Escolar Documentos
Profissional Documentos
Cultura Documentos
In Support of
Assessment Report for Five Source Water Protection Areas (SPAs) Under
Ontario’s Clean Water Act
(Niagara, Hamilton – Halton, CTC, Trente & Quinte SPAs)
Report for Lake Ontario Collaborative (LOC)
By Dr R Dewey
Affiliation –Modelling Surface Water Ltd
45 Wigmore Drive
Toronto , Ont M4A 2E6
December 2011
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Abstract
The Lake Ontario Collaborative used the event‐based modelling approach for the identifying
significant threats to Lake Ontario drinking water intakes in the study area (Coastal Zone of Lake
Ontario) between Wellington and Niagara. Under this approach, the Source Protection
Committee (SPC) decides, based on local knowledge, what activities it wants evaluated through
modelling, to identify Threats from Activities in Intake Protection Zone 3 (IPZ3).
In this report, the various "spill scenarios" defined by the SPC, were examined to determine the
impact of the different spills on the water quality at the municipal water treatment plant intakes
along the north shore of Lake Ontario.
The spills were modelled with a calibrated three‐dimensional hydrodynamic and water quality
computer model. The MIKE‐3 software, based on the whole of Lake Ontario, was used in nested
mode to simulate the currents in the nearshore areas of interest for the Lake Ontario
Collaborative. An extensive set of trials were made to calibrate the model, both hydrodynamics
and advection‐dispersion, using observed currents from several different locations, and a
specific spill (a tritium spill). The vector variance of speed and direction or Fnorm value was to
used as the measure of agreement between the model predictions and observed data, values as
low as 0.62 (zero is perfect match) were achieved over several months of simulation results.
Temperature serial correlations as high as 0.95 were achieved.
Areas, modelled in fine detail, ranged along the Lake Ontario shoreline from the Town of
Wellington, through Cobourg, Port Hope, Durham Region, City of Toronto, Peel Region, Halton
Region, City of Hamilton, Grimsby, to St. Catharines and Niagara Region. Spill scenarios
addressed:
• Disinfection failures at the wastewater treatment plants ‐ high E. coli levels released
through outfall/diffusers,
• Tritium releases at shoreline from Pickering and from the Darlington NGS diffuser,
• Oil pipeline breaks, transporting refined gasoline, with subsequent release of Benzene
to tributaries of Lake Ontario
• Tank Farm releases of gasoline , with subsequent release of benzene to tributaries of
Lake Ontario,
• Sanitary trunk sewer breaks, which release raw sewage to tributaries of Lake Ontario,
causing impacts of E. coli and TSS,
1‐2
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
• Industrial Food processing plant, lagoon break and subsequent release of E. coli to
tributary of Lake Ontario,
• Tanker truck spill of gasoline and subsequent release of benzene to Lake Ontario,
The LOC determined that an activity poses a significant drinking water threat, “where the spill
has the potential to reach surface water intake(s) at a sufficient concentration to cause
deterioration in water quality (the impact)”. Criteria based on Drinking water standards and
other approaches were used as the threshold for determining deterioration. Peak concentration
values and general spatial persistence of pollutants were used as methods for summarizing,
impacts at intakes, in this report.
1‐3
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Table of Contents
1‐4
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
1‐5
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
1‐6
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
7.6.3 Newcastle and Bowmanville Intakes ‐ Gasoline Spills from Wilmot Graham and
Bowmanville Creeks ........................................................................................................... 7‐55
1‐7
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
List of Figures
Figure 2‐1 Lake Thermal Process.................................................................................................. 2‐3
Figure 3‐3 NOAA produced Direction TS at Pickering ‐ Blue is ADCP data, Black is Model
Prediction ..................................................................................................................................... 3‐6
Figure 3‐4 NOAA produced Speed TS at Pickering – Blue line is Model prediction, Red is ADCP
data............................................................................................................................................... 3‐7
Figure 3‐8 2 km grid of Lake Ontario with ADCP Locations.......................................................... 3‐9
Figure 3‐9 ADCP CM4 Direction comparison.............................................................................. 3‐12
Figure 3‐10 ADCP CM4 Speed comparison – many episodes of speeds above 0.5 m/s, model is
unable to reproduce these events ............................................................................................. 3‐12
Figure 3‐12 ADCP CM2 Speed – many episodes of speeds above 0.25 m/s, model is unable to
reproduce these events.............................................................................................................. 3‐13
Figure 3‐16 Pearson Isopleths ‐ MIKE 3...................................................................................... 3‐16
Figure 3‐18 Temperature Isopleths Comparison........................................................................ 3‐18
Figure 3‐19 Temperature Time Series Comparison.................................................................... 3‐19
Figure 3‐20 Surface Vector & Temperature Comparison........................................................... 3‐20
1‐8
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 3‐23 Extended 270 m Grid............................................................................................... 3‐22
Figure 4‐1 270 m Map with Intake Locations............................................................................... 4‐1
Figure 4‐2 Horgan & Harris........................................................................................................... 4‐3
Figure 4‐3 Island Intakes............................................................................................................... 4‐4
Figure 4‐5 Hamilton Intake........................................................................................................... 4‐5
Figure 4‐9 Trenton Winds Island Intakes...................................................................................... 4‐7
Figure 4‐10 Trenton Winds Clark to Oakville................................................................................ 4‐8
Figure 4‐11 Trenton & Kingston winds at Hamilton..................................................................... 4‐8
Figure 4‐12 Pearson and Trenton Airport direction time series – the 7th has a deviation and more
between the 10th and 12th........................................................................................................... 4‐9
Figure 4‐13 Pearson and Trenton Airport speed time series – no major variances are apparent. 4‐
10
Figure 4‐16 Day 7 from release ‐ current reversal – Trenton Winds.......................................... 4‐11
Figure 4‐17 Day 7 from release ‐ Pearson Winds – note slower reversal – no separation – up to
this time the plumes are very similar......................................................................................... 4‐11
1‐9
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 4‐19 Day 8 from release ‐ Pearson Winds – note small separation and slower speed of
plume.......................................................................................................................................... 4‐12
Figure 4‐20 Day 10 – Trenton Winds.......................................................................................... 4‐12
Figure 4‐21 Day 10 ‐ Pearson Winds used – plume has hardly moved compared to Trenton case
above .......................................................................................................................................... 4‐12
Figure 4‐22 Day 12 Trenton Winds Plume impacting Clark, Lakeview, Lorne Park and Oakville4‐13
Figure 4‐23 Day 12 Pearson Winds, plume seems stalled.......................................................... 4‐13
Figure 4‐24 Day 13 Trenton winds used – plume at Hamilton................................................... 4‐13
Figure 4‐26 Etobicoke Compass Rose......................................................................................... 4‐16
Figure 4‐27 Pickering Compass Rose.......................................................................................... 4‐17
Figure 4‐28 90m nested grid of Inner Harbour with ADCP locations......................................... 4‐18
Figure 4‐29 Western Gap Surface Direction............................................................................... 4‐19
Figure 4‐31 Western Gap Surface Speed.................................................................................... 4‐20
Figure 4‐37 GP Surface Directions.............................................................................................. 4‐22
Figure 4‐40 GP 3m deep Speeds................................................................................................. 4‐23
Figure 4‐41 Surface Temperatures 2007.................................................................................... 4‐23
1‐10
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 4‐45 Gibraltar Point ADCP Location Speed Time Series in Oshawa‐Hamilton Main Spill G 4‐
29
Figure 4‐46 Gibraltar Point ADCP Location Direction Time Series in Oshawa‐Hamilton Main Spill
Grid ............................................................................................................................................. 4‐29
4‐47 MOE ADCP Locations for Credit River nested Grid Calibration.......................................... 4‐30
Figure 4‐48 #3541 speed time series......................................................................................... 4‐30
Figure 4‐49 #3541 direction time series..................................................................................... 4‐31
Figure 4‐51 #3223 direction time series at surface layer........................................................... 4‐32
Figure 4‐54 80 layer surface TS for speed ADCP 1269 ‐ Grimsby............................................... 4‐33
Figure 4‐61 #3172 speed time series......................................................................................... 4‐37
Figure 4‐62 #3172 direction time series.................................................................................... 4‐37
Figure 5‐2 Ajax Intake impact....................................................................................................... 5‐2
1‐11
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 5‐4 Oshawa Intake impact................................................................................................. 5‐3
Figure 5‐7 Extended Easterly 2430 grid........................................................................................ 5‐6
Figure 5‐8 Extended 810 m grid with intakes............................................................................... 5‐6
Figure 5‐9 Extended 270m grid with Intakes................................................................................ 5‐6
Figure 5‐13 Bowmanville Intake Plume........................................................................................ 5‐8
Figure 5‐14 Newcastle Intake Plume............................................................................................ 5‐8
Figure 5‐17 Cobourg Intake Plume............................................................................................... 5‐9
Figure 6‐1 Lake Ontario 2430m grid............................................................................................. 6‐2
Figure 6‐4 Predicted current direction at Outfall......................................................................... 6‐3
Figure 6‐5 Predicted current direction at Intake.......................................................................... 6‐4
Figure 6‐6 Predicted E. coli levels at Intake.................................................................................. 6‐4
1‐12
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 6‐11 E. coli with Decay for East Intakes............................................................................. 6‐7
Figure 6‐12 Ammonia for West Intakes........................................................................................ 6‐7
Figure 6‐13 E. coli for West Intakes.............................................................................................. 6‐7
Figure 6‐15 Easterly extent for E. coli........................................................................................... 6‐8
Figure 6‐18 Westerly extent of E. coli with decay........................................................................ 6‐9
Figure 6‐19 Lakeview E. coli time series ‐ Winter Period – note smaller scale on right axis for
Oakville and Lorne Park.............................................................................................................. 6‐10
Figure 6‐20 Lakeview E. coli time series ‐ Summer Period – note smaller scale on right axis is for
Oakville and Lorne Park.............................................................................................................. 6‐10
Figure 6‐22 Horgan Intake E. coli levels – predicted level in black, observed levels in red (FC) and
blue (E. coli) ................................................................................................................................ 6‐18
Figure 6‐24 TSS time series for Mega event............................................................................... 6‐21
Figure 6‐28 West Intakes TSS time series................................................................................... 6‐24
1‐13
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 6‐29 E. coli time series..................................................................................................... 6‐27
Figure 6‐32 Grimsby impact at 100#/100mL.............................................................................. 6‐28
Figure 7‐2 Benzene time series at intakes.................................................................................. 7‐32
Figure 7‐4 Oakville Tank Spill ‐ 0.05 mg/L Benzene Isopleth...................................................... 7‐33
Figure 7‐11 Benzene time series for Don River spill................................................................... 7‐37
Figure 7‐18 Benzene Plume at Hamilton Intake......................................................................... 7‐43
1‐14
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 7‐20 Burlington, Burloak & Oakville Intakes time series................................................. 7‐45
Figure 7‐21 Easterly Plume July 15............................................................................................. 7‐46
Figure 7‐22 Easterly Plume July 16............................................................................................. 7‐47
Figure 7‐24 Easterly benzene time series................................................................................... 7‐48
Figure 7‐25 Port Hope & Cobourg Intake Map........................................................................... 7‐49
Figure 7‐27 Port Hope May 18 event ‐ the yellow scale is 2.3 to 2.6 mg/L Benzene................. 7‐51
Figure 7‐28 Spatial Extent of Spill............................................................................................... 7‐51
Figure 7‐30 Cobourg Intake Impact from Ganaraska event ‐ the red scale is above 0.89 mg/L
Benzene ...................................................................................................................................... 7‐52
Figure 7‐31 Cobourg Intake Impact from Cobourg Brook.......................................................... 7‐53
Figure 7‐33 Spatial Extent of Cobourg Brook Spill...................................................................... 7‐54
Figure 7‐36 Bowmanville to Newcastle Grid Map...................................................................... 7‐55
1‐15
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 7‐44 Bowmanville Intake ‐ Impact from Wilmot Creek................................................... 7‐59
Figure 7‐47 Newcastle Intake plume.......................................................................................... 7‐61
Figure 7‐50 Bowmanville Intake Plume from Graham Creek..................................................... 7‐62
Figure 7‐53 Lakeview time series from Credit River................................................................... 7‐64
Figure 7‐56 Lorne Park surface plume........................................................................................ 7‐65
Figure 7‐57 Clark surface plume................................................................................................. 7‐65
Figure 7‐58 Humber River Spill time series – note smaller scale on right axis for Harris and Island
Deep intakes ............................................................................................................................... 7‐66
Figure 7‐59 Don River Spill time series – note smaller scale on right axis for Clark, Lakeview and
Island Deep intakes .................................................................................................................... 7‐67
Figure 7‐63 Duffins Creek Spill time series................................................................................. 7‐71
Figure 7‐64 Intakes and 16 Mile Creek Locations....................................................................... 7‐72
Figure 7‐65 16 Mile Creek time series........................................................................................ 7‐72
1‐16
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 7‐66 Oshawa Creek Grid.................................................................................................. 7‐73
1‐17
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
List of Tables
Table 3‐1 Fnorm Scores ................................................................................................................ 3‐8
Table 3‐2 FNORM Scores for MIKE‐3.......................................................................................... 3‐11
Table 3‐6 Roughness Values....................................................................................................... 3‐27
Table 4‐1 Percent of time Posted............................................................................................... 4‐25
Table 6‐3 Skyway & Woodward Impacts.................................................................................... 6‐14
Table 6‐4 Cobourg, Port Hope, Corbett, Harmony and Courtice Impacts.................................. 6‐15
Table 6‐7 Mega Event Peak Levels for STS Breaks Occurring Randomly Over the Summer Period
.................................................................................................................................................... 6‐24
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Table 7‐1 Oakville Fuel Storage Tank Farm Benzene Spill.......................................................... 7‐31
Table 7‐4 Peak Level Benzene from Spill location at Intake (Units are mg/L) and Duration of
typical event with Benzene above 0.005 mg/L .......................................................................... 7‐75
GLOSSARY OF TERMS
ADCP – Acoustic Doppler Current Profiler, a meter that measures current velocity at variable
depths.
Algorithm ‐ A sequence of instructions, calculations or logical decisions used to define a
step‐by‐step procedure for solving a problem, often used for calculation and
data processing.
Alongshore Currents ‐ Currents flowing approximately parallel to the shoreline.
Bathymetry ‐ Depth measurements of a body of water at various points which define the
bottom surface.
Coastal Zone ‐ the area from the shoreline to the 30 m depth Contour.
Diffuser ‐ A submerged structure consisting of a manifold with many ports through
which the effluent is discharged as turbulent jets into the receiving water at
high velocity to promote initial mixing.
Dispersion ‐ Spreading of mass during transport resulting from both physical mixing and
molecular diffusion from areas of high concentration to low concentration.
Downwelling ‐ an effect of stratified layers – warm surface water is pushed downward, opposite
to an upwelling event.
Point Source ‐ A pollution source occupying a very small area and having very little initial
dilution.
Residence Time ‐ the time required to replace a volume of water, sometimes referred to as
Flushing time, usually based on the water flowing into a system and the system volume.
Assumes plug flow and no recirculation or back missing.
Time of Travel ‐ the time required for a particle of water to travel from point A to B.
TSS – Total Suspended Solids (mg/L)
1‐19
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Upwelling‐ an effect of stratified layers – cold water from lower layers are pushed upward and
on shore – usually the results a strong wind along the south shore of the lake. Internal waves
can also cause them.
Units
° = degrees of angle
°C = degrees Celsius
cm/s = centimetres per second
g/cm/day = grams per centimetre per day
g/m³ = grams per cubic metre
g/s = grams per second
hrs = hours
km = kilometres
m = metres
m/s = metres per second
m² = square metres
m²/s = square metres per second
m³ = cubic metres
m³/s = cubic metres per second
m³/d = cubic metres per day
μg/L = micrograms per litre
Bq/L = Becquerels per litre
ppm = parts per million
units/L = arbitrary units per litre
1‐20
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
1 Introduction and Model Description
1.1 Introduction
The Lake Ontario Collaborative used the event‐based modelling for the identification of
significant threats to Lake Ontario drinking water intakes in the study area. This approach and
the applicable guidance is further described in the Appendix of the Assessment Report
(CTC,2011b). Under this approach, the Source Protection Committee decides, based on local
knowledge, what activities it wants evaluated through modelling to identify Threats from
Activities in Intake Protection Zone 3.
The Lake Ontario Collaborative used an impact assessment method to determine if an activity
poses a significant drinking water threat by determining “whether a spill has the potential to
reach surface water intake(s) at a sufficient concentration to cause deterioration in water
quality (the impact)”.
This report describes the lake modelling performed to assess the impact of spill events on the
water intakes in the study area. The report evolved over the study period as spill scenarios were
developed and questions were asked as to "what ifs". This report was prepared to support the
analysis and questions posed by the Lake Ontario Collaborative.
The source of water for Lake Ontario based municipal drinking water intakes, is in this coastal
zone, which is defined as water depths of 30 m or less (CTC,2011b). With one exception, the
intake pipes are all located within this near‐shore zone (their intake crib is located at distances
ranging from 0.5 – 5 kilometres off‐shore). But the focus of the modelling assessment provided
in this report is both the Coastal Zone and the main lake off shore waters.
In a large lake system such as Lake Ontario, water quality and the sources and processes that
influence water quality are not the same for the near shore area (coastal zone) as compared to
that found further offshore (main lake area). In Lake Ontario the coastal zone is considered as
the area from the shoreline out to the 30 m depth contour. In the coastal zone, water quality is
influenced by land‐based discharges (such as rivers, streams, wastewater treatment plants, and
groundwater) which mixes at the boundary of the zone with the off‐shore main lake waters. The
rate at which this mixing of the coastal and main lake water occurs is subject to hydrodynamic
forces such as prevailing wind speed and direction, water and air temperatures and the
bathymetry. Other processes in the lake can cause taste and odour problems for example.
The quality of water in the main lake area is established largely by water flowing from the
upstream Great Lakes (Erie, Huron, Michigan, and Superior) through the Niagara River into Lake
Ontario and direct rainfall and atmospheric fallout to the lake’s surface together with
biochemical processes that occur within Lake Ontario.
1‐21
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
In carrying out this work, events were modelled based on large releases of contaminants
associated with existing activities on land that might result in deterioration of water quality to
the point that it is unsuitable for use as a source of drinking water. A number of spill scenarios
were modelled to determine if certain land‐based activities could pose a potential drinking
water threat to these intakes. Any scenario that identifies conditions under which a contaminant
could exceed a threshold in the raw water is identified as a significant drinking water threat. The
events that were modelled were: disinfection failures at each municipal waste water treatment
plant; accidental large scale release of tritiated water from nuclear power plants; product of
waste spills from industrial facilities; and spills from a petroleum pipeline as it crosses major
tributaries.
In terms of governance of the Modelling work, the list of events was developed in consultation
with municipal staff responsible for water and wastewater, conservation authority staff, and
some industrial representatives, under the Leadership of Mr Rodney Bouchard (Peel Region),
who acted as Project Manager for the Lake Ontario Collaborative.
1‐22
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
2 Modelling Objectives and Selection
2.1 Modelling Objectives
The following points outline the objectives of the 3‐D hydrodynamic and water quality modelling
present herein:
• Develop a dynamic 3‐D hydrodynamic model that accurately simulates the currents
and water temperature conditions along the north shore of Lake Ontario;
• Calibrate the model based on observed currents and temperature data;
• Calibrate the model advection‐transport module based on observed spill along the
north shore of Lake Ontario;
• Calibrate the model to simulate the bacteria levels from stormwater, CSOs and
tributaries;
• Use the calibrated model to assess impacts on the water quality at water treatment
plant intakes from numerous spill scenarios;
• Establish a methodology for sampling the spectrum of lake conditions.
2.2 Model Selection
The modelling software selected for this application was the MIKE‐3 package developed by the
Danish Hydraulic Institute (DHI). MIKE‐3 is a professional engineering software package for 3‐D
free surface flows.
2.3 Mike 3 Model NonTechnical Description
A whole lake model is required to predict the water currents in the nearshore area of interest,
say the Toronto Waterfront, or the lake offshore of the Town of Wellington. MIKE‐3 uses the
full three‐dimensional representation of water motion, including thermodynamics. It accurately
simulates the seasonal thermal conditions and summer stratification that affects the circulation
pattern in Lake Ontario, which is required for accurate predictions of water currents. The model
is based on a number of separate software modules which are invoked for various tasks such as
hydrodynamic simulation, advection‐dispersion of dissolved substances, particle tracking for oil‐
spills, ecological & biological processes. All advanced modules depend on the basic
hydrodynamic module which supplies the information about the lake currents and temperature.
2‐1
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
2.3.1 Mike3 HD module
The Hydrodynamic (HD) module simulates the 3‐D incompressible Reynolds averaged Navier‐
Stokes equations with Boussinesq assumptions. The model is based on continuity of mass,
momentum, temperature, density and salinity equations, with a turbulence closure scheme.
The effects of source and sinks, river discharges, surface water elevations are included.
The temperature is affected by solar input through air temperature, relative humidity and sky
clearness.
2.3.2 Mike3 AD Module
The advection‐dispersion (AD) module is coupled with the HD module. Transport (advection) of
scalar quantities such as temperature, salinity and tracers is handled in the AD module.
Dispersion is the second purpose of this module – it covers the spreading, longitudinally,
laterally and vertically.
Tracers can be conservative or have a first order decay.
2.3.3 Lake Hydrodynamics and Heat Exchange
The major forcing function used to drive the currents in the model is wind stress.
The thermal stratification of Lake Ontario begins after the spring thaw. Water near the
shoreline warms up first and the zone of warmer water slowly spreads out as the heating from
the sun increases. Water temperatures start out at 4°C and warm from there. The maximum
density of water occurs at 4°C and is the major factor in the formation of the thermal
stratification. Cold water at 4°C will sink below warmer water. Wind mixing of the upper water
column is only sufficient to keep the top 20 to 35 m or so well mixed during the summer period,
causing the water this depth to remain at 4°C. There will be a structured thermal distribution in
the water column. Typically the water column would be 20°C from the surface to say 20 m, over
the next 10‐m or so the temperature decreases non‐linearly to 4°C and from 35 m downward
the water is a constant 4°C. A thermal bar first forms in the spring and slowly extends out into
the lake – Figure 2‐1 . The spatial distribution of the layers is not even, typically a dome forms in
the lake with the warm layer thinnest in the center of the lake and thickest at the shoreline –
Late June.
2‐2
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Halton-
Halton-Hamilton Source Protection Region
CTC Source Protection Region
Trent Conservation Coalition Source Protection Region
12
Lake Ontario Workshop Dec. 9, 2011
Figure 2‐1 Lake Thermal Process
When the lake is stratified, wind stress affects the lake differently than when the lake is
isothermal as in the spring and fall. Upwelling and downwelling events occur during
stratification, which cause cold deep lake water to flow toward the north shore displacing
warmer water with clean fresh cold water, downwelling has the opposite effect. These events
are not predicted by two‐dimensional models which is why three dimensional models are used.
In order to cause warming and cooling of the water in the lake, a thermodynamic balance is
required. The heat balance is controlled by latent heat loss by thermal radiation to outer space
and evaporation and heat gain by solar radiation (long wave and short wave) and conduction
from surface air.
2.4 Meteorological Data
The MIKE‐3 HD module requires the surface wind speed and direction to drive the water
currents. Wind speed and direction time series at an hourly frequency were used to provide
the surface wind stress.
Additional parameters include air temperature, relative humidity and sky clearness. These data
are usually measured at major airports 24/7 on an hourly basis.
2‐3
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Atmosphere and Environmental Services (AES) also operate lake buoys that collect wind and
temperature related data. While this data would seem to be a natural source of the required
data, the model performance based on the data is very poor compared to the land‐based
Airport data.
Airport data from Toronto’s Pearson, Trenton, Kingston, Buffalo, Rochester and the weather
station at NWRI Burlington were tried in various simulations during the calibration process.
A new source was recently introduced – NOAA (National Oceanographic and Atmospheric
Administration) has produced a 5 km based 2‐D wind field based on the North American
Mesoscale Atmospheric Model. The hourly data is available for 2002 and onwards.
2‐4
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
3 Section 2 – Model Description and Calibration
3.1 Hydrodynamic Model Calibration
The Mike 3 model is based on a mathematical formulation known as the finite difference (FD)
method. The lake is represented by a grid of squares with vertical layers. The whole lake is
divided up into squares with edges 2430m long. Equal length vertical layers are used to
represent the water depth. The vertical distribution of depths is known as the Z‐coordinate
system. Other models use a FD grid with a constant distance between layers, such as the Sigma‐
coordinate system ‐ with variable layer thickness, and there are others based on variations of
this method. Still other models are based a triangular grid known as a Finite Mesh method and
can have either a Z, Sigma coordinate system or mixed.
The method MIKE‐3 uses for nesting ‐ in order to obtain fine scale resolution of study area ‐
consists of a whole lake model, in our case based on a resolution of 2430 m, and nested grids at
a ratio of 3:1, or 810, 270, 90 m respectively. The nested grids are computed simultaneously
along with the whole lake grid. This provides a top‐down, bottom‐up feedback system so that
fine details in the smaller grids can influence the larger grids calculations, and similarly large
scale motions are felt by the finer grids. Tributaries can be fed into the computations at any
scale of the nesting grids. For example the Niagara River flows into the large whole lake model,
and the water levels at Kingston are used to control the whole lake elevation. Smaller
tributaries or spill discharges are usually released into the smallest grid. Sometimes it is
necessary to use the outer nested grid to discharge tributaries that are remote to the study area
but may have an effect of the area, this helps to keep the simulation times short as larger fine
grids vastly increase computer times.
Other models do not have this flexibility ‐ POM, ELCOM, CANDIE, etc.. Their method would be
to run a whole lake model and then the velocities and temperature and other pertinent
variables would be stored along boundary lines that form the edges of a nested grid. Then the
nested grid would be run with the stored data providing the forcing at the boundaries of the
finer grid. The finer grid would "feel" the impact of the Niagara River, but it cannot feed back to
the whole lake model any influences it may have on the whole lake circulation. For example the
Toronto Inner Harbour, the East Headland, Humber Bay and other finer resolution shoreline
features would not be represented in a whole lake model, the headland and Toronto Islands
may be just a blip on the shoreline. The flows in and out of the eastern and western gaps would
not be felt by the whole lake mode. In the MIKE‐3 world all these features would be modelled
at each time step.
Another significant benefit of simultaneous nested grids is that a plume that moves beyond the
finest scale grid is still within the computational domain, it is still in the model! In a two mode
3‐1
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
model, like POM, if a plume moves beyond the boundaries of the fine grid, it is lost. So a plume
that moves out the model domain, and then a current reversal occurs that would bring the
plume back into the fine grid, doesn't happen in the non‐ nested approach of models like POM.
The MIKE‐3 system can model plumes that move outside of even all 3 sub‐nests, into the 2430
m whole lake model, and then move back into the say 90 m grid, although there will be some
loss of definition when going from the coarse to fine grids.
A disadvantage of a nested grid model, run in a finite difference format, is the computation time
step ‐ the time step of the lake wide model is set by the grid size of the smallest mesh. In the
2430 m grid a time step of 600 seconds is possible, however when the nested grids are used the
time step has to decrease to about 30 seconds for a 90 m grid and 90 seconds for the 270 m
grid.
Boegman and Yerubandi, 2006, showed 2 km resolution was insufficient to resolve some of the
fine scale processes, topographic and Kelvin waves, for example, and that grids smaller than 1
km are required. For MIKE‐3 this will require increasing the number of grids by at least 4 times
for the whole lake model and consequently much smaller time steps which will require longer
simulation times. We are limited by computer resources and other time constraints at this time.
This is a compromise that must be considered in future spill modelling.
3.2 Calibration Parameters
The calibration process involves selecting the appropriate grid sizing, time step, vertical
resolution, wind source and other driving forces, and then adjusting the model parameters (fine
tuning) to make the model predictions agree with observed data. Normally current data
collected with instruments deployed in the lake are used to calibrate the hydrodynamic module.
Temperature data collected at water intakes are also valuable in this process.
3.3 Turbulence Models
MIKE‐3 has 6 optional turbulence models, but the user can select only one. They are:
• the very simple "no eddy viscosity"
• "constant eddy viscosity"
• Smagorinsky formulation
• κ model formulation
• κ‐ε model formulation
• highest order accuracy ‐ Mixed κ‐ε/Smagorinsky.
3‐2
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
The simple models do not develop a thermal stratification, the Mixed κ‐ε/Smagorinsky was used
throughout all simulations as it is the highest order accuracy scheme and produces better
correlations with the observed data.
3.4 Bathymetry
The bathymetry data was obtained from the NOAA Lake Ontario CD at a resolution of 3 arc
seconds (NOAA 1999). This grid was interpolated to a 90m resolution with the MIKE‐3
bathymetry editor. The whole lake resolution was selected as 2430m.
The whole lake model resolution of 2430m is too coarse to study the near‐shore dynamics of the
lake, so a method of nesting is used to provide currents and plume travel information in greater
detail.
3.5 Nested Grids
The Mike‐3 model uses a 3:1 nesting with smaller grids based on 810m, 270m, 90m, 30m and so
on. Most of the nested models used in the report, are based on a 270m and 90m grid. The
computational time increases with smaller grid sizes so it is important to economize grid layout.
3.6 Vertical Resolution
Just as important, to minimize computer time, is the choice of vertical resolution. Lake Ontario
is over 250m deep, with the thermocline extending down to about 60m at the time of fall over‐
turn; adequate representation is required of the upper 60m while the area below the
thermocline can be simulated as a single layer of water. This is one limitation of the Z‐
coordinate system. Many trials have shown that the optimal number of vertical layers lies
between 20 and 40 layers with thickness of 4 to 2m respectively or a vertical coverage to about
80 m ‐ Section 3.15.2 shows a sensitivity test of layers. Earlier experience with MIKE‐3 in
studies such as the City of Toronto WWFMMP waterfront modelling report tested the number
of layers and found similar results.
Figures 3‐1 and 3‐2 show a typical lake grid system, with the lake rotated so that the nested grid
area has a smooth shoreline parallel to one axis. Orientations with large areas of land or water
on side of the grid (triangular distribution) are un‐economical or wasteful in terms of computer
resources.
3.7 ADCPs
Acoustic Doppler Current Profilers (ADCP) are current meters which measure current velocites
at different depths in the water column of the lake where the current meter is moored. ADCP’s
use sound waves to measure the water velocity. Many hundreds of “pings” per minute are
made and the time for the sound wave to reflect off of suspended particles in the water and
return to the sensors provides a measure of depth, Three sensors on the meter measure the
scatter of the wave and defines the current speed and direction in the “bins”. Bins are the
3‐3
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
selected depths in the water column; usually 1m intervals are chosen and data are averaged
over 15 minutes to one hour. Temperature is collected at the ADCP deployed depth.
Figure 2 shows the location of Acoustic Doppler Current Profilers (ADCP) operated by Ontario
Power Generation at the Pickering and Darlington NGS. The data collected by these ADCPs are
available to use in calibration of the model. Additional ADCP data has recently been found and
will be used to verify some the model grids – although the data may not be in the same years as
the spill scenarios were run.
Other ADCP locations included Toronto Inner Harbour and NWRI deployments in Lake Ontario.
3.8 Thermistor Chains
Environment Canada operated temperature thermistor chains at a few stations in the lake and
this data has also been made available for use in the calibration.
3.9 Water Elevation
Water levels in Lake Ontario vary over the season, rising in the spring from the snow melt and
dropping over the summer due to drying conditions. This effect is added for increased accuracy.
3.10 Niagara River
The large flow from the Niagara River is included in the whole lake model to also increase
accuracy. While this river flow does not directly impact circulation on the north‐shore it does
have an impact along the south shore and is included for completeness.
3.11 Easterly Pickering Spill Grid Calibration
The first in a series of calibration exercises is the Easterly Pickering spill 270 m grid extending
from Pickering to Darlington – Figure 3‐2.
3‐4
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 3‐1 2430 m whole lake grid with nested grids
Figure 3‐2 270 m nested grid with ADCP locations
3.11.1 Application of Wind and other Forces
The next step in model development is the selection of winds. Single station data such as
airports are used to provide a uniform wind over the whole lake. There has been limited success
with combining several airports data, by some form of bilinear interpolation, to produce a 2
dimensional (2d) wind field.
3‐5
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
NOAA can provide a 5km grid of their North American Mesoscale Atmospheric model at 1 hour
intervals. The NOAA model is a weather prediction tool which uses observed data at stations
throughout North America and is considered the most accurate 2d wind field available for
model use. The data is available from 2002 onwards.
The model was driven by NOAA wind field for 2006 and Pearson Airport wind for both 2006,
2007. ADCP data were available at Pickering for 2006 and 2007, and Darlington ADCP had data
only for 2006. Figures 3‐3 & 3‐4 show a comparison of the model predictions of the direction
and speed time series at Pickering ADCP. The model used 30 vertical layers with 3m thickness.
The simulation started May 15 with the water temperature set to 4°C throughout the water
column in all nested grids and the whole lake model. For years prior to the availability of NOAA
data single station airports were used.
Water elevation data from the Kingston gauge or Olicot NY gauge is applied.
The Niagara River is included as an hydraulic force, the water temperature is set to observed
data from the Lake Erie westerly Buoy (Port Colborne) operated by Environment Canada.
(http://www.meds-sdmm.dfo-mpo.gc.ca/isdm-gdsi/waves-vagues/search-
recherche/list-liste/data-donnees-
eng.asp?medsid=C45142&s1=7/1/1994&s2=10/19/2009)
Figure 3‐3 NOAA produced Direction TS at Pickering ‐ Blue is ADCP data, Black is Model Prediction
3‐6
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 3‐4 NOAA produced Speed TS at Pickering – Blue line is Model prediction, Red is ADCP data
3.12 Fnorm Scores
The model results appear highly correlated with the data. The method used to measure model
accuracy is based on a vector average that uses both speed and direction called the Fourier
Norm (FN). The calculation involves the average difference between the two vector
components of velocity. In model terms the velocity is computed as a V component in the
onshore‐offshore (east‐west) direction and the U component in the alongshore (north‐south)
direction. Equation (1) describes the Fourier Norm as:
r r 1
vo , vc NΔt 2
r r 1 r r 2
FN =
r
where vo , v c =
N
∑ vo − vc
t =1
[1]
v o ,0
and vo = observed data and vc = computed data.
The Fnorm scores are tabulated in Table 3‐1. The 2006 data spanned the entire year with a few
weeks missing in July, while the 2007 data covered the period of October and November only. A
Fnorm score of zero is a perfect match, a score of 1.0 mean the difference between the
computed vector and the observed vector (error size) is the same as the vector size, which could
3‐7
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
be 180° direction difference, or small computed speed, large observed speed. These scores
represent a very good correlation with the ADCP data. Fnorm scores were determined from the
surface layer ‐ usually referred to as the TOP layer, addition layers were used, here the 5m
depth from the surface ADCP data was compared with the equivalent model layer.
The NOAA data perform well in 2006, having a slightly better score than the Pearson results. In
2007 the two wind sources are tied at the surface but Pearson scores slightly better at the
deeper layer. Similar results have been found in other years, sometimes the NOAA scores better
than Pearson or other airports.
Table 3‐1 Fnorm Scores
3.13 Temperature Correlations
Figure 3‐5 and 3‐6 show the temperature prediction at the bottom layer where the ADCPs
measure the temperature. Again the data are very well correlated ‐ serial correlations range
from 0.9 to 0.95. There are episodes when the model has large variance with the observations
and this is usually attributed to the limited depth resolution and course vertical thickness.
Figure 3‐5 NOAA produced Temperature TS at Pickering ADCP ‐ bottom layer ~ 9m
3‐8
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 3‐6 NOAA produced temperatures at Darlington ADCP ‐ bottom layer ~ 10 m
In terms of the hydrodynamics of the model, the NOAA wind field will be used when available as
it provides a good correlation with the ADCP data. When NOAA winds are not available the
most appropriate airport station will be used.
3.14 Comparison with other Hydrodynamic Codes
A series of simulations using a 2 km grid of Lake Ontario were performed by Huang et.al (2010)
at CCIW and the results were used to compare their codes with the MIKE‐3 code. Huang
compared the three codes (POM, ELCOM, and CANDIE).
Figure 3‐ 7 shows the model grid with ADCP locations.
Figure 3‐7 2 km grid of Lake Ontario with ADCP Locations
CCIW deployed ADCPs in 2006, at the locations shown in Figure 3‐ 7.
3‐9
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
CM4 (1270), CM2 (1268) had data from April 12 to Nov 11 at 30 min intervals.1269 had data to
July 24 and 1266 had data from July 25 to Nov 11.
MIKE‐3 vertical resolution was 40 layers at 2 m thickness. Multiple trials early on found this
configuration to be optimum for the grid size.
Other Model parameters – Quickest‐Sharp transport scheme applied. Constant wind stress
coefficient – 0.0026. Hydraulic input of Niagara River with Lake Erie AES buoy temperature
applied. Kingston water level also applied.
The Fourier Norm was used to score the various model trials.
Initial trials used Pearson, Hamilton and the AES buoy off of Grimsby, with Hamilton and
Pearson out scoring AES buoy.
CCIW provided a synthesized wind field based on measured winds at 6 stations in the lake,
spanning the period May 1 to Sept 22.
The NOAA North American Mesoscale model 5 km wind field was also available for our tests.
This data was scaled to match the 2 km grid.
Three models were used in the final evaluation, Pearson, CCIW and the NOAA wind fields.
ADCP data available at surface, 5 m and 10 m and were used in the evaluations, or depths that
were close enough.
The bottom roughness was left as the default value of 0.05 in all model runs. Wind stress
coefficient default value was 0.0026 was usually used, if a different value was used in a scenario
it will be noted.
Table 3‐3 lists the Fnorm score and the RMSE for each vector for ADCP CM4, CM2 and 1269.
ADCP 1266 had typical scores greater than 1.3 so it was not included here.
3‐10
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Table 3‐2 FNORM Scores for MIKE‐3
CM4 (1270)
NOAA 0.89 0.14 0.14 0.91 .007 0.1 0.96 .006 .009
CCIW 0.92 0.12 0.12 .097 .07 .10 0.98 .02 .006
Pearson 1.0 0.15 0.15 1.0 .008 .11 0.97 .007 .10
CM2 (1268)
NOAA 0.95 0.007 0.007 1.0 .005 .004 1.02 .004 .004
CCIW 0.95 0.008 0.007 1.01 .005 .005 1.01 .005 .005
Pearson 0.99 0.008 0.007 1.01 .005 .004 0.96 .007 .006
1269
NOAA 0.91 0.16 0.13 .091 .006 .005 0.98 .006 .006
CCIW 1.02 0.008 0.006 1.01 .007 .005 1.06 .008 .004
Pearson 1.08 0.16 0.14 1.09 .007 .006 1.07 .007 .004
The NOAA wind field has better FNORM scores. It is noted that the model appears to be able to
better simulate the directions of currents in the spring and fall period outside of the thermal
stratification period. The CCIW data is shorter term May start compared to April 12 and ends in
September. The NOAA results were tested for May to Sept 15 and had FNORM scores of 0.91,
0.92 and 0.98 for each depth category, which still outscored the CCIW results.
Figure 3‐7 and 3‐8 show the surface layer NOAA and CCIW directions and speeds compared to
the ADCP CM4 data. It is clear that the ADCP data has high speed events which the models
could not represent. The directional accuracy is fairly good. Figure 3‐9 and 3‐10 show the CM2
results. Once again the speeds are very high, making the FNORM scores poor. The quality of the
ADCP data is suspect as speeds of 0.5 m/s in Lake Ontario are very rare! Lower layers or bins in
the data file might be better for comparison; however, they were not available.
3‐11
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Huang .et.al, on Lake Ontario and were able to achieve FNORM scores between 0.5 and 0.8 with
these ADCP data sets. Golders (2009) stated the MIKE‐3 calibrated model used for the
Darlington environmental assessment was better at representing the near shore dynamics and
was not got at mid‐lake simulations. This shortcoming is likely due to the vertical resolution not
covering the entire water column in multiple layers.
Figure 3‐8 ADCP CM4 Direction comparison
Figure 3‐9 ADCP CM4 Speed comparison – many episodes of speeds above 0.5 m/s, model is unable to
reproduce these events
3‐12
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 3‐10 ADCP CM2 Directions
Figure 3‐11 ADCP CM2 Speed – many episodes of speeds above 0.25 m/s, model is unable to reproduce
these events
Temperature chains were also deployed and the surface temperature at 1269 has been plotted
in Figure 3‐11 for the NOAA and CCIW wind field results.
3‐13
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 3‐12 1269 Surface Temperatures
Temperature Isopleth plots for each trial is shown in Figure 3‐12 to3‐ 14 for the 752 thermistor
chain – Figure 3‐15 shows the observations. A system to score the isopleths has not been found
yet. The overall behaviour is rather good when compared to the observed data. The date units
for the MIKE 3 results are 2 days per tick mark, starting April 1 for Figure 12 to 14 and May 1 for
Figure 3‐15.
3‐14
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 3‐13 NOAA Isopleths – MIKE 3
3‐15
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 3‐14 CCIW Isopleths from Huang et.al, 2010
Figure 3‐15 Pearson Isopleths ‐ MIKE 3
3‐16
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 3‐16 752 Observations
NOAA data for air temperature and dew point are available and have not yet been incorporated.
3.14.1 Additional Comparisons with Huang et.al.
Some figures were extracted from Huang and used to compare with the MIKE‐3 results. We
could not provide a quantitative measure as the raw data from Huang were not available.
Figure 3‐16 shows the isopleths of temperature produced from the four different codes.
Figure 3‐17 shows the comparison of time series of temperature at the surface for the four
codes.
Finally Figure 3‐18 shows the surface velocity vectors and temperature produced by the four
codes. Figures 3‐19 and 3‐20 are enlarged versions for closer comparisons.
3‐17
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Temperature Isopleths
• Mike 3 using NOAA
winds @ site 752
Figure 3‐17 Temperature Isopleths Comparison
3‐18
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Temperature TS
• Mike 3 @ 1269
Figure 3‐18 Temperature Time Series Comparison
3‐19
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 3‐19 Surface Vector & Temperature Comparison
3‐20
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 3‐20 MIKE‐3 Vectors and temperatures September 1 2006
Figure 3‐21 POM Vectors and temperature same day as Figure 3‐20
3‐21
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
3.15 Calibration of the 270 m Grid Hamilton to Oshawa
This model domain was most frequently used in the spill scenarios. Additional calibration tests were
performed to address model parameters and model algorithms normally outside of the more dominant
model parameters ‐ Figure 3‐22 shows the 270 m grid.
Figure 3‐22 Extended 270 m Grid
The following sensitivity trials were run using 30 layers at 3 m thickness.
3.15.1 Dispersion in MIKE3
The internal code of MIKE‐3 determines the range of the dispersion coefficients for temperature and
advection dispersion, based on the grid size and bathymetry and possibly other parameters that are not
obvious in the user interface. Setting these values outside of the defaults usually causes a simulation to
not start or fail.
For example, the 270 m nested grid used for the majority of spills along the north shore, using 30 layers
and 3 m thickness, Table 3‐4 has the following model defines dispersion values for the temperature.
Table 3‐3 Dispersion Values
3‐22
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Four trials were run with different horizontal and vertical coefficients; the results for the Fnorm and
temperature correlation are listed in Table 3‐5. The Fnorm scores used the Pickering ADCP data shown
in Figure 3‐3 and 3‐4. While there are not large differences in the scores, the B scenario has a slightly
better temperature score. The B scenario was used in the spill simulations. The 40 layer ‐ 2 m layer
thickness simulation achieved higher FNORM scores dues to the higher vertical resolution.
Table 3‐4 Horizontal and Vertical temperature Coefficients
ADCP Layer A B C D
3‐23
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
From the MIKE‐3 users manual:
Each grid also has an entry for the horizontal and vertical dispersion factors. The default values are 0.1.
Trials with different horizontal values did not seem to make any differences. The vertical value does
impact the temperature stratification ‐ a value of between 0.01 and 0.0005 seems to be optimal for
good temperature correlation with ADCP data and water intake temperature data.
3‐24
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
There is also an option to set the "formulation" based on;
• proportional to eddy viscosity
• proportion to the velocity
• tensor dispersion.
Tests have shown the first two options produce almost the same velocity time series. The tensor
dispersion has been found to have slightly better Fnorm scores, but the temperature time series is
poorly correlated with observed data. The tests have not been exhaustive and further tests are
required. For example when selecting the tensor option the user interface now provides setting the
dispersion factor for three values per grid, the default values are all 0.1. The user manual does not
specify what the values represent so it is very difficult to know what to set – (DHI personal
communication states the values are for the X, Y, Z axis). Trials have found the default values will not
complete a run. A trial with all values set to 0.001 works, a trial with the first two set to 0.1 and the
third to 0.001 does not finish.
3.15.2 Model Sensitivity to Vertical Layers and layer thickness
Experience with the model has found that generally good agreement with temperature and ADCP data is
achieved when the model layer depth exceed the bottom of thermocline ‐ about 60 m. The one
drawback of the MIKE‐3 model is that the layer thickness is constant, and for deep lakes such Lake
Ontario, maximum depth ~280 m, means that to completely cover the vertical resolution many layers
are required which will slow down scenarios. It is thought that the layers under the thermocline
generally move as one whole layer and that representing the waters below the thermocline as one layer
is a good approximation. In the spill modelling and calibration layers to a depth of 80 to 90 m were
used, with either 40 layers at 2m thickness or 30 m at 3 m thickness.
A small number of trials (three) were made using the same general conditions but with 3 different layer
schemes. They were:
• Case A 40 layers 2.5 m thickness ‐ covering 100 m depth
• Case B 20 layers 5 m thickness ‐ covering 100 m depth
• Case C 10 layers 10 m thickness ‐ covering 100 m depth.
The temperature response of the model at the Pickering ADCP location is shown in Figure 3‐20.
3‐25
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
It is observed that Cases A and B perform very well, but Case C does not have a good correlation with
the ADCP data. The correlations for each case are:
Case A 0.93
Case B 0.92
Case C 0.69.
Model accuracy increases with layers and finer thicknesses improve the overall response of the model.
Figure 3‐23 Temperature response to variable layers
3.15.3 Wind Stress Coefficient
The surface winds provide the major forcing for the hydrodynamics of the Lake. The wind speed and
direction are transferred to the water surface by a wind stress equation which requires setting the value
of the wind stress coefficient. The default value is 0.0026, which is standard in many of the codes.
There is also an option to provide a variable wind stress function. This is best described by "5 ‐ 26, 15 ‐
36", which is executed as:
3‐26
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
• winds less than 5 m/s coefficient is 0.0026,
• winds between 5 and 15 m/s, a linear gradient from 0.0026 to 0.0036 is used,
• winds greater than 15 m/s coefficient is 0.0036.
A series of simulations were made with NOAA 2006 wind field and different wind stress values, and the
model predictions were tested with the Pickering NGS ADCP data. Table 3‐6 lists the results, the
variable scenarios provide a slightly higher correlation. The constant 0.0026 value was used in most of
the spill scenarios.
Table 3‐5 Wind Stress Coefficient
5‐26, 15‐36 5‐32, 15‐42
3.15.4 Bottom Roughness
The lake bottom roughness default value is 0.05 m. Three trials were made with different values. Table
3‐7 lists the results. Larger roughness values have a slightly better correlation. The 0.05 value was used
in all spill modelling.
Table 3‐6 Roughness Values
3‐27
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
3.15.5 Transport Schemes
There are four different schemes to describe the transport computations, Table 3‐7 lists the results for
each scheme, based on roughness of 0.5 and NOAA winds with variable stress 5‐26,15‐32. The Simple
Upwind and 3D upwind have better Fnorm scores at the bottom layer but the temperature correlations
are poorer than both Ultimate‐Quickest and Quickest‐Sharp. The Quickest‐Sharp scheme was used in all
the spill modelling.
Table 3‐7 Transport Schemes
3‐28
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
4 AdvectionDispersion Model Calibration
The calibration of the advection‐dispersion module uses a known discharge and observations of the
passage of the plume. The usual method is to release a known quantity of dye or some other easily
measured substance and track the plume over several hours or days. There is a historical event in which
a spill of tritium from the Pickering NGS was tracked from Oshawa to Hamilton. This is an ideal
calibration event.
4.1 August 1992 Tritium Spill
The Lake Ontario model was set up with an expanded 270m grid to cover the area from Hamilton to
Oshawa, see Figure 4‐1. The entire lake was modelled at a resolution of 2430m.
Figure 4‐1 270 m Map with Intake Locations
The Tritium spill was released on August 2 at 0400 hours. The spill lasted six hours at a release rate of
0.000119 m3/s – total release volume ~2900 kg. The estimated Tritium concentration was 7.9 E+11
Bq/kg = Bq/L. It is understood that the spill was mixed into the cooling water channel and was
completely well mixed by the time it entered Lake Ontario. Tritium has a half‐life of some 12 years and a
first order decay rate of this order was applied. The discharge was applied to either the surface layer if
the vertical resolution was coarse or two layers if allowable.
Ontario Hydro Nuclear (OHN) measured Tritium levels at the water intakes and shoreline locations along
the north shore for several weeks after the event. These observations and were reported in Report
NA44‐REP‐03483.2‐0021‐R00, 1994, OHN. The data has been used to compare the model predictions.
4‐1
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Initially the tritium plume moved eastward impacting the Ajax , Whitby and Oshawa intakes and then
the winds shifted and the plume reversed course and then traveled west in one single movement and
Tritium was measured at all intakes as far as Hamilton.
A number of simulations were made in the calibration process. Historical winds were available at the
larger Airports situated around the lake, NOAA 2 d wind field data was not available until 2002. The
vertical resolution of the model was varied from 20 layers to over 80 at intervals from 1 m to 6 m in
layer thickness.
4.1.1 Dispersion Coefficients for AD Module
The default values were used along with the Eddy velocity scheme. If the time of travel is not in
agreement with the observations then fine tuning of the AD dispersion coefficients will not correct the
timing issue ‐ it is most likely the winds that are not correct.
4.1.2 Pearson Winds
When Pearson hourly winds were applied, with 2 discharge locations at split vertical layers, 30 layers at
4 m resolution, Figures 4‐2 to 4‐5 show the observed and model predictions for the western intakes.
Horgan & Harris amplitudes are very close, with initial peak times in good agreement. The tail of the
plume is too high in amplitude and lingers too long.
4‐2
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 4‐2 Horgan & Harris
The rest of the intakes show a large lag in the plume reaching the intakes, while the amplitudes are
quite close.
4‐3
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 4‐3 Island Intakes
Figure 4‐4 Clark to Oakville Intakes
4‐4
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 4‐5 Hamilton Intake
There were periods in the Pearson data with zero speed and direction, usually late at night and early
morning, this could be caused by very calm conditions or the airport was closed and no readings were
taken. A combination of Pearson and Buttonville airports, Buttonville data was substituted for the
missing Pearson data, this method had less success.
4.1.3 Kingston Winds
Kingston airport winds were next applied, with somewhat better success, Figures 4‐6 & 4‐7.
Figure 4‐6 Kingston Winds Harris‐Horgan
4‐5
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 4‐7 Kingston Winds Clark to Oakville
4.1.4 Trenton Winds
Trenton Airport winds were tried next. The time of arrival at most intakes was very close. The amplitude
was low at Harris‐Horgan, which could be increased through some fine tuning. The Hamilton match was
very good. An additional trial was made with winds from Buffalo, and the results were very poor as this
station did not reproduce the current shift to the west. Several other trials were made and the final one
with 40 layers and 2 m thickness are presented. The time of travel (TOT) for Horgan (Figure 4‐8) is
about 24 hours late and Harris is late by about 20 hours. The TOT to the shallow island (Figure 4‐9) is
very close, however the peak is over‐predicted. The westerly intakes from Clark to Oakville (Figure 4‐10)
have good TOT and the amplitudes are also in good agreement. The Hamilton plume (Figure 4‐11) has
good TOT however the peak levels do not reach the observed values.
4‐6
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 4‐8 Trenton Winds Harris‐Horgan
Figure 4‐9 Trenton Winds Island Intakes
4‐7
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 4‐10 Trenton Winds Clark to Oakville
Figure 4‐11 Trenton & Kingston winds at Hamilton
The model dispersion coefficients were left as default values ‐ Section 4.2.2.1 has additional dispersion
coefficient testing. The AD module is considered calibrated and the only adjustments required for other
discharges are the decay rate and discharge concentrations.
4‐8
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
4.1.4.1 August 2 1992 Tritium Spill Detailed Surface Plume Behaviour based
on Trenton Winds
A series of figures are provided here to show the behaviour and evolution of the surface plume as it
moves from Pickering, first to the east and then westerly to Hamilton. The following set of figures show
the overall plume behaviour from release to impact at Hamilton Figures 4‐12 to 4‐22. Interspersed with
the Trenton winds simulations are four figures from the simulation using Pearson Winds. These figures
are provided to show the different behaviour of the plume when different wind sources are used. When
the surface temperature of the lake is compared with these figures – 4‐15 for example the large blue
patch of water intruding into the red zone “horseshoe” shape is a warm water plume which appears to
push or follow the tritium plume for several days. In addition, just before the spill there was a large
upwelling event which dropped the nearshore temperatures down to the 4° to 6°C range.
The time series of speed and direction at Pearson and Trenton Airports are shown in Figures 4‐12 and 4‐
13. The directions are similar until August 9‐10 where there is a shift between the two – which may
cause the different plume behaviours seen in the figures.
Figure 4‐12 Pearson and Trenton Airport direction time series – the 7th has a deviation and more between the
10th and 12th
4‐9
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 4‐13 Pearson and Trenton Airport speed time series – no major variances are apparent
Figure 4‐14 Day 2 from release
Figure 4‐15 Day 3 from release – Trenton Winds
4‐10
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 4‐16 Day 7 from release ‐ current reversal – Trenton Winds
Figure 4‐17 Day 7 from release ‐ Pearson Winds – note slower reversal – no separation – up to this time the
plumes are very similar
Figure 4‐18 Day 8 from release ‐ plume pushed westerly –Trenton Winds
4‐11
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 4‐19 Day 8 from release ‐ Pearson Winds – note small separation and slower speed of plume
Figure 4‐20 Day 10 – Trenton Winds
Figure 4‐21 Day 10 ‐ Pearson Winds used – plume has hardly moved compared to Trenton case above
4‐12
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 4‐22 Day 12 Trenton Winds Plume impacting Clark, Lakeview, Lorne Park and Oakville
Figure 4‐23 Day 12 Pearson Winds, plume seems stalled
Figure 4‐24 Day 13 Trenton winds used – plume at Hamilton
4‐13
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
As this was an event, it was necessary to try different wind sources to make the plume travel as it must
have. There was a current reversal which suggests that a cold or warm front passed through the region
and each airport registered it differently, with Trenton records appearing to have tracked the front the
best. Buffalo Airport winds did not cause a current reversal.
4.1.5 Tritium Levels at AjaxWhitbyOshawa
The model over predicted the concentration at these intakes, the reasons are still unknown, and
ongoing work is trying to resolve this issue.
4.1.6 Other Model Predictions
Hurdawar‐Castro (2007) used the IDOR 3‐D model to simulate the tritium spill. The times of travel were
close at some intakes (model area limited to Harris – Oshawa WTP intakes), and the amplitudes were
also reasonable. The discharge amount did not agree with the published data.
The MOE – Peter Nettleton presented their MIKE‐3 finite Mesh model results at a recent workshop on
the Tritium spill (Candu Owners Group – Jan‐19 2012) . The model was able to match the observations,
however while the spill was only 6 hours long, they spread the spill over 60 hours to achieve their
results. They speculate that there was some re‐circulation activities that spread out the plume. The
MOE version used a bi‐linear interpolation of 8 wind stations around the lake.
The MIKE‐3 model was run with a 60 hour spill – and with cooling water and intake active – using
Trenton winds. The amplitudes at Horgan and Harris doubled, the westerly intakes time of travel were
off, Hamilton results were about the same as the original results.
4.1.7 Ambient Tritium levels in Lake Ontario
The model assumed zero background levels of Tritium. An extensive review by Fairlie, 2007, reports
ambient levels are thought to be 2 Bq/L, based on Lake Superior measurements in 1997/1998 ‐ where
no nuclear facilities are located (King et al, 1998). Lake Ontario was measured at 7.1 Bq/L by King et al.
Over 3000 samples of drinking water across Ontario had an average value of between 5 and 10 Bq/L
(Health Canada 2006).
Levels of tritium before the 1992 spill measured at the 4 WTP of Toronto had levels between 9 and 16
Bq/L in the treated water (Ontario Hydro Nuclear, 1994).
4.1.8 Sensitivity to Wind Sources
Two simulations were run for the 2008 season, using the single station Trenton Airport data and the
NOAA 2D wind field data. All other model parameters were identical. The 1992 tritium spill at Pickering
was used as the source, with a release on August 2 2008.
4‐14
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
The times series of tritium at the Ajax and Harris intakes were compared, shown in Figure 4‐19. The
model response is very different!
The Trenton wind simulation results were:
Ajax peak 40000 Bq/L and time of travel of peak was about 2 days
Harris peak was 200 Bq/L and time of travel of peak was about 15 days.
The NOOA wind simulation results were:
Ajax peak 0.01 Bq/L and time of travel of peak was 24 days,
Harris peak was 2000 Bq/L and time of travel of peak was 5 days.
The Trenton wind simulation had the plume move easterly for a long period and a reversal to reach
Harris ‐ while the NOAA wind had the plume move westerly initially, with an eventual reversal to reach
Ajax.
4‐25 Trenton ‐ NOAA Wind Sensitivity
4‐15
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
4.1.9 Lake Currents Different Basins
Lake Ontario has been characterized as having separate distinct current patterns in the western basin
(Hamilton to the Toronto Islands) and the central basin (Toronto Islands to Prince Edward County). The
lake currents predicted by the model for the January to May period were analyzed and the resulting
current‐speed distribution was plotted as a compass rose. Figure 4‐20 shows the current distribution
offshore of Etobicoke and Figure 4‐21 shows the currents offshore of Pickering. The Etobicoke currents
are generally equally distributed to east and west currents with higher speed events flowing westward ‐
possibly due to the larger fetch from the east. The equal distribution would indicate that there is not a
stable eddy in the western basin. The Pickering currents are biased to easterly flows in the majority and
with stronger speeds over the period. This current distribution with the major easterly flow would
indicate a clockwise eddy in the central basin.
Figure 4‐26 Etobicoke Compass Rose
4‐16
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 4‐27 Pickering Compass Rose
4.2 Toronto Inner Harbour Hydrodynamic and E. coli Calibration
4.2.1 Toronto Inner Harbour Hydrodynamic Calibration 20072008
The Inner Harbour model was calibrated with 2008 ADCP data, Figure 4‐22 shows the 90m nested grid.
ADCP data were available from October 30 to December 11, 2008.
The model was driven with NOAA wind field hourly data along with NOAA air temperature in 2
dimensions. Additional hydraulic forcing was supplied by the flow from the Niagara River with
temperature collected at the Lake Erie AES buoy offshore of Port Colborne. The Don River flow as
predicted by consultants involved with the Don Truck Sewer Project (MMM) was included along with
temperature.
Vertical resolution was 2m thickness and 40 layers.
4‐17
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
The shoreline geometry is very complex, not as simple as the large scale north shore of Lake Ontario,
which is fairly uniform straight shoreline. The Inner Harbour, Islands, East Headland and Humber Bay
cause complex currents that the model is not always able to resolve.
Figure 4‐28 90m nested grid of Inner Harbour with ADCP locations
Model was initiated on May 1 with 4˚ C uniform temperature.
Figures 4‐23 to 4‐34 to show the direction and speed as observed and predicted at each ADCP location
for the surface layer and mid‐depth. The ADCP directions in the Western Gap as consistent at both
layers, while the speed show some slowing at depth. For most of the data period the model tracks
direction with the exception of the first 7 days of November and the episodes of high frequency current
reversals. The model speeds are quite well matched changing with amplitude for the higher speed
events. The Fnorm score for the Western Gap surface layer was good overall with a value 0.94. Deeper
layers achieved better scores with a 0.86 for Nov. 7th to the end and the RMSE of 7 mm/s for both
components.
4‐18
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
The Eastern Gap ADCP data has high frequency current reversals and 2 layer flow for some periods,
(upper layer flowing opposite to bottom layer). The speed events are somewhat similar to those in the
Western Gap. The model predicts high speed events for both Gaps, but the Eastern Gap observed
events are much smaller in amplitude. The Fnorm scores are poor with values well above 1.5, but with
RSME of 8 mm/s for each component. The location of this instrument was not in the middle of the
channel, but on the southwest corner, which may be the cause of the poor correlations. Ship traffic and
lack of clearance in the channel precludes the deployment of the ADCP in a better location.
Gibraltar Point (GP) had much better success with surface Fnorm scores of 1.03 – (after Nov 7 ‐ 0.69),
for the deeper layers and 1.09 with higher RMSE of 6.9 mm/s for each component (0.7 after Nov 7). The
ADCP surface direction shows high frequency “jitters” or variations around a mean direction, while the
deeper layer shows a more dominant current direction. The same behaviour is seen in the speeds. Thus
we have better Fnorm scores at depth due to a more consistent current regime.
Figure 4‐29 Western Gap Surface Direction
Figure 4‐30 Western Gap 2m deep Direction
4‐19
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 4‐31 Western Gap Surface Speed
Figure 4‐32 Western Gap 2m deep Speed
Figure 4‐33 Eastern Gap Surface Direction
4‐20
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 4‐34 Eastern Gap 3m deep Direction
Figure 4‐35 Eastern Gap Surface Speed
Figure 4‐36 Eastern Gap 3m deep Speed
4‐21
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 4‐37 GP Surface Directions
Figure 4‐38 GP 3m deep Direction
Figure 4‐39 GP surface speeds
4‐22
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 4‐40 GP 3m deep Speeds
Surface temperature data were collected in 2007 and Figure 4‐35 shows the model results with the
observations, there is a good correlation, however, the model results have a few episodes with over
predictions of up to 6°C.
Figure 4‐41 Surface Temperatures 2007
4.2.2 Inner Harbour E. coli Calibration
As part of the Don Trunk Sewer Project, the Inner Harbour (IH) model was used to predict the water
quality impacts from over 50 discharges along the Toronto Waterfront. The discharges ranged from
large and small CSO and stormwater outfalls to major tributaries and large wastewater treatment
plants. The time series of flow and pollutants were developed for existing conditions and were used to
calibrate the IH advection‐dispersion module for E. coli. Intensive monitoring of E. coli over the summer
swimming period in the IH during 2007 and 2008 were available for the calibration.
4‐23
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
The model first order decay rate for E.coli was adjusted during the calibration phase to match as best as
possible with the observed data in 2007 ‐ (a first order decay of 1.1 x 10‐5/s or 60% die‐off per day, T90
of 57 hrs). Other studies and model calibrations have quoted decay rates between T90 of 24 to 48
hours.
Table 4‐1 lists the model predictions for the IH locations along with the 2007 & 2008 observations for
the period June 1 to August 31, for the percent of time the locations (Figure 4‐36) is above the PWQO
for E. coli (100#/100mL). The assessment of the “model accuracy” was difficult as the model over
predicted the E. coli posting time at some locations and under predicted the times at other locations.
Adjusting the decay rate would improve the accuracy in some areas but make it worse in other locations
so an overall average was used as a final measure. The observations of percent time posted indicate
that 2007 was a dry year while 2008 was a wet year.
During dry years the locations influenced by the Don River (yellow) have similar posting rates, while
those not influenced tend to have lower posting rates. In a wet year (green) the Don River locations are
predicted to have higher posting levels than observed, while the non‐Don locations are predicted to be
about the same. This would indicate that the Don River loadings may be too large during wet weather
years.
4‐24
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Table 4‐1 Percent of time Posted
4‐25
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 4‐42 IH Sampling Locations
4.2.2.1 Sensitivity Analysis of AdvectionDispersion Coefficients
A series of simulations were made with different coefficients and “formulation schemes” – either based
on Eddy velocities or model vector velocities. The values were used in each test are shown in Table 4‐2.
The existing scenario posting rate – as shown in Table 4‐1 were used as the basis and the resulting
posting rates predicted by the new coefficient settings are compared in Table 4‐3 as percent change
from the status quo. The most variance is found at IH‐8 and IH‐11, the two stations farthest from the
Don Mouth. Trial D and E had identical results. The velocity formulation D had same overall average to
the Eddy formulation of trial B. Trial C had almost identical average to the existing scenario, did the
smaller values compensate each other?
The alternative dispersion coefficients do not produce large changes at all stations, some locations have
up to 10% differences from the status quo. If we look at stations IH‐10, the 2007 & 2008 prediction
value is over the observed value and setting the dispersion coefficient to schemes a, b, d, would lower
the predicted values. However that would affect the IH‐11 results which had the 2007 over predict and
the 2008 under predict, using any scheme would improve the 2007 but make the 2008 result worse.
The same would happen at IH‐8. Consequently the default values appear to be acceptable. The
formulation selected also does not seem to cause major changes in the predictions.
4‐26
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Table 4‐2 Dispersion Coefficient and Formulation used in trials
Table 4‐3 AD dispersion coefficient tests – Percent change from existing scenario
a b c d e
61E 0.3 1.4 3.7 2.7 2.7
IH‐1 0.0 0.5 0.7 1.1 1.1
IH‐10 ‐5.6 ‐5.0 ‐1.7 ‐5.5 ‐5.5
IH‐11 ‐7.6 ‐9.9 ‐5.7 ‐10.3 ‐10.3
IH‐12 ‐1.3 ‐1.7 0.9 ‐3.5 ‐3.5
IH‐13 0.8 2.6 4.2 2.9 2.9
IH‐2 ‐1.6 0.7 1.4 1.4 1.4
IH‐3 ‐2.4 ‐2.9 ‐0.7 ‐2.6 ‐2.6
IH‐4 ‐1.0 ‐0.6 ‐0.3 ‐0.9 ‐0.9
IH‐5 0.0 0.0 0.6 0.0 0.0
IH‐6 ‐1.8 2.3 3.2 1.1 1.1
IH‐7 ‐0.4 ‐1.5 ‐1.2 ‐1.0 ‐1.0
IH‐8 ‐6.7 ‐6.3 ‐5.4 ‐5.8 ‐5.8
IH‐9 ‐2.7 ‐4.3 ‐2.7 ‐4.7 ‐4.7
L‐10 ‐1.4 0.2 2.8 1.4 1.4
L‐12 0.1 ‐2.4 ‐1.1 ‐2.6 ‐2.6
MOE 1364 ‐2.1 ‐2.1 0.2 ‐2.5 ‐2.5
Average ‐1.6 ‐1.4 0.2 ‐1.4 ‐1.4
4‐27
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
4.2.3 Calibration of other Nested Model Domains
This section describes simulations made after the spill modelling was completed. During the spill
modelling period ADCP data was not available to use for correlations, and some spills had to be released
in years without ADCP data. ADCP data has been found and made available from the MOE. Figure 4‐37
shows Lake Ontario with the various 270m nested grids used in the spill modelling. The270 m grids that
were re‐run for ADCP correlation were Grimsby‐Port Dalhousie (Figure 6‐30), Burlington Ship Channel
(Figure 7‐), Port Hope‐Cobourg (Figure 5‐9, 7‐25), Oshawa to Hamilton (Figure 3‐19 – main Spill grid),
Credit River (Figure 7‐52). The various Fnorm scores and time series are provided below.
Figure 4‐43 Lake Ontario 270 m nested grids locations ‐ for additional ADCP comparison
4.2.3.1 Gibraltar Point ADCP in the larger OshawaHamilton Grid
The Oshawa‐Hamilton 270 m grid was re‐run for 2008 and the ADCP data collected at Gibraltar Point
was used to compare model results. The initial model configuration of 30 layers and 3 m thickness did
not produce a good Fnorm score although the Fnorm score of 40 layers and 2 m thickness did produce a
good value when run in the Inner Harbour configuration. The number of layers were increased to 80 at
4‐28
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
2 m thickness and the Fnorm score improved to a value of 0.88. Figures 4‐38 and 4‐39 show the time
series of speed and direction.
Figure 4‐44 MOE and City ADCP locations used in the calibration of the nested grids
Figure 4‐45 Gibraltar Point ADCP Location Speed Time Series in Oshawa‐Hamilton Main Spill G
Figure 4‐46 Gibraltar Point ADCP Location Direction Time Series in Oshawa‐Hamilton Main Spill Grid
4‐29
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
4.2.3.2 Credit River 270m Nested Grid
The MOE provided two data sets of nearshore ADCP deployments in 2004 – the locations are shown on
Figure 4‐44. The Fnorm scores were poor for the 30 layer model and increasing the number of layers to
80 and 2 m thickness did not improve the performance.
4‐47 MOE ADCP Locations for Credit River nested Grid Calibration
The #3541 surface layer had a Fnorm score of 1.29 at the original 30 layer configuration, when the layers
were increased to 80 the score was 1.34, Figures 4‐40 and 4‐41 show the time series of speed and
direction.
Figure 4‐48 #3541 speed time series
4‐30
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 4‐49 #3541 direction time series
ADCP #3223 had a surface layer score of 1.23 in the surface layer, and the 80 layer model had a score of
1.28, Figures 4‐42 and 4‐43 show the time series.
Figure 4‐50 #3223 speed time series.
4‐31
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 4‐51 #3223 direction time series at surface layer
4.2.3.3 GrimsbyPort Dalhousie 270 m nested grid Calibration
The CCIW ADCP data from 2006 – discussed in Section 3‐14 – was used to test this model domain –
Figure 4‐51 shows the locations of the ADCPs.
The Grimsby‐Port Dalhousie Grid was re‐run with 60 and 80 layers at 2 m thickness. The Fnorm scores
were not good at all. The best scores were 1.27 for 60 layers and 1.38 for 80 layers for ADCP 1269.
ADCP 1268 scored 1.68 and 1.8 respectively. Figures 4‐44 and 4‐45 show the directions and speed time
series for 1269. The model amplitude for speeds are agreeable with the ADCP data, so time of travel of
spills will be reliable. The directions are somewhat agreeable although there are periods of circular
motion.
Figure 4‐52 ADCP locations for Grimsby nested grid calibration
4‐32
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 4‐53 80 layer surface TS for direction ADCP 1269 Grimsby
Figure 4‐54 80 layer surface TS for speed ADCP 1269 – Grimsby
4.2.3.4 Burlington Ship Channel nested grid calibration
The CCIW ADCP data were used to test this domain – Figure 4‐33 shows the locations in the nested grid.
The Burlington Ship Channel grid had similar poor results as the Grimsby case. The Fnorm scores were
1.28 for 1269 and 1.62 for 1268 in the 80 layer setup. Figure 4‐46 and 4‐47 show the time series for
ADCP 1269. The overall amplitudes of speed are agreeable with the ADPC. The spill scenarios were
based on observed directions so the time of travel should be reasonable.
4‐33
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 4‐55 ADCP locations
4‐34
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 4‐56 ADCP 1269 Direction TS for 80 layers
Figure 4‐57 ADCP 1269 Speed TS for 80 layers
4.2.3.5 Port Hope Cobourg nested grid calibration
MOE provided ADCP data offshore of Port Hope #3514 ‐ April to November 2008, and offshore of
Cobourg #3172 ‐ same period – Figure 4‐44 shows the locations in the 270 m nested grid. The 30 layer
3m thickness model was run for 2008 using NOAA winds. The Fnorm score for #3514 was 0.89, the time
series of speed and direction are provided in Figures 4‐48 and 4‐49. The Fnorm score for #3172 was only
1.15, the time series are shown in Figures 4‐50 and 4‐51. The depth at #3514 was 10m and #3172 was
20m.
4‐35
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 4‐58 ADCP locations
The layers were increased to 80 and the Port Hope #3514 Fnorm score went up to 0.99. The bottom
layer had a score of 1.75, the original 30 layer score was 1.35. The bottom speeds from the model were
significantly higher than the ADCP data. Similarly the Cobourg score went from 1.15 to 1.29 in the top
layer for the 30 and 80 layer model.
Figure 4‐59 #3514 Speed time series
4‐36
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 4‐60 #3514 Direction time series
Figure 4‐61 #3172 speed time series
Figure 4‐62 #3172 direction time series
4‐37
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
4.3 Calibration Summary by Model Domains
The model calibration of the hydrodynamics was limited to a few areas due to lack of data of observed
currents. Table 4‐2 lists the nested grids used in the spill modelling and the calibration.
Table 4‐4 Summary of model grids calibration
4‐38
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
4‐39
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
4.4 Conclusions
The MIKE 3 hydrodynamic model was calibrated with ADCP and thermistor data collected from several
sites on Lake Ontario. The simulations have a range of accuracies, some great, some not so good.
Generally, the model can produce accurate simulations if the vertical resolution is deeper than the
thermocline, usually 30 layers at 3 m or 40 m at 2m resolution is sufficient for good correlations with
observations. The complex shoreline of the Toronto Inner Harbour was often beyond the abilities of the
model to resolve the currents at the resolution used in this set of simulations.
The advection‐ dispersion module was also calibrated with observed data from a spill and longer term
continuous simulations. The model was able to simulate the spill time series at most locations, while
the long term simulations had good correlations with the observed data at most locations.
A summary of the model calibration parameters is presented in Table 4‐3.
By comparison, Golders (2009) used the MIKE‐3 system to model thermal cooling at the proposed new‐
build at Darlington, they published the model parameters that best fit their observations.
4‐40
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Table 4‐5 Model Calibration Parameters
Temperature Dispersion Scheme Eddy velocity relationship
Heat Exchange Coefficients
Displacement (Day light saving ‐1
time)
4‐41
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
AD Module
Substances Both conservative and first order
decay
Initial ambient conditions zero
Decay rates Tritium – half life some 12 years
E. coli T90 57 hours
Benzene – both conservative and
first order decay
Dispersion Coefficients default
Dispersion scheme Eddy velocity relationship
4‐42
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
5 Tritium Spill Scenarios
5.1 Tritium Spill Event to the East
What would be the impact on the water intakes to the east of Pickering if the plume had traveled east
without a current reversal?
The existing Provincial Water Quality Objective for Tritium is 7000 Bq/L. The August 2 1992 event had
peak levels below 700 Bq/L at the Toronto Intakes. There is a proposal to make the Ontario Drinking
Water Standard (ODWS) as low as 20 Bq/L. What levels would be observed at the easterly water intakes
if this event occurred during a long term easterly current?
The 1992 spill event ‐ 2900 kg at 7.9 x 1011 Bq/l in 6 hours ‐ was released during a period of easterly
currents, in this case on May 17, 2006 at 1200 hrs, see Section 3.11 ‐Figure 3‐1 for nested grid ‐ Figure 3‐
3 for current event. The temperature was set to 35 °C as it was released into the cooling water channel
of the plant, assumed to be running at high power generation rates due to summer demand. The lake
was still cold, between 4 and 8 °C having had an upwelling event just prior to the release. Figure 5‐1
shows the time series of Tritium predicted at the WTP intakes. Ajax has a peak of 2000 Bq/l and the
plume takes about 24 hrs to pass. Whitby has a peak of 12,000 Bq/L and passage takes about 24 hrs.
Oshawa's peak is 20,000 Bq/L and passage is about 48 hrs. The Darlington site, (no intake‐ just a point
for comparison), peaks at 6000 Bq/L and also has a 48 hr passage time. The Oshawa intake is shallow
and close to shore and therefore receives the highest level as the plume is still attached to the coastline.
Figures 5‐2 to 5‐ 5 show the surface plume as it passes each intake.
The plume reverses and come back across the intakes, with levels of between 2 and 100 Bq/L and these
levels persist for some time, as shown in Figure 5‐6, where the scale has been changed to logarithmic to
show the range.
5‐1
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 5‐1 WTP intakes ‐ Tritium levels
Figure 5‐2 Ajax Intake impact
5‐2
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 5‐3 Whitby Intake impact
Figure 5‐4 Oshawa Intake impact
Figure 5‐5 Darlington site impact
5‐3
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 5‐6 Long Range time series at the intakes.
5.1.1 Tritium Plume Impacts on Intakes Between Bowmanville and
Wellington
The MIKE‐3 model grid was extended further east to encompass the five intakes along the north shore
of Lake Ontario. Figures 5‐7 to 5‐9 show the model domains and WTP intakes in this simulation. In
order to get fine near shore resolution, it was necessary to extend the 810m grid from the spill location
all the way to east of Cobourg. The Wellington Intake was modelled with the 2430 grid. The smaller
resolution grid will provide a bit more "accuracy" with respect to the peak levels of Tritium, the courser
resolution results are not that far off for the Tritium spill.
Figures 5‐10 to 5‐12 show the time series of the Tritium plume passage. Table 5‐1 lists the model results
for each intake. The duration of initial plume was based on the time from first impact to when Tritium
levels fell below 20 Bq/L. on the first wave. Port Hope and Cobourg show the two results for each grid
size. Port Hope is predicted to have a peak level of 755 Bq/L based on the 270 m grid, while Cobourg can
expect a peak of around 530 Bq/L. The time to peak for each grid resolution is not significant, however
the duration is doubled for Port Hope and larger for Cobourg at the finer resolution. Wellington will
experience a very small plume with negligible impact. Bowmanville and Newcastle will have a second
extended wave of at least 9 days above 20 Bq/L.
5‐4
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Table 5‐1 Easterly Tritium Spill
755 ‐ 270 16 days (384 hrs)
525 ‐ 270 7.5 days (180 hr)
Figures 5‐13 to 5‐17 show the surface plume of Tritium around the time of peak transit at each intake.
The Port Hope Intake experiences a second pass and that event has been included. The Wellington
plume is not shown ‐ the plume initially spreads south across the lake around the time it passes
Bowmanville and the southern tongue of the plume eventually moves counter‐clockwise around the
lake until it gets to the Wellington intake.
5‐5
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 5‐7 Extended Easterly 2430 grid
Figure 5‐8 Extended 810 m grid with intakes
Figure 5‐9 Extended 270m grid with Intakes
5‐6
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 5‐10 810m Time Series
Figure 5‐11 270m Time Series ‐ Port Hope & Coburg
5‐7
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 5‐12 2430m Time Series – Wellington
Figure 5‐13 Bowmanville Intake Plume
Figure 5‐14 Newcastle Intake Plume
5‐8
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 5‐15 Port Hope Intake Plume ‐ First pass
Figure 5‐16 Port Hope Intake Plume ‐ Second pass
Figure 5‐17 Cobourg Intake Plume
5‐9
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
5.1.2 Tritium Spill at Darlington NGS
A spill equal to the August 1992 event at Pickering was modelled as occurring at the Darlington cooling
water discharge diffuser. The cooling water flow is about 150 m3/s with a 35 °C temperature, through a
diffuser 800 m offshore of the site ‐ the diffuser was modelled as 270m in length ‐ shorter than the
exiting unit. The tritium spill was again repeated at regular intervals (about 7 days apart) to sample the
currents to obtain both easterly and westerly events. The spatial extent of the plume at 7000 Bq/L and
20 Bq/L are shown in Figures 5‐18 and 5‐19. The 7000 Bq/L plot is based on the 810 m grid that covers
the shoreline from Pickering to Cobourg. The 20 Bq/L plot is based on the whole lake grid at 2430, to
show the extent of the plume.
The time series of Tritium at each Easterly intake is shown in Figure 5‐20.
The Peak levels are listed in Table 5‐2 ‐ along with the Pickering predictions ‐ for comparison. Table 5‐3
lists the results for the Westerly Intakes
Table 5‐2 Tritium Levels at Easterly Intakes
5‐10
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 5‐18 810 m Grid 7000 Bq/L map
Figure 5‐19 2430 m Grid 20 Bq/L map
5‐11
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 5‐20 Intake time series
5‐12
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Table 5‐3 Tritium Levels at Westerly Intakes
Island 500 (shallow Intake)
Note: Pickering data based on 270 m grid, Darlington based on 2430 grid results.
5‐13
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
6 Wastewater Discharge Spill Scenarios
The first three spill scenarios (Sections 6.1 to 6.3) examined the impact from individual wastewater
treatment plants on the local water intakes. These were "exploratory" scenarios to determine if the spill
of high levels of bacteria and ammonia would be a threat to the water intakes. Modelling results
indicated that ammonia was not a significant parameter and it was not considered further in the
modelling process. Bacteria were considered a threat so more extensive modelling was undertaken. In
addition to the impact of E. coli, other pathogens are associated with wastewater, so that if E. coli levels
are high, then it is assumed that other public health risks are possible.
6.1 Wellington Intake Water Quality Impact from Wastewater Discharge
The same model setup was used to simulate a spill of sewage at the outfall and predict the water quality
impacts at the intake.
Figures 6‐1 and 6‐2 show the whole lake model and the 90m nested grid in Wellington Bay. The location
of the outfall (from Canadian Hydrographic chart 2060) and the intake (Stantec report ‐ Rodney
Bouchard private communication) are shown in Figure 6‐2.
A spill was modelled as continuous at a rate of 1500 m3/day with a E. coli concentration of 5,000,000
#/100mL (raw sewage) and ammonia of 15 mg/L (raw sewage). The sewage temperature was set at
20°C. Both parameters were modelled as conservative substances ‐ no decay.
The predicted water temperature at the intake ‐ Figure 6‐3, predicted current directions at outfall and
intake ‐ Figures 6‐4 and 6‐5 show when events will be responsible for transporting (advection) the
wastewater plume from the outfall to the intake. In this case currents flow towards the east (in the
range of 80 °N), with some reversals swinging the plume around to cross over the intake.
Figures 6‐6 and 6‐7 show the predicted levels of E. coli and Ammonia at the intake. The peak E. coli level
is about 1000 #/100mL while the ammonia peak at 0.003 mg/L ‐ both parameters have 5000:1 dilution.
The peak events coincide with currents in the 80°N direction.
Figure 6‐8 is a compilation of the event of April 22, the peak level of E. coli in each grid cell for the 96
hours after the event was plotted, this shows the maximum level that will be experienced at each grid,
or a spatial extent of the event.
6‐1
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 6‐1 Lake Ontario 2430m grid
Figure 6‐2 90m grid with intake and outfall locations
6‐2
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 6‐3 Predicted lake temperature at Intake
Figure 6‐4 Predicted current direction at Outfall
6‐3
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 6‐5 Predicted current direction at Intake
Figure 6‐6 Predicted E. coli levels at Intake
6‐4
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 6‐7 Predicted Ammonia levels at Intake
Figure 6‐8 Spatial extent of plume for April 22 peak
6‐5
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
6.2 Ashbridges Bay Treatment Plant Bypass Spill
Ashbridges Bay Treatment Plant has the largest discharge of municipal treated wastewater to Lake
Ontario. A continuous spill of untreated wastewater was modelled with the 270m grid. The discharge
rate was set to 15 m3/s with ammonia level set to 18 mg/L and E. coli set at 5,000,000 #/100ml. Both
parameters were set as conservative. Additional simulations for E. coli discharges were made, with a
first order decay of 1.1 x 10‐5/s or 60% die‐off per day (T90 of 57 hrs), based on the Toronto Inner
Harbour calibration (Section 4.2).
Figures 6‐9 to 6‐11 show the model predictions for intakes east of ABTP and while Figures 6‐12 to 6‐14
show results for intakes west of ABTP. The spatial extent of the E. coli plumes are shown in Figures 6‐15
to 6‐18. These isopleth plots are generated by scanning the hourly data over a three week period to
determine the spatial extent of the plume. The ammonia plumes were similar, just a change in
amplitude and therefore are not provided in this report.
Figure 6‐9 Ammonia for East Intakes
Figure 6‐10 E. coli for East Intakes
6‐6
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 6‐11 E. coli with Decay for East Intakes
Figure 6‐12 Ammonia for West Intakes
Figure 6‐13 E. coli for West Intakes
6‐7
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 6‐14 E. coli with Decay for West Intakes
Figure 6‐15 Easterly extent for E. coli
Figure 6‐16 Easterly extent for E. coli with decay
6‐8
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 6‐17 Westerly extent of E. coli
Figure 6‐18 Westerly extent of E. coli with decay
6.3 Lakeview E. coli Spill
A long term spill of effluent from the plant was simulated. Bacteria level was 5,000,000 #/100mL and
first order decay was used. Model runs were made for a winter period from January 1, 2006 to May 1
2006 and for a summer period from May 1 2006 to August 31 2006.
The time series of E. coli at the closest intakes are shown in Figure 6‐19 for the winter period and 6‐20
for the summer period. The peak levels for each intake and dilution ratio are shown in Table 6‐1.
Table 6‐1 Lakeview Bypass E. coli Peaks and Dilution Ratios
6‐9
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 6‐19 Lakeview E. coli time series ‐ Winter Period – note smaller scale on right axis for Oakville and Lorne
Park
Figure 6‐20 Lakeview E. coli time series ‐ Summer Period – note smaller scale on right axis is for Oakville and
Lorne Park
6‐10
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
6.4 Wastewater Treatment Plant Disinfection Failure
Two alternative forms of Disinfection failure were considered: (i) total plant failure and (ii) loss of the
disinfection process at the WWTP. The abstraction used to develop the spill scenario for WWTPs (SPC,
2011b) in this report, are based on a total plant failure. This is abstracted from the total plant failure at
the Peel (/G.E. Booth/Lakeview) WWTP (SPC, 2011b) but could also be caused by a prolonged electrical
power interruption; for this abstraction, it is assumed that there are no removal of E Coli through the
WWTP, ‐ hence the effluent E Coli density equals the influent E Coli density (5,000,000 E Coli / 100 ml).
An alternative process failure, rather than a total plant failure, would be only disinfection system failure;
in this abstraction, limited monitoring information available to the SPC team suggested that a typical
primary – secondary WWTP achieves approximately an order of magnitude reduction in E Coli level’s
(hence the effluent E Coli level would be of the order of 500,000 / 100 ml) . A third alternative cause
would be a plant bypass ( including a disinfection system bypass) associated with a major Wet Weather
event to the WWTP; because one or more of the WWTPs of the LOC are designed and built so that such
a bypass can not occur, this scenario has not been evaluated to date.
Each of the WWTPs along the north shore was simulated as having a total disinfection failure in the
plant and all effluent discharged through their respect outfall/diffusers. In the scenario that was
modelled, a two‐day total plant failure was assumed at each of the wastewater treatment plants, and
the two‐day event was modelled as successive events for the normal weather conditions found during
May to October so that different lake circulation patterns and speeds could be assessed. The
simulation date was April 25 to August 31, 2008. Each of the simulation results was enumerated to
determine the peak level at the various intakes upstream/downstream from them. The 270 m tritium
spill grid was used. Although a prolonged period with no disinfection is un‐likely, the extended spill was
used to sample the seasonal conditions in the lake, from isothermal to summer thermal stratification.
What is useful is the peak values and durations of events predicted, and these would occur even if the
spill was only for a few days.
Each WWTP was simulated at the design CofA flow rate and the E. coli level in the discharge was set
constant at 5,000,000 #/100mL ‐ with first order decay in the lake. The Mid‐Halton plant was discharged
through the Oakville SW outfall; a new outfall for this plant is being planned as discussed in an EA
(Region of Halton, April 2010).
Table 6‐2 lists the resulting peak levels at each intake of interest and the dilution ratio from the WWTP
outfall.
The Skyway and Woodward WWTP discharge inside Hamilton Harbour. Both discharges cause hydraulic
flushing from the Harbour to the Lake; the discharge E Coli Level was set at 5,000,000 E. coli/100mL.
Flows into the Harbour from Spencer Creek and Grindstone Creek were also included to mimic hydraulic
flushing caused by watershed runoff into Harbour; their bacteria levels were set at zero. Lake – bay
6‐11
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
exchange cause by lake seiches were calculated internally by the model. The model grid was the same as
the Burlington ship canal benzene spill. Table 6‐3 lists the results.
The WWTPs in Durham Region east of Duffins Creek, Corbett Creek, Harmony Creek and Courtice were
run using the Easterly Tritium grid ‐ Table 6‐4. The Port Hope and Cobourg WWTPs were run with the
270 m Port Hope‐Cobourg grid, Cobourg has two plants ‐ labelled east and west.
The percent of time the intakes would experience E. coli above the PWQO are listed in Table 6‐5.
6‐12
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Table 6‐2 WWTP Bypass ‐ Peak level ‐ Average Level
WWTP: Duffin Creek Highland ABTP Humber Lakeview Mid‐Halton Oakville SE Oakville SW Clarkson
Intake: Peak Mean Peak Mean Peak Mean Peak Mean Peak Mean Peak Mean Peak Mean Peak Mean Peak Mean
Units are
(#/100mL)
Horgan 2470 173 10471 810 1373 52 100 3 45 1.2
Harris 450 21 1308 66 4911 200 216 15 110 6
Island West
14 0.12 3 0.03 68 1 28 1.1 41 0.3
Deep
Clark 23 0.43 32 0.6 2671 80 11688 334 55600 5500 32 1 52 2 35 1.3 1400 42
Lakeview 37 0.8 780 40 2906 100 83800 1600 62 2 58 3 46 2 1426 59
Lorne Park 13 0.3 756 16 734 33 38000 2400 248 11 539 26 216 14 5600 529
Oakville 2 0.05 108 2 78 2 3070 70 5756 766 1456 105 12168 1820 9950 593
Burloak 56 1.5 66 1.4 1000 22 1367 33 265 9 637 60 889 50
6‐13
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
WWTP: Duffin Creek Highland ABTP Humber Lakeview Mid‐Halton Oakville SE Oakville SW Clarkson
Intake: Peak Mean Peak Mean Peak Mean Peak Mean Peak Mean Peak Mean Peak Mean Peak Mean Peak Mean
Units are
(#/100mL)
Burlington 11 0.1 6 0.1 20 0.5 6153 425 103 1.7 1050 40 623 9
Table 6‐3 Skyway & Woodward Impacts
6‐14
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Table 6‐4 Cobourg, Port Hope, Corbett, Harmony and Courtice Impacts
Intake: Peak Mean Peak Mean Peak Mean Peak Mean Peak Mean Peak Mean
(#/100mL) (#/100mL) (#/100mL) (#/100mL) (#/100mL) (#/100mL) (#/100mL) (#/100mL) (#/100mL) (#/100mL) (#/100mL) (#/100mL)
* NOTE: Bowmanville & Newcastle are estimates based on similar distance from Courtice to Oshawa (Bowmanville) and Courtice to Whitby (Newcastle)
6‐15
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Table 6‐5 Percent of Time E. coli above PWQO 100#/100mL
Intake/Source Cobourg Cobourg Port Courtice Harmony Corbett Duffins Highland ABTP Humber Lakeview Mid‐ Oakville Oakville Clarkson Skyway Woodward
East West Hope CK Halton SE SW
Island Shallow
Island Deep
6‐16
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Intake/Source Cobourg Cobourg Port Courtice Harmony Corbett Duffins Highland ABTP Humber Lakeview Mid‐ Oakville Oakville Clarkson Skyway Woodward
East West Hope CK Halton SE SW
* NOTE: Bowmanville & Newcastle are estimates based on similar distance from Courtice to Oshawa (Bowmanville) and Courtice to Whitby (Newcastle)
6‐17
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
6.5 Sanitary Trunk Sewer Impacts
6.5.1 Highland Creek Sanitary Trunk Sewer break
During the August 19, 2005 rainstorm, a major storm event, the bed of Highland Creek was scoured out
where a trunk sewer crosses and the sewer pipe broke, allowing a release of raw sewage to the
tributary. The estimated hydrograph, from HEC‐RAS modelling, of the flow is shown in Figure 6‐21.
The NOAA winds for 2005 were applied to the model, with Pearson data used for the heat exchange.
Figure 6‐21 Highland Creek Discharge Rate
The effluent was modelled as having a TSS level of 1000 mg/L and the E. coli level was 1,000,000
#/100mL. City staff estimated the discharge rate was 0.6 m3/s from the broken main and lasted three
days. Figure 6‐22 shows the E. coli levels predicted at the Horgan Intake, the dilution was 14285:1. The
peak level arrived at the intake about 34 hours after the peak in the river flow. The raw water bacteria
sampling detected one (1) E. coli/100mL on Aug 21 at 0200 hours. There was another spike in bacteria
at the intake on August 25, which may have been caused by the storm event on a nearby tributary
Figure 6‐22 Horgan Intake E. coli levels – predicted level in black, observed levels in red (FC) and blue (E. coli)
6‐18
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
6.5.2 Mega Event for STS Breaks Centered on the Toronto Area
Scenario 1 – Mega Event from Intense Aug 2005 rainstorm
This scenario assumes an intense rain event/wind storm similar in magnitude to the August 19‐20th
2005 event. During the August event, a number of roads and a trunk sewer were washed out due to a
fast peaking event. This event will involve simultaneous main trunk breaks crossing each of the main
tributaries along the Toronto Waterfront. The break was assumed to last 24 hours.
The daily average flows for each tributary were obtained from the Canada Water Survey Branch of
Environment Canada. The Highland Creek gauge was not operating during most of 2005 so an estimated
time series was provided by TRCA.
The trunk sewer flows discharges to receiving waters, were estimated as a portion of the design flow of
each WWTP that they service, the flows are listed below.
Rouge R (York‐Durham‐YDSS) 1.8 m3/s
Highland Creek 0.6 m3/s
Don River 2.2 m3/s
Humber River 1.77 m3/s
Etobicoke Creek(Booth S T S ) 1.4 m3/s.
The Booth (formerly Lakeview WWTP) STS break was assumed to occur at a crossing of Etobicoke Creek
and discharge at the mouth.
The sewer flows were added to the tributary flows at the river mouths ‐ no dilution nor time of travel
were considered for the upstream portions.
The simulation for E. coli assumed the ambient level was 1000 #/100mL in each tributary and during the
trunk sewer break, the trunk sewers would contribute 5,000,000 #/100mL. The ambient lake conditions
were assumed zero E. coli and TSS ‐ first order decay of E. coli was applied.
The simulation for TSS assumed 10 mg/L TSS for ambient conditions. Environment Canada estimated
the total loads of TSS for the following tributaries;
Humber River 264879 MT
Don River 49918 MT
Highland Creek 15507 MT
6‐19
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
The event mean concentration was calculated to be:
Humber River 15500 mg/L
Don River 4814 mg/L
Highland Creek 1794 mg/L
Rouge River 5260 mg/L (based on Highland Creek)
Lakeview 2000 mg/L (assumed raw sewage level).
Figures 6‐23 and 6‐24 show the time series for each parameter.
The peak levels at each intake are tabulated in Table 6‐6.
Table 6‐6 Mega Event Peak levels at Intakes
6‐20
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 6‐23 E. coli time series for Mega event
Figure 6‐24 TSS time series for Mega event
6‐21
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
(b) Scenario 2 – Mega Event occurring randomly over Summer Period
A second scenario was run with the trunk sewer breaks occurring at 5 to 6 day intervals over the
simulation period, in order to represent other current conditions in the lake. During these events the
tributary flows were based on observed data, which meant that STS breaks could occur during dry
weather or other flow conditions in the tributaries. Figures 6‐25 to 6‐28 show the time series for the
intakes.
Table 6‐7 lists the peak values for E. coli and TSS. The peak values are larger, indicating that different
current conditions can create high levels at the intakes, even when the tributaries are not flowing at
extreme storm flow rates.
Figure 6‐25 East Intake E. coli time series
6‐22
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 6‐26 West Intake E. coli time series
Figure 6‐27 East Intakes TSS time series
6‐23
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 6‐28 West Intakes TSS time series
Table 6‐7 Mega Event Peak Levels for STS Breaks Occurring Randomly Over the Summer Period
Horgan 299 5
Harris 175 4
Island Shallow 28 2
Burlington 6 1.3
6‐24
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Table 6‐7 does not define which tributary caused the peak levels at each intake. Separate simulations
were made with for tributary providing the high E. coli levels, while all others were at normal (non –
impacted ) levels. Table 6‐8 lists the individual contributions to each intake, the impacting tributary is
highlighted, and summarized as:
• Horgan is impacted by Highland Creek ,
• Harris is impacted by the Don River,
• Clark is impacted by Etobicoke Creek and then to a lesser extent by the Humber River,
• Lakeview ‐ same as Clark,
• Lorne Park is impacted by Etobicoke Creek,
• Oakville is impacted by Etobicoke Creek.
Table 6‐8 Multiple Mega Event ‐ Separate Tributary Contributions
6‐25
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
6.6 Industrial Food Processor Lagoon Spill
A spill of the total contents from a lagoon to the upper reach of the Credit River was simulated. Credit
River Conservation provided the time of travel and subsequent dilutions of the plume as it travelled
down the river ‐ from HEC‐RAS model.
The spill details are:
Volume release 52,800 m3 in a 24 hour period.
E. coli concentration in lagoon ‐ assumed 5,000,000 #/100mL
River at flood stage ‐ as per CVC model.
E. coli concentration at mouth of river 25.33 #/100mL (concentration input to Lake Model).
As in the gasoline pipeline spills, the spill was released a number of times over the simulation period to
sample most of the possible current regimes. The Credit River spill grid was used for the model. Figure
6‐29 shows the time series of E. coli at the relevant intakes. The peak levels are below one E. oli/100mL..
6‐26
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 6‐29 E. coli time series
6.7 Grimsby and Port Dalhousie WWTP Disinfection Failure on Grimsby
Intake
The Grimsby and Port Dalhousie WWTP were simulated with disinfection failure and E. coli levels in the
discharge at 5,000,000 #/100mL. Figure 6‐30 shows the 270 m grid and locations of the Grimsby Intake
and wastewater discharges.
Figure 6‐30 270 m grid of Niagara Outfall & Intake
6‐27
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
The bypass event was continuous and the time series of E. coli at the Intake is shown in Figure 6‐31.
Figure 6‐31 Intake time series
The Grimsby discharge causes a peak level at the intake of 3100 #/100mL, while the Port Dalhousie
discharge only causes a peak of 10 #/100mL. The percent time E. coli was above the PWQO of
100#/100mL was 15% for Grimsby and zero for Port Dalhousie. The spatial extent of the plumes at 100
#/100mL are shown in Figures 6‐32 and 6‐33. These plots were generated from the hourly data and
storing the locations where the E. coli exceeded the criteria.
Figure 6‐32 Grimsby impact at 100#/100mL
6‐28
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 6‐33 Port Dalhousie impact at 100#/100ml
This is the end of the spill scenarios examining the impacts of bacteria (E. coli) on the water intakes.
6‐29
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
7 Benzene Related Spill Scenarios
This section focuses on the impacts from spills that release benzene, a known carcinogen. Benzene is
added to gasoline to increase the octane number, generally in the range of 0.5 to 1 %. Any spill of
gasoline will have some small amount of Benzene, and Benzene is completely miscible with water.
7.1 Oakville Fuel Storage Tank Farm Benzene Spill
A spill of benzene at the Oakville Fuel Storage Tank Farm was simulated with the Lake Ontario Mike‐3
Model.
The waterfront from Hamilton to Whitby at a resolution of 270 m is shown in Figure 7‐1, with the spill
locations and the water intakes that could be impacted.
Figure 7‐1 270m Waterfront Grid
The storage tanks are rated at 26 million litres of gasoline. It was assumed that the Benzene
concentration was 1% and the spill lasted 6 hours. Therefore the benzene spill simulates an equivalent
release of 260,000 L over a six hour period. It is assumed the discharge is released to the lake in the
surface layer of the water after flowing down Bronte Creek. Benzene is fully miscible with water and will
subsequently be transported along the shoreline with the currents. The ODWS for benzene is 0.005
mg/L.
The method used for this, and other spills, is several random spills, about 5 to 6 days apart. This allows
sampling of lake currents over a range of lake conditions / events, both easterly and westerly. Benzene
was modelled as a first order decay of 6.E‐6/s, which also prevents a build up of benzene in the near‐
shore zone. This method differs from other simulations (such as for tritium) which tried to capture
specific events ‐ east or west ‐ with limited success.
The model was started May 15 2006 with isothermal conditions of 4° C and run until August 10.
The time series of Benzene at each intake was extracted from the hourly simulation results, the peak
values are listed in Table 7‐1, Figure 7‐2 shows the benzene levels at the impacted intakes.
7‐30
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Table 7‐1 Oakville Fuel Storage Tank Farm Benzene Spill
Intake Peak Benzene
Whitby e‐12 mg/L
Ajax e‐11 mg/L
Horgan e‐5 mg/L
Harris 0.0005 mg/L
Island Deep 0.002 mg/L
Clark 0.014 mg/L
Lakeview 0.5 mg/L
Lorne Park 1.25 mg/L
Oakville 9.0 mg/L
Burloak 0.67 mg/L
Burlington 11.0 mg/L
Hamilton 0.84 mg/L
7‐31
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 7‐2 Benzene time series at intakes
The benzene plume is calculated to persist for several days at each intake. Burlington has two big
events in June, which have levels above 0.4 mg/L for 3 days. Other intakes have levels above 0.5 mg/L
for up to 2 days.
The spatial extent of the plume at the 0.005 mg/L and 0.05 mg/L isopleths are shown in Figures 7‐3 and
7‐4.
Figure 7‐3 Oakville Tank ‐ 0.005 mg/L Benzene Isopleth
7‐32
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 7‐4 Oakville Tank Spill ‐ 0.05 mg/L Benzene Isopleth
7.2 Mini Spill Associated with a Ship at Oakville Tank Farm
A small spill from a ship loading gasoline at the pier was modelled. Again, benzene at 1%, was released
during the event.
Three scenarios were simulated:
1. 20000 L released in 15 minutes or 200 L of Benzene,
2. 50000 L released in 15 minutes or 500 L of Benzene,
3. 100000 L released in 15 minutes or 1000 L of Benzene.
The peak levels at each intake are tabulated in Table 7‐2. The results are almost linear as peak levels are
almost five times and double the base level from a 200 L spill.
Table 7‐2 Mini Spill Benzene levels
7‐33
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Burlington intake will experience the highest levels, the time above 0.005 mg/L is about 2 hours for #1,
10 hours for #2 and 13 hours for #3.
Figures 7‐5 & 7‐6 show the isopleths (0.005 and 0.05 mg/L) for the period May 15 to June 6, for the 200
L spill, Figures 7‐7 &7‐8 for the 500 L spill and Figures 7‐9 & 7‐10 show the 10000 L spill.
Figure 7‐5 Scenario #1 200 L spill, Benzene at 0.005 mg/L
Figure 7‐6 Scenario #1 200 L spill, Benzene at 0.05 mg/L
7‐34
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 7‐7 Scenario #2 500 L spill, Benzene at 0.005 mg/L
Figure 7‐8 Scenario #2 500 L spill, Benzene at 0.05 mg/L
Figure 7‐9 Scenario #3 1000 L spill, Benzene at 0.005 mg/L
7‐35
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 7‐10 Scenario #3 1000 L spill, Benzene at 0.05 mg/L
7.3 Keele Tank Farm Spill
A spill from a tank equivalent to the Oakville Tank Farm spill was simulated at the Keele Tank Farm
(Keele & 407). The storage tanks are rated at 26 million litres of gasoline. It was assumed that the
Benzene concentration was 1% and the spill lasted 6 hours. Therefore the benzene spill simulates an
equivalent release of 260,000 L over a six hour period. It is assumed the discharge is released to the
lake in the surface layer of the water at the river mouth.
In these simulations, one a release to the upper Don River and one to the upper reach of the Humber
River, multiple releases were used, spaced at 5 to 7 day intervals. The Canada water Survey gauge data
for each tributary was also used as hydraulic forcing at the river mouth. The scenarios were run from
May 15 to August 15.
In these simulations Benzene has been set as a first order decay parameter, with a decay rate of 90%
reduction in 4 days.
7.3.1 Don River Spill Event
The time series of benzene at each of the intakes are plotted in Figure 7‐11.
7‐36
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 7‐11 Benzene time series for Don River spill
The peak levels for each intake are:
Ajax 0.004 mg/L
Horgan 0.038 mg/L
Harris 0.059 mg/L
Island Deep 0.009 mg/L
Clark 0.004 mg/L
Lakeview 0.003 mg/L.
The spatial extent of the plume for the period May 15 to June 6 at the peak level of Benzene at 0.005
mg/L in each grid cell is plotted in Figure 65. Table 7‐12 lists the complete intake results.
7‐37
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 7‐12 Benzene 0.005 mg/L isopleths from Don River spill
7.3.2 Humber River Spill
The time series for the intakes are shown in Figure 7‐13.
Figure 7‐13 Humber River Spill time series
7‐38
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
The peak levels at the intakes are:
Horgan 0.001 mg/L
Harris 0.0055 mg/L
Island Deep 0.015 mg/L
Clark 0.55 mg/L
Lakeview 0.31 mg/L
Lorne Park 0.078 mg/L
Oakville 0.002 mg/L.
Two spatial extent plots are presented, one for easterly intakes (Figures 7‐14 & 7‐15) and one for
westerly intakes ‐ Figure 7 ‐ 68. The easterly intakes are based on the time period May 15 to June 3 and
the westerly intakes are for the time period July 4 to July 17. The complete list of impacts at intakes are
found in Table 7‐3.
Figure 7‐14 Benzene 0.005 mg/L at easterly Intakes
7‐39
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 7‐15 Benzene 0.005 mg/L at westerly intakes
Table 7‐3 Benzene levels at Intakes
7‐40
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
7.4 Burlington Ship Canal Benzene Spill
A tanker trunk contains about 34,000 l of benzene. A spill from a tank lasting one hour was modelled
with the discharge to the Burlington Ship Canal.
A new model grid was used with nesting focused on the Western end of Lake Ontario. Figure 7‐16
shows the 810 and 270 m grids. The 810 m grid contains the Hamilton, Grimsby, Burlington and Oakville
intakes while the 270 m grid only has the Hamilton Intake.
Figure 7‐16 Spill & Intake Locations ‐ 810 m Grid
7‐41
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Hydraulic forcing from Hamilton Harbour is provided by the discharges from the Skyway and Hamilton
Woodward wastewater treatment plants and from Spencer and Grindstone Creeks. The simulations also
include lake – Bay exchange, caused by lake seiches. The model grid and related data were available for
2008, so 2008 NOAA winds were applied to the Lake.
The model was run and current events that could potentially transport the plume towards the south,
impacting Hamilton and Grimsby Intake, or towards the northeast impacting Burlington and Oakville
Intakes, were evaluated. Suitable events were found on June 4 to the south and July 17 to the
northeast.
The Hamilton and Grimsby Intake benzene time series is shown in Figure 7‐17. The peak level at
Hamilton is over .25 mg/L and remains above the PWQO for about 24 hours. The peak at Grimsby is
0.035 mg/L. The ODWS for benzene is 5 μg/L for drinking water.
Figure 7‐17 Hamilton & Grimsby Intake Benzene time series ‐ note the logarithmic axis.
7‐42
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
The surface plume impacting the two intakes is shown in Figures 7‐18 and 7‐19.
Figure 7‐18 Benzene Plume at Hamilton Intake
7‐43
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 7‐19 Benzene plume at Grimsby Intake
7‐44
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Three intakes along the north shore were impacted by the spill released on July 13. The time series for
the event is shown in Figure 7‐20. The peak levels and time of travel to peak predicted by the model
are:
The surface plume on July 15 and 16 are shown in Figures 7‐21 & 7‐22.
Figure 7‐20 Burlington, Burloak & Oakville Intakes time series
7‐45
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 7‐21 Easterly Plume July 15
7‐46
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 7‐22 Easterly Plume July 16
7.5 Credit River Benzene Spill
A spill of benzene from a tanker truck over a bridge of the Credit River was simulated similar to the
Burlington Bridge event – see section 7.7 for grid domain. The spill was made during a westerly and
easterly current episode. The spill is released at the river mouth, having been transported down the
river ‐ with no loss to the environment.
7‐47
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
The Westerly event was released on August 12 0000 hrs, the resulting time series of benzene at the
intakes is shown in Figure 7‐23. The Lorne Park intake is impacted first with the peak arriving about 60
hours after the release and then a second higher pulse during a current reversal after 3 days. Lakeview
and Clark intakes are then impacted with peaks at 3.6 and 5.2 days respectively. The peak level at the
Lorne Park intake almost reaches the 100 ug/L level for benzene.
Figure 7‐23 Westerly benzene time series
The Easterly event was released on July 26 at 0200 hours and the time series of benzene at the intakes
are shown in Figure 7‐24. The peak level at Lakeview is over eighty times the ODWS. The plume is
predicted to extend as far as the Horgan intake.
Figure 7‐24 Easterly benzene time series
7.6 Gasoline Spills in Ganaraska River, Cobourg Brook, Graham Creek,
Wilmot Creek and Bowmanville Creek
Background
The next series of spill scenarios is based on a break in the pipeline carrying gasoline from Montreal to
Sarnia. This pipeline crosses many of the tributaries along the north shore of Lake Ontario and a break
would release gasoline along with Benzene to Lake Ontario. The simulations are broken down by
regions based on model grids.
7‐48
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Spills into Ganaraska River through Bowmanville Creek
A gasoline spill from the petroleum pipeline which crosses the Ganaraska River and Cobourg Brook was
modelled with the Lake Ontario version of MIKE‐3. The 270 grid, which spans the area Port Hope to
Cobourg, was used ‐Figure 7‐25.
Figure 7‐25 Port Hope & Cobourg Intake Map
The simulation year was 2006. The daily flow in the river was obtained from the Canada Water Survey
database. The wind forcing was the NOAA 2D wind field, other met data were from Pearson Airport.
The simulation period was April 15 to July 7.
The pipeline flow was based on the average flow rate of 450 m3/hr, or 0.125 m3/s. Information,
available to the study team, indicated that regular gasoline, 87 Octane, has between 0.5 and 1%
benzene, added to increase the octane number. Assuming a 1% concentration, then 0.00125 m3/s of
pure benzene would be spilled during a pipe rupture. The pipeline flow was mixed with the river flow
and discharged at the mouth of the river. Benzene is fully miscible in water and it is assumed that the
benzene in the gasoline will fully mix in the river water ‐ there was no loss assumed from evaporation,
decay, or other losses.
The temperature in the tributaries were set constant at 20 °C, as was the gasoline temperature in the
pipeline. The lake temperature was predicted by the model, starting from 4 °C isothermal at start up
and developing the summer stratification.
The pipeline break was modelled as a six‐hour event. The event dates were randomly chosen ‐ usually
about 36 hour apart. This method provides a typical lake response, and does not rely upon selected
directional events.
7‐49
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
7.6.1 Ganaraska River Spill
7.6.1.1 Port Hope Intake
Figure 7‐26 shows the predicted benzene levels at the depth of the intake over the simulation period.
The maximum level was over 3 mg/L, which is well above the drinking water standards. The dilution is
over 330,000:1, but there is still a high level of benzene at the intake. Figure 7‐27 shows the surface
plume at the May 18 peak event ‐ which was a westerly current event.
Figure 7‐28 is a compilation of the event of May 18. The peak level of benzene in each grid cell for the 96
hours after the spill was determined and plotted; this shows the maximum level that are calculated in
each grid, or a spatial extent of the spill event.
7.6.1.2 Cobourg Intake
Figure 7‐29 shows the predicted benzene levels at the depth of the intake over the simulation period.
The maximum levels are over 1.0 mg/L. Figure 7‐30 shows the May 21 plume which was an easterly
current event from the Ganaraska River.
Figure 7‐26 Port Hope Intake Benzene levels
7‐50
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 7‐27 Port Hope May 18 event ‐ the yellow scale is 2.3 to 2.6 mg/L Benzene
Figure 7‐28 Spatial Extent of Spill
7‐51
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 7‐29 Cobourg Intake Benzene from Ganaraska River
Figure 7‐30 Cobourg Intake Impact from Ganaraska event ‐ the red scale is above 0.89 mg/L Benzene
7‐52
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
7.6.2 Cobourg Brook Spill
The same spill event scenario was run with the discharge from Cobourg Brook.
7.6.2.1 Cobourg Intake Impact
Figure 7‐31 shows the time series of predicted Benzene at the Cobourg Intake, The peak levels are above
3 mg/L. Figure 7‐32 shows the surface plume for the April 20 event and Figure 7‐33 shows the spatial
extent.
Figure 7‐31 Cobourg Intake Impact from Cobourg Brook
Figure 7‐32 Cobourg Intake impact from Cobourg Brook
7‐53
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 7‐33 Spatial Extent of Cobourg Brook Spill
7.6.2.2 Port Hope Intake
Figures 7‐34 and 7‐35 show the time series and surface plume on April 22 event from Cobourg Brook,
which is the same spill event from April 18; a current reversal brought the plume to Port Hope.
Figure 7‐34 Port Hope Intake ‐ Impact from Cobourg Brook
Figure 7‐35 Port Hope Intake ‐ Impact from Cobourg Brook
7‐54
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
7.6.3 Newcastle and Bowmanville Intakes Gasoline Spills from Wilmot
Graham and Bowmanville Creeks
Figure 7‐36 shows the 270 grid centered on Bowmanville and Newcastle.
Figure 7‐36 Bowmanville to Newcastle Grid Map
The same wind and other meteorological data were applied to this grid as the Port Hope model. The
only difference was the simulation period, the Bowmanville Creek data was only available after August
21, so the simulation started later ‐ May 30.
7.6.3.1 Bowmanville Creek Spill
Figures 7‐37 and 7‐38 show the time series and surface plume at the Bowmanville Intake, due to the
spill from Bowmanville Creek. The Bowmanville Intake has peak levels above 10 mg/L, while the
Newcastle Intake has peaks of about a factor of 10 less, or 1.0 mg/L. Figures 7‐39 and 7‐40 show the
Newcastle Intake time series and surface plume. The same event causes the plume at each intake. The
spatial extent is shown in Figure 7‐41.
7‐55
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 7‐37 Bowmanville Intake Time Series
Figure 7‐38 Bowmanville Intake ‐ Impact from Bowmanville Creek
7‐56
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 7‐39 Newcastle Intake Impact from Bowmanville Creek
Figure 7‐40 Newcastle Intake ‐ Impact from Bowmanville Creek
7‐57
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 7‐41 Spatial Extent of Bowmanville spill
7.6.3.2 Wilmot Creek Spill
Figures 7‐42 & 7‐43 show the time series and surface plume at the Newcastle Intake due to a spill on
Wilmot Creek. The peak levels are above 3 mg/L.
Figure 7‐42 Newcastle Intake Time Series from Wilmot Creek Spill
7‐58
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 7‐43 Newcastle Plume from Wilmot Creek
Figures 7‐44 and 7‐45 show the impact on the Bowmanville Intake from a spill in Wilmot Creek, and
Figure 7‐46 shows the spatial extent.
Figure 7‐44 Bowmanville Intake ‐ Impact from Wilmot Creek
7‐59
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 7‐45 Bowmanville Intake plume from Wilmot Creek Spill
Figure 7‐46 Wilmot Creek Spatial extent
7‐60
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
7.6.3.3 Graham Creek Spill
The Graham Creek flow was pro‐rated by drainage area with the Wilmot Creek flow data.
7.6.3.4 Newcastle Intake
Figure 7‐47 shows the model grid for the Graham Creek spill and the surface plume for the April 21
event at the Newcastle Intake.
Figure 7‐47 Newcastle Intake plume
Figures 7‐48 & 7‐49 show the time series for Benzene at the Newcastle and Bowmanville Intakes. Peak
levels are above 3.0 mg/L.
Figure 7‐48 Newcastle Intake time series from Graham Creek
7‐61
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 7‐49 Bowmanville Intake time series from Graham Creek
Figure 7‐50 shows the plume from the April 29 event at Bowmanville Intake and Figure 7‐51 shows the
spatial extent.
Figure 7‐50 Bowmanville Intake Plume from Graham Creek
7‐62
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 7‐51 Graham Creek Spatial extent
7.7 Gasoline Pipeline Break on Credit River
The same approach was applied to the Credit River, with impacts at Lakeview, Lorne Park and Clark
Intakes. Figure 7‐52 shows the model grid with the Lakeview surface plume event. Figures 7‐53 to 7‐54
show the times series of Benzene at each intake, and Figures 7‐55 & 7‐56 show the surface plumes at
the other intakes. The Lakeview peak level is predicted to be 0.37 mg/L, Lorne Park peak is 2.4 mg/L and
Clark peak is 0.15 mg/L.
Figure 7‐52 Model grid and Lakeview Intake surface plume
7‐63
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 7‐53 Lakeview time series from Credit River
Figure 7‐54 Lorne Park time series from Credit River
Figure 7‐55 Clark time series from Credit River
7‐64
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 7‐56 Lorne Park surface plume
Figure 7‐57 Clark surface plume
7.8 Gasoline Pipeline Break on Humber and Don Rivers
The 270 m grid for these scenarios is shown in Figure 7‐1 with the intake locations.
The same method ‐ Credit River, Ganaraska etc. was used, with flows in the rivers from recent City of
Toronto studies, generated from watershed models.
The time series of Benzene predicted at each of the intakes are shown in Figures 7‐58 and 7‐59, for the
Humber River and Don River respectively.
7‐65
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 7‐58 Humber River Spill time series – note smaller scale on right axis for Harris and Island Deep intakes
7‐66
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Figure 7‐59 Don River Spill time series – note smaller scale on right axis for Clark, Lakeview and Island Deep
intakes
The peak levels for a spill from the Humber River are:
Clark Intake 0.79 mg/L
Harris Intake 0.101 mg/L
Lakeview Intake 0.30 mg/L
Island Shallow Intake 0.40 mg/L
Island Deep Intake 0.015 mg/L.
The peak levels for a spill from the Don River are:
Clark Intake 0.035 mg/L
Harris Intake 0.316 mg/L
Lakeview Intake 0.023 mg/L
Island Shallow Intake 1.0 mg/L
Island Deep Intake 0.01 mg/L.
7‐67
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
7.9 Gasoline Pipeline Break on Highland Creek, Rouge River and Duffins
Creek
Map is shown in Figure 7‐60.
Figure 7‐60 270 m Grid with intake and tributary locations
Flows in the tributaries were supplied by Environment Canada for the 2008 season.
7‐68
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
7.9.1 Highland Creek Spill
The model predicted time series of Benzene at each intake depth is shown in Figure 7‐61.
The peak values for each intake are:
Harris 0.088 mg/L
Horgan 0.29 mg/L
Ajax 0.01 mg/L
Whitby 0.008 mg/L.
Figure 7‐61 Highland Creek Spill time series
7‐69
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
7.9.2 Rouge River Spill
The time series for the intake Benzene levels are shown in Figure 7‐62.
The peak level for each intake is:
Harris 0.045 mg/L
Horgan 0.27 mg/L
Ajax 0.011 mg/L
Whitby 0.006 mg/L.
Figure 7‐62 Rouge River Spill time series
7‐70
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
7.9.3 Duffins Creek Spill
The time series for the intakes are shown in Figure 7‐63.
The peak level for each intake is:
Harris 0.047 mg/L
Horgan 0.075 mg/L
Ajax 0.061 mg/L
Whitby 0.011 mg/L.
Figure 7‐63 Duffins Creek Spill time series
7‐71
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
7.10 Gasoline Spill 16 mile Creek
The intake locations are shown in Figure 7‐64.
Figure 7‐64 Intakes and 16 Mile Creek Locations
The times series at nearby intakes are shown in Figure 7‐65.
Figure 7‐65 16 Mile Creek time series
7‐72
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
The peak levels at each intake are:
Hamilton 0.007 mg/L
Burlington 0.035 mg/L
Burloak 0.014 mg/L
Oakville 0.12 mg/L
Lorne Park 0.012 mg/L.
7.11 Gasoline Spill at Oshawa Creek
The 270 m grid is shown in Figure 7‐66 with the Creek and intake locations.
Figure 7‐66 Oshawa Creek Grid
7‐73
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
The time series of Benzene at each intake is shown in Figure 7‐67.
Figure 7‐67 Oshawa Creek time series
The peak levels of Benzene at each intake are:
Oshawa Intake 1.4 mg/L
Ajax Intake 0.14 mg/L
Whitby Intake 0.32 mg/L.
Table 7‐4 lists the peak levels for all intakes from the spill locations modelled. Also included in Table 7‐4
is an estimate of the duration of the time the benzene is above 0.005 mg/L. The duration varies from
intake to intake, depending on lake circulation patterns and current reversals. The multiple releases
causes some overlapping of plumes which make it difficult to discern individual events.
7‐74
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Table 7‐4 Peak Level Benzene from Spill location at Intake (Units are mg/L) and Duration of typical event with Benzene above 0.005 mg/L
Discharge: Cobourg Ganaraska Wilmot Graham Bowmanville Oshawa Duffins Rouge Highland Don River Humber Credit 16 Mile
Creek River Creek Creek Creek Creek Creek River Creek River River Creek
Intake:
(48 hrs) (36 hrs)
(37 hrs) 60 (hrs)
Oshawa Beyond Grid Beyond Grid Beyond Grid 1.4 Varies 36 Varies 36 Varies 36 Varies 36 Varies 36 Varies 36 Varies 36
to 72 hrs to 72 hrs to 72 hrs to 72 hrs to 72 hrs to 72 hrs to 72 hrs
(48 hrs)
(48 hrs)
7‐75
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Discharge: Cobourg Ganaraska Wilmot Graham Bowmanville Oshawa Duffins Rouge Highland Don River Humber Credit 16 Mile
Creek River Creek Creek Creek Creek Creek River Creek River River Creek
Intake:
(30 hrs)
7‐76
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
8 References
Boegman and Yerubandi, Process Oriented modelling of Lake Ontario Hydrodynamics, 2006 ISEH
Proceedings.
City of Toronto WWFMMP report – MIKE‐3 modelling report. – full reference required
CTC Source Protection Region (CTC, 2011a) Updated Assessment Report Toronto and Region Protection
Area. Volume 1: Assessment Report Main Body. [http://www.ctcspa.ca/FINAL_TPSPA_UAR_
for_MOE_websized.pdf]
CTC Source Protection Region (CTC, 2011b) Updated Assessment Report Toronto and Region Protection
Area. Volume 2: Appendices. Appendix E6: Threats from Activities in Intake Protection Zone 3 (IPZ‐3) .
[http://www.ctcspa.ca/FINAL_TPSPA_UARAppendices_for_MOE_websized.pdf]
Dewey (2003 b) Waterfront Modeling Report, using DHI model. City of Toronto WWFMMP Report.
Prepared in support of City of TORONTO (2003). WWFMMP Waterfront Report.
Fairlie, I. Tritium Hazard Report: Pollution and Radiation Risk from Canadian Nuclear Facilities, June
2007.
Golders Associates, 2009, Surface Water Environment Assessment of Environmental Effects Technical
Support Document New Nuclear‐ Darlington Environmental Assessment NK054‐REP‐07730‐00012
Rev000. Prepared for Ontario Power Generation.
Hatch Mott MacDonald Group Mid‐Halton Wastewater (sewage) Treatment Plant Environmental Study
Report, Appendix B, April 2010.
Huang, A., Yurabandi, R., Lu, Y. and Zhao, J. Hydrodynamic Modelling of Lake Ontario, An
intercomparison of three hydrodynamic models, J. Geophys. Res, 2010.
Health Canada 2006 ‐ Radiological Characteristics of Drinking Water. Document for Public Comment.
Prepared by the Federal‐Provincial‐Territorial Committee on Drinking Water.
Hurdowar‐Castro D., Tsanis I. Simanovski I., 2007, Application of a Three‐dimensional Wind Driven
Circulation Model to Assess the Locations of New Drinking Water Intakes in Lake Ontario, IAGLR, 33 ‐1,
pp 232‐252.
King et al 1998 ‐ Tritium if the Great Lakes in 1997 ‐ AECL Report RC‐1981.
8‐1
Spill Scenario Modelling for Lake Ontario Intakes
Report for Lake Ontario Collaborative
Ontario Hydro Nuclear, Environmental and Radiological Impact of a large Release of Heavy water from
Pickering Generating Station in August 1992. Report NA44‐REP‐03482.2‐0021‐R00 ‐ December 1994.
Region of Halton (2010) Mid Halton Wastewater (Sewage) Treatment Plant Phase IV and V Expansion.
Class Environmental Assessment – Environmental Study Report Appendix B. Environmental
Investigations (Appendix prepared by Hatch Mont MacDonald Group April 2010)
8‐2