Multi-sensor analysis of the effects of a wildfire in an Alaskan black spruce forest

摘要

Wildfires play a central role in the ecology of boreal forests throughout the circumpolar region. Fires result in distinct and characteristic signatures on all remote sensing imagery. The large proportion of total area burned in boreal forests occurs in large fire events-the majority of area burned occurs in fires greater than 10,000 ha in size. These fires result in signatures which are not only spatially heterogeneous, but also have distinct temporal signatures. The authors present the results of a study which investigated the use of a variety of systems to monitor the effects of fire on remote sensing data collected over a large (40,000 ha) forest fire near Tok, Alaska. This fire occurred during the summer of 1991. Visible and near-IR region (AVHRR, SPOT) as well the microwave region (the ERS-1 and JERS-1 SARs) observations were made. The studies have shown that in the visible and near-IR data sets, the remotely-sensed signatures result from variations in the levels of burn intensity and patterns of vegetation regrowth. These patterns can clearly being interpreted through examination of vegetation indices derived from the AVHRR and SPOT data sets. On the other hand, the analyses of the microwave imagery collected by the ERS-1 SAR show that the spatial patterns of radar backscatter are primarily due to variations in surface roughness due to variations in burn intensity as well as variations in soil moisture, while the temporal patterns of radar backscatter are due to seasonal variations in soil moisture. The authors present examples of the remotely-sensed data sets along with ground-truth measurements which illustrate the sources of variations observed in the satellite imagery.