Page 1
© NERC All rights reserved
Integrated Environmental Modelling applied to the Thames Basin, UK: Linking models
using OpenMI to allow multi-scale simulation of groundwater processes
Hughes (with lots of other contributors, but notably: Bricker, Jackson, Mackay, Mansour, Peach, and Upton)
OpenMI day – Delft Software Days 31st October 2014
Page 2
© NERC All rights reserved
Study Area
Page 3
© NERC All rights reserved
Hydrogeology
• Thames river basin consists of 20 groundwater systems • Focus on:
• Cretaceous Chalk • Jurassic Limestone
Page 4
© NERC All rights reserved
The Chalk Aquifer Chalk • Highly permeable with
fractures and solution enhancement
• Palaeogene in the central part leading to groundwater confinement
• Rivers are sustained by groundwater input
• Use of fully distributed groundwater model
Page 5
© NERC All rights reserved
The Limestone Aquifer Limestone
• Characterized by its complex fractured structure
• Very responsive to recharge
• Well drained by rivers • Groundwater discharge
runs over non-aquifers and onto the Chalk
• Groundwater discharge is simulated using semi-distributed groundwater model
Page 6
© NERC All rights reserved
Page 7
© NERC All rights reserved
Page 8
© NERC All rights reserved
Page 9
© NERC All rights reserved
Page 10
© NERC All rights reserved
Page 11
© NERC All rights reserved
Page 12
© NERC All rights reserved
Page 13
© NERC All rights reserved
Hydrogeological complexity
Page 14
© NERC All rights reserved
Role of the Palaeogene and karst in the Chalk groundwater system in the Pang-Lambourn
Page 15
© NERC All rights reserved
Importance of the Palaeogene deposits?
Chalk
London Clay
Lambeth Group
Page 16
© NERC All rights reserved
Colne Valley - Adits
Note alluvium, sands and gravel, and river terrace deposits
Page 17
© NERC All rights reserved
Adit Abstraction Systems Modelling Wall Hall Pumping Test
Distance from shaft (m)
Modelled levels after 100 days pumping
Page 18
© NERC All rights reserved
Cotswolds limestones
Upper Jurassic
Middle Jurassic
Lower Jurassic
Oxford Clay
Great Oolite (GO)
Inferior Oolite (IO)
Lias
N
S
LIAS
IO FE GO
FullersEarth
Page 19
© NERC All rights reserved
Great Oolite GW level Great and Inferior Oolite GW level
Geology in 3D using GSI3DTM
Page 20
© NERC All rights reserved
Conceptual understanding Cotswolds Cotswolds
Page 21
© NERC All rights reserved
• Semi-distributed model of aquifer system: single head calculated in each “bucket”.
• Designed to be incorporated into a multi-aquifer model of Thames catchment.
• Code is OpenMI compliant to link with other components.
Numbering shows ID of each “bucket” Rivers are represented as distributions of level for each “bucket”
Cotswolds – modelling approach
Page 22
© NERC All rights reserved
Linking multiple models in OpenMI
Page 23
© NERC All rights reserved
0
50
100
150
200
250
300
350
400
450
500
Jan-91 Sep-93 Jun-96 Mar-99 Dec-01 Sep-04 Jun-07 Mar-10 Nov-12
Cell 14Churn at Perrot's Brook
0
50
100
150
200
250
300
350
Jan-91 Sep-93 Jun-96 Mar-99 Dec-01 Sep-04 Jun-07 Mar-10 Nov-12
Cell 13Coln at Fossebridge
Page 24
© NERC All rights reserved
0
100
200
300
400
500
600
700
800
Jan-93 Jun-94 Oct-95 Mar-97 Jul-98 Dec-99 Apr-01 Sep-02 Jan-04 May-05 Oct-06
Cell 5Thames at West Mill Cricklade
0
1000
2000
3000
4000
5000
6000
7000
8000
Jan-93 Sep-95 Jun-98 Mar-01 Dec-03 Sep-06 Jun-09 Mar-12
Cell 1Thames at Eynsham
Page 25
© NERC All rights reserved
Need for Model Linking
A
R
Q R A
Q
GW Abstraction
Return flow to Thames at Oxford
River flow at Reading
Treated wastewater returned to Thames at Oxford
Groundwater abstraction regulated by flow at Reading
River flow
Pumping rate
50 Ml/d
150 Ml/d
Page 26
© NERC All rights reserved
River Flow
Abs
tract
ion
rate
Overland flow Recharge
Baseflow
Current Suite of Models
Chalk •2D TV distributed model •Inputs: recharge, spring flow, abstractions •Outputs: baseflow, groundwater head
Recharge •2D distributed model •Inputs: climate (rainfall and PET) •Outputs: overland flow, soil moisture, recharge
Limestone •Semi-distributed model (variably sized units) •Inputs: recharge •Outputs: baseflow, groundwater head
River Thames •1D river routing algorithm •Inputs: overland flow, baseflow •Outputs: total flow
WM module •Inputs: river flow Outputs: pumping rate, discharge to river
Page 27
© NERC All rights reserved
River Flow
Abs
tract
ion
rate
Overland flow
Recharge
Recharge
Thames
Baseflow
The Open Modelling Interface (OpenMI) Standard
Page 28
© NERC All rights reserved
Results
A
R
Q
Page 29
© NERC All rights reserved
Boreholes
• Groundwater typically exploited using them
• Complex situation – lots of mechanism
• Scale issue: regional flow 10s km, but boreholes operate on cm scale
• Challenge in terms of modelling them
Page 30
© NERC All rights reserved
Multi-scale Groundwater Modelling
Radial Flow Model ZOOMQ3D Radial-Cartesian Coupling
OpenMI
Page 31
© NERC All rights reserved
Multi-scale Groundwater Modelling
Radial Flow Model
• Finite difference approximation • Darcy-Forchheimer (non-linear flow) • Logarithmic radial node spacing • Vertical layering • Vertical & horizontal heterogeneity
• Partially or fully penetrating borehole • Borehole storage • Borehole casing & screening • Seepage face development
Page 32
© NERC All rights reserved
Multi-scale Groundwater Modelling
Radial Flow Model Radial-Cartesian Coupling
SPIDERR Flow Model •Hybrid radial-Cartesian method applied in petroleum reservoir models • Limitations of grid construction BUT • Quick, simple & user friendly
Page 33
© NERC All rights reserved
Multi-scale Groundwater Modelling
Radial Flow Model Radial-Cartesian Coupling
ZOOMQ3D OpenMI
• OpenMI standard for linking models • Data exchange maintains consistency between two models • Quick and easy to link several borehole models • Make use of existing regional models
Page 34
© NERC All rights reserved
Multi-scale Groundwater Modelling
SPIDERR Flow Model ZOOMQ3D OpenMI
Pumping Rate Parameterisation
Boundary Flows Recharge & Leakage
Page 35
© NERC All rights reserved
Study Area
Page 36
© NERC All rights reserved
Model Application
1. Calibration of SPIDERR Flow Model to pumping test data 2. Coupling of SPIDERR Flow Model with ZOOMQ3D regional model
3. Historic simulation and comparison with operational data 4. Abstraction scenarios to inform DO assessment
to Thames Water operational supply borehole to inform assessment of Deployable Output
Page 37
© NERC All rights reserved
Model Application 1. Calibration of SPIDERR Flow Model to pumping test data
Modelled and observed drawdown for a step- drawdown and constant rate test at the
abstraction borehole
Page 38
© NERC All rights reserved
Model Application 2. Coupling of SPIDERR Flow Model with ZOOMQ3D regional model
3. Historic simulation and comparison with operational data
Modelled and observed groundwater levels in the abstraction borehole
Modelled (coupled and uncoupled ZOOMQ3D ) and observed baseflow in
the River Kennet close to the abstraction borehole
Page 39
© NERC All rights reserved
Model Application 4. Abstraction scenarios to inform DO assessment
Modelled groundwater levels in the borehole under increasing rates of abstraction
Modelled baseflow in the river under increasing rates of abstraction
Updated DO assessment using modelled data suggests the borehole may be capable of sustaining
a higher rate of abstraction under drought.
Page 40
© NERC All rights reserved
Conclusions • This a demonstration of how OpenMI technology
offers flexibility to link models o Models can be of different structure (Fully
distributed versus semi-distributed groundwater models)
o Models can be of different nature (groundwater models, river models, management modules)
o Models can operate at very different scales (regional groundwater flow, river reach, borehole)
• While OpenMI technique is not straight forward to use, it is a more pragmatic approach than building one model that represents all processes