Methods of fire situation analysis

Pic. 1. Area of Sakhalin Island covered in fire analysis. Clic picture for detale map (45 Êá)

NOAA series images with standard algorithms were used to identify active burn areas. Since this technique can produce errors, every outbreak of fire was verified by a number of subsequent images. Burned areas were identified using distant infrared spectral zones of NOAA images, and then verified by a series of subsequent images. The following images were used for the fire map of Sakhalin and part of the Far East: NOAA images dated September 4, 11, 23, 25, October 13, 22, 26, 27, 28, November 16, 17, 1998. Fire dynamics in a number of cases were verified and corrected by images of 2 spectral zones, of medium and high resolution of Resource-01 satellites, dated June 27, July 1, 7, 14, 17.

Due to the specific nature of capturing satellite imagery and complicating weather conditions, not to mention the haze which hung over Sakhalin for most of the time, we were able to analyze only part of the area. Picture 1 shows the area of Sakhalin analyzed in this report (approximately 87 % of the island's area).

In developing the map, we decided not to include fires on non-forest areas, (e.g. bogs, wetlands, forestless flood plains, stone slopes with no or little forest cover, previously burned territories lacking subsequent growth, logging sites with no growth, etc.) Neither does the applied technology allow to identify and register the areas, covered by ground fires of low intensity and by minor burning, as well as by "grass" fires, for instance spoil banks of railways and agricultural lands.

It is necessary to point out that early-morning distant infrared spectral zone satellite imagery, does not clearly detect sites which have completely stopped burning and smoldering. Therefore, the total estimate of burning areas is somewhat lower due to underreported extinguished fire sites.

The specific features of images and deciphering technologies do not always allow one to accurately identify remaining unburned area within major fire sites. This situation concerns most of all a cross-country terrain. However, this overestimation of particular fire sites is fully compensated for by those areas burnt by ground fires and not included into the total forest area of analysis due to the above mentioned reasons.

Thus, the estimates of the areas, burnt by fire in Sakhalin in 1998, appear to be minimal.

The work resulted in a computerized map of Sakhalin in MapInfo 4.5 showing marked fire sites, as well as a burn intensity classification: from the first (crown fires) to the fifth (final stage of smoldering).