Regional And Mesoscale Meteorology Team
Daily Satellite Discussion
Tuesday July 15, 1997
Today's imagery looks at fire detection from GOES satellites. Colorado and much of the west now has areas with extreme fire risk at the peak of the summer tourist season. In fact, it was the lead news story last weekend in the Denver TV market.
As with any phenomena, GOES can be best used when all of the channels are considered. The primary GOES channel for fire detection is 3.9 um. This is because the radiance at 3.9 um increases more rapidly than at 10.7 for a given brightness temperature. Thus, if a small percentage of a pixel contains a hot fire, then there is a much greater radiance from the GOES 3.9 um detectors than 10.7. For small fires, a difference image between 3.9 um and 10.7 will show large differences in temperature. This is the reason we also look at the reflectivity product. However, at this time of year there can be problems. The maximum detectable temperature for the GOES-9 3.9 um channel is 325 K (GOES-8 is 10 K warmer), and pixels may become saturated during the summer. We noticed on Monday that the GOES-9 3.9 um imagery was showing a large area saturated at the "warm" end of the brightness temperature range. Recall that "warm" does not always refer entirely to temperature in 3.9 um because a portion of the signal is from reflected solar radiation during the daytime. So, since albedo and temperature contribute to the 3.9 um signal, contributions from highly reflective and warm/hot surfaces add to saturate the sensor as shown above (left) in the orange areas.
The next product that we would look at is the reflectivity product. The reflectivity product (a process by which we attempt to isolate the reflected portion of the 3.9 um signal) was also saturated in much of the west. Fires during the day in these areas may become difficult if not impossible to distinguish. The reflectivity product is useful in fire applications for at least three reasons.
1) It helps to show the reflected contribution in the 3.9 um imagery. Since reflective surfaces (shown as white in above-right) can also saturate the sensor, it is important to remove these saturated pixels from consideration as a possible fire.
2) It helps to show non-reflective land surfaces which typically are significantly vegetated or forested. In forest fire situations, the reflectivity product shows a significant contrast because the aforementioned background appears dark and the temperature contribution is still significant enough that it appears "saturated".
3) For small fires or initiation, the reflectivity product will likely indicate a fire before it becomes obvious in the 3.9 imagery alone. This is because the difference between the two channels adds to the detection capability. Sub-pixel hot spots cause 3.9 to appear much warmer than 10.7.
Finally, for large fires, it is also useful to look at the 1 km visible imagery. Smoke plumes can often be seen and be indicative of the conditions in the layers above the fire region.
Check the fire detection section of the RAMMT 3.9 um tutorial for more information and additional examples. There's also an excellent fire weather site hosted by the NWS in Boise which includes products from the forest service and NCEP.
Brian Motta