attenuation of pollutants in the hyporheic zone: river tame, west midlands michael rivett...

Attenuation of Pollutants in the Hyporheic Zone: River Tame, West Midlands

Michael Rivett M.O.Rivett@bham.ac.uk

School of Geography, Earth & Environmental Sciences

Groundwater modellers’ Forum: Groundwater – Surface Water Interaction Modelling Workshop Birmingham, UK. 28 Mar 2007

Outline

• Pollutant natural attenuation in the hyporheic zone

• River Tame field studies

• Issues / questions relevant to modelling

• Summary discussion points

Pollutant Natural Attenuation (NA)• ‘Natural Attenuation’ (NA) refers to the

“naturally occurring physical, chemical or biological processes that act within an aquifer to reduce contaminant mass, concentration, flux or toxicity” (Environment Agency, 2000)

m - Xylene

CH3

CH3

Data from: J.F. Barker (Univ. Waterloo, Canada)

Attenuation – Hyporeheic Zone (HZ)

• Infiltration of oxygenated SW

• High organic carbon, nutrients

• High microbiological activity

• Steep redox gradients• Dynamic flows

(USGS Circular 1139, 2001)

River

Hyporheic Zone – Groundwater – surface water mixing

• “Last gasp” remediation of groundwater contaminant plumes

Attenuation – Hyporeheic Zone (HZ)

• “HZ natural attenuation (NA) capacity”: Contaminant flux that can be “treated” by attenuation processes during transport through the HZ

• “Treatment”: Reduction in contaminant concentration and transformation to benign products. ( MNA strategies)

HZ: the reality is complex

Some issues:• Defining an “attenuation capacity” that is…

– Temporally and spatially predictable and sufficient

• Receptor definition: riverbed, benthic life, river …

• Scales of measurement, regulation, modelling

(Conant, 2000)

• Spatial variable

• Temporally dynamic

River Tame field studies

River Tame - Birmingham aquifer

• Rivett, M.O., Ellis, P.A., Greswell, R.B., Ward, R.S., Roche, R.S., Cleverly, M., Walker, C., Conran, D., Fitzgerald, P.J., Willcox, T., Dowle, J., in subm. Cost-effective mini drive-point piezometers and multilevel samplers for monitoring the hyporheic zone. In submittal to Quarterly Journal of Engineering Geology & Hydrogeology.

• Ellis, P.A., Mackay, R., Rivett, M.O., 2007. Quantifying urban river–aquifer fluid exchange processes: A multi-scale problem. Journal of Contaminant Hydrology 91, 51-80.

• Ellis, P.A., Rivett, M.O., 2007. Assessing the impact of VOC-contaminated groundwater on surface-water at the city scale. Journal of Contaminant Hydrology 91, 107-127.

• Rivett, M.O., Greswell, R.B., Mackay, R., Lydon, C., Conran, D.J., Ellis, P.A., 2007. Natural attenuation potential of the urban hyporheic zone: Foundational studies to the River Tame (Birmingham, UK) dipole field experiments. In: Proceedings of the First SWITCH Scientific Meeting, Birmingham, UK, 9-10 Jan 2007.

• Shepherd, K.A., Ellis, P.A., Rivett, M.O. 2006. Integrated understanding of urban land, groundwater, baseflow and surface-water quality – The City of Birmingham, UK. The Science of the Total Environment 360, 180-195.

• Ellis, P.A., 2003. The impact of urban groundwater upon surface water quality: Birmingham – River Tame study, UK. PhD thesis, School of Geography, Earth & Environmental Sciences, University of Birmingham, UK, 360 pp.

• Ellis, P.A., Rivett, M.O., Henstock, J, Dowle., Mackay, R., Ward, R. and Harris, R., 2002. Impacts of contaminated groundwater on urban river quality – Birmingham, UK. In: Groundwater quality: Natural and enhanced restoration of groundwater pollution. IAHS publication No. 275, 71-77.

• Ellis, P.A., Rivett, M.O. and Mackay, R. 2004. Estimation of groundwater-contaminant fluxes to urban rivers. In: Hydrology: Science & Practice for the 21st Century, British Hydrological Society (Publ.) 272-279.

Triassic Sandstone aquifer effluent to 7km reach of the

River Tame

• Qmin~180 Ml/d• Baseflow ~7% Qtotal

• Quality Class E/F

• K ~ 2 m/d• n ~ 0.27• Sy ~ 0.1• Chlorinated solvents,

metals

Methods:

• Drivepoint piezometers

• Drivepoint multilevel samplers

(c) (d)

(e) (f)

(a) (b)

Methods:

MLS Transect #1

(g) (h)

(i) (j)

Issue 1: assessment scale(s)

• Field investigation– City reach 7 km– 400 m reach– 50 m reach– Single multilevel profile

• Attenuation process investigation• Regulatory compliance• Up-scaling issues• ? Models developed at appropriate

scale(s)

0.1

1

10

100

5 11 12 4 2 1 13 14 6 19 15 10 8 9 16 7 17 18

Profile number

Con

cent

ratio

n (u

g/l)

PCE

TCE

cDCE

tDCE

1,1-DCE

1,1,1-TCA

1,1-DCA

TCM

0.1

1

10

100

5 11 12 4 2 1 13 14 6 19 15 10 8 9 16 7 17 18

Profile number

Con

cent

ratio

n (u

g/l)

PCE

TCE

cDCE

tDCE

1,1-DCE

1,1,1-TCA

1,1-DCA

TCM

0.5 m below riverbed

0.1

1

10

100

5 11 12 4 2 1 13 14 6 19 15 10 8 9 16 7 17 18

Profile number

Con

cent

ratio

n (u

g/l)

PCE

TCE

cDCE

tDCE

1,1-DCE

1,1,1-TCA

1,1-DCA

TCM

0.1

1

10

100

5 11 12 4 2 1 13 14 6 19 15 10 8 9 16 7 17 18

Profile number

Con

cent

ratio

n (u

g/l)

PCE

TCE

cDCE

tDCE

1,1-DCE

1,1,1-TCA

1,1-DCA

TCM

0.5 m below riverbed

Profile Number

Bank 1

Bank 2

VOCs in groundwater baseflow (7 km reach)

7 km Reach

400 m reach scale

• TCA is parent solvent– DCE is abiotic degradation product– DCA is biotic degradation product

0.1

1

10

100

1000

375350325300275250225200175150125100755025

Longitudinal reach distance (m)

Con

cent

ratio

n (u

g/l)

1,1,1-Trichloroethane

1,1-Dichloroethene

1,1-Dichloroethane

TCADCEDCA

Cross river transect:

• Evidence of negligible (if any) NA in HZ

Modelling at various scales• Based of River Tame studies:

Ellis, P.A., Mackay, R., Rivett, M.O., 2007. Quantifying urban river–aquifer fluid exchange processes: A multi-scale problem. Journal of Contaminant Hydrology 91, 51-80.

Modelling• Examine GW-SW interations / mixing

zone• 4 models at various scales (+ aquifer

model)

Sandstone

Gravel

Made Ground

River

Cross Sectional Model

300 m wide

100 m deep

Riverbed

11 12 13 # # # # # # 96 97 98 99 100

Piezom eter P 10 Piezom eter P 11

River

Sandstone

W eathered Sandstone

River bed Sedim ents

Sands and Gravels

Made Ground

Column Number

RIVER

11 12 13 # # # # # # 96 97 98 99 100

Piezom eter P 10 Piezom eter P 11

River

Sandstone

W eathered Sandstone

River bed Sedim ents

Sands and Gravels

Made Ground

Column Number11 12 13 # # # # # # 96 97 98 99 100

Piezom eter P 10 Piezom eter P 11

River

Sandstone

W eathered Sandstone

River bed Sedim ents

Sands and Gravels

Made Ground

Column Number11 12 13 # # # # # # 96 97 98 99 100

Piezom eter P 10 Piezom eter P 11

River

Sandstone

W eathered Sandstone

River bed Sedim ents

Sands and Gravels

Made Ground

Column Number

RIVER

Sandstone

Gravel

Made Ground

River

Cross Sectional Model

300 m wide

100 m deep

Riverbed

11 12 13 # # # # # # 96 97 98 99 100

Piezom eter P 10 Piezom eter P 11

River

Sandstone

W eathered Sandstone

River bed Sedim ents

Sands and Gravels

Made Ground

Column Number

RIVER

11 12 13 # # # # # # 96 97 98 99 100

Piezom eter P 10 Piezom eter P 11

River

Sandstone

W eathered Sandstone

River bed Sedim ents

Sands and Gravels

Made Ground

Column Number11 12 13 # # # # # # 96 97 98 99 100

Piezom eter P 10 Piezom eter P 11

River

Sandstone

W eathered Sandstone

River bed Sedim ents

Sands and Gravels

Made Ground

Column Number11 12 13 # # # # # # 96 97 98 99 100

Piezom eter P 10 Piezom eter P 11

River

Sandstone

W eathered Sandstone

River bed Sedim ents

Sands and Gravels

Made Ground

Column Number

RIVER

Model 2: 300 m transect

Issue #2 How to improve our understanding of flow nuances

• Dominance of lateral flows into the river through the river sides, e.g. central 50% of the river bed contributes only 25 % of total baseflow

• River water entering the river bed during the passage of the flood wave occurs only for very short times (<10 mins) and in negligible quantities - aquifer head changes derive mostly from damming of water entering the river rather than flow reversal

• After passage of flood wave a persistent (rather than large short-term) release of groundwater occurs as the stored water progressively drains to the river

• Vertical pressure gradients through the bed of the river are higher at the river channel edge than near the centre

• Local scale geological heterogeneity (particularly clay lenses) substantially influences the flow geometry beneath the river.

-1.6

-1.4

-1.2

-1.0

-0.8

-0.6

-0.4

-0.2

0.0

Sp

ecif

ic d

isch

arge

to

the

rive

r m

3d

-1

(neg

ativ

e in

dic

ates

dis

char

ge

to t

he r

iver

)

0

10

20

30

40

50

60

70

80

90

100

Dry Weather Flow Conditions, R24

Seepage face

Cumulative discharge (2nd y axis)

River boundary cells in order starting at the top of the Eastern bank down across the river bed and up to the top of the Western bank, breaks between vertical and horizontal faces.

River Bank (East) River Bank (West)River Bed

Cell DimensionsHorizontal 0.3 mVertical 0.1 m

River boundary cells

North Bank

South BankRiverbed

Surface water – Groundwater mixing• River Tame chloride: a convenient

tracer.0

0.2

0.4

0.6

50 100 150 200

De

pth

(m

)

Chloride (mg/l)

23-Jul-06 (MLS-1)

02-Aug-06 (MLS-1)

08-Aug-06 (MLS-1)

14-Aug-06 (MLS-2)

0

0.2

0.4

0.6

50 100 150 200 250 300

De

pth

(m

)

Chloride (mg/l)

23-Jul-06 (MLS-3)

08-Aug-06 (MLS-3)

14-Aug-06 (MLS-3)

Proposed

dipole site

Transect 8

Transect 11

T8-ML1T8-ML3

T11-ML2T11-ML4

Riverbed monitoring network

Proposed

dipole site

Transect 8

Transect 11

T8-ML1T8-ML1T8-ML3T8-ML3

T11-ML2T11-ML2T11-ML4T11-ML4

Riverbed monitoring network

• Test hypothesis that local-scale mixing can be explained by bed material heterogeneity and undulations in bed form (Conant, 2004)

• 10-m reach model populated with slug test hydraulic conductivity data. The modelling indicated that:– a laterally persistent shallow HZ of less than 10 cm only should exist; – where flows are concentrated, e.g., near river edges, HZ existence is

very unlikely– HZ is physically created if the bed material is protected from significant

vertical flow.

• These results are not entirely consistent with field-observed deeper mixing zones, may be due to:– low permeability structures deeper in the profile are having a significant

influence in generating the apparently high hydraulic gradients; – other processes are operative, e.g. in the summer aquatic plants are

more common and may induce locally steeper river gradients; – upper layer material structure is radically different to that modelled, e.g.

due to bioturbation or more likely river reworking during the commonly observed armouring process leading to the winnowing of fine material in the near riverbed surface.

Surface water – Groundwater mixing

• Impacts on the local mixing in the river bed may occur due to – flow variations arising from rapid pressure perturbations at the base of the river

caused by turbulent eddies and surface wave forms (with time periods on the order of a second).

– Penetration of pressure transients to reasonable depth (up to 0.5 m) in the river bed on these timescales may be hypothesized and that enhanced chemical diffusion may operate in this zone as a consequence of fluctuating velocity induced mixing.

• Whether such pressure transients can operate at levels that may cause enhanced diffusion and thereby enhance the depth of penetration of the stream aquifer mixing zone was evaluated my momentum exchange modelling.

• The results confirmed that: – The momentum exchange increases the depth of penetration of reverse flows

into the river bed although the magnitudes are lower. – The flow reversals extend to a depth of 50 cm for the same system with

momentum suggesting that enhanced diffusive mixing into the river bed profile could occur against a general upflow.

– This does lend support to the concept that small scale pressure transients could be influential in increasing the mixing at depth which when combined with flow mixing created through the heterogeneity patterns in the river bed may explain HZ (mixing zone) thicknesses that are observed to be much greater than ~10 cm.

Surface water – Groundwater mixing

Issue #3: Complexity, heterogeneity & predictability of attenuation capacity (reactions)

Attenuation of chlorinated VOCs• Anaerobic or aerobic pathways (5 sub-

types)• DOC, DO, Eh, H2, pH, bacteria type, electron

acceptor(s)

0

50

100

150

200

-105 -75 -45 -15 15 30 45

Conc

entr

ation

(μg/

l)

Distance downstream from Transect #1 (m)

TCE

cDCE

VC

150 m Reach

0

200

400

600

800

0 1.67 3.33 7.5 10 15 20 25 30 35 40 45 50

Conc

entr

ation

(μg/

l)Distance downstream (m)

TCE

0

200

400

600

800

0 1.67 3.33 7.5 10 15 20 25 30 35 40 45 50

Conc

entr

ation

(μg/

l)

Distance downstream (m)

cDCE

0100200300400500600700800

0 1.67 3.33 7.5 10 15 20 25 30 35 40 45 50

Conc

entr

ation

(μg/

l)

Distance downstream from Transect #1 (m)

VC

50 m Sub-reach

0.1 m

0.5 m Depth

profile0.3 m

Geologic controls ?

• Grain-size distribution of surface sediments

cis-DCE/TCE ratios with depth

0

10

20

30

40

50

0 5 10 15 20

cis-DCE/TCE concentration (ug/L)

De

pth

be

low

riv

erb

ed

(c

m)

ML 1

ML 3

ML 4

ML 5

ML 7

ML 12

ML 14

ML 2

TCE conc with depth

0

10

20

30

40

50

60

0 500 1000 1500

TCE conc (ug/L)

De

pth

be

low

riv

erb

ed

(c

m)

ML 1

ML 4

ML 5

ML 8

ML 2

• ~ 120 sample points in riverbed

Cross river transect

TCE cDCE VC

-0.1

0

0.1

0.2

0.3

0.4

0.5

0.6

0 1 2 3 4 5 6 7 8 9 10

Distance from river bank (m)

De

pth

be

low

riv

erb

ed

(m

)

477 73 1

757 102 8

275 241 54

35 323 191

2 71 132

0.9 nd nd

523 56 nd

99 11 nd

33 2 nd

27 3 nd

19 2 nd

nd nd nd

4 1 nd

1 3 nd

5 2 nd

4 2 nd

4 2 nd

nd nd nd

nd 1 nd

nd 1 nd

nd 2 nd

nd 1 nd

nd 1 nd

nd nd nd

nd nd nd

nd nd nd

nd nd nd

1 nd nd

1 1 nd

nd nd nd

Concentration (μg/l): TCE cDCE VC

Issue #3: Complexity, heterogeneity & predictability of attenuation capacity (reactions)

• Sophisticated kinetic, multi-process, NA models exist as research tools (e.g. Gerhard, Univ. Edinburgh)

• In practice, modeling of HZ natural attenuation is fairly limited (simple half-life approach?)

• How sophisticated do our (site) NA models need to be ?

Issue #4: By passing high attenuation zones

• Contaminated groundwater may by-pass “high attenuation capacity” zones (silty/organic carbon-rich)

• Majority of flow may be via more permeable “low-attenuation capacity”, zones (sands and gravels) of low residence time

Modelling must have a role here ?

Issue #5: Prediction of NAPL migration

• Little multi-phase flow modelling has been applied to the groundwater – surface-water interface Temporal behaviour NAPL property (density, viscosity...)

controls Preferential pathways

Issue #6: Communication of modellers with the “rest of hydro-world”

You want what in your model ?

Attenuation of Pollutants in the Hyporheic Zone

Summary of modelling issues

#1 - Appropriate scale(s) for the problem(s)#2 - Flow nuances: complexity required#3 - Attenuation reaction: complexity required#4 - By passing of high attenuation zones #5 - NAPLs: controls on migration#6 - Communication of model needs/outputs