Wildland fires have been regulating the evolution, productivity and biodiversity of natural plant communities and ecosystems since millennia. Many vegetation communities have adapted to fire dynamics and developed as a response to fires leading to a higher species-richness. However, since humans started to set artificial fires and change natural fire frequencies and timing, these have become one of the most devastating hazards worldwide. They destroy environment and property, impact air quality, contribute to global warming, and even threaten lives.
The TIMELINE Hot Spots product shows pixels where the automated algorithm detected active fires. Mainly very hot fires (which can cover only a small part of an AVHRR pixel), or bigger and cooler fires covering larger areas can be detected by the developed algorithm.
The TIMELINE Burnt Areas product contains information about burn scars. As burn scars can be detected over a certain time after the fire event, it is even possible to detect burnt areas, which were covered by clouds during the time the fire was active. The minimum detectable size of a burnt area depends on the spatial resolution of the imagery.
The fully automated Hot Spot detection processor is based on a further development of the contextual algorithm of Giglio et al. (1999). To reduce the possibility of false (over) classification, unsuitable land cover classes and disturbing objects are excluded prior to the detection of the fire Hot Spots.
The first step of the Hot Spot processor is a selection of possible fires. Different thresholds are used for deriving different probability levels. Within a contextual approach the brightness temperature values of the fire candidate pixels are compared with the values of valid background pixels around these Hot Spot candidates. A set of statistical measurements decides whether a Hot Spot candidate pixel is finally selected as a fire Hot Spot pixel. More detailed information is available in Plank et al. (2017).
Burnt Area pixels are detected via a change detection approach between a pre- and a co(post)-event from AVHRR acquisitions. More precisely, an AVHRR dataset of pre-event imagery, based on data of up to 30 days prior to the event, is used to generate a cloud-free pre-event mosaic. The burnt area itself is derived by a combination of a set of indices. Automated thresholding techniques are applied for the final classification of the burnt area. Urban and desert area and area covered by water or cloud are excluded from the processing of burnt areas.
The results of the Hot Spot processor were tested with simulated fire data. Moreover, the processing results of the AVHRR imagery were validated with five different datasets: MODIS Hot Spots, visually confirmed MODIS Hot Spots, fire-news data from the European Forest Fire Information System (EFFIS), burnt area mapping of the Copernicus Emergency Management Service (EMS) and data of the Piedmont fire database. For details see Plank et al. (2017). The figure below shows a validation example. The result of the developed AVHRR Hot Spot detection processor shows a good agreement with the information derived from Landsat-8 and MODIS. The Burnt Area processor is currently being enhanced and further automated.
Plank, S; Fuchs, E-M; Frey, C (2017) A Fully Automatic Instantaneous Fire Hotspot Detection Processor Based on AVHRR Imagery—A TIMELINE Thematic Processor. Remote Sensing, 9 (1), 30.
Giglio, L.; Kendall, J.D.; Justice, C.O. (1999) Evaluation of global fire detection algorithms using simulated AVHRR infrared data. Int. J. Remote Sens. 20, 1947–1985.