Silapaswan, C.S., D. Verbyla, and A.D.
McGuire. 2001.
Land cover change on the Seward Peninsula: The use of remote
sensing to evaluate potential influences
of climate change on historical vegetation dynamics.
Canadian Journal of
Remote Sensing. 27:542-554..
Abstract. Vegetation on the Seward Peninsula, Alaska,
which is characterized by transitions from tundra to boreal forest, may
be sensitive to the influences of climate change on disturbance and species
composition. To determine the ability to detect decadal-scale structural
changes in vegetation, Change Vector Analysis (CVA) techniques were evaluated
for Landsat Thematic Mapper (TM) imagery of the Seward Peninsula from 1986
to 1999. Scenes were geographically corrected to sub-pixel accuracy
and then radiometrically rectified. Between the 1986 and 1992 satellite
scenes, the CVA detected changes in direction and magnitude of the two
indices (TM Band4/TM Band 3, TM Band 5). For Row 14, change was detected
for 135,518 ha and for Row 15, change was detected for 111,831 ha.
Between the 1992 and 1999 scenes, change was detected by CVA for 93,278
ha. Across Path 78 Row 14 and 15, the unsupervised classification
detected that 55% of the pixels changed between 1986 and 1992. Overall,
approximately 759,610 ha changed to a class with a more developed canopy
and only 268,132 ha changed to a class with a less developed canopy. CVA
results and photo interpretation together suggest that shrub advance is
approximately 100 m in valleys north of the Bendeleben Mountains and that
shrubs have increased along riverbed bottoms. Thus, the change detection
analysis based on the unsupervised classification indicates that land-cover
change on the Seward Peninsula was predominantly in the direction of increased
shrubbiness. Taken together, our comparison of CVA results, unsupervised
classification results, and visual interpretation of aerial photographs
suggests that shrub cover may be increasing on the Seward Peninsula.
The use of both CVA and unsupervised classification together provided a
more powerful interpretation of change than either method alone in transitional
regions between tundra and boreal forest.
Email: D.Verbyla@uaf.edu
Last updated: August 2001