Section 04-02 Nonpoint Source Pollution and ... - Groundwater

4/22/2015

1

Nonpoint Source Pollution Assessment:Framework, Vulnerability Analysis, and Modeling

Thomas Harter

Outline• Driver: regulatory framework

• Understanding nonpoint source transport dynamics (v. point sources)

• Vulnerability assessment

• Scoring methods

• Statistical methods

• Modeling

Dubrovsky et al., USGS, 2010 Dubrovsky et al., USGS, 2010

Nitrate: Impacted regions within the Central Valley

red dots: wells above MCL for nitrate

CVSALTS, Tasks 7 and 8 – Salt and Nitrate Analysis for the Central Valley FloorFinal Report, December 2013

Figure 7-14

All Water Systems

Estimated locations of the area’s roughly 400 regulated community public and state-documented state small water systems and of 74,000 unregulated self-supplied water systems. Source: Honeycutt et al. 2012; CDPH PICME 2010.

4/22/2015

2

Regulating Water Pollution Sources

Surface Water Quality

Ground Water Quality

Point Sources of Pollution

Nonpoint Sources of Pollution

1970s ‐ nowClean Water Act:

NPDES Permits







NPDES Permits


TMDL







NPDES Permits


TMDL

1980s ‐ nowSuperfund, TSCA, RCRA, FIFRA

RCRA Groundwater Monitoring

• Affected parties:

o TSDFs (transport, storage, and disposal facilities)• Permitted facilities vs. Interim facilities (existed prior to RCRA rules)

o MSWFs (municipal solid waste landfills)

• Detection monitoring

o 1 or more monitoring wells upgradient

o 3 or more monitoring wells downgradient

o Objective: SSI (statistically significant increase)?

• Compliance monitoring / Assessment monitoring

o Objective: groundwater protection standards exceeded?

• Corrective Action

o Treatment

o Cleanup

o Cease and desist

o ….

http://www.epa.gov/osw/hazard/tsd/td/ldu/financial/gdwater.htmhttp://www.epa.gov/solidwaste/nonhaz/municipal/landfill/financial/gdwmswl.htm

Regulatory Approaches toGroundwater Protection and Monitoring

Modified from: EOS, Transactions, AGU 2001

Regulatory Approaches toGroundwater Monitoring

from: Parker, Beth L., Cherry, John A. & Swanson, Benjamin J., 2006. A Multilevel System for High-Resolution Monitoring in Rotasonic Boreholes.Ground Water Monitoring & Remediation 26 (4), 57-73.doi: 10.1111/j.1745-6592.2006.00107

from: http://www.ems-i.com

4/22/2015

3







NPDES Permits


TMDL


Monitoring w/ Monitoring Wells

???

Modified from: EOS, Transactions, AGU 2001

Farm Contaminant Sources: Regional Scale

• Source of N (2007) in CA:o Fertilizer use (varies with farm / farming practices) 740,000 tons

o Animal Manure 240,000 tonso Septic leach fields 27,000 tons

o Irrigation water source & mgmt.

o Treated municipal effluent 31,000 tons

Farm Contaminant Sources: Farm Scale

• Sources of N:o Feedloto Lagoono Storage areaso Manured fieldso Fertilized fieldso Various cropso Septic system

Dairy Farm Contaminant Sources: Management Units

Farm Sources: Field Scale

4/22/2015

4

Why is Nonpoint Source Pollution Different from Point Source Pollution of Groundwater?

• Scale

o Millions of acres vs. 1‐10 acres

• Intensity

o Within ~1 order magnitude above MCL vs. many orders of magnitude

above MCL

• Hydrologic Function

o Recharge vs. non‐leaky

• Frequency

o Ongoing/seasonally repeated vs. incidental

• Heterogeneity & Adjacency

CHALLENGE 1: there is ALWAYS recharge in agriculture, EVERYWHERE

r=0

q

Non-recharging source, incidental release

q

Recharging source: planned/frequent release

r

Target:UST

Target:Agriculture/Land Application

Source Area of a Monitoring Wellin a Recharge Area

Aquifer hydraulic conductivity: K

Slope of the water table: i

Monitored source length, s = d * q/r

Horizontal flow: q = K * i (Darcy’s law)

Vertical flow: r (recharge)

s

r

qd

Source Area of a Monitoring Well

Source Area of a Monitoring Wellin a Recharge Area

Horizontal flow: q = K * i (Darcy’s law)

Vertical flow: r (recharge)

Monitored source length, s = d * q/r

• Recharge rate, r: 1 ft/yr = 0.003 ft/d

• Horizontal gradient, i: 0.3% = 0.003

• Length of screen below water table, d: 20 ft

• K (ft/d) ‐ q (ft/d) ‐ s (ft)

1 0.003 2010 0.03 20020 0.06 40050 0.15 1,000

100 0.3 2,000 500 1.5 10,000

Monitoring Design for Varying Water Table Depth

(a) water level high

(b) water level intermediate

(c) water level deep

gw flow

source area

source area

source area

4/22/2015

5

CHALLENGE: There is NO floating product – redefine “First Encountered Groundwater”

regionalgw flow

monitored source area (a few feet long)

(a) Screen (length ~ 20’) located at water table, but not intersecting sand layer

(b) Screen (length ~ 20’) located in sand layer

regionalgw flow

monitored source area (several hundred feet long)

loamy clay

sand

sand

sand

regionalgw flow

MW Well Design: Varying Water Table in Heterogeneous Aquifer

regionalgw flow

monitored source area

(a) Screen (length ~ 20’) located at water table, but not intersecting sand layer

(b) Screen (length ~ 20’) located in sand layer

regionalgw flow

monitored source area

loamy clay

sand

sand

sand

regionalgw flow

CHALLENGE: There is NO PLUME to chase – nitrate, nitrate everywhere, all the time!......

• Regional variations in landuse / hydrogeology

• Farm‐to‐farm differences in management

• 1 Farm = Many Sources (Management Units)

• Within‐source/field variability (soil, irrigation)

4/22/2015

6

Farm Groundwater Monitoring:Tile Drain Monitoring

V7

V8

V9

W9

W8

V1

V2

V4

V5

V6

W1

W2

W3

W4

W5

W6

W7

W27

W28

W29

V10

W31

W30

V24

V22

W11

W10

W12

V20

V21

V19V15

V16

V17

V18

V14

V13

W23

W22

W17

W16

W18

W19W15

W14

V25 Average N of two tile drains

Nov

-19

95

Fe

b-1

99

6M

ay-1

99

6Ju

l-19

96

Se

p-1

99

6Ja

n-1

99

7A

pr-

199

7Ju

l-19

97

Oct

-19

97

Jan

-19

98

Fe

b-1

99

8M

ay-1

99

8Ju

n-1

99

8A

ug

-19

98

Nov

-19

98

Fe

b-1

99

9A

pr-

199

9Ju

n-1

99

9S

ep

-19

99

Dec

-19

99

Ap

r-20

00

Jul-

200

0O

ct-2

00

0Ja

n-2

00

1A

pr-

200

1Ju

l-20

01

Oct

-20

01

Jan

-20

02

Ap

r-20

02

Jul-

200

2S

ep

-20

02

Jan

-20

03

Ap

r-20

03

sampling date

20

30

40

50

60

70

80

90

100

tota

l N [m

g/l]

Mean

Average N of monitoring wells on dairies with tile drains

No

v-1

99

5F

eb-1

99

6M

ay-

19

96

Jul-

19

96

Sep

-19

96

Jan

-19

97

Ap

r-1

99

7Ju

l-1

99

7O

ct-1

99

7Ja

n-1

99

8F

eb-1

99

8M

ay-

19

98

Jun

-19

98

Aug

-19

98

No

v-1

99

8F

eb-1

99

9A

pr-

19

99

Jun

-19

99

Sep

-19

99

De

c-1

99

9A

pr-

20

00

Jul-

20

00

Oct

-20

00

Jan

-20

01

Ap

r-2

00

1Ju

l-2

00

1O

ct-2

00

1Ja

n-2

00

2A

pr-

20

02

Jul-

20

02

Sep

-20

02

Jan

-20

03

Ap

r-2

00

3

sampling date

20

30

40

50

60

70

80

90

100

tota

l N [m

g/l]

Mean

Wells Drains

Source Area of a Barn / Irrigation Well

source area

domestic well

regional gradient

recharge

effective gw flow direction

Domestic Well Monitoring

source area

barn well /irrigation well regional gradient

recharge

effective gw flow direction

Production Well Monitoring

Cross-section

2 milesx

200 ft

Plan-view

2 milesx

2 miles


• Water flow is horizontal & vertical

• Horizontal travel distances are generally MUCH longer than travel vertical distances

• Different depths of the well screen capture different water!





4/22/2015

7






Deeper Groundwater= Older Groundwater

Recharge Rate (color map) & Reverse Particle Paths from 100m wellpoints

Groundwater Age [Years]at 30 m (top), 100 m (bottom)

Future Impacts Likely Increase:Delay of Impact due to GW Age

Age at 100 ft (30 m) depth

Age at 300 ft (100 m) depth

Harter et al., 2009

Measured Groundwater Age in Multilevel Groundwater WellsTritium/Helium‐3 Groundwater Age (2‐20 yrs)

one-half mile

N

1

2

3

4

5

6

7

1B: 2 yr

2B: 5 yr

3B: 20 yr

6B: <2 yr

4: 4-20 yr

The DairyDTW: 80-120 ft bgsWells: multi-level

Observations Young groundwater present Age increases with depth in

both multi-level wells & across the site

No significant saturated-zone denitrification in monitor wells Courtesy, Brad Esser & Jean Moran, LLNL, 2009







NPDES Permits


TMDL


4/22/2015

8

Focus: Enforcement Monitoring

Example of Working with a Regulation: Speed Limit

Management Tool:Brakes

Feedback:Speedometer

Enforcement:Radar Controls

Responsible Party:Driver


Applying Point Source Approach to Nonpoint Source:

Management Tool:$$$ “agronomic”

Feedback:missing

Enforcement:Monitoring Wells

Responsible Party:Landowner


Alternative Monitoring Approach to Nonpoint Source:

Management Tool:Water and Nutrient Management

Feedback:Nutrient/Water Monitoring

& Assessment

Enforcement:Annual Nitrogen Budget

+Management Practice

Assessment+

Regional Trend Monitoring








NPDES Permits

1980s – nowCA pesticide contamination

prevention act 2010s ‐ futureCA Porter‐Cologne:

Dairy OrderILRP/Ag OrdersCV‐SALTS


TMDL



4/22/2015

9

Assessment: Field Trials & Modeling Transport and Fate of Nitrate and Salts=> improved management practices

0

200

400

600

800

1000

1200

1993 1994 1995 1996 1997 1998 1999 2000

Crop Year

lbs

N p

er a

cre

fertilizer N (commercial)

org-N (manure)

NH4-N (manure)

Plant N uptakeN surplus

106 mg/l NO3-N

43 mg/l

10 mg/l

0

20

40

60

80

100

120

140

5/1

/93

10

/30

/93

5/1

/94

10

/31

/94

5/1

/95

10

/31

/95

5/1

/96

10

/30

/96

5/1

/97

10

/31

/97

5/2

/98

10

/31

/98

5/2

/99

11

/1/9

9

5/1

/00

10

/31

/00

Sampling Date

Ave

rag

e sh

allo

w g

rou

nd

wat

er n

itra

te-N

[m

g/l]

Measured

Modeled

VanderSchans et al., Ground Water, 2009for further publications: http://groundwater.ucdavis.edu/gw_201.htm


Alternative Monitoring Approach to Nonpoint Source:

Management Tool:Water and Nutrient Management

Feedback:Nutrient/Water Monitoring

& Assessment

Enforcement:Annual Nitrogen Budget

+Management Practice

Assessment+

Regional Trend Monitoring


Groundwater Assessment / Vulnerability Analysis => prioritize planning/enforcement

Vulnerability Analysis: Overview Vulnerability Analysis: Some Modeling Examples

Another Vulnerability Scheme: Nitrate Hazard Index

Dzurella, Pettygrove et al.,Journal Soil Water Conservvation, 2015

Based on:

SoilCropIrrigation

Nitrate Contamination Study Area

4/22/2015

10

Explaining the Mass Balance Approach to Estimate N Leaching to Groundwater

Synthetic fertilizer N+

Wastewater effluent N+

Biosolids N+

Dairy manure N+

Atmospheric deposition N+

Irrigation water N

=

Atmospheric losses N+

Harvested N+

Surface runoff N+

Leaching N to groundwater+

Storage Change in N in root zone

Mass balance requires that:

Explaining the Mass Balance Approach to Estimate N Leaching to Groundwater

Synthetic fertilizer N+

Wastewater effluent N+

Biosolids N+

Dairy manure N+

Atmospheric deposition N+

Irrigation water N

=Atmospheric losses N

+Harvested N

+Surface runoff N

After setting “storage change in N” to zero and rearranging the mass balance equation, we obtain the following formula to estimate N leaching to groundwater:

Leaching N to

groundwater-

+ +

+ +

+

-

=

+● 0.9

Scale:50 m

50 m (1 acre) scale

1940 1950 1960 1970 1980 1990 2000 2010

1M ac

2M ac

3M ac

4M ac

110,000

220,000

330,000

440,000

Cropland Area

Cropland Area(without Alfalfa)

tons N/yr

study area scalePrevious Slide: spatial resolution ‐ 50 m x 50 m (~1 acre)Below: spatial resolution ‐ study area total

Irrigation water

Atmosphere

SyntheticFertilizer

Biosolids

Effluent

Poultry, Swine

Dairy Manure

Atmosphere

Runoff

Leaching to Groundwater

Harvest

18

Total Nitrogen Inputs:420,000 tons N/yr

Total Nitrogen Outputs:420,000 tons N/yr

Scale:Study Area

4/22/2015

11

Role of thick unsaturated soils/sediments in nonpoint source transport & travel time

Alluvial Fan Stratigraphy

Modeling Approach

• Model 2 “Heterogeneous ‐SF”

– Given the information for the spatial correlation structure of the scaling factors, a value of scaling factor at each grid is generated

Results (Velocity)

Homogeneous Hetero - SF Hetero - VG

4/22/2015

12

Results (Conc.)


Nitrate in a 16 m thick alluvial unsaturated zoneFresno, California

Results (High subplot in kg N)


Input N 2400 2400 2400

Root Uptake 835 838 827Leaching to GW 956 839 916

Stored in RZ 93 95 89Stored in Deep VZ 515 565 543

Measured N = 87 Kg/ha

Predicted N (MB) = 478 kg/ha

• Depth to the water table+

• Cropland water budgets (deep percolation)+

• Soil type d

Vadose Zone Residence Time of Nitrate

Vadose Zone Travel Time Vadose Zone Travel Time

4/22/2015

13

Nonpoint Source Groundwater Modeling:Key Challenge (why MODPATH is not enough)

spatio‐temporally distributed sources:loading to water table

spatially distributed sinks: wells

High Resolution Flow Field

adaptive mesh grid refinement

Kourakos et al., Water Resour. Res, 2012Kourakos and Harter, Env. Simulation, 2014Kourakos and Harter, Comp. Geosciences, 2014

Adaptive Mesh Refinement Finite Element Grid


Water Table Distribution

Kourakos et al., Water Resour. Res, 2012Kourakos and Harter, Env. Simulation, 2014Kourakos and Harter, Comp. Geosciences, 2014 Kourakos and Harter, 2012, 2014, 2014

4/22/2015

14

Streamlines: Coarse vs. Fine Discretization

COARSEFINE

“NPS Assessment Tool”

Kourakos et al., WRR, 2012

“mSim” NPS Assessment Tool

Matlab code available at:

http://groundwater.ucdavis.edu/mSim(to be updated soon with adaptive mesh refinement code)

TRILINOS

Adaptive Mesh Refinement (DEAL.II)


Unit Response Functions: Examples

Source Loading + Transfer Function to Wells =Simulated Nitrate Concentration History

spatio‐temporally distributed sources:loading to water table

spatially distributed sinks: wells

Validation

4/22/2015

15

Predictions UsingGroundwater Nitrate Loading

Exceedance Probability,Nitrate above 45 mg/L (MCL)

Eastern Tulare Lake Basin

Deep Percolation of Salinity

N

San Joanquin river

Alluvial Fan Boundary

Location of study area.

A

A’

Zhang, 2005; Zhang, Harter, and Sivakumar, WRR 2006

Hydrogeologic cross section of study area [Belitz and Phillips,1995]


1

2

3

45

1—upper fan 2—inter-fan3—distal fan II4—distal fan I5—Sierran Sand

Multi-unit division of the study area.

Zhang, 2005; Zhang, Harter, and Sivakumar, WRR 2006 Zhang, 2005; Zhang, Harter, and Sivakumar, WRR 2006

4/22/2015

16

Upper fan realization.

Zhang, 2005; Zhang, Harter, and Sivakumar, WRR 2006 Zhang, 2005; Zhang, Harter, and Sivakumar, WRR 2006

-1

-0.5

0

0.5

1

1.5

0 500 1000 1500 2000

time

cum

ula

tive d

istr

ibution c

urv

e

averaged breakthroughcurve of wells in upfanmean+1.96stdev

mean-1.96stdev

-1

-0.5

0

0.5

1

1.5

0 500 1000 1500 2000

averaged breakthrough curve of wells ininterfanmean+1.96stdev

mean-1.96stdev

-0.5

0

0.5

1

1.5

0 500 1000 1500 2000

averaged breakthrough curve ofwells in distal fan Imean+1.96stdev

mean-1.96stdev

-0.5

0

0.5

1

1.5

0 500 1000 1500 2000

averaged breakthrough curve ofwells in distal fan Imean+1.96stdev

mean-1.96stdev

-0.5

0

0.5

1

1.5

0 500 1000 1500 2000 2500

averaged breakthrough curve ofwells in Sierra Sandmean+1.96stdev

mean-1.96stdev

0

100

200

300

400

500

0 50 100 150 200 250

depth of top screen

arri

val

tim

e o

f fi

rst

10%

mas

s

upfan

inter-fan

distal fan I

distal fan II

Sierra Sand

0

100

200

300

400

500

0 30 60 90 120

length of well screen

arri

val

tim

e o

f fi

rst

10%

mas

s

upfan

inter-fan

distal fan I

distal fan II

Sierra Sand


Breakthrough curves of salinity: probability of arrival


Microbial nonpoint source pollution

4/22/2015

17

Fallow Field During Pre‐Irrigationwith Manure

Upgradient monitoring well

Downgradient monitoring well

Groundwater flow direction

Fallow Field During Pre‐Irrigationwith Manure

Upgradient monitoring well

Downgradient monitoring well

Groundwater flow direction

Conceptual Model

Transport Equation

Advection(Flow)

Filtration

Dispersion

Modeling Source Loading

HOMOGENEOUS

HETEROGENOUS:GAUSSIAN

Modeling Source Loading

HETEROGENOUS:POISSON

(High Intensity)


(IntermediateIntensity)


(Low Intensity)

4/22/2015

18

Principle of Poisson Loading

• Defined by 4 simple Poisson distributed parameters

• High, intermediate and low density considered (10 realizations each).

• Poisson loading consistent with zones of preferential infiltration, or multiple distributed point sources

Poisson Pulse DensityParameter Description High Inter-

mediateLow

Λc Expected number of Clusters

50 25 10

γn Expected number of pulses per

cluster

10 5 2

ρr Expected radius of

pulse from cluster centre

(m)

3 3 3

φr Radius of a pulse (m)

1 1 1

Mean spatial coverage (%)

53 22 4

Heterogeneous Aquifer Model

1 of 10 random realizations

Facies‐Specific Parameters

FaciesGravel Coarse Sand Medium Sand Sandy Loam Clay

Volumetric Proportion (%)

21 17 31 26 5

Mean Length (m) 2.56 1.7 Background 2.47 1.69

Hydraulic Conductivity (m/day)

100 50 10 1 0.001

Collector diameter (mm) 10 1.0 0.3 0.03 3.3 x 10-4

Collector Efficiency ηc*

Enterococcus 1.01 0.85 1.80 14.01 716.4Escherichia Coli 5.98 3.70 5.89 25.44 1.36 x 104

Salmonella 4.38 2.75 4.44 20.38 1.23 x 104

Campylobacter 3.99 2.66 4.58 24.93 1.46 x 104

Collision Efficiency αc 4.5 x 10-5 1 x 10-4 1 x 10-3 5 x 10-3 0.5

Filtration Coefficient λ* (m-1)

Enterococcus 4.75 x 10-

3

8.93 x 10-2 6.31 2453 1.02 x 1010

Escherichia Coli 2.83 x 10-

2

0.39 20.63 4452 2.13 x 1010

Salmonella 2.07 x 10-

2

0.29 15.56 3566 1.93 x 1010

Campylobacter 1.88 x 10-

2

0.28 16.05 4363 2.29 x 1010

Heterogeneous Aquifer Model

1 of 10 random realizations

One of 14,040 hypothetical monitoring well locations within this modeling domain (excluding areas near the simulation domain boundary).

High loading rate

Medium loading rate

Low loading rate

Homogeneous Aquifer Heterogeneous Aquifer

Probability of Prevalence in Time

Homogeneous Aquifer

Heterogeneous Aquifer

Loading:Homogeneous

4/22/2015

19


Homogeneous Aquifer


Loading:Gaussian


Homogeneous Aquifer


Loading:Poisson

- high intensity

Generation of additional transport “modes” owing to downgradienttransport and secondary loading for a spatially discontinuous source. For repeated loading, observed concentration at any point is a function of the aquifer transport properties and loading density


Homogeneous Aquifer


Loading:Poisson

- high intensity


Homogeneous Aquifer


Loading:Poisson

- intermediateintensity


Homogeneous Aquifer


Loading:Poisson

- low intensity

4/22/2015

20

http://www.whymap.org/

Conclusions

• fine grained alluvial aquifers: strong attenuation of

fecal microorganism, disease outbreaks events

most likely occur from loading immediately

adjacent to wells / faulty well construction

• High permable/low attenuation strata: large

transport distances of fecal microorganisms

• Poisson pulse source term - consistent with

sporadic observations of fecal microorganisms in

groundwater, even adjacent to persistent sources of

contamination

• Further research: spatial variation in fecal sources;

remobilization of microbes at field/farm scale =>

better risk model parametrization

• Additional processes: vadose zone, non-ideal

behavior

Section 04-02 Nonpoint Source Pollution and ... - Groundwater

Documents