Lake Pepin Watershed Full CostLake Pepin Watershed Full Cost Accounting Project: 

Optimum Landscape Allocation of C i P i f W Q liConservation Practices for Water Quality 

and Ecosystem Service Valuation.

Lake Pepin Watershed Full CostLake Pepin Watershed Full Cost Accounting Project: 

Brent Dalzell1 Steve Polasky2 Brent Dalzell , Steve PolaskyD. Mulla1, D. Pennington2, and S. Taff2 

UNIVERSITY OF MINNESOTAUNIVERSITY OF MINNESOTA1Department of Soil, Water, and Climate

2D t t f A li d E i2Department of Applied EconomicsFunding: Minnesota Pollution Control Agency

Lake Pepin (birth place of waterskiing – 1922)

Minnesotans Love Their Water Resources

 Lake Pepin Watershed

MINNESOTA

Upper Mississippi River BasinNORTH

DAKOTA

St. Croix River Basin

WISCONSIN

Minnesota River Basin

Cannon River Basin

SOUTH DAKOTA

Lake Pepin

IOWALake Pepin BasinsBASIN

Cannon River Basin

Minnesota River BasinSt C i Ri B iSt. Croix River BasinUpper Mississippi River Basin

HUC 07040001Major RiversMetro Area

Feature Area ( Kilometers )Lake Pepin Watershed 122,575 Minnesota 218,480Lake Pepin Watershed 105,368with in Minnesota

2

Minnesota Pollution Controal Agency

Sediment and phosphorus loading to Lake Pepin is occurring at an accelerated rate

SCWRS Fact Sheet 2002‐02; (Engstrom and Almendinger)

Lake Pepin TMDL: What’s the problem?

Kelley and Nater, 2000.

What s the problem?

•Minnesota River has always been a disproportionate source of sediment to Lake PepinMinnesota River has always been a disproportionate source of sediment to Lake Pepin but its role has increased:

•87 to 90% of the current sediment load originates in the MN River Basin.

TMDL GoalTMDL Goal• The Lake Pepin TMDL is expected to require from  25 50% d i f P d di f25 to 50% reductions of P and sediment from current levels

• MPCA is obligated (Minn Stats 114D 25) to• MPCA is obligated (Minn. Stats. 114D.25) to expand the scope of its TMDL analyses:– Examine the impacts of meeting TMDL goals on theExamine the impacts of meeting TMDL goals on the area’s water quality, carbon budget, habitat, and agricultural production 

b f d d h– Estimate economic benefits and costs associated with attainment of water quality standards resulting from changes in management in the Lake Pepin watershed g g p

How To Meet TMDL Goals?How To Meet TMDL Goals?

• There are many potential changes in land useThere are many potential changes in land use and land management that could meet TMDL goals

• We are assessing alternative land use/management scenarios to achieve wateruse/management scenarios to achieve water quality improvements 

• Developing a comprehensive assessment of the net economic benefits associated with alternative scenarios for achieving TMDL goals

Marine & Terrestrial Ecosystem ServicesnaturalcapitolPROJECT

• Ecosystem services are the ycontributions of nature to the provision of goods and services that contribute to human well‐beingbeing

– Provisioning (food, fuel, fiber…)

– Regulating (carbon sequestration, water quality…)Cultural (aesthetics– Cultural (aesthetics, recreation…)

Marine & Terrestrial Ecosystem ServicesnaturalcapitolPROJECT

Recreation  Sediment retention

Fisheries

Aquaculture Water purification

Crop pollinationCoastal Protection

Wave Energy

Crop pollination

HydropowerWave Energy

Aesthetic QualityAgricultural prod’n

Irrigation water

Habitat Quality

Water Quality

Flood control

Commercial timber

Carbon Sequestration

Schematic of ecosystem services research

(1) IncentivesDecisions by firms 

d i di id lPolicy 

decisions

(2) Actions

and individualsdecisions

Other Ecosystems(6) Economic

(4) Biophysical tradeoffs

Ecological

considerations Ecosystems(6) Economicefficiency

tradeoffs

(3) Ecological        production functions

Benefits Ecosystem

(3)

(5) Valuationand costs

Ecosystem services

Polasky & Segerson Annual Review of Resource Economics 1: 409‐434.

Evaluating Trade OffsgWater quality improvements

Biological habitat Carbon sequestration

Agricultural productionq p

s MP

/land

scape

veBM

e scen

arios/

500 m Nativ

Veg Bu

ffers

Alternative

–win”

5 V

A

“win –

Study Area: Small Watersheds

Important non‐field sources of sediment at the upland/ravine interface.

Soil & Water Assessment Tool (SWAT)Plant Growth, Hydrology, Erosion, Sediment and Nutrients export

SWAT inputs

SWAT d lSWAT model

Subtitle Weather Inputs•Precipitation•Precipitation•Temperature•Wind Speed

Land Cover (NLCD)Soils ‐ SSURGO p

•Relative Humidity•Solar Radiation

Digital Elevation ModelManagement Practices

“InVEST”Integrated Valuation of Ecosystem 

Services and Tradeoffs

http://www.naturalcapitalproject.org/InVEST.html

SWAT(biophysical model)

InVEST(ecosystem service and 

valuation model)

SWAT d lMarket valuation

•Crop Yield / Biomassvaluation model)

SWAT modelMarket valuation:•CropsField Sources of:

•Sediment

Non‐market valuation:•Sediment

•Phosphorus•Water

Subtitle •Phosphorus•Carbon SequestrationNon‐Field Sources of:

•Sediment•Phosphorus

Alternative Scenarios and Landscape Options

•Conservation Tillage: Chisel and disk tillage practices are replaced with a conservation tillage practice that leaves 30% residue at time of planting Field cultivators are still used before plantingtime of planting.  Field cultivators are still used before planting.

•Reduced P Fertilizer Application: Fall application of P fertilizer is reduced by 50% from current levels.  Manure application is unchanged.

•Cropland Conversion to Grassland: Biomass is harvested. Previous tile drainage systems remain intactPrevious tile drainage systems remain intact.

•Cropland Conversion to Switchgrass: Biomass is harvested. Previous tile drainage systems remain intact.

•Cropland Conversion to Forest: Previous tile drainage systems remain intact.

Model calibrationGetting the water balance right…

Getting the water balance right…8

7

Illinois ‐ FermiLab (2004‐2009)

South Dakota ‐ Brookings (2004‐2010)Grasslands

5

6 SWAT ‐Minnesota (2002‐2006)

4

ET (m

m/day)

2

3

E

1

00 50 100 150 200 250 300 350

day of year//public.ornl.gov/ameriflux

Getting the water balance right…8

7

8

SiouxFalls (2008‐2009)

FermiAg(2006,2008)

Corn

5

6 SWAT (2002‐2006)

4

ET (m

m/day)

2

3

E

1

//public.ornl.gov/ameriflux

00 50 100 150 200 250 300 350

day of year

Grasslands‐Corn ET comparison

Data from SWAT outputcomparison p

6

5 per. Mov. Avg. (SWAT Grassland (2002‐2006))

4

55 per. Mov. Avg. (SWAT Corn (2002‐2006))

3

4

(mm/day)

2

ET 

0

1

00 50 100 150 200 250 300 350

day of year

Seven Mile Creekmodel outputsmodel outputs

Site 1: 37 km2

Sit 2Site 2: 35 km2

Water/wetlandsRoadsResidential/developed

Site 3: 90 km2

ForestHay/PastureRow Crops

NLCD 2001

Flow – watershed outlet04.5 0

504.0

4.5

100

1503.0

3.5

mm)

m3/sec)

monthly precipObserved Flow ‐ site 3Predicted Flow ‐ site 3

200

2502.0

2.5

thly Precip. (m

onthly Flow (m

ValidationObserved Mean = 0.58 m3/secPredicted Mean = 0.66 m3/secNSE = 0.89

300

3501 0

1.5 Mon

t

Mean Mo NSE   0.89

350

4000.5

1.0

4500.0

Dec‐01

Mar‐02

Jul‐0

2

Oct‐02

Jan‐03

Apr‐03

Aug‐03

Nov‐03

Feb‐04

Jun‐04

Sep‐04

Dec‐04

Mar‐05

Jul‐0

5

Oct‐05

Jan‐06

May‐06

Aug‐06

Nov‐06

Feb‐07

Jun‐07

Sep‐07

Dec‐07

Apr‐08

Jul‐0

8

Oct‐08

Sediment – upland site 10

50300

350

100

150250

m)t (tons) monthly precipObserved Sediment ‐ site 1P di d S di i 1

200

250150

200

ly Precip. (m

m

dimen

t Export Predicted Sediment ‐ site 1

CalibrationObserved Mean = 29.3 tPredicted Mean = 23.1 tNSE = 0 80

ValidationObserved Mean = 7 8 t 250

300100

150

Mon

th

Mon

thly Se d NSE = 0.80 Observed Mean = 7.8 t

Predicted Mean = 10.7 tNSE = 0.47

350

40050

Overall Means:Observed = 16.75 tonsPredicted = 15.92 tons

4500

Dec‐01

Mar‐02

Jul‐0

2

Oct‐02

Jan‐03

Apr‐03

Aug‐03

Nov‐03

Feb‐04

Jun‐04

Sep‐04

Dec‐04

Mar‐05

Jul‐0

5

Oct‐05

Jan‐06

May‐06

Aug‐06

Nov‐06

Feb‐07

Jun‐07

Sep‐07

Dec‐07

Apr‐08

Jul‐0

8

Oct‐08

Sediment – watershed outlet0

506000

7000

100

1505000

m)(ton

s) monthly precipObserved Sediment ‐ site 3P di d S di i 3

200

250

4000

y Precip. (mm

imen

t Export  Predicted Sediment ‐ site 3

ValidationObserved Mean = 461 2 t 250

3002000

3000

Mon

thly

Mon

thly Sed

Observed Mean = 461.2 tPredicted Mean = 94.7 tNSE = 0.23

350

4001000

4500

Dec‐01

Mar‐02

Jul‐0

2

Oct‐02

Jan‐03

Apr‐03

Aug‐03

Nov‐03

Feb‐04

Jun‐04

Sep‐04

Dec‐04

Mar‐05

Jul‐0

5

Oct‐05

Jan‐06

May‐06

Aug‐06

Nov‐06

Feb‐07

Jun‐07

Sep‐07

Dec‐07

Apr‐08

Jul‐0

8

Oct‐08

Important non‐field sources of sediment at the upland/ravine interface.

Phosphorus – upland site 10

501200

1400

100

1501000

mm)

xport (kg)

monthly precipObserved TP ‐ site 1P di d TP i 1

Overall Means:

200

250600

800

thly Precip. (m

Phosph

orus Ex Predicted TP ‐ site 1

CalibrationObserved Mean = 138 kgPredicted Mean = 98 kgNSE = 0 90

ValidationObserved Mean = 45 kg

Observed P = 83.9 kgPredicted P = 65.1 kg

300

350

400

Mon

t

Mon

thly P NSE = 0.90 Observed Mean = 45 kg

Predicted Mean = 41 kgNSE = 0.42

400

4500

200

4500

Dec‐01

Mar‐02

Jul‐0

2

Oct‐02

Jan‐03

Apr‐03

Aug‐03

Nov‐03

Feb‐04

Jun‐04

Sep‐04

Dec‐04

Mar‐05

Jul‐0

5

Oct‐05

Jan‐06

May‐06

Aug‐06

Nov‐06

Feb‐07

Jun‐07

Sep‐07

Dec‐07

Apr‐08

Jul‐0

8

Oct‐08

Phosphorus – watershed outlet0

504000

4500

100

1503000

3500

m)ort (kg) monthly precip

Observed TP ‐ site 3P di d TP i 3

200

2502000

2500

y Precip. (mm

osph

orus Expo Predicted TP ‐ site 3

Validation 250

3001500

2000

Mon

thly

Mon

thly Pho

ValidationObserved Mean = 362 kgPredicted Mean = 259 kgNSE = 0.88

350

400500

1000

4500

Dec‐01

Mar‐02

Jul‐0

2

Oct‐02

Jan‐03

Apr‐03

Aug‐03

Nov‐03

Feb‐04

Jun‐04

Sep‐04

Dec‐04

Mar‐05

Jul‐0

5

Oct‐05

Jan‐06

May‐06

Aug‐06

Nov‐06

Feb‐07

Jun‐07

Sep‐07

Dec‐07

Apr‐08

Jul‐0

8

Oct‐08

Flow‐Sediment relationship at watershed outlet

7000

8000Watershed outlet

y = 548.02x1.9216R² = 0.98516000

tons)

4000

5000

dimen

t loa

d (

Watershed outlet

Power (Watershed outlet)

2000

3000

mon

thly sed

1000

00.0 1.0 2.0 3.0 4.0 5.0

monthly mean flow (m3 sec‐1)

Comparison of observed and regression‐predicted monthly sediment loadspredicted monthly sediment loads.

8000

Testing the regression

6000

7000

oad (ton

s)

Testing the regression

Linear (Testing the regression)

y = 1.1194x ‐ 58.94R² = 0.9729

4000

5000

ed sed

imen

t lo

2000

3000

ssion‐pred

icte

0

1000Regres

0 1000 2000 3000 4000 5000 6000 7000Observed Monthly Sediment load (tons)

SWAT‐predicted sediment from field sources07000

506000

100

1505000

mm)

ort (tons)

monthly precip

Observed Sediment ‐ site 3

200

2503000

4000

thly Precip. (m

Sedimen

t Expo

Predicted Sediment ‐ site 3

ValidationObserved Mean = 461.2 tPredicted Mean 94 5 t

300

350

2000

Mon

t

Mon

thly S Predicted Mean = 94.5 t

NSE = 0.23

350

4001000

4500

Dec‐01

Mar‐02

Jul‐0

2

Oct‐02

Jan‐03

Apr‐03

Aug‐03

Nov‐03

Feb‐04

Jun‐04

Sep‐04

Dec‐04

Mar‐05

Jul‐0

5

Oct‐05

Jan‐06

May‐06

Aug‐06

Nov‐06

Feb‐07

Jun‐07

Sep‐07

Dec‐07

Apr‐08

Jul‐0

8

Oct‐08

SWAT‐predicted phosphorus from field sources0

504000

4500

100

1503000

3500

m)ort (kg)

monthly precip

200

2502000

2500

y Precip. (mm

osph

orus Expo monthly precip

Observed TP ‐ site 3

Predicted TP ‐ site 3ValidationOb d M 362 k d di d 250

3001500

2000

Mon

thly

Mon

thly Pho Observed Mean = 362 kg

Predicted Mean = 260 kgNSE = 0.88

under predicted by 28%

350

400500

1000

4500

Dec‐01

Mar‐02

Jul‐0

2

Oct‐02

Jan‐03

Apr‐03

Aug‐03

Nov‐03

Feb‐04

Jun‐04

Sep‐04

Dec‐04

Mar‐05

Jul‐0

5

Oct‐05

Jan‐06

May‐06

Aug‐06

Nov‐06

Feb‐07

Jun‐07

Sep‐07

Dec‐07

Apr‐08

Jul‐0

8

Oct‐08

Summary of sediment and phosphorus sources7000

Sediment4000

5000

6000

7000

nt Loa

d (ton

s)

Non‐Field Sources (Regression‐predicted)

Field Sources (SWAT‐predicted)

0

1000

2000

3000

nthly Sedimen

0

Jan‐02

Apr‐02

Jul‐0

2

Oct‐02

Jan‐03

Apr‐03

Jul‐0

3

Oct‐03

Jan‐04

Apr‐04

Jul‐0

4

Oct‐04

Jan‐05

Apr‐05

Jul‐0

5

Oct‐05

Jan‐06

Apr‐06

Jul‐0

6

Oct‐06

Jan‐07

Apr‐07

Jul‐0

7

Oct‐07

Jan‐08

Apr‐08

Jul‐0

8

Oct‐08Mon

h h7000

Phosphorus4000

5000

6000

P load

 (kg) Non‐Field Sources (Regression‐predicted)

Field Sources (SWAT‐predicted)

1000

2000

3000

Mon

thly P

0

Jan‐02

Apr‐02

Jul‐0

2

Oct‐02

Jan‐03

Apr‐03

Jul‐0

3

Oct‐03

Jan‐04

Apr‐04

Jul‐0

4

Oct‐04

Jan‐05

Apr‐05

Jul‐0

5

Oct‐05

Jan‐06

Apr‐06

Jul‐0

6

Oct‐06

Jan‐07

Apr‐07

Jul‐0

7

Oct‐07

Jan‐08

Apr‐08

Jul‐0

8

Oct‐08

Average distribution of sediment and phosphorus sources

Sediment Phosphorus

field sources non‐field 

non‐ field sources

sources

field sources

sources

Data preparation schematic for optimization input

Single

SWAT‐predicted sediment loss

Adjustment based on comparison of HRU and Reach outputs. (channel deposition)

Single model unit

(HRU)

Flow‐sediment relationship

field sediment

HRU water yield

non‐field sediment

HRU data for optimization

Optimization MethodsOptimization Methods

The quest for the Efficiency Frontier

SWAT(biophysical model)

InVEST(ecosystem service and 

valuation model)

SWAT d lMarket valuation

•Crop Yield / Biomassvaluation model)

SWAT modelMarket valuation:•CropsField Sources of:

•Sediment

Non‐market valuation:•Sediment

•Phosphorus•Water

Subtitle •Phosphorus•Carbon SequestrationNon‐Field Sources of:

•Sediment•Phosphorus

Optimization Methods• The goal of the analysis is to find land‐use patterns that maximize the watershed TMDL reductions for athat maximize the watershed TMDL reductions for a given watershed economic return, and vice versa

• We combine results from the biophysical and economic models to search for these efficient land‐

ttuse patterns

• By finding the maximum TMDL reduction for a fixed• By finding the maximum TMDL reduction for a fixed economic score, and then varying the economic score over its entire potential range, we trace out the p gefficiency frontier 

Max environ.

benefitTotal economic

75% reduction

benefitreturns = $4.4 million

50% reduction ($100 ha-

1)

25% reduction

Max economic returns

Water Budget ComparisonWater Budget Comparison

Water Budget ‐ Baseline (row crop)

Water Budget ‐ Switchgrass

Surface Runoff

Lateral Soil Flow

Tile Flow

Evapotranspiration

Surface Runoff

Lateral Soil Flow

Tile Flow

Evapotranspiration

D

BaselineE

C

F = 25 m grassland buffer

C

B G = 250 m grassland buffer e sc

enar

ios

B

GH

G 250 m grassland buffer

Alte

rnat

ive

AFH

I H = High erosion areas

I = 250 m grassland buffer around wildlife refuge

Efficiency frontiers for Sed reduction constrained by market returns and by total value

Adding non-market ecosystem service value:- carboncarbon

sequestration- Sediment

reduction- Phosphorus osp o us

reduction

E

Current Market Returns

Current Market Returns + ES

Current Market Returns

Current Market Returns + ES

Baseline

D‐$1,951,710-$3,115,260

C‐$826,000-$2,206,000

C

‐$8 600$912 000

Additional content in final report:

Includes multiple landscapeB

$8,600-$912,000 Includes multiple landscape images (and valuation) for different points along various frontiers.

A$310,500-$158,000

$359,618$171,657

Efficiency frontiers for Sed reduction constrained by market returns and by total value

E DBaseline

CSeven Mile

B

Seven Mile Creek -

PhosphorusB Phosphorus

A

Efficiency frontiers for P reduction constrained by market returns and by total value

E D E DWhat’s the

C C

benefit of considering non-market ecosystemecosystem service value?

B B

A

A

Data preparation schematic for optimization input

Single

SWAT‐predicted sediment loss

Adjustment based on comparison of HRU and Reach outputs. (channel deposition)

Single model unit

(HRU)

Flow‐sediment relationship

field sediment

HRU water yield

non‐field sediment

HRU data for optimization

Changes in streamflow ‐ climate

Wang and Hejazi, WRR, 2011

Changes in streamflow ‐ humans

Wang and Hejazi, WRR, 2011

Changes in streamflow ‐ total

Wang and Hejazi, WRR, 2011

Conclusions

• Modest gains in water quality are possible i h d i iwithout reducing current economic returns (but 

standard best management practices will not achieve TMDL goals)

50% d ti ibl b t t b t ti l• 50% reductions are possible, but at substantial cost in terms of reduced economic returns

Conclusions

• The failure to incorporate the value of pecosystem services in land use planning can result in poor outcomes• Low level of ecosystem services• Low value of total goods and services from g

landscape

• Important to include value of ecosystem services in land use planning and incentives to landowners

Thank you!