Transcript SWOT measurements for improving our

SWOT Measurements for
Improving Understanding of
Mid-Latitude Hydrology
Franklin W. Schwartz
School of Earth Sciences
The Ohio State University
Acknowledgements: National
Science Foundation, Ganming
Liu, Bo Zhang, Jerry Allen
September 15, 2008
• Example of pothole lake in South Dakota in
Waubay Lakes chain
• Typically, a product of an extremely hummocky,
glacial terrain
• Lakes are commonly found in closed basins,
often saline, surrounded farmland
• Thousands of small lakes
• Pothole lakes and wetlands occur together
with few large recreational lakes
• Entire watershed area is hydrologically
closed
Prairie Pothole Region
• Unique lake system
Canada and USA
• Upwards of 6 million
pothole lakes
• Most located
around edges with
more rainfall
• Important farming
impacts
Hydrology Prairie Pothole Lakes
• Water primarily from snowmelt runoff
with ground water and summer rains less
important
• Water levels fluctuate tremendously
depending upon variable climate
• Continental climate of prairies cycles
between drought and deluge
Waterfowl
• Prairie Pothole region produces 50%
primary game ducks in North America
• For seven species – e.g., mallard, bluewinged teal, redhead, and canvasback
- region home to >60% N.A. breeding
population
• Populations of some species of ducks
rise and fall in response to deluges and
droughts
Why Study these Lakes?
• Hydrology and biology of these lakes and
wetlands well understood
- more than 30 years of study at
Cottonwood Lakes Study Area
• Important role ground-water and surface
water interactions
• Long-term monitoring at a few sites
explained how pothole lakes responded
to periodic drought and deluge
New Challenges
• Emerging challenge for hydrologists is
describing and understanding processes
in large complex systems
• Conventional monitoring approaches
inadequate and not commonly available
• Tremendous potential in linking regionalscale models, and space geodetic and
remote sensing techniques
• SWOT provides important new
capabilities
Regions of Interest
M
P
• Pothole lakes not
uniformly distributed
• Prairie Coteau and
Missouri Coteau
Lakes and Climate Variability
• Study area – tip of
Prairie Coteau in SD
• Climate affects on
water on landscape
- change in numbers of
lakes, size, volume
- 1988-92 2rd drought
century
- 1993-1997 greatest
deluge
- observable by Landsat
Precipitation Waubay Lakes Area
60
Snow (cm)
50
40
30
20
10
0
25
2nd Drought
Rainfall (cm)
20
1st Deluge
15
10
5
0
86
87
88
89
90
91
92
93
94
95
Year
96
97
98
99
00
01
02
03
Lake Occurrences – 1990 vs. 1997
Lakes and Power Laws
• Known for many years that areas of lakes
followed a power-law distribution
• e.g. 2500 lakes by Kent and Wong [1982]
• Now commonly applied in global
assessment
• What pattern of organization of lake
systems? Can we use for analysis?
• Powerful because lake/wetland
complexes rationalized by few parameters
Lakes and Power Laws
• Developed area versus frequency curves
- one curve for each Landsat image – Spring
- straight line
- boundaries
4/23/1987
4/15/1990
5/06/1992
5/04/1997
5/18/2002
Regression lines
1000
800
600
Count of lakes
400
200
100
80
60
40
20
10
8
6
6
8
10
Small Lakes
20
40
60
Area of lakes (Landsat pixels)
80 100
Large Lakes
Seasonal Effects
• Within any year considerable variability
- spring to summer – small lakes impacted
4/15/1990
8/05/1990
6/16/2001
8/27/2001
5/18/2002
7/29/2002
9/05/2002
Regression lines
1000
800
600
Count of lakes
400
200
100
80
60
40
20
10
8
6
6
8
10
20
40
60
Area of lakes (Landsat pixels)
80 100
Additional Imagery
• Lines extend 1.5 orders magnitude in area
• Colored digital aerial photography
• 1 meter resolution lets us measure lakes
areas of the order of 100 m2
Develop Test Area
• Landsat
- Low res over big area
• DOQQ
- Hi res over small area
• Next Step
- power law for DOQQ
- small area, fewer lakes
Are Lake Areas Self-Similar?
5000
3000
Counts of lakes from DOQQ
Normalized counts from DOQQ
Counts of lakes from Landsat
Regression lines
1000
700
500
300
Count of lakes
log( y )  6.61 - 1.29 log( x )
log( y )  7.07 - 1.41 log( x )
100
70
50
30
10
7
5
3
log( y )  4.78 - 1.29 log( x )
1
100
300
600
1000
3000
Area of lakes (m2)
6000 10000
30000
Dust Bowl Drought – 1930s
• Aerial photographs commonly available
1939
2003
Extrapolate 1939 Photography
3000
Line in 1939 (normalized)
Line on 8/05/1990
Line on 5/06/1992
Line on 5/04/1997
Regression lines
log( y )  7.39 - 1.50 log( x )
1000
700
500
log( y )  5.72 - 1.22 log( x )
log( y )  7.60 - 1.49 log( x )
Counts of lakes
300
log( y )  4.64 - 0.89 log( x )
100
70
50
30
10
7
5
3
1
100
300
600
1000
3000
Lake areas (m2)
6000 10000
30000
Conceptual Model
• Area small lakes changes rapidly – season
• Area large lake changes slowly - cycles
Extensions
• Modeling now underway to simulate
behavior of a lake complex 100,000 lakes
• Ganming Liu able to calibrate to power
laws and long-term records for individual
lakes
• Work will be helped when SWOT mission
comes along
- changes in storage great opportunity to
recast power laws
Sample Simulation Results
• 100-year simulation of a pothole lake
complex along Missouri Coteau, ND
• Stochastic analysis ~106 lake basin
realizations to provide power laws
100
04/1992
09/1992
04/2002
09/2002
1000
Number of lakes
Number of Lakes
08/1939
08/1986
08/1990
08/2002
100
10
10
0.6
0.8
1
2
Lake area (ha)
4
6
8
10
0.4
0.6
0.8
1
2
Lake area (ha)
4
6
8
Important Findings
• Like others found that areas of lakes
obey a power-law function – 3.5 orders
• No single power law because rapid shifts
as a function of climate
- seasonal effects important
• Small lakes and large lakes respond to
different climate signals
• For this reason, small lakes could be
robust for small periods in a long drought