Appendices for the FIRECLIM project 11/19/12 Appendix A. Procedures for delineating the WUI Delineation of the WUI is based on Stewart et al. (2003; 2007) and the Buffer from Structures method outlined by Platt (2010). Platt’s method uses parcels as the basis for WUI designation, while Stewart et al.’s (2003; 2007) methods are much more widely used. Melding elements from both methods would provide a WUI designation that serves two purposes: (1) Identification of structures at risk from wildfire; and (2) prioritization of forest treatments to reduce wildfire risk. This approach would include the following steps: 1. Remove Flathead and Smith Valley areas from the study area, thus making them unavailable for WUI designation. Use the internal 2010 GCS WUI boundary as the extent of the Flathead and Smith valleys. Parcels separated by the internal boundary of the GCS WUI should be included in the steps below if their parcel centroid is within the WUI boundary. This method provides consistency with existing Flathead County targets for the WUI. 2. Determine structures at a density greater than 1 structure per 40 acres in the CAMA data and RECID2 output for subperiods. a. Assign parcel centroids to any developed parcel in the study area containing a residential structure. The centroid will serve as the location of the residential structure in that parcel. For parcels containing more than one residential structure, randomly assign the additional residential structures to a location in the parcel. b. Calculate the distances between residential structures that are at a density greater than 1 structure per 40 acres, which is the density requirement1. All parcels containing residential structures that meet the distance criterion are included in the WUI (the entire parcel) pending additional steps. 3. Remove LANDFIRE pixels in the study area that do not contain wildland vegetation or human development (i.e. residential structures) (Platt 2010; Wilmer and Applet 2005). 4. Determine presence of wildland vegetation near residential structures. a. Buffer .5 miles around all residential structure centroids that meet the distance criterion described in step 2. The .5 mile buffer is based on the Healthy Forests Restoration Act guidelines for WUI structure/community protection (Theobald and Romme 2007; Stewart et al. 2009). The area of .5-mile buffer is approximately 502 acres. b. Determine the presence and extent of wildland vegetative cover within the residential structure buffers created in step 4a. Buffers containing at least 50% wildland vegetation are designated as potential intermix WUI areas. Residential buffers containing less than 50% wildland vegetation are designated as interface WUI areas if they are within 1.5 miles of an area larger than 1,235 acres that is heavily vegetated. An area is considered heavily vegetated if at least 75% of its area is in wildland vegetation. These definitions are consistent with the existing FIRECLIM WUI methodology. 1Platt (2010) provides the following process for determining structures at a density greater than 1 structure per 40 acres: Starting from the locations of residential structures (i.e. parcel centroids and, if applicable, the random location(s) of additional residential structures in the parcel), determine if any other residential structures are located within 1,890 ft. Structures within 1,890ft of other structures are at a density of greater than 1 structure per 40 acres. 1 5. Impose additional vegetation requirements on potential intermix WUI areas. a. To be included in the intermix WUI, a potential intermix WUI area (as defined in step 4) must be: (1) surrounded by or directly adjacent to a large body of continuous2 wildland vegetation (at least 1,235 acres) containing at least 50% wildland vegetation OR; (2) part of an aggregate of buffers that contain a large body of continuous wildland vegetation totaling at least 1,235 acres (Stewart et al 2007; 2009). Any potential intermix WUI area that does not meet this additional vegetation requirement is not considered part of the WUI. Completion of step 5 will identify all developed residential parcels that are used in the WUI calculations for the FIRECLIM project. No more parcels will be added to the WUI for that subperiod. The full FIRECLIM project also requires identification of WUI areas targeted for forest treatments. Step 7 below provides that identification. Areas identified in step 7 will be used for the simulation of forest treatments by Fire-BGCv2. 6. Impose a 1.5 mile WUI buffer extending from the border of any developed WUI parcel (interface or intermix). a. Only private lands within the 1.5 mile buffer are included in the WUI, not the entire parcel, unless the entire parcel is engulfed during the buffering process. The 1.5 mile buffer applies only to areas that contain wildland vegetation. b. The 1.5 mile buffer cannot extend into the Flathead or Smith valleys. This completes the WUI area designation. 7. Implement additional steps used by GCS Research (2010) to designate the 2010 Flathead WUI. a. Eliminate any gaps in the 1.5 mile WUI area buffer that are less than 40 acres b. “…the external and internal edges of the overall WUI were joined together and made into a continuous edge to reflect their experience with wildland fire in Flathead County.” c. Buffer 1.5 miles from the border of public lands (subject to the valley constraint in step 1) as was conducted in the most recent CARES WUI definition. The WUI concept was created primarily to help identify private structures at risk from wildland fire. Less emphasis is placed on private lands not containing structures. The full FIRECLIM project may seek to identify any private property as part of the WUI regardless of the presence of structures. If project members decide that the full FIRECLIM model should include private lands that do not contain structures, another set of criteria are needed to define such intermix or interface properties. The following steps would be used to assign private property without structures to either the interface or intermix WUI. 1. Parcels that do not contain a residential parcel are assigned a parcel centroid and are subject to the same process as steps 1-5 of the previous section above. References GCS Research. 2010. WUI methodology for Flathead County. Unpublished manuscript. Platt, R.V. 2010. The wildland-urban interface: Evaluating the definition effect. Journal of Forestry, 108(1), 9-15. 2A continuous body of wildland vegetation requires connectivity among LANDFIRE pixels identified as wildland vegetation. 2 Stewart, S.I., Wilmer, B., Hammer, R.B., Aplet, G.H., Hawbaker, T.J., Miller, C. and Radeloff, V.C. (2009). Wildland-urban interface maps vary with purpose and context. Journal of Forestry. 107(2), 78-83. Stewart, S.I., Radeloff, V.C., Hammer, R.B. and Hawbaker, T.J. 2007. Defining the wildland-urban interface. Journal of Forestry, 105(4), 201-207. Stewart, S.I., V.C. Radeloff, and R.B. Hammer. 2003. Characteristics and location of the wildland-urban interface in the United States. In: Proceedings of the Second International Wildland Fire Ecology and Fire Management Workshop, November, 16-20, Orlando, FL. Theobald, D.M., Romme, W.H. 2007. Expansion of the US wildland-urban interface. Landscape and urban planning, 83, 340-354. Wilmer, B., and Aplet, G. 2005. Targeting the community fire planning zone: Mapping matters. The Wilderness Society, Washington, DC. Available at: http://wilderness.org/content/targeting-community-fireplanning-zone-mapping-matters Appendix B. Procedure for prioritizing fire district treatments by landownership, subperiod, and county-wide priority The 2010 Flathead County CWPP process included collaborative designation of county-wide and individual fire district “priority areas” for wildfire risk and fuels reduction. Some fire district priority areas are nested within the county-wide priority areas and others are not (see attached map). The full FIRECLIM model will use both countywide and fire district priority areas to determine where fuels treatment will occur during subperiods. Countywide priority areas include five categories of decreasing importance: (1) Extreme; (2) High; (3) Medium High; (4) Medium and; (5) Medium Low. Individual fire district priority areas are ranked in order of importance for fuel reduction and can be found in the “Flathead County Community Wildfire Fuels Reduction/Mitigation Plan” document beginning on page 45. The full FIRECLIM model will include three forest treatment scenarios. The number of forest treatment scenarios in the full model was decreased from three to five following recent FIRECLIM project meetings. Remaining forest treatments for the full model include: 1. CWPP priority areas only 2. WUI areas (includes CWPP areas) 3 WUI, CWPP and non-WUI areas Prioritization for scenario 1 (CWPP priority areas) Step 1 Prioritization in forest treatment scenario #1 will use the ranking procedure outlined in Appendix B of the test case, however, treatments in different priority rankings will be concurrently allocated forest treatments acres using a weighted system. A disproportionate amount of available treatment acres will be randomly allocated to CWPP areas based on the county-wide priority zone they exist in or intersect (i.e. “extreme,” “high,” “medium high,” “medium,” and “medium low”). For example the following proportions of forest treatments scheduled to be done by subagent s during subperiod t could be: 3 WUI Priority zone of CWPP area Extremely high High Medium high Medium Medium Low Proportion of total forest treatment acres .52 .26 .13 .06 .03 As you can see the proportion of forest treatments allocated to CWPP areas in each priority zone doubles as the priority zone increases. Subagents that do not have CWPP acreage in a given WUI priority zone will allocate those acres to the highest priority zone in which they have CWPP areas in their landownership. Step 2 Once the proportion of acres allocated to CWPP priority area treatments within individual county wide priority treatments is set, allocation will proceed based upon the ranking of priorities within individual fire districts. This is similar to the test case allocation. In the event that the landowner manages multiple CWPP areas: (1) In the same county wide priority area; (2) that contain multiple fire districts; and (3) where the fire district ranking of those CWPP areas is the same, those acres are split among the CWPP areas. See step 1,a,ii in Appendix B of the test case for examples of the above scenario. Prioritization for scenario 2 (WUI areas, including CWPP areas) Prioritization in forest treatment scenario #2 will used the weighted proportions outlined in option 1 of forest treatment scenario #1 to disproportionally allocate acreage across county wide priority areas. Acreage for treatments will be allocated randomly in the priority area as determined by Fire_BGCv2. Subagents that do not have acreage in a given WUI priority zone will allocate those acres in the highest priority zone in which they have acres in their landownership. Acreage allocated to CWPP areas would vary among landowners based on the results of the land and wildfire management agencies survey currently being conducted. Allocation of forest treatments among CWPP treatment areas would be the same as that outlined in Option 2 of forest treatment scenario #1. Prioritization for scenario 3 (WUI areas, CWPP areas and non-WUI areas) The proportion of available acres in forest treatments allocated to CWPP, WUI and non-WUI areas will be obtained using the land and wildfire management agencies survey currently being conducted. Individual subagents of Agenta can have different proportions of treatments allocated to CWPP, WUI and non-WUI areas. Prioritization of WUI forest treatment in scenario #2 will use the weighted proportions outlined in option 1 of forest treatment scenario #1 to disproportionally allocate acreage across county wide priority areas. Acreage of treatments will be allocated randomly in the priority area based upon the average size of forest treatments during the past 10 years. Subagents that do not have acreage in a given WUI priority zone will allocate those acres in the highest priority zone in which they have acres in their landownership. Allocation of forest treatments among CWPP treatment areas would be the same as that outlined in Option 2 of forest treatment scenario #1. 4 Forest treatments in non-WUI areas will be randomly allocated on the landscape. The size of forest treatments in non-WUI areas will be drawn randomly from the range of forest treatments (in acres) conducted during the past 10 years. Appendix C. Procedure for estimating conditional probabilities of structure losses from wildfire 1. The CAMA parcel data enumerates the following classes of exterior wall finishes and roof materials for residential structures parcels: Exterior wall finish codes (dominate type of exterior wall finish): 0 OTHER - indicates any other type of exterior finish markedly different in appearance and which cannot be equated to one of the below. 1 STUCCO - stucco over frame or Permastone or Formstone 5 2 SHINGLE - shingles or shakes 3 MASONITE - hardboard of any type 4 ASBESTOS - asbestos shingles of plank 5 ALUMINUM/VINYL/STEEL - aluminum, vinyl, or steel siding or sandwich panels 6 WOOD SIDING or SHEATHING - solid board 7 STONE - cut stone veneer, or reinforced concrete 8 BRICK - solid masonry, brick on block, or veneer 9 BLOCK - concrete block, structural clay tile, stucco over concrete block 10 LOG - large diameter timbers that serve as both the foundation and exterior of the structure Roof material codes (dominate type of roofing material): 1 WOOD SHINGLE - roof material composed of small sheets of redwood, white or red cedar machined to a uniform thickness and size 2 SLATE - natural, durable, stone used as a shingle. It is a fireproof, but brittle surface, about 3/16" thick and milled into various shapes. Colors range from grey to various hues of red, green, or purple 3 TILE - usually half-rounded product of either clay or concrete which has been kiln baked to a hardness which gives a wearing surface that needs no paint. It usually has various red shades. 4 COPPER - roof made of copper 5 METAL - roof made of sheets of corrugated galvanized metal, flat, standing seam or batten seam plates. 6 WOOD SHAKE - very similar to wood shingle, except for various thicknesses and slightly irregular shapes due to splitting rather than cutting the wood. 7 COMPOSITION ROLE - made of compressed fiber or paper material saturated with asphalt and rolled out over wood sheathing. 8 BUILT UP TAR AND GRAVEL (ROCK) - roofing that is built up by laying roofing felt with overlapping seams, then sealed by mopping with hot tar or roofing compound. The final coat is the most common roof material for flat roofs and underground homes. 9 ASBESTOS - a rigid, grey, non-flammable natural mineral fiber shingle formed into a diamond or hexagonal shape (a barrel shaped cap raises from the roof ridge). 10 ASPHALT SHINGLE - a flexible composition or fiberglass shingle applied over roofing felt (most commonly used in new construction). 11 OTHER - roof material not listed above, which includes enamel metal shingles, thin membrane terne, or builtup concrete. 2. Define low, moderate, and high flammability of exterior walls and roofs based on expert review of the following CAMA structural material codes: Exterior wall flammability classes: Low: Codes 1, 7, 8, 4, 9, Moderate: 5, 0, 10 High: Codes 2, 6, 3 Roof flammability classes: Low: Codes 2, 3, 4, 5, 9, 10 Moderate: 7, 8, 11 High: Codes 1, 6 3. Define low, high, and very high structure ignition classes for residential structures in terms of combinations of the three exterior wall flammability classes and three roof flammability classes, as follows: 6 Roof flammability Low Moderate High Exterior wall flammability Low Moderate High low (L) high (H) very high (VH) high (H) high (H) very high (VH) very high (VH) very high (VH) very high (VH) Assign conditional probabilities of residential structure losses from wildfire to the three structure ignition classes based on the results of Structure Ignition Assessment Model (SIAM) assessments of homes in western Montana (Stockmann et al 2010) as follows: Low 0.80 High 0.95 Very high 1.0 The first row shows the structure ignition classes and the second row gives the conditional probability specified for each class. Conditional probabilities reflect the common structure ignition potential of homes in each class. 4. Partition the residential structures added in each subperiod into two categories: (a) those located in subdivisions; and (b) those not located in subdivisions. The CAMA parcel data for July 2010 indicate that 55% of the residential structures in Flathead County are in subdivisions. Therefore, 55% of the residential structures added during each subperiod are randomly assigned to the subdivision category and the remainder is assigned to the non-subdivision category, subject to the constraint that residential structures located in the same parcel are in the same category.3 Residential structures in the subdivision category that were built after August 2007 are assumed to be compliant with the current WUI subdivision regulations that require the use of Firewise measures in subdivision design and Class A and Class B fire-rated roofing material (low flammability class). Siding materials of residential structures in subdivisions are also assumed to be Class A and Class B fire-rated (low flammability category). Residential structures in the subdivision category are assigned the conditional probability of the low structure ignition class (i.e., 0.80). Residential structures in the non-subdivision category are randomly assigned to structure ignition classes based on probabilities derived from the distribution of existing structures among material ignition classes in 2010 determined from the CAMA parcel data. For example, suppose the distribution is as follows: 0.6 0.4 0.2 0 L H 3 VH This procedure allows residential structures in one parcel to be allocated to the subdivision category and residential structures in an adjacent parcel to be allocated to the non-subdivision category. In some cases, this may not be realistic. RECID2 does not simulate whether or not the residential structures added in each subperiod are in a subdivision. 7 Based on this distribution, the probabilities of assigning residential structures to the three structure ignition classes are 0.25 for the low class, 0.45 for the high class, and 0.30 for the very high class. The conditional probability of residential structure losses from wildfire in the non-subdivision category equals the conditional probability corresponding to a randomly-selected material ignition class. For example, non-subdivision residential structures randomly assigned to the high material ignition class are assigned a conditional probability of 0.95 based on material flammability. (A variation of this procedure allows the probability distribution of material ignition classes to change over subperiods in response to a higher risk of wildfire losses due to climate change and other factors.) 5. Define four fuel reduction classes in the Home Ignition Zone4 for land developers/homebuilders and homeowners. The four classes are: (1) full fuel reduction; (2) heavy fuel reduction; (3) light fuel reduction or; (4) no fuel reduction. Land developers/homebuilders select a fuel reduction class for residential structures during the initial subperiod, which is the subperiod in which the structure is built. Homeowners select a fuel reduction class for structures during subsequent subperiods, which are the subperiods that follow the initial subperiod. Full fuel reduction refers to the complete conversion of wildland vegetation in zone 1 of the Home Ignition Zone (30 feet or more) to a manicured landscape and fuel reduction activities, such as pruning and thinning, in zone 2 (30-100 feet) and (if possible) zone 3 (100-200 feet). These standards follow Firewise and National Fire Protection Association Guidelines. Heavy fuel reduction refers to full Firewise thinning and brush clearing standards in zone 1 of the home ignition zone (30 feet), but without additional fuels reduction in zones 2 or 3 of the home ignition zone. Light fuel reduction refers to small-scale brush clearing, thinning and defensible space creation that does not meet Firewise standards for any zone in the Home Ignition Zone. No fuel reduction means none of the fuel reduction activities described above is used. 6. Assign the following conditional probabilities of structural losses from wildfire to fuel reduction classes: Structures in the full fuel reduction class are assigned a conditional probability of 0.36. This value is adapted from Stockmann et al. (2010) and reflects a second SIAM assessment to gauge the average reduction in structural loss probabilities resulting from varying levels of vegetation management. Structures in the heavy fuel reduction class are assigned a conditional probability of 0.55, which is the average of the SIAM probability reduction factors for full fuel reduction and light fuel reduction classes. Structures in the light fuel reduction class are assigned a conditional probability of 0.89. This value is adapted from Stockmann et al. (2010) and reflects a second SIAM assessment to gauge the average reduction in structural loss probability resulting from varying levels of vegetation management. Structures in the no fuel reduction class are assigned a conditional probability equal to the structural material ignition probability (see step 3, above). 7. Residential structures in the subdivision category that were built after August 2007 are assumed to be compliant with the current subdivision regulations for Flathead County that require the use of Firewise 4 The Home Ignition Zone is the area that principally determines the ignition potential of a structure (Cohen 2008). It includes a structure and 100 feet to 200 feet of its surroundings, depending on local factors such as fuel type(s), fuel loadings, slope, wildfire intensity and firebrand potential. Research by Cohen (2000) indicates that the characteristics of the Home Ignition Zone and their response to burning objects and firebrands are a primary determinant of structure ignition during wildfires. 8 measures. Such structures are assumed to have full or heavy fuel reduction in the initial period; allocation between the two classes is determined in consultation with the stakeholder panel. 8. Determine whether to assign each structure a conditional probability of wildfire losses equal to the probability based on structural ignition class or the probability based on fuel reduction class. During the initial subperiod, if the conditional probability based on fuel reduction is less than the conditional probability based on structure ignition class, then the former is used. Otherwise, the latter is used. The following decision tree summarizes the procedure for determining whether to assign a conditional probability based on structure ignition class or fuel reduction class: *If light fuel reduction is performed and the structure is assigned to the low structure ignition class, it is assigned a conditional probability of 0.80. 9. Determine the probabilities of fuel reduction performance and fuel reduction class during the initial and subsequent subperiods based on the following assumptions: (a) structures in subdivisions have full or heavy fuel reduction and use materials that ensure the “low” structure ignition class (see steps 4 and 7) and; (b) land developers/homebuilders select the level of fuel reduction for structures during the initial subperiod and homeowners select the level of fuel reduction for structures during subsequent subperiods. Initial subperiod. There are no current or historical data with which to evaluate land developers’/homebuilders’ decisions regarding vegetation management during the initial subperiod. For that reason, the test case assigns equal probabilities to whether or not fuel reduction has been performed and selects the fuel reduction class for residential structures NOT in subdivisions as follows: 1a. Was fuel reduction been performed in the initial subperiod? i. Yes: 0.50 ii. No: 0.50 2a. If yes, what level of fuel reduction was performed? i. Full fuel reduction: 0.33 ii. Heavy fuel reduction: 0.33 iii. Light fuel reduction: 0.33 9 2b. If no, the structure ignition class is randomly selected and the structure ignition probability associated with that class is used (see step 4). Flathead County subdivision regulations pertaining to the WUI warrant using different assumptions and probabilities for whether or not fuel reduction was performed and selection of fuel reduction class during the initial subperiod for residential structures in subdivisions, as follows: 1a. Was fuel reduction been performed in this subperiod? i. Yes: 1.0 ii. No: 0 2a. What level of fuel reduction was performed? i. Full fuel reduction: 0.25 ii. Heavy fuel reduction: 0.75 iii. Light fuel reduction: 0 Subsequent subperiods. There are no current or historical data with which to evaluate homeowners’ fuel reduction decisions in subsequent periods. Members of the stakeholder panel were asked to assess the likelihood that Flathead County homeowners perform vegetation management in the home ignition zone. Results of that assessment were used to specify the probabilities that homeowners perform different levels of fuel reduction. There is no information for evaluating the level of fuel reduction selected by homeowners during subsequent subperiods. Probabilities for structures NOT in subdivisions in subsequent subperiods are as follows5: 1a. Was fuel reduction performed during this subperiod? i. Yes: 0.65 ii. No: 0.35 2a. If yes, what level of fuel reduction was performed? i. Full fuel reduction: 0.33 ii. Heavy fuel reduction: 0.33 iii. Light Fuel reduction: 0.33 2b. If no, the structure ignition class is randomly selected and the structure ignition probability associated with that class is used (see step 4). Probabilities for structures in subdivisions are: 1a. Was fuel reduction performed during this subperiod? i. Yes: 0.80 ii. No: 0.20 2a. If yes, what level of fuel reduction was performed? i. Full: 0.20 ii. Heavy: 0.45 iii. Light: 0.35 5 Structures built before the first study subperiod or subdivision structures built before August 2007 will be assigned a conditional probability based on the probabilities outlined for subsequent subperiods. 10 2b. If no, the structure ignition class is set as the conditional probability for the low ignition class (i.e. 0.80) 10. Account for the longitudinal or carry-over effect of fuel reduction choices on the conditional probability of structural losses. Full or heavy fuel reduction is expected and light fuel reduction is not expected to reduce the conditional probability of structural losses for more than one subperiod (10 years). Additional procedures are needed to account for: (1) the variable amount of effort needed to maintain full and heavy fuel reduction across subperiods; and (2) how the benefits of full and heavy fuel reduction diminish when they are not maintained. Both procedures are described below. Less effort is required to maintain the benefits of full and heavy fuel reduction than to initiate them. Homeowners can choose to maintain the benefits of the full and heavy fuel reduction for a set number of subperiods following the subperiod in which they were initiated.6 The effort required to initiate and maintain the benefits of light fuel reduction are the same; thus the benefits of light fuel reduction do not carry over to future subperiods. Failure to perform fuel reduction in subsequent subperiods results in regrowth of vegetation and a corresponding increase in the conditional probability of structural losses from wildfire. This probability is increased in each subperiod during which fuel reduction is not performed until it equals the structural ignition probability. The regrowth rate of vegetation in parcel j and the corresponding increase in probability of structure ignition (i.e., the regrowth factor) varies depending on the presence of dry or mesic forest associations on the site. The conditional probability of parcels in dry forest types (i.e. Ponderosa Pine, mixture of Douglas Fir and Ponderosa Pine) where fuel reduction has been performed (heavy or full) will be increased by a regrowth factor of 0.20 during the first subperiod of inactivity and by a regrowth factor of 0.50 in subsequent subperiods. The conditional probability of parcels in mesic forest types (i.e. Douglas Fir, Larch, and Subalpine Fir) where fuel reduction has been performed (heavy or full) will be increased by a regrowth factor of 0.50 in all subperiods. The following equation is used to calculate the conditional probability of wildfire losses for structure i in parcel j for fuel reduction class v (v = full or heavy vegetation management) during subperiod t + 1 when fuel reduction is not performed during that subperiod: pij(t+1) = (1 + Rjft)(pijvt) 6 The level of fuel reduction required in subsequent subperiods in order to maintain the fuel reduction benefits achieved during the initial or current subperiod is determined as follows: Maintenance of full fuel reduction benefits (i.e., lower conditional probability) requires at least heavy fuel reduction during the first and second subperiods following original full fuel reduction. Maintenance is not an option in the third subperiod following full fuel reduction because heavy and light fuel reduction only addresses regrowth in zone 1 of the Home Ignition Zone, not regrowth in zones 2 and 3. Approximately 20 years of no treatment in zones 2 and 3 of the Home Ignition Zone will result in a young, thick forest that will influence the probability of wildfire impacts to the parcel and structure. For that reason, full fuel reduction is required in the third subperiod following original full fuel reduction in order to retain the benefits of fuel reduction and address regrowth in zones 2 and 3 of the home ignition zone. Maintenance can then be performed in later subperiods as explained above. Maintenance of heavy fuel reduction requires at least light fuel reduction during the first subperiod following the original fuel reduction. Maintenance is not an option in the second subperiod following heavy fuel reduction because light fuel reduction is not adequate enough to restrict regrowth of young trees and shrubs in or adjacent to zone 1 of the home ignition zone or manage the canopy of trees in that zone. Heavy fuel reduction is required in the second subperiod following the original heavy fuel reduction in order to retain the benefits (i.e., lower conditional probability) of heavy fuel reduction. Maintenance can then be performed in later subperiods as explained above. 11 where: pijvt = conditional probability of wildfire losses for structure i in parcel j for fuel reduction class v in subperiod t; and Rjvt = regrowth factor for parcel j for forest type f in subperiod t. Failure to perform fuel reduction will not increase the conditional probability of structure losses from wildfire beyond the probability based on structural ignition class. Therefore: pijt ≤ pijct Likewise, failure to maintain the previous fuel management class will not result in a conditional probability that exceeds the conditional probability based on fuel reduction class performed during the current subperiod. Therefore: pijt ≤ pijv Example: a. A new residential structure is built with wood siding and composition roll roofing. It is not located in a subdivision and is located in a dry forest type. The structure is assigned to the “high” structure ignition class, which has conditional probability of burning of 0.95. b. The developer/homebuilder performs full fuel reduction in the home ignition zone during the initial subperiod. The initial conditional probability for full fuel reduction is 0.36. Therefore: pijt = 0.36, where t refers to the initial subperiod. c. In the next subperiod (t + 1), the homeowner does not perform any fuel reduction. That homeowner’s conditional probability for subperiod t + 1 is as follows: pij(t+1) = (1 + 0.20)(0.36) = 0.43 The final conditional probability of structure loss from wildfire during subperiod t + 1 is 0.43. d. The homeowner does not perform fuel reduction in the subperiod t + 2. That homeowner's conditional probability for the subperiod t + 2 is as follows: pij(t+2) = (1 + 0.50) (0.43) = 0.65 The final conditional probability of structure loss from wildfire during subperiod t + 2 is 0.65. 11. Select the conditional probability of structure losses from wildfire in later subperiods given the various decisions and procedures described in steps 1-10. The following decision tree describes this process: 12 *If light fuel reduction is performed and the structure is assigned to the low structure ignition class, it is assigned a conditional probability of 0.80. 12. Determine probabilities of fuel reduction and fuel reduction classes during subsequent subperiods (after the second subperiod). Probabilities may differ slightly from those assigned in the initial subperiod because responsibility for fuel reduction decisions shifts from land developers/homebuilders to homeowners in subsequent subperiods. Structure ignition class is determined using existing data on structures in the county (see step 4, above). Two separate sets of probabilities are described below because structures in subdivisions are assumed to have full or heavy fuel reduction and use materials that ensure the “low” structure ignition class (see steps 4 and 7). Probabilities for whether or not homeowners conduct fuel reduction during subsequent subperiods are the same as those outlined step 9. Probabilities that are contingent on previous actions are assigned based on the probabilities reported in step 9. Equal probabilities are assigned to decision outcomes when there is no prior data on which to base these probabilities. The conditional probabilities for structures NOT in subdivisions are as follows: 1a. Has fuel reduction been performed during this subperiod? i. Yes: 0.65 ii. No: 0.35 2a. If yes, was full or heavy fuel reduction performed in the previous subperiod? i. Yes: 0.66 13 ii. No: 0.34 3a. If yes, has fuel reduction been maintained? i. Yes: 0.50 ii. No: 0.50 4a. If no, calculate conditional probability based on time since fuel reduction (step 9). 2b. If no, what level of fuel reduction has been performed? i. Full fuel reduction: 0.33 ii. Heavy fuel reduction: 0.33 iii. Light fuel reduction: 0.33 1b. If no was full or heavy fuel reduction performed in the previous subperiod? i. Yes: 0.66 ii. No: 0.34 5a. If no, the structure ignition class is randomly selected and the structure ignition probability associated with that class is used (see step 4). 5b. If yes, calculate conditional probability based on time since fuel reduction (step 9) The conditional probabilities for structures in subdivisions are as follows: 1a. Has fuel reduction been performed during this subperiod? i. Yes: 0.80 ii. No: 0.20 2a. If yes, was full or heavy fuel reduction performed in the previous subperiod? i. Yes: 0.65 ii. No: 0.35 3a. If yes, has fuel reduction been maintained? i. Yes: 0.60 ii. No: 0.40 4a. If no, calculate conditional probability based on time since fuel reduction ( (step 9). 2b. If no, what level of fuel reduction has been performed? i. Full fuel reduction: 0.20 ii. Heavy fuel reduction: 0.45 iii. Light fuel reduction: 0.35 1b. If no, was full or heavy fuel reduction performed in the previous subperiod? i. Yes: 0.65 ii. No: 0.35 5a. If no, the structure ignition class is randomly selected and the structure ignition probability associated with that class is used (see step 4). 5b. If yes, calculate conditional probability based on time since fuel reduction (step 9) 13. Examples of conditional probability of structure losses. a. A new residential structure is built with wood siding and composition roll roofing. It is not located in a subdivision. The structure is assigned to the “high” structure ignition class, which has conditional probability of burning of 0.95. 14 b. The developer/homebuilder performs full fuel reduction in the home ignition zone during the initial subperiod. The final conditional probability for full fuel reduction is 0.36. c. In the following subperiod, the homeowner performs heavy fuel reduction. This level of vegetation management is sufficient to maintain the full fuel reduction performed in the earlier subperiod. As such, the structure retains the 0.36 conditional probability of structure ignition in the second subperiod. d. Another homeowner whose house was built using the same materials as the first and whose developer/homebuilder performed full fuel reduction in the initial subperiod did not perform any fuel reduction in the second or third subperiods. That homeowner’s conditional probability for the third subperiod is as follows: pijt = 0.36 + [(0.20)(2)] = 0.76 The final conditional probability of structure loss from wildfire during the third subperiod is 0.76. 14. Examples of unconditional probability of structure losses. a. If the probability that parcel j burns in the initial subperiod is 0.10 and the conditional probability of wildfire losses for residential structure i in parcel j, structure ignition class c, vegetation management class v, and subperiod t is 0.36, then the unconditional probability of wildfire losses is puijt = pbjt * pijt = (0.10)(0.36) = 0.036. b. If the probability that parcel j burns in the second subperiod remains constant at 0.10 and the homeowner does not perform fuel reduction, then the conditional probability of wildfire losses to residential structure i in parcel j, structure ignition class c, vegetation management class v, and subperiod t is 0.56. The unconditional probability of wildfire losses is puijt = pbjt * pijt = (0.10)(0.56) = 0.056. c. If the probability that parcel j burns in the third subperiod increases to 0.23 and the homeowner does not perform fuel reduction, then the conditional probability of wildfire losses for residential structure i in parcel j, structure ignition class c, vegetation management class v, and subperiod t is 0.76. The unconditional probability of wildfire losses is puijt = pbjt * pijt = (0.23)(0.76) = 0.17. d. If the probability that parcel j burns in the fourth subperiod remains constant at 0.23 and the homeowner performs heavy fuel reduction, then the conditional probability of wildfire losses for residential structure i in parcel j, structure ignition class c, vegetation management class v, and subperiod t is 0.55. The unconditional probability of wildfire losses is puijt = pbjt * pijt = (0.23)(0.55) 15. Determine whether to assign each structure built prior to the initial study subperiod a conditional probability of wildfire losses equal to the probability based on material structural ignition class or the probability based on fuel reduction classes and/or regrowth. This process expands the probabilities assigned in step 12 and applies them the decision tree in step 11 to determine the carryover effect of any fuel reduction or maintenance performed by land/developers or homeowners of existing structures. More specifically, the procedure randomly assigns the following characteristics: (1) Whether fuel reduction has occurred in the previous three subperiods; (2) The number of subperiods since the most recent incidence of fuel reduction and; (3) Whether the previous fuel reduction treatment was maintenance. The assignment of these attributes can be paired with the processes and limitations outlined in step 10 to determine what the existing probability of a structure might be during the initial subperiod. Assignment of fuel reduction probabilities includes the up to five subperiods to account for the maximum time that a combination of fuel reduction treatments could affect the conditional probability of a structure during the initial study subperiod. 15 There are no current or historical data with which to evaluate land developers’/homebuilders’ decisions regarding vegetation management before the initial subperiod. Subdivisions regulations outlined in step 4 were implemented after August 2007 and do not apply to structures built before the initial subperiod. For that reason, the test case assigns the following probabilities to structures built before the initial subperiod. Probabilities are accompanied by an example to demonstrate the progression of the assignment process: 1a. Has fuel reduction ever occurred? i. Yes: 0.50 ii. No: 0.50 2a. If yes, how many subperiods since last treatment? Structures built in 2000-2009 i. Previous subperiod: 1.0 Structures built in 1990-1999 i. Previous subperiod: 0.5 ii. 1 subperiod: 0.5 Structures built before 1990 i. Previous subperiod: 0.33 ii. 1 subperiod: 0.33 iii. 2 subperiods: 0.33 3a. What was the last level of fuel reduction performed during the subperiod defined in 2a)? i. Full: 0.33 ii. Heavy 0.33 iii. Light: 0.33 4a. What is the likelihood that last fuel reduction was maintenance (only applies to heavy or light fuel reduction)? Heavy fuel reduction i. Maintenance: 0.50 ii. Not maintenance: 0.50 Light fuel reduction i. Maintenance: 0.50 ii. Not maintenance: 0.50 5a. If maintenance, What is the likelihood it is the first or second occurrence (applies only to heavy fuel reduction in subperiod immediately preceding initial study period)? i. First maintenance: 0.50 ii. Second maintenance: 0.50 1b. If no, assign material structure ignition class probability based on CAMA data. Example: a. An existing structure was built in 1970 with cedar shingles and wood panel siding. As such, it is placed in the "very high" structure ignition category (ignition probability of 1.0). The structure is located in a dry forest type. b. Fuel reduction has occurred on this parcel since the structure was built (step 1a). c. It has been 2 subperiods since fuel reduction has been performed (step 2a). d. Heavy fuel reduction was last performed (step 3a). 16 e. Fuel reduction outlined in step (d) was maintenance, meaning a conditional probability of 0.36 (not the 0.55 normally associated with heavy fuel reduction) during that subperiod. f. Use the regrowth equation in step 10 of Appendix G to calculate increase in conditional probability during previous 2 subperiods: pij(t+1) = (1 + 0.20)(0.36) = 0.43 pij(t+2) = (1 + 0.50) (0.43) = 0.65 Thus, the conditional probability of the existing structure going into the first study subperiod is 0.65. Example with maintenance: a. An existing structure was built in 1990 with asphalt shingle roofing and brick exterior walls. As such it is placed in the "low" structure ignition category. The structure is located in a mesic forest type. b. Fuel reduction has occurred on this parcel since the structure was built (step 1a). c. Fuel reduction was performed during the previous subperiod (step 2a). d. Heavy fuel reduction was last performed (step 3a). e. Fuel reduction outlined in step d was maintenance (step 4a), meaning a conditional probability of 0.36 during that (previous) subperiod. This is lower than usual for the heavy fuel classification (0.55) because it was used to maintain full fuel reduction in a prior subperiod. f. Maintenance performed in the previous subperiod was the second instance of maintenance (step 5a), meaning no additional maintenance can occur. Thus, the conditional probability of the existing structure going into the first study period is 0.36 g. During the first study subperiod, the homeowner again performs heavy fuel reduction. The conditional probability for this period is the lowest value between the current fuel reduction treatment (0.55) and the regrowth factor calculation: pij(t+2) = (1 + 0.50) (0.36) = 0.54 Thus, the conditional probability of the structure for the first subperiod is 0.54. It IS NOT 0.36 because maintenance of this stand can no longer occur. References Cohen, J. 2008. The Wildland-Urban Interface fire problem: A consequence of the fire exclusion paradigm. Fire History Today Fall: 20-26. Cohen, J.D. 2000. Preventing disaster: Home ignitability in the wildland-urban interface. Journal of Forestry 98(3): 15-21. Stockmann, K., J. Burchfield, D. Calkin, and T. Venn. 2010. Guiding preventative wildland fire mitigation policy and decisions with an economic modeling system. Forest Policy and Economics 12(2): 147-154. 17 Appendix D. Land ownership in study area The study area is the yellow-shaded area shown below. Southwestern portion of county Southwestern spur of county Acres Percent Montana DNRC (state trust lands) 110,060.82 7.36 National Park Service 275,432.89 18.43 Plum Creek Timber Company 185,689.72 12.42 Other private land 196,848.20 13.17 716,938.22 47.96 18 US Forest Service Appendix E. Determining maximum acres in forest treatments One of the data inputs required by Fire-BGCv2 is maximum acres in forest treatments (i.e., commercial harvesting (ch), mechanical fuel reduction (mf), and prescribed burning (pb)) for each of the six landownerships. Maximum acres were specified based on: (1) a triangular probability distribution of acres treated; (2) average acres per treatment during the 10-year period preceding 2010; (3) the percent by which maximum acres treated in future subperiods exceeds average acres treated during the 10-year period preceding to 2010; (4) annual growth in sales for the wood products manufacturing and residential construction industries during the evaluation period given by the RECID2 model; and (5) annual growth in the size of the WUI. The following specifications and assumptions were used in determining maximum acres: 1. Triangular distribution of acres treated. Acres in forest treatments are assumed to have triangular probability distributions. The triangular probability distribution for a random variable x is T(a, b, c) = [2(x - a)/(c - a)(b - a)] for a ≤ x ≤ b and T(a, b, c) = [2(c - x)/(c - a)(c - b)] for b ≤ x ≤ c, where a in the minimum value, c is the mode, and b is the maximum value of x. The mean of the triangular distribution is (a + b + c)/3 and the variance is (a2 + b2 + c2- ab - ac - bc)/18. Graphically, T(a, b, c) is: Probability Acres in forest treatment k a c b The procedure does not require specifying the parameters of the triangular distribution; only that the distribution is triangular. 2. Ratio of maximum acres to average acres. The ratio of maximum acres to average acres for treatment k (k = ch, mf, pb) when acres treated are triangularly distributed is: rk = bk/[(ak + bk + ck)/3]. (1) For example, rk = 1.63 for ak = 10, bk = 36, and ck = 20, which implies maximum acres are 63% greater than average acres for treatment k. Equation (1) can be modified to allow rk to vary over subperiods (i.e., rkt = bkt/[(akt + bkt + ckt)/3]). For simplicity, the test case assumes rk is constant over subperiods. The value of rk for the three treatments is specified in collaboration with the land and wildland fire management agencies’ panel. 3. Maximum acres per treatment. Maximum acres in treatment k for economic growth scenario j through the end of year t is: bjkt = (ãk)(rk)(sjkt) (2) where ãk is average acres in treatment k in the study area during the 10-year period preceding 2010, rk is defined in step 2, and sjkt is a scaling factor for the growth in maximum acres in treatment k through the end of year t (t = 1, …, 50) for economic growth scenario j (j = L, M, H). Multiplying ãk by rk converts 19 average acres to maximum acres. Multiplying (ãk)(rk) by sjkt adjusts the maximum acres over subperiods as described in step 4. 4. Adjusting maximum acres. The scaling factors for commercial harvesting (ch) and growth scenario j through the end of year t are: so(chi)t = 1 when maximum acres in commercial harvesting do not increase over time (case 1) (3) (1 + gj(wpr))t when annual maximum acres increase at the rate gj(wpr) (case 2) gj(wpr) is the average annual growth rate for the wood products manufacturing and residential construction industries (designated wpr) for growth scenario j. Annual growth rates for these industries come from the IMPLAN model, which is part of the RECID2 model. The scaling factors for fuel reduction treatments mf and pb for growth scenario j through the end of year t are: sjt = 1 when maximum acres do not increase over time (case 1) (4) (1 + gjwt)t when annual maximum acres treated increase at the rate gjwt (case 2) gjwt is the average annual growth rate in the aerial extent of the WUI (W) through the end of year t for growth scenario j (in decimal equivalent). The specification of sjt implies the scaling factors for mf and pb are the same for a given year and growth scenario. Case 1 represents a situation in which the maximum acres in mf and pb do not increase over years, and hence subperiods, due to budget constraints. Case 2 represents a situation in which the maximum acres in mf and pb increase over time, and hence subperiods, at a rate equal to the annual rate of growth in the aerial extent of the WUI. Case 2 corresponds to the condition in which fuel reduction treatment budgets are sufficient to support the maximum acres treated. 5. Example of total and subperiod maximum acres for commercial harvesting. Consider how to calculate the maximum acres in commercial harvesting for the five subperiods and moderate economic growth scenario (M) for case 2 when: (1) 10,000 acres were commercially harvested in the 10-year period preceding 2010 (ãch = 10,000); (2) maximum acres are 63% higher than average acres (rch = 1.63); (3) rch does not vary over subperiods; and (4) the average annual growth rate in the wood products manufacturing and residential construction industries for the moderate economic growth scenario is 2.03% (i.e., gM(ch) = 0.0203). Maximum acres commercially harvested during the 10-year period preceding 2010 (designated by a zero subscript) and through the end of each 10-year subperiod from 2010 to 2059 based on the above values are: bM(ch)0 = (10,000)(1.63) = 16,300 acres; bM(ch)10 = (10,000)(1.63)(1 + 0.0203)10 = 19,928 acres; bM(ch)20= (10,000)(1.63)(1 + 0.0203)20 = 24,364 acres; bM(ch)30 = (10,000)(1.63)(1 + 0.0203)30 = 29,787 acres; bM(ch)40= (10,000)(1.63)(1 + 0.0203)40 = 36,417 acres; and bM(ch)50 = (10,000)(1.63)(1 + 0.0203)50 = 44,523 acres. 20 (5) These maximum commercially harvested acres are entered into the Fire-BGC model. A similar procedure is used to determine the maximum acres treated with mechanical fuel reduction and prescribed burning for each subperiod. The above procedure is used to derive the maximum acres in forest treatments for each subperiod and land ownership. Steps 6 through 9 describe the procedure was used to estimate the maximum acres in forest management treatments for the six landownerships by subperiod. The information obtained and calculations used to determine maximum acres are based on the procedures outlined in steps 1-5 above. 6. Obtain test case values of rk and ãk from landowners. Multiple versions of a “maximum acres questionnaire” were created and sent to representatives from each of the six landowners in the test case (Private, US Forest Service, Glacier National Park, Montana DNRC, Plum Creek Timber Company, and Other). The “Other” category is represented by The Confederated Salish and Kootenai Tribes because they are the largest landowner in the category and other landowners are unlikely to conduct large-scale forest treatments. Subcategories were created within the “private” land designation to reflect the variety of agencies and organizations conducting or facilitating (i.e., state and federal assistance programs) fuel reduction on this landownership. These included: (1) Resource Conservation and Development (RC & D) and Montana DNRC; (2) fuel agreements filed by private landowners; (3) USDA National Resources Conservation Service (NRCS); and (4) private contractors. Modified “maximum acres questionnaires” were sent to representatives from (1)-(3). These questionnaires prompted agencies/organizations to report forest treatments on their lands (where appropriate) and those the agency/organization helped facilitate on private lands. The procedure used to calculate maximum acres in the “private contractors” subcategory (4) is outlined in step 7. 7. Obtain information on fuel reduction performed by private contractors through a telephone survey. The sample frame for the telephone survey was a list of registered forest contractors in a 3-county area (Lake, Lincoln, and Flathead) compiled by RC & D. Contractors with non-working phone numbers or who did work in Flathead County were eliminated from the sample. A simple random sample of remaining contractors was contacted via telephone to help approximate maximum acres treated on private lands (n=16, approximately 50% of sample frame). Contractors were asked how many acres by treatment (ch, mt, pb) they had conducted during the previous year, the average maximum acres (by treatment) conducted in one year during the previous 10-year period, the number of years they had been working in the county and whether funding for the thinning was conducted using state or federal funds. Answers were reported on an excel spreadsheet. Acreage reported as using state or federal funds was not included in the final aggregation of max acres treated by private contractors because it was already reported by other agencies (see step 6). A probability distribution of the survey results was used to approximate the acreage treated by private contractors not included in the survey. Final results of this approximation were then used to calculate an average rk and ãk for private contractors. 8. Calculating max acres for private land. Results of the max acres questionnaires from “private” subcategories and the private contractor survey were entered into another excel spreadsheet. Maximum acres treated per organization/group in each of the three forest treatments and during each 10-year subperiod of the project were calculated using the equations for bjkt outlined in step 3 and the growth rate (gM(ch) = 0.0203) specified in step 5. The sum of maximum acres conducted by these various organizations/groups represents the total max acres in the “private” landownership and was applied to the excel spreadsheet described in step 9. 21 9. Calculating max acres for other landownerships. Final inputs from the maximum acres questionnaires and the results of step 8 (above) were entered into a final excel spreadsheet. Two cases of maximum acres by landownership, forest treatment and subperiod are presented as outlined in step 4: (1) when maximum acres do not increase over time; and (2) when annual maximum acres treated increase at the rate gjwt. Both cases were calculated using the equation for (bjkt) given in step 3. Case 2 uses the growth rate (gM(ch) = 0.0203) specified in step 5. The growth rate gM(ch) (0.0203) is only applied to landowners whose practices are linked to market values (US Forest Service, Plum Creek, and Private Lands). The growth rate does not apply to other landowners due to agency/organization policy or availability of harvest (Other, DNRC and Glacier National Park). 22 Appendix H: Study area neighborhoods 23 Appendix E. Calculating proportion of total properties added during each subperiod The ABM decision rules for Agentp require calculation of the proportion of total properties simulated during each subperiod. This requirement makes it possible to calculate the total costs of subdivision regulations. Data to complete calculations for subdivision proportions were obtained from The Flathead County GIS department. These data included files outlining: 1) the number and location of subdivisions for all properties and; 2) a master dataset of properties that existed at the end of each year (separate files) from 2005-2011. Personnel at the Flathead County GIS Department This turned out not to be the case as the field that they suggested for linking the data (Assessor ID) did no suggested matching up data from these two sets to determine counts of subdivisions relative to total developed properties. This proved to be impossible as the field they suggested for linking the data (Assessor ID) did not occur in both datasets and was not unique to each property. Members of the FIRECLIM project team ended up approaching the counts in two ways, which are outlined below. Results from the two methods were then averaged to determine the final proportion of total properties added during each subperiod. Method 1 1. Determine the amount of overall properties subdivided or added between two years of data (e.g. from 2005-2006). 2. Query out the number of subdivided properties during the year of interest from the subdivision layer. 3. Create a new layer from the data of #2. 4. Properties in the subdivision layer are often repeated, as evidenced by the replication of their unique Tract ID or unique polygon layer. These are not the same as multiple subdivided parcels. For that reason I used the dissolve tool to eliminate repetition in the results from step 3. The result is the number of properties subdivided during that year. 5. The difference between step 1 and step 4 is the number of properties that did not go through the subdivision process for that year. 6. Determine the yearly proportion of properties that did or did not go through the subdivision process. 7. Average proportions across the years available (i.e. 2005-2011). Results: 49.6% of new parcels during the 2005-2011 time period went through the subdivision process; 50.4% did not go through the subdivision process. Method 2 1. Open the 2011 master dataset for existing properties (e.g. those that went through or did not go through the subdivision process). This includes all properties that existed at the end of 2011. 24 2. Query out any properties that are not privately owned (i.e. residential/commercial). I can do this by using a query string relating to a specific data column. 3. Query out any properties that were subdivided (or not subdivided) using a unique identifier in the Tract ID (i.e. a set of 'XXX' letters embedded in a larger number/letter string denotes that the property was not the result of a subdivision. Results: 53% of properties in Flathead County have gone through the subdivision process; 47% of properties did not go through the subdivision process. The average of results from methods 1 and 2 are: 51.3% of properties went through the subdivisions process; 48.7% percent of properties did not go through the subdivision process. These are the values that will be used in the FIRECLIM ABM. Appendix F. Determining Agenth (Homeowners) Costs for Fuel Reduction Levels 1. Cost guide for vegetation removal (costs per acre) Table 1. RC&D cost guidelines for vegetation removal Low Medium (<600 tpa) (600-1100 tpa) Hand thin $240 $360 Machine thin<45% $170 $300 Handpile $360 $600 Machine Pile $120 $240 Chipping $600 $780 Mulching $480 $600 Burning Piles $120 $180 Commercial product $300 $300 from Biomass Pruning $100 $150 High (1100 tpa) $480 $360 $840 $360 $1,100 $840 $240 $300 Choice $200 4 1 1 2 2 3 3 3 3 2. Determine which actions are required for different levels of fuel treatment a. The FIRECLIM test case description provides guidelines for fuel reduction treatment requirements in the Home Ignition Zone (HIZ). See Appendix A at the end of this document. Choices referred to in the following description are outlined in the far right column of Table 1 above. Choice options were determined by FIRECLIM project members by aggregating actions that serve similar purposes (e.g. aggregating byproducts of vegetation removal, disposing of materials from vegetation removal, etc.) Full fuel reduction requires: One action from choice 1 for zones 1 and 2 of the HIZ; One action from choice 2 for zones 1 and 2 of the HIZ; one action from choice 3 for zone 1 of the HIZ; choice 4 for zones 1 and 2 of the HIZ. Heavy fuel reduction requires: One action from choice 1 for zone 1 of the HIZ; One action from choice 2 for zone 1 of the HIZ; one action from choice 3 for zone 1 of the HIZ; choice 4 for zone 1 of the HIZ. Light fuel reduction can be: One action from choice 1 OR choice 4 for zone 1 of the HIZ; One action from choice 2 for zone 1 of the HIZ: One action from choice 3 for zone 1 of the HIZ. 25 3. Converting costs for use in the Home Ignition Zone Zone 3 HIZ Zone 2 HIZ Zone 1 HIZ 3a. Conversion of HIZ to acreage 100 feet 70 feet Residence 30 feet 3b. Area of different segments of the Home Ignition Zone 1. Home Ignition Zone 1 = 2,826 square feet = 0.065 acres 2. Home Ignition Zone 2 = 15,386 square feet = 0.35 acres 3. Home Ignition Zone 3= 31,400 square feet= 0.72 acres 4. Total area of Home Ignition Zones 1 AND 2=41,400 square feet= 0.72 acres 5. Total area of Home Ignition Zones 1, 2 AND 3= 125,600 square feet= 2.88 acres 4. Selecting treatments conducted by members of Agenth during FIRECLIM model runs The parameters outlined in section 2 provide initial guidance as to the actions performed by members of Agenth when considering the potential costs of different fuel reduction levels. However, members of agenth can choose between multiple options regarding the actions and costs that comprise the parameters of section 2. For instance, homeowners can choose between chipping, mulching, burning piles or obtaining commercial product from biomass as a means to remove the byproducts of fuel reduction. Each action comes with different costs. For this reason additional procedures are required to select between options and costs comprising different fuel reduction levels as outlined in section 2. Table 1 and section 2 of this document outline the actions homeowners are required to perform under different fuel reduction levels. No data exists to approximate the proportions of actions that existing members of Agenth currently employ regarding options outlined in Table 1. For that reason ‘within choice’ options relevant to fuel actions required of the homeowner (e.g. choice 2 is either handpile or machine pile) will be randomly selected 26 for the homeowner. The associated cost of that choice option will be applied to their overall costs for treatments, as outlined in section 4. Insufficient data also exists for approximating the density of vegetation around each individual WUI structure. For this reason the density of vegetation necessary to clear is randomly selected from the options in Table 1. 4. Calculating updated costs for fuel reduction The total cost for any given fuel reduction level can be determined using one of the following equations: CRlt= (CO1d) (1 + λ)r(AZ1l)+(CO2d ) (1 + λ)r (AZ2l)+(CO3d) (1 + λ)r (AZ3l)+(CO4d) (1 + λ)r (AZ4l) (t = 1,…,5) (1) Where: CRl= Cost of fuel reduction level l during subperiod t CO1=Randomly selected cost of fuel reduction option choice 1 with randomly selected vegetation density d λ=Is an inflation rate for costs associated with fuel reduction options AZ1l= Area of the HIZ required to be treated with option choice 1 under fuel reduction level l CO2=Randomly selected cost of fuel reduction option choice 2 with randomly selected vegetation density d AZ2l=Area of the HIZ required to be treated with option choice 2 under fuel reduction level l CO3=Randomly selected cost of fuel reduction option choice 3 with randomly selected vegetation density d AZ3l=Area of the HIZ required to be treated with option choice 3 under fuel reduction level l CO4=Cost of fuel reduction option choice 4 with randomly selected vegetation density d AZ4l=Area of the HIZ required to be treated with option choice 4 under fuel reduction level l Only one value of vegetation density d is selected per property. r = 10 for t = 1, r = 20 for t = 2, r = 30 for t = 3, r = 40 for t = 4, and r = 50 for t = 5 Equation 1 can be used to calculate costs for the full, heavy or light fuel reduction categories. Light fuel reduction requires at least one action from choice 1 or choice 4, meaning that costs associated with one of these options will always be zero in equation 1. 5. Example A homeowner will perform full fuel reduction during subperiod 2020-2029 of the FIRECLIM simulation. Random selection of choice options and vegetation density for that homeowner is as follows: Choice 1: Machine thinning Choice 2: Hand pile Choice 3: Chipping Choice 4: Pruning 27 Vegetation density: medium 600-1100 tpa The cost to the homeowner in performing full fuel reduction for the 2020-2029 subperiod would be calculated as follows: CRl= ($300)(1+.035)10(0.72)+($600 ) (1+.035)10 (0.72)+($780)(1+.035)10 (0.065)+($150) (1+.035)10 (0.72) CRl = ($304.78) +($609.55)+($71.54)+($152.39) CRl=$1,138.26 Appendix G. Relationships between models and processes used to simulate E(RLW) FireBGCv2 RECID2 •Simulates vegetation growth and succession •Simulates forest treatments from landowners •Incorporates climate change impact on weather •Simulates residential development • Determines the number, size and location of new properties added each subperiod FSIM Home and property values •Calculates burn probabilities given vegetation and weather inputs from FireBGCv2 •Calculated by drawing from values in each of the existing 22 study area neighborhoods •Appreciated over time Parcel burn probabilities WUI Designation •Procedure for calculating weighted burn probabilities for each parcel •Determines properties to be included in E(RLW) calculation •Broad designation of exposure to risk Percentage of aesthetic value lost Structure burn probabilities •Decision tree process assigns the probability that structures burn given choices about vegetation management and building materials. •Procedure for calculating weighted burn probabilities for parcel E(RLW) calculations (for subperiod) Procedure or process Model 28 29