Transcript Document

Walnut Creek: Monitoring, Modeling, and
Optimizing Prairie Restoration
Sergey Rabotyagov1, Keith Schilling3, Manoj Jha2, Calvin
Wolter3, Todd Campbell2
Affiliations: 1. College of Forest Resources, University of Washington, Seattle, 2. Center
for Agricultural and Rural Development, Iowa State University, Ames, IA. 3. Iowa DNRGeological Survey
Gulf Hypoxia Workshop
Ames, Iowa
October 16, 2008
Walnut Creek Watershed Restoration
• The project was established in 1995 in
relation to watershed restoration activities
at Neal Smith National Wildlife Refuge
located near Prairie City, Iowa
• Large areas of the Walnut Creek
watershed have been converted from row
crop to native prairie by the US Fish and
Wildlife Service
• Paired watershed approach - Walnut
Creek is 12,890 ac (treatment watershed)
and Squaw Creek is 11,714 ac (control
watershed)
• Watersheds share a basin divide and have
similar basin characteristics
Watershed Info
• Since 1993, 3,023 ac of prairie planted
in Walnut Creek watershed – most
located in core of watershed between
two stream gauges (23% of watershed)
• 3.7% of watershed – rented to area
farmers
• From 1992 to 2005: row crop land use
decreased from 69 to 54% in WC and
increased from 71 to 80% in Squaw
Creek
• Nitrogen applications reduced 21%;
Pesticide use reduced by 28%
1990 Land Cover
69-71% row crop
2005 Land Cover
54.5% row crop in Walnut Creek
80.6% row crop in Squaw Creek
Nitrate Concentrations and
Loads
WNT2 range 0.5 to 14 mg/l
10000
SQW2 range 2.1 to 15 mg/l
16
W NT2
1000
12
8
D isch a rg e (cfs)
10
4
1
0 .1
0
10000
16
SQW 2
1000
12
100
8
10
4
1
0 .1
0
1996
1998
2000
2002
W a te r Y e a r
2004
N itra te -N C o n ce n tra tio n (m g /l)
100
Exceeded 10 mg/l (MCL) 32.8%
in Walnut Creek 51.5% in Squaw
Creek
Similar temporal pattern of
detection – higher in spring and
early summer
Subbasin WNT3
35.7% prairie
Subbasin WNT5
45.9% prairie
Subbasin WNT6
14.3% prairie
Annual Changes in Nitrate
20
20
20 W NT
20
3
15
15
15
10
10
10
5
5
5
N O 3 -N C o n c e n tra tio n s (m g /l)
5
W NT 1
0
2 01 5
5
W NT 2
1 01 5
SQW 1
SQW 4
SQW 2
1 51 0
51 0
10
5
5
0
52 0
SQW 5
W NT 6
2005
2003
2004
2002
2000
2001
10
1999
2005
2003
2004
2002
2000
2001
1999
01 5
1998
10
0
1997
15
1998
0W N T
1 52 0
0
20
1997
0
20
1996
N O 3 -N C o n c e n tra tio n s (m g /l)
10
1996
N O 3 -N C o n ce n tra tio n s (m g /l)
15
0
20
SQW 3
5
5
W a te r Y e a r
0
0
1996 1997 1998 1999 2000 2001 2002 2003 2004 2005
W a te r Y e a r
1996 1997 1998 1999 2000 2001 2002 2003 2004 2005
W a te r Y e a r
Conclusions from monitoring



Project results indicate that prairie
reconstruction can improve water quality in
agricultural watersheds
Many years are needed to detect changes in
nitrate due to slow groundwater flow velocities
in glacial till catchment
Much more in Schilling and Wolter’s work
Questions we would
like to address
1.
2.
Given the location of prairie restoration, what
does water quality modeling tell us about the
“prairie effect”: the impact of prairie
restoration on nutrient loadings?
If we wish to achieve nutrient loading
reductions at least cost, where should we have
put the prairie?
The “Prairie Effect”

We wish to isolate the effect of prairie
restoration



Land use has changed in the rest of the watershed,
which confounds the impact of the restoration
Create a “counterfactual” scenario by overlaying
2005 prairie area onto the 1990 land use map of
the watershed
Run the SWAT model for the actual 1990 land
use and the counterfactual to isolate the impact
of the prairie
The Prairie Effect and
Cost-Effectiveness


For example, suppose
prairie restoration is
predicted to reduce
nitrate loadings from N0
to N1
Can (could) one do
better?


Either achieve the same
level of nutrient reductions
at lower cost or
Achieve higher nutrient
reductions at the same cost
Cost
R2
Current: R
R1
N
N1
N0
Why is this important?
1.
2.
3.
We are looking for a modeling confirmation of
the effectiveness of restoration
We are looking to develop the capability to
efficiently locate future prairie restoration (or
other conservation practices) in the watershed
We are looking to inform restoration policies
elsewhere
Fundamental Questions

To select the mix and location of agricultural
conservation practices to meet water quality
improvement objectives at least cost




Here we focus on prairie restoration
What are the trade-offs between costs and water quality
improvements?
Conceptually, we wish to solve a multiobjective
problem:
min (Cost, Pollutant 1, … , Pollutant K)

Subject to

Conservation technology and physical constraints
Tradeoff Frontier:
“Conservation PPF”
Cost


The solution is a set of
prescriptions for
location of conservation
practices which yield
Pareto-efficient
outcomes in (Cost,
Pollutant 1, … ,
Pollutant K) space
For convenience, call
this frontier of outcomes
a “conservation PPF”
R2
Current: R
R1
N
N1
N0
Solution Framework





We wish to approximate the solution to:
min (Cost, Nitrate, Phosphorus)
Looking for a 3-dimensional conservation PPF
Of the 3 objectives to be minimized only cost can be
readily computed (as cost of land retirement)
Nutrient loadings need to be simulated
Combine:



An evolutionary algorithm, SPEA2
Hydrologic model, Soil and Water Assessment Tool (SWAT)
Sometimes referred to as simulation-optimization
framework
One possible watershed
configuration (a candidate solution)
d
a
b
c
a
a
d
a
a
a
Practice options = (Leave As Is,
Convert to Prairie)
Population = set of configurations
b
b
a
13 Fields
2 conservation practices
213 (8192) possible configurations
We end up with over 1300 hundred “fields”
Algorithm progression
Results

“Prairie Effect” is estimated to be:
28% reduction in Nitrate-N
 18% reduction in Total P


Preliminary findings suggest that
It could be possible to achieve the same nutrient
reductions for about 30% cheaper
 It could be possible to obtain up to an additional
14% reduction in N and 10% reduction in P for the
cost of existing prairie

Where could the prairie be located to
achieve same reductions at lower cost?
Where could the prairie be located to
achieve higher reductions at the same cost?
Preliminary Conclusions and Future Work



Preliminary modeling suggests that restoration is
indeed quite effective in reducing nutrient
loadings
We could do “better” if our only objectives were
nutrient reductions (but prairie restoration has
other goals!)
We develop a framework which
Can suggest the cost-effective placement of
additional prairie or other conservation practices in
Walnut Creek
 Can accompany future restoration efforts
