Abstract:

Forest burned scar area is an important factor for monitoring forest fire and assessing the losses. The traditional method quickly extracting burned scar area is based on satellite images with low-middle spatial resolution and cannot fully meet the need of the accuracy of assessment. This paper presents a new approach which takes advantage of high spatial resolution of CBERS/CCD and high temporal resolution of MODIS. Firstly, the vegetation coverage derived from CBERS/CCD was used to calculate NDVI with the Linear Special Mixed Model. Based on the data, the vegetation coverage data with the MODIS spatial resolution was obtained by calculating the average. Secondly, the burned scar pixels were derived by using the NDVI difference between the MODIS data before and after the forest fire. Finally the burned scar region area can be calculated based on the vegetation coverage data and the burned scar pixels. The validation, conducted in Yakeshi City in Inner Mongolia Autonomous Region in the late of May,2006, showed that the new method enhanced the accuracy in estimating the burned scar area over the traditional one.

Key words: multi-source remote sensing data, forest fire, burned scar region