APPLICATION OF REMOTE SENSING FOR MAPPING AREAS AFFECTED BY WILDFIRES AND MONITORING VEGETATION CHANGE IN SOUTHEAST IDAHO

The Idaho National Engineering and Environmental Laboratory (INEEL), located in southeast Idaho, U.S.A., was established in 1949 and is designated as one of seven D.O.E. National Environmental Research Parks (NERP). INEEL’s NERP designation allows INEEL to serve as an outdoor ecological research laboratory. Over the past several years, the INEEL and surroundings land holdings have experienced numerous large wildfires that have impacted the biotic and physical landscape of the area.

Mutlispectral Landsat TM data is used to identify and map the vegetative cover prior to, and following, the onset of wildfires on INEEL and surrounding areas. . Our studies indicate increases in reflectance in Landsat band 7 and decreases in reflectance in Landsat band 4 in areas of fire burn scars. Therefore, a ratio of bands 7 and 4 was tested. This ratio proved to be the most effective measure of burn severity and delineation of burn perimeters, in comparison to other ratios and band combinations.

Like most of the western U.S., the sagebrush ecosystem in southeast Idaho sustains many wildlife species. Research has shown that historically, sage-steppe type environments burn approximately every 60-100 years; whereas cheat grass infestations burn every 3-5 years. The exotics increase fuel loads, resulting in increased fire reoccurrences. Thus, future wildfire prevention depends heavily on mapping these exotics. For this purpose airborne hyperspectrtal remote sensing data (HyMap) was utilized to map the cheat grass.

As a first step, the NDVI (Normalized Difference Vegetation Index) was used to show to the robustness of vegetation. The NDVI was calculated using bands 4 and 3 from Landsat TM data and wavelengths 0.76-0.9 micrometers and 0.63-0.69 micrometers from the HyMap data. The NDVI results show good classification between dead versus live vegetation.

Finally supervised and unsupervised classification techniques were also used to map the cheat grass. Our preliminary classification results showed little success in using supervised classifications because of the limited available field data. Alternatively, the unsupervised classifications were found to be useful in classifying sagebrush versus senescent and live grasses. We are collecting additional field (spectral and GPS) data to cross check this classification techniques.