractionsBP
Model
Other
1919
Application of BP Model in an Application of BP Model in an Extreme Fire Weather YearExtreme Fire Weather Year
Study area
Application of the Model
2020
Location of Location of Romeo Malette Forest Romeo Malette Forest
(RMF)(RMF)
2121
FBP Fuel TypesFBP Fuel Types
2,028,224 ha
C114%
C225%
C3 7%Other
3%
Mixedwood32%
NonFuel8%
Water11%
C1
C2
C3
Other
Mixedw ood
NonFuel
Water
2222
Fire HistoryFire History
From 1976 to 1999 there are 909 fires.
The average is 37.875/year.
The average area burned a year is 1136.15 ha.
2323
Historical Ignition PatternsHistorical Ignition Patterns
2424
Annual Burn ProbabilityAnnual Burn Probability
0.0567%
2525
Burn Probability in an Extreme Burn Probability in an Extreme Weather YearWeather Year
What is an EXTREME Fire Weather Year?What is an EXTREME Fire Weather Year? The year that has most ESCAPED firesThe year that has most ESCAPED fires
Burned more areaBurned more area
1991 1991 Average Number -75 (real number)Average Number -75 (real number)
Daily WeatherDaily Weather
LOP - 90.7% (Uniform response time)LOP - 90.7% (Uniform response time)
2626
Burn Probability under Extreme Burn Probability under Extreme Weather Conditions (1991)Weather Conditions (1991)
0. 258%
2727
Burn Fractions by Fuel TypeBurn Fractions by Fuel Type
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
C1 C2 C3 C4 C5 D1 S1O1a
M12
_25
M12
_50
M12
_75
NonFue
l
Water
Total
FBP Fuel Type
Bu
rn F
ract
ion
s (%
)
2828
Applications of Burn Probability Model
BP ModelFuel
Ignition
Response time(LOP)
Assessment &Decision
FireSmartHarvest
People-caused ignition control
Fuel Management
FireSmart Roads
2929
Used in other modelsUsed in other models
Forest Management Models Forest Management Models (FireSmart)(FireSmart)
Burn Probability by StandsBurn Probability by Stands
Burn Fractions by species, standBurn Fractions by species, stand
WUI Fire Management WUI Fire Management
Wildlife Habitat Suitability AssessmentWildlife Habitat Suitability Assessment
3030
Limitations and Future Limitations and Future WorkWork
Spotting is not included in fire spreadSpotting is not included in fire spread - - Prometheus with spotting capability Prometheus with spotting capability will be used in futurewill be used in future
Use of Regional Climate ModelUse of Regional Climate Model
Produce more BP maps!Produce more BP maps!
3131
Acknowledgement
Kelvin Hirsch, Victor Kafka, Marc-André Parisien, Bernie Todd
Canadian Forest Service, Northern Forestry Center
Jennifer Johnson, Mike B. Wotton, Ana C. Espinoza, Mariam Sanchez G, Justin Podur and Jennifer Beverly
Faculty of Forestry, University of Toronto
Sustainable Forest Management Network
Jim Caputo, Robert McAlpineOntario Ministry of Natural Resources
Tembec
3232
Thanks !Thanks !
Comments & Comments & Questions?Questions?
Comments & Comments & Questions?Questions?