The RAVG program provides assessments of vegetation conditions (burn severity) following large wildland fires on forested National Forest System (NFS) lands. Staff at the USGS Earth Resources Observation and Sciences (EROS) Center conduct similar assessments for select fires on Department of the Interior (DOI) lands.
Scope
Analysis under the Forest Service RAVG program is routinely conducted for wildland fires that include at least 1,000 acres of forested National Forest System (NFS) lands. (Since 2016, the threshold has been 500 acres for fires in the Eastern and Southern NFS regions.) Although the number of smaller fires is much higher, these large fires account for the majority of the area burned.
The RAVG process was developed and first implemented in NFS Region 5 (California). It was adapted by GTAC for nationwide implementation in 2007. Under current protocols, RAVG products for the majority of qualified fires are made available within approximately 45 days after fire containment, provided that suitable imagery is available. A common exception is for fires that burn into late fall or occur in deciduous cover in leaf-off conditions. These fires are typically mapped the following summer, near the peak of greenness. Note: Post-fire data based on imagery acquired within a few weeks after the fire are called "initial assessments"; those based on imagery acquired near the following peak of green are termed "extended assessments".
Process
RAVG products are generated using a two-date change detection process and regression equations that relate imagery-derived burn severity indices to field-based burn measures. RAVG analysis starts with a pair of moderate-resolution multi-spectral images (e.g., Landsat imagery), one from before the fire and one from after the fire. The image pair is used to derive a burn index such as the Relative Differenced Normalized Burn Ratio (RdNBR, Miller and Thode 2007), which is sensitive to vegetation mortality resulting from the wildfire event.
The RAVG program relies primarily on imagery from Landsat sensors (Landsat 8 and 9 Operational Land Imager (OLI) and, previously, Landsat 7 Enhanced Thematic Mapper (ETM+) and Landsat 5 Thematic Mapper (TM)) and from the European Space Agency's Sentinel-2 MultiSpectral Instrument (MSI). Other multi-spectral sensors can be used provided they have sufficient resolution and the necessary spectral bands. The preferred bands are the near infrared (NIR) and short-wave infrared (SWIR, around 2.2 micrometers), which are ideal for detecting the change from healthy green vegetation to dead vegetation, bare soil, and ash. The two bands are used to calculate the Normalized Burn Ratio (NBR) for each image, and multiple "differenced" indices, such as the Differenced NBR (dNBR, the change in NBR from the pre-fire image to the post-fire image), the Relative dNBR (RdNBR, a modified dNBR that accounts for pre-fire vegetation density), and, for some fires, the Relative Burn Ratio (RBR, another modified dNBR).
Models are used to determine three burn measures--percent change (loss) in live basal area (BA), percent change in canopy cover (CC), and the standardized Composite Burn Index (CBI)--from the imagery-based indices. The models were derived from relationships between field data and contemporaneous imagery-based indices. For the original models--the default, national models--field data consisted of CBI and individual tree data (tree mortality data by species and size class) collected from many fires in the Sierra Nevada and northern California and the index (RdNBR) was derived from Landsat imagery. In recent years, regional models were developed from field data collected on burned areas in the Southwest (Arizona and New Mexico) and the Northwest (Oregon, Washington, northern Idaho, western Montana, and northwestern Wyoming). Beginning in 2024, the Southwest models are applied by default to fires in Arizona and New Mexico; the Northwest models are applied to extended assessments in the Northwest. As older fires are reprocessed, the analyst may choose to apply the regionally appropriate model. When one of the regional models is applied to a RAVG dataset for a given fire, it will be noted in the associated metadata.
Other standard products are derived from the burn metrics. Classified (thematic) versions of each measure are created by binning the continuous values. Summary tables and maps are produced by combining the burn data with a vegetation layer derived from the LANDFIRE Existing Vegetation Type (EVT) layer and administrative data (NFS ownership and designated wilderness areas).
Products
RAVG products include the following for each dataset (nominally, each fire):
- Geospatial products, including imagery and derived data
- Satellite imagery (Landsat, Sentinel-2 or similar)
- Pre-fire scene (spatial subset)
- Post-fire scene (spatial subset)
- Normalized burn ratio and related indices
- Burn severity measures derived from pre- to post-fire change
- Percent basal area loss (continuous and 4- and 7-class classified versions)
- Percent canopy cover loss (continuous and 5-class classified versions)
- Composite burn index (continuous and 4-class classified versions)
- EVT-based vegetation classes
- Satellite imagery (Landsat, Sentinel-2 or similar)
- User-friendly visualizations
- Summary table of affected area by vegetation class, ownership class, and burn class
- Ancillary data
- Fire perimeter (shapefile)
- Masked areas, if any (shapefile)
- Metadata (text)
Additional details about each product follow. For the national model, equations for derived raster data are included in the metadata for each fire.
- Burn severity measures. The primary geospatial products are raster datasets (TIFF format) representing burn measures.
- Percent basal area (BA) loss represents the change in live basal area relative to the pre-fire condition. For the continuous version, values range from 0 to 100%. The 7-class version (BA-7) includes the following classes:
- Class 1: 0%
- Class 2: >0% - <10%
- Class 3: 10% - <25%
- Class 4: 25% - <50%
- Class 5: 50% - <75%
- Class 6: 75% - <90%
- Class 7: 90% - 100%
- A 4-class version (BA-4) is created by recoding the BA-7 classes:
- Class 1: 0%
- Class 2: >0% - <25%
- Class 3: 25% - <75%
- Class 4: 75% - 100%
- Note that a different recoding is used for the four BA loss classes in the PDF maps and tabular summaries:
- Class 1: 0% - <25%
- Class 2: 25% - <50%
- Class 3: 50% - <75%
- Class 4: 75% - 100%
- Percent canopy cover (CC) loss represents the change in canopy cover relative to the pre-fire condition. For the continuous version, values range from 0 to 100%. The 5-class version (CC-5) consists of the following classes:
- Class 1: 0%
- Class 2: >0% - <25%
- Class 3: 25% - <50%
- Class 4: 50% - <75%
- Class 5: 75% - 100%
- The Composite Burn Index (CBI) is a standardized fire severity rating based on a composite of effects to the understory vegetation (grass, shrub layers), midstory trees, and overstory trees. Values range from 0 (unchanged) to 3 (highest severity). The classified product included in the RAVG dataset has the following four classes:
- Note: In all of the burn condition raster datasets, areas that are masked due to clouds, cloud shadows, smoke, active fire, or other reasons, are indicated with a negative value (-9 or -9999) for continuous data or 9 for classified data.
- Percent basal area (BA) loss represents the change in live basal area relative to the pre-fire condition. For the continuous version, values range from 0 to 100%. The 7-class version (BA-7) includes the following classes:
- Other raster data include a subset of the multi-spectral imagery (e.g., pre- and post-fire Landsat imagery) used for the assessment, the associated indices (pre-fire NBR, post-fire NBR, and one or more of dNBR, RdNBR, and RBR), the RAVG vegetation classes, and a RAVG tree/non-tree dataset.
- The burn boundary (perimeter) and masked areas, if any, are supplied in vector form (shapefiles).
- A map (PDF) of the burned area portrays the post-fire imagery and classified burn severity.
- A Google Earth file (KMZ) allows for interactive exploration of the perimeter, classified burn severity and imagery in the context of high-resolution imagery and other data available in the Google Earth application.
- A spatial summary table (Excel) lists affected area (acres) by vegetation class, ownership class and burn severity class. Slope data (percent rise) are also included in the "raw" data sheets.
Applications
RAVG products are intended primarily for use in assessing fire-related reforestation needs. RAVG data help staff on local units prioritize areas for further assessment and support reforestation funding requests and decisions. They facilitate post-fire vegetation management decision-making by reducing planning and implementation costs. RAVG data also serve a variety of related Agency objectives, such as wildlife habitat analysis and salvage harvest planning.
Note: The RAVG regression equations are not calibrated to non-forest vegetation. RAVG burn severity measures should be interpreted in light of existing (pre-fire) vegetation.
Related programs
RAVG is one of three post-fire programs at GTAC. The others are the Burned Area Emergency Response (BAER) Imagery Support program and the Monitoring Trends in Burn Severity (MTBS) program. Although the three programs have many similarities, they differ in their methods and protocols, as well as in their intended audiences.
- The BAER Imagery Support program supports BAER teams performing emergency assessments and soil stabilization treatments immediately following select wildfires. It is a cooperative effort between GTAC and the U.S. Geological Survey (USGS) Earth Resources Observation and Science (EROS) Center. Its objective is to provide rapid delivery of satellite imagery, Burned Area Reflectance Classification (BARC) data, and related geospatial data to Forest Service and Department of the Interior (DOI) BAER teams. The BAER is a dNBR-based map that helps the BAER team prioritize its work and serves as the primary input to the final soil burn severity map produced by the BAER team. BAER Imagery Support is considered an emergency assessment; analysts typically deliver the program’s products to BAER teams within hours of receiving useable post-fire imagery. The BAER program’s first priority is to address emergency soil stabilization needs to prevent further damage to life, property, and natural and cultural resources.
- MTBS is a multi-year project designed to consistently map the burn severity and perimeters of large fires across all lands of the United States since 1984. The data generated by MTBS are used to identify national trends in burn severity, providing information necessary to monitor the effectiveness and effects of the National Fire Plan (NFP) and the Healthy Forests Restoration Act. The Wildland Fire Leadership Council (WFLC), a multi-agency oversight group responsible for implementing and coordinating the NFP and Federal Wildland Fire Management Policy and Program Review, sponsors MTBS. The project is conducted through a partnership between EROS and GTAC. The MTBS project maps burn severity using the dNBR, with the RdNBR used to help determine appropriate burn severity thresholds. Fires that occur in higher biomass vegetation (forests and dense shrublands) are typically mapped as extended assessments, allowing the products to capture the effects of delayed vegetation mortality.
Changes
Various aspects of RAVG scope, processes, and products have changed over time and are subject to change in the future. Users are encouraged to review the metadata included in each RAVG data bundle for descriptions of included data.