Burned area mapping
Burned areas tend to have higher surface temperatures than surrounding non-affected vegetation shortly after a fire. This is explained by the fact that charcoal and ash absorb more energy than vegetation and also because there is no cooling from the vegetation formerly covering the burned area and the loss of soil moisture. LST data from sensors onboard polar platforms for burned area mapping have been used in several studies.
This increase in LST over burned areas has been used in algorithms that combine remote-sensed information on the near- and thermal infrared regions to distinguish burned areas. Figure 45 presents results for Portugal on 4 August 2003, obtained using fuzzy logic methods. This approach utilizes information from near- and thermal infrared channels from the AVHRR sensor onboard NOAA satellites, and computes the probability (on a scale from 0 to 1) that a given pixel is within a burned area.
Fig. 46: The summer of 2003 was characterized by very warm and dry conditions in Europe, especially in the west. In particular, it was the short-lived heatwave that occurred in the first fortnight of August that was responsible for the worst fire occurrences ever recorded in continental Portugal. According to official data, the burned area reached a total amount of 453,097 ha, 304,182 ha of which (i.e. 66% of the total) were recorded in the first two weeks of August, and 91,439 ha (i.e. 22% of the total) were recorded on 4 August. It is worth emphasizing that wildfires in August were responsible for the death of 21 people and an estimated loss of 15.5 million euros. The left panel represents an RGB (channel 4 - thermal infrared, channel 2 - near-infrared, channel 1 - red) of an AVHRR/NOAA image on 4 August 2003. The right panel represents the output of the neuro-fuzzy system (Calado et al., 2005) that utilizes information from the near- and thermal infrared region to produce a probability from 0 (no possibility that a pixel is burned) to 1 (full possibility that a pixel is burned).
Another important variable related to fires is the Fuel Moisture Content (FMC) of vegetation, since it is a driving factor of wildfire susceptibility and wildfire behavior. FMC of a sample is determined by dividing the difference between the wet and dry weights by dry weight of that sample. Potential improvements in live FMC estimations have been gained by incorporating information about satellite-derived surface temperature, as this variable would be expected to increase in drier plants due to decreased evapotranspiration.