Fires ignited by people or through natural causes have interacted over time with ecosystems. These fires alter fire regimes and pose great threats to biodiversity, sustainability, and public and firefighter safety.

The Nature Conservancy found that at least 43% of priority conservation area identified to date in the United States is moderately to severely at risk of significant degradation due to fire exclusion or the threat of unnaturally severe wildfires. Large fires not only wreak havoc on ecosystem health and endangered firefighter safety, they also cost millions of dollars to control. As a result, fire departments, legislators, land management agencies and partners need to employ efficient, cost-effective and scientifically credible methods to prioritize areas for restoration treatments.

The National Fire Plan, together with the 10-year Cohesive Implementation Strategy, engages states and local communities in a coordinated effort using a variety of fuel reduction treatments. In the short term, the focus is on areas near communities and interface areas where hazardous fuel reduction will reduce future fire risk and suppression costs. However, there remains a need to develop sound tools to prioritize limited resources between landscapes, within and across regions.

It's difficult to accurately and consistently quantify fire regime conditions because succession, fuel accumulation, and fire exclusion and climate change are highly variable. However, each time an extreme fire year makes its mark, there's a demand for data to respond to questions and to support pressing multi-scale fire management and prioritization decisions. Scientists must find effective fire regime condition assessment approaches that balance the needs for scientific credibility, spatial continuity and quick delivery with those of data availability and analysis time constraints.

The USDA Forest Service recently provided national-level, coarse-resolution data on the degree and nature of departure of current vegetation and fuels from historical conditions. This fire regime condition class data represents a significant leap forward in the integrating, classifying and mapping of biophysical, vegetation and fuel-characteristic data to prioritize resource allocation for restoration, fuels treatment and conservation.

CLASS SYSTEM

Fire regime condition class is an index used for allocation of fire funding and resources, prioritization of fuels and restoration treatments, and evaluation of the successes and failures of historical wildfire management activity. Fire regime and associated national FRCC mapping have provided key spatial data for development of the Forest Service and USDI cohesive strategies for restoration of fire-adapted ecosystems and for addressing the goals and objectives of the National Fire Plan. Such assessments provide core data used in prioritizing restoration and conservation, and the FRCC concept is readily being adopted by Congress and land management decision-makers as a useful tool to account for the success of hazardous fuels and ecosystem restoration projects nation-wide. However, the current coarse-scale data has been misused for regional- and project-level prioritization and planning.

Current FRCC data only addresses prioritization between large regions and groups of states, not between individual projects. New field FRCC procedures support project-by-project assessment, but don't provide the spatial data and models necessary for nationwide assessments. Finer-scale FRCC data is being developed through the LANDFIRE project, www.landfire.gov, using remote sensing and gradient modeling, but it's currently in a prototype stage and is not yet charged with analysis of the contiguous lower 48 states. Rapid mid-scale FRCC assessments are needed to bridge the gap between the available but limited-use coarse data and the fine-scale data that won't be widely available for five to 10 years.

PROJECT OBJECTIVES

A rapid mid-scale process provides spatial fire regime condition class data at resolutions appropriate for multi-area planning, assessment of potential long-term effects of alternative conservation strategies, prioritization of projects, development of fire management plans, and revision and amendment of agency forest and resource land management plans. It uses existing spatial and remotely sensed data and quantitative state-transition models to provide FRCC maps by vegetation type at 30-square-meter pixel resolution.

This data will help planners coordinate the tasks of conservation, restoration and hazard reduction across multi-ownership landscapes when combined with information on other prioritization and management criteria, such as housing density, sensitive species, fire occurrence, transportation networks and management constraints.

The Rapid FRCC Assessment process was developed to directly address key priorities for the Forest Service, DOI and TNC, including those for forest and rangeland health, the National Fire Plan, the Cohesive Implementation Strategy, and biodiversity conservation. More specifically, the rapid assessment process was designed to:

• Assess fuel and vegetation-fuel conditions across public and private lands.
• Prioritize hazardous fuels reduction across public and private lands where the negative effects of wildland fire to communities, ecosystems and biodiversity health are greatest.
• Provide a framework for development of biodiversity conservation strategies.
• Restore healthy, diverse and resilient ecological systems to minimize uncharacteristically severe fires on a priority watershed basis through long-term restoration.
• Promote the development and use of the best available science.
• Monitor projects and facilitate multi-party adaptive implementation.

The prototype process was applied using available data from northern New Mexico. In this region, public land managers, tribes and The Nature Conservancy are working to restore altered fire regimes, particularly in the Jemez Mountains, where in May 2000 the Cerro Grande Fire caused 18,000 residents to be evacuated, cost about $1 billion and burned 48,000 acres.

The Rapid FRCC Assessment process is designed to use the most recent data available, calculating departure of current information from reference conditions. Mapping objectives for accurate fine-scale (40 hectares or less) pixel or polygon spatial typing require input layers of high spatial accuracy and substantial ground-truthing, or testing of geophysical anomalies. In contrast, mapping objectives to achieve accurate composition of types within a much larger polygon can be achieved with relatively coarse spatial inputs. In addition, accurate relationships can be mapped using low-resolution data through use of relative classifications such as “low,” “moderate” or “high” risks and non-species — specific map legends (such as early, mid-closed, mid-open, late-open and late-closed forest stages). Assessment steps include:

1. Identify biophysical types and model reference conditions.

2. Map vegetation types using existing spatial data.

3. Classify and map seral or withering/structural stages and uncharacteristic types.

4. Calculate and map departure in fuels and fire frequency/severity.

DYNAMIC DEVELOPMENT

Immediate availability of interim continuous spatial FRCC and associated data will accelerate coordination of the tasks of restoration and fire hazard reduction across multi-ownership landscapes.

1. Identify biophysical types and model reference conditions. Potential natural vegetation types, or PNVTS, are one type of biophysical classification based on plant species that are indicators of the natural disturbance regime, site climate and soil relationships. Biophysical characteristics that to a large extent control fire regimes are reflected in the distribution of vegetation. PNVTS are the foundation for stratification of reference and vegetation-fuel conditions, the development of reference models and calculation of departures between reference and current conditions.

   Researchers used potential natural vegetation types classified by the Rocky Mountain Research Station Fire Modeling Institute as a foundation. Quantitative state-transition models for each vegetation type found in the western United States were developed in conjunction with the Project Scale FRCC project using the Vegetation Dynamics Development Tool. In the northern New Mexico prototype area, types included pine forest, pine — Douglas fir, southwest mixed conifer, spruce-fir, juniper-piñon, sagebrush, desert shrub, shinnery, plains grassland, desert grassland, and alpine meadows/barren.

   For each vegetation type, literature reviews and expert input were used to estimate successional transition times, fire frequency and severity, and chances of disturbance between a relatively simple set of historical structural stages. In most cases, structural stages were identified as early seral, mid-seral open, mid-seral closed, late-seral open, and late-seral closed, or a subset thereof. This simple classification was consistent with mid-scale spatial data available for structure and composition. VDDT models were then put in parameters with fire disturbance probabilities and run for 500 to 1,000 years, or until PNVT stage composition stabilized.

2. Map PNVTs using existing spatial data. The Rapid FRCC Assessment process requires the use of available spatial data on current land cover, vegetation and seral/structural stage to refine spatial data layers to mid-scale resolutions. Mapped PNVTS must coincide with the models, fire regime characteristics and reference conditions developed in step 1. Lands converted to other uses, such as agriculture or urban development, can be classified as such, or a determination of the PNVT prior to conversion can be made.

   For the prototype area, we obtained current mid-scale resolution vegetation cover for New Mexico from the U.S. Geological Survey's Gap Analysis Program and Land Use/Land Cover data, and cross-walked them with data from the Rocky Mountain Research Station's FMI. Classification of GAP and Land Use/Land Cover types by PNVTS allowed for refinement and cross-checking of PNVT data to 30-square-meter resolution. The resulting vegetation layer remains to be peer-reviewed and validated in the field.

   ELEVATED INFO

   Consistent, science-based measures of opportunities and risks across all land ownerships are a prerequisite for successful collaborative, multi-partner watershed-scale fire planning.

3. Classify and map seral or withering/structural stages and uncharacteristic types. Landsat Thematic Mapper satellite imagery can be processed at 30-meter spatial resolution to develop a preliminary seral/structural stage map for vegetation in the assessment area. Seral/structural stages must coincide with classes used to model reference conditions in step 1. For fire regime condition classification purposes, grasslands and shrublands are generally classified based on the degree of shrub or tree encroachment.

   Classification of vegetation structure involves using thematic stratification, unsupervised classification techniques, spatial modeling and manual editing. Any existing GIS data pertaining to current seral stage, size class, structure or any other pertinent vegetation characterization is evaluated and used to map structure where appropriate. For the remainder of the assessment area where existing GIS data is unavailable, an unsupervised classification of the Landsat imagery results in spectral classes that can be evaluated using aerial imagery, field-based plot data, or any other available ancillary data to determine the relationship between the spectral reflectance characteristics from the imagery and current structure/seral stages. Spectral classes are then labeled as “early,” “mid,” or “late” seral stage. If available, other ancillary GIS data can be used to further refine. These models may include the use of elevation/aspect zones and current vegetation-fuel types to further stratify the spectral classes for more accurate labeling of structure.

   Also, for areas exhibiting spectral anomalies or known errors that can't be efficiently and effectively corrected through further automated image-processing techniques, manual editing can enhance the thematic accuracy of the final structure classification. For the northern New Mexico prototype, we used existing USGS Land Use/Land Cover classification data to stratify the landscape into forest, shrub, water and other categories. An unsupervised classification of the Landsat imagery resulted in spectral classes that that were labeled as early, mid, or late seral stage based on texture and information on land-use history, such as the known year of large fires and expected structural classes at the time of the imagery. Young stands of regenerating trees or stands recently affected by fire or mechanical manipulation present a unique spectral characteristic not often confused with more mature mid- and late-seral stage forests.

   Similarly, much older, mature forests with typically multi-storied structure and the presence of large trees also possess unique spectral reflective properties that distinguish them from younger forests. Uncharacteristic types, such as areas converted to invasive species that represent conditions not found in reference models, also should be identified and classified as “uncharacteristic.” Future work will use existing ancillary GIS data sets to refine the classification of structure and uncharacteristic communities.

4. Calculate and map departure in fuels and fire frequency/severity. The departure in vegetation or fuels and fire frequency and severity is calculated by comparing reference seral/structural stage compositions and fire frequency/severity by PNVT to current conditions. Combined vegetation and fire regime dissimilarities ranging from 0-33% are FRCC 1, “intact” or unaltered. Departures ranging from 34-66% and 67-100% are classified as FRCC 2, “moderate,” or FRCC 3, “high,” departure, respectively. FRCC then can be mapped by PNVT to provide mid-scale information on conditions regionally.

Other criteria such as housing density, community proximity to natural areas, sensitive species, biodiversity value, risk of erosion or mass wasting, natural or human-caused fire occurrence, degree of landscape fragmentation, or road density may be used in conjunction with FRCC to develop priorities for management. Future development of the New Mexico prototype will incorporate criteria important for regional and national project prioritization and planning, as determined by decision-makers at multiple scales.

DISCUSSION AND CONCLUSIONS

This Rapid FRCC Assessment process can provide critical ecological data for use in prioritization of fire use, fuel restoration and maintenance projects, development of multi-agency fire management plans, development of conservation area strategies, tracking success of restoration strategies, and revision and amendment of forest and resource land management plans. Specifically, this assessment process can:

1. Provide a mid-scale, intermediate intensity assessment of fire regime conditions.
2. Provide interim data that can be used for national or regional prioritization and planning.
3. Provide input to fine-scale, high-accuracy, longer-term fire regime assessment projects.
4. Integrate existing ecological spatial data and expert input across all ownerships.

The development of the Rapid FRCC Assessment process and New Mexico prototype took approximately six months and could be applied easily to other geographic areas. The greatest challenges for future application will be development of reference conditions, especially for ecosystems where little is known of natural fire regimes or vegetation mosaics. Great care needs to be taken in developing and using reference conditions. Historical variability is a complex result of natural and human-induced change. Reference conditions are extremely useful as indicators of ecosystem function and sustainability, but do not necessarily represent desired future conditions, or sustainable conditions under current climate, land use or managerial constraints.

Certainly, as in any mapping effort using remotely sensed data, some errors will result in less than perfect classification results. It's anticipated that the greatest errors occurring from this type of analysis will be errors of omission from the late seral stage class in low density forest stands of 40% canopy cover or less. In these circumstances, reflectance from the existing forest canopy is often overwhelmed by other lower stature ground cover or the sparsely vegetated ground itself. Coupled with a forest density classification, current vegetation data, potential natural vegetation regimes and empirically derived fire regime dynamics models, the forest structure classification will facilitate the broad assessment of fire regime conditions.

The nature of Rapid FRCC Assessments represents a trade-off between using the best available consistent data that's less than one year old and using highly accurate data that's not currently available at large spatial scales. This trade-off necessitates the use of peer review, and model and field validation to the extent necessary to ensure that data are as robust as possible.

Over time, each assessment step can be adjusted to enhance logic and potential accuracy based on review. Overall, this methodology will provide a reliable, consistent characterization of FRCC across regions or states in a relatively short period of time. Errors will be consistent and often quantifiable with moderate analysis of results, and this process will result in a consistent, effective and efficient representation of FRCC.

Ayn J. Shlisky is a landscape ecologist for The Nature Conservancy Global Fire Initiative in Boulder, Colo. She currently leads TNC's role in the collaborative LANDFIRE project, in partnership with the Department of the Interior and USDA Forest Service. Her expertise lies in integrating landscape, ecoregional and national level applied fire ecology, ecological modeling, succession, disturbance ecology, and forest and range ecology and management. Prior to joining TNC, Shilsky worked for the Forest Service for 12 years as an ecologist, range conservationist and forester. She has a bachelor's degree in forest management, a master's degree in range management and a Ph.D. in ecosystem sciences from the University of California — Berkeley.

Wendel J. Hann is a landscape fire ecologist for the USDA Forest Service, Fire and Aviation Management at the Gila National Forest in Silver City, N.M. His primary responsibilities are in national ecosystem and fire risk assessment, technology development and transfer in fire and landscape ecology, and in land management planning and implementation. Hann has a bachelor's degree in range and wildlife management and a master's degree in forest and watershed science from Washington State University — Pullman; and a Ph.D. in Forest, Wildlife, and Range Ecology from the University of Idaho — Moscow.