Cyril Garneau1, Sauvage S. 1, Probst A.1, Lofts S.2, Tipping E.2, Brito D.3, Neves R.3, Sanchez-Perez J-M. 1

1 Université de Toulouse, CNRS, France2 Centre for Ecology and Hydrology, Lancaster Environment Center, Bailrigg, Lancaster, U.K.

3 Maretec, Instituto Superior Técnico, University of Lisbon, Portugal

Trace metals transfer in rivers: a semi-empirical formulation to describe a complex sorption – desorption process to be

implemented in SWAT model.

SWAT 2014, Pernambuco, Brazil

River = integrator of pollution

Atmospheric deposition

Urban release andindustries

Mining activities

Geological background

Trace Metal problematic in natural waters

• Exhibits a wide range of behaviour in natural waters

• Controlled by variousenvironmental factors

• difficult to model

2

SWAT 2014, Pernambuco, Brazil

Transfer of trace metals in rivers

• Advection• Dispersion• Transient storage

• Sédimentation• Érosion

𝐾𝑑 =𝑇𝑀𝑝𝑎𝑟𝑡

𝑇𝑀𝑑𝑖𝑠𝑠

µ𝑔𝑘𝑔µ𝑔𝐿

Dissolved phaseParticulated phase

Physico-chemistry

[Metal]

Nature of SSSWAT 2014, Pernambuco, Brazil

Modelling approaches : Sorption models

MINTEQA2 (Allison et al., 1991)

WASP (Ambrose et al., 1993)

WHAM (Tipping, 1993)

Trace elementFree adsorption site

Occupied adsorption site

SWAT 2014, Pernambuco, Brazil

• Different levels of modelling simplicity

– SWAT: the TM are fully dissolved and transported from the inlet to the outlet.

– Two phases separation of TM: The TM exist either in a dissolved form or in a particulate one.

– Two phases separation with dynamic equilibrium: The two TM phases can interact to maintain an equilibrium based on the physico-chemistry.

5

Modelling approach

𝑇𝑀𝑑𝑖𝑠𝑠𝑜𝑢𝑡 = 𝑇𝑀𝑑𝑖𝑠𝑠𝑐ℎ ×𝑉𝑜𝑢𝑡𝑉𝑐ℎ

𝑇𝑀𝑝𝑎𝑟𝑡𝑐ℎ = 𝑇𝑀𝑝𝑎𝑟𝑡𝑐ℎ,𝑖 + 𝑇𝑀𝑝𝑎𝑟𝑡𝑒𝑟𝑜𝑑𝑒𝑑 − 𝑇𝑀𝑝𝑎𝑟𝑡𝑑𝑒𝑝𝑜𝑠𝑖𝑡𝑒𝑑

𝑇𝑀𝑝𝑎𝑟𝑡𝑜𝑢𝑡 = 𝑇𝑀𝑝𝑎𝑟𝑡𝑐ℎ ×𝑉𝑜𝑢𝑡𝑉𝑐ℎ

𝐾𝑑 = 𝑓 𝑝ℎ𝑦𝑠𝑖𝑐𝑜 − 𝑐ℎ𝑒𝑚𝑖𝑡𝑟𝑦 =𝑇𝑀𝑝𝑎𝑟𝑡

𝑇𝑀𝑑𝑖𝑠𝑠

µ𝑔𝑘𝑔µ𝑔𝐿

𝑇𝑀𝑑𝑖𝑠𝑠𝑐ℎ =𝑇𝑀𝑡𝑜𝑡𝑎𝑙𝑐ℎ

1 + 𝐾𝑑 × 𝑆𝑃𝑀

𝑇𝑀𝑝𝑎𝑟𝑡𝑐ℎ = 𝑇𝑀𝑡𝑜𝑡𝑎𝑙𝑐ℎ − 𝑇𝑀𝑑𝑖𝑠𝑠

SWAT 2014, Pernambuco, Brazil

Objectives

• Propose an alternative modelling option for trace metal

– Use two TM phases: dissolved and particulate

– Use a dynamic Kd formulation based on available variables and quantities in SWAT

6

First step : we have to identify key parameters involved in Kd formulation

SWAT 2014, Pernambuco, Brazil

Complex modelling

MOHID- Full St-Venant equations- Advection – diffusion equations- Transient storage- Erosion – deposition equations

WHAM

- Simulates upstream river and tributaries

- Watershed dynamic outputs

- SPM concentration

- Chemical speciation of TM

- Specific reaction with organic matter

SWAT 2014, Pernambuco, Brazil

Study case

• The Garonne River sector– Reach of 80 km (watershed 13800 km²)

– Monthly average discharge 75 to 341 m3/s

• 4 campaigns in 2005 (13 sites)• Nutrients

• Major ions

• Physico-chemistry

• 15 Trace metals

SWAT 2014, Pernambuco, Brazil

Study caseChoise of two trace elements- Lead (Pb)- Arsenic (As)- Copper (Cu)

Low

wat

ers

Hig

h w

ater

s

𝑓𝐶𝑂𝑃 = min0,60

𝑆𝑃𝑀 − 1.0+ 0.021; 0.24

+

SWAT 2014, Pernambuco, Brazil

TM temporal concentration evolution

10

TM c

on

cen

trat

ion

(µ

g/l

)

Date Date

SWAT 2014, Pernambuco, Brazil

TM spatial concentration evolution

11

Low waters High waters

SWAT 2014, Pernambuco, Brazil

Key results

• The particulate phase is important to describe sources and sinks of trace metals

12

41% of copper trapped on the riverbed

76% of lead trapped on the riverbed

SWAT 2014, Pernambuco, Brazil

Key results

• Physico-chemistry influence

13

Log1

0(K

d)

and

pH F

low

(m3.s

-1)

pH

𝐾𝑑ld

𝐾𝑑𝐶𝑜

Flow

SWAT 2014, Pernambuco, Brazil

Key results

14

5.4

5.6

5.8

6

6.2

6.4

6.6

6.8

7 7.5 8 8.5 9 9.5

log1

0(K

d)

pH

Lead

Verdun

5.4

5.6

5.8

6

6.2

6.4

6.6

6.8

0 50 100 150

log1

0(K

d)

SPM

Verdun

4.3

4.4

4.5

4.6

4.7

4.8

4.9

5

5.1

7 7.5 8 8.5 9 9.5

log1

0(K

d)

pH

Copper

4.3

4.4

4.5

4.6

4.7

4.8

4.9

5

5.1

0 50 100 150

log1

0(K

d)

SPM

SPM controlled Kd

SWAT 2014, Pernambuco, Brazil

Key results

15

5.4

5.6

5.8

6

6.2

6.4

6.6

6.8

0 1 2 3 4

log1

0(K

d)

POC

Lead

Verdun

4.3

4.4

4.5

4.6

4.7

4.8

4.9

5

5.1

0 1 2 3 4

log1

0(K

d)

POC

Copper

SWAT 2014, Pernambuco, Brazil

Proposition

• Metal input generation– Anthropic sources: point-sources, atmospheric

deposition, diffuse sources (fertilizers application)– Geogenic source: Based on soil maps, residence time

and mixing models (Wade et al. 1999)

• Modelling requirements– Fixed Kd model: efficient for high Kd trace metal – Variable Kd model: efficient for low Kd– A simple empirical variable Kd model can be built on

limited inputs variables (pH, SPM and POC concentrations)

16

SWAT 2014, Pernambuco, Brazil

Thank you for your attention

SWAT 2014, Pernambuco, Brazil