Heather K. Rogers

A thesis submitted in partial fulfillment of the requirements for the degree of
Master of Science
University of Washington

2003

Program Authorized to Offer Degree: College of Forest Resources

 *RTI has made Heather Roger's Thesis Defense Presentation available on-line. The defense video is synchronized with the corresponding presentation at two different magnitudes. Details are below.

*** This video presentation only works on Internet Explorer with Windows Media Player *** (Will NOT work on a Macintosh or with Netscape)
	Recommended for dial-up modems (good for any modem greater than 33 kbps). Click on the small video icon to view the presentation with the smaller video file. The presentation is cued up with the speaker's commentary, and will change slides automatically.
	Recommended for high-speed connections. Click on the large video icon to view the presentation with the larger video file. The presentation is cued up with the speaker and will change slides automatically.

Here's a link to the PDF version of this Thesis

Acknoledgements

I really enjoyed working with Kevin Ceder, Larry Mason, Dr. James McCarter, Bruce Lippke, Jeff Comnick, Luke Rogers and Matt McLaughlin on this research. I learned a great deal through their patience and support

Table of Contents

List of Figure

List of Tables

1. Introduction
      1.1 Background
      1.2 Research Objective

2. Methods
      2.1 Literature and Government Report Review
      2.2 Local Interviews
      2.3 Technical Tools
      2.4 Data and Information
      2.5 Growth, Treatment, and Wildfire Simulation
      2.6 Analysis of Fire Risk
      2.7 Analysis of Economics

3. Case Study Site Descriptions
      3.1 Fremont National Forest
      3.2 Okanogan National Forest

4. Results
      4.1 Fire Risk Results
      4.2 Economic results
      4.3 Cost to fight fire on the Fremont and Okanogan National Forests

5. Discussion
      5.1 Fire Risk
      5.2 Economics

6. Conclusion

7. Future work

Bibliography

Appendix A Fremont National Forest Map and Results
Appendix B Okanogan National Forest Map and Results

List of Figures

List of Tables

1. Introduction

1.1 Background

Dry, mid to high elevation federal forests in the inland West are overstocked with unhealthy, suppressed, small trees (Cooper 1960, Pyne 1997, Arno 2000). This is a result of decades of grazing, fire suppression, and logging (Pyne 1982, Arno 1996). Historically 10-25 year fire return intervals maintained a healthy ecosystem in these forests (Agee 1993). Many of these forests have grown under suppressed conditions which leave them more susceptible to insect and disease (Stewart 1988, Cochran and Barrett 1998, Pfilf et al. 2002). Due to lengthened fire intervals the dominant species in these forests have transitioned from fire tolerant to fire intolerant species (Pfilf et al. 2002). In forests now dominated by fire intolerant species the fuel load available to burn is more vulnerable to more intense fires (Huggard and Gomez 2001). Due to the high fuel load and altered species composition from tolerant to intolerant tree species, the fires may burn out of control as intense wildfires (USDA Forest Service 2002a). Fire suppression has caused these naturally or mechanically reseeded stands to grow into densely stocked forests (Pyne 1997, Everett et al. 2000, Strohmaier 2001). Removal of surplus fuel is needed to restore frequent low intensity fire regimes (Omi and Martinson 2002).

High fire risk forests in the United States (US) are concentrated on public lands including the national forests (USDA Forest Service U.S. Dept. of Interior and Western Governors' Association 2001). The majority of United States Forest Service (USFS) land is concentrated in the Western half of the US. Figure 1 shows the distribution of large fires for 2002 alone. There are many overlaps between the USFS land and the 2002 fires as shown by comparison of Figure 1 A and B. Concentrations of high fire risk forests appear in very similar locational distributions as do the large forest fires in 2002 in the Western United States (Powell et al. 2001).

Figure 1 A and B 2002 Large Fire Locations and USFS ownership

The extent of the problem nationally is evidenced by the trend of acres burned over the last 10 years and the number of acres in high fire risk (United States National Interagency Fire Center 2002). Figure 2 shows acres burned in wildland fires 1993-2002. Note particularly the large number of acres burned in 2000 and 2002. The USFS manages 192 million acres of which 56 million acres have been determined to be in high fire risk conditions (Powell et al. 2001). Roughly half of the high fire risk acres on National Forest lands are located in the Pacific Northwest (Western Governors Association 2001, 2002).

Figure 2 Acres Burned in Wildland Fires Reported 1993-2002 (NICC 2002)

There are many potential impacts from forest fires to rural communities that are located near high fire risk forests. Impacts of wildfire include decreased forest health, air pollution, loss of wildlife habitat, destruction of private property, and decline in the quality of life for people who live near this forestland. The federal government is developing strategies for fuel reduction treatment in at-risk forests through the National Fire Plan (Western Governors Association 2001, 2002, The Office of the President 2002). Fuel reduction treatments are being tested through innovative contracting mechanisms and research. Fuel removal thinning costs may be considered as an investment in reducing the expense to the government of suppressing fire. Currently, the federal budget spent on fire suppression well exceeds the allocation for fuel reduction activities. In recent years record costs have been incurred by federal agencies to fight forest fires, $1.4 billion in 2000 and $2.2 billion in 2002 (The Office of the President 2002). There are also costs to local areas business, and states, counties, and cites when forest fires occur (Babbitt and Glickman 2000).

1.2 Research Objective

The objective of this study is to evaluate the cost and/or revenue associated with fire risk reduction effectiveness for different management alternatives on two National Forests (NF) in order to support more effective hazardous fuels reduction strategies. This assessment analyzes fire risk and net revenue associated with hazardous fuels reduction treatment simulations. The two forests that were chosen both have high fire risk overstocked forest areas dominated by small diameter suppressed trees and, as a result, both forests recognized to be are in need of fuel removal thinning treatments to reduce fire risk (Fremont National Forest 2003, Okanogan National Forest 2003). On the other hand while these forest share a similar risk reduction challenge there are many aspects of these forests that are quite different such as the harvesting and manufacturing infrastructures, the forest types, moisture and weather patterns, etc.
Hypothesis: Positive net revenues from logs sales will result from harvest activities designed to reduce fire risk in most of the moderate to high risk inland west forests.

Null Hypothesis: There is little potential for positive net revenue from logs sales that will result from harvest activities designed to reduce fire risk in moderate to high risk inland west forests.

Also worthy of consideration in the context of fuel removal treatment evaluations is a broader set of costs and benefits expanded to include an assortment of market and non-market values that result from risk reduction investments. Simply put, it may be that, even if fuel removal harvests don't generate positive revenues from log sales, the many other benefits resulting from resource protection and utilization will outweigh any government costs for service contracts to remove surplus fuel loads.

An additional goal of this investigation is present a technical methodology to assist forest managers in quick evaluation of forest fire risk distributions and potential treatment alternatives across broad landscapes. Without technology to assist such analysis, the complexity of forest planning for hundreds of thousands of acres is daunting. It is the intention of this investigation to demonstrate technical modeling capabilities beneficial to foresters that must communicate management choices to involved lay publics so that they might better understand the consequences of treatment alternatives or no action.

2. Methods

This project has been designed as a parametric sensitivity analysis to be used in tandem with existing forestry modeling capabilities to assess the effectiveness of varying degrees of harvest intensities for fire risk reduction against operational costs and revenues associated with alternative fuels reduction strategies. To support this project, information needed to gain better understanding of the opportunities and obstacles associated with fuel removal activities on federal lands has been gathered from the scientific literature, government publications, and personal interviews with forestry professionals and community representatives.

2.1 Literature and Government Report Review

A review of pertinent elements of the scientific literature and various government reports was completed in order to achieve several informational goals identified as important to the results of this project. In addition to general background information on the history and magnitude of wildfire risk associated with overstocked forests, other information including but not limited to logging and hauling costs, forest product types and values, Forest Service administration costs, Forest Service contracting authorities, community demographics and infrastructures, etc. has been assembled to best inform this investigation.

2.2 Local Interviews

There is a wealth of experience and knowledge held by members of local communities (USDA Forest Service USDI and Western Governors' Association 2001). Interviews were conducted with public agency professionals, loggers, foresters, consultants, and economic development organizations. These interviews provided information on local costs, values, and perceived opportunities. Interviews were conducted by telephone and in person. The interviews covered questions regarding quantitative information such as costs of operation and qualitative discussions including experiences of treating overstocked stands. Respondents shared opinions and commentary on how to improve and streamline current fuel removal techniques and contract options. In order to respect respondent privacy, specific details and data will not be associated with specific contractors or organizations, however, a list of interviewees is presented within the bibliography (local interviews 2002).

Many individuals generously contributed information founded upon their professional judgment and personal experiences. For example, operational cost estimates provided by private contractors served to enrich the quality of cost data from other sources. Suggestions from local people on how to customize Forest Service contract offerings for increased efficiencies proved essential for better understanding of operational possibilities. The benefits provided to this project from personal interviews served to underscore a recurring theme in this project: successful fuel reduction strategies will likely be based upon integration of anecdotal and institutional knowledge. Through this investigation interview data, research information from the literature, and results of modeling analysis are presented together by topic of interest.

2.3 Technical Tools

Evolving forestry and mapping technologies can help foresters design complex fuel reduction strategies that integrate multiple management goals. This project has developed a parametric sensitivity analysis that utilizes available technologies to analyze a range of conditions that apply to customized local needs. Software capabilities employed for use in this project include the Landscape Management System (LMS), Forest Vegetation Simulator (FVS), Fire and Fuels Extension (FFE), Microsoft Access , Microsoft Excel , and ESRI ArcView . Customized software functionality created for use in this project included spreadsheets and databases to assess performance metrics.

     The Landscape Management System

The effects of forest management alternatives on fire risk reductions, forest product outputs, and economic metrics, were simulated using the Landscape Management System (LMS). LMS is an evolving computer-based, landscape-level forestry analysis software tool developed at the University of Washington College of Forest Resources (McCarter 1997, McCarter et al. 1998, McCarter 2001). LMS offers a software platform for the integration of component software capabilities that include growth and yield models, interactive stand treatment simulation programs, tabular and graphical analytical outputs, and stand and landscape visualization programs. Data sources necessary for LMS include stand inventory information (tree-based measurements), landscape data (slope, aspect, elevation, site quality), and Geographic Information System (GIS) spatial data (stand boundaries, streams, roads, etc.). Using LMS it is possible to simulate stand and landscape inventory growth projections forward in time to create predictions of potential stand and landscape forest conditions. Simulations may be created that include harvesting, regeneration, and wildfire. The graphical user interface (GUI) within LMS is designed to provide a user-friendly "click and go" command format. The intended result is that this powerful forestry software is available for use by individuals with minimum computer skills and limited financial resources. Consequently, LMS has proven to be useful not only as a powerful analysis support tool for forestry professionals but also as a communication tool for use with stakeholder groups embarked on the often conflict-vulnerable process of consensus building. LMS is available for download and provided at no charge through a forestry research partnership between the University of Washington and Yale University. The web site address is http://lms.cfr.washington.edu/.

     Forest Vegetation Simulator

The Forest Vegetation Simulator (FVS) is an individual-tree, distance-independent growth and yield model (Crookston 1990, Van Dyck 2000). FVS will simulate growth and yield for most major forest tree species, forest types, and stand conditions. FVS can simulate a wide range of silvicultural treatments. Variants of FVS provide growth and yield models for specific geographic areas of the United States. Prognosis (Stage 1973) is the original model that evolved into the Forest Vegetation Simulator. Stage developed Prognosis for use in the Inland Empire area of Idaho and Montana. In the early 1980s, the National Forest System's Timber Management Staff selected the individual-tree, distance-independent model form as the nationally supported framework for growth and yield modeling. Over the following years, the Forest Management Staff's Growth and Yield Unit incorporated much of the Prognosis modular structure and capabilities into the national model framework. This model framework is the Forest Vegetation Simulator, or FVS (Wykoff et al. 1982).

There are 21 different FVS variants. Each is calibrated to a specific geographic area of the United States. Various extensions are available for some of the variants. These extensions provide the ability to estimate the influence of other agents upon tree growth (such as insects, disease, and fire), extend FVS modeling capabilities, and permit multiple stand simulation. For the simulations needed for this investigation the East Cascades Variant (EC) of FVS and the South Central Oregon and Northeastern California Variant (SORNEC) of FVS were selected for use within LMS to contribute growth-modeling capabilities for the Okanogan National Forest and the Fremont National Forest respectively. More information and a suite of FVS regional variants are available from the USFS web site at: http://www.fs.fed.us/fmsc/fvs/.

     Fire and Fuels Extension to the Forest Vegetation Simulator

The Fire and Fuels Extension to the Forest Vegetation Simulator (FFE-FVS) links the existing FVS, models that represent fire and fire-effects, with newly developed fuels dynamics and crowning submodels (Beukema et al. 1997, Scott and Reinhardt 2001). The Fire and Fuels Extension (FFE) has been developed to assess, risk, behavior, and impact of fire in forest ecosystems (Beukema et al. 2002). FFE can produce reports of changes in various indices of potential fire severity as a result of alterations to stand characteristics resulting from simulated management alternatives. More information and downloadable FFE for use with selected variants of FVS are available from the USFS web site at: http://www.fs.fed.us/fmsc/fvs/.

The 35-year simulations selected for this project were performed using the East Cascades (EC) and Southern Oregon and Northeast California (SORNEC) variants of the Forest Vegetation Simulator (FVS) with the Fires and Fuels Extension (FFE) within the Landscape Management System (LMS). 35 years was used as a time horizon to consider the durability of post harvest risk reduction for each simulation into the future.

This project uses the capabilities of the Fire and Fuel Extension to the Forest Vegetation Simulator (FVS). The Fire and Fuel Extension (FFE) was developed to integrate the FVS growth model with elements of existing models of fire effects and behavior (Beukema et al. 2002). The FVS growth model was used to simulate tree growth and silvicultural treatments. FFE was used to assess fire risk for all scenarios and fire impacts for the wildfire simulation. Both tool were run from the Landscape Management System to allow additional analytical procedures which addressed other research questions. Inventory information from 1995 was used to simulate stand growth to 2030, with treatments performed in 2000.

2.4 Data and Information

     Current Vegetation Survey

Conditions on the Fremont and Okanogan National Forests were represented, simulated, and analyzed using the Current Vegetation Survey (CVS) Occasion 1 data sets. Data for these national forests was collected during the period from 1994 to 1996. Re-measurements of many plots occurred during successive panels of CVS Occasion 2, but full re-measurement data was not available for both forests. As a result, CVS Occasion 1 data, with a base year of 1995, was selected to provide the forest inventory information used to undertake the simulation analysis required for this study. The 1995 data were "grown" forward within FVS for one growth period to 2000 to bring data close to present time before treatment simulations were conducted.

The Fremont National Forest contains 601 total CVS plots. Plots with dominant species by basal area of lodgepole pine (Pinus contorta), ponderosa pine (Pinus ponderosa), or white fir (Abies concolor) were used in the analysis. Plots with other dominant species associated with higher-elevation long duration fire cycles or non-forested plots associated with grasslands, rocky outcrops, or water were not considered in this analysis. For the Fremont National Forest 61 plots were dominated by juniper (Juniperus occidentalis). While these areas may well benefit from fuel reduction, presently there is no growth model for this species. For this reason the plots dominated by juniper were discarded from the data set as well. It is worthy of mention, however, that juniper harvests could augment feedstock supplies for biomass-to-energy projects and that juniper removals will likely reduce overall forest fire risk (Swan 2002). A total of 502 plots or 84% of the total plots were selected as forested areas to be evaluated for treatment simulations.
A total of 663 CVS plots were available from the Okanogan National Forest. Plots used in the analysis were those in which the dominant species, determined by basal area, was ponderosa pine, lodgepole pine, Douglas fir (Pseudotsuga menziesii), or western larch (Larix occidentalis). Plots with other dominant species associated with higher elevation long duration fire cycles and non-forested plots were considered not suitable and were removed from the data set used for this analysis. The number of plots used in the simulations for the Okanogan National Forest was 413 or 62% of the total available CVS plots.

The selected 502 plots (FNF) and the 413 plots (ONF) from the CVS database were used to create two forest inventory datasets representative of the variety and distribution of forest age classes, densities, tree species, tree sizes, and crown characteristics present in the ONF and the FNF that would be subject to consideration for hazardous fuel reduction treatments. For purposes of conducting forest-wide simulations, the data from each plot has been assumed to represent the inventory of a one-acre forest stand. Subsequently, the simulated FNF will have a 502 acre "forest" and the simulated ONF will have a 413 acre "forest". To expand per acre volumes from CVS data for landscape inventory estimates, one would use 1849.6 as an expansion factor resulting from the 1.7 mile grid used to systematically distribute CVS sampling point locations.

2.5 Growth, Treatment, and Wildfire Simulation

To analyze the relative effectiveness of alternate harvest treatment intensities on fire risk reduction and the subsequent economic results, four silvicultural prescriptions were developed to conduct harvest simulations for each CVS plot for the Fremont and Okanogan National Forests. A NO Action simulation of growth without disturbance and a Wildfire simulation where all acres were ignited were conducted to represent opposite ends of a control spectrum to evaluate do nothing verses the consequences of potential fire disturbances verses effectiveness of the risk reduction treatments. The four harvest treatments were selected to span a range of removal intensities, removing various categories of trees from the very small to the very large and with both fixed and variable density targets. The treatment alternatives were selected, as well, to be readily comparable to simulation findings emerging from other fire risk reduction research projects. All harvest simulations growth projections were done using variants of FVS within LMS. The East Cascades (EC) Variant of FVS was used for the Okanogan inventories, and the South Central Oregon and Northeastern California (SORNEC) Variant of FVS was used for the Fremont inventories. Simulated treatments were conducted in 2000. Post treatment inventories were grown forward to 2030 using 5-year growth simulation periods. A set of results were developed with and without ingrowth. Alternatives modeled included no action (no treatment or disturbance within the study period), four different harvest treatments, and a wildfire simulation.
The six treatment prescriptions that were developed to investigate the response of different forest types to different treatment strategies include:

• No Action (No action). This prescription assumes no harvest activities and no wildfire for the duration of the simulation period. While no wildfire seems an unrealistic expectation, this simulation is valuable to display increases in risk for the forest landscape over time.
  
  
• Remove 9" and Under, (9 and under). This prescription harvests all trees 9" in diameter at breast height (DBH) and smaller. This treatment represents an approach in use by the Forest Service and recommended by Babbitt and Glickman in 2000.
  
  
• Remove 50% BA, From Below (Half BA). This treatment is a removal of half of the total basal area (BA)/acre by removing the smallest trees (thinning from below).
  
  
• Leave 45 sqft BA, From Below (BA 45). This treatment is intended to simulate restoration of savannah-like conditions that are similar to what has been described in literature as the pre-settlement open-stand conditions that resulted from frequent but low wildfires (Agee 1993). In the FNF, all ponderosa pine were left standing, while in the Okanogan both ponderosa pine and western larch were favored as leave trees. In both cases, these species were selected for retention in order to help restoration these forests to what is considered to be an open pre-settlement condition dominated by thick bark fire tolerant species. For an example of what BA 45 means as a management target consider that if trees are approximately 12" DBH then at BA 45 approximately 57 trees per acre (TPA) would be left after harvest.
  
  &TPA = BA/DBH²*.005454
  
  
• Remove 12" and Greater, From Above (12 and over). This treatment is to simulate harvest designed to maximize profit by taking the largest and most valuable trees that are 12" DBH and larger. This practice was commonly known as high grading in the first half of the twentieth century. This simulation conservatively estimates the value of stand inventories at risk from wildfire.
  
  
• Wildfire Simulation (Wildfire). This simulation is undertaken to demonstrate the levels of mortality for different stand inventories that might be associated with wildfire. The wildfire was simulated using the FFE extension within FVS. Burn conditions to be specified in the model were a temperature of 70 degrees Fahrenheit, a wind speed at 20 feet in the stand of 20 miles per hour, and nominal moisture levels of "very dry" (Crookston et al. 2002).

All treatments and the wildfire simulation occurred in year 2000. The four thinning treatments modeled included a removal of all trees with a DBH less than or equal to nine inches (9 and under); a thin from below removing 50% of the original basal area (Half BA); a thin from below with a residual basal area target of 45 square feet favoring ponderosa pine and western larch (BA 45); and a removal of all trees with a DBH greater than or equal to 12 inches (12 and over).

Results were produced for each alternative with and without regeneration to simulate either controlled burn fuel removal or fire risk impacts associated with accumulating fuel loads from ingrowth. Simulations with regeneration were modeled to have a stocking level of 500 trees per acre 4 years after a treatment or wildfire. The distribution of species for the new seedlings was based on the distribution of species by basal area in the residual stand. No action simulations received no regeneration. All simulations including No action utilize FVS to "grow" existing inventories (including regeneration where applicable) forward through time to the end of the simulations period at 2030.

2.6 Analysis of Fire Risk

Simulations that were treated but not assigned regeneration inventories suggest outcomes in 2030 that might be achieved if controlled burns or other treatments remove additions to fuel loads that might otherwise result from ingrowth. Simulations that were treated and assigned post-treatment regeneration inventories (ingrowth) are conducted to demonstrate that effective long term risk reduction will likely require some type of post treatment fuels control.
Each alternative was analyzed to examine the impact on fire risk and economic net revenue. Initial forest conditions were analyzed for each national forest for fire risk, forest structure, and species composition. Additionally, each CVS plot was analyzed at each projection cycle for fire risk.

     Fire Risk Classification

High, moderate, and low fire risk was estimated for each CVS plot in the simulation dataset based on the Severe Crowning Index assessment from the Potential Fire Report produced by FFE. The Crowning Index indicates the estimated wind speed in miles per hour (mph) at 20 feet off the ground that would initiate an active crown fire assuming ignition of a surface fire. Assumptions required by the model include a temperature of 70 degrees Fahrenheit and 'very dry' moisture conditions (Crookston et al. 2002). Results from the crowning index estimates for each stand were sorted into one of three risk classes. Lower wind speeds indicate greater risk. If the crowning index was less than 25 mph, then the plot was considered to be in the high fire risk category. Moderate risk stands were those with a Severe Crowning Index between 25 and 50 mph. Low fire risk stands were those with a crowning index greater than 50 mph. It should be noted here that risk classifications are arbitrary thresholds useful and necessary for comparative analysis but that they may very well understate the risk at the margins. The word moderate properly segments a risk difference between high and low although the risk of a fire from wind speeds only slightly higher than 25 MPH might not be considered a moderate risk by many publics. It is in part for this reason that performance comparisons for treatment alternatives were reported in this investigation for high and moderate as separate risk classes and then combined as the total area to be considered for risk reduction treatments.

The initial 1995 fire risk distribution for the Fremont and Okanogan National Forests was reported as the percentage of CVS plots in each of the fire risk categories prior to any treatment or growth simulation. Fire risk distribution was similarly reported for projected and treated plot inventories at each growth cycle. For purposes of simulations to demonstrate comparisons between treatment alternatives only the plots with high and moderate initial classification were treated. Low risk areas did not receive treatment simulations since treatments of low risk areas would logically be considered unnecessary or of low priority. While some low risk areas may experience increases in risk over time most low risk areas appear to be either very young small diameter forests or rangeland/forest interface with sparse distributions of forest inventories that are unlikely to require fuels reductions at the time of this study.

     Forest Structure

Forest structure was determined using an approach utilized by the Business Bureau of Economic Research in Montana in "A strategic assessment of fire hazard in Montana" (Fiedler et al. 2001). This canopy structure classification system identifies stands as being single-layered, two-layered, multi-layered, or scattered. Five potential layers could be present in a stand, based on a minimum amount of basal area in a diameter size class. The sapling size class required at least 5 square feet of basal area in trees with less than 5" DBH to be present. The pole, medium, large, and very large size classes included trees within a DBH range of 5-9", 10-15", 11-20", and greater than 20", respectively. These size classes required at least 10 square feet of basal area to be present to be considered as a canopy layer. Stands classified as single- and double-layered had one and two size class layers present, respectively. Multi-layered stands had more than two layers present. Scattered stands had no layers present and at least 25 square feet of basal area in the stand. Using this system, each plot was classified in 1995. The initial stand structure distribution for each landscape was determined as the percentage of plots in each category.

     Forest Type

Forest type for the Fremont National Forest was determined based on criteria provided by the Sue Puddy, the Silviculturist at the Fremont National Forest. This classification system identified plots by dominant species and structure. The categories were Ponderosa Pine Closed, Ponderosa Pine Moderate, Ponderosa Pine Open, Ponderosa Pine Very Open, Juniper, Lodgepole Pine, Mixed Closed, and Mixed Open. Plots with at least 12 ponderosa pine tpa with a DBH greater then 14" were classified as one of the Ponderosa Pine types. The canopy closure algorithm by Crookston and Stage (1999) was then used to distinguish Ponderosa Pine Closed (greater than 50% canopy closure), Ponderosa Pine Moderate (35-50%), Ponderosa Pine Open (25-35%), and Ponderosa Pine Very Open (less than 25%). Plots with greater than or equal to 70% of the tpa in juniper were classified as Juniper. The Lodgepole Pine forest type was defined by plots with greater than or equal to 50% of the total tpa in lodgepole pine and less than 15% of the total tpa in ponderosa pine trees with a DBH greater than 10". Plots in the Mixed forest types were classified as not meeting any of the above criteria. Mixed Closed plots had greater than 40% canopy closure. Mixed Open plots had less than or equal to 40% canopy closure. Forest type distribution for both the Fremont and Okanogan National Forests were reported as the percentage of plots in each structure type in 1995.

Forest type was reported for each national forest for conditions in the initial time period. Forest type for the Okanogan National Forest indicated the percentage of plots in Cold Dry, Dry, Mesic, and Moist conditions. Cold dry forests typically have mixed mortality fires in an elevation range from 6000-7200 ft. Dry forests have 7.5-50 year fire return intervals and are found from 1,200 to 5000 ft. Mesic forests experience weather driven catastrophic fire events every 100 or more years found in a wide elevation range from 1800-6000 ft. Moist forests are 100 to 300 year fire return interval found in mid elevations of 3000-4500 ft (Northeastern Cascades Late-Successional Reserve Assessment Team 1998). The forest type was determined using plot locations, which are UTM coordinates, for each CVS plot which were "joined" in the GIS with a forest type layer provided by John Townsley, the Silviculturalist at the Okanogan National Forest.

2.7 Analysis of Economics

Each of the four harvest alternatives were also analyzed to examine the positive or negative economic net revenue that resulted when estimated harvest and hauling costs were subtracted from the gross revenues from sale of estimated log yields. Interviews provided a range of primary data regarding local logging and hauling costs and log values by grade and species per thousand board feet (MBF). Secondary data was also gathered from available Forest Service documents and market reports that show the current market opportunities and trends of historic of log prices. The collected cost information for operational costs/acre, and average log values/MBF that were incorporated into the economic evaluation of the treatment alternatives examined in this project are described later in the text. Within LMS, a bucking algorithm was used to optimize estimates of log segments that result from trees harvested in simulated silvicultural treatments. Estimated volumes of logs by grade and species from harvest simulations were multiplied by delivered log prices to estimate gross harvest revenue for each stand (plot). Microsoft Access™ and Excel™ were used to compute the gross and net revenue per acre based upon subtraction of local logging costs, and hauling costs from local market log values by species and grade. In some cases, effective fuel reductions required the removal of non-merchantable small diameter trees. Where this was the case an additional operational cost/acre referred to in this study as pre-commercial thinning (PCT) was charged against gross revenues to complete the economic analysis for each stand. The gross log value/acre minus the logging, hauling, and (PCT) costs equals net economic return per acre. Estimates of preparation, administration and litigation costs to the USFS are not considered in this analysis but have historically been significant as noted in the USFS publication "The Process Predicament" (USDA Forest Service 2002b).

     Conversions

To utilize the specific logging costs, hauling costs, and log value estimates that were gathered from interviews and publication, some numbers required conversion from tons to thousand board feet. Forestry professionals from both National Forests were interviewed for the appropriate conversion factor to use. Weight to volume conversion factors are by nature variable due to water content in log, tree species, and time since the log has been felled. A conversion rate of 7 tons/MBF was agreed to be most generally representative and was selected to be used to convert some costs and values based on tonnage into $/MBF. Table 1 shows the range of conversion factors that resulted from local interviews (local interviews 2002).

Table 1 Tons Per Thousand Board Feet (MBF) for Eastern Washington and Oregon

     Logging and Hauling Costs

Eighteen logging contractors were interviewed in Oregon and Washington. These loggers were willing to share information on the logging equipment mixes that they have, the costs to log with their equipment, and the cost to haul the wood to the mill. The haul costs were assigned based on the interview results according to the average haul the loggers suggested for each forest. Harvest operations costs estimates collected from these contractors include both cable and ground based logging operations. Table 2 shows the high, low and average logging hauling and PCT costs per acre. These costs were assigned by calculating an average of all the high and low operations costs collected from contractors for each forest. These figures were used for economic valuation of thinning simulations for the FNF and ONF. The PCT costs are included to estimate the range of costs required to thin some of the non-merchantable stems in conjunction with the removal of any merchantable material. A low PCT cost of $300/acre and a high of $500/acre were used to simulate treatment of non-merchantable material as part of fuel reductions in any stand with greater than 200 TPA 6" in diameter or smaller. The interviews with contractors and USFS employees suggested 200 TPA, of submerchantable material as the threshold of when PCT costs become a realistic addition to logging costs. PCT costs include removal of submerchantable material to the road or landing. This material could be used as biomass fuel for energy generation, but has historically not been economically feasible to remove to a conversion site.

Table 2 FNF and ONF Low and High Logging, Hauling/MBF and PCT Costs/Acre

There is a high degree of variability in logging and hauling costs suggested by interview respondents. To demonstrate a representative range of potential operations costs, simulated harvest yields were analyzed for both high and low cost for the four thinning treatments. Interviews with many employees in the Forest Service, Department of Natural Resources, and Oregon Department of Forestry served to confirm contractor cost estimates and validate the range of costs per acre. Several factors including equipment, terrain, contract specifications, and density of stand are known to influence operating cost variability. In addition, many of those interviewed commented on their experiences logging for the USFS compared to logging on private land. Some contractors reported that higher operations charges were necessary to profitably operate on federal forests as opposed to private or state owned forest lands. Other contractors reported that as a result of unfavorable experiences with USFS contracts that they only work on private land now. Interview comments suggested that the many complicated factors regarding contract requirements for harvest activities on federal lands have made such operations difficult and expensive.

     Mill Log Values

Logs that are removed during fuel reduction thinnings, can include a mixture of non-merchantable trees, pulp logs and sawlogs. Interviews with mills around the FNF and ONF were combined with log price market reports to estimate delivered log prices. Prices in this study are current as of August 2002. Table 3 shows the prices collected from nine mills and three regional log value reports.

Table 3 Regional Log Sort Values $/MBF Used for Economic Valuation

     Net Revenue Calculation

Volumes of harvest were simulated for alternative thinning treatments and divided to estimate potential yield volumes by species and grade estimated by the bucking algorithm in LMS. Estimates of merchantable volumes of pulp and sawlogs were divided by species and grade. Each species and grade volume was multiplied by the assigned price per MBF. Gross estimated revenues from log sales were determined for each stand and for each treatment alternative by summing species and grade returns for each stand. Treatment costs were subtracted from gross revenue from log sales to determine net revenue. The average net revenues for each treatment type by risk class were calculated for comparison with risk reduction success resulting from each treatment alternative. Risk reduction and the associated economic results when compared for each treatment alternative are presented to offer dual measures of effectiveness. Such comparisons are valuable to foresters planning treatments for maximum risk reduction at least cost. Since harvesting and hauling contractor costs are subtracted from gross revenues, the resultant positive or negative net return from each treatment simulation may be considered indicative of either potential timber sale revenues (theoretical bid maximums in excess of operational costs in the case of a positive returns) or stewardship costs for risk reduction (while logs may not be profitable from a timber sale perspective they do have sufficient value to discount fuel reduction costs when included in a combined goods and services transaction). Neutral returns mean that the value of the logs harvested will cover the risk reduction treatment costs but do not have sufficient value to warrant a timber sale offering. In some cases foresters may want to combine treatments and stands such that the positive revenues available from one stand fuel reduction harvest can be used to offset the negative revenues associated with another stand fuel reduction harvest. This may be done to create a service contract that is cost neutral or a timber sale where the harvest value of some stands carries the cost of fuel reductions for other stands and still yields positive revenue.

3. Case Study Site Descriptions

The Fremont National Forest (FNF) in Oregon and the Okanogan National Forest (ONF) in Washington were selected as study areas for this project. Both of these National Forests are located within the dry interior portion of the western United States. Both of these forests are thought to have had frequent fire return cycles, prior to European settlement, that created many areas that were dominated by open stands of ponderosa pine (Fremont National Forest 2003, Okanogan National Forest 2003). Both the FNF and the ONF contain substantial acreages of overstocked forests that are considered to be at risk from wildfire Table 4. Both National Forests have experienced destructive wildfires in recent years. The communities surrounding these two forests provide a variety of infrastructure options to remove fuel from overstocked stands. However, the logging, transportation, and processing of low value smaller wood has historically not been profitable. The rural communities surrounding these forests have experienced double-digit unemployment and economic declines due to job losses associated with reductions in federal timber harvest volumes. Individuals, organizations, and businesses from both areas demonstrated interest in this investigation and contributed valuable reference information through personal interviews. The citizens of these local Oregon (OR) and Washington (WA) communities appear eager to meet the challenge to remove the fuel from forests with a high fire danger.

Table 4 Acres in Initial Fire Risk Class for Forests on FNF and ONF

3.1 Fremont National Forest

The FNF is in the south central dry interior of the state of Oregon. There are several rural communities surrounding the FNF boundary. The forest lies roughly between the towns of Lakeview, Klamath Falls and Bend, Oregon just North of the California/Oregon border Figure 3. Also see Appendix D for forest maps.

Figure 3 Fremont National Forest Boundary

The majority of the 1,198,301 acres within the boundary of the FNF are in Lake County which is 8,359 square miles. The population of Lake County is 7,470 and neighboring Klamath County is 64,116 (US Census Bureau 2003). The town of Lakeview is at the Southeastern corner of the FNF close to the California border and has a population of 2,800 (Fremont National Forest 2003). Lake County has .9 people per square mile. The unemployment rate for 2002 was 8.7% and it was 10.4% in 2001. The high for the 1990's was an unemployment rate of 12.2% (Lake county 2003). "Lake County was also the only county in the state that experienced a net job loss during the 1990's" (Kauffman 2001).

The forest includes wildlife and tree species adapted to the climate and elevation variation from 5,000 to 7,000 feet with mild terrain on slopes roughly 40% and less. About half the FNF is a mixed open forest type, see Figure 4, with multi-structured canopies.

Figure 4 FNF Forest Type Distribution

Half of the forest plots have greater than 500 Trees per acre (TPA) and a Basal Area (BA) ranging from 50-150. The quadratic mean diameter (QMD) is 12" and less for over 90% of the forestland. The major tree species include ponderosa pine (Pinus ponderosa), juniper (Juniperus occidentalis), lodgepole pine (Pinus contorta), and at higher elevations white fir (Abies concolor). Most of these trees are adapted to summer drought and extreme temperature fluctuations due to the nature of the arid region (Fremont National Forest 2003). The 10-20 inches average precipitation occurs from the autumn through the spring and as a result the summers are dry and hot. See the Appendix A for detailed charts of the original forest conditions. Of the 502 plots on the FNF the fire risk distribution is 154 high, 236 moderate, 112 low risk.

Figure 5 shows high moderate and low fire risk by percentage of total forest.

Figure 5 FNF Fire Risk Distribution

Fire risk analysis conducted with FFE indicate that 77.8% of FNF is presently in a moderate to high risk condition with 30.7% of the FNF considered to be high fire risk forests. The FNF has experienced destructive wildfires in recent years. In 2002 over 125,000 acres of the FNF burned due to wildfire (local interviews 2002).

In 2001 22 MMBF were harvested from FNF (local interviews 2002). There is one mill in Lakeview, Oregon. However, as many as 4 mills in the surrounding area receive logs from the FNF. The Fremont National Forest also has a sustained yield unit, and network of local efforts and non-profit organizations working to maintain forestry infrastructure.

3.2 Okanogan National Forest

The Okanogan National Forest (ONF) is located in North central Washington. In 2000, the ONF was merged with the Wenatchee National Forest to become the Okanogan-Wenatchee National Forest (OWNF). Okanogan is the Northern portion of what is now the Okanogan-Wenatchee NF.

Figure 6 Okanogan National Forest boundary

Figure 6 shows the original Okanogan National Forest boundary. Also see Appendix B for forest map. The Okanogan National Forest consists of 1,226,550 acres total that are spread across four counties including Skagit, Whatcom, Okanogan, and Chelan counties. The population is most concentrated around the towns of Omak and Okanogan. Several of the other towns close to the forest include Oroville, Tonasket, Twisp, Brewster, Winthrop, Chelan, and Leavenworth. The population in 2001 for Okanogan County was over 20,100 and for Chelan County 67,000 (US Census Bureau 2003). The unemployment rate for the past ten years has been over 10% for Okanogan County. In 2000 it was 11% and 2001 it was 10.8%.

     ONF is predominately dry forest type ranging in elevation from 3000 to 6000 feet in elevation

Figure 7. The Okanogan National Forest is dominated by multistructured Douglas-fir (Pseudotsuga menziesii), lodgepole pine (Pinus contorta), ponderosa pine (Pinus ponderosa), and Western larch (Larix occidentalis) with a QMD of less than 12". The BA of the forest plots is concentrated from 50 to 200 sqft. The tree densities for the majority of the forest plots range from 250 to 4000 TPA. Most of these trees in the ONF are adapted to summer drought; high summer temperatures of 90 degrees Fahrenheit are not uncommon (Okanogan National Forest 2003). The 20-40 inches average precipitation occurs from the autumn through the spring and as a result the summers are dry and hot (local interviews 2002). See Appendix B for detailed charts of the original forest conditions.

Figure 7 ONF Forest Type Distribution

FFE analysis of the ONF data in Washington indicates that of the 413 stands total, 117 (28.3%) are classified as high risk, 200 stands (48.4%) are moderate risk, and 96 stands (23.2%) are low fire risk Figure 8. There have been large fire years in the past decade in 2000 and 1994 on the ONF.

Figure 8 ONF Fire Risk Distribution

The mills and infrastructure surrounding the ONF are further distances away from the forest compared to the mills around the FNF increasing the haul distance and cost required to transport the wood to be processing facilities. There are also a wider variety of species in ONF potentially available for harvest that may require hauling trees to a wider variety of mills. The total harvest at Okanogan for 2001 was .1 MMBF. Whereas in 1989 the 10 year average harvest volume was 71 MMBF for the Okanogan National Forest (Okanogan National Forest 1989).

4. Results

4.1 Fire Risk Results

     Fremont National Forest

As shown in

Figure 5, 30.7% of the 502 plots on the FNF are in high fire risk classification. There appear to be some common characteristics of high fire risk stands on the FNF. High risk stands have thin bark species and multi-layered canopies that are indicators of past fire suppression (Agee 1993). Figures 9 and 10 display the occurrence of these common traits of high fire risk stands on FNF. The majority of the stands designated as high fire risk are white fir dominated (53.0%) and have multi-layered canopies (94%). The presence of white fir, a thin barked shade tolerant and fire intolerant species, as well as, multi-layered canopies would indicate that wildfire has been successfully suppressed and that the current condition is not reflective of historic frequent ground fire conditions (Hopkins 1981).

A	B
Figure 9 A and B FNF High Fire Risk Species and Structure Distributions

Conversely, the low risk stands are dominated by ponderosa pine and scattered structure canopy. The scattered structure would suggest that many low risk stands could be low density rangeland/forest interface. See the Appendix A for a full set of tables and charts that display initial and post-treatment forest conditions.

The simulations results indicate that the thinning treatments, Half BA and BA 45, may be the most effective in reducing fire risk in high and moderate risk forests. After the Half BA treatment, 55.4% of high and moderate risk forests were transitioned to low risk while after the BA 45 treatment, 63.8% of high and moderate risk forests were transitioned to low risk. This is compared with the 9 and under treatment which results in only 22.1% of the stands in the low fire risk category after treatment. Somewhat unexpectedly the 12 and over treatment results showed a transition of only 20.5% to low risk from high and moderate risk stands. See Appendix A for detailed tables showing risk transitions from one risk class to the next.

Stands designated as in a high risk forest condition logically represent those forested areas with the most opportunity for fire risk reduction. Figure 10 shows that the greatest reduction of risk resulting from fuel removal treatments occurs with the BA 45. Figure 10 shows the response to treatment with regeneration included in the simulations. This is equivalent to thinning and not planning any follow up fuels reduction treatments such as future thinning or burning to control fuel build up from regeneration ingrowth. Subsequently, 15 to 20 years after fuel reduction treatments, fire risk begins to increase dramatically with accumulation of new trees in the understory, suggesting that entries for ingrowth removals should commence 10-15 years after treatment to prevent future risk.

Conversely, Figure 11 is intended to display the forest risk through time as it might be with a control burn program to remove risk from ingrowth. Subsequently reductions in risk are maintained into the future by excluding regeneration from the simulations. Amongst the high risk stands both the BA 45 and the Half BA treatments reduce the risk in most of the stands; 151 or (98.1%) of the BA 45 and 144 or (93.5%) of the Half BA treated stands moved from high risk to either moderate or low risk status. However, the BA 45 treatment resulted in many more stands dropping from high to low risk status than the Half BA treatment. 110 plots (71.4%) went from high to low risk for the BA 45 treatment while only 42 plots (27.3%) went from high to low risk as a result of the Half BA treatment. The post-treatment growth simulations without regeneration for stands originally at high risk indicate that stands in low risk classification actually increase slightly (average 4-5% depending on treatment) over time for the FNF, see Appendix A. This may be due to reductions to ladder fuels associated with growth of leave trees. See Appendix A for the full set of fire risk response results for FNF and Appendix B for ONF.

Figure 10 FNF High Fire Risk Response to Six Simulations with Regeneration

Figure 11 FNF High Fire Risk Response with No Regeneration after Treatment

All three thinning from below treatments result in substantive risk reduction as does the Wildfire simulation Table 5. The 12 and over treatment does result in some risk reduction but most change is within the moderate risk. No action simulations result in net risk increases on high and moderate stands in the FNF. The Wildfire simulations result in near total mortality from crown fires in both high and moderate fire risk stands on FNF.

Table 5 FNF fire risk reduction treatment metrics

     Okanogan National Forest

High fire risk stands comprise 28.3% of the Okanogan National Forest stand data examined in this investigation. The most prevalent species in ONF stands is Douglas-fir (64.2%). While Douglas-fir is not necessarily a thin barked species in inland west ecosystems it functions as a late seral shade tolerant species in ONF forests that is present when fire return cycles become infrequent (Agee 1993). Within high fire risk stands in the ONF, the dominant forest species is Douglas fir (72.7%) and multistructured canopies (89.7%) are the dominant forest structure. Both FNF and ONF appear to have high risk forest characteristics of multi-storied canopy and late seral dominant species that are likely a result of prolonged fire exclusion. While low risk stands in the ONF are dominated by Douglas-fir they appear to have a higher percentage of ponderosa pine than high and moderate risk areas as well as a scattered canopy structure. See Appendix B for the full set of fire risk response results for ONF.

A	B
Figure 12 A and B ONF High Fire Risk Species and Structure Distributions.

The response of the high and moderate risk classes to treatment simulations indicates that the treatment, BA 45, was the most effective in reducing the risk. Post-treatment results for BA 45 show 72.5% of high and moderate stands transitioned to low risk status. The Half BA treatment resulted in 56.2% of stands going to low risk after treatment. This is compared with the 9 and under which resulted in 35% of the stands in the low fire risk category. The 12 and over treatment resulted in 17.4% of high and moderate risk stands going to low risk, however, as with the FNF, most of these stands originate as moderate risk. No action resulted in increased numbers of stands in high risk over time. Due to the climate and inventory differences between the ONF and the FNF, the Wildfire simulation on the ONF did not result in near total mortality to all forest inventories as it did in the FNF. High and Moderate risk forests in the ONF retained a mean TPA of 40 after wildfire with 71.3% reduced to low risk.

The high fire risk stands represent the highest level of fuel and the most critical opportunity for fire risk reduction. Simulations indicate that risk reduction under such conditions is maximized with the BA 45 treatment. Figure 13 shows the response to treatment with regeneration included in the simulations. This is equivalent to thinning and not planning any follow up maintenance of fire risk in the form of thinning or burning included in the management plan. Elevated risk levels begin to reappear in treated stands on the ONF after 15-25 years suggesting that future fuel removal entries are necessary to control fire risk.

Figure 14 shows the results when regeneration is excluded from the simulations. Exclusion of regeneration is intended to demonstrate treatment effectiveness when combined with future control of fuel build-up from ingrowth. BA 45 initially reduced high fire risk to low for 62.4% of the stands but, by the end of simulation period in 2030, 46% remained in low risk indicating a 16% increase in risk. Very few stands returned to high risk, however Error! Reference source not found. In 2030, only 5.1% of stands initially designated as high risk with BA 45 remained high risk. While there appears to some change over time from low to moderate risk without regeneration for Half BA and 9 and under very few stands regain high fire risk classification during the study period. This trend is different than what was observed on the FNF where slight reductions in risk occurred over time. The exception is the 12 and over stands that began as high risk. By the end of the simulation period 92% remained as high risk. Forests that were not treated, No action, remained as high risk. See Appendix B for the full set of fire risk response results.

One difference between Okanogan and Fremont is that Okanogan high risk stands that are treated do not return to high risk as soon after treatment as Fremont stands. The wildfire simulation on the ONF does not completely burn up all the high risk stands as does the wildfire simulation for the FNF. Some of the forest stands on ONF are also further west and have a cooler and moister climate. As a result, the fuel moisture content in many high fire risk ONF stands is higher and therefore is less susceptible to complete combustion from wildfire than the stands in the FNF.

Table 6 ONF fire risk reduction treatment metrics