High
spatial resolution multi-spectral imagery. High spatial resolution
multi-spectral imagery is useful in numerous natural resource and
environmental
applications.
We've been using such
imagery (aerial photographs, 1-m ADAR, 1-m Quickbird, 4-m IKONOS)
over time in several projects: to map tree mortality caused by
disease and beetle damage, to map wetland vegetation, and to extract
vegetation classes. The
maps produced
are
valuable to examine ecological function: e.g. disease establishment
and spread, vegetation succession, and habitat fragmentation. Analysis
of high spatial
resolution
imagery presents technical challenges different from those
produced by coarse resolution imagery analysis. For example, traditional
pixel-based classifiers do not work well with high spatial
resolution imagery when the mapping target (e.g. an oak tree) is
larger than the
pixel size. In addition, airborne
imagery introduces georeferencing problems not encountered with
satellite imagery, making pixel-by-pixel change detection
problematic. We are investigating these classification and
georeferencing issues in a number of projects (e.g. in our Sudden
Oak Death and Wetlands
research), and using and developing and utilizing new tools to
meet these challenges.
New
Sensors. We are also interested in applying imagery from new
hyperspectral and
active (e.g. LiDAR) sensors to our ecological work. Previous
research investigating the ability of hyperspectral data to
distinguish between different health levels in oak trees was
inconclusive, yet the tool remains a powerful one for vegetation
discrimination. LiDAR might provide valuable information about
structural changes to oak woodlands experiencing disease, as well
as
provide more detailed topographic information for ecological modeling.
