Transcript NSLakes_Poster2

Remote Sensing of Frozen Lakes
on the North Slope of Alaska
Nancy French, Sean Savage,
Robert Shuchman, Robert Edson
Altarum Institute
Ann Arbor, Michigan, USA
[email protected]
Introduction
Development of a remote sensing approach for monitoring
the freeze condition of freshwater lakes on the North Slope
is valuable because ground surveys of all lakes in the
region is not feasible. SAR has been used to map the
freeze condition of lakes to distinguish between lakes
which are frozen to the lakebed and those which contain
liquid water [2-6]. For lakes frozen to the bed, the SAR
return is low and the lake appears dark, while for lakes that
are not completely frozen, the SAR return is high and the
lake appears bright. Based on these existing techniques,
we used synthetic aperture radar (SAR) images from the
ERS-2 remote sensing satellite to map the freeze condition
of lakes in a small area of Alaska’s North Slope.
Comparison of September and March
ERS images of North Slope Lakes.
Bright lakes are not frozen to the lake
bed while lakes with dark returns are
frozen to the lake bed and have no
liquid water under the ice. Lake
boundaries derived from the Landsat
classification are shown in red and
were used in the analysis.
Abstract
We used synthetic aperture radar (SAR) images from the ERS-2 remote sensing satellite
to map the freeze condition of lakes on Alaska’s North Slope, the geographic region to
the north of the Brooks Range. An image from March 1997, to coincide with the period of
maximum freeze depth, was used for the frozen lake mapping. Emphasis was placed on
distinguishing between lakes frozen to the lakebed and lakes with some portion unfrozen
to the bed (a binary classification). The result of the analysis is a map identifying lakes
as frozen to the lakebed and lakes not frozen to the lakebed. This analysis of one SAR
image has shown the feasibility of a simple technique for mapping frozen lake condition
for supporting decision making and understanding impacts of climate change on the
North Slope.
Summary of Approach
Landsat 7 ETM+ from 2002 and and ERS-2 SAR imagery from the
1996/97 winter season was obtained for an area in the Northwest NPRA
(south of Point Barrow).
A supervised classification of a Landsat 7 ETM+ image was generated to:
– Provided clear discrimination of lake boundaries.
– Serve as a lake mask in the analysis of the SAR image.
A March SAR image was used to determine frozen to lakebed or not
frozen to lakebed using established techniques
Emphasis was placed on distinguishing between lakes frozen to the
lakebed and lakes with some portion unfrozen to the bed (a binary
classification).
– The histogram of pixel values was analyzed to determine the
appropriate threshold pixel value:
• If maximum pixel value > 78 = not frozen to lakebed
• If maximum pixel value < 79 = frozen to lakebed
REFERENCES
[1] NRC, “Cumulative environmental effects of oil and gas activities on Alaska's North Slope,” 288 pp., National Academies Press, Washington DC, 2003.
[2] C.R. Duguay and P.M. Lafleur, “Determining depth and ice thickness of shallow sub-Arctic lakes using space-borne optical and SAR data,” Int. J. Rem.
Sensing, 24(3), 475-489, 2003.
[3] W.R. Rouse, and others, “Effects of climate change on the freshwaters of arctic and subarctic North America,” Hydrologic Processes, Vol II, 873-902, 1997.
[4] W.F. Weeks, A. J. Gow, R. J. Schertler, “Ground-truth observations of ice-covered North Slope lakes imaged by radar” Tech. Report No. Report 81-19, Cold
Regions Research and Engineering Laboratory, 1981.
[5] W.F. Weeks, P. V. Sellmann, W. J. Campbell, “Interesting features of radar imagery on ice-covered North Slope lakes,” J. Glaciology, 18(78), 129-136, 1977.
[6] M.O. Jeffries, K. Morris, G. E. Liston, “A method to determine lake depth and water availability on the north slope of Alaska with spaceborne imaging radar
and numerical ice growth modeling,” Arctic 49, 367-374, 1996.
The Altarum Institute is a nonprofit research and innovation institution with more than 55 years of success in researching, developing, and
deploying advanced information technology and decision support systems solutions that promote the sustainable well-being of society.
Altarum serves government and commercial customers in the national security, healthcare, and environment sectors.
Landsat 7
ERS-2 SAR
July 2001 geocoded image
March 1997 image
Geocoded
using L-7
image
Image classified into 4
classes: water, ice,
wetland, tundra
Lake areas extracted
(masked) using Landsatderived map of lakes
Compared to BLM deep
lakes map
Statistics
extracted
Recoded to 2
classes: lake, land
Raster image
converted to
vector map of
lakes
Lakes tagged as frozen to bed
or not frozen to bed using
maximum pixel value threshold
Avail. for NS
decision support
Image filtered to
exclude small
areas of water
Results
500
Frozen
to lake
bed
400
350
Not frozen to lake bed
300
Frequency
Most lakes in Example 1 (top) show a
bright SAR signature, which indicates
that they are not frozen to the lake
bed. Several larger lakes in the
March image for Example 2 (bottom)
are dark, indicating that they are
frozen to the lakebed.
John Payne
Alaska State Office
Bureau of Land Management
Anchorage, Alaska, USA
250
200
15 0
10 0
50
0
20
30
40
50
60
70
80
90
10 0
78
110 12 0 13 0 14 0 15 0 16 0 17 0 18 0
Pixel value
19 0 2 0 0 2 10 2 2 0 2 3 0 2 4 0 2 5 0 2 6 0
Distribution plot of the maximum
pixel value within each lake.
Lakes with all pixels less than or
equal to 78 are considered to
be frozen to the lake bed.
Lakes which are not frozen to
the lake bed have at least one
pixel greater than 78.
450
400
350
300
Frequency
The North Slope of Alaska is located in the far north,
bordered by the foothills of the Brooks Range to the south
and the Arctic Ocean to the North. The region
encompasses 89,000 mi2, comprised of diverse and unique
ecosystems. This area, is well known for its oil fields, which
are thought to have the greatest remaining oil potential of
any onshore area in the United States.
One of the most prominent features on the North Slope is
the mosaic of freshwater lakes produced through
thermokarst activity and other cryogenic processes, many
of which are very shallow (< 2 meters depth). The
maximum freeze depth for lakes of in the North Slope
region is 1.8 m [1], typically occurring during March [2].
Lakes less than 1.8 m deep freeze to their beds, while
lakes greater than 1.8 m deep will have liquid water
throughout the winter.
Edward Josberger
Washington Water Science Center
U.S. Geological Survey
Tacoma, Washington, USA
250
200
150
100
50
0
250-500
500-999
1,000-9999
10,000-99,999
100,000-999,999
1,000,000+
Area (m2)
Results of the March 1997
analysis of frozen lakes
showing lakes frozen to the
lakebed in blue, and those
not frozen to the lake bed in
yellow. Results are overlaid
on the March 1997 ERS-2
SAR image from which the
map was derived.
Lakes frozen to bed
Lakes not frozen to bed
Out of 2,350 lakes mapped, 767 are frozen to their
beds (33%), and 79 are 1km2 or larger. The
distribution of lake area by freezing condition shows
that, generally, smaller lakes will freeze to their bed
while larger lakes will not. There are some exceptions
with 4 lakes >1 km2 freezing to their beds and 68
lakes <1,000 m2 remaining unfrozen to their bed.