Sub-pixel fractional area of wildfires from satellite measurements: retrieval, validation, and potential applications

Detailed validations of retrieved fire area fractions are now possible in the western United States via the multispectral, high-resolution data (3-50 meters) obtained from the Autonomous Modular Sensor (AMS), flown aboard an Unmanned Airborne Vehicle (UAV). Using this validation tool, the potential for obtaining sub-pixel fire (hot spot) information for fire pixels detected by the MODerate Resolution Imaging Spectroradiometer (MODIS) is investigated. A two-component model (Dozier method) for retrieving sub-pixel fire area fraction and temperature has been available since 1981. However, modifications are made to the retrieval to account for atmospheric effects by implementing output from a radiative transfer model at 3.96 and 11 µm (MODIS fire detection channels). Using several fire events in the Western United states, a comparison between the retrieved fire fractions for MODIS and the AMS flight scans is preformed. Specific results from a large fire event in southern California suggest that the retrieval may be possible for fires covering a fractional area greater than 0.003 (corresponds to a 3000 m² fire within a 1 km pixel). The pixel background temperatures for each channel are also investigated and slight deviations are found to dramatically affect the retrieved fire area fraction. If successful, this sub-pixel retrieval will not only provide a valuable step for improving emissions estimates and plume height forecasts, it will also allow the meteorological effects on fire intensity (e.g., fire radiative power normalized to the fire size instead of pixel area) to be investigated. From an operational perspective, this may prove valuable for determining the size of a fire front, spread rate, and fire intensity. Future satellite sensors, such as the VIIRS instrument on the JPSS and the ABI on GOES-R, could also benefit from similar sub-pixel retrievals.