Integrated Modelling as a Tool for Assessing Groundwater Sustainability under Future Development and Drought
1. 1
Integrated Modelling
as a
Tool for Assessing
Groundwater Sustainability
under Future Development and Drought
in
York Region, Ontario, Canada
CWRA 2014
Earthfx Incorporated
Toronto, Ontario, Canada
2. 2
Presentation Objectives
► Overview of Presentation:
Overview of Study Area
Technical background, goals and challenges
Modelling Approach
Modelling Result Highlights
► Emphasis on the unique technical aspects of this project
► Special thanks to all the staff at Earthfx and our study team partners
for their efforts on this project.
3. 3
Region of York Study Area
► Region of York
Population 1.03 million (2011)
840,000 urban residents
► West Holland Marsh Ag. Area
40% Marsh
60% Agriculture (3x more
productive/acre than Ontario average)
► Study area ranges from highly urban
to highly productive farmland
► York Municipal Water Supply
41 York Municipal Wells
19 Other Municipal Wells
► Key geologic features:
Oak Ridges Moraine
Subglacial tunnel valley systems
4. 4
Tier 3 Water Quantity Risk Assessment Objectives
► Evaluation of 4 sub-watersheds
identified at the Tier 2 stress level
► Delineation of Vulnerable Areas
WHPA-Q1/Q2
► Risk Assessment/Wellfield
Sustainability Scenarios
Existing Land Use and Takings
Allocated Demand and Future
Land Use
Drought Conditions –
Existing/Future
► Impacts on Other Uses
Cold Water Streams and Wetlands
► Significant GW Recharge Areas
Municipal Wells and Stressed Catchments
5. 5
York Region: Water Use
► Municipal Water Supply: 41% of total GW taking
41 York Municipal Wells
19 Other Municipal Wells
► Other Water Takings
248 permitted non-municipal GW combined GW/SW takings
286 non-permitted known takings
432 permitted SW takings
► All SW and GW sources simulated using actual daily values, including
peaking rates, so as to fully assess drought sustainability
6. 6
York Region Study Area Challenges
► Geologic Issues
Complex conceptual model, with erosional tunnel valley features
► Hydrogeologic Issues
Multiple aquifers with variable aquifer confinement
Over 1000 SW and GW takings
► Significant agricultural and golf course water use
► Fluctuations in municipal water use
► Surface Water and Hydrology Issues
Hummocky topography – focused recharge
Urbanization
Lowland areas with significant Dunnian GW feedback
► Integrated SW/GW issues
Significant GW/SW interaction including springs, wetlands, intermittent
reaches, and stream leakage in the welllfield areas
7. 7
York Region Model: Technical Foundation
► 2002 MOE GW Protection Fund Work produced:
ORM Database/York Region Sitefx Database
Oak Ridges Moraine Regional Model (GSC Surfaces)
YPDT “Core Model” (Earthfx Surfaces)
► 8 Layer Conceptual Model
► Steady State MODFLOW Model
Many technical insights and applications
► Since 2004
Many applications of the database, model and understanding (sewer construction, etc.)
Additional transient data compilation (York Region and PGMN network)
Evolving conceptual understanding of the till stratigraphy
Improvements in integrated modelling
2002 Models used extensively for Tier 1 and 2 SWP assessments
► 2010: Start of the Tier 3 Study
Some resistance to doing a major model update: was it necessary?
Legend:
Halton Till
Oak Ridges Complex
Northern/Newmarket Till
Thorncliff Fm.
Sunnybrook Fm
Scarborough Fm
(Note: Formation name
or equivalent)
Scale: (metres)
0 5000 10000 15000
ORM
Laurentian
River Valley
Newmarket Till
Tunnel
Channel
Thorncliff Fm
North South
Lake Ontario
North South Section:
Yonge Street
0 20000 40000
Sec tionD is tance
0100200300
Elevation
8. 8
York Tier 3: Technical Goals and Improvements
► Database Driven Integrated Modelling
Conversion of York Region GW group to a comprehensive SQLServer database
Extensive review and “mining” of reports compiled since development of the Core Model
Compilation and assessment of over 1000 surface water and groundwater takings
Compilation and calibration to over 100 million water levels, stream flow and climate
measurements
► Conceptual geologic model review and refinement:
Complete re-assessment of the shallow subsurface layering: where SW and GW interact
Subdivision of the Oak Ridges Aquifer into three layers to represent ORM silts and perched WT
Subdivision of the Newmarket Till into 3 layers
► Development of a fully integrated, fully distributed model
Hydrology: Fully distributed, dynamic simulation of 3D hydrology (precip., runoff and interflow)
► Complete simulation of focused recharge on hummocky topography of the moraine
► Snowpack simulation to evaluate spring freshet recharge processes
► Full simulation of urban development and changes in imperviousness
Hydraulics: Continuous simulation of stream network routing and GW/SW interaction throughout
the entire 4,450 km stream network
Groundwater: Actual daily SW and GW water takings, including York Region peak pumping
► In short: a significant technical leap from a steady state GW platform
9. 9
Why choose an Integrated Approach?
► Simulation of the complete water budget:
Guaranteed Accountability: All water inputs and outputs (precipitation,
SW and GW takings, streamflow and GW discharge)
Dynamics: An integrated approach is necessary because of the significant
fluctuations in the water budget elements
► Seasonal changes
► Summer daily peaking rates (pumping fluctuations)
► Growth in some areas, reductions due to new pipeline supply in other areas.
► Other water use: Complex combined SW/GW takings
► Tier 3 Applications:
Well sustainability under long term drought conditions
Full simulation of reductions in recharge, runoff and streamflow leakage
(both due to drought and urbanization)
Ecological issues – stream leakage near wellfields, wetland impacts
11. 11
Integrated GW/SW Modelling
► Water simply does not care
what we call it (SW or GW) and
it moves seamlessly between
domains
► Our experience is that
integrated modelling provides
insights that simply cannot be
obtainable from uncoupled
models
Integrated models are 10x tougher
to build, but 100x more insightful!
► Integrated modelling forces you
to look at your “blind spots”
14. 14
GSFLOW Stream
Interaction
► Streams are represented as a
network of segments or channels
Streams can pick up precipitation,
runoff, interflow, groundwater and pipe
discharges
Stream losses to GW, ET, channel
diversions and pipelines
► GW leakage/discharge is based on
the dynamic head difference
between aquifer and river stage
elevation
Similar to MODFLOW rivers, but the
stage difference is based on total
flow river level
River Loss
River Pickup
15. 15
Full Stream Network Simulation
► All streams are represented as the smallest Strahler Class 1 streams represent
the greatest total stream length and have the greatest baseflow pickup (i.e.
from springs and seeps)
Strahler
Class
No. of
Segments
Total
Length
(km)
% of Total
Length
Total
Discharge
(m3/s)
% of Total
Discharge
1 4213 2185 43% 3.65 26%
2 2118 1186 23% 2.75 19%
3 1083 832 16% 3.15 22%
4 529 431 8% 2.07 15%
5 29 266 5% 1.43 10%
6 16 112 2% 0.61 4%
7 7 66 1% 0.6 4%
Total 7995 5078 14.26
Strahler Classes Baseflow Pickup
16. 16
GSFLOW Total Flow Routing
► White-blue gradation indicates
total streamflow
Green-orange gradation indicates
topography
► All streams, including key
headwater springs are simulated
Click for Animation
17. 1717
Aquifer Head vs. Stream Stage
• Groundwater
discharging to the
stream, except during
large flow events
• Example stream gauge
18. 18
Benefits of Integrated Stream Routing
► Head dependent leakage based on total flow stream levels
In a GW only model, the leakage is based on baseflow levels only
High stream levels after a storm can drive SW into the GW system
► Upstream flow can infiltrate downstream to the GW system
Full 3D “routing” of both SW and GW
► Analysis of the entire water budget, including SW takings, SW
discharges and stream diversions
► Model calibration to a field measurable parameter (total streamflow)
No need to guesstimate baseflow
► Direct baseflow measurement is nearly impossible (seepage meters?)
► Baseflow separation is, at best, an unscientific empirical estimate
19. 19
GW Feedback:
Surface Discharge and Saturation Excess
Rejected Recharge
Soil water
Unsaturated
zone
Precipitation
Evapotranspiration
StreamStream
Evaporation
Precipitation
Infiltration
Gravity drainage
Recharge
Ground-water flow
Soil-zone base
Surface Discharge
► Surface Discharge is the movement of water from the GW system to
the soil zone, where it can become interflow or surface runoff
► Saturated soils can reject recharge: groundwater feedback
21. 21
GSFLOW Conclusions
► GSFLOW features:
Streams can be incised in the GW system layers
Interaction is conceptually similar to MODFLOW Rivers, but with total flow routing
Streams can dry up and later rewet
Every component of the stream flow can be identified and visualized
► Limitations: Stream routing simplified when compared to storm water models
Timing and channel flow representation not ideal for peak flow or flood modelling
(However, GW interaction is likely not significant during peak flow analysis)
► Overall benefits for water budgeting and cumulative impact:
Full accounting of gains and losses to the stream network
Ideal for simulation of impact during low flow conditions
Allows calibration to total measured streamflow at the gauge
► Much more direct than trying to calibrate to a baseflow estimate
22. 22
YORK TIER 3 MODEL
DEVELOPMENT AND
CALIBRATION: OVERVIEW
22
23. 23
York Tier 3 Model Development Phases
► Step 1: Steady State MODFLOW and PRMS model: initial calibration
Objective is to get the models up and running and internally consistent
► Step 2: Fully integrated transient calibration
Core calibration period included average (2006), dry (2007) and wet (2008) years
► Good water use, water levels, climate and streamflow data for calibration
Dry/Wet year transition provides insight into both seasonal and longer term storage
► Tier 3 Applications: 10 Year Drought Simulation: 1958-1967
Multiple scenarios with different takings and land use (each scenario is 1 TB in size!)
Each 10 year run is a “Scenario” with historic climate and current water taking
Results processed to evaluate both water level and stream sensitivities and Tier 3
issues
► Ecological impacts assessment of future water use and land develop
Simulation outputs include all components of accumulated total streamflow
(baseflow and runoff) throughout the entire steam network
24. 24
Conceptual Geologic Model Update
► Updated Conceptual Model:
Description accompanied by
schematics of key geologic settings
and processes
► Updated 3D model surfaces
considered:
New boreholes, seismic data,
geophysical logs
Earlier conceptual models
(GSC/CAMC/Earthfx)
► All surfaces completely re-gridded
and rebuilt, with:
ORAC silts
Upper/Lower Newmarket Till
N-S Section along Bayview Ave
25. 25
Step 1: Steady-State GW Model
► Model inputs include average
pumping at municipal and private
wells.
► Steady state recharge based on
results of long-term average of
PRMS step 1 simulation
► Model calibrated to match static
water levels in WWIS database and
average heads in wells with
continuous record.
► Model matched observed water
levels and groundwater flow
patterns well
Simulated heads in INS/Lower ORAC
26. 26
Step 1: Steady-State Baseflow Simulation
► Steady-state model only routes
baseflow
► Model was calibrated to match
estimated baseflow at EC gauges
► Red zones show areas of surface
discharge
Simulated groundwater discharge to streams and wetlands
27. 27
PRMS: 3D Hydrology Simulation
► Cascade routes overland flow and
interflow downslope to streams
Allows infiltration of run-on
► Used a modified SCS CN method for
Hortonian flow estimate
Initial abstraction calculated by PRMS.
CN values updated daily based on
antecedent moisture conditions
► Dunnian runoff calculated based on soil
moisture
Overland flow network from 100-m DEM
28. 28
Distributed Modelling - PRMS
► Soil water balance calculated on a cell-
by cell basis.
► Unique inputs for each model cell
Climate data interpolated over grid
Topography from DEM
► slope and slope aspect
► Parsimony
Regionally consistent values for
vegetative cover,
% impervious for land use classes
Regionally consistent values for soil
properties by surficial geology class
Land Use Class
Assigned to Grid
% Impervious based
on Land Use Class
29. 29
PRMS Model Results
► Model calibrated to match flows at
EC Gauges
► Daily outputs for each cell
Can be averaged monthly,
annually, and over study period
Hydrographs can be generated for
each cell.
Net Precipitation
Cascade Flow
Actual ET
GW RechargeDischarge to Streams
30. 30
Recharge Change
► Future land use
% impervious and vegetative
cover were modified
Results subtracted to show areas
with significant change to GW
Recharge and other water balance
components
Change in GW Recharge - Future Land Use
31. 31
Step 2: Integrated GSFLOW Stream Gauge Calibration
► All mapped streams in York/TRCA
area represented in model
► Model calibrated to observed total
flows measured at EC gauges
32. 32
GSFLOW Stream Response
► Gradational Stream Color: Total
accumulated stream flow along
reach
► Blue shading: Overland runoff
from rainfall events
► Animation shows headwater
tributaries flowing after a storm
and then drying up during the dry
periods
► Storm of August 19, 2005
produces large overland and
stream flows
Click for Animation
33. 33
GSFLOW Comparison to TRCA Sport Flows
► Check of simulated summer flows
to low flows measured by TRCA
in 2002
► Gradational Stream Color: Total
accumulated stream flow along
reach. Note log scale
► Colour-coded diamonds show
measured flows.
Comparison of mid-September 2005 to TRCA baseflows
35. 35
Risk Assessment: Vulnerable Areas
► Scenario G(2) looked at changes
in heads due to future pumping
(municipal and non-municipal
consumptive use)
► WHPA-Q1 defined by 1-m
drawdown from no-pumping
condition
► Simulated steady-state heads with
future pumping subtracted from
heads with no pumping. The
simulated drawdown cone is
continuous.
► Change in land use had no effect
on extent of WHPA-Q1
Maximum extent of 1-m drawdown due to all takings
36. 36
Risk Assessment Scenarios
► For example, Scenario G(2)
looked at incremental changes
in heads due to future increases
in municipal pumping
► Simulated steady-state heads
with future pumping subtracted
from heads with existing
pumping.
Extent of 1-m drawdown in the TAC
37. 37
Impact on Other Uses
► Scenario G(2) also looked at
incremental changes in baseflow
due to future increases in
municipal pumping
► Simulated baseflow with future
pumping subtracted from
baseflow with existing pumping.
► Change occurs mostly within 1-m
drawdown cone
% decrease in baseflow due to increase in municipal pumping
38. 38
Impact on Other Uses
► Changes above 20% of baseflow in
coldwater streams caused by planned
systems is considered significant risk
► Changes above 10% of baseflow in
coldwater streams caused by increase
from existing to allocated demand for
existing systems is considered moderate
risk
► Reaches with 50% decrease in flow to
warm water streams (red circle)
► Also looked at 1-m decrease in heads
below wetlands and at other permitted
takings
% decrease in baseflow due to increase in municipal pumping
39. 39
SGRA Analysis
► Tier 3 model to estimate average
groundwater recharge
► Clipped and infilled areas based on
procedures followed in LSRCA and
TRCA Tier 1 studies
SGRAs defined for LSRCA and TRCA
41. 41
Drought Analysis
► Simulations considered the
10-year drought of WY1957-
WY1966. Added two years
for model startup
► Scenario D simulated
drought with existing
pumping and land use
► Scenario H(1) simulated
drought with increased
pumping and land use
change
► Low heads in Summer 1965.
Simulated heads – Location D – Scenario D
42. 42
Drought Analysis
► Model run starts with a
steady-state Scenario C
simulation.
► Two year simulation
(average years) run to set up
transient model conditions
(i.e. get soil moisture to
average levels etc.)
► Drought reference level -
September 1956 - provides
reasonable average
conditions.
► Drawdowns are change from
simulated heads at start of
drought to heads on worst
date
Decrease in TAC heads due
to 10-year drought –
Scenario D
Decrease in TAC heads due
to 10-year drought –
Scenario H(1)
43. 43
Drought Analysis
► Also looked at changes in streamflow
under drought conditions
► Change primarily occur in headwater
streams
% decrease in streamflow due to 10-year drought
45. 45
Summary
► The York Tier 3 project is complete
with Peer Review sign-off
► Project report: 953 pages
Warning: may cause drowsiness
► The project represent a significant
improvement over the previous Core
Model, and should be an excellent
foundation for York and TRCA moving
forward.
► Special thanks to all the staff at
Earthfx, our partner agencies and peer
reveiwers!
Click for Animation
Monthly average flows – Scenario H(1)