Skip Navigation Link



Wildfire Automated Biomass Burning Algorithm (WF_ABBA)

The WF-ABBA product builds on the original work of Prins and Menzel [1992] to detect satellite image thermal anomalies typically associated with actively burning vegetation fires. The algorithm uses a dynamic thresholding-contextual approach primarily driven by the mid-wave (3.9 ?m) and long-wave (11.2 ?m) infrared data from the NOAA Geostationary Operational Environmental Satellite series. The WF-ABBA product provides routine active fire detection information, in addition to sub-pixel fire size and mean temperature estimates using the bi-spectral method of Dozier [1981], and fire radiative power (FRP) retrievals using the radiance method of Wooster et al. [2003]. The product has been generated operationally at NOAA/NESDIS since the early 2000s where it is used in the Hazard Mapping System (HMS) as well as other applications relating to biomass burning emissions. Data validation studies have been implemented for the WF-ABBA fire data over areas in North and South America [Prins et al., 1998; Schroeder et al., 2008a/b, 2010]. The currently available data record starts in 1997 with GOES-8 and extends until present time. Some historical data is available here.

The WF-ABBA algorithm has been implemented and further customized for the new generation GOES-R series, and renamed to Fire Detection and Characterization (FDC) (users may also find the alternative naming convention: Fire/Hot Spot Characterization). The higher saturation temperature (400K) of the mid-wave infrared (channel 7) data makes the GOES-R series especially suited for active fire detection applications. Similar to the WF-ABBA product, the FDC data provide fire detection and sub-pixel fire characterization. For a detailed description of the algorithm implemented for GOES-R series, users are referred to the Algorithm Theoretical Basis Document. Currently, the FDC product has a provisional maturity status. Users are highly encouraged to consult the FDC product status information and look for the associated “Provisional Readme” document.

The WF-ABBA/FDC text (ASCII) data files are saved in comma-delimited format compatible with Geographic Information Systems (GIS). Each file provides information describing the geographic location (pair of latitude/longitude coordinates) of individual fire pixels along with basic attributes describing: (i) the observed 4?m and 11?m brightness temperature [K], (ii) sub-pixel fire characterization (fire size [km2], temperature [K], and FRP [MW]), (iii) ecosystem classification (see table of definitions derived from the Global Land Cover Characterization (GLCC) data base), and (iv) the fire pixel classification flag. The latter is divided into six sub-groups as follows:

Fire pixel classification
Fire Pixel Flag Description Sub-Pixel Fire Retrieval Availability
Size Temperature FRP
0 Fully-processed
1 Saturated pixel      
2 Cloud-contaminated      
3 High probability    
4 Medium probability    
5 Low probability      

*Low probability fire pixels are discarded from the data base to minimize false alarm proliferation

GOES-16/Advanced Baseline Imager (ABI) Fire Detection and Characterization (FDC) product animation describing the 15-min full-disk data over a 24h period on 01 June 2018 (Julian day 152). Primary pixel classification categories are colored as follows: green = land; blue = water; white = clouds; red = fire; black = block-out zones (circular areas of glint and high solar elevation) and unclassified data.

Read more about important data quality considerations

For questions concerning the WF-ABBA/FDC product, you may contact the fire team at SPSD or the algorithm developer.