Toronto GO Train Flood Avoidable July 8, 2013 - Worse May 29, 2013 Flood Ignored

Could the stranded GO Train in Toronto's Don River valley have been avoided on July 8, 2013? Could use of known flood risk mapping and real-time flood level monitors avoided the near catastrophe? Yes. And yes and yes. In fact, on May 29, 2013 flooding was deeper but the peak at 5 a.m. missed the trains!

It is likely that if there had been fatalities among the 1400 GO Train passengers, Metrolinx employees or Toronto's first responders, the Ontario government would have called an inquiry (see clipping at right). The Ministry of Labour, who recognizes the occupational health and safety hazards of flooding,  would likely have found that Metrolinx failed to identify known workplace hazards or identify safe work practices for its employees.

And that provincial inquiry would have revealed that:

GO Train Stranded Flood Toronto
Mock Go Train Drowning Inquiry newspaper clipping.  This near miss could
have been fatal to passengers, Metrolinx workers, or first responders.
i) flood risk of the rail line was known to be frequent from available flood hazard maps and reports (best and most frequently updated hydrologic modelling of flows and hydraulic modelling of flood levels in Canada actually),

ii) July 8, 2013 observed flow rates and flood levels were not rare or unexpected from a flood risk management perspective - i.e., could have been anticipated - in fact, on May 29, 2013, just 40 days and forty nights before the incident, flood levels peaked 20 cm higher,

iii) monitoring real-time flood levels were rising rapidly upstream of and beside the flood site could have guided train dispatchers to not send the 5:30 train (GO Transit 835) into the flooded Don River valley,

iv) train frequency increase (more than rain frequency) contributed to the near tragedy, and increased risk exposure as the number of Richmond Hill trains on this line doubled since 1996,

v) any tragedy that occurred was avoidable.

Inquiries get to the bottom of risk issues following deaths. For example, the Walkerton Commission of Inquiry has resulted in a comprehensive risk management strategy for managing drinking water risks in Ontario. But this was only after Walkerton’s drinking water system was contaminated with deadly bacteria, seven people died, and more than 2,300 became ill. The goal of the Inquiry was to answer questions:
GO train Flood
GO Train Flood July 8, 2013 - Toronto Don River Valley
  • What actually happened?
  • What were the causes?
  • Who was responsible?
  • How could this have been prevented?
  • How do we make sure this never happens again?
Let's look at the same questions in the context of the Metrolinx Richmond Hill train flood incident in 2013.

What Actually Happened?

The GTA Don River Watershed (correction to original post Jan.2, 2016) received up to 126 mm of rainfall on July 8, 2013.  That is a lot of rain and runoff was high, resulting in flood warnings and high flood levels at the Don River flood monitoring stations.

River levels at the upstream East Don River flood level monitoring site (TRCA's East Don at York Mills HY022 gauge) were already rising when the previous by "5:00 pm" train level Union Station. Upstream flood levels were even higher when the stranded "5:30" train, Go Transit 835, left Union Station - this warning sign was not used by Metrolinx to assess the risk to their route further downstream.  Graphs are generated from TRCA's online archive.

Go Train Flood Toronto
Recorded East Don River at York Mills Flood Levels July 8, 2013 (per TRCA)
5:30 Richmond Hill GO Train departed Union Station after upstream flood levels has already risen rapidly.

Why did the 5:30 train leave when the upstream flood levels pointed to potential risks?
What Were the Causes?
Metrolinx GO Train frequency has doubled from 3 to 6 return
trains per day from Union Station to Richmond Hill since 1996.

Cause 1). Train frequency has doubled over the past 19 years, increasing exposure to flood events.

The Richmond Hill GO Train service started in 1978 and frequency has increased.  In 1996, service was cut to 3 trains per day while currently there are 6 trains during the afternoon / evening period (a 100% increase in risk exposure).

GO Train flood Toronto
Train operation was halted or detoured during previous Don River floods
as documented in the 1981 Flood Inquiry Report.
Cause 2). The track floods frequently due to summer storm events (or spring melt).

The Keating Channel Flood Inquiry Report for Premier Davis in 1981 documented frequent flooding decades earlier, even when urbanization of the Don River Watershed was not as intense as it is now.

Cause 3). Water levels rose to expected design flood levels on July 8, 2013, but measurements were ignored / not considered.

The key concern here is that design flood levels experienced on July 8, 2013 were not rare (see footnote 1 on correlation of storm vs. flood severity).  The 2-year return period flood level is 79.45 metres near the Evergreen Brick Works, across from the flooded GO Train site. In common terms, this flood level can be expected every 2 years, and so over a long period, it has a very high risk of occurring (about 17 times between the start of service in 1978 and the flood of 2013).

Higher safety-risk flood levels of one metre higher are not rare from a flood risk management perspective.  Even a higher 10-year flood level (80.41 metres) has a 88% risk of happening at least once over 20 years. This is illustrated on the graph below which compares expected design flood levels, risk of reaching these levels over time, and actual real-time flood levels recorded July 8, 2013. These design flood levels are from a Evergreen Brick Works presentation and, a site located beside the stranded train location - the Brick Works has a comprehensive flood risk management strategy.
GO Train Flood Toronto
Recorded Don River Flood Levels at Todmorden July 8, 2013 (per TRCA).
Design Flood Levels per MMM (2008) per Evergreen’s Urban Watershed Forum 2015.
Flood risks were known. Upstream flood levels had risen. Local flood levels were rising.

So, the causes include more frequent trains travelling through a highly flood prone part of the Don River valley, and the operator Metrolinx having ignored measured flood levels.

Surprisingly - even higher flood levels were recorded on May 29, 2013 but trains missed this peak due to timing. The May 28, 2013 Union Station departures occurred before the rise in flood levels and the May 29, 2013 Richmond Hill trains arrived near Todmorden after the peak flood levels has subsided (GO Transit may have modified schedule). The maximum track flooding on May 29, 2013 appears to be 20 cm deeper than on July 8, 2013 - rainfall in East York was over 40% greater.
Worst GO Train Flood Toronto
May 29, 2013 flooding is deeper than July 8, 2013 flooding based on TRCA Todmorden Flood Monitoring Gauge.
The May 29, 2013 flood would have flooded tracks by almost 2 metres.
GO Trains appear to have missed peak flood depths by departing before and after peak flood level occurred.

Worst Metrolinx Flood
May 29, 2013 flood flow rates at the incident site higher than July 8, 2013 flood flows by 10 cubic metres per second.
TRCA from Todmorden Flood Monitoring Gauge
GO Train Flood
May 29, 2013 Flood Levels Worse Than July 8, 2013
But even May 29, 2013 flood levels were not unprecedented according to Toronto Police.  That flood closed the Don Valley Parkway, left it covered in mud and according to reports by City News:

“I haven’t seen flooding on the Don Valley Parkway like this,” Toronto police Staff. Sgt. Brian Bowman told Breakfast Television. “One of my officers had, back in 1986. He saw it reach the top of the [concrete] jersey barriers, so it’s not unprecedented.”

Not only were flood levels July 8, 2013 not rare, flows were below common design flow rates (see footnote).

Who Is Responsible?

In Ontario, everyone is responsible for safety in the workplace. Employers and employees have a role. Reports suggest that Metrolinx, the employer, did not receive TRCA flood warnings and did not make use of monitored flood levels in the watershed to inform its operations. Perhaps it did not understand the workplace hazards on the Richmond Hill line.

GO Train Flood Safety
Ontario Ministry of Labour identifies drowning and other risks associated
with flooding.
After the incident, Metrolinx dismissed the value of TRCA flood warnings saying to the National Post “Because the TRCA flood warning was fairly general, it only has limited value,” said Greg Percy, the Metrolinx vice president of infrastructure. It is understandable that weather warnings do not always translate into actual flood conditions.  But Metrolinx did not acknowledge the use or value of measured flood levels, including those upstream on the East Don tributary and those adjacent to the flood-prone tracks at the nearby Todmorden flood monitoring gauge.

Evergreen Brick Works beside the stranded GO Train site has a flood
emergency plan and its building are flood proofed. Brick Works reported
higher flooding in May 2013 than July 2013 on their site.
It would appear Metrolinx was responsible for not assessing workplace hazards for its employees and customers.  Also, it did not develop safe work practices and procedure that would allow it to identify flood risks and take appropriate action (cancel service, substitute buses instead of trains, divert to Barrie line, etc.). Employers like Evergreen Brickworks have had flood emergency planning in place for many years as the location is known to be highly flood prone.

TheStar.Com reported two days after the July 8, 2013 incident that stranded customers were offered $100 as compensation and Mary Proc, GO vice-president for customer service said "It is an exceptional gesture for an unprecedented circumstance." She did not acknowledge that while the inconvenience to passengers was unprecedented, the flood levels were not. She added “That was a night of firsts for us: The first time we had a month of rain in one night; the first time that any customer had to wait seven hours to be moved off a train, and the first time we deployed boats to take our customers off a train.” She did not acknowledge that more rain, more runoff volume, higher flow rates and higher flood depths occurred on May 29, 2013, or that a month of rain in one night is not an uncommon design event for prudent flood risk management. GO Transit (@GOTransit) did Tweet about Richmond Hill line disruptions at 3 a.m. on May 29, 2013 suggesting perhaps they are aware of flood conditions on the line before scheduled morning trains:

How Could This Have Been Prevented?

The East Don flood level graph above showed that the Don River had risen quickly by 5:30 p.m.. The graph below shows that flood levels where the train was stranded had risen nearly half way between the normal water level and the track level by 5:30 p.m..  By the time the stranded GO Train approached the Brickworks at 5:45 p.m., the real time level monitoring would have shown the Don River flood levels approaching the track - see graph below.  But these levels were not used by Metrolinx.

Go Train Flood
Track level is below the frequent flood zone (2-year flood). The 5:30 p.m. Richmond Hill GO Train departed when upstream and local flood levels were rising. The GO Train was stranded at 5:45 p.m. when flood levels were at frequent flood levels - not even up to the 2-year flood level of 79.45 m.

This accident could have been prevented by having operational procedures included checking monitored flood levels and modifying service to avoid sending trains onto flooded tracks.

How Do We Make Sure This Never Happens Again?

GO Train Flood
Environmental Commissioner mistakenly links GO Train flooding to
climate change, ignoring known risk factors, watershed conditions,
and operational gaps contributing to flood risk and damages.
According to the Environmental Commissioner of Ontario, the commuter train flooding is a symptom of climate change (Feeling the Heat: Greenhouse Gas Progress Report 2015 Media Release).

Obviously that is not an informed statement. By linking the GO Train incident to climate change and suggesting that risk mitigation efforts relate to emissions controls, the fundamental causes, responsibilities, and potential solutions to the near tragedy have all been grossly ignored.

To ensure this never happens again Metrolinx has had to suffer through 'near miss' in occupational safety terminology. Putting on a positive spin at their September 10, 2013 customer service update, Metrolinx suggested they received as many commendations as complaints (perhaps the commendations were directed to first responders?).  Metrolink also highlighted the 126 mm of rainfall in the "Toronto area", failing to note that record was at Pearson Airport in the Etobicoke Creek watershed, not the Don River watershed where GO Transit 835 was stranded. Also Metrolinx would look at several areas of concern including:

  • "identifying high risk areas,
  • improving customer messaging systems, and
  • upgrading the early warning storm warnings."

GO Train Flood Toronto Metrolinx
Metrolinx Customer Service Report September 2013 refers to 'massive' and 'record' storm as causes to the serious incident.

Metrolinx could identify high risk areas by reading the Flood Inquiry report from 1981 (a real Inquiry for Premier Davis ... not the mock one at the start of this post).  It identifies railway line damage and flooding during ice-free conditions (i.e., no ice blockage of bridges) during the "Great Flood" on September 13, 1878, the spring of 1914, and during two storms in 1980 (March and April). Limited flooding was also reported on May 11, 1981 including the area of the Bayview Extension and the Toronto Brick Yards (the stranded GO Train location near the Evergreen Brick Works).  The report indicates that train operation has halted, or trains were detoured during floods, including December 25, 1979, January 11, 1980, March 21, 1980, April 14, 1980, February 11, 1981 and May 11, 1981.

go train flood
Example historical flooding in vicinity of stranded GO Train during July 8, 2013 flood (1981 Inquiry report Table 1).
According to the 1981 Flood Inquiry report, during the March 21, 1980 flood "Part of the CNR track flooded to the north and east of a point south of Bloor Street."  The 20 hour low intensity spring storm produced high runoff as ground in the watershed was partly saturated and frozen.  This area north of Bloor Street corresponds to the location of the July 8, 2013 Go Train flood.  During the April 14, 1980 flood, after a 5-6 hour period of rain, CNR tracks at the Bloor Street ramp were flooded. The GO Train was stranded just east of where the Bloor Street ramp connects to the Bayview Ave.

GO Train Flood Passenger Swimming
GO Train / GO Transit 835 passenger swimming from
stranded train in Don River valley.
The following summer on June 26, 2014 InsideToronto.Com reported "Metrolinx is crediting recent improvements to its flood monitoring protocol in lessening the impacts to morning GO Transit commuters following a severe rainstorm." Metrolix reported that "Three GO trains en route to Union Station were also diverted from the Richmond Hill to the Barrie rail corridor to avoid the flooded tracks.".  This was after a moderate storm described as an "event which dropped some 50 millimetres of rain in the Don Valley alone, according to Environment Canada.". A 50 millimetre rainfall event is less severe than a common 5 year / 12 hour event, or  10 year / 6 hour event, meaning the track did, does and will flood frequently.

GO Train Flood
It is unlikely that Metrolinx would admit that the July 8, 2013 flood was a common design condition for flood risk management, and that it was perhaps just lucky in its complacency / ignorance - May 29, 2013 flooding less than 6 weeks earlier was worse. Fortunately complacency is no longer the case and it is rerouting trains after even moderate design storms such as on July 26, 2014, June 22/23, 2015, October 28, 2015.......

A post below to the blog has a comment from an apparent operator indicating that the flooding of the tracks was indeed frequent and that the operating procedures have now changed.

Anonymous said...

Back when I started as long as we could still see the rails we used to drive through that like it was nothing. Now the second the water starts touching the rails an automatic system declares an 'emergency' and the whole line is shut down. Not saying that's the right or wrong way, but its interesting how times have changed.

1 - Correlations of Storm Severity vs. Flood Level Severity.  How could a record amount of rain have been recorded on July 8, 2013 but not a rare design flood level in the Don River flood plain?  The answer is that record amounts of rain did not occur everywhere across the Don River watershed - conservatism in flood risk management and planning assumes extreme rainfall occurs over the entire watershed. The record rain occurred over one small part of the watershed, such that the resulting flow rates were not extreme, having less than 10 year return periods. So while the record rainfall had a 100 year return period somewhere beyond the Don River watershed (Pearson Airport in the Etobicoke Creek watershed), on average, overall, July 8, 2013 was less severe over the large Don River watershed. The video at the very bottom of the post shows the July 8, 2013 storm pattern from historical radar over the approximate Don River watershed area.

GO Train Worst Flood
Worse Flooding May 29, 2013
The May 29, 2013 flood resulted in a higher flood level at the Todmorden monitoring station.  As reported by the Globe and Mail, less than half the July 8, 2013 rainfall fell on Toronto that day "In all, Toronto and the area north received up to 60 millimetres of rain before the downpour eased at around 6 a.m. No injuries or serious traffic accidents were reported." Our correction to that report is that at the Toronto East York Dustan climate station (ID 6158751) a higher 73.6 millimetres of rainfall was recorded (late on May 28, 2013 before the flood levels and flows crested the next day).

Half the rainfall and worse flooding? Yes. Welcome to the world of hydrology where the anecedent soil moisture conditions, and the temporal and spatial patterns of storms over watersheds affect the cumulative runoff volume which can influence the peak flow and peak water levels as much as any local 'spot' measurement of rainfall.

Toronto North York July 8, 2013
Rainfall Just Half Mississauga
Record Rainfall
On July 8, 2013, the rainfall across the Don River watershed was variable with much lower totals in the mid portion of the watershed. The Environment Canada historical climate archives show that the Toronto North York climate station (ID 615S001) measured only 65.8 mm of rain on July 8, 2013, just over half the Pearson record; the Toronto East York Dustan climate station (ID 6158751) measured only 51.4 mm of rain, less than half the Pearson total. Summary tables are shown below and at right (click to enlarge).

Despite this fact of lower, non-record Toronto-proper rainfall, the CBC reported the July 8, 2013 GO Train incident as a top weather story of 2013 saying:

"When you look at the amounts of rain that fell ... it was like Toronto was the bull's eye," said Phillips, who described it as "a direct hit with a drenching rain storm."

Toronto East York July 8, 2013
Rainfall Just Half Mississauga
Record Rainfall
The CBC News has mistaking described record rainfall in Mississauga (Pearson climate station location) and in the Etobicoke Creek watershed with moderate rainfall in Toronto in the Don River watershed (where the GO Train was stranded).  A watershed map is provided below.

The May 28-29, 2013 storm that caused higher flows and higher flooding on the GO Train tracks dropped 43% more rain than July 8, 2013 in the mid portion of the watershed according to the Toronto East York Dustan climate station records.

The CBC News story also fails to recognize the frequent flood-prone nature of the rail tracks where the GO Train was stranded. Such reporting is an example of "anchoring or focalism", the cognitive bias per Nobel Memorial Prize in Economics winner Daniel Kahneman, in which people rely too heavily on the first piece of information present (Pearson/Mississauga rainfall record) when it is in fact irrelevant to Toronto/Don Watershed flooding.

Here is a summary of mid watershed rainfall (East York), and flows and water levels (Todmorden monitoring site near the stranded GO Train site) for the May 2013 and July 2013 floods:

Date                      East York Rainfall      Peak Flow     Peak Flood Level
May 28-29, 2013           73.4 mm               190+ cms              80.9 m
July 8-9, 2013                51.4 mm               180+ cms              80.7 m

From a design storm perspective the recorded flow rates on May 29 and July 8 were low to moderate. The chart below from TRCA's Don River hydrology report shows a range of design flows from drought conditions, to average yearly peaks, up to rare storms considered for flood hazard management and design in Ontario. At less than 200 cms, both the 2013 peak flood flows were less extreme than the one-in-five-year design flow, a frequent design flow that has a 20% (1/5) chance each year of being exceeded.