Remote Sensing of Total and Surface Burn Ratios Following a Wildfire in East Siberia using 30 m–1 km Resolution Images

We evaluated the estimation of the area ratios of the land categories including the total-burn/wither and surface-burn areas following a wildfire in East Siberia. We obtained the land classification from 30-m resolution Landsat ETM+ image data and used it for evaluating the estimation of the area ratios of the land categories in each pixel by using 500-m and 1-km resolution images, which were generated to simulate six of the MODIS visible-to-shortwave-infrared bands. The 30-m resolution classification image was effective for the evaluation since the classification was consistent with the interpretation of the two IKONOS images obtained prior to and following the fire. The spectral mixture analysis (SMA) and multiple linear regression analysis (MLRA) were used from the viewpoints of the area ratio estimation of each land category in one pixel and in the entire study area. As a result, the 500-m resolution image provided more congruent results than the 1-km resolution image in both the SMA and MLRA. The area ratios of the total burn and surface burn were estimated using the SMA with the 500-m resolution image with 6 % and 3 % errors, respectively. The total-burn area and surface-burn area in the study area were estimated to be 3.8 (%) and 28.6 (%), which was an underestimation of 3 % an