| Objective 1: Improve Categorical Resolution and Accuracy of
Forest Cover Type and Condition Mapping and Change Detection
Forest industry, counties, and state and federal government
agencies must assess the location, extent, and condition of forests
over large areas. These data form the foundation of planning,
for example, for mill expansion, allowable harvest, or response
to fires or insect or disease damage. Currently, such planning
is done either with aggregate data from several sources, with
differences in categories, accuracy and spatial resolution, or
infrequent statewide inventories. While more than 20 years of
research has been devoted to landcover classification with Landsat
data and there have been many improvements in image processing
systems, there is currently no data and analysis system that
produces accurate, categorically detailed forest cover type classifications
over large areas. However, we believe that is largely because
insufficient attention has been given to fully utilizing the
best data and analysis methods that are available.
The synoptic coverage of Landsat data is a major advantage of
Landsat data, but most research and classifications of the data
have been over small test sites. Therefore, to effectively use
the data in an operational setting, the problems of large area
classification must be addressed. A second observation is that
most studies do not seem to incorporate and take advantage of
the best combination of strategies for forest and land cover
classification. Research has often been with a single factor
rather than attempting to apply and evaluate the best combination
of data and procedures. The goal of this objective is to develop
strategies, with emphasis on integrating the best possible data,
algorithms and analysis methods, to maximize classification accuracy
and categorical detail. Possible methods include multitemporal
image classification, image stratification, combining optical
and radar data in classifications, selection of classifier, and
use of ancillary data.
The successful launch of IKONOS-2
satellite in the summer of 1999 opened several research opportunities
addressing forest management on the local level. All of the
project collaborators from industry, county land departments,
and the agencies believe this kind of high resolution imagery
has considerable potential in forest inventory and/or management.
We therefore added a suite of forest management related applications
which are listed below. These projects maintain the initial
objectives, which are to find the best possible algorithms
and analysis methods to maximize classification accuracy and
categorical detail using this new type of imagery.
Link to Objective
1 Results
|
