Resort Municipality of Whistler

Wildfire Risk Management System

 

 

 

 

Submitted by

 

 

 

 

 

Submitted to

 

Bruce Hall, Fire Chief

Resort Municipality of Whistler

4325 Blackcomb Way

Whistler, B.C.

V0N 1B4

 

 

November 2005

 

			B.A. Blackwell & Associates Ltd.

 

Table of Contents

 

1   Introduction. 5

2   Wildfire Risk Management 5

3   The Resort Municipality of Whistler and surrounding area. 6

4   Methods. 7

4.1   Overview.. 7

4.2   Development of Probability Theme. 11

4.2.1       Probability of Ignition Component 11

4.2.2       Fire Behaviour Component 14

4.2.3       Suppression Response Capability Component 17

4.3   Development of Consequence Theme. 20

4.3.1       Recreation Use Component 20

4.3.2       Air Quality Component 22

4.3.3       Visual Quality Component 25

4.3.4       Urban Interface Component 26

4.3.5       Biodiversity Component 28

5   Results. 30

5.1   Overview.. 30

5.2   Data Quality Issues. 33

5.2.1       Fuel Typing. 33

6   Applications in the Fire Management Plan for the RMOW... 36

8   Acknowledgements. 37

9   References. 37

Appendix 1:      WRMS model subcomponent rating scales and weights. 39

Probability Component Tables. 39

Consequence Component Tables. 41

Appendix 2: The Wildfire Ignition Probability Prediction System ( WIPPs ) 44

 

Lists of Figures

 

Figure 1.  Conceptual representation of risk assessment/management as the resultant of two factors, Probability and Consequence. 6

Figure 2.  Overview of the study area. 7

Figure 3.  RMOW Wildfire Risk Management System (WRMS) model structure. 8

Figure 4.  Component level rating example: Suppression Response Capability. 9

Figure 5.  RMOW Wildfire Risk Management model interface. 11

Figure 6.  Probability of Ignition component and associated subcomponents. 13

Figure 7.  Fire Behaviour component and associated subcomponents for the 90% percentile July/August weather conditions applying a windspeed of 9 kilometers. 15

Figure 8. Graphic that shows factors affecting spotting. 16

Figure 9. Map showing the polygon assignment of spotting distances based on fuel type, and fire behaviour potential for a 16 kph windspeed. 17

Figure 10.  Suppression response capability component and associated subcomponents. 19

Figure 11.  Recreation Use component and associated subcomponents. 21

Figure 12.  Air Quality component and associated subcomponents. 24

Figure 13.  Visual Quality component and associated subcomponents. 26

Figure 14.  Urban Interface component and associated subcomponents. 28

Figure 15.  Biodiversity component and associated subcomponents. 30

Figure 16.  Summary mapping outputs from the RMOW Wildfire Risk Management System. 31

Figure 17.  Final overlay of probability and consequence. 32

Figure 18.  Initialized weights on all components. 33

Figure 19.  Comparison of original MOF fuel typing (top) and updated fuel typing (bottom) for the RMOW    35

 

 

Lists of Tables

 

Table 1. Overview of Methods, Databases and Sub-Models for each Subcomponent of the RMOW Wildfire Risk Management System.. 10

Table 2. Summary of spotting distance applied in the analysis by species and windspeed. 16

 

1          Introduction

 

In 2004, the Resort Municipality of Whistler (RMOW) began the development a Wildfire Risk Management System (WRMS) for the Municipality. Wildfire is a natural disturbance agent in the forests that surround the RMOW and has the potential to negatively impact social and economic stability, and environmental quality.  Historically these areas have been exposed to low frequency (300-600 years), high severity stand replacement fires that have the potential to significantly alter the forests adjacent to Whistler.  Although the probability of large wildfires within this community is generally considered low, the consequences associated with a large wildfire could be devastating. This report documents the methods and results of the WRMS analysis for the RMOW.

 

This project builds on the wildfire threat analysis methodology that was initially pioneered in Australia (Muller 1993, Vodopier and Haswell 1995) and has since been adapted for use in British Columbia in a number of different contexts and scales (Hawkes and Beck 1997, Blackwell et al. 2003).  In previous applications, all fire related factors (fire risk, suppression response capability, fire behaviour, and values at risk) were related equally without consideration of formal risk management theory.  The revised system developed for this project adopts a risk management approach to guide the quantification of separate and discrete landscape-level probability and consequence ratings, using the same underlying data attributes.  As part of the development of the new GIS-based WRMS system, a user interface has been incorporated enabling fire managers to develop and analyze fire management scenarios and test real time weather conditions. The resultant WRMS better enables fire and municipal managers to design strategies and tactics for fire management in the RMOW. The WRMS tool has been used to determine the wildfire risk profile of Whistler and forms a spatial foundation on which to build a Community Wildfire Protection Plan.

 

2          Wildfire Risk Management

 

Definitions of the term “risk” and all its derivatives (i.e., risk management, risk assessment, risk evaluation) are inconsistent in the wildfire literature, perhaps as a legacy of the fact that most wildfire research has been broken down into specialty topics such as fire behaviour, fire effects, and fire history/occurrence.  For the purposes of the WRMS, wildfire risk is defined as the probability and consequence of wildfire at a specified location under specified conditions.  This definition is consistent with the generic definition of risk and its derivative terms being adopted in many jurisdictions worldwide (Canadian Standards Association 1997, Council of Standards Australia/New Zealand 1999, International Standards Organization 2002).

 

Analytically, the WRMS approach to wildfire risk assessment provides a spatial characterization of risk based on probability and consequence ratings.  In other words, the WRMS can indicate, at any given location and under specified conditions, what the probability of wildfire occurring is and, for a given wildfire behaviour, what the potential consequences on valued resources are.

 

In other fields of risk management (e.g., hazardous materials management), a single resultant quantification of probability and consequence is often derived mathematically. However, in the case of wildfire risk assessment it has been found (as in Bachmann and Allgower 1998) more useful to keep these elements separate, since they may imply different management approaches spatially.  Figure 1 shows how various combinations of probability and consequence can imply the basic management strategies.  In practice, the implementation of this risk management approach requires a detailed spatial examination of assessment results across a full continuum from low to high ratings.

Figure 1.  Conceptual representation of risk assessment/management as the resultant of two factors, Probability and Consequence

 

3          The Resort Municipality of Whistler and surrounding area

 

The project study area is 111,365 hectares and spans the Highway 99 Sea to Sky corridor from just north of Squamish to just south of Pemberton (Figure 2).  Elevations range from 130 to 2500 meters.  A comprehensive ecological inventory of ecosystems and wildlife habitat exists for the RMOW boundary (Blackwell et. al. 2004).  Forests in the lower elevations include western hemlock, amabilis fir, western red-cedar, and Douglas-fir.  With increasing elevation, yellow cedar and mountain hemlock become dominant tree species.  In the harsh climate of the highest elevations, vegetation consists of herbs, lichens, and scattered low alpine shrubs and trees.

 

Wildfire is a natural disturbance agent in a portion of this heavily forested, coastal landscape. Historically these areas have been exposed to low frequency (300-600 years), high severity stand replacement fires (Green et al. 1998).  Although the probability of large wildfires within the study area is considered generally low, the consequences associated with a large wildfire could be devastating to both the RMOW and the adjacent urban interface communities along the corridor.  Air quality, urban interface, recreation use, visual quality and biodiversity are important values that must be considered in a wildfire risk assessment of the RMOW and surrounding area.

 

Figure 2.  Overview of the study area

 

In recent years, fire management within the study area has focused on initial attack and all wildfires have been actively suppressed.

4          Methods

4.1       Overview

 

The purpose of this WRMS was to create a spatial representation of all factors that influence the probability and consequence of wildfire in the study area.  The basic model structure was similar to the one used in 2004 in the Greater Vancouver Watersheds (GVWD); it was further developed and refined through a series of workshops with RMOW Staff.  The model was implemented in a GIS environment using ArcMap 8.2.1 (ESRI) and ArcInfo 8.0.2 (ESRI) using a raster grid at 50m by 50m cell resolution. 

 

The final WRMS model structure is portrayed in Figure 3.  The final spatial probability rating was derived from three major components: Ignition Probability, Fire Behaviour, and Suppression Response Capability.  The final spatial consequence rating was derived from five major components that were significant within the study area: Air Quality, Urban Interface, Recreation Use, Visual Quality and Biodiversity.  Each main model component was in turn derived from several subcomponents as shown in the Figure 3.

 

Figure 3.  RMOW Wildfire Risk Management System (WRMS) model structure.

 

At the subcomponent level, individual ratings for each raster cell were developed on 0-10 scales based on existing biophysical databases and, in some cases, the application of sub-models (e.g., rate of fire spread calculated using the Canadian Fire Behaviour Prediction System and spatial fuel inventory data).  An overview of each subcomponent method, database source and/or sub-model is provided in Table 1.

 

At the component level, the rating for each raster cell was calculated as a weighted sum of all its subcomponents.  Figure 4 provides an example of the rating scales and subcomponent weighting for the Suppression Response Capability component.  All other components were derived in a similar manner (see Appendix 1).  Similarly, at the overall rating level for probability or consequence, the rating for each raster cell was calculated as a weighted sum of all its components.

Figure 4.  Component level rating example: Suppression Response Capability.

 

Table 1. Overview of Methods, Databases and Sub-Models for each Subcomponent of the RMOW Wildfire Risk Management System

	Component	Subcomponent	Overview Method	Database/Sub-Model
Probability Rating	Probability of Ignition	Ignition Potential	Calculation based on fuel type and fire weather indices	·- Wildfire Ignition Probability Predictor1
		Lightning Caused Fire	Inverse distance weighted interpolation of the number of lightning fire ignition points (since 1950) within a 500m buffer	- ESRI Spatial Analyst2 ·- GVWD/Ministry of Forests fire records
		Human Caused Fire	Inverse distance weighted interpolation of the number of human fire ignition points (since 1950) within a 500m buffer	- ESRI Spatial Analyst ·- GVWD/Ministry of Forests fire records
	Fire Behaviour	Fire Intensity	Calculation using fire weather, fuel type and topography	·- Fire Behaviour Predictor 973
		Rate of Spread	Calculation using fire weather, fuel type and topography	·- Fire Behaviour Predictor 97
		Crown Fraction Burned	Calculation using fire weather, fuel type and topography	·- Fire Behaviour Predictor 97
	Suppression Response Capability	Constraints to Detection	Average elevation above valley bottom of forest inventory polygon	·- TRIM
		Proximity to Water Sources	Buffer distance from determinant streams and lakes	- TRIM
		Air Tanker Arrival Time	Measured flight time (concentric) from air tanker base	·- Protection Branch data
		Terrain Steepness	Average slope of forest inventory polygon	·- TRIM
		Proximity to Roads/Helipads	Buffer distance from roads, helipads, and alpine tundra/parkland	·- TRIM and RMOW spatial data
Consequence Rating	Recreation Use	Parks	Provincial and municipal park boundaries	·- RMOW and BC Parks spatial data
		Special Features	100m buffer around feature	·- RMOW spatial data
	Air Quality	Proximity to Population	Buffer distance from urban interface	·- TRIM
		Smoke Production Potential	Smoke production as a function of seral stage	·- TRIM
		Smoke Venting Potential	Average elevation above valley floor of forest inventory polygon	·- TRIM
		Smoke Venting Index	Smoke dispersion rating based on long-term monthly averages.	·- GVWD’s Ambient Air Analyst
	Visual Quality	Visual Quality	Areas delineated as visually sensitive from local vantage points	·- RMOW existing visual quality rating
	Urban Interface	Interface	Buffer distance from interface areas	·- TRIM
		Infrastructure	Buffer distance from urban interface	·- RMOW & BCTC spatial data
		Watersheds	Buffer distance from transmission lines	- Provincial & RMOW spatial data
	Biodiversity	Red & Blue Listed Elements	Areas containing red and/or blue listed species or ecosystems	·- RMOW TEM & CDC data
		Protected Area Network	Spatial dataset created from terrestrial ecosystem mapping	·- RMOW TEM
		Other High Value Biodiversity Areas	100m buffer around feature	- RMOW spatial data

¹FORTester v1.0 (Canadian Forest Service 2002);  ²ESRI Spatial Analyst 8.1.2 (ESRI 2001);  ³Fire Behaviour Predictor 97 (Remsoft, 1997)

From a functionality perspective, the WRMS interface provides several unique features (Figure 5).  Managers are able to turn on or off various subcomponents or components in the model, which allows the quick and efficient exploration of model interdependencies and the ability to analyze individual spatial ratings.  There is also the ability to enter and test various combinations of weights at both the component and subcomponent level, which allows the exploration of specific fire management questions (e.g., what if we increase our suppression resources) and the systematic sensitivity testing of underlying professional judgements that went into the model structure (see Section 4 below).

 

Finally, the model was developed with four user-input functions to support specific analytical requirements.  Users can select the 1:00 pm windspeed and direction in combination with the calculated daily FFMC and BUI obtained from local fire weather stations to evaluate the current a daily fire risk profile for the Municipality.

 

Figure 5.  RMOW Wildfire Risk Management model interface.

 

4.2       Development of Probability Theme

4.2.1    Probability of Ignition Component

 

The probability of ignition component was divided into three subcomponents: fires caused by lightning, fires caused by human activity and ignition potential (Figure 6). The subcomponent rating scales and assigned initial weights are shown in Appendix 1.

 

Lightning and Human Caused Fire

 

The first two subcomponents, lightning and human caused fires, were based on historical fire frequency and cause in the study area from 1950 to 2004.  Fire history records from the Ministry of Forest Protection Branch were translated into spatial points within the GIS framework.  Five hundred meter radius buffers were then created around every fire location point.  This buffer distance was chosen because some older fire location data was only considered accurate to the nearest kilometer and represented fire ignition origin, and not fire perimeter.  The number of fire location points within these new buffer polygons was totaled.  ESRI Spatial Analyst (2001) was then used to determine the final probability of ignition through the application of inverse distance weighted interpolation.  The purpose of interpolation was to predict the value of cells that lack actual points. The simplest form of inverse distance weighted interpolation is sometimes called "Shepard's method" (Shepard 1968). The equation used is as follows:

 

where: n is the number of scatter points in the set; f_i are the prescribed function values at the scatter points (e.g. the data set values), and; w_i are the weight functions assigned to each scatter point.

The classical form of the weight function is:

 

where: p is an arbitrary positive real number called the power parameter (typically, p=2), and; h_i is the distance from the scatter point to the interpolation point, or

where: (x,y) are the coordinates of the interpolation point, and; (x_i,y_i) are the coordinates of each scatter point. 

The weight function varies from a value of unity at the scatter point to a value-approaching zero as the distance from the scatter point increases.  The weight functions are normalized so that the weights sum to unity.

The effect of the weight function was that the surface interpolated each scatter point and was influenced most strongly between scatter points by the points closest to the point being interpolated.

Ignition Potential

 

The third subcomponent, ignition potential, was an indicator of the potential for fire ignition based on fuel type and 90^th percentile fire weather conditions (historic fire weather representing 90% of the most extreme conditions recorded).  It was calculated using the Wildfire Ignition Probability Predictor (WIPP), a tool from FORTester v1.0 (Lawson et al. 1993, Bernie Todd personal communication.). The model determined the probability of sustained ignition from simulated people-caused fire brands (matches and camp fires) and predicted, in broad classes (“no-fire day” less than 50% probability of sustained ignition and “fire day” greater than 50% probability), from readily available indicators of fire danger based on benchmark fuel type groups applicable to British Columbia (Appendix 2).  Ignition probabilities expressed on an area basis provided a measure of people-caused fire potential from simple fire danger rating system components.

 

Figure 6.  Probability of Ignition component and associated subcomponents.

 

4.2.2    Fire Behaviour Component

 

The WRMS model developed for the RMOW was designed to accept a range of fire weather conditions. The fire behaviour component estimated how wildfire would behave under historic weather conditions that have occurred over the recorded climate record for the Municipality.  Information was compiled that related stand-level fuel types, slope, aspect, and fire weather for the study area.  The resulting data was processed through the FBP97 (Fire Behaviour Predictor 97) program.  Fire Behaviour Predictor 97 is a Windows™ based version of the Canadian Fire Behaviour Prediction System (Forestry Canada 1992) developed by Remsoft Inc.  The fire behaviour outputs of FBP97 include: fire intensity; rate of spread, and; crown fraction burned. These outputs form the subcomponents of the fire behaviour component (Figure 7).

 

The Canadian Fire Behaviour Prediction System uses 16 national benchmark fuel types to predict fire behaviour.  For the WRMS, seven of the 16 fuel types were selected to estimate fire behaviour based on species composition and stand structure attributes.  The provincial fuel type database was adjusted to reflect changes in forest cover over the past eight years (since 1997) and to correct fuel typing areas that match with fuel types verified by both field checking and aerial photography review.

 

Weather information was derived from historic records collected from five weather stations associated with the study area (station #1040390, 1040400, 1040420, 1048898 and 1108988. Depending on the element measured, the period of record was 1931 to 2005. Data for temperature and precipitation was only continuous from 1950.  A look up table, with computed fire weather indices summarized by station and Biogeoclimatic Unit, was developed specifically for the RMOW. This look-up table allows computation of fire percentiles for all possible permutations and combinations of fire weather indices for the period of record.

 

Fire weather data (temperature, relative humidity, precipitation, and wind speed) was used to calculate Fine Fuel Moisture Code (FFMC) and Build-Up Index (BUI).  Fire behaviour was subsequently modeled in FBP97 using upslope winds calculated from the relevant aspect. The subcomponent rating scales and assigned initial weights are shown in Appendix 1.

 

Fire Intensity

 

The fire intensity subcomponent was a measure of the rate of heat energy released per unit time per unit length of fire front.  It was based on the rate of spread and predicted fuel consumption of the fire, and was expressed in kilowatts per meter (Pyne 1984).

 

Rate of Spread

 

The rate of spread subcomponent was a measure of the speed at which fire expands its horizontal dimensions at the head of the fire.  This was based on the hourly Initial Spread Index (ISI) value and was expressed in meters per minute.  The rate of spread was adjusted for steepness of slope and interactions between slope direction and wind direction determined from the Build-Up Index (BUI).

 

Crown Fraction Burned

 

The crown fraction burned subcomponent was a measure of the proportion of the tree crowns consumed by fire and was expressed as a percentage value.  It was based on rate of spread, crown base height and foliar moisture content.

 

Figure 7.  Fire Behaviour component and associated subcomponents for the 90% percentile July/August weather conditions applying a windspeed of 9 kilometers

 

Calculation of Spotting Distances

 

The calculation of spotting distance for individual forest polygons was based on the predictive spotting models contained within BEHAVE (USDA Forest Fire Behaviour prediction software). Spotting models were originally devised to predict the maximum distance burning embers would travel over flat and regularly undulating terrain. The balance between particle size, burnout rate, and time or distance traveled determines maximum spotting distance (Figure 8). Smaller particles are lofted higher and transported further, but burnout sooner than larger particles.

 

Forest polygon size was an important consideration in determining the threshold of fuel necessary to create spotting. For the purpose of this analysis within the urban area, forest polygons (parks and greenways) less than 10 hectares in area were not included in the spotting assessment. For forest polygons outside of the urban area, areas less than 20 hectares were not included in the assessment.  The purpose of this analysis was to compute the maximum spotting distance over complex landscapes, for a given windspeed and fuel type, that particles of different sizes would travel (Table 2, Figure 9). The spotting distance across the interface assumed that wind direction was down slope and into the urban area. In general, it is believed that these models are conservative and underestimate the actual spotting distances under conditions of running crown fire.

 

Figure 8. Graphic that shows factors affecting spotting.

 

Table 2. Summary of spotting distance applied in the analysis by species and windspeed.

 

 

Figure 9. Map showing the polygon assignment of spotting distances based on fuel type, and fire behaviour potential for a 16 kph windspeed.

 

4.2.3    Suppression Response Capability Component

 

Ability to suppress wildfire was dependent on the speed of detection, terrain, accessibility and availability of resources.  Five subcomponents were used to determine overall suppression response capability.  These included constraints to detection, proximity to water sources, air tanker arrival time, steepness of terrain, and proximity to roads and helipads (Figure 10). The subcomponent rating scales and assigned initial weights are shown in Appendix 1.

 

Constraints to Detection

 

In British Columbia, fires are detected by three primary methods that include a provincial lightning location system, aircraft, and or by the public.  Due to the unpredictability of flight frequency and public response, it was not possible to quantify the speed of detection.  Detection is primarily a function of visibility limitations associated with high elevation cloud in specific parts of the study area. A storm front with varying amounts of precipitation typically follows an active lightning period. This storm front creates cloud and fog within higher elevations zones of the study area during a 12 to 24 hour period following the storm. This cloud and fog cover inhibits the critical detection period; since most fire ignitions within the study area occur during the transition from a high to low-pressure weather system. The constraints to detection subcomponent were therefore based on elevation classes.  The higher the elevation, the more likely detection will be constrained by cloud and fog cover.  Elevation classes were assigned in increments of 500 m and were measured from the valley bottom of the RMOW.  Elevations greater then 1000 m were given the highest rating.

 

Proximity to Water Sources

 

Proximity to water sources was delineated using the hydrological base and only included determinant (perennial) water sources.  Proximity to water sources for fire suppression (an indicator of the ability to access water quickly for fire fighting) was evaluated by creating a 100 m and 300 m buffer around all determinant rivers, creeks and lakes.  Areas outside of the 300 m buffer were given the maximum subcomponent rating.

 

Air Tanker Arrival Time

 

The air tanker arrival time subcomponent was determined based on the distance from the closest air tanker base to the study area, the Abbotsford base.  The ratings increased with greater distance from the base.

 

Terrain Steepness

 

Steepness of terrain influences the ability of a ground crew to build fireguards and carry out ground suppression.  Average slope class was determined from the terrain data and ratings were assigned according to slope class. 

 

Proximity to Roads

 

Proximity to roads was used to evaluate the accessibility of suppression resources reaching areas within a given landscape unit.  It was evaluated based on a bush-walking rate of 1 km/h.  Proximity to roads and helipads was rated by creating buffers around all roads and helipads in the study area and assigning weights relative to walking time from these areas.  Alpine tundra was included as area accessible by helicopter.

Figure 10.  Suppression response capability component and associated subcomponents.

 

4.3       Development of Consequence Theme

4.3.1    Recreation Use Component

 

Providing recreation opportunity is an important mandate of the RMOW.  Although the probability of lightning caused forest fires within the study area is considered low, human caused fires present a substantial threat to the community. Overall, the consequence of fire impact on recreation use would be considerable.

 

The recreation use consequence component was developed using information on two recreation-related sub-components that encompassed important areas for recreation (Figure 11). The subcomponent rating scales and assigned initial weights are shown in Appendix 1.

 

Parks

 

All municipal parks within the RMOW (Alpha Lake, Balsam, Brio, Dream river, Emerald, Eva Lake, Fits Creek Skateboard, Fitzsimmons Creek, Green Lake, Lakeside, Lost Lake, Meadow, Millar’s Pond, Myrtle Philip School, Rainbow, Snowflake, Spruce Grove and Wayside Parks) were given the maximum rating of 10, while Provincial Parks (Blackcomb Glacier, Brandywine Falls, Callaghan Lake, Garibaldi, and Nairn Falls) were given a rating of five.

 

Special Features

 

Several important recreation areas outside of the municipal boundary would be put in jeopardy by wildfire.  These include the ancient cedars near the trophy lakes, Alexander Falls, the Interpretive Forest, the kayak hotspots along the Cheakamus River, and the section of Green River, from Green Lake to Wedge Creek, used for white water rafting.  All areas were buffered by 100m and given the maximum weighting.

 

Figure 11.  Recreation Use component and associated subcomponents.

 

4.3.2    Air Quality Component

 

Wildfire within, and or, adjacent to the RMOW has the potential to substantially impact the air quality of the entire valley bottom within the community. Heavy wildfire caused smoke could force a large-scale evacuation of the RMOW lasting several days to a week. Smoke related air pollution is not a problem that can be confined to one location; it must be examined at a broader, landscape level.  Because the forest landscape is in close proximity to the populated areas of the community, smoke and forest fire related emissions have the potential to notably impact regional air quality. The air quality component of the WRMS system was developed considering a number of related factors including proximity to population, smoke production potential, and smoke venting potential (Figure 12). 

 

The WRMS system is considered useful for identification of potential air quality impacts of wildfires.  However, given the complex topography of the area, actual air quality impacts from wildfires are difficult to accurately predict without detailed knowledge of airflow and other atmospheric parameters (i.e. stability and mixing height) in the region, particularly near the areas of smoke release and the surrounding airshed. The subcomponent rating scales and assigned initial weights are shown in Appendix 1.

 

Proximity to Population Centers

 

The proximity to population subcomponent was based on distance to population centers (urban interface).  The ratings in this subcomponent were assigned with the assumption that wildfire in close proximity to residential areas would have more potential to impact air quality (with smoke emissions, ash and embers) than wildfire occurring far from residential areas. 

 

Smoke Production Potential

 

Smoke production is based on several factors including the moisture content of the fuel, the heat of combustion and, most importantly, the amount of fuel present on a given site.  Available biomass (a function of structural stage) was used as a surrogate for smoke production potential.  It was assumed that higher amounts of biomass (forest floor and dead and living vegetation) contributed to increased amounts of smoke production.  Smoke production potential was greatest in old forest of the Coastal Western Hemlock (CWH) zone, followed by young forest, old forest of the Mountain Hemlock (MH) and Engelmann Spruce –Subalpine fir (ESSF) zones, and finally, shrub herb.  Old forest in the CWH was delineated from the ESSF and MH old forest because the amount of available biomass that contributes to flaming combustion is substantially lower in the ESSF and MH zones compared to the CWH zone. 

 

Smoke Venting Potential

 

The ability of the atmosphere to disperse and transport smoke is commonly estimated using the ventilation index (VI), which is forecast daily by Environment Canada..  Smoke venting potential is an indicator of potential smoke dispersion based on mixing height during poor VI days.  Within the RMOW WRMS, the smoke venting potential was rated as a function of elevation; where higher elevations had a higher smoke venting potential than lower elevations.  Typically, fires that are sufficiently upslope of the valley bottom have a greater likelihood of transporting the smoke plume above the mixed layer and or the valley re-circulations, thereby allowing smoke to be mixed to higher elevations without being transported down the valley into nearby communities.

 

Monthly Smoke Venting Index

 

On any given day any range of ventilation conditions can occur, however, there is some seasonality to the ventilation index that makes the occurrence of good to poor ventilation index days more likely depending on the time of year.  This subcomponent was included in the air quality component to provide a relative monthly comparison of smoke venting potential.  During the fire season, September and October have poor venting conditions compared to May and June when the venting index is generally good.  For the hotter months of July and August, smoke venting potential is average compared to other times during the year. 

 

 

Figure 12.  Air Quality component and associated subcomponents.

 

4.3.3    Visual Quality Component

 

Visual quality within the RMOW is considered fundamental to the maintenance and integrity of the Community of Whistler as a world-class resort. Large-scale fire has the potential to blacken much of the landscape, which would impair visual quality and therefore impact the resort experience. The visual quality component provided a rating of the impact of a fire on visual quality from regional and local vantage points in the RMOW.  Visual quality information was obtained from the Ministry of Forests, local licensees and the Whistler Interpretative Forest.  The High Elevation component of the Protected Area Network (PAN) was also included as a visually sensitive area.  Any areas delineated as a visual sensitivity units were given the maximum rating of ten and those areas not designated as visually sensitive were given a rating of zero (Figure 13).

 

Figure 13.  Visual Quality component and associated subcomponents.

 

4.3.4    Urban Interface Component

 

The Urban Interface component provided a rating of the potential for fire to pose a direct threat to people and property located in and around the RMOW.  It contained three subcomponents: interface, infrastructure and watersheds (Figure 14). The subcomponent rating scales and assigned initial weights are shown in Appendix 1.

 

Interface

 

The interface subcomponent was an indicator of threat to property and was based on structure density determined using TRIM. All anthological building features were extracted from this data set and buffered such that structural classes could be assigned based on their density on the landscape. It was recognized that interface wildfire risk is influenced by other factors out of RMOW control, including building materials (i.e. unrated roofing materials), defensible space around structures, access, water availability and vegetation within the proximity of homes.

 

Interface density classes were delineated as follows:

Undeveloped 0-1 structures/km²

Isolated =1-10 structures/km²

Mixed = 10-100 structures/km²

Developed = 100-1000 structures/km²

Urban = > 1000 structures/km²

 

The urban class was assigned a maximum rating, while area with no structures was assigned a rating of zero.

 

Infrastructure

 

The infrastructure subcomponent was an indicator of fire risk to key infrastructure within the RMOW.  Only infrastructure of immediate importance during a wildfire was included.  For example, the hydro right-of-way was included because the shutdown of this transmission line could cause considerable economic loss and or unacceptable social consequences to all communities along the sea to sky corridor and in areas of the lower mainland.  Key infrastructure was comprised of the health care center, reservoir pumping stations, fuel storage, the hydro right-of-way, and the fire hall.  All point locations were buffered by 500m and given maximum ratings.

 

Watersheds

 

The watershed subcomponent was developed as an indicator of the risk to water supply and water quality. Whistler’s water supply is serviced by surface water from key drainages adjacent to the Municipality. Given the importance of water to the community, watersheds within the study area were given maximum ratings.

 

Figure 14.  Urban Interface component and associated subcomponents.

 

4.3.5    Biodiversity Component

 

The ecosystem integrity component was developed using three subcomponents; red and blue listed elements (Conservation Data Center, Victoria), Whistler’s Protected Area Network (PAN) (RMOW 2005) and other high value biodiversity areas identified by Municipal staff (Figure 15). The subcomponent rating scales and assigned initial weights are shown in Appendix 1.

 

Red and Blue Listed Elements

 

The area outside the RMOW but inside the study area contained only two noted locations of one element at risk; the blue listed nodding semaphoregrass (Pleuropogon refractus).  However, these data are considered incomplete.  There is a high probability that the area contains other important habitat for other elements at risk. Within the RMOW, Whistler Terrestrial Ecosystem Map (TEM) data (B.A. Blackwell and Associates Ltd. 2004) was used to determine locations of important habitat at risk within the RMOW.  For wildfire, this included one red listed species Keen's long-eared myotis, and five blue listed species; coastal tailed frog, red-legged frogs, great blue heron, bull trout and Dolly Varden trout.  TEM polygons rated as potential habitat for at risk plants were also included.  Red listed species were given maximum ratings and blue listed species were given a rating of five.

 

Protected Area Network

 

Through the Whistler Environmental Strategy, the RMOW has established a Protected Area Network (PAN). The strategic goal of the PAN is to protect, within Municipal boundaries, an ecologically viable network of critical areas (Whistler Environmental Strategy 2004). Utilizing information on ecosystems from the TEM inventory, the Municipality has identified and protected unique and sensitive habitats such as streams, lakes, wetlands, old growth forests, alluvial forests, riparian areas, and the corridors connecting them. The lower elevation PAN areas are ranked above  higher elevation PAN areas; primarily due to the limited area of important ecosystems in the valley bottom and their relative rareness compared to higher elevation protected areas. A summary of the ranking can be found in Appendix 1.

 

Other High Value Biodiversity Areas

 

Several other high value biodiversity areas outside of the RMOW were included to account for the absence of TEM data.  These included the Soo and Pinecrest wetlands, Cougar Mountain, the riparian train wreck, identified deer winter range, important stream networks within the interpretive forest and spotted owl SRMZs. All were given maximum ratings within the WRMS framework.

 

Figure 15.  Biodiversity component and associated subcomponents.

 

5          Results

5.1       Overview

 

A schematic compilation of mapping outputs from the initial implementation of the RMOW Wildfire Risk Management System is presented in Figure 16. The mapping outputs parallel the description of the model in the previous section. In other words:

·         Subcomponents maps are generated using 0-10 rating scales derived from existing GIS databases and/or sub-model outputs;

·         Component maps are generated using user-defined weights on each subcomponent (see Appendix 1);

·         Final probability and consequence rating maps are generated using user-defined weights on each component (Figure 16), and;

·         A final probability/consequence overlay map is generated by overlaying the final rating maps (Figure 17).

Figure 16.  Summary mapping outputs from the RMOW Wildfire Risk Management System.

Figure 17.  Final overlay of probability and consequence

 

As shown in Figure 16 and all of the maps in Section 4, each component and subcomponent map applies a similar white-green-yellow-orange-red colour scheme depicting ratings on 0 – 10 scales. An expanded color scheme was used to show all probability/consequence combinations for the final interpretation. In this manner the final probability/consequence overlay map in Figure 17 reflects the full range of risk spatially, within the municipality, from ‘low probability- high consequence’ areas through to ‘high probability-low consequence’ areas for extreme fire conditions.

 

These final mapping outputs are the result of multiple interactive workshops, during which the project team evaluated the accuracy and consistency of each subcomponent and component.  The initial weights used to generate these outputs at the component level are shown in Figure 18. 

 

In overview, the area of highest consequence is located within the interface areas of the RMOW.  This was expected given the identified values at risk.  In terms of wildfire probability, there is a relatively large area of moderate to high fire probability in the valley bottom where the town of Whistler is located. This probability declines with increasing elevation out of the valley bottom.  These probability ratings are driven largely by human ignitions and the fire behavior potential of 30 to 40 year old second growth plantations located within the valley bottom. The probabilities of ignition and fire behavior are offset by good suppression capability afforded by roads, water sources and the gentle terrain associated with the valley bottom.

 

Figure 18.  Initialized weights on all components.

 

5.2               Data Quality Issues

 

The data provided by the TEM inventory was fundamental to the development of many of the underlying spatial GIS databases.  The TEM inventory databases included substantial validation efforts and therefore we are confident in the accuracy of this data.

 

However, one concern directly related to the fire management data used as part of this project was fuel typing. Since these data sources are fundamental to the development of the fire behaviour themes, we expand on these concerns below.

5.2.1    Fuel Typing

 

As part of the provincial fuel type classification program, the Ministry of Forests Protection Branch completed fuel typing of the RMOW (Hawkes et. al. 1995).  This classification applies the Canadian Fire Behavior Prediction (FBP) System fuel type classification using a detailed algorithm that relates specific attributes of standard forest cover inventory data to specific fuel types within the FBP classification scheme (Taylor et. al. 1997).  The fundamental concern when applying the FBP system in a coastal application relates to associating classical boreal fuel types with temperate coastal forests. 

 

For some of the fuel types present in the watershed there is a reasonable fit with FBP types. For example, we used C4 for pole sapling forests and C3 for young forest, which have worked well.  The qualitative attributes of these FBP fuel types are similar and representative of the structural attributes present within these forest types of the RMOW.

 

However, for other fuel types, the relationship is considered poor.  In particular, the old forests surrounding the RMOW, which represent a significant portion of the total area, do not correspond well with any of the FBP types. Applying the provincial fuels algorithm to the RMOW results in large parts of the old growth landscape being classified as an M2 fuel type.  From previous work in the Vancouver watersheds, we have established that, where M2 has been applied, modeled fire behavior does not correspond well with wildfire observations.  Typically, in all the scenarios where this fuel type classification is used, the model appears to over-predict fire behaviour in terms of high fire intensity and high rates of spread.  It is suspected that the reason for this over prediction of fire behavior components is because M2 is based on the C2 fuel type (Boreal Black and White Spruce) with a reduced percentage of deciduous. This C2 fuel type, under similar weather conditions, has some of the highest rate of spread, fire intensity, and crown fraction burn measures when compared to other FBP fuel types. In discussion with CFS and the Ministry of Forests Protection Branch fire behaviour specialists, it was determined that C5 should be substituted for M2 in an attempt to alter these fire behaviour outputs to levels considered more realistic for both the weather and fuel conditions present within the watersheds.  The substitution of C5 for M2 resulted in an improved result, particularly for the old forest types. Based on this information, C5 was substituted for M2 in the RMOW. Figure 19 provides a comparison of fire behaviour outputs using M2 and C5 fuels types for the same weather conditions.

 

In addition to the issue identified for M2, the provincial fuel type inventory incorrectly classified a significant area of forest plantation as deciduous fuel type D1. This was primarily the result of changes in the forest cover between 1997 and the present. The original classification was based on a 1997 inventory, at which time many of the plantations were dominated by a shrub and deciduous component. Over the past eight years, many of these plantations have become dominated by coniferous species resulting in a shift from a deciduous fuel type to a coniferous fuel type. This change was identified as part of ground and data checking that occurred during the project.

 

 

 

 

Figure 19.  Comparison of original MOF fuel typing (top) and updated fuel typing (bottom) for the RMOW

 

6          Applications in the Fire Management Plan for the RMOW

 

The development of the RMOW WWRMS has benefited from the collaborative approach of the Municipality. This translated to a willingness to learn and explore the fire risk elements that could impact the Municipality.  The process has allowed interested parties to participate in the various phases of model development and has created a dynamic learning environment for overall fire management planning. The model provides useful outputs to assist in developing strategic fire management strategies. 

 

The ROMW WRMS provides a comprehensive assessment of the wildfire risk within and adjacent to the community. The assessment can be used to further develop strategic fire management zones for the fire management program as described within the Fire Management Plan.

 

Fire protection resources can undergo a detailed evaluation of suppression response capability. The level of risk, as identified by the fire management zones, can prioritize efficient use of these resources. By improving the fire suppression capability, the risk of wildfire can be reduced. This may require the acquisition of more resources (water delivery systems) or modification of existing practices (helicopter contract response times).

 

Full implementation of the WRMS system within the RMOW will require some level of staff training to fully utilize the capability of the fire risk system, especially if used to develop more advanced strategies (adjusting weightings and input choices). Further work could include the development of “real time” watershed zoning capability with linkages to daily weather and forecasts. Other considerations could include incorporating people and lightning fire occurrence prediction into a “live” system. This would move the WRMS closer to a tactical tool that could be applied in daily operations.

 

As mentioned above, the WRMS identifies information gaps that are not only integral to the model but also to the existing fire management program. Additional collection of the fire weather indices at strategic locations within the watersheds and an effective data management system would improve the existing fire management program.

 

The current WRMS has utilized the most appropriate fuel types from the Canadian Fire Behaviour Prediction System for RMOW lands. Modified stands from fuel hazard treatments (wildland/urban interface) or other disturbance such as insects (mountain pine beetle) produce a unique fuel type in the short-term. The development of new model algorithms is required to properly assess how fire behavior and ignition potential in these stands would change in the future. A change in fire behavior and ignition potential may reflect a higher or lower risk of wildfire.

 

8          Acknowledgements

 

We acknowledge the active participation of Bruce Hall, Don McLaurin, Heather Beresford, Paul Beswetherick, Bill Barrett, and Rob Witton from the Resort Municipality of Whistler.  Judi Beck, BC Ministry of Forests Protection Branch provided provincial fuel typing algorithms.  Fiona Steele of B.A. Blackwell and Associates Ltd. implemented the system.  Claire Tweedsdale of Forest Ecosystem Solution Ltd. did the GIS programming.  Brian Macintosh and Phil Taudin-Chabot (Coastal Fire Center) and Ed Korpela (Regional Fire Specialist) provided helpful commentary during a review of the work.

 

9          References

 

Blackwell, B.A. and D.W. Ohlson. 2004. GVRD Watershed Wildfire Risk Management System. Contract Report to the Greater Vancouver Regional District. 9 pages.

 

Blackwell, B.A., Gray, R.W., Steele, F.M., Needoba, A.J., Green, R.N., and K. MacKenzie. 2003. A wildfire threat rating system for the Birkenhead and Gates Landscape Units, British Columbia. in: R.T. Engstrom and W.J. de Groot (eds.) Proceedings of the 22nd Tall Timbers Fire Ecology Conference: Fire in Temperate, Boreal, and Montane Ecosystems. Tall Timbers Research Station, Tallahassee, FL.

 

Canadian Standards Association, 1997. Risk Management: Guideline for Decision-Makers: A National Standard of Canada. CAN/CSA-Q850-97. Etobicoke, Canada.

Council of Standards Australia / Council of Standards New Zealand, 1999. Risk Management. AZ/NZS 4360:1999. Strathfield, Australia.

 

Green, R.N., B.A. Blackwell, K. Klinka, and J. Dobry. 1998. Partial reconstruction of fire history in the Capilano watershed. Contract report to the Greater Vancouver Water District.

 

Hawkes, B. and Beck, J. 1997. A wildfire threat rating system. 1997. Forest Service, Pacific Forestry Centre, Victoria, BC. Technology Transfer Note 01.

International Standards Organization, 2002. Risk Management Vocabulary, Guidelines for use in Standards. First Edition. ISO/IEC Guide 73:2002(E/F). Geneva, Switzerland.

 

Hawkes, Brad C., David Almost Goodenough, Bruce Lawson, Alan Thomson, Olaf Niemann, Peter Fuglem, Judi Beck, Bryan Bell, and Phil Symington. "Forest Fire Fuel Type Mapping Using GIS and Remote Sensing in British Columbia." GIS `95 Conference Proceedings. Fort Collins: GIS World, Inc., 1995. 2:647-656.

 

Lawson, B.D., O.B. Armitage, and G.N.  Dalrymple. 1993. Ignition probabilities for simulated people-caused fires in British Columbia’s lodgepole pine and white spruce-subalpine fir forests. Paper presented at the 12 International Conference on Fire and Forest Meteorology, Oct. 26-29, 1993, Jekyll Island, Georgia.

 

Muller, C. 1993. Wildfire threat analysis: A decision support system for improved fire

management. Paper presented at "The Burning Question: Fire Management

in NSW", Coffs Harbour, Aug. 1993, UNE, Armidale, NSW, Australia.

 

Pyne, S.J. 1984.  Introduction to Wildland Fire: Forest Management in the U.S.  John Wiley and Sons.  New York, NY.

 

Taylor, S. R.G. Pike, and M.E. Alexander. 1997. Field guide to the Canadian Forest Fire Behavior Prediction (FPB) System. Special Report 11. Fire Management Network. Canadian Forest Service, Northern Forestry Center. 60 p.

 

Vodopier, J.; Haswell, D. 1995. The application of wildfire threat analysis in forests of southwestern Australia. In: BUSHFIRE ’95, Australian Bushfire Conference,

Sept. 27-30, 1995, Hobart, Tasmania (no pagination in proceedings).

 

Appendix 1:    WRMS model subcomponent rating scales and weights

Probability Component Tables

 

Consequence Component Tables

 

 

 

 

 

Appendix 2: The Wildfire Ignition Probability Prediction System ( WIPPs )

 

(1)  Format of the Standard WIPP Equation is :

 

P = 1 / {1 + exp[ B0 + B1*FFMC + B2*DMC + B3*DC + B4*BUI + B5*FWI + B6*ISI ] }

 

(2) Standard Association of FBP Fuel Types and  WIPP Equations:

 

Table 1 provides the suggested standard association of WIPP equation to FBP Fuel types.

 

(3)  Possible Association of WIPP Equations to FBP Fuel Types

 

The option exists to change the choice of the WIPP equation, which is used for each FBP fuel type. The default option, which is the first equation listed, and the subsequent possible options are listed in Table 2. These possible associations are from Lawson and Armitage (1997)

 

(4)  Relationship of WIPP Equations to General Fuel Type and Provincial  Experimental Sites

 

Table 3 details the general fuel types and provincial test sites that were used to create the individual WIPP equations.

           

References

 

Lawson, B.D., O.B. Armitage, and G.N. Dalrymple. 1994a. Ignition probabilities for simulated people-caused fires in B.C.’s lodgepole pine and white spruce-alpine fir forests. Pages 493-505 in Proc.12^th Conf. On Fire & Forest Meteorology. Oct 26-28, 1993. Jekyll Is. GA., Soc. Am. Foresters. Bethesda, MD.

 

Lawson, B.D., O.B. Armitage. 1997. Ignition Probability Equations for some Canadian Fuel Types.  Report submitted to the Canadian Committee on Forest Fire Management. (Draft report ).

 

 

 

Table 1 : Standard Association of FBP Fuel Types and  WIPP Equations

 

 

 

 

               

 

Table 2:  Possible Association of WIPP Equations to FBP Fuel Types

 

     Table 2 :   Possible Association of WIPP Equations to FBP Fuel Types  ( cont’d )

 

 

 

Table 3  :  Relationship of WIPP Equations to General Fuel Type