Background > Fire M3 > Data Sources and Methods

Background Information

Fire Monitoring, Mapping, and Modeling (Fire M3)

Data Sources and Methods

AVHRR Hotspots 

For the identification of actively burning fires, Fire M3 acquires visible and infrared satellite imagery from the Advanced Very High Resolution Radiometer (AVHRR) sensor on board the US National Oceanic and Atmospheric Administration (NOAA) polar-orbiting satellite series. The data are received from two satellite receiving stations, located at Prince Albert, Sask. and Mont-Joli, Que. Automated daily processing of the images is done at the Canada Centre for Remote Sensing (CCRS) in Ottawa. This includes geocoding, which determines the precise location and orientation of the image on the earth's surface, and compositing, which involves combining individual images to produce a single image covering all of Canada. Where two images overlap, the pixels with the smallest zenith angle are selected.

Fire detection algorithms (Li et al. 1997, 2000; Abuelgasim and Fraser 2002) are then applied to the single-day and night composite images to identify active wildfires. Hotspots (pixels identified as containing fire) are recognizable in the mid-infrared channel because of the heat generated by the fire. However, other factors such as warm surfaces, sun glint from water bodies, and highly reflective clouds can produce the same signal. A series of threshold tests or filters is used to remove false hotspots. Water bodies are masked out, as are nonforested areas.

The major limitation of satellite-based fire monitoring is the inability to identify active fires under cloud or heavy smoke. Fires can also be missed if they are not active at the time of satellite overpass or if they are small, of low intensity, or burning on the surface under the tree canopy. In addition, a small number of false hotspots are generated in spite of the filtering process. Many of these come from industrial sites or noise (errors) in the satellite images.

MODIS Hotspots

The Moderate-Resolution Imaging Spectroradiometer (MODIS) is a sensor on board the National Aeronautical and Space Administration's (NASA's) Terra and Aqua satellites, launched in 1999 and 2002, respectively. MODIS acquires data in 36 spectral bands, some of which are ideal for fire detection (Kaufman et al. 1998). Hotspot locations are determined by the MODIS Land Rapid Response System, a collaboration between NASA, NOAA , the University of Maryland, and the USDA Forest Service. Hotspots are provided to the Fire M3 project courtesy of the Fire Chemistry Project at the USDA Forest Service's Rocky Mountain Research Station.

The fire detection strategy incorporates both a fixed-threshold algorithm and a contextual (relative to background) algorithm. If a pixel fails to meet the criteria for the threshold algorithm, it is tested further by the contextual algorithm. The criteria are different for daytime and nighttime imagery. The daytime process also includes an additional test to filter out pixels contaminated by sun glint.

MODIS is expected to have better fire detection potential than AVHRR because the main fire detection channel is at a more sensitive frequency and saturates at a much higher temperature than the equivalent AVHRR channel.

Smoke

The smoke in the Fire M3 images is identified from AVHRR imagery using an algorithm developed by CCRS. Unfortunately, it is frequently impossible to distinguish smoke from cloud. The best way to determine if smoke has been correctly identified is if it can be associated with nearby hotspots. In other words, where there's fire, there's smoke.

Cloud

Cloud is also identified by CCRS from AVHRR imagery. Because the final national composite image is made up of many images taken at different times, the cloud cover depicted in the images does not represent the actual cloud cover at any particular time. Rather, the final image gives an indication of which areas of the land surface were visible to the sensor at some time during that day. Fires burning under cloud cover cannot be identified from satellite imagery.

Ecozones

The ecozones classification (Wiken 1986) used in the hotspot reports was produced by Environment Canada. It is the most general of four levels of classification dividing Canada into zones, provinces, regions, and districts on the basis of physiographic and ecological factors.

References

Abuelgasim, A.A.; Fraser, R.H. 2002. Day and night-time active fire detection over North America using NOAA-16 AVHRR data. Pages 1489-1491 in IEEE International Geoscience and Remote Sensing Symposium, Toronto, ON. 24-28 June 2002. Volume 3.

Kaufman, Y.J.; Justice, C.O.; Flynn, L.P.; Kendall, J.D.; Prins, E.M.; Giglio, L.; Ward, D.E.; Menzel, W.P.; Setzer, A.W. 1998. Potential global fire monitoring for EOS-MODIS. Journal of Geophysical Research 103:32215-32238.

Li, Z.; Cihlar, J.; Moreau, L.; Huang, F.; Lee, B.S. 1997. Monitoring fire activities in the boreal ecosystem. Journal of Geophysical Research 102:29611-29624.

Li, Z.; Nadon, S.; Cihlar, J. 2000. Satellite detection of Canadian boreal forest fires: development and application of the algorithm. International Journal of Remote Sensing 21(16):3057-3069.

Wiken, E.B., compiler. 1986. Terrestrial ecozones of Canada. Environment Canada, Hull, QC. Ecological Land Classification Series 19.

Fire M3 Summary

Fire M3 Frequently Asked Questions

Last updated: 2006-07-08

Important Notices