HR: 1330h
AN: A22B-1072    [PDF]
TI: A Statistical Analysis of Automated and Manually Detected Fires Using Environmental Satellites
AU: * Ruminski, M G
EM: mark.ruminski@noaa.gov
AF: NOAA/NESDIS, 5200 Auth Rd, Camp Springs, MD 20746 United States
AU: Mcnamara, D
EM: donna.mcnamara@noaa.gov
AF: NOAA/NESDIS, 5200 Auth Rd, Camp Springs, MD 20746 United States
AB: The National Environmental Satellite and Data Information Service (NESDIS) of the National Oceanic and Atmospheric Administration (NOAA) has been producing an analysis of fires and smoke over the US since 1998. This product underwent significant enhancement in June 2002 with the introduction of the Hazard Mapping System (HMS), an interactive workstation based system that displays environmental satellite imagery (NOAA Geostationary Operational Environmental Satellite (GOES), NOAA Polar Operational Environmental Satellite (POES) and National Aeronautics and Space Administration (NASA) MODIS data) and fire detects from the automated algorithms for each of the satellite sensors. The focus of this presentation is to present statistics compiled on the fire detects since November 2002. The Automated Biomass Burning Algorithm (ABBA) detects fires using GOES East and GOES West imagery. The Fire Identification, Mapping and Monitoring Algorithm (FIMMA) utilizes NOAA POES 15/16/17 imagery and the MODIS algorithm uses imagery from the MODIS instrument on the Terra and Aqua spacecraft. The HMS allows satellite analysts to inspect and interrogate the automated fire detects and the input satellite imagery. The analyst can then delete those detects that are felt to be false alarms and/or add fire points that the automated algorithms have not selected. Statistics are compiled for the number of automated detects from each of the algorithms, the number of automated detects that are deleted and the number of fire points added by the analyst for the contiguous US and immediately adjacent areas of Mexico and Canada. There is no attempt to distinguish between wildfires and control or agricultural fires. A detailed explanation of the automated algorithms is beyond the scope of this presentation. However, interested readers can find a more thorough description by going to www.ssd.noaa.gov/PS/FIRE/hms.html and scrolling down to Individual Fire Layers. For the period November 2002 thru August 2003 64% of the total number of fires were added manually. This ratio has shown a seasonal fluctuation with a smaller percentage of fires being added manually during the summer fire season and a larger percentage added during the winter and spring when agriculture and control burns dominate. This is due to the shorter duration and cooler depiction of the agricultural fires in the satellite imagery and the limitations and conditions specified in the algorithms. For the various algorithms 23% of the total number of fires were from ABBA, 5% were from FIMMA and 16% were from MODIS. However, there was a wide discrepancy in the percentage of automatically detected fire points that the analysts deleted for each of the algorithms. The MODIS points were most reliable with 84% passing the editing phase and this has been a fairly consistent ratio through the period. 53% of the ABBA points were retained for the final analysis. However, a much larger percentage of points were deleted during the summer season (May-August). This was due in large part to a greater number of false alarms caused by high surface temperatures and high solar reflectivity off clouds near sunrise/sunset. For the FIMMA only 39% of the points were retained. While there are a number of areas for improvement with the algorithm, the single greatest cause for the large number of deletions has been not accurately detecting noise in the imagery which is interpreted as a fire. It is hoped that this statistical information will be a useful tool in making adjustments to the algorithms that will lead to a greater number of fire detects with a smaller percentage of false alarms.
UR: http://www.ssd.noaa.gov/PS/FIRE/
DE: 3360 Remote sensing
DE: 9350 North America
SC: Atmospheric Sciences [A]
MN: 2003 Fall Meeting