GOES-R ABI Active Fire Algorithm Validation

The GOES-R/ABI Active Fire algorithm will provide high frequency fire detection and characterization data for the Western Hemisphere at a nominal spatial resolution of 2km. The validation strategy for the fire algorithm builds on the use of near-coincident high spatial resolution airborne and spaceborne reference data sets to quantify sub-pixel fire activity imaged by ABI. Candidate airborne reference data sets include multispectral linescanners and/or focal plane array instruments operating at wavelengths in the VNIR-SWIR and/or MIR-LWIR spectrum, and at saturation temperatures above 600K. The spatial resolution of these instruments varies typically between 5-50m depending on the aircraft's height above ground. Candidate spaceborne reference data sets may be divided into two groups: Landsat-class and moderate spatial resolution (approx. 350m) data covering the mid-infrared spectrum. Both airborne and spaceborne reference data sets will provide a fire classification product (binary mask identifying fire-affected pixels) to validate fire detections, whereas complementary fire retrievals (e.g., fire radiative power estimates) may only be derived from a small subset of those data due to saturation issues and other sensor limitations. Here we discuss these alternatives and describe the proposed science algorithms to be used to generate the corresponding fire reference data and to support the scaling-up analyses for the validation of the ABI fire product. We present the latest test results using GOES-Imager and fire reference data derived from NASA's AMS and MASTER airborne linescanners flown over wildfires in the Western United States.