Using the VIIRS Day/Night Band to Improve Nocturnal Fire Detection

Abstract:

Wildfires are a serious threat to life and property that has exacted greater costs in recent years, despite improving warning systems. In addition to local impacts, the smoke produced by wildfires and biomass burning can travel thousands of kilometers downwind, impacting visibility and health far from the source. Using the recently launched Visible Infrared Imaging Radiometer Suite (VIIRS) aboard the Suomi-NPP satellite, this study investigates the adjustment of fire pixel selection criteria to include visible light signatures at night, allowing for greatly improved detection of smaller and cooler fires from satellite observations. VIIRS scenes with coincident Advanced Spaceborne Thermal Emission and Reflection (ASTER) overpasses are examined by applying the operational VIIRS fire product algorithm, and including a modified “candidate fire pixel selection” approach, which lowers the 4 µm brightness temperature thresholds but includes a minimum DNB radiance. This approach is applied while leaving the contextual tests unchanged; to be flagged as fires, fire pixels must pass the existing operational tests that compare them to background temperatures. While the detection of small agricultural fires in the Central United States remain problematic because of the coarse spatial resolutions of the 750 meter (M)oderate resolution bands, a large increase in the number of detected fire pixels is observed with small non-agricultural wildfires. Quantitative use of the DNB to improve detection of these smaller fires could lead to reduced warning and response times as well as provide more accurate quantification of biomass burning emissions at night.