The Role of Vegetation in Minimising GW Recharge – application to mining and waste management industries Derek Eamus, C Macinnis-Ng, I Yunusa, M Zeppel.

The Role of Vegetation in The Role of Vegetation in Minimising GW Recharge – Minimising GW Recharge –

application to mining and waste application to mining and waste management industriesmanagement industries

Derek EamusDerek Eamus,,

C Macinnis-Ng, I Yunusa, M ZeppelC Macinnis-Ng, I Yunusa, M Zeppel

Terrestrial Ecohydrology Research GroupTerrestrial Ecohydrology Research Group

University of Technology, SydneyUniversity of Technology, Sydney

Outline of the talkOutline of the talk

• Why do we need to minimise GW recharge?Why do we need to minimise GW recharge?

• How can we minimise it?How can we minimise it?

• Design of store-release caps requires Design of store-release caps requires knowledge of rate of water use by knowledge of rate of water use by vegetationvegetation

• How might we model veg water use?How might we model veg water use?

Outline of the talkOutline of the talk

• A modified Jarvis-Stewart modelA modified Jarvis-Stewart model

• Using a Soil-Plant Atmosphere modelUsing a Soil-Plant Atmosphere model

• A case study of a site in NSWA case study of a site in NSW

Minimising GW recharge is Minimising GW recharge is important for:important for:

• Minimising development of dryland salinity Minimising development of dryland salinity

• Preventing leachates from waste storage Preventing leachates from waste storage dumps poisoning an aquiferdumps poisoning an aquifer

• Preventing acid drainage from mine-site Preventing acid drainage from mine-site rock dumpsrock dumps

How can we minimise How can we minimise GW recharge?GW recharge?

• Use vegetation to transpire rain back to the Use vegetation to transpire rain back to the atmosphere before it percolates beyond the atmosphere before it percolates beyond the root zoneroot zone

• Therefore need to design a “cap” on the siteTherefore need to design a “cap” on the site

• This raises many questions of design, eg: This raises many questions of design, eg: • How deep should the soil be on the clay cap? How deep should the soil be on the clay cap? • How do we know how much water the vegetation How do we know how much water the vegetation

will transpire on a daily/seasonal/annual basis? will transpire on a daily/seasonal/annual basis?

Water use by vegetation – Water use by vegetation – a remindera reminder

• Water use by vegetation is determined by:Water use by vegetation is determined by:• Solar radiation inputSolar radiation input• Soil moisture contentSoil moisture content• Atmospheric water content (humidity or Atmospheric water content (humidity or

vapour pressure deficit)vapour pressure deficit)• Leaf area index of the vegetationLeaf area index of the vegetation

• Current models require parameterisation Current models require parameterisation for each site individually – eg SPA, for each site individually – eg SPA, VADOSE, Penman-MonteithVADOSE, Penman-Monteith

• Models have a large number of input Models have a large number of input variables – SPA has about 15, VADOSE variables – SPA has about 15, VADOSE has far too many, the P-M has 6has far too many, the P-M has 6

We can model veg water useWe can model veg water use

• Parameterising models for every site and Parameterising models for every site and vegetation type is too slow and expensivevegetation type is too slow and expensive

• We have developed a modified Jarvis-We have developed a modified Jarvis-Stewart model that can be used for any Stewart model that can be used for any ecosystem dominated by woody ecosystem dominated by woody vegetationvegetation

We can model veg water useWe can model veg water use

Modelling tree water use – Modelling tree water use – the original JS approachthe original JS approach

)/1( ca

apnc GG

DGCRE

)()()( 321 fDfRfGG SMaxcc

)()(ˆ)( 321 fDfRfEE SMaxcc

Penman-Monteith Equation and Jarvis-Stewart Model1. Needs measurements of Gc

2. Circular, Complex and Time Consuming

A modified Jarvis-Stewart Model1. Measurements in Ec

2. Retains Mechanistic understanding of processes

Model Functional DependenciesModel Functional Dependencies

)exp()(ˆ 232 DkDkDf

)exp()( 22 DkDf

f1(RS )RS1000

1000 k1RS k1

f3()0

wC w1

, W,W C, C

Dependence of Gc and Ec onchanging solar radiation

Dependence of Gc on changingvapour pressure deficit

Dependence of Gc and Ec onchanging soil moisture content

Dependence of Ec on changingvapour pressure deficit

Veg Veg water use water use varies as varies as a function a function

of light, of light, VPD and VPD and

soil soil moisturemoisture

0 200 400 600 800 1000 1200 1400

0.00

0.05

0.10

0.15

0.20

0.25

0.30

0 1 2 3 4 5 6 7 8 8 9 10 11 12 13

0 200 400 600 800 1000 1200 1400

0.000

0.001

0.002

0.003

0.004

0.005

0.006

0 1 2 3 4 5 6 7 8 8 9 10 11 12 13

0.0

0.2

0.4

0.6

0.8

1.0

0.0

0.2

0.4

0.6

0.8

1.0

Summer Winter

f)e)d)

b)a)

Solar Radiation (W m-2)

Stan

d Tr

ansp

irat

ion

(mm

hr

-1)

c)

Vapour Pressure Deficit (kPa)

Soil Moisture Content (mm3 mm-3)

Solar Radiation (W m-2)

Can

opy

Con

duct

ance

(m

m h

r-1

)

Vapour Pressure Deficit (kPa)

Soil Moisture Content (mm3 mm-3)

Ec/

Ec m

ax G

c/G

c m

ax

How well does the modified How well does the modified Jarvis-Stewart model perform?Jarvis-Stewart model perform?

We compared it to the standard We compared it to the standard P-M approach and to an P-M approach and to an artificial neural network artificial neural network

statistical modelstatistical model

1 Jan 2 Jan 3 Jan 4 Jan 5 Jan 6 Jan 7 Jan

0.00

0.05

0.10

0.15

0.20

5 Feb 6 Feb 7 Feb 8 Feb 9 Feb 10 Feb 11 Feb 12 Feb

0.00

0.05

0.10

0.15

0.20

0.25

14 Jul 15 Jul 16 Jul 17 Jul 18 Jul 19 Jul 20 Jul 21 Jul

0.00

0.05

0.10

0.15

0.20

9 Sep 10 Sep 11 Sep 12 Sep 13 Sep 14 Sep 15 Sep 16 Sep

0.00

0.05

0.10

0.15

0.20

0.25

Sta

nd

Tra

nsp

irat

ion

(m

m h

r-1)

Sapflow PM Jarvis ANN

Summer Winter

• With our modified JS model, the slope of With our modified JS model, the slope of the regression for observed and modelled the regression for observed and modelled is close to oneis close to one

A A comparison comparison

of three of three sites – all sites – all

three three behave behave similarlysimilarly

Do we need to parameterise Do we need to parameterise the model independently the model independently

for each site?for each site?

• If average parameter values work, this If average parameter values work, this would be a massive saving in effort.....would be a massive saving in effort.....

The The modified JS modified JS

using using average average

parameter parameter values values

does very does very well well

(Paringa data(Paringa data))

Using an Using an averaged set averaged set of parameter of parameter values allows values allows

us to us to generate daily generate daily rates of water rates of water use from just use from just a set of met a set of met

datadata

Applying a modelling approach to Applying a modelling approach to testing the mechanismtesting the mechanism

• We applied the Soil-Plant-Atmosphere We applied the Soil-Plant-Atmosphere model of Williams model of Williams et al.et al. 2001 to the problem 2001 to the problem

• The SPA model is a detailed mechanistic The SPA model is a detailed mechanistic model that calculates C fluxes, water fluxes, model that calculates C fluxes, water fluxes, leaf water potential and GPP of landscapesleaf water potential and GPP of landscapes

The The SPASPA models water flux from models water flux from SSoil, oil, through the through the PPlant to the lant to the AAtmospheretmosphere

Plant data

Soil data

Soil water uptake

Sap flow

Leaf water potential

Met data

Stomatal

conductance

Photosynthesis

GPP

Transpiration INPUTS

OUTPUTS

The SPA The SPA model model

does well does well in in

modelling modelling sap flow at sap flow at

our siteour site

(a) Spring

0.0 0.1 0.2 0.3

Mod

elle

d sa

p flo

w (

mm

3 w

ater

hr

-1 m

m-2

grou

nd a

rea)

0.0

0.1

0.2

0.3

Regression of measured & modelled sap flow1:1 line

(b) Summer

Measured sap flow (mm3 water hr-1 mm-2 ground area)

0.0 0.1 0.2 0.3

Mod

elle

d sa

p flo

w (

mm

3 wat

er h

r-1

mm

-2 gr

ound

are

a)

0.0

0.1

0.2

0.3

Regression of measured & modelled sap flow1:1 line

(b) Summer

Day of study period

85 90 95 100 105 110 115 120

Sap

flow

(mm

day

-1)

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

1.8

2.0

2.2

ModelledMeasured

(a) Spring

Day of study period0 5 10 15 20 25 30Sa

p flo

w (m

m d

ay-1

)

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

1.8

2.0

ModelledMeasured

Veg water use Veg water use was independent was independent

of the water of the water content of the content of the

upper 80 cm of upper 80 cm of soil – highlighting soil – highlighting the importance the importance of deep roots in of deep roots in the clay layerthe clay layer

Daily

Et (

mm da

y-1

)

0.0

0.5

1.0

1.5

2.0

day - spring vs measured sap flw - spring x column 1 vs y column 1

80 cm soil moisture storage (mm)

Day of study period

0 20 40 60 80 100 120

Soil m

oistur

e (mm

)

80

120

160

200

240

280

(a)

(b)

ConclusionsConclusions

• The modified JS model allows quantification of The modified JS model allows quantification of water use from basic met data and using average water use from basic met data and using average parameter valuesparameter values

• The SPA model is a detailed model that allows us to The SPA model is a detailed model that allows us to examine the mechanisms underlying observed examine the mechanisms underlying observed behaviourbehaviour

• Management of deep drainage through vegetation is Management of deep drainage through vegetation is a realistic option for the waste and mining industriesa realistic option for the waste and mining industries

Published by CSIRO 2006 ISBN 0 643 06834 1Published by CSIRO 2006 ISBN 0 643 06834 1