Fire risk, active fire detection and fire damage assessment
Land and forest fires present an increasing problem in many tropical forest regions. Remote sensing might deliver useful information before (fire risk), during (daily active fire detection) and after (damage and regrowth assessment) fire events.
Fire Risk Assessment
There are clear indications that radar can be useful in support of mapping fire susceptibility by using the stability of the radar backscatter level in the pre-fire period as an indicator. Consequently, fire risk/hazard maps can be made in order to plan preventive actions before forest fires start. An example for an area in East Kalimantan is given in figure 1 below. See also the fire damage assessment section below.
Active Fire Detection
The regular detection of locations of active fire is possible using various sensors, such as Envisat AATSR (or its predecessor ATSR at ESAs ERS satellite), MODIS, and NOAA AVHRR, even on a daily basis.
Active fires in Central Kalimantan detected at night-time using ERS ATSR (image courtesy of ESA, CNES)
Active NOAA fire hotspots detected during 2004 in combination with a regional scale monitoring land cover / use map based on SPOT Vegetation 1 km resolution data, and the borders of a pulp and paper plantation in West Kalimantan (grey). Fire has clearly been used in the northern sector of the plantation (see concentration of red dots), most probably to clear the area.
Fire Damage Assessment
The assessment of fire damage may be facilitated by remote sensing data at several scales using the results of the medium and /or high resolution monitoring systems.
The direct impact of fires on Borneo during the 1997-1998 fire event as visible on a NOAA image (left, note huge coverage of yellow smoke. Image courtesy of Meteorological Service Singapore) and a SPOT Vegetation composite of 1998 (right, note the burned areas in pink/purple).
In the ERS radar damage assessment images below several features are striking. An evaluation of the pre-fire period fire risk and the actual damage done, for example, reveals that all forest under these extremely dry conditions actually burned except for the protected forest areas Wanariset, Bukit Bangkirai and Sungai Wain, where active fire fighting took place and for small areas close to the mangroves. The latter may be related to more favourable terrain drainage conditions. Good forest, not actively protected from fires, did burn. An example is the stretch of forest in the lower-left corner between forest plantations. These areas are extremely vulnerable and apparently the fire easily spreads.
ERS SAR data fragments of an 18 x 20 km2 area near Balikpapan (East-Kalimantan) located directly west of the Wanariset research station. (a) SAR image of 13 April 1998, (b) multi-temporally segmented SAR image of 2 June 1997 (red), 13 April 1998 (green) and 18 May 1998 (blue), (c) pre-fire period classification, (d) Independent post-fire period classification. INDREX-1996 campaign, East Kalimantan, Indonesia.
Pre-fire and post-fire classification accuracy of multi-temporal ERS SAR data.