Fire today ManagementVolume 69 • No. 2
Wind Models Fire Research Fuel Models Fire Research Forecasting Suppression Costs Smoke Management
United States Department of Agriculture Forest Service This Issue…
This issue provides a glimpse into the role that research and technology play in the management of fires today and into the future. Over the years, the Forest Service and the interagency fire community have considered not only the science behind fire itself, but also the science of predicting fires and what is likely to happen when a fire-start occurs. Many aspects of fire management—fuels, wildland-urban expansion, and environmental factors among them—are different today than they were even a decade ago, making it more critical than ever to use emerging science and state-of-the-art methods of prediction to keep firefighters and the public safe. The articles in this issue reflect just a few of the models, tools, and approaches that are currently shaping and advancing the science and management of fire to achieve that end.
—Tory Henderson, Issue Coordinator
Erratum In Fire Management Today vol. 69, no. 1 [Winter 2009], the caption for the photo of snow geese near a Marsh Master in the “Myth Busting about Wildlife” article gave an incorrect credit. It should have credited Drew Wilson, Virginia Pilot.
Fire Management Today is published by the Forest Service of the U.S. Department of Agriculture, Washington, DC. The Secretary of Agriculture has determined that the publication of this periodical is necessary in the transaction of the public business required by law of this Department.
Fire Management Today is for sale by the Superintendent of Documents, U.S. Government Printing Office, at: Internet: bookstore.gpo.gov Phone: 202-512-1800 Fax: 202-512-2250 Mail: Stop SSOP, Washington, DC 20402-0001
Fire Management Today is available on the World Wide Web at
Tom Vilsack, Secretary Melissa Frey U.S. Department of Agriculture General Manager
Thomas L. Tidwell, Chief Karen Mora Forest Service Managing Editor
Tom Harbour, Director Editors Fire and Aviation Management EMC Publishing Arts
The U.S. Department of Agriculture (USDA) prohibits discrimination in all its programs and activities on the basis of race, color, national origin, age, disability, and where applicable, sex, marital status, familial status, parental status, religion, sexual orientation, genetic information, political beliefs, reprisal, or because all or part of an individual’s income is derived from any public assistance program. (Not all prohibited bases apply to all programs.) Persons with disabilities who require alternative means for communication of program information (Braille, large print, audio- tape, etc.) should contact USDA’s TARGET Center at (202) 720-2600 (voice and TDD). To file a complaint of discrimi- nation, write USDA, Director, Office of Civil Rights, 1400 Independence Avenue, S.W., Washington, D.C. 20250-9410, or call (800) 795-3272 (voice) or (202) 720-6382 (TDD). USDA is an equal opportunity provider and employer.
September 2009
Trade Names (FMT) The use of trade, firm, or corporation names in this publication is for the information and convenience of the reader. Such use does not constitute an official endorsement of any product or service by the U.S. Department of Agriculture. Individual authors are responsible for the technical accuracy of the material presented in Fire Management Today.
Fire Management Today 2 Fire Management today Volume 69 • No. 2 On the Cover: Contents Anchor Point—Wildland Fire Decision Support System Tools . . . . .4 Tom Harbour The Key Decision Log: Facilitating High Reliability and Organizational Learning ...... 5 Anne E. Black A Multi-Disciplinary Approach to Fire Management Strategy, Suppression Costs, Community Interaction, and Organizational Performance ...... 11 Anne E. Black, Krista Gebert, Sarah McCaffrey, Toddi Steelman, and Janie Canton-Thompson Efforts To Update Firefighter Safety Zone Guidelines ...... 15 Flames run through a stand of trees Bret Butler as winds push a fire forward during the Skinner Fire, Tusayan Ranger Modeling Fuel Succession ...... 18 District, Kaibab National Forest, Brett Davis, Jan van Wagtendonk, Jen Beck, and Kent van Wagtendonk Arizona. Photograph by Mike Uebel, Fuel Age and Fire Spread: Natural Conditions Versus Opportunities Forest Service, Kaibab National Forest, 2005. for Fire Suppression ...... 22 Richard W. Halsey, Jon E. Keeley, and Kit Wilson
The USDA Forest Service’s Fire and Aviation Fuel Treatment Guidebook: Illustrating Treatment Effects Management Staff has adopted a logo on Fire Hazard ...... 29 reflecting three central principles of wildland Morris Johnson, David L. Peterson, and Crystal Raymond fire management: A Suite of Fire, Fuels, and Smoke Management Tools ...... 34 • Innovation: We will respect and value Roger D. Ottmar, Clint S. Wright, and Susan J. Prichard thinking minds, voices, and thoughts of those that challenge the status quo while Using Wind Models To More Effectively Manage Wildfire ...... 40 focusing on the greater good. Brian Potter and Bret Butler • Execution: We will do what we say we Assessing Changes in Canopy Fuels and Potential Fire Behavior will do. Achieving program objectives, improving diversity, and accomplishing Following Ponderosa Pine Restoration ...... 47 targets are essential to our credibility. John Paul Roccaforte, Peter Z. Fulè, and W. Wallace Covington • Discipline: What we do, we will do well. Fiscal, managerial, and operational discipline are at the core of our ability to short Feature fulfill our mission. Web Site on Fire ...... 10
Firefighter and public safety is our first priority.
Volume 69 • No. 2 3 by Tom Harbour Anchor Director, Fire and Aviation Management Point Forest Service
Wildland Fire deCision support system tools
he Forest Service has always new improvements to the Wildland to engage at another time because had a symbiotic relationship Fire Situation Analysis (WFSA), the the risks outweigh the benefits. Tbetween the practitioner and Wildland Fire Implementation Plan the scientist. Throughout the years, (WFIP), and Long-Term Incident When we look to the future, the this relationship has aided the wild- Planning (LTIP) to create one sup- role of science will continue to be land firefighter in becoming a bet- port tool. This system provides very important to the firefighter ter, more efficient, safer firefighter agency administrators and incident on the ground. It will provide on the ground. commanders the ability to use newer, better tools to enhance per- scaleable decision support tools formance, be more cost-effective, Over the past three decades, when considering fire behavior and will provide improved predict- wildland fires have dramatically modeling, economic principles, and ability for line officers and fire increased in size and complex- information technology, and will management officials, allowing ity. Although fires have changed, result in decisions that are safer them to manage fire—regardless the tools in our decision support and more effective for the firefight- of complexity—in the safest, most toolbox have remained relatively efficient, most cost-effective man- constant. The major tool avail- ner. The WFDSS exemplifies our able to line officers and incident When we look to the dedication to working toward this managers—initially, the Escaped future, the role of future for fire management. Fire Situation Analysis (EFSA) and science will continue to now the Wildland Fire Situation Analysis (WFSA)—has not changed be very important to for nearly three decades. It is time the firefighter on the Further Information to adapt! ground. InciWeb may be accessed at
Fire Management Today 4 the Key deCision log: FaCilitating high reliability and organizational learning Anne E. Black
f you were involved in the 2008 fire season in the West, you We asked two questions: “How do you I may have heard the term “Key define success?” and “How do you know Decision Log” or “KDL.” This article describes the KDL concept, when you’re achieving it?” it’s intent (past and present), how it was applied in 2008, and where the “How do you know when you’re Costs, Community Interaction, and practice is heading. achieving it?” The most common Organizational Performance, in this response—meeting the line officer’s issue). We grounded our thinking The KDL’s purpose is to facilitate objectives—was also acknowledged on concepts of high reliability, high continuous learning in fire man- as difficult to impossible to assess performance, and organizational agement processes and outcomes. because, in the current system, learning. It arose out of a dual desire to there often isn’t a measurable continually improve organizational objective to track, some measures From organizational learning performance and to meet societal are difficult to obtain in the midst theory (Garvin 2000) comes the demands for transparency in deci- of a multiteam incident, and it is understanding that learning has sionmaking. Its development is a difficult to assess or gain insight two parts: attention and delibera- story of innovation and feasibility: into the more-effective versus the tion. It doesn’t simply happen of a mix of ‘ivory tower,’ grounded less-effective choices and actions itself. Organizations that learn practice, learning theory, and hard, without also capturing the ratio- successfully do so because they cold reality. KDL takes “applied nale behind them. Moreover, the establish explicit protocols to docu- research”—applying and developing key reasons for embarking on such ment critical processes and then research-based knowledge to under- an effort are to improve current use these in structured reflection, stand a problem—into the realm of incident outcomes, improve next such as in After Action Reviews, to “action research”—an iterative, col- year’s actions, and provide a way uncover the root causes of success- laborative effort taken by research- to describe outcomes, not to grade ful and unsuccessful outcomes. ers and managers to understand participants in a given event. This and improve organizational process requires a standardized and central- From high-performance theory and culture. ized process that promotes reflec- comes the recognition that, while tion and wide dissemination of les- success depends on effective inter- Concept Development sons learned. nal business practices (risk man- The content and structure of agement), achieving and sustaining KDL stem from initial conversa- We combined our 2007 insights high performance also requires tions among geographic area and into a collaborative Joint Fire consistent and effective attention to national incident commanders Science Project involving Forest financial management (operational (ICs), Forest Service fire staff, line Service researchers, academics, efficiency and transparency), work- officers, and researchers during resource agencies’ staffs, and board ing relationships (communication the 2007 fire season in Idaho and members of the Northern Rockies and collaboration), and innovation Montana. We asked two questions: Coordination Group (NRCG) and learning (experimentation, “How do you define success?” and to develop and test a means for error detection, reflection, etc.) capturing, tracking, and under- (Norton and Kaplan 2000). Anne Black is a social scientist/ecolo- standing “progress” and “success” gist for the Forest Service’s Aldo Leopold in incident management (see A Operating with high reliability Wilderness Research Institute and the Rocky Mountain Research Station, Multidisciplinary Approach to Fire (Weick and Sutcliffe 2007) requires Missoula, MT. Management Strategy, Suppression noticing and acting quickly and
Volume 69 • No. 2 5 decisively on small deviations. perspective in which the context and check alignment and progress High reliability theory reminds us and behavior of individual and during an incident (daily check-ins to periodically question whether small group actors, the patterns between line officers and ICs, for we are focusing on the most tell- that emerge from the interaction example). ing information and how well our of these actors across boundaries, interpretation of that informa- and the drivers of these patterns However, many of these procedures tion matches the actual situation. (some of which undoubtedly arise are not consistently practiced, par- Organizational performance theo- from outside the local context) are ticularly as a means to compare ries remind us that where we focus identified. intentions for specific incident attention and how we frame the outcomes. Recent investigations world are as much determined by Theory suggests, then, a process into who conducts AARs, for organizational policies and culture that can track alignment among instance, indicate that AARs aren’t as by individual experience. Thus, partners, reveal the focus of atten- consistently conducted (at least in it becomes critical to track and tion and patterns of focus, and cap- frequency; “quality” is a separate periodically revisit both operational ture the criteria used to interpret topic), nor is there a way to build effectiveness (“Are we noticing all that we need to, and are we inter- preting what we see effectively?”) Continuous improvement in decisionmaking and organizational effectiveness considers both the cultural frame of (“What do our patterns of focus reference that focuses decisionmakers’ reveal about perceived organiza- tional priorities and conflict resolu- attention on specific aspects of a situation tion, and are these the most effec- more than others and the way in which tive?”). this information is interpreted, weighed, and integrated to arrive at a particular decision. Beyond this, we wanted to remain cognizant of “practical drift” in organizational life (Snook 2000): what is noticed in terms of risk these individual and small group as experts of our local systems, management, financial manage- insights into a corporate or collec- we adapt corporate protocols and ment, and partner/stakeholder rela- tive perspective. So the need is to procedures to better match local tions. The process needs to be com- build on current practice by defin- conditions. This works well until pleted at the team, host, and inci- ing a common practice, consistent local adaptations begin to collide dent levels but be consistent across method, and central location so or offset each other. Especially in incidents and centrally collected to that we can better see how (and multilevel and complex situations, allow for developing insights and why) our system operates to pro- periodically checking to ensure that lessons at both operational and duce the outcomes it does at team, all parts of the system are in align- organizational levels. incident, and organizational levels. ment and not working at crosspur- The KDL is intended to do this. poses to each other becomes criti- cally important. From Theory The 2008 Pilot to Practice Practical drift is particularly dif- Working with members of the The fire community is not start- ficult to detect when “minor” NRCG, we developed data collection ing at ground zero; there are a adaptations cause mismatches forms for a “balanced scorecard,” number of practices in place that across boundaries—geographic capturing information on intent support organizational learning. areas, functions, time, or organiza- (in-brief), actions (daily), and out- For instance, unit or incident logs tional scales—because, by defini- comes (closeout) from members of often capture key decisions, we’ve tion, a boundary marks the edge the host agency (agency adminis- begun to build After Action Reviews of our local expertise or attention. trator, fire staff, resource advisor, or (AARs) into our business model, Assessing practical drift requires public affairs officer) and Incident and there are a number of ways in cultivating a corporate (broader) Management Team (IC, operations which key players communicate
Fire Management Today 6 staff, or public information person- Despite the numerous variations Throughout the nel). of the form, all KDLs included six basic questions: process, critical In May 2008, the Forest Service’s • “Who was the decisionmaker?” feedback and input have Washington Office asked the • “What was the decision?” come from all levels research team if we could also • “What alternatives/risks were of the fire community: capture key decisions and decision considered?” the NIMO teams, rationales. After initial testing on • “What was the rationale for the the Indians Fire in California, the decision?” the Forest Service resulting “significant cost KDL” • “What were the cost implica- National Leadership focused on decisions with signifi- tions?” and Team, users of the cant financial implications using • “With whom was the decision system, and members a simple, table-based form to be shared or discussed?” completed as key decisions were of the interagency High made. The concept was quickly Depending upon the version used, Reliability Organizing adopted—and locally adapted—and the definitions of a “key decision” community of practice. numerous, slightly different ver- included: sions began circulating. To address • A decision that has a significant this, we created a Web page at the impact on cost, sociopolitical as type 1, 2, 3, 4, and 5; prescribed Lessons Learned Centers’ Incident conditions, or resource alloca- fire events; incidents under single Management Team (IMT) site tion; and home unit jurisdiction; and (
Volume 69 • No. 2 7 database. The Web version popu- • Accountability—tracking or Discussion lated our central Oracle database assigning responsibility appropri- An often-heard adage concerning directly. KDLs submitted via email ately; and decisionmaking is that decisions were manually copied into the Web • Task requirements—identifying are only as good as the perception database. It was then possible to and addressing the current situ- of reality behind them. We know use qualitative analysis procedures ation (for example, demobilizing from organizational theory that to categorize responses. crews when those resources are perception is profoundly influenced no longer needed or ramping by organizational culture as refined Narrative analysis of submissions up in response to changing fire by individual expertise and experi- indicates that the most frequent behavior). ence. Continuous improvement in types of decisions entered can decisionmaking considers both the be classed into an intuitive and Less frequently cited rationales cultural frame of reference that relatively small set: choice of tac- include (in no particular order): focuses decisionmakers’ attention tic, choice of resource type, size • Operations—actions taken to on specific aspects of a situation (team size or number of a given influence incident duration or more than others and the way in resource, ramp up, or ramp down), size (and thereby, costs) to pro- which this information is inter- and choice of strategy. Decisions tect values at risk, to achieve preted, weighed, and integrated made specifically to curb costs were natural resource benefits, or to to arrive at a particular decision. less frequent, and most of these influence tactical impacts on KDLs capture a slice of this percep- revealed consideration of small— natural resources; tion, offering the potential for both but recurring—issues, such as • Planning—anticipating responses individual teams and the organiza- decisions concerning camp cater- to potential fire development or tion as a whole to reflect upon the ers. Entries describe decisions and obtaining information in order to conceptual models in use and how actions that increase as well as make the next decision; these influence what factors are decrease costs. Few entrants were • Policy—following mandated pro- considered and, of those, what fac- able to quantify short- or long-term cedures as they define (and some- tors are incorporated into the deci- cost implications, though most times limit) strategies; sionmaking process. were able to predict whether these • Relationships and communi- would be positive or negative. cation—the quality of working When KDLs are uploaded to the relationships (poor or good) or central server, they provide the data Most frequently cited rationales for communications as they influ- necessary to cultivate an objective decisions and actions reveal a focus ence coordination among staffs organizational perspective, facili- on (in no particular order): and personnel; tate organizational learning, and • Efficiency—the decision is most • Availability and training— improve corporate effectiveness. efficient among alternatives; “making do” with what’s available They provide a window into the • “Right resource at right cost”— and creating training opportuni- broader organizational culture and matching the work with its cost ties; and structures that create the opera- (for example, considering “exclu- • Complexity—addressing multiple tional “decisionspace” and atten- sive use” instead of “call-when- goals in tactics and number or dant constraints within which the needed” or use of Federal crews type of resources selected. individual units and teams must instead of contract crews); operate. Results from 2008 provide • Safety—limiting exposure, ensur- Many teams recorded decisions a window into how fire managers ing medical support, and reduc- made by staff members beyond ICs, see their decisionspace and the ing travel time; line officers, and agency adminis- range of rationales upon which • Probability of success—selecting trators, including decisions made they based decisions. options with the greatest proba- by command staff and area com- bilities of success and minimizing mand personnel. It was also stated In terms of the “balanced score- an insistence on holding a fire that any decisions that affect cost card” concept, the KDL entries line when there is a probability should be noted, including those reveal a significant focus on opera- that the line would be breached; made by dispatch personnel and the tional efficiency. The expansion of regional office. entrants from the originally target-
Fire Management Today 8 ed ICs and agency administrators to any set of circumstances and that 4 and 5), as well as on larger fires all command and general staff areas all of these serve to improve under- (types 1, 2, and 3). and from an expected suite of line- standing and promote learning. The • Including the KDL in a Letter based decisionmakers to general NIMO exercises, for example, use of Delegation encouraged cre- staff on both team and host units the KDLs as a jumping-off point ation of a complete incident log. indicates a desire to record and for discussions about effective risk Key decisions were most often communicate details of incident management for future fire sea- discussed and identified during management. sons. evening planning meetings with command and general staffs, with Safety and probability of success KDL 2009— one person given responsibility were two of the highest profile Next Steps for data entry. Still, KDL efforts aspects of risk management record- initiated during 2008 were some- ed. Much less is revealed about The 2008 effort provided significant times preempted by higher priori- the role of working relationships information for identifying value ties. (particularly external to the host and critical processes, as well as • The current KDL system should IMT), experimentation, and learn- identifying weaknesses in the pilot be streamlined, avoid duplicat- ing opportunities. Whether these effort. Feedback received from field ing other programs, and allow are systemwide patterns or particu- users noted a variety of benefits in entrants editing capability. lar to the few teams and units who the process: Future development must pro- participated in our pilot project • Transparent documentation of vide access to the interagency cannot be answered until we get a decisions provides value, offering community and include a user’s more extensive dataset. a way to review decisions made guide, more training, and out- in realtime, identify trends, and reach. KDL should also be linked By creating one form in which key make planning corrections. to WFDSS and document imple- incident participants can note their • The process was useful for creat- mentation decisions. decisions and subsequent actions, ing a final fire narrative, tracking • There is followup needed on the the KDL can help facilitate learning large and small decisions, com- 2008 KDL decisions. The agency and improve operational effective- municating between the agency needs an analysis of project prog- ness at the local level. Ideally, each administrator and IMT, and ress, including an assessment of KDL reveals the line of reasoning providing upper echelons with a working relationships, innovation and choices seen and taken by more explicit record of the com- and learning, financial manage- the key players on an incident to mitment to cost management. ment, and other intended out- achieve the desired outcomes cap- • The process was valuable as a tool comes of the KDL process. tured in Delegations of Authority to capture IMT staff decisions, and strategic direction documents during the season to provide Based on this feedback, analysis of (Wildland Fire Situation Analysis insight into consequences of entries, and lessons learned from [WFSA] or Wildland Fire Decision alternatives and during postsea- 2008, a steering committee of rep- Support System [WFDSS]). As son reviews. resentatives from research, National such, they can be used to link and Forest System management (fire improve alignment of incident Feedback also provides advice for staff, IMTs, and line officers), and objectives and intent to actual out- further developing the process: interagency partners has developed comes. • Decisions made at national, a revised version of the KDL. This regional, national forest, and IMT second version combines the “bal- Several teams and at least one levels directly impact resource anced scorecard” and KDL efforts host unit report using their KDLs availability, strategy, tactics, into a single, streamlined form that as input for AARs, and the NIMO duration, and costs of incidents. can be completed as needed during teams used KDLs to build case To meet the objective of improv- an incident. This effort is a stand- studies of large fires for incorpora- ing organizational “Situational alone program built to capture tion into 2009 training. Such exer- Awareness,” KDL must capture decisions flowing from strategic cises are particularly useful when the full spectrum of perspectives. decisions made and documented the approach recognizes that there • KDL was useful for recording in an incident’s guiding document are multiple rational responses to decisions on smaller fires (types (WFSA, Wildfire Implementation
Volume 69 • No. 2 9 Plan, or WFDSS). The 2009 ver- aspects of the incident (safety, References sion provides the opportunity for ecology, cost). We expect this Garvin, D. 2000. Learning in action. more rigorous analysis through to affect our incident only, or Cambridge, MA: Harvard Business School use of structured responses in addi- other incidents as well. Press. Kaplan, R.S.; Norton, D.P. 1992. The bal- tion to narrative. Additionally, the anced scorecard: measures that drive combination of WFSA/WFDSS and The KDL can be accessed by performance. Harvard Business Review. KDL will capture the decision flow authorized users through the Fire Jan.–Feb.: 71–80. from land management direction and Aviation Management Web Reason, J. 1997. Managing the risks of organizational accidents. Burlington, VT: to fire management plan, to inci- Applications (FAMWEB) Web site, Ashgate Publishing. 252 p. dent objectives and strategy, and which provides secure access by Snook, Scott A. 2000. Friendly fire: the through implementation. As such, interagency partners. accidental shootdown of U.S. Black Hawks over northern Iraq. Princeton, NJ: it builds a story: Princeton University Press. 257 p. I/we have made a decision, Acknowledgements Weick, K.; Sutcliffe, K. 2007. Managing the taken an action, or raised unexpected: resilient performance in an This effort has been funded by an issue in order to move age of uncertainty. 2nd ed. Hoboken, NJ: Joint Fire Science Project, National John Wiley & Sons, Inc. 194 p. towards meeting an objec- Fire Plan, and Forest Service Fire tive. We need to do this now and Aviation Management. Our because of some pressing or information technology partner is emerging issues or concerns. Digital Visions Enterprise Unit. We expect this to impact key
Web Site on Fire A Hub for Fire Information
The Web site at
Links in the Web site connect the viewer to specialized areas for indepth information. These include familiar topics and some unexpected resources: InciWeb is an interactive list of all recorded fire incidents through- out the United States over the last fire season, and back issues of Fire Management Today can be viewed and printed, from the latest to the first issue of December 1936.
The site serves as a comprehensive source of information on fire incidence and response from historic records up to the present and into the emerging future.
Fire Management Today 10 a multi-disCiplinary approaCh to Fire management strategy, suppression Costs, Community interaCtion, and organizational perFormanCe Anne E. Black, Krista Gebert, Sarah McCaffrey, Toddi Steelman, and Janie Canton-Thompson
ildland fire management is the new emphasis? Is it affect- Furthermore, highly contextual must balance the multiple ing the bottom-line, and if so, for observations, such as perceptions Wobjectives of protecting life, whom? How successful are we of community relations, quickly property, and resources; reduc- regarding land management objec- deteriorate with time. ing hazardous fuels; and restoring tives or safety? How well are we ecosystems. These Federal policy communicating intent with our key In this article, we describe our imperatives, varied yet connected, partners and the public, and what current work to assess the utility must be met under an increas- message is being received? of available data to reflect on the ingly constrained budget. A key to performance of fire management. management success is effectively To date, the available data—which exercising the full range of manage- Research related to include objectives and strategies ment flexibility in responding to captured by decision documents, wildland fire. community and public daily plans, final narratives, and understanding of fire GIS imagery—have not been sys- Over the past several fire seasons, management during tematically captured or analyzed. there has been increasing empha- a fire event is limited, sis on strategies to achieve fire particularly as that Using a “Balanced management objectives using less Scorecard” than full perimeter control, such understanding relates to as more prescribed burning and the use of alternatives In summer 2008, we embarked on a Joint Fire Science Program (JFSP) focused point and area protection. to full suppression. While the strategies and tactics project to develop and field-test a themselves are not new, wider use “balanced scorecard” of organiza- by Federal agencies, particularly tional performance suggestive of on multi-jurisdiction events and in Answering these questions in a gen- outcome, impact, and trend. The areas adjacent to private lands, has eralizable way has its challenges. “balanced scorecard” framework raised concerns among partners For example, the current financial outlines four critical aspects of and stakeholders. How effective system can inhibit accurate capture management necessary for develop- of daily costs. Incident managers ing a robust picture of an organi- Anne Black is an interdisciplinary ecolo- have different interpretations of zations’ performance: customers, gist/social scientist at the Forest Service’s financial health, internal business Aldo Leopold Wilderness Research Institute, wildland fire response strategies Missoula, MT. Krista Gebert is an econo- and different meanings for “cost perspective, and innovation and mist at the Forest Service Rocky Mountain effectiveness.” The fire community learning (Kaplan and Norton 1992). Research Station in Missoula, MT; Sarah lacks agreed-upon metrics to mea- For use with fire incidents, we McCaffrey is a social scientist at the Forest translated those aspects to commu- Service Northern Research Station in sure performance (such as degree Evanston, IL. Toddi Steelman is an associ- of success in meeting objectives: nity relations and public sentiment; ate professor in the Department of Forestry resource benefits, protection, effi- fiscal efficiency; safety, ecological and Environmental Resources at North health, and tactical effectiveness; Carolina State University in Raleigh, NC. ciency, and internal and external Janie Canton-Thompson is a social scien- relations, for example), which and organizational learning. tist at the Forest Service Rocky Mountain inhibits comparison across cases. Research Station in Missoula, MT.
Volume 69 • No. 2 11 We sought to use this “balanced tactics. Therefore, it is critical to uses quantitative and qualitative scorecard” to investigate how understand how agency/commu- methods to address the following tradeoffs are made in the decision- nity interactions shape both public questions: making process, including how acceptance and managers’ percep- • Do outreach and interaction community interaction increases or tions of what is acceptable. Much with the public prior to the fire, decreases the opportunity to exer- of the recent research related to including involvement in land cise different responses to wildland community response to wildfire management and fire manage- fire and how management flexibility has focused on prefire mitiga- ment planning and hazard may or may not contribute to cost tion actions on public and private mitigation, increase acceptance and organizational performance. In land. This research indicates that of alternative strategies (i.e., less close partnership with the Northern increased understanding of the pur- than full suppression) during a Rockies Coordination Group pose and effectiveness of mitigation fire and decrease post-fire con- (NRCG) and the National Incident measures is associated with greater flict? Management Organization (NIMO) acceptance (McCaffrey 2006). • How does provision of informa- teams, we are collecting informa- tion during a fire, both in terms tion running the gamut from tangi- Partnerships and collaboration of content and presentation, ble and measurable (safety, ecology, with communities and agencies influence acceptance of alterna- and costs) to less tangible but still are also tied to effective mitigation tive strategies? Are there particu- perceptible (suppression effective- measures (Steelman and Kunkel lar types of information that are ness, community interactions, and 2004; Steelman and others 2004). more or less important in shap- efficiency) and fully process-based Additional research indicates that ing acceptance? (organizational learning) through fire mitigation efforts are more • How does public reaction to fire interviews, cost analyses, and the effective when wildland managers management efforts during an incident key decision log (KDL, see not only understand and address event affect fire management The Key Decision Log: Facilitating the factors that may influence decisions? High Reliability and Organizational stakeholders’ acceptance of man- Learning, in this issue). agement practices but also work In 2008, we collected data on com- to engage them in risk manage- munity outreach, interactions, and The benefits derived from this ment decisions (Winter and oth- responses from three fires: the Gap research will include: (1) a clearer ers 2002; Zakzek and Arvai 2004). Fire on the Los Padres National description for fire managers of Unfortunately, research related to Forest in California (full suppres- the relationships among costs (to community and public understand- sion with perimeter control), the Federal, State, and local entities), ing of fire management during a Cascade Fire on the Custer National community interaction, safety, ecol- fire event is limited, particularly as Forest in Montana (modified sup- ogy, organizational performance, that understanding relates to the pression), and the Gunbarrel Fire responses to wildland fire, and fire use of alternatives to full suppres- on the Shoshone National Forest management strategies; and (2) a sion. Consequently, we do not know in Wyoming (prescribed burn- protocol that allows monitoring of much about how agencies reach ing). We interviewed participants and learning from organizational out to the public, what communi- from incident management teams, performance trends, both process ties or the public understand about host agencies, and key commu- (agency-team relations) and out- strategies and tactics, or how the nity members. We plan to test the come (the impact of a given fire on public can facilitate or obstruct the importance of some of the dynam- human safety, values at risk, com- use of different options for respond- ics identified in these discussions munity relations, and ecosystems). ing to wildland fire. on a wider audience as the research continues. Community Relations One specific objective under the and Public Sentiment JFSP project is to gain an under- Fiscal Efficiency standing of the relationships Many policy and decisionmakers Fire managers must take into between pre-fire and during-fire hoped that the use of less-aggres- account both public expectations community interaction and fire sive suppression strategies, where and the degree of public accep- management flexibility. To meet appropriate, in 2007 would result tance for different strategies and this objective, our research design
Fire Management Today 12 in lower costs (OIG 2006; USDA Preliminary economic work sug- sionmakers frame their decision and USDI 2007; NRCG 2007). gests that the data and regression space and how they weigh poten- However, the interplay of wild- models used to derive the “stratified tially conflicting objectives within land fire management decisions cost index” (a current performance that space. Indepth field interviews and cost containment is not well measure for both the Forest Service in the fall of 2008 helped us to understood, and research designed and U.S. Department of the Interior better understand the full context to assess the factors affecting sup- Bureau of Land Management, and impact of other factors on the pression expenditures has been Gebert and others 2007) may also relationship between response and limited due to a lack of data. Early be useful for assessing the effect cost, due to the fact that contex- research by Gonzalez-Caban (1984) of responses to wildland fire on tual information is not captured in estimated suppression expenditures suppression expenditures per acre. financial records. These interviews based on the number and type of Early analysis of interviews shows focused on understanding: the different resources used on that State and local cooperators, • How does an increasing emphasis the fire. More recently, Donovan in most cases, are quite concerned on cost containment influence and others (2004) used regression about their costs increasing as the strategies and tactics used on analysis to identify variables affect- Federal policy shifts toward more wildland fires? ing suppression expenditures for prescribed burning, monitoring, • How does the use of the new 58 fires that occurred in Oregon and point protection strategies. Wildland Fire Decision Support and Washington in 2002. The The general assumption is that System (WFDSS) influence the Forestry Sciences Lab in Missoula less-aggressive strategies will result strategies/tactics or costs of sup- has conducted research on large in cost savings for the agency. pressing fires? fire suppression costs since fis- However, some interviewees argue • How do decisionmakers weigh cal year 1998 (Gebert and others that less-aggressive strategies cause potentially competing fire man- 2007; Canton-Thompson and oth- ers 2006; Canton-Thompson and others 2008). Still, much remains An assumption is that less-aggressive strategies to be learned about fiscal efficiency will result in cost savings. However, some argue related to fire management. that less-aggressive strategies cause fires to last We hope to extend this work by longer, so the savings may be minimal at best. focusing quantitative and qualita- tive assessments on how wildland fire management strategies and fires to last longer and increase agement objectives and risks tactics influence wildland fire costs chance of escape to private land, (long- versus short-term risks, and vice versa. Quantitative tech- culminating in higher private costs safety, cost, probability of suc- niques include economic assess- and prolonging costs to local stake- cess, and public opinion), and are ments of previous fires. Qualitative holders, particularly with respect to there patterns to these weights assessments include interviews personal health and tourism losses. across strategies? with agency officials, cooperators, We are hoping that our analyses and stakeholders about their experi- (both qualitative and quantitative) We are also building and testing ences with the interplay of response will shed light on this issue or at a Web-based system in which fire strategies and costs. Our two major least that we will start to collect the managers (both line officers and questions are: type of information necessary to team members) can document • Does point protection/monitor- answer this question over the next their key decisions and decision ing, rather than full perimeter few years. rationales. This effort recognizes control, affect suppression costs that while Federal fire management for Federal agencies? Risk Management and policy (1995) outlines four major • Do strategies and tactics aimed at Organizational Learning objectives—safety, property and less than full perimeter control resource protection, hazard reduc- reduce the costs of fire manage- To fully understand the mechan- tion, and ecological restoration—as ment or simply shift the cost bur- ics of the current decisionmaking yet, there are no consistent assess- den to non-Federal entities? process, we need to know how deci- ment criteria or data collection
Volume 69 • No. 2 13 protocols through which to mea- urban interface—are resulting in Garvin, D. 2000. Learning in action. sure progress. Not only is a system longer fire seasons and wildfires Boston, MA. Harvard Business School Press. 254 p. important for program reporting; it that are increasingly resistant to Gebert, K.M.; Calkin, D.E.; Yoder, J. 2007. is critical for organizational learn- control and expensive to suppress. Estimating suppression expenditures for ing (Garvin 2000). Thus, we also Under these conditions, the tac- individual large wildland fires. Western Journal of Applied Forestry. 22(3): want to evaluate: tics relied upon in the past now 188–196. • How consolidated and compre- often pose unacceptable risks to Gonzalez-Caban, A. 1984. Costs of fire- hensive a story of fire manage- firefighter safety. This has led to fighting mopup activities. Research Note PSW-RN-367. Berkley, CA: Forest ment can we tell by capturing the call for changes in the way we Service, Pacific Southwest Forest and currently available incident data manage wildfires. Flexible manage- Range Experiment Station. 5 p. in a central location and through ment means less emphasis on put- Kaplan, R.S.; Norton, D.P. 1992. The bal- a log of key decisions during an ting fires out, no matter what the anced scorecard: measures that drive performance. Harvard Business Review. incident? cost, and more emphasis on using Jan.–Feb.: 71-80. • How effective is maintaining the our scarce firefighting resources to McCaffrey, S.M. 2006. The public and wild- log at facilitating organizational effectively meet multiple objectives. land fire management: Social science findings for managers. Gen. Tech. Rep. learning? To help land managers and policy NRS-GTR-1. Newtown Square, PA: Forest makers succeed in this new arena, Service, Northern Research Station. 202 In consultation with NRCG and information is needed on the inter- p. NRCG (Northern Rockies Coordination NIMO, we developed and launched play of fire management strategy, Group). 2007. Appropriate management a Web-based database for use by suppression costs, Forest Service/ response summary for the Northern incident and management units. community interaction, and organi- Rockies. Missoula, MT: Northern Rockies Originally envisioned for use only zational performance. We hope this Coordinating Group. Available online at
Fire Management Today 14 eFForts to update FireFighter saFety zone guidelines Bret Butler
ne of the most critical deci- sions made on wildland fires When fires are located on slopes or ridges, Ois the identification of suitable convective energy transfer may reach distances safety zones for firefighters during daily fire management operations. equal to several flame lengths ahead of the fire To be effective (timely, repeatable, front. This implies that the current safety zone and accurate), these decisions rely guidelines may be inadequate in situations where on good training and judgment, but the safety zone and/or fire are on slopes. also on clear, concise guidelines. This article is a summary of safety zone guidelines and the current on flat terrain, convective energy that radiant energy transfer is the research efforts focused on improv- transfer is primarily upward in the dominant energy transfer mode. ing firefighter safety zones, and a plume and radiant energy transfer Experimental measurements verify request for input from firefighters is directed in all directions, includ- the accuracy of this assumption, and fire managers. ing out ahead of the fire front. but also indicate that when fires Therefore, firefighter safety guide- are burning on slopes, convective How Far Should We Be lines are based on the assumption energy transfer can be significant. From the Flames? Guidelines for the minimum dis- tance a firefighter should be from a Distance from flame (ft) flame are discussed in training cur- 650 ricula and published in the Incident 600 Response Pocket Guide (IRPG) and Fireline Handbook (fig. 1). The 550 No injury current safety zone guidelines are 500 Burn injury probable based on the distance at which 450 0.5 exposed human skin will develop a 400 second-degree burn in less than 90 Burn injury limit 350 1 Btu/ft2/s seconds (the distance correspond- 300 ing to a radiant incident energy flux 2 level of 7.0 kW-m-2). An approxima- 250 tion of this model used in the field 200 indicates that a firefighter should 150 keep a minimum distance of four 100 times the flame height. For a cir- 50 5 cular safety zone, this would be the 0 safety zone radius. 0 50 100 150 200 250 300 350 Flame height (ft) Mann South Current safety zone guidelines were Gulch Canyon and Fire Butte Fire developed based on fires located Battlement on flat terrain. When fires burn Creek Fires
Rule-of-thumb=4 x maximum flame height Bret Butler is a research mechanical engi- neer for the fire, fuel, and smoke science program at the Forest Service Missoula Figure 1—Results from original safety zone analysis. Model results were verified by Fire Sciences Laboratory in Missoula, MT. comparison against past fires.
Volume 69 • No. 2 15 Intuition, professional observations, and the few experimental measure- ments that have been reported indi- cate that, when fires are located on slopes or ridges, convective energy transfer may reach distances equal to several flame lengths ahead of the fire front. This implies that the current safety zone guidelines may be inadequate in situations where the safety zone and/or fire are on slopes. It is also clear from visits to designated safety zones on numerous wildland fire incidents that considerable ambiguity exists regarding identification or cre- ation of true “safety zones,” versus “deployment zones.” Fire researchers installing a video acquisition box. Photo by Bret Butler, Forest Service. New Research for Variable Terrain The Joint Fire Science Program is A new research project supported by supporting a new research proj- the Joint Fire Science Program is underway ect to measure convective energy to measure convective energy transport transport and use those measure- and use those measurements to develop ments to develop safety zone guidelines for slopes. We expect the safety zone guidelines for slopes. guidelines will depend on multiple variables, including fire charac- cm) by 5.9 inches (15 cm) by fire behavior but also provides teristics (flame length or height), 7.1 inches (18 cm) and pro- insight into our data analysis. site characteristics (e.g., slope), vides an insulated protective Camera(s) are housed within and relative location (i.e., chimney, enclosure for a datalogger, 3.9-inch (10 cm) by 7.1-inch ridge, midslope, ridgetop). We place sensors, and other electronics. (18 cm) by 7.5-inch (19 cm) special emphasis on the effect of The standard instruments con- fireproof enclosures. The boxes “chimneys” and other terrain fea- sist of radiometers that mea- have a double-lens configura- tures that produce dangerous levels sure total and radiant energy tion: one lens of high tempera- of convective heating ahead of the fluxes; small-gauge thermo- ture Pyrex glass and a second fire front. The approach for this couples—36 gauge (0.13 mm lens of hot mirror coated glass study has several phases: diameter) wires—that sense that reflects infrared radiation 1. Measure convective and radiant flame and air temperature; and (heat) while allowing visible energy transport in fires across pitot-static type velocity probes light to pass through. The cam- a range of slopes and fuel types. that sense the magnitude and eras can be turned on manually An experienced team has start- direction of airflow before, dur- or set to trigger and record ed gathering measurements ing, and after the fire passes. through a wireless link to the at various sites over a range of FBP sensor packs. slopes and fire intensities using The Video Acquisition Box pro- 2. Use measurements to “tune” two types of instrument pack- vides digital video imagery, an and validate theoretical heat ages: the Fire Behavior Package integral component of our field transfer models. Tools include and the Video Acquisition Box. campaigns. Collecting video software that simulates heat-
imagery not only allows us to ing rates due to conduction; The Fire Behavior Package observe actual footage of the radiation; convection (steady- (FBP) measures 11 inches (27
Fire Management Today 16 attributes of an effective safety zone may depend on more than one variable. Some potential tools for presenting and using Steep Slope multidimensional information are pocket cards, nomograms (graphic representations), and computer models. Figure 2 Medium Slope presents one possible method for presenting the results— here, slope steepness and fire intensity dictate the minimum Safety Zone Size safe separation distance from Low Slope the flames. Interviews and consultations with IMTs and fire crews during the field deployment phase of the study will guide the selection and development of one or more Flame Height products for delivering the new safety zone guidelines. Figure 2—A possible format for Safety Zone guidelines on slopes. How Firefighters state or transient); and the flow radiant energy distributions Can Help field that develops ahead of a around the fires are being com- This effort depends heavily on par- fire over the expected range of pared to human burn injury ticipation from the wildland fire- flame sizes, fire front location limits to determine appropriate fighting community. The research relative to slope, slope steep- separation distances and associ- team requires assistance during all ness, and slope shape (i.e., ated safety zone sizes. phases, and particularly in identify- concavity). Fire growth is being 3. Consult with incident manage- ing potential measurement sites explored using existing spa- ment teams (IMT), fire crews, and developing the best format tial fire models like FARSITE and others from the wildland for presenting new guidelines. If and the new Wildland Fire fire community to determine you have suggestions for potential Dynamics Simulator developed the best methods for deliver- opportunities to measure convec- at the National Institute for ing new slope/safety zone tive flow, either on prescribed Standards and Technology. The guidelines. We expect that the burns or ongoing wildland fires, or measurements collected in the results will indicate that slope how best to convey the new guide- field efforts are being used to steepness and concavity may lines (pocket card, nomograms, “tune” and validate theoretical be nearly as critical to safety and/or computer models, etc.), heat transfer models. Measured zone characteristics as fire please contact the author. Your and modeled convective and intensity. Consequently, the input will be greatly appreciated.
Volume 69 • No. 2 17 modeling Fuel suCCession Brett Davis, Jan van Wagtendonk, Jen Beck, and Kent van Wagtendonk Surface fuels data are of critical importance for supporting fire inci- dent management, risk assessment, and fuel management planning, but the development of surface fuels data can be expensive and time con- suming. The data development pro- cess is extensive, generally begin- ning with acquisition of remotely sensed spatial data such as aerial photography or satellite imagery (Keane and others 2001). The spa- tial vegetation data are then cross- walked to a set of fire behavior fuel models that describe the available fuels (the burnable portions of the vegetation) (Anderson 1982, Scott and Burgan 2005). Finally, spatial fuels data are used as input to tools such as FARSITE and FlamMap to Figure 1—Successional pathway diagram for the Timber Litter 6 (TL6) fire behavior fuel model current and potential fire model. spread and behavior (Finney 1998, Finney 2006). fire simulation based on these data Fuel Model become less accurate as the data Development The capture date of the remotely age. Because of the amount of labor sensed data defines the period for and expense required to develop To characterize fuels and fuel suc- which the vegetation, and, there- these data, keeping them updated cession, we started with the fire fore, fuels, data are most accurate. may prove to be a challenge. behavior fuel models described The more time that passes after in “Standard Fire Behavior Fuel the capture date, the less accurate In this article, we describe the Models: A Comprehensive Set for the data become due to vegetation Sierra Nevada Fuel Succession Use with Rothermel’s Surface Fire growth and processes such as fire. Model, a modeling tool that can Spread Model” (Scott and Burgan Subsequently, the results of any quickly and easily update surface 2005). These models provide a fuel models with a minimum of more detailed representation of Brett Davis is a geographic informa- additional input data. Although it fuels than the standard 13 Northern tion specialist and fire modeler for the was developed for use by Yosemite, Forest Fire Laboratory (NFFL) fuel Forest Service Aldo Leopold Wilderness models (Anderson 1982) and, there- Research Institute, Rocky Mountain Sequoia, and Kings Canyon Research Station, in Missoula, MT. Jan van National Parks, it is applicable to fore, gave us the flexibility to depict Wagendonk is a research forester emeritus much of the central and southern different successional stages. We for the U.S. Geological Survey, Western then developed crosswalks between Ecological Research Center, Yosemite Field Sierra Nevada. Furthermore, the Station, El Portal, CA. Jen Beck is a fire methods used to develop the model current vegetation data and fire ecologist for the National Park Service at have national applicability. behavior fuel models in a series of Redwood National Park, Orick, CA. Kent collaborative meetings with fuels van Wagtendonk is a geographer for the National Park Service at Yosemite National experts from the parks and the U.S. Park, El Portal, CA. Geological Survey (USGS). Among
Fire Management Today 18 Table 1—Fuel models represented in Yosemite, Sequoia, and Kings Canyon National 2007) data to obtain estimates of Parks. fire severity on most of their larger Fuel Model Description fires for fires that have burned in the past 30 years. We used these Grass 1 (GR1) Short, sparse, dry climate grass estimates of fire severity, usually Grass 2 (GR2) Low load, dry climate grass classified into low, moderate, high, Grass 4 (GR4) Moderate load, dry climate grass and unburned categories, to model post-fire transitions among fuel Grass-Shrub 1 (GS1) Low load, dry climate grass-shrub models. Grass-Shrub 2 (GS2) Moderate load, dry climate grass-shrub Shrub 1 (SH1) Low load, dry climate shrub Fuel Model Transitions Shrub 2 (SH2) Moderate load, dry climate shrub For each of the 18 unique surface fuel models developed for park Shrub 5 (SH5) High load, dry climate shrub vegetation types, we created a state- Shrub 7 (SH7) Very high load, dry climate shrub and-transition model, represented Timber Litter 1 (TL1) Low load, compact conifer litter by a successional pathway diagram (fig. 1). A “state” is a standard fire Timber Litter 2 (TL2) Low load, broadleaf litter behavior fuel model, and “transi- Timber Litter 3 (TL3) Moderate load, conifer litter tions” are the changes to the fuel model that occur as a result of Timber Litter 4 (TL4) Small downed logs either fuel accumulation or low, Timber Litter 6 (TL6) Moderate load, broadleaf litter moderate, or high severity fire. Timber Litter 7 (TL7) Large downed logs Transition times associated with each state represent the waiting Timber Litter 8 (TL8) Long needle litter period before the current state Timber Understory 1 Low load, dry climate timber-grass-shrub changes to the subsequent state (TU1) (fuel model) in the absence of fire. Timber Understory 5 Very high load, dry climate timber-shrub We based the state, transition, and (TU5) transition time selections on the distribution of the underlying vege- the three parks, we classified veg- Fire Severity: The tation for each fuel model assigned etation into 18 unique surface fuel Major Model Input in the crosswalk. models (table 1) and then we devel- oped successional pathways for Fire severity and time since last There are two general categories of each (fig. 1). fire are the key inputs that dictate fuel model transitions: (1) transi- which successional pathway the tions resulting from low, moder- Our end result is a deterministic model will follow. We defined fire ate, or high severity fire, and (2) model of fuel succession, based severity as the degree of post-fire transitions resulting from fuel on expert local knowledge, that change that would be seen from a accumulation. Transitions result- accounts for both fuel accumula- remotely sensed (aerial) perspec- ing from fire are to a temporarily tion and consumption. The model tive, a definition compatible with unburnable state—several years predicts how fuels—represented by Normalized Burn Ratio techniques may be required before enough fuel the fire behavior fuel models—can for assessing fire severity (Key accumulates to support another be expected to change over time. and Benson 2005, Thode 2005, fire and thereby transition back to Rules governing transitions from Miller and Thode 2007). Sequoia a burnable fuel model. Transitions one fuel model to another reflect and Kings Canyon National Parks based on fuel accumulation reflect our best judgment about how have used delta Normalized Burn the rate of post-fire recovery or quickly fuels accumulate and how Ratio (dNBR) (Thode 2005) and fuel accumulation that occurs in different vegetation types respond Yosemite National Park has used the absence of fire; these rates vary to fires of various severities. Relative delta Normalized Burn with the severity of the fire and Ratio (RdNBR) (Miller and Thode
Volume 69 • No. 2 19 the accumulation potential of the have very different fuel accumula- Conclusions post-fire vegetation. For example, tion rates and responses to fire, One must keep in mind that, like if the Timber Litter 6 (TL6) fuel and, therefore, separate succession all models, the Sierra Nevada Fuel model burns in a moderately severe diagrams were developed for each Succession Model is subject to the wildfire, it becomes unburnable type (fig. 3, fig. 4). quality of the underlying data and for 5 years. After 5 years, enough the accuracy of the predictions of fuel will have accumulated for it to Updating Fuels Data the fire and fuel modeling experts. transition back to the original TL6 The Sierra Nevada Fuel Succession It is also subject to all assump- fuel model (fig. 1). Model provides a simple, practical, tions of the models and classifica- and easy way to keep surface fuels tion systems used to generate its We developed 22 diagrams, one data current. It extends the useful inputs, specifically the vegetation for each of the 18 original fuel life of these expensive and labor- and severity classifications based models plus four “special cases” intensive data and should improve on remotely sensed imagery. In (see below). Most diagrams are self- the predictive accuracy of fire addition, fire is the only landscape- contained, reflecting successional modeling tools such as FARSITE scale process modeled—the roles pathways that are either circular and FlamMap. To keep fuels data of other processes such as insects, or dead-end in a “final” fuel model current, the model should be run diseases, blow-downs, or avalanches (fig. 2). But in a few cases, a fuel each year at the end of the fire sea- are not accounted for in this ver- model transition reflects an under- son, starting the first year after the sion of the model. lying vegetation type change that creation of the fuels data (gener- necessitates a transition to a new ally the vegetation data’s capture As the Sierra Nevada Fuel succession diagram. For example, date). Model outputs include a fuel Succession Model is put into use if fuel model TL6 burns in a high model grid representing the next at Yosemite, Sequoia, and Kings severity fire, it will eventually year’s fuels (prior to next year’s fire Canyon National Parks, the transi- become a Grass-Shrub 1 (GS1) fuel season) and inputs for the following tions and transition times in the model and continue to follow the year’s succession run. successional pathways will be vali- successional path illustrated in the
TL6 diagram (fig. 1). On the other hand, if it remains unburned for 50 years, it will transition to fuel model Timber Litter 8 (TL8) as a result of an underlying vegetation type change, and the succession model will then follow the succes- sional pathways in the TL8 dia- gram.
The four “special cases” represent situations where multiple vegeta- tion types were similar enough in their expected fire behavior to be crosswalked to a single fuel model but were markedly different in their accumulation rates or response to fire. In these special cases, we developed multiple distinct succes- sional diagrams for the fuel model. For example, fuel model Timber Litter 2 (TL2) was crosswalked from evergreen oaks (Quercus spp.) as well as a variety of deciduous tree Figure 2—Fully self-contained successional pathway diagram for the Timber Understory 1 species. These two vegetation types (TU1) fire behavior fuel model.
Fire Management Today 20 dated and adjusted in response to field observations. Currently, the model is only applicable to the cen- tral and southern Sierra Nevada, but the techniques should be gen- erally applicable to any fire-prone landscape.
References Anderson, H.E. 1982. Aids to determining fuel models for estimating fire behavior. Gen. Tech. Rep. INT-GTR-122. Ogden, UT: Forest Service, Intermountain Forest and Range Experiment Station. 22 p. ESRI (Environmental Systems Research Institute). 2005. ArcGIS. Version 9.1. Redlands, CA: ESRI. Finney, M.A. 1998. FARSITE: Fire Area Simulator-model development and evalu- ation. Res. Pap. RMRS-RP-4. Ogden, UT: Forest Service Rocky Mountain Research Station. 47 p. Finney, M.A. 2006. An overview of FlamMap modeling capabilities. In: Andrews, P.L.; Butler, B.W., comps. 2006. Fuels Management—How to Measure Success: Figure 3—Successional pathway diagram for the Timber Litter 2.1 (TL2.1 - evergreen Conference Proceedings. 28–30 March oaks) fire behavior fuel model. 2006; Portland, OR. Proceedings RMRS-P-41. Fort Collins, CO: Forest Service, Rocky Mountain Research Station. 809 p. Keane, R.E.; Burgan, R.; van Wagtendonk, J. 2001. Mapping wildland fuels for fire management across multiple scales: Integrating remote sensing, GIS, and biophysical modeling. International Journal of Wildland Fire. 10: 301–319. Key, C.H.; Benson, N.C. 2005. Landscape assessment: Remote sensing of severity, the Normalized Burn Ratio. In: Lutes, D.C.; Duncan C.; Keane, R.E.; Caratti, J.F.; Key, C.H.; Benson, N.C.; Sutherland, S.; Gangi, L. J., eds. FIREMON: Fire effects monitoring and inventory system. Gen. Tech. Rep. RMRS-GTR-164-CD. Fort Collins, CO: Forest Service, Rocky Mountain Research Station. 1 CD. Miller, J.D.; Thode. A.E. 2007. Quantifying burn severity in a heterogeneous land- scape with a relative version of the delta Normalized Burn Ratio (dNBR). Remote Sensing of Environment. 109: 66–80. Scott, J.H.; Burgan, R.E. 2005. Standard fire behavior fuel models: a compre- hensive set for use with Rothermel’s surface fire spread model. Gen. Tech. Rep. RMRS-GTR-153. Fort Collins, CO: Forest Service, Rocky Mountain Research Station. 72 p. Thode, A.E. 2005. Quantifying the fire regime attributes of severity and spatial Figure 4—Successional pathway diagram for the Timber Litter 2.2 (TL2.2 - deciduous) complexity using field and imagery data. fire behavior fuel model. Dissertation. Davis, CA: University of California-Davis.
Volume 69 • No. 2 21 Fuel age and Fire spread: natural Conditions Versus opportunities For Fire suppression Richard W. Halsey, Jon E. Keeley, and Kit Wilson
ildfires are driven and younger aged vegetation can play in increasing in size. This has allowed restrained by an interplay assisting fire suppression efforts are us a unique opportunity to com- Wof variables that can lead frequently intertwined when they bine quantitative analysis of fuels to many potential outcomes. As are actually addressing two differ- with observations of actual wildfire every wildland firefighter learns in ent issues: the role that younger behavior and the results of actions basic training, the ability of a fire aged vegetation may play in stop- taken by firefighters. to spread is determined by three ping fire spread naturally versus its basic variables: fuel type and con- role in assisting firefighting efforts. Although firefighters can provide dition, weather, and topography. The fact that fires in younger, light- extremely valuable information Fire suppression obviously plays a er fuels are typically easier for fire- about wildland fire, this informa- significant role in determining fire fighters to extinguish than those in tion is generally underutilized by spread as well, so firefighter activity heavier, older fuels does not neces- the scientific community. To take becomes an additional variable. sarily correlate with what would advantage of their observations, we occur under natural conditions conducted extensive interviews with In southern California, where wild- without active fire suppression. those who were on scene when the fires occur predominately in shru- Cedar Fire burned into the study bland ecosystems, such as chapar- The Study area. These firefighters provided ral, there is continual debate over us with the tactical details of their To examine these issues and better the relative roles of weather and suppression efforts, as well as the understand the interplay of vari- fuels in fire spread. During Santa weather conditions they encoun- ables that determine fire spread, we Ana wind conditions, often char- tered. While Remote Automated examined four sites in San Diego acterized by single digit humidity, Weather Stations (RAWS) can pro- County, California, that burned in temperatures over 90 ºF (23 ºC), vide a general view of prevailing the January 2001 Viejas Fire and and 80 mile-per-hour (130 km/h) weather conditions, they are fre- were adjacent to (but unburned by) sustained wind speeds, weather quently too far away from the site the October 2003 Cedar Fire. typically overwhelms the influence of interest to offer the data needed; of fuel type. During milder condi- this was especially true during the The four study sites are particu- tions, fuel type becomes much Santa Ana wind condition that larly interesting for two reasons. more important. existed during the Cedar Fire, as First, they provide a test case for wind speeds varied dramatically how young fuels respond during a Questions about the chaparral’s within relatively short distances. wildland fire. After the Cedar Fire natural fire regime and what role jumped Interstate Highway 8, spot fires ignited the Viejas Fire scar’s The Location Richard Halsey is director of the California Chaparral Institute, Escondido, CA. Jon E. 3-year-old fuels and continued The sites are located in a portion Keeley is a research scientist for the U.S. burning for a considerable amount of the Cleveland National Forest, Geological Survey, Western Ecological of time. Secondly, the sites are in California, in what could be classi- Research Center, in Sequoia National Park, Three Rivers, CA. Kit Wilson is a geo- close proximity to the location fied as a fire corridor, a site where graphic information specialist consultant of an important fire suppression wildfires frequently start or burn with Open Spaces GIS, Escondido, CA. The action taken by Forest Service through due to local topographical authors would like to thank Bill Howell, firefighters that likely prevented and weather conditions. Four of Nicholas Halsey, and Jake Halsey for assis- tance in data collection. the Cedar Fire from dramatically San Diego’s largest fires have either
Fire Management Today 22 The Fuel Estimates of the young fuels in the 2001 Viejas Fire scar that were available but unburned during the October 2003 Cedar Fire were obtained during the summer of 2004. Although these samples were taken 9 months after the Cedar fire, they would have been representa- tive of the fuels burned in 2003 because there was a severe drought in 2004 and observations revealed relatively little growth by most spe- cies.
All plant material was collected from ten 1-meter squares equally spaced along the periphery of a 66- Figure 1—Looking down (southwest) from the Alpine Overlook into the Sweetwater River by 164-foot (20-by-50 m) plot. This canyon. Many of the largest fires in San Diego County have burned through this area. sampling protocol was repeated at The 2003 Cedar fire scar (“2003”) is the blackened ground in the middle and right of the four sites. The sites were on flat or photo. Two of the study sites where vegetation in the 2001 Viejas fire scar (“2001”) was sampled are located near the “X” mark. The Hotshot handline cut to the large boulder south-facing slopes within the 2001 outcropping (B) and down into the canyon starts at the lower middle of the photo. The Viejas Fire scar between Interstate stain from red fire retardant can also be seen near the boulders. Note how close the Cedar Highway 8 and the Sweetwater fire came to the dense chaparral stand that last burned in 1970. Photo by Richard W. River. Halsey. Prior to the 2001 fire, the plots burned through or bordered this acre (70,992 ha), wind-driven were covered by relatively sparse location. The focal point of the area blaze that raced 30 miles (48 km) chamise chaparral. The average is the Alpine Overlook, a highway within 24 hours to the outskirts 3-year-old fuel component con- overlook off Interstate 8, about 25 of El Cajon, CA. The nearby Viejas sisted of approximately .75 kg/m2 miles (40 km) east of San Diego. and Cedar Fire scars complete the coarse fuel (largely shrub skeletons About a quarter mile (.4 km) below patchwork (fig. 1). resulting from the 2001 Viejas Fire) the overlook lies the canyon bot- and a similar level of fine dead fuels tom of the Sweetwater River, only When the Cedar Fire made contact (<0.25 in [0.6 cm] in diameter) a trickling stream most of the year. with the area on October 27, 2003, arising from annuals, herbaceous At 3,000 feet (914 m), with an aver- the overall weather conditions were perennials, and short-lived peren- age annual rainfall of 17 inches (43 conducive to fire spread. The near- nials with annual dieback (fig. 2). cm), the area supports an interest- est RAWS station recorded relative Several of the meter squares had ing mix of vegetation, featuring humidity range of 9–12 percent more than a total of 2 kg/m2 of several differently aged patches of for October 27, between 0800 and fuel due to the presence of large, chaparral. The south-facing slope, 1630 hours. Temperature ranged resprouting chamise and sugar immediately below the overlook, from 69 to 78 ºF (67–83 ºC) with bush (Rhus ovata) and substantial is covered by a sparse stand of 16–24 mph (26–39 km/h) northeast skeletons of coarse fuels (fig. 3a). chamise (Adenostoma fascicula- wind speeds. Topography reduced Much of the cover at the site was tum) chaparral partially burned the winds in the canyon below the contributed by short-lived plants in a 42-acre (17 ha) blaze in 1982. overlook, although they were still that died back each summer, Across the river, on the north- blowing an estimated 10–15 mph resulting in a marked relationship facing slope, is a dense stand of old- (16–24 km/h) from the northeast between fine fuels and plant cover growth mixed chaparral last burned during the suppression action. (fig. 3b). in the 1970 Laguna Fire, a 175,425-
Volume 69 • No. 2 23 Table 1—Comparison of fuel loads in various plant communities common in southern California.
Plant Community Metric tons/hectare (acre) Chamise chaparral (study site) 10/24 (59) Mature chaparral1 Light 22/34 (84) Medium 34/67 (166) Figure 2—Fuel types. The study area’s Heavy 65/90 (222) 3-year-old fuel component was an equal Mature coastal sage scrub2 7/25 (62) mixture of burned sticks (coarse) and fine fuels, such as short-lived perennials. Annual grassland2 1/5 (12)
1Dodge 1975 2Green 1981 stands away from roads and fuel unable to attempt any fire suppres- breaks with fire return intervals sion efforts. Ten- to twenty-foot greater than 20 years are generally (3–6 m) flame lengths blasting up free of non-native weeds (Keeley from below the overlook, unburned 2006). Therefore, the fine-fuel com- fuel on both sides of the fire, ponent was dominated by native and steep, rocky terrain made an ground cover, especially deer- approach too dangerous. In addi- weed (Lotus scoparius), rush-rose tion, estimated 30 mph (48 km/h) (Helianthemum scoparium), and northeastern wind gusts were whip- resprouting chamise. Deerweed is a ping the fire into unpredictable prolific, post-fire species composed directions. of a multitude of thin stems that die back during seasonal drought. This was a critical juncture in the The amount of plant material avail- Cedar Fire’s southeastern expan- able as fuel in the study area is sion. The flames had jumped comparable to similar, unburned Interstate 8 and ignited the 3-year- plant communities in southern old fuels west of the overlook at California (table 1). approximately 0800 hours. By the time the Hotshots arrived, the The Fire fire had moved southeast through the young fuels, across slope, and After returning from fighting the against the wind into the older Figure 3—Fuels analysis. Some of the Old Fire in the San Bernardino samples (a) had significant amounts chamise chaparral stand directly National Forest and a quick brief- of course fuels due to resprouting below the overlook. The flames had shrubs. However, much of the cover was ing on the Cedar Fire, the crew also spread to the southwest, hav- contributed by short-lived plants, (b) of Engine 47 arrived at the Alpine resulting in a marked relationship between ing consumed a significant amount Overlook on Monday, October 27, fine fuels and plant cover. of the 3-year-old vegetation. The 2003, at about 1400 hours. “The winds in the canyon below the fire had already jumped the high- overlook were weaker than those way when we got there and was Although small amounts of non- near the top, moving an estimated burning directly below the over- native grasses, primarily red brome 10–15 mph (16–24 km/h) when the look parking lot,” Engineer Jerry or foxtail chess (Bromus madriten- crews began working the fire. sis), were present in some of the Amador recalled (J. Amador, pers. comm. 2005). meter-squares, they did not make a The potential danger was obvious to significant contribution to the fine- everyone on scene. The 33-year-old Before Engine 47 arrived, the Vista fuel mix. This was expected because fuels across the Sweetwater River Grande Hotshots had been on scene relatively undisturbed chaparral would have created extremely dan- for less than an hour but were
Fire Management Today 24 The Black Mountain Hotshots took over beyond a large rock out- cropping and continued cutting the handline all the way to the Sweetwater River, arresting the fire’s southeastward expansion. As the crew of Engine 47 had run out of hose at this point, the Hotshots worked without water support. The southwest flank, burning through the 3-year-old fuels, was extinguished by helicopter water drops. The fire was cold by approxi- mately 1630 hours. After jumping Interstate 8, the fire burned approx- imately 75 acres (30 ha), moving 0.5 miles (0.8 km)over a period of 8 hours (fig. 5).
Analysis Figure 4—The Alpine Overlook looking toward the northeast. The 2003 Cedar Fire was stopped at the visible Hotshot handline (A) running downslope through the center of Modeling of expected fire behav- the image. The fuel within the blackened ground immediately above the handline (B) ior using field measures of fuels was sparse chamise. The lighter burned area to the far left (C) was composed of lighter vegetation that had recovered after the 2001 Viejas fire. Note the previous Viejas fire from these young seral stands was handline (D) between the darker and lighter burned areas. The boulder outcropping (E) done with Behave Plus, a PC-based shown in Figure 1 is also visible. Photo by Randy Lyle. Windows software application used gerous conditions if ignited, pre- venting firefighters from continu- ing their fire suppression efforts. “If we had any hope of controlling the fire at this location,” Battalion Chief John Truett said, “it had to happen before it jumped into those heavy fuels across the canyon. There’s no way we’d send firefight- ers into that stuff” (J. Truett, pers. comm. 2005).
After a load of fire retardant was dropped by a helicopter, the crew of Engine 47 began laying hose below the overlook and along a barbed-wire fence, extinguishing the flames as they went. Following close behind, the Hotshot crew cut a cold trail through the chamise Figure 5—Infrared aerial overview of the Alpine Overlook area. The Cedar fire scar is chaparral on the southeast flank. shown in black, with unburned areas in red. Interstate 8 is shown at the bottom of the More aerial support was provided by photo with the overlook at the far lower left. The boulder outcropping shown in Figures two twin-engine S-2 tankers, which 1 and 4 is to the right of “A.” Two sample sites are at “B.” The approximate location dropped one load of red fire retar- where the Cedar fire jumped Interstate 8 is marked “C.” The Sweetwater River canyon is at the top of the photo. North is at the bottom. Image acquired and processed by Furgo dant, each, near the lower flank of EarthData, Inc. the fire burning into the chamise (fig. 4).
Volume 69 • No. 2 25 Conclusions The 3-year-old fuels within the Viejas Fire scar did, however, We have found that 3-year-old, make it possible for firefighters to post-fire native vegetation in a approach the area with acceptable recovering chamise chaparral stand risk in order to conduct fire sup- is capable of carrying a fire under pression activities. By 1800 hours moderate weather conditions. on Monday afternoon, the northeast Under more severe conditions, such winds had begun shifting to the fuels are capable of generating sub- northwest. The flames were already stantial flame lengths and spread Figure 6—Fuels status. Dead fuels backing southeast downslope dominated the samples within the 2001 rates. Consequently, these younger toward the heavy fuels across the Viejas Fire scar. fuels would not likely provide the river prior to the wind shift. If the necessary barrier to stop wildland fire had not been extinguished fires in chaparral systems under to predict wildland fire behavior within the short, 4-hour time win- natural conditions. This was dem- (see
Fire Management Today 26 fire spread, other factors (such as weather and topography) can domi- nate the outcome. For example, the western expansion of the 2003 Cedar Fire burned through numer- ous younger aged patches of veg- etation before hitting the commu- nity of Scripps Ranch in San Diego (Halsey 2006). Interestingly, this edge of the fire finally stopped in extremely dense chaparral without the aide of fire suppression efforts. This abrupt perimeter edge was likely caused by a combination of higher fuel moisture content, slow- ing of Santa Ana winds, and the influx of a marine moisture layer.
Other research has also suggested there is not a strong relationship Figure 8—2003 and 2007 San Diego County fire overlap map. The map shows the perimeters of the 2003 Paradise Fire (A), Cedar Fire (B), and Otay Fire (C). 2007 fires are between hazard of burning and labeled to the right of their perimeters: Poomacha Fire (D), Witch Creek Fire (E), and fuel age (Moritz and others 2004) Harris Fire (F). Major roads are represented by dark lines. Note the significant areas of and that large fires are not depen- overlap between the 2007 and 2003 fires. From Halsey (2008). dent on old age classes of fuels but spreading once they reach younger Our research does not lend sup- rather extreme weather conditions aged patches. Consequently, large port to this hypothesis as we found (Keeley and others 1999). In addi- shrubland wildfires over 10,000 that younger aged fuels of the tion, the relatively low lightning acres (4,047 ha) are viewed as type found in recovering chaparral frequency that occurs in southern “unnatural,” byproducts of past stands can carry fire, even under California, especially at lower eleva- fire suppression efforts that have moderate weather conditions. tions, would not likely ignite the allowed large, older stands of chap- Although fuel type and condition number of fires required to create arral to develop. play a critical role in determining small, mixed-aged patches of veg-
Table 2—Wildlands burned in 2003 and 2007 firestorms. The baseline data for this table was drawn from San Diego County vegetation data, 2003 fire scar perimeter, and 2007 fire perimeter. All results are expressed in thousands and rounded off to reflect estimated values and are intended only as general indicators of habitat loss, as some areas within mapped perimeters may not have burned. Compiled by Kit Wilson.
Total Area Area Area Percent Percent Percent Percent Habitat Burned Burned Burned burned Burned Burned Burned Area 2007 2003 Twice 2007 2003 Twice in 4 Yrs (acres/ha) (acres/ha) (acres/ha) (acres/ha) Chaparral 856 (346) 117 (47) 14% 173 (70) 20% 27 (11) 3% 31% Coastal + 290 (117) 94 (38) 32% 79 (32) 27% 34 (14) 12% 48% Scrub Grasslands 161 (65) 21 (8.5) 13% 17 (7) 11% 2 (1) 2% 22% Riparian + 64 (26) 9 (4) 14% 6 (2) 9% 2 (1) 3% 20% Marsh Woodland 112 (45) 33 (13) 30% 27 (11) 24% 10 (4) 9% 44% Forest 86 (35) 13 (5) 16% 20 (8) 23% 1 (0.4) 1% 37%
Volume 69 • No. 2 27 etation (Keeley 1982, Christensen fires may cause to native plant Ecology and Management Conference. 1985). communities. In conjunction with San Diego, CA: Association for Fire Ecology. this research, we have been investi- Halsey, R.W. 2008. Fire, chaparral, and sur- While we seriously doubt that gating the health of the study site’s vival in southern California. San Diego, southern California shrubland fires chaparral ecosystem after it was CA: Sunbelt Publications. 232 p. Keeley, J.E. 1982. The distribution of can be naturally constrained by subjected to the fires of 2001 and lightning and man-caused wildfires in younger aged vegetation, strategi- 2003. Preliminary data have shown California. In Conrad, C.E., and Oechel, cally placed fuel treatments can that there is significant mortality of W.E., eds. Proceedings, Symposium on Dynamics and Management of play a supporting role in fire sup- chamise burls, reduction of several Mediterranean-type Ecosystems. Gen. pression efforts. Our data support native plant species, and spread of Tech. Rep. PSW-58: 431–437. the observation that younger fuels invasive, non-native grasses such as Keeley, J.E. 2006. Fire management in post-fire chaparral ecosystems foxtail chess (Bromus madritensis). impacts on invasive plant species in the western United States. Conservation can provide greater opportuni- By adding more fire to the land- Biology. 20: 375–384. ties for firefighters to establish scape through rotational burning Keeley, J.E.; Fotheringham, C. J.; Morais, fire suppression anchor points, in order to create mosaics, type- M. 1999. Reexamining fire suppres- sion impacts on brushland fire regimes. especially under moderate weather conversion of native shrubland sys- Science. 284: 1829–1832. conditions and along the flanks of tems to non-native grasslands will Minnich, R.A. 1983. Fire mosaics in wind-driven fires. This explains the likely increase. southern California and northern Baja California. Science. 219: 1287–1294. desire of some land managers to Minnich, R.A. 1989. Chaparral fire his- establish an artificial “mosaic” of References tory in San Diego County and adjacent northern Baja California: an evaluation age classes on a landscape scale to Christensen, N.L. 1985. Shrubland fire of natural fire regimes and effects of sup- prevent the development of large, regimes and their evolutionary conse- pression management. In Keeley, S.C., quences. In Pickett, S.T.A.; White, P.S., contiguous old-age fuel beds (P. ed. The California Chaparral: Paradigms eds. The Ecology of Natural Disturbance. Reexamined. No. 34 Science Series. Scully pers. comm. 2008). However, New York: Academic Press: 85–100. Natural History Museum of Los Angeles beyond the strategic application of Dodge, J.M. 1975. Vegetational changes County: 37–47. associated with land use and fire history fuel manipulations at the wildland/ Moritz, M.A.; Keeley, J.E.; Johnson, in San Diego County. Ph.D. dissertation. urban interface, we have only lim- E.A.; Schaffner, A.A. 2004. Testing a Riverside, CA: University of California- basic assumption of shrubland fire ited understanding of the efficacy Riverside. management: How important is fuel Green, L.R. 1981. Burning by prescription of such treatments on the broad age? Frontiers in Ecology and the in chaparral. General Technical Report landscape. Environment. 2: 67–72. GTR-PSW-51. Albany, CA: Forest Service, Zhou, X.; Weise, D.; Mahalingam, S. 2005. Pacific Southwest Forest and Range Experimental measurements and numer- Another important factor to con- Experiment Station. 36 p. ical modeling of marginal burning in live Halsey, R.W. 2006. Weather, fuels, and sup- sider regarding the artificial cre- chaparral fuel beds. Proceedings of the pression during the 2003 Cedar Fire: Combustion Institute. 30: 2287–2294. ation of mixed-aged mosaics is the Which variables made the critical differ- resource damage that additional ence? Proceedings, 3rd International Fire
Fire Management Today 28 Fuel treatment guidebooK: illustrating treatment eFFeCts on Fire hazard Morris Johnson, David L. Peterson, and Crystal Raymond he Guide to Fuel Treatments for fuel treatments that consider (Johnson and others 2007) ana- The goal was to link desired future conditions for mul- Tlyzes potential fuel treatments information from tiple resources (e.g., wildlife, water, and the potential effects of those and timber production). Developers treatments for dry forest lands in silviculture and fire determined the final structure of the Western United States. The science to provide the guide after reviews by scientists guide examines low- to mid-ele- quantitative guidelines and resource managers and two test vation dry forest stands with high for fuel treatment that efforts involving national forests. stem densities and heavy ladder fuels, which are currently common consider desired future The scientific basis for fuel treat- due to fire exclusion and various conditions for multiple ments is documented in recent land management practices, such resources. syntheses (Graham and others as timber harvesting. These stands 2004, Peterson and others 2005) are the focus of potential manage- and numerous publications (Agee ment activities intended to modify protocols, and tools; writing peer- 1996, 2002; Brown and others forest structure and fuels to reduce reviewed documents that synthe- 2004; Carey and Schuman 2003; crown fire hazard on public lands. size and integrate the ecological Fitzgerald 2002; Kalabokidis and The guide is intended for use by and social science relevant to fuels Omi 1998; Keyes and O’Hara 2002; fire managers, silviculturists, and treatments; and delivering these Pollet and Omi 2002; Sandberg and other resource specialists who are products in a user-friendly format others 2001; Scott and Reinhardt interested in evaluating the effects to community leaders and educa- 2001; Weatherspoon 1996). The of fuel treatments on dry forest tors, fuels management specialists guide provides quantitative guide- ecosystems. and resource specialists, National lines for treatments based on the Environmental Policy Act (NEPA) scientific principles in these docu- Development of the planning team leaders, and line ments and is intended to cover a Guide officers in the Forest Service and broad range of possible treatments the Department of the Interior. and stand conditions. However, the In April 2003, the Forest Service Information derived from this effort representative cases in the guide initiated the Fuels Planning: is applicable to categorical exclu- are not comprehensive, and inter- Science Synthesis and Integration sion documents, environmental pretation and application of quanti- project (known as the Fuels impact statements, environmental tative output will typically need to Synthesis Project) to accelerate assessments, and other NEPA docu- be adjusted based on local condi- the delivery of research informa- ments. tions and objectives. tion to fuels specialists and others involved in project planning. The Scientists at the Pacific Wildland Analytical Tools geographic focus of the project was Fire Sciences Laboratory, Pacific on the dry forests of the Western Northwest Research Station, devel- The Fire and Fuels Extension of United States. Project goals includ- oped the guide in cooperation Forest Vegetation Simulator (FFE- ed developing accessible analyses, with other scientists and resource FVS) (Reinhardt and Crookston managers throughout the Western 2003) was used to prepare the Morris Johnson is an ecologist and David guide. This tool links forest growth Peterson is a research biologist with the United States. The goal was to link Forest Service’s Pacific Northwest Research information and data from silvicul- modeling with fire behavior model- Station, Pacific Wildland Fire Sciences ture and fire science in order to (1) ing to produce information relevant Laboratory, in Seattle, WA. Crystal to management of forest stands, Raymond is a researcher and Ph.D student assist decisionmaking about fuel at the Fire and Mountain Ecology Lab, treatments in dry forest stands and fuels, and fire. FVS has been widely Department of Forest Resources, University (2) provide quantitative guidelines used by resource managers and of Washington, Seattle, WA.
Volume 69 • No. 2 29 scientists for over two decades, The guide provides data were obtained from resource has been programmed to cover managers on national forest units many of the major forest types in quantitative guidelines throughout the Western United the United States, and is regarded for treatments based on States. Stands selected for analysis as a credible tool for applications the scientific principles had high stem densities and had in forest management (Dixon in these documents and not experienced recent fire or thin- 2002). Integration of fire concepts is intended to cover a ning. In the guide, only stands at is a recent and valuable exten- relatively low elevations and slopes sion of the FVS approach to forest broad range of possible of less than 40 percent were con- stand simulation, but it has not treatments and stand sidered as potential candidates for been available long enough to be conditions. fuel treatment. Fuel treatment thoroughly tested. However, it is scenarios are organized according the only analytical tool currently to Forest Service regions in the available that quantitatively links tion, forest structure and fuels are Western United States. stand dynamics and fire science. calculated for 50 years posttreat- At a minimum, FFE-FVS requires ment at 10-year increments, so that Fuel Treatments input of forest stand attribute data long-term stand conditions can be Fuel treatment scenarios analyzed (species, diameter at breast height assessed and users can determine in the guide to Fuel Treatments [d.b.h.], and height), but fuels data when additional fuel treatments were determined with extensive are extremely helpful. might be needed. Users familiar feedback from Federal resource with FFE-FVS have the option of managers. These scenarios cover In the guide, the effects of fuel running their own simulations a range of potential thinning and treatments are quantified for forest to calculate site-specific effects of surface fuel treatments that would structure, surface fuels, and poten- treatments. be reasonable and appropriate alter- tial fire behavior. FFE-FVS was natives for NEPA analysis and simi- used to calculate a variety of fuel Scenarios displayed in the guide lar documentation. The scenarios treatment combinations (includ- are intended to represent a range illustrate representative situations ing the no-action alternative, four of dry forest types in the Western that might be encountered in oper- levels of thinning, three types of United States, specifically those for- ational management and planning surface fuel modification, and pre- ests dominated by ponderosa pine and do not illustrate all possible scribed fire alone) for each of 25 (Pinus ponderosa Dougl. ex Laws), treatments. representative forest stands (fig. mixed conifers—often including 1). FFE-FVS runs are summarized Douglas-fir (Pseudotsuga menziesii Thinning from below (or low thin- for each stand with visualizations [Mirb.] Franco) as a codominant— ning) refers to removal of stems and extensive tabular data (not and pinyon-juniper (Pinus spp. starting from smallest to increas- included in this article). In addi- and Juniperus spp.). Specific stand ingly larger stems until the target density is reached. In practice, thinning from below often has a No action Prescribed Thin from Thin from Thin from Thin from d.b.h. limit below which no stems fire only below to 50 below to 100 below to 200 below to 300 tpa, 18-in tpa, 18-in tpa, 18-in tpa, 18-in are harvested, with that lower limit d.b.h. limit d.b.h. limit d.b.h. limit d.b.h. limit set to reduce costs and maximize value of harvested material.
In guide scenarios, stem harvesting begins with trees smaller than 1 in (2.5 cm) d.b.h. and then proceeds No action Pile and Prescribed to larger stems. For all thinnings, burn fire no trees larger than 18 in (44 cm) d.b.h. are harvested. This limit is intended to retain larger, more Figure 1—Matrix of thinning and surface fuel treatments implemented for each stand in the guide to Fuel Treatments. Each stand was projected through a series of 14 potential fire-resistant individuals. In prac- treatments. tice, this upper d.b.h. limit could
Fire Management Today 30 Table 1—Summary of values and assumptions used in FFE-FVS for surface fuel treatments.
Surface fuel treatment FFE FVS values and assumptions FVS keywords All boles greater than 6 in diameter at breast height (d.b.h.) are removed from stand. The No action entire tree (branch and bole) and branch mate- Yardloss rial from trees greater than 6 in d.b.h. are left in stand. All boles greater than 6 in d.b.h. are removed from stand. The entire tree (branch and bole) and branch material from trees greater than Yardloss Pile and burn 6 in d.b.h. are left in stand. 80 percent of the PileBurn remaining fuel from the entire stand is concen- trated into piles that cover 10 percent of the stand area. No tree mortality will result. All boles greater than 6 in d.b.h. are removed from stand. The entire tree (branch and bole) and branch material from trees greater than 6 in d.b.h. are left in stand. Windspeed at 20 ft above Yardloss Prescribed fire vegetation = 10 mph. FVS predefined moisture SimFire group (3) selected to represent fuel moisture percentages for prescribed fires. Temperature equals 70 °F. Note: predefined moisture values are specific to FVS variants. be higher or lower depending on acre (TPA) (741 trees per hectare tiveness. Assumptions regarding local harvest specifications and [TPH]) rarely changes fuel condi- slash disposal, material left on site, resource objectives. Thinning from tions enough to modify fire hazard area affected, and effectiveness of below is the most commonly used significantly from initial stand con- treatments are summarized in table approach to modify stand struc- ditions. 1. Prescribed fire is considered to ture, density, and fuels, although be a broadcast burn that covers the many other silvicultural approaches Some managers prefer to use basal entire treatment area. are available (Graham and others area (BA) as a target for thinning. 1999). Thinning as used within FVS This measurement may be more The following is an example of sce- is applied equally across a given appropriate for even-aged stands narios derived from the guide. stand. In practice, variable-density with relatively low variability in thinning—a spatial pattern of tree tree size. BA is calculated for each clumps and openings—can be used thinning treatment, so both BA and to achieve the same final tree den- stem density are available for all sity but attain greater heterogeneity scenarios. in stand structure. Variable-density thinning cannot be represented In practice, techniques used for in FVS and is, therefore, not con- modification of activity fuels and sidered here. For target densities residual surface fuels vary consid- different than those in the guide, erably, as does the effectiveness of users can interpolate or extrapo- those techniques. Options included Figure 2—Computer simulation of forest late the results found in tables and in the guide are intended to cap- structure prior to the four thinning visualizations. Exploratory runs of ture the more common approaches treatments in the Forest Vegetation Simulator. Stand visualization taken from FFE-FVS indicate that thinning to currently used in the field and to stand data for the Bitterroot National densities greater than 300 trees per represent moderately high effec- Forest.
Volume 69 • No. 2 31 50 trees per acre 100 trees per acre TPH, respectively) treatments have a more closed canopy and fire behavior is influenced less by grass fuels, so flame lengths and poten- tial BA mortality are lower than the more open stands. Activity fuels are reduced by the pile-and-burn treatment and, to a greater extent, by the prescribed fire treatment, 200 trees per acre 300 trees per acre which also reduces litter and duff, but flame lengths and BA mortality remain high owing to grass fuels.
Silvicultural and Surface Fuel Treatments: Long-Term Effects Although the prescribed fire-only treatment does not reduce crown fire potential in the short term, the Figure 3—Computer simulation of forest structure following the four thinning treatments predicted fire type is surface fire in the Forest Vegetation Simulator: thinning to 50, 100, 200, and 300 trees per acre. Stand visualization taken from stand data for the Bitterroot National Forest. after 10 years. Crown fire poten- tial continues to decline as canopy Initial Conditions/No-Action Silvicultural and Surface Fuel base height increases and flame Trajectory Treatments: Immediate Effects lengths decrease. In all thinning This stand (fig. 2) has a high tree According to results from FFE-FVS, treatments, flame lengths decrease density of 2,345 TPA (5,795 TPH) the prescribed fire-only treatment over time as canopy cover increases primarily composed of grand fir and decreases canopy bulk density and fuel model assignment shifts Douglas-fir with a ponderosa pine and slightly increases canopy base from predominantly fuel model 2 overstory. Woody fuel loading is 9 height, but not enough to prevent to predominantly fuel model 9. The tons/ac (20,175 kg/h), and litter and passive crown fire for severe fire 200 TPA treatment has the great- duff loading is 7 tons/ac (15,692 weather. This treatment reduces est long-term effect on crown fire kg/h). Canopy bulk density is surface fuels in all size classes, but potential, with a predicted surface 0.0087 lb/ft3 (0.14 kg/m3), and can- flame lengths increase after treat- fire type for 50 years with pile-and- opy base height is 3 ft (0.91 m), so ment owing to grass fuels associ- burn or no surface fuel treatment ladder fuels are sufficient to enable ated with the use of fuel model 2. and 40 years with prescribed fire passive crown fire, but canopy fuels Grass fuels are not tracked in FFE treatment. The 50 TPA (124 TPH) are not sufficient to enable active and may or may not be the primary treatment had the most short-lived crown fire spread. Crowning index fuel following prescribed fire. effect on crown fire potential, with is 19 and severe weather wind speed regeneration causing a drop in can- is 17 mph (27 km/h), so although All thinning treatments reduce opy base height in 30 years regard- this stand is not classified as active canopy bulk density and increase less of surface fuel treatment. crown fire, crown fire hazard is canopy base height; the greater high. Potential BA mortality is 97 the thinning, the greater is the References percent for severe fire weather. change in forest structure (fig. 3). Agee, J.K. 1996. The influence of for- With no action, flame lengths, sur- The predicted fire type after treat- est structure on fire behavior. In: face fuels, and canopy base height ment is surface fire for all thinning Sherlock, J.; Landram, M.; Gray, S., eds. Proceedings, 17th annual forest vegeta- increase slightly over time, with options, but the more open stands tion management conference, Redding, crown fire potential decreasing in are characterized predominantly CA. 52–68. 20 years and then increasing again by fuel model 2, so flame lengths Agee, J.K. 2002. Fire behavior and fire-resil- ient forests. In: Fitzgerald, S., ed. Fire in 40 years. Crown fire potential increase and potential BA mortality in Oregon’s forests: risks, effects, and and flame lengths remain low for remains above 20 percent regard- treatment options. Portland, OR: Oregon moderate fire weather for the entire less of surface fuel treatment. The Forest Resources Institute: 119–126. Anderson, H.E. 1982. Aids to determining 50-year projection. 200 and 300 TPA (494 and 741 fuel models for estimating fire behav-
Fire Management Today 32 ior. GTR-INT-122. Ogden, UT: Forest Graham, R.T.; Harvey, A.E.; Jain, T.B.; Tonn, Peterson, David L.; Johnson, M.C.; Agee, Service, Intermountain Forest and Range J.R.. 1999. The effects of thinning and J.K.; Jain, T.B.; McKenzie, D.; Reinhardt, Experimental Station. 22 p. similar stand treatments on fire behavior E.D. 2005. Forest structure and fire haz- Brown, R.T.; Agee, J.K.; Franklin, J.F. 2004. in western forests. Gen. Tech. Rep. PNW- ard in dry forests of the Western United Forest restoration and fire: principles GTR-463. Portland, OR: Forest Service, States. Gen. Tech. Rep. PNW-GTR-628. in the context of place. Conservation Pacific Northwest Research Station. 27 p. Portland, OR: Forest Service, Pacific Biology. 18: 903–912. Graham, R.T.; McCaffrey, S.; Jain, T.B.. Northwest Research Station. Carey, H.; Schuman, M. 2003. Modifying 2004. Scientific basis for changing forest Pollet, J.; Omi, P.N. 2002. Effect of thin- wildfire behavior—the effectiveness of structure to modify wildfire behavior and ning and prescribed burning on crown fuel treatments: the status of our knowl- severity. Gen. Tech. Rep. RMRS-GTR-120. fire severity in ponderosa pine forests. edge. National Community Forestry Fort Collins, CO: Forest Service, Rocky International Journal of Wildland Fire. Center, Southwest Region Working Paper Mountain Research Station. 43 p. 11: 1–10. 2. Santa Fe, NM: Forest Trust, National Johnson, Morris C.; Peterson, David L.; Sandberg, D.V.; Ottmar, R.D.; Cushon, Community Forestry Center. 26 p. Raymond, Crystal L. 2007. Gen. Tech. G.H. 2001. Characterizing fuels in the Dixon, G.E., compiler. 2002. Essential FVS: Rep. PNW-GTR-686. Portland, OR: Forest 21st century. International Journal of a user’s guide to the Forest Vegetation Service, Pacific Northwest Research Wildland Fire. 10: 381–387. Simulator. Internal Rep. Fort Collins, Station. 322 p. Scott, J.H.; Reinhardt, E.D. 2001. Assessing CO: Forest Service, Forest Management Kalabokidis, K.D.; Omi, P.N. 1998. crown fire potential by linking models Service Center. 196 p. Reduction of fire hazard through thin- of surface and crown fire behavior. Res. Fitzgerald, S.A. 2002. Fuel-reduction and ning/residue disposal in the urban inter- Pap. RMRS-RP-29. Fort Collins, CO: restoration treatments for Oregon’s face. International Journal of Wildland Forest Service, Rocky Mountain Research forests. In: Fitzgerald, S., ed. Fire in Fire. 8: 29–35. Station. Oregon’s forests: risks, effects, and treat- Keyes, C.R.; Morgan, V.J. 2007. Putting out Weatherspoon, C.P. 1996. Fire-silviculture ment options. Portland, OR: Oregon fire with gasoline: pitfalls in the silvi- relationships in Sierra Forests. Sierra Forest Resources Institute: 127–138. cultural treatment of canopy fuels. Gen Nevada Ecosystem Project: Final Tech. Rep. PNW-GTR-2003. Portland, Report to Congress. Assessments and OR: Forest Service, Pacific Northwest Scientific Basis for Management Options. Research Station. Davis, CA: University of California- Davis, Centers for Water and Wildland Resources.
Volume 69 • No. 2 33 a suite oF Fire, Fuels, and smoKe management tools Roger D. Ottmar, Clint S. Wright, and Susan J. Prichard
he Fire and Environmental • Consume 3.0. Consume provides The tools can be used individually Research Applications Team users the ability to estimate fuel or in combination to support a T (FERA) of the Forest Service, consumption and emissions variety of management situations. Pacific Northwest Research Station, from fuelbeds burned during For example, for a fuel reduction is an interdisciplinary team of prescribed and wildland fires. See project, managers may need to first scientists that conduct primary
Fire Management Today 34 include a hurricane damage series The photo series are important with development of new fire- and for the Southern United States and land management tools that can be natural resource-based software a volume focusing on relevant fuel- used to assess ecological landscapes applications, highlighted the need bed types in the San Francisco Bay through appraisal of living and dead for an electronic version of the area. woody material, vegetation biomass, photo series. DPS was the result and stand characteristics. Once (see Wright and others, in this vol- Each volume contains up to four an ecological assessment has been ume, for a more detailed descrip- photo series for 1 to 17 sites. Series completed, stand treatment options, tion). DPS provides easy access to for each site include standard, such as prescribed fire or harvest- data and images from all of the vol- wide-angle, and stereo-pair photo- ing, can be planned and implement- umes, series, and sites in the NFPS. graphs and inventory data summa- ed to better achieve desired effects Information presented in this new rizing: while minimizing negative impacts format can be used for planning • vegetation composition, struc- on other resources. fuels treatments or other manage- ture, and loading; woody material ment actions and as inputs to fire loading; Digital Photo Series behavior and fire effects models and • density by size class, forest floor applications. DPS has the ability to NFPS was developed primar- depth, and loading; and grow as new photo series are devel- ily for field-based assessments. • various site characteristics. oped and as the priorities and needs Technological advances, coupled of fire and fuels managers change and evolve.
Published volumes of the Natural Fuels Photo Series.
Region Fuelbed Type(s) Volume Pacific Northwest Mixed-conifer with mortality, western juniper, sagebrush, grassland I1 Alaska Black spruce, white spruce II1 Alaska Hardwoods with spruce IIa1 Rocky Mountains Lodgepole pine, quaking aspen, gambel oak III1 Southwest Pinyon-juniper, chaparral, sagebrush IV1 Midwest Red and white pine, northern tallgrass prairie, mixed oak V1 Lake States Jack pine Va1 Southeast Longleaf pine, pocosin, marshgrass VI1 Southeast Sand hill, sand pine scrub, hardwoods with white pine VIa1 West Coast Oregon white oak, California deciduous oak, mixed-conifer with shrub VII1 Northeast Hardwood, pitch pine, red spruce/balsam fir VIII1 Southwest Oak/juniper IX2 Montana Sagebrush with grass, ponderosa pine-juniper, X2 Hawaii Grassland, shrubland, woodland, forest N/A2 Brazil Cerrado N/A2 Mexico Montane subtropical forests, temperate forests, montane shrublands N/A2 Southeast Hurricane damaged pine N/A3
1 Photo series can be purchased from the National Interagency Fire Center in Boise, ID, for a nominal charge. 2 Photo series can be requested free of charge from the Fire and Environmental Research Applications Team. 3 This photo series is in preparation; expected publication is spring 2009.
Volume 69 • No. 2 35 Application of NFPS and DPS NFPS and DPS are useful tools in several branches of natural resource science and management. Inventory data provided by these tools can be used as inputs for evaluating animal and insect habi- tat, nutrient cycling, and micro- climates. Fire managers will find these data useful for predicting fuel consumption, smoke production, fire behavior, and fire effects dur- ing wildfires and prescribed fires. In addition, the photo series can be used to appraise carbon sequestra- tion, an important factor in predic- tions of future climate, and to link remotely sensed signatures to live and dead fuels on the ground. Photo series photograph from the new hurricane damage photo series that is in preparation. FCCS FCCS enables land managers to create and catalog fuelbeds for Fuelbed strata and categories fuels and fire planning. It contains in the FCCS searchable fuelbed data sets that represent much of North America and were compiled from scien- tific literature, natural fuels photo series, fuels data sets, and expert opinion. The system allows custom- ization of these fuelbeds or creation of new fuelbeds to represent a par- ticular situation or scale of interest. FCCS reports assigned and calcu- lated fuel characteristic for each of six fuelbed strata, including the canopy, shrubs, nonwoody, woody, litter-lichen-moss, and duff.
FCCS calculates the relative fire hazard of each fuelbed, includ- ing surface fire behavior, crown fire, and available fuel potentials, scaled on an index from 0 to 9. The FCCS also uses a modified ver- sion of the Rothermel surface fire behavior equations (Rothermel 1972, Sandberg and others 2007) to predict actual surface fire behavior,
Fire Management Today 36 including reaction intensity (Btu Acronyms -2 -1 ft min ), flame length (ft), and DPS Digital Photo Series rate of spread (ft min-1), and based on both benchmark and user-spec- FCCS Fuel Characteristic Classification System ified environmental conditions. By FEPS Fire Emision Production Simulator comparing predicted flame length FERA Fire and Environmental Research Applications Team and rate of spread, FCCS provides a crosswalk to any of the origi- FIREMON Fire Effects Monitoring and Inventory Protocol nal 13 Fire Behavior Prediction FOFEM First Order Fire Effects Model System fuel models (Albini 1976) FVS Forest Vegetation Simulator and to any of the 40 standard fire behavior fuel models (Scott and NFPS Natural Fuels Photo Series Burgan 2005). FCCS also reports carbon storage by fuelbed stratum, fuelbed data, summarizing and category, and subcategory and pre- Protection Agency is developing a calculating fuel characteristics, and dicts the amount of combustible national air pollutant and carbon predicting surface fire behavior, carbon based on selected fuel mois- emission inventory based on FCCS crown fire behavior, and available ture scenarios. Finally, the system fuelbeds (fig. 1). LANDFIRE (a proj- fuel for consumption. FCCS also reports in English and metric units, ect producing consistent and com- provides the necessary inputs to provides the capability to upload prehensive maps and data describ- run fuel consumption and emis- photos to represent fuelbeds, and ing vegetation, wildland fuel, and sion production models, such as can be run in a batch mode to pro- fire regimes across the United Consume and FEPS. vide outputs for multiple fuelbeds States ) (Rollins and others 2006) is simultaneously. also developing a 30-meter resolu- FCCS fuelbeds are being mapped tion map layer of FCCS fuelbeds for on the Okanogan-Wenatchee and the United States. Application of the FCCS Deschutes National Forests to allow FCCS facilitates the mapping of managers to evaluate fire hazard FCCS was introduced to managers fuelbed characteristics and fire haz- and maximize fuel treatment effec- and scientists during 15 national ard assessment by storing realistic tiveness. The U.S. Environmental workshops and through 8 pub-
Figure 1—One-km resolution FCCS fuelbed data are available for the continental United States (map available through
Volume 69 • No. 2 37 The photo series are important land Consume predicts fuel consump- management tools that can be used to tion, pollutant emissions, and heat release based on a number of vari- assess ecological landscapes through appraisal ables, including fuel loadings, fuel of living and dead woody material and vegetation moisture, and other environmental biomass and stand characteristics. factors. Using these predictions, resource managers can determine when and where to conduct a pre- lished papers in a special section consumption focused on prescribed scribed burn or to plan for a wild- of the Canadian Journal of Forest burning following logging in for- land fire to achieve desired objec- Research (Berg 2007; Ottmar and ested ecosystems. FERA’s new fuel tives while reducing the impact on others 2007; Riccardi and oth- consumption studies in natural other resources. ers 2007a; Riccardi and others environments (developed with sup- 2007b; Sandberg and others 2007a; port from the Joint Fire Science Consume allows land managers and Sandberg and others 2007b; Schaaf Program and the National Fire researchers to input fuel character- and others 2007; McKenzie and Plan) have improved our under- istics, lighting patterns, fuel condi- others 2007). Source data refer- standing of fuel consumption in tions, and meteorology to more ences for each fuelbed, as well as shrublands (including chaparral, accurately predict fuel consump- supplementary fuelbeds useful to sagebrush, and palmetto-galberry tion and emissions. Consume can specific locations and purposes, types), hardwood forests (includ- import data from the FCCS, and its can be found on the FCCS Web site ing southern and eastern regions reports are formatted to feed other (
Fire Management Today 38 Consume algorithms to estimate Application of FEPS Riccardi, C.L.; Ottmar, R.D.; Sandberg, emissions to predict smoke impacts D.V., Andreu, A.; Elman, E.; Kopper, K.; Hourly emission and heat release Long, J. 2007a. The fuelbed: a key ele- across landscapes. data for wildland fires in fuel types ment of the fuel characteristic classifica- tion system. Canadian Journal of Forest in the United States produced by Research. 37(12): 2,394–2,412. Fire Emissions FEPS 1.1 can be fed into dispersion Riccardi, C.L.; Prichard, S.J.; Sandberg, Production Simulator models, such as the BlueSky smoke D.V.; Ottmar, R.D. 2007b. Quantifying physical characteristics of wildland fuels Modeling the impact of emissions modeling framework (O’Neil and using the fuel characteristic classifica- from wildland fire on visibility and others 2003) and V-Smoke (Lavdas tion system. Canadian Journal of Forest public health requires the rates 1996) for assessing smoke impacts Research. 37(12): 2,413–2,420. Rollins, M.G.; Frame, C.K., tech. eds. as well as the total amount of fuel from wildland fire. 2006. The LANDFIRE prototype project: consumption, heat release, and nationally consistent and locally relevant emissions production. These rates References geospatial data for wildland fire manage- ment. Gen. Tech. Rep. RMRS-GTR-175. are required inputs for smoke dis- Albini, F.A. 1976. Estimating wildfire Fort Collins: Forest Service, Rocky persion models for assessing poten- behavior and effects. Gen. Tech. Rep. Mountain Research Station. 416 p. tial visibility and health impacts of INT-30. Ogden, UT: Forest Service, Rothermel, R.C. 1972. A mathematical Intermountain Research Station. 92 p. smoke at a distance from the fire. model for predicting fire spread in wild- Berg, E. 2007. Characterizing and clas- land fuels. Res. Pap. INT-115. Ogden, UT: FEPS, an update of the Emissions sifying complex fuels: A new approach. Forest Service, Intermountain Research Production Model (EPM) (Sandberg Canadian Journal of Forest Research. Station. 37(12): 2381–2382. and Peterson 1984), models the Sandberg, D.V.; Riccardi, C.L.; Schaaf, M.D. Dixon, G.E., comp. 2003. Essential FVS: 2007a. Fire potential rating for wildland characteristics of prescribed burns A User’s Guide to the Forest Vegetation fuelbeds using the Fuel Characteristic and wildland fires. FEPS signifi- Simulator. Internal Rep. Fort Collins, Classification System. Canadian Journal cantly improved the usability, appli- CO: U.S. Department of Agriculture, of Forest Research. 37(12): 2456–2463. Forest Service, Forest Management Sandberg, D.V.; Riccardi, C.L.; Schaaf, M.D. cability, and accuracy of EPM. FEPS Service Center. 193 p. 2007b. Reformulation of Rothermel’s 1.1 includes the fuels data from the Lavdas, L.G. 1996. VSMOKE: Users manual. Wildland Fire Behaviour Model for most popular fuelbeds in the FCCS Gen. Tech. Rep. SRS-006. Asheville, Heterogeneous Fuelbeds. Canadian NC: Forest Service, Southern Research Journal of Forest Research. 37(12): and produces hourly emission and Station. 156 p. 2438–2455. heat release data for prescribed and McKenzie, D.; Raymond, C.L.; Kellogg, Sandberg, D.V.; Peterson, J. 1984. A source wildland fires. It can also accept L.-K.B.; Norheim, R.A; Andreu, A.G.; strength model for prescribed fire in Bayard, A.C.; Kopper, K.E.; Elman. E. fuel consumption data generated by coniferous logging slash. Presented to 2007. Mapping fuels at multiple scales: the 1984 Annual Meeting, Air Pollution the FOFEM (Reinhardt and others landscape application of the Fuel Control Association, Pacific Northwest 1997), Consume 2.1, and Consume Characteristic Classification System. Section. Reprint #84.20. Portland, 3.0. Canadian Journal of Forest Research. OR: Forest Service, Pacific Northwest 37(12): 2421–2437. Research Station. 10 p. O’Neil, S.M.; Ferguson, S.A.; Peterson, J.; Schaaf, M.D.; Sandberg, D.V.; Schreuder, FEPS distributes total fuel con- Wilson, R. 2003. The BlueSky smoke M.D.; Riccardi, C.L. 2007. A concep- sumption amounts over the life of modeling framework. 5th Symposium on tual framework for ranking crown fire Fire and Forest Meteorology, Orlando, potential in wildland fuelbeds. Canadian the burn to generate hourly emis- FL. Available at
Volume 69 • No. 2 39 using Wind models to more eFFeCtiVely manage WildFire Brian Potter and Bret Butler
Fire and Wind urface wind is often the domi- This resolution of wind information can be nant environmental variable useful to fire models simulating fire Saffecting wildland fire inten- growth in very specific locations, sity and spread. Traditionally, fire analysts and managers have such as individual drainages or ridges. depended on local measurements and site-specific forecasts to deter- mine winds influencing their fire. vided at fine scales (tens of yards yield useful projections include: However, advances in computer or meters). Fire behavior analysts • Broad-scale pressure variations hardware, increased availability of (FBANs), long-term fire analysts • Topography electronic topographical data, and (LTANs), and incident meteorolo- • Atmospheric stability advances in numerical methods for gists (IMETs) on fire management • Surface heating computing winds have led to the teams use all the technology • Surface friction development of new tools capable available to them to prepare their of simulating surface wind flow. forecasts. To do this accurately and Broad-scale pressure variations Several options for estimating efficiently, they need to know the (such as pressure ridges, troughs, winds across the landscape now strengths and weaknesses of the and fronts) influence surface exist, and they have the potential available wind models—knowing wind flow primarily as the pres- to dramatically improve fire growth where and how to obtain wind sure engine that drives the flow. estimates. Their benefits include information for a specific site at a Topography can modify the pres- improved firefighter and public fine scale can be a challenge. sure-driven flow of winds, changing safety and more efficient use of fire both direction and speed, especially management resources. Some wind models are more appro- near the ground. Stability, defined priate for research, some for daily as the tendency for air to move Wind modeling for fire manage- forecasts, and some for field appli- up or down, is caused by vertical ment presents unique challenges. cations. The goal of this article is changes in temperature and mois- Fire managers must develop new to provide an overview of models ture and is a key factor in deter- tactics within a matter of a few that can be used to support opera- mining gustiness. Surface heating hours while considering changing tional fire management. We briefly differences arise from variations in weather conditions and the poten- describe how the different models soil types, vegetation, and water/ tial for fire to move into new areas. work and what they provide in land boundaries, and are responsi- Often, to be relevant to short-term terms of output. ble for phenomena such as onshore fire management tactics, wind and offshore breezes. Surface fric- speed and direction must be cal- Wind Models tion caused by the size and orienta- tion of vegetation and other surface culated within an hour or two of a All physics-based wind models use features (e.g., forest, brush, grass, weather forecast and must be pro- some derivation of Newton’s laws of and open water) create more or motion and thermodynamics and less drag on the wind. This drag, Brian Potter is a research meteorologist need some starting data, such as or friction, can force the air to rise, on the AirFire Team at the Forest Service the wind’s direction and speed in or Pacific Wildland Fire Sciences Laboratory fall, or flow around an area. The near the area being modeled. These in Seattle, WA. Bret Butler is a research fire also has a significant influence mechanical engineer for the fire, fuel, data are commonly called bound- on local wind flow, but no exist- and smoke science program at the Forest ary conditions or initial conditions. Service Missoula Fire Sciences Laboratory ing wind model includes the effect Further information needed to in Missoula, MT. of the fire on the winds around it
Fire Management Today 40 (except for a few currently appro- sides and tops of hills. However, The most widely known prognos- priate only as research tools). the wind estimates degrade on the tic models are those used by the lee-sides of hills because turbulence National Weather Service (NWS) The art of wind modeling lies in is not included in the model and it for daily weather forecasting. The determining how the various wind is necessary to accurately simulate NWS uses several models, the most models treat these processes and the effects of hilltops or terrain widely known perhaps being the their interactions. Proper use of curvature on the surface winds. North American Meso (NAM) and wind models requires that the WindWizard, another tool devel- the Global Forecast System (GFS) user recognize what features are oped over the last 8 years, includes models. NAM output is available at influencing the situation at hand. the processes found in WindNinja 7.5-mile (12-km) resolution, while As a general rule, the more physi- but also includes the effects of the highest resolution GFS output cal processes the model includes, nonbuoyant (mechanically) gener- is 50 miles (81 km). Another prog- the more accurately it estimates ated turbulence (Forthofer 2007). nostic model, the NCAR/Penn State their influence, but also the slower Because of this added complexity, Mesoscale Model 5 (MM5), is pri- and/or more computer-intensive outputs from WindWizard show marily a research model, designed the model will be. Similarly, the better accuracy, including moderate to be run on large computers for more extensive the geographic area being modeled, the more computer memory and time the model needs Wind variations on a similar scale can cause to complete a simulation. sudden and dramatic changes in fire behavior; in other words, fine-scale variations in winds can All wind models fall into one of two categories: diagnostic or prognos- significantly influence fire growth at larger scales. tic. Diagnostic wind models give an estimate of winds at one specific time and do not provide any infor- success in predicting lee side winds. scientific research. Spatial resolu- mation on potential change. Such Both of these models have been tion of MM5 data depends on the models essentially assume that the shown to simulate wind in stable source, but grid resolution is rarely forces producing the winds are and neutral atmospheric conditions finer than 2.5 miles (4 km). The not changing quickly over time, well, in a qualitative sense, with newest prognostic model is the which make the calculations much WindWizard providing the better Weather Research and Forecasting simpler. Prognostic wind models, accuracy of the two. Lastly, the (WRF) model. WRF is a faster, taking into account changing con- CALMET model (Scire and others updated version of MM5 capable of ditions, estimate how the winds will 2000) was originally developed for running with a grid as fine as 330 behave over time. Prognostic mod- air pollution studies and includes feet (100 m), although the compu- els are the more accurate of the effects of turbulence, slope flows, tational requirements are huge. All two types, but they require much and flow over land and water sur- of these prognostic models simulate more computer time, more data to faces. It is the most computation- the atmosphere in all three dimen- start the calculations, and generally ally intensive of the three models sions as it changes over time. To do more expertise to run. The farther named here, and requires substan- this requires powerful computers out in time the prognostic model tially more expertise to operate. In and large amounts of time—they forecasts, the higher the likelihood comparison to prognostic models, cannot be used on a laptop in the the simulations will deviate from these diagnostic models typically field and require users with exten- reality. have relatively cheap computational sive specialized skills. However, the requirements (from seconds to a advent of remote computing has Diagnostic models vary in com- couple of hours per run on a single provided access to the output from plexity: for example, WindNinja, a processor computer). The price for these models for meteorological recently released wind simulation cheap computing is incorporation specialists working in fire camps. tool, considers a minimum number of fewer parameters and a wind of physical factors (Forthofer 2007). field prediction capability of only Generally, diagnostic models can This tool generally gives good one point in time. provide predicted winds at the 330- estimates of winds on the upwind foot (100-m) scale and, in some
Volume 69 • No. 2 41 cases, at even finer scales. This res- In general, the model sequential weather patterns and to olution of wind information can be provide fine spatial resolution that useful to fire models simulating fire must provide output better accounts for fine scale topo- growth in very specific locations, data on the same graphical effects. Think of this as such as individual drainages or or finer scale than the equivalent of taking individual ridges. The relatively coarse scale of the issue of greatest image frames from a video and most prognostic models can limit enlarging or enhancing them for their use on wildland fires: a large concern. greater detail. fire might be represented by only a few grid points (predicted values) In addition to wanting winds data necessary wind data. If the issue of and many fires by possibly just one. for existing fires, fire managers concern is average rate of spread The coarse resolution also means often are interested in “what if” on a fire complex, then wind data that the models approximate the scenarios that explore the effects of on a scale of a few miles or kilo- effects of larger features, such as various wind scenarios on planned meters may be adequate—though mountain ranges, but smooth over fire events (prescribed fires) or this might be too wide an area for small-scale terrain features. As a for the analysis of long-term fire typical diagnostic modeling, yet too result of such smoothing, the mod- growth potential or to understand limited for most prognostic mod- els cannot resolve the winds modi- what caused a fire to burn as it did els. In such a situation, computer fied by individual ridges, valleys, or (e.g., accident investigations). For access and time constraints are mountains. In many cases, these example, it is useful to know what more likely to dictate the modeling small-scale features have a great might happen to a prescribed fire option selected. influence on the local winds influ- under various potential weather encing a fire and should be consid- scenarios or to project potential fire The ideal model for a given appli- ered (Butler and others 1998). behavior at specific locations if an cation must include the physical unplanned ignition should occur. processes acting on the air at the Matching the Model same time scales as the issue of It is not feasible for the average to the Application concern. If one needs to estimate fire modeler to obtain multiple, winds for a period during which There are two critical issues that custom prognostic weather simu- there is considerable change in the influence the selection of a wind lations. Even if it was feasible, it regional pressure pattern or when model to support management of would be difficult to specify the the surface heating is changing a specific fire: the management initial and boundary conditions for rapidly (as when the sun crosses a objectives and the scale of input the simulations, as these are typi- ridge near the fire front) or when (and therefore output) data avail- cally obtained from coarser-scale there are strong differences in able to address those concerns. In global models and weather station surface heating and weak regional general, the model must provide measurements not generally avail- pressure variations (such as near a output data on the same or finer able and require the expertise of an lake or ocean shore in the absence scale than the issue of greatest con- experienced meteorologist to inter- of a cold front), then a prognostic cern. When the concerns relate, for pret. In cases where time, compu- model may be necessary. Most of instance, to the rate of fire spread tational resources, and technical the rest of the time, one need not up a valley or over a ridge, the expertise are abundant, prognostic include all of the physical processes model must show winds at numer- models could be appropriate; but to get a good estimate of winds. If ous locations in the valley or along in general, given typical wildland high resolution is also important, the ridge—typically at a resolu- fire management time and resource then the best solution might be tion of about 330 feet (100 m) or constraints, diagnostic models are to use the relatively coarse-scale less between points. This suggests the only practical option available wind information generated by selecting a diagnostic model if for such pre-planning needs. model results are needed quickly. prognostic models as input to a diagnostic model. The diagnostic If, in contrast, the issue relates to Model Application smoke dispersion across a region model can be run multiple times of one or several States, then only from the prognostic simulation to Recent research and anecdotal a prognostic model can provide the capture alternative scenarios or observations suggest that variations
Fire Management Today 42 winds at high resolution; thus, they cannot resolve the fine-scale wind variations near the ground surface.
Forthofer (2007) has shown that the use of winds generated by the WindNinja and WindWizard diagnostic models increased the accuracy of fire spread simulations for the South Canyon Fire (Butler and others 1998) over those based on uniform wind fields (figs. 1 and 2). The winds from the diagnostic models significantly improved the accuracy of growth projections for fires burning in complex terrain. Also, the fire spread simulations for both Mann Gulch and South Canyon were more sensitive to ini- tial wind direction when run with uniform winds than when they were run with a diagnostic wind model. Forthofer interprets this to mean that local terrain features, such as the individual turns and twists of drainages, spur ridges, or Figure 1—Fire growth simulation based on uniform wind speed and direction. White and rock outcroppings, tend to affect black dashed lines represent actual fire perimeters prior to and after the large fire spread event that occurred July 6, 1996. Pink dashed lines are firelines constructed by crew. The and control near-surface wind speed wind used for this simulation was steady from the west at 32 mph. The fire growth was and direction, sometimes regardless simulated using FARSITE. of the upper-air wind direction. in slope and vegetation on scales of 3 to 30 feet (1-10 m) can trigger the transition of a surface fire to a crown fire. Wind variations on a similar scale can cause sudden and dramatic changes in fire behavior; in other words, fine-scale variations in winds can significantly influence fire growth at larger scales. While regional pressure drives air flow, surface topography (i.e., individual terrain features) and the relative position, shape, and size of vegeta- tion direct and channel that flow Figure 2—Wind field and associated fire growth prediction for the South Canyon Fire at the fine scale. While prognos- on July 6, 1996, based on FARSITE simulations using winds produced by the WindNinja tic models use most or all of the model (a) and by the WindWizard model (b). Winds used for the boundary condition elements mentioned above, they were west at 49 mph (79 km/h) measured at the Rifle, CO, remote automatic weather station at the time of the fire and according to witness statements. The dashed white line require detailed data at larger scales represents the fire perimeter prior to the cold front passage, and the solid black line is and across large spans of time to the fire perimeter that existed after the cold front passage. Pink dashed lines are firelines compute even short forecasts of constructed by crew. This perimeter was reconstructed from witness statements and other evidence. The orange solid area is the projected fire growth during the cold front event.
Volume 69 • No. 2 43 Shows promise as CFD type model, but not readily available or document - ed. Output file type unclear.
Fire Management Today 44 Accuracy in fire growth simulation analysts working on a laptop. Their m). But these are currently only improves when more detailed local utility is increased when supported research models not generally wind fields are included. The utility by good prognostic models. available for real-time application. of these winds to fire planning and In the meantime, current research management has been documented. Future Research efforts both within land manage- For example, Butler and others Directions ment agencies and the National (2006) and Stratton (2006) present Weather Service promise continued several cases in which WindNinja There are many areas of active improvement in existing wind- and WindWizard can be used in research related to winds and fire, modeling tools. prescribed fire planning, wildland and a number deal directly with fire accident investigations, fire wind models. Work is underway to There is ample evidence that newly management, firefighter and public develop easier methods whereby the developed wind simulation tools safety, and emissions monitor- output from prognostic models can can provide fire managers with ing. But even the simplest of wind be used to initialize the flow calcu- valuable information. Both diag- models requires experience: Weise lations in diagnostic models. This nostic and prognostic models have and others (2007) have shown that approach could provide users with their particular strengths, weak- the use of diagnostic models does the benefits of the numerous vari- nesses, and roles in supporting not guarantee accurate fire growth ables handled by prognostic models wildland fire management. Using simulations—an indication of the along with the computational speed them effectively requires identi- “art” required in wind modeling. and fine-scale resolution of diag- fication of the processes affecting nostic models. winds on a fire and the model(s) There is active debate in the sci- most appropriate for the situation entific community about whether Ongoing work seeks an under- so that the data and models avail- diagnostic model winds (such as standing of how wind moves able complement each other. Table from WindWizard or WindNinja) through forest canopies and how 1 summarizes the various wind are accurate or even the best infor- variations in vegetative cover cause modeling alternatives. mation available to fire behavior microscale changes in wind flow. analysts. This discussion is entirely Other research seeks to better The human component is never appropriate, as it is the key to understand how winds high above far from the technical component. motivating scientific research the ground might influence fire Once necessary data are identified, and likely will lead to improved spread. Updrafts and downdrafts accurate input is required from tools and methods for determin- caused by fire, topography, or solar lookouts, spotters, and observers. ing wind information. However, heating can carry winds from hun- After model execution, it is criti- for operational purposes, there dreds of feet above ground down to cal that winds from any computer is one fundamental question: do the surface; these winds could be model should be reviewed by an any of these models provide fire very different in speed or direction experienced, knowledgeable IMET. behavior analysts with better winds from those at the surface and so The output can then be passed to data than they currently get in a produce otherwise unexpected fire analysts and fire managers, who timeframe that helps them do their behavior. Research is still needed can use it to lay out effective strate- jobs? Compared to using single- to understand conditions under gies. This coordination of software point wind data to predict fire which upper atmospheric winds and decisionmakers is ultimately growth over the entire area of a are most likely to drop near the what makes wind and fire modeling fire, the cases described above show surface, how far they might drop, software valuable. quite clearly that fine-scale predic- and on what horizontal scale these tions from diagnostic models yield processes occur. Acknowledgements improvements in fire growth mod- We would like to thank Dr. Tara eling. WindWizard and WindNinja New research prognostic models Strand of the AirFire Team at the are currently the only models are in development and testing that Pacific Northwest Research Station readily available in North America include the capability to account for helpful conversations during capable of providing high-resolu- for some surface drag due to veg- the writing of this paper and Mr. tion wind data in a timely fashion etation and can simulate flows Jason Forthofer of the Missoula for operational purposes to fire at scales of 30-300 feet (10-100
Volume 69 • No. 2 45 Fire Science Laboratory for techni- C.; Pellicane, Patrick J.; Burns, Denver P.; Stratton, R.D. 2006. ‘Guidance on spatial cal advice. Draggan, Sidney, eds. Monitoring Science wildland fire analysis: models, tools, and Technology Symposium: Unifying and techniques.’ Gen. Tech. Rep. RMRS- Knowledge for Sustainability in the GTR-183. Fort Collins, CO: Forest References Western Hemisphere. 2004 September Service, Rocky Mountain Research 20-24; Denver, CO. Proceedings RMRS- Station. 15 p. Butler, B.W.; Bartlette, R.A.; Bradshaw, L.S.; P-42CD. Fort Collins, CO: Forest Service, Weise D.R.; Chen, S.; Riggan, P.J.; Fujioka, Cohen, J.D.; Andrews, P.L.; Putnam, T.; Rocky Mountain Research Station. 990 p. F.M.; Jones, C. 2007. Using high- Mangan, R.J. 1998. Fire behavior associ- Forthofer, J.M. 2007. Modeling wind in resolution weather data to predict fire ated with the 1994 South Canyon Fire on complex terrain for use in fire spread spread using the FARSITE simulator: Storm King Mountain. Res. Pap. RMRS- prediction (thesis). Fort Collins, CO: a case study in California chaparral. RP-9. Ogden, UT: Forest Service, Rocky Colorado State University. Proceedings, Seventh Symposium on Mountain Research Station. 82 p. Scire, J.S.; Robe, F.R.; Fernau, M.E.; Fire and Forest Meteorology, Bar Harbor, Butler, B.W.; Forthofer, J.M.; Finney, M.A.; Yamartino, R.J. 2000. A User’s Guide ME. Available at:
Fire Management Today 46 assessing Changes in Canopy Fuels and potential Fire behaVior FolloWing ponderosa pine restoration John Paul Roccaforte, Peter Z. Fulé, and W. Wallace Covington
In 1995, the Ecological Restoration Institute (ERI) at Northern Arizona The reduction in canopy fuels is visually University, the U.S. Department evident in the treated area, where surface fuels, of the Interior Bureau of Land Management (BLM), and the once dominated by forest floor and coarse Arizona Game & Fish Department woody debris, now consist of an abundant (AZGFD) began a collaborative herbaceous understory. effort to implement landscape- scale restoration treatments in a ponderosa pine ecosystem at Mt. • Compare the output from two fire Estimating Trumbull, located in northwestern behavior models; and Canopy Fuels Arizona. The primary goal of the • Use the comparisons to assess the Canopy fuels are critical inputs project was to restore forest struc- effectiveness of landscape-scale for models that predict crown fire ture and ecosystem processes with- restoration treatments on reduc- (Scott and Reinhardt 2002) but in the historical ranges of variabil- ing crown fire hazard. they are rarely measured directly. ity (Moore and others 1999) with There are various methods for an adaptive management approach. Using tree-ring analysis, we recon- estimating CFL and CBD, and the Other project objectives included structed a prefire-exclusion stand resulting estimates can vary widely. reducing fuel loads, disrupting structure for the year 1870. In 1996 Many methods rely on allometric fuel continuity, and reducing the and 1997, we installed permanent equations that estimate the mass of likelihood of stand-replacing crown monitoring plots and collected pre- foliage and fine twigs based on tree fires by implementing mechanical treatment forest structure data for diameter. We examined three com- thinning followed by prescribed fire estimating canopy fuel load (CFL) mon techniques that are based on (Moore and others 2003, Roccaforte and canopy bulk density (CBD) the following equations: and others 2008). The project also across the ~3,000 acre study area • Fulé and others’ (2001) allomet- aimed at providing research oppor- (fig. 1). Approximately half of the ric equations for foliage and fine tunities in a southwestern ponde- study area, a contiguous, densely- twigs of ponderosa pine developed rosa pine ecosystem. treed area, was left untreated and used as a control. By 2003, most of in Arizona, • Brown’s (1978) allometric equa- We initiated a study to examine the other half had received restora- tions for foliage and fine twigs of canopy fuels and potential fire tion treatments—thinned and/or ponderosa pine developed in the behavior during three time periods: burned with prescribed surface fire northern Rocky Mountains, and 1870 (prefire-exclusion), 1996/97 (Roccaforte and others [in press]). • Cruz and others’ (2003) stand- (pre-treatment), and 2003 (post- We collected post-treatment data scale equations based on tree treatment). Our goals were to: for each plot in the summer of density and basal area. • Compare three common canopy 2003. Stand structure data were fuel estimation approaches; used to derive CFL and CBD in the study area for all three time peri- Changes in ods. Those estimates were then fed Canopy Fuels John Paul Roccaforte is a research special- into two fire models to compare For all of the time-treatment com- ist, Peter Z. Fulé is the director of opera- historic fire behavior with potential binations, Brown’s (1978) equations tions, and W. Wallace Covington is the fire behavior on the control site director for the Ecological Restoration always produced the highest value Institute at Northern Arizona University in and the site that received restora- for average CFL, the Fulé and oth- Flagstaff, AZ. tion treatments.
Volume 69 • No. 2 47 els under scenarios of extremely dry conditions and a range of windspeeds to simulate the severe weather under which uncontrol- lable crown fires most commonly spread.
Changes in Potential Fire Behavior Initial modeling results for the three CBD levels showed that crown fire activity was correlated with CBD for both FlamMap and NEXUS. However, the FlamMap simulations were sensitive to CBD, showing little active crown fire for any of the time-treatment combinations when the low and intermediate CBD values were used. Therefore, we restricted our analysis to results from the two fire behavior models using the highest CBD values (i.e., from Cruz and Figure 1—A map of the study site (~3,000 acres) showing permanent plot locations (black others 2003). dots). The treated area is in the northern part of the study area; the control is in the southern part. FlamMap predicted that active crown fire would not occur within ers (2001) estimate was always low- Modeling Potential Fire the study area in 1870 even with est, and the Cruz and others (2003) Behavior 43 mph (70 km/h) windspeeds. By estimate always produced inter- We used two common fire behav- 1996/97, nearly 90 percent of the mediate values (fig. 2a). For CBD, landscape was classified as either Fulé’s equation again produced the ior models to predict potential fire behavior and evaluate treatment passive or active with 43 mph (70 lowest estimate, but Cruz’s estimate km/h) windspeeds. After treatment, was highest and Brown’s estimate effectiveness: • FlamMap, a spatially explicit the percent of the landscape in the was intermediate in all but the treated area susceptible to active treated area in 1870 (fig. 2b). model that assesses fuel hazards by simultaneously predicting fire crown fire was reduced from 46 percent to less than 5 percent com- Regardless of which equation was behavior for each individual pixel on the raster landscape (Stratton pared to the control, which showed used, CFL and CBD values were little change. relatively low over the entire 2004), and • NEXUS, another hazard model, study area in 1870, with dramatic NEXUS predicted that some active increases across the entire land- which uses plot- or stand-scale data to predict potential fire crown fire would occur within the scape by 1996/97. By 2003, treat- study area in 1870 with windspeeds ment lowered CFL and CBD by behavior (Scott 1999; Scott and Reinhardt 2001). greater than 31 mph (50 km/h), about 40-60 percent compared with with up to 17 percent of the land- slight increases in the control (fig. These fire behavior models are scape supporting active crown fire 2). The reduction in canopy fuels is when modeled with 43 mph (70 visually evident in the treated area, designed for assessing fuel hazards rather than fire growth; we selected km/h) windspeeds. NEXUS pre- where surface fuels, once dominat- dicted that 80 percent of the pre- ed by forest floor and coarse woody them because both are ideal for evaluating treatment effects on treatment landscape would support debris, now consist of an abundant active crown fire with 43 mph (70 herbaceous understory (fig. 3). fire behavior. We ran both mod-
Fire Management Today 48 (a) Canopy Fuel Load change in the control There are two key differences 0.75 over the same time peri- between FlamMap and NEXUS. Fulé and others (2001)
Brown (1978) od. NEXUS also predict- First, in FlamMap, the model Cruz and others (2003) ed a substantial increase inputs are interpolated or calculat- ) 2 0.5 in crowning index (the ed across the plot grid to produce windspeed required to output across the landscape, where-
CF L (lbs/ft 0.25 sustain active crown as in NEXUS, inputs are calculated fire) in the treated area for each plot and outputs are inter- (fig. 4). polated across the landscape. Thus, 0 Control Treated Control Treated Control Treated even though the fundamental fire 1870 1996/97 2003 behavior predictions are nearly the Year Model Comparisons same in the two models, the dif- (b) Canopy Bulk Density ferent approaches to modeling fire 0.02 The purpose of model- across the landscape lead to some- Fulé and others (2001) ing fire behavior at Mt.
) Brown (1978)
3 what different results. Second, only 0.015 Trumbull was to use Cruz and others (2003) NEXUS accounts for the occurrence the output as a way of of conditional crown fire, the situa- 0.01 comparing potential CBD (lbs/ft tion where passive crown fire is not fire behavior between predicted to occur due to a high 0.005 three time periods and canopy base height even though between the control and active crown fire could occur due 0 Control Treated Control Treated Control Treated treated areas following to high CBD (Scott and Reinhardt 1870 1996/97 2003 restoration treatments. 2001). Year One should always Figure 2—Although canopy fuel load (a) and canopy bulk interpret model output Although FlamMap and NEXUS dif- density (b) estimates were variable, all methods showed with caution, and these fer, predicted outcomes were con- low values in 1870, marked increases by 1996/97, and model runs were not substantial reduction in the treated area by 2003. sistent: under extreme drought and expected to accurately wind conditions, the proportion of km/h) windspeeds. By 2003, NEXUS predict the behavior of an actual the landscape susceptible to active predicted that active crown fire fire. Using the output from two crown fire decreased in the treated was reduced from 82 percent to 48 different fire models is a way of area. In contrast, the models show percent in the treated area with 43 validating each of their treatment little change in crown fire hazard mph (70 km/h) windspeeds with no comparisons. in the control over the same time period.
Figure 3—This before and after photo series illustrates the reduction in canopy fuels and consequent increase in herbaceous surface fuels. The upper photo was taken in 1996 prior to treatment; the lower photo was taken in 2003 after the area was thinned and burned. Arrows indicate the same trees for reference. Photos: John Paul Roccaforte, Ecological Restoration Institute, Northern Arizona University, Flagstaff, AZ 1996, 2003.
Volume 69 • No. 2 49 International Journal of Wildland Fire. 12: 39–50. Fulé, P.Z.; McHugh, C.; Heinlein, T.A.; Covington, W.W. 2001. Potential fire behavior is reduced following forest restoration treatments. In: Vance, R.K; Covington, W.W.; Edminster, C.B; Blake, J.B., eds. 2000. Proceedings of the Ponderosa Pine Ecosystems Restoration and Conservation: Steps Toward Stewardship Conference; 25-27 April; Flagstaff, AZ: Proceedings RMRS-P-22. Ogden, UT: Forest Service, Rocky Mountain Research Station: 28–35. Fulé, P.Z.; Verkamp, G.; Waltz, A.E.M.; Covington, W.W. 2002. Burning under old-growth ponderosa pines on lava soils. Fire Management Today. 62(3): 47–49. McGlone, C.M.; Springer, J.D.; Covington, W.W. 2009. Cheatgrass encroachment on a ponderosa pine forest ecological restoration project in northern Arizona. Ecological Restoration. 27: 37–46. Moore, K.; Davis, B.; Duck, T. 2003. Mt. Trumbull ponderosa pine ecosystem restoration project. In: Omi, P.N.; Joyce, L.N., eds. Proceedings of the Fire, Fuel Treatments, and Ecological Restoration Figure 4—Fire behavior modeling output from NEXUS shows that higher windspeeds are Conference. Proceedings RMRS-P-29. necessary to sustain active crown fire in the treated area in 2003; hence, active crown Fort Collins, CO: Forest Service, Rocky fire hazard was reduced following treatment. Alternatively, little change in potential fire Mountain Research Station. 117–132. behavior occurred in the control over the same time period. Moore, M.M.; Covington, W.W.; Fulé, P.Z. 1999. Reference conditions and ecological restoration: a southwestern Management The ERI, BLM, and AZGFD will ponderosa pine perspective. Ecological Implications continue to address these and other Applications. 9: 1266–1277. challenges in the future. Roccaforte, J.P.; Fulé, P.Z.; Covington, W.W. This study provided a quantitative 2008. Landscape-scale changes in canopy fuels and potential fire behavior follow- evaluation of the effectiveness of The Mt. Trumbull ecosystem will ing ponderosa pine restoration treat- landscape-scale restoration treat- never be “fireproofed.” Maintenance ments. International Journal of Wildland ments on canopy fuels and crown of the surface fire regime will be Fire. 17: 293–303. fire hazard for the Mt. Trumbull Roccaforte, J.P.; Fulé, P.Z.; Covington, W.W. vital to retaining open forest condi- In press. Monitoring landscape-scale pon- landscape. Although canopy fuel tions and relatively low crown fire derosa pine restoration treatment imple- estimates and fire behavior predic- hazard into the future. Although mentation and effectiveness. Restoration tions varied depending on which Ecology. the Mt. Trumbull ponderosa pine Scott, J.H.; Reinhardt, E.D. 2002. models were used, all modeling ecosystem is not yet “restored,” Estimating canopy fuels in conifer for- scenarios resulted in substantially restoration treatments have been ests. Fire Management Today. 62(4): lowered canopy fuels and crown 45–50. successful at substantially reduc- Scott, J.H. 1999. NEXUS: A system for fire hazard in the treated area. This ing crown fire hazard and creating assessing crown fire hazard. Fire suggests that restoration treat- more sustainable forest conditions. Management Notes. 59: 20–24. ments were an effective manage- Scott, J.H.; Reinhardt, E.D. 2001. Assessing ment strategy. crown fire potential by linking models of References surface and crown fire behavior. Research Paper RMRS-RP-29. Fort Collins, CO: Brown, J.K. 1978. Weight and density of Forest Service, Rocky Mountain Research It should be noted that treatments crowns of Rocky Mountain conifers. Station. 59 p. have also resulted in the loss of Res. Pap. INT–197. Ogden, UT: Forest Stratton, R.D. 2004. Assessing the effective- Service, Intermountain Forest and Range some old trees from prescribed ness of landscape fuel treatments on fire Experiment Station. 56 p. fire activities (Fulé and others growth and behavior. Journal of Forestry. Cruz, M.G.; Alexander, M.E.; Wakimoto, 102: 32–40. 2002) and the spread of cheatgrass R.H. 2003. Assessing canopy fuel stra- (Bromus tectorum), an invasive tum characteristics in crown fire prone species (McGlone and others 2009). fuel types of western North America.
Fire Management Today 50 Contributors Wanted
We need your fire-related articles and photographs for Fire Management Today! Feature articles should be up to about 2,000 words in length. We also need short items of up to 200 words. Subjects of articles published in Fire Management Today include:
Aviation Incident management Communication Information management (including systems) Cooperation Personnel Ecosystem management Planning (including Equipment/Technology budgeting) Fire behavior Preparedness Fire ecology Prevention/Education Fire effects Safety Fire history Suppression Fire science Training Fire use (including prescribed fire) Weather Fuels management Wildland-urban interface Firefighting experiences
Submitting Articles: Send electronic files by e-mail or traditional or express mail to: USDA Forest Service Attn: Karen Mora, Managing Editor 2150 Centre Avenue Building A, Suite 300 Fort Collins, CO 80526 Tel. 970-295-5715 Fax 970-295-5885 e-mail:
Contributor guidelines are posted at
Volume 69 • No. 2 51 To fax your orders: 202-512-2104 To phone your orders: 202-512-1800 or 1-866-512-1800 For subscription cost and to Order on Line: http://bookstore.gpo.gov
Mail To: U.S. Government Printing Office - New Orders P.O. Box 979050 St. Louis, MO 63197-9000
Fire Management Today 52