RiskRisk--Based Decision Analysis in Ground Water Based Decision Analysis in Ground Water Quality ManagementQuality Management

Jagath KaluarachchiProfessorCivil and Environmental EngineeringUtah State University

BackgroundBackground

Ground water is an important natural resource providing valuable water supply to most users.

Even in abundance supply, poor ground water quality is of limited use.

Ground water quality is typically affected by land use activities producing both point and non-point source pollution.

Impacts of poor ground water quality include public health effects to economic damages.

NonNon--Point Source Pollution Point Source Pollution

Common pollutants include heavy metals, nitrogen, and organic chemicals.

Common chemicals used in agricultural activities are• nitrogen in fertilizers• pesticides, insecticides, and herbicides

Unlike on-site remediation with point-sources, best management practices (BMPs) are implemented to minimize non-point source pollution.

Nitrate in Ground WaterNitrate in Ground Water

Nitrate is commonly found in ground water in background concentrations of 1 to 5 ppm.

Excessive nitrate concentration in ground water above 10 ppm (as N) can cause health impacts including the potential for cancer.

Heavy nitrate concentrations in ground water is found due to nitrogen-rich fertilizer and septic systems.

Nitrate is a concern in agro-well areas of Sri Lanka including northern and eastern coastal aquifer regions.

What is the spatial distribution of sustainable on-ground N loading to maintain public health in a system of agricultural watersheds?

Which BMPs should be considered to reduce nitrate pollution in ground water if loadings are high?

What are the individual economic costs incurred due to the adoption of each BMP?

What is the tradeoff between competing environmental and economic goals in adopting BMPs and how to prioritize the BMPs accordingly?

Research QuestionsResearch Questions

Management OptionsManagement Options

Change of land use practicesManure applicationFertilizer applicationCrop rotationBetter designed septic systems

Change of land useAgricultural to residentialAgricultural to industrial

Conceptualization and Model Development

On-Ground Nitrogen Loading

Soil Nitrogen Dynamics

Fate and Transport of Nitrate

Nitrate Leaching

•Advection•Dispersion•Reaction

Water Table

SourcesDairy manureFertilizersSeptic systemsDairy farm lagoonsWet and dry depositionLawnsIrrigation water

LossesVolatilizationRunoff

•• Mineralization• Immobilization• Nitrification• Denitrification• Plant Uptake

Water Supply Well

Conceptual Model

SoilFertilizer

Organic N

64

4

5 5

5

7

1

3

108 8

32

9

10

11

Water Table

To Ground water

1. Mineralization 2. Nitrification 3. Immobilization4. Fertilization 5. Manure Application 6. N Fixation7. Crop Residue 8. Plant Uptake 9. Denitrification10. Volatilization 11. Leaching

N2, N2O

ManurePlant Residue

AmmoniumAmmonia

Nitrate

N2

Plant Organic nitrogen

Sustainable N-loading (for each watershed)

No

Determine potential BMPs

Multi-criteria decision analysis(decision criteria, utility theory, and ranking)

Ranking of BMPs

Select BMPs

Decision A

nalysis

Simulations

Soil-N dynamics and fate & transport of NO3

Maximize N-loadings subject to health risk constraints

Existing N-loading < sustainable N-loading

On-ground N loading

Optim

ization

Yes,no action needed

Integrated Analysis

Demonstration ExampleDemonstration Example

SumasSumas--Blaine Aquifer, WABlaine Aquifer, WA

%

%

% %

%

%

%

%

%

$

$

$

$

$

$

$SumasBlaine

Lynden

Everson

Ferndale

NooksackBirch Bay

1

4

10

313

38

6

2

33

7

516

39

23

3132

15

20

30

14

36

17

26

18

118

27

29

19

28

9

12

34

24

22

37

3521 25

C

D

E F

G

H

I

B

A

US/Canada border

1 0 1 2 Miles

N

1 Bertrand2 Blaine3 Breckenridge4 California5 Cherry Point6 Dale7 Deer8 Fazon9 Fingalson10 Fishtrap

11 Fourmile12 Haynie13 Johnson14 Jordan15 Kamm16 Lake Terrell17 Lower Anderson18 Lower Dakota19 Lummi Peninsula East20 Lummi Peninsula West

21 Lummi River Delta22 Nooksack Channel (water)23 Nooksack Deming to Everson24 Nooksack River Delta25 North Fork Anderson26 North Fork Dakota27 Saar28 Sandy Point29 Schell30 Schneider

31 Scott32 Semiahmoo33 Silver34 Smith35 South Fork Anderson36 South Fork Dakota37 Swift38 Ten Mile39 Wiser Lake/Cougar Creek

Sumas-Blaine AquiferModel domain

$Cities% Boundary condition points

BackgroundBackground

Area of 963 square kmMostly agriculture but scattered residential and industrial activities. Serious nitrate contamination over the past two decades; sometimes more than 150 ppm. Low water table and high vulnerability to nitrate leaching.Heavy agricultural activities

8th in the US for dairy production5th in the world for raspberry production

Land Cover ClassificationLand Cover Classification(NLCD from the USGS)(NLCD from the USGS)

1 0 1 Miles

N

NLCD grid11 Open Water12 Perennial Ice/Snow21 Low Intensity Residential22 High Intensity Residential23 Commercial/Industrial/Transportation31 Bare Rock/Sand/Clay32 Quarries/Strip Mines/Gravel Pits33 Transitional41 Deciduous Forest42 Evergreen Forest43 Mixed Forest51 Shrubland61 Orchards/Vineyards/Other71 Grasslands/Herbaceous81 Pasture/Hay82 Row Crops83 Small Grains84 Fallow85 Urban/Recreational Grasses89 Dairy91 W oody W etlands92 Emergent Herbaceous W etlands

Selected ResultsSelected Results

NLCD class

Dai

ry m

anur

e

Wet

dep

osit

ion

Dry

dep

osit

ion

(reg

iona

l)

Dry

dep

osit

ion

(dai

ry)

Irri

gati

on

Fert

ilize

r

Law

ns

Legu

mes

Low Intensity Residential • • • High Intensity Residential • • • Commercial/Industrial/Transportation Bare Rock/Sand/Clay • • Quarries/Strip Mines/Gravel Pits • • Transitional • • • • Deciduous Forest • • Evergreen Forest • • Mixed Forest • • Shrubland • • Orchards/Vineyards/Other • • • • Grasslands/Herbaceous • • • • Pasture/Hay • • • • • Row Crops • • • • Small Grains • • • • Fallow • • • • Urban/Recreational/Grasses • • • Dairy Farms • • • • • Woody Wetlands • • Emergent Herbaceous Wetlands • •

Simulation ModelsSimulation Models

On-ground loading of N • actual land use information

Soil-N transformations• in-house model similar to the NLEAP

Flow in ground water• MODFLOW

Fate and transport in ground water• MT3D

Optimization• Genetic algorithm combined with artificial neural network

Multi-criteria decision analysis• Importance order of criteria method

OnOn--ground N Loadingground N Loading

23%

20%

9%7%

5%

4%

4%

4%

24% Bertrand

Fishtrap

Johnson

Breckenridge

Kamm

Ten Mile

South Fork Dakota

California

Remaining drainages

Dairy Manure (56%)

Fertilizers (31%)

Atmospheric deposition (7%)

Legumes (2%)

Irrigation (1%)

Dairy Lagoon (2%)

Septic Systems (1%)

Total Nitrogen LoadingTotal Nitrogen Loading

0

1

2

3

4

5

6

Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

Time (months)

Mas

s of n

itrog

en (1

0^6)

On-ground Leaching

Transient soil nitrogen balanceTransient soil nitrogen balance

0

18

36

54

72

90

108

126

Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

Time (months)

Mas

s of n

itrog

en (1

03 lbs)

0.0

0.4

0.8

1.2

1.6

2.0

2.4

2.8On-ground Leaching Recharge

Health Risks for different NHealth Risks for different N--LoadingsLoadings

5

7

9

11

13

15

1 6 11 16 21 26 31 36 41 46 51 56

Receptor

NO

3-N

(mg/

L)

Existing OptimalMCL

Existing and Existing and optimal Noptimal N--loadings from loadings from fertilizer and fertilizer and manuremanure

0.0

0.2

0.4

0.6

0.8

1.0

Four

mile

TenM

ile

Cal

iforn

ia

Sout

h Fo

rk D

akot

a

Schn

eide

r

Lum

mi P

enin

sula

Wes

t

Fish

trap

Bre

cken

ridge

Dal

e

John

son

Jord

an

Ber

trand

(Can

ada)

Fish

trap

(Can

ada)

Sum

as R

iver

(Can

ada)

Drainage

Fert

ilize

r (1

0^6

lbs

N)

ExistingOptimal

0.0

0.5

1.0

1.5

2.0

2.5

Four

mile

Cal

iforn

ia

Schn

eide

r

Fish

trap

Dal

e

Jord

an

Fish

trap

(Can

ada)

Drainage

Man

ure

(10^

6 lb

s N

)

ExistingOptimal

110.0750.075240.2960.391Johnson (Canada)

210.2400.240221.0931.406Fishtrap (Canada)290.2310.231261.1251.516Bertrand (Canada)40.0300.030130.3110.359Jordan

280.1600.160181.1051.344Johnson00.0900.090180.5670.694Dale

110.0950.107251.7502.347Breckenridge

40.3320.346271.3451.838Fishtrap00.0280.028180.0690.085Lummi Peninsula00.0830.083280.1150.159Schneider

00.0320.032241.0401.376S. Fork Dakota00.1100.110200.5230.653California170.1030.124420.6191.070Tenmile

00.0600.060220.3190.410Fourmile

Reduction (%)

SustainableExistingReduction (%)

SustainableExisting

Fertilizer Loading (x106 lbs.yr)Manure Loading (x106 lbs.yr)Drainage

Mean Cost(x106 $)DescriptionBMP

14.4Manure composting/exporting + fertilizer application reduction + adopt a feeding strategy for dairy cattle

9

2.1Manure composting/exporting + adopt a feeding strategy for dairy cattle

8

21.0Manure composting/exporting + fertilizer application reduction7

10.6Adopt a feeding strategy for dairy cattle + fertilizer application reduction

6

-1.7Adopt a feeding strategy for dairy cattle5

12.3Fertilizer application reduction4

8.7Manure composting/exporting3

43.8Dairy cattle head reduction2

0Do-nothing1

MultiMulti--criteria decision analysiscriteria decision analysis

Criteria• Summation of concentration deviations above MCL• Number of receptors exceeding MCL• Net cost• Cost per unit concentration reduction• Nitrate buildup in the ground water• Nitrogen buildup in the soil• Cumulative nitrate flux to the surface water• Nitrate leaching• Total on-ground nitrogen loading• On-ground nitrogen runoff losses• On-ground nitrogen volatilization losses

-100

100

300

500

700

1 2 3 4 5 6 7 8 9BMP

CPC

R

0

14

28

42

56

1 2 3 4 5 6 7 8 9

BMP

EMC

L

Values of Decision Criteria

Efficiency of BMPsEfficiency of BMPs

9.0

10.0

11.0

12.0

0 20 40 60 80 100 120

Time (months)

Nitr

ate

conc

entr

atio

n (m

g/L

)

12, 3, or 84567 or 9MCL

Importance Order of Criteria MethodImportance Order of Criteria Method

0.475

0.905 0.905

0.354

0.7030.780

1.0000.905

1.000

0.237

0.659

0.818

0.265

0.555

0.685

0.8890.838

0.922

0.000

0.412

0.731

0.175

0.406

0.589

0.779 0.7710.843

0.00.10.20.30.40.50.60.70.80.91.0

Alt. 1 Alt. 2 Alt. 3 Alt. 4 Alt. 5 Alt. 6 Alt. 7 Alt. 8 Alt. 9

BMP

Tot

al u

tility

scor

e

Ranking of BMPsRanking of BMPs

1419

4148

2557

5626

6265

3334

8883

7772

9991MinimumAverageMaximum

Utility Score of BMPRanking

DecisionDecision--Support SystemSupport System

BenefitsBenefits

Site-independent soil nitrogen dynamics model provides spatial and temporal distribution of nitrate leaching to ground water.

The decision model predicts the sustainable on-ground nitrogen loading that satisfies the health risk constraints.

The decision model can be used in predicting aquifer vulnerability to nitrates under a variety of land use classes and practices.

Evaluate and prioritize management options under a variety of economic and environmental decision criteria.

ConclusionsConclusions

In agriculture-dominated watersheds, high nitrate leaching is due mainly to agricultural practices. The difference between the nitrate leaching and on-ground nitrogen loading is usually substantial and signifies a soil buildup of nitrogen.

Accounting for the spatial distribution of on-ground nitrogen loadings and nitrate leaching is essential for reliable modeling of nitrate fate and transport in ground water.

The proposed integrated modeling framework allows for the accurate simulation of the outcome of the current land use practices and the proposed BMPs.