Improving Nocturnal Fire Detection With the VIIRS Day–Night Band
| Autor: | L. Ellison, Jun Wang, Charles Ichoku, Edward J. Hyer, Thomas N. Polivka |
|---|---|
| Rok vydání: | 2016 |
| Předmět: |
Visible Infrared Imaging Radiometer Suite
010504 meteorology & atmospheric sciences Pixel Meteorology Fire detection 0211 other engineering and technologies 02 engineering and technology 01 natural sciences Subpixel rendering Brightness temperature Radiance General Earth and Planetary Sciences Radiometry Environmental science Electrical and Electronic Engineering Image resolution 021101 geological & geomatics engineering 0105 earth and related environmental sciences Remote sensing |
| Zdroj: | IEEE Transactions on Geoscience and Remote Sensing. 54:5503-5519 |
| ISSN: | 1558-0644 0196-2892 |
| DOI: | 10.1109/tgrs.2016.2566665 |
| Popis: | Building on existing techniques for satellite remote sensing of fires, this paper takes advantage of the day–night band (DNB) aboard the Visible Infrared Imaging Radiometer Suite (VIIRS) to develop the Firelight Detection Algorithm (FILDA), which characterizes fire pixels based on both visible-light and infrared (IR) signatures at night. By adjusting fire pixel selection criteria to include visible-light signatures, FILDA allows for significantly improved detection of pixels with smaller and/or cooler subpixel hotspots than the operational Interface Data Processing System (IDPS) algorithm. VIIRS scenes with near-coincident Advanced Spaceborne Thermal Emission and Reflection (ASTER) overpasses are examined after applying the operational VIIRS fire product algorithm and including a modified “candidate fire pixel selection” approach from FILDA that lowers the 4- $\mu\text{m}$ brightness temperature (BT) threshold but includes a minimum DNB radiance. FILDA is shown to be effective in detecting gas flares and characterizing fire lines during large forest fires (such as the Rim Fire in California and High Park fire in Colorado). Compared with the operational VIIRS fire algorithm for the study period, FILDA shows a large increase (up to 90%) in the number of detected fire pixels that can be verified with the finer resolution ASTER data (90 m). Part (30%) of this increase is likely due to a combined use of DNB and lower 4- $\mu\text{m}$ BT thresholds for fire detection in FILDA. Although further studies are needed, quantitative use of the DNB to improve fire detection could lead to reduced response times to wildfires and better estimate of fire characteristics (smoldering and flaming) at night. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|