PE&RS August 2002

Highlight Article

Fire Detection and Monitoring Products at the National Oceanic and Atmospheric Administration
Donna McNamara, George Stephens, Bruce Ramsay, Elaine Prins, Ivan Csiszar, Chris Elvidge, Ruth Hobson, and Chris Schmidt

Within the National Oceanic and Atmospheric Administration (NOAA), the National Environmental Satellite, Data, and Information Service (NESDIS) is responsible for acquiring, launching, and operating the nation's civil operational environmental satellites; two Geostationary Operational Environmental Satellites (GOES) over the United States (GOES-8 over the eastern U.S., GOES-10 over the western U.S.), and two Polar-orbiting Operational Environmental Satellites (POES), one in a morning orbit and one in an afternoon orbit. In addition, NOAA operates the U.S. Air Force weather satellite, the Defense Meteorological Satellite Program (DMSP), and acquires data from international environmental satellites.

While the original mission of these satellites was focused on meteorological and oceanographic observations, they have evolved into systems that provide a wealth of data on many other environmental parameters. One environmental observation that has seen increased interest is the detection of hot spots, such as those caused by wild fires.

NOAA's capability to support the fire management community through the detection and monitoring of wild fires from its environmental satellites is growing and improving. Additionally, the requests for these space-based products and services is increasing. As a result, NOAA's efforts in this arena have seen the development of specialized products and services to meet the ever-growing demand. The following is a compilation of information regarding four products developed at NESDIS. It is a brief overview, so the reader should refer to the NESDIS fire product page for more information; http://www.ssd.noaa.gov/.

To best address the fire community's requirements, NESDIS developed the Hazard Mapping System (HMS) which is an interactive processing system that allows the trained satellite analysts to manually integrate data from various automated fire detection algorithms (tuned for each of the satellite sensors), from the GOES sensor, POES sensor, and Advanced Very High Resolution Radiometer (AVHRR). The result is a quality-controlled display of the locations of fires detected by meteorological satellites within the 48 conterminous states and Alaska, with Hawaii to be added soon. The end product is generated in text, geographic information system (GIS), and graphic formats that are user friendly.

Within NESDIS, the Satellite Services Division (SSD) is running this improved fire and smoke product, integrating new remote sensing products, in a preoperational demonstration mode. The product was initially demonstrated during the 2001 fire season and many refinements have been made to support the 2002 fire season. The HMS product is made twice a day (approximately 4:00 pm and 11:00 pm Eastern time). Verification of these products is ongoing, but the products are not considered fully operational.

For the fire managers, the layers that go into the quality controlled HMS product are made automatically as soon as the data are received, despite the problem with false detects. Among the automated layers is the recently developed Fire Identification Mapping and Monitoring Algorithm (FIMMA) product developed within the NESDIS Office of Research and Applications (ORA) using POES data. The product is generated from the AVHRR instrument using NOAA14 passes, and NOAA16 nighttime passes, which have 3.7 micrometers measurements (sensitive to such heat sources). The FIMMA system includes a navigation correction, and a software package designed to search the images for known geographic points. It then "rubbersheets" the image to fit those points, thus improving the navigation accuracy to within about 1km. The orbital data are available near real time, typically 36 hours after satellite overpass. Comparing the potential fire pixel to the neighboring fire-free pixels is also done to eliminate false fire alarms due to high background temperature and reflection from bright surfaces. Additional tests, such as the screening for reflection from small water bodies in sunglint conditions, is also included in the processing. Fire locations represent the approximate location of the fire pixel and do not represent the actual fire size. The AVHRR instantaneous field of view at nadir is 1 by 1 km, but the fire phenomena may be much smaller than that. If no fires are detected for a given pass, a file is not placed on the server.

This product was recently transitioned from the research domain to the operational suite of fire products in the SSD. The product is generated by an automated algorithm and placed on the operational fire web page as soon as available. It should be noted that the product is not quality controlled. For a quality-controlled product, users are referred to the Hazard Mapping System (HMS) product.

The Wild fire Automated Biomass Burning Algorithm (WFABBA) is another automated product employing GOES Imager data developed by researchers in ORA in collaboration with the Cooperative Institute for Meteorological Satellite Studies (CIMSS) at the University of Wisconsin. In the raw WFABBA product, a large number of low possibility false detections have been noted around sunrise and sunset due to reflection off clouds. The algorithm uses multispectral GOES imagery (visible, 3.9 and 10.7 micrometers). The product has been filtered against detects noted in the past 12 hours. A new detect is held until it is seen at least twice by either satellite. This has greatly reduced the number of pixels potentially identified as fires. False detects are also noted during summer when surface temperatures can reach extreme values. Fire locations represent the approximate location of fire pixels and do not represent the actual fire size. The GOES data, which comes in every 15 minutes, has an instantaneous field of view at nadir of 4 by 4 km, which is quite large, but the fire phenomena are usually much smaller than that. Studies at CIMSS have shown that the minimum detectable fire size at the subsatellite point, and smoldering at 450K, is approximately 0.5 to 1 acre in size in relatively noncloudy conditions. This product can be easily read into a geographic information system (GIS) as table data. The product is run halfhourly upon receipt of imagery. Recently, the WFABBA software was transitioned to the operational fire product site at the SSD to be run operationally. The product is generated by an automated algorithm and is placed on the web page in near real time, and it has not been quality controlled. Again, for a quality-controlled product, users are referred to the Hazard Mapping System (HMS) product. The WFABBA product should become fully operationally by August 2002.

Work is also ongoing using fire detection from Defense Meteorological Satellite Program/Operational Linescan System (DMSP/OLS). The DMSP/OLS fire detection algorithm was developed and is being refined at NOAA's National Geophysical Data Center (NGDC) in Boulder, Colorado and is provided to SSD. Ingest of DMSP data, and preprocessing/geolocation of the data used by SSD to make the fire product, originate at NGDC. The algorithm uses the nighttime visible data from the OLS instrument and subtracts the stable (city) lights. Ideally, the lights that remain are caused by fires. Cloudy areas are screened to reduce reflection from clouds. Lakes and rivers are also filtered. An analyst manually screens the data to eliminate false detects and to look for obvious fire signatures. Users should note, the SSD product looks for fires only in the conterminous U.S. west of 95 degrees west longitude. The eastern U.S. is not analyzed at this time due to the large number of city lights. This technique cannot be used in Alaska during the summer months because there is too much sunlight when the satellite passes over. The analysis is done by 10:00 am Eastern time, using data from the DMSP/F15 satellite from the previous evening (equator crossing time about 9:20 pm). Detecting true fires from false detects, especially in high moonlight situations, can be difficult. This product is not yet considered operational and is undergoing validation. Users are referred to the Global Fire Detection web page at NGDC http://www.ngdc.noaa.gov/dmsp/fires/globalfires.html.

It should also be noted that NESDIS acquires data from the Moderate Resolution Imaging Spectroradiometer (MODIS) instrument on the NASA Terra satellite. MODIS fire products are from NOAA's MODIS Near Real Time Processing System, and the fire algorithm was developed by the MODIS Fire and Thermal Anomalies team (NASA and University of Maryland). MODIS is also a polar-orbiting instrument, so it provides two views per day of the U.S. fires. This will be increased to four looks per day as fire products from the MODIS AQUA instrument become available later this fall.

Donna McNamara, George Stephens, Bruce Ramsay, Elaine Prins, Chris Elvidge, and Ruth Hobson are with the National Oceanic and Atmospheric Administration/National Environmental Satellite Data and Information Service.

Ivan Csiszar is with the University of Maryland, Department of Geography.

Chris Schmidt is with the University of Wisconsin-Madison, Cooperative Institute for Meteorological Satellite Studies.