Urban Heat Island - A Local Climate Cause for Urban Flooding?

Urban flooding is often caused by high intensity rainfall, typical
during summer thunderstorms.

The Framework Convention on Climate Change points to climate change impacts on extreme weather events:

Article 8
1. Parties recognize the importance of averting, minimizing and addressing loss and damage associated with the adverse effects of climate change, including extreme weather events and slow onset events, and the role of sustainable development in reducing the risk of loss and damage.

As we move ahead, its important to quantify trends in extreme rainfall events, particularly as these relate to urban flooding. Recently Northern Illinois University (NIU) explored the effects of local urban heat island on storms.  We review this in the context of trends in Toronto-area extreme rainfall, as well as trends across Ontario and Canada.

NIU research suggests that local heat island effects increase the incidence of thunderstorms.  They found that for any given day in the 17-year study period (1997-2013) there was about a 5 percent greater chance that Atlanta would experience a thunderstorm compared to a similar sized rural area. As convective summer thunderstorms are the typical cause of urban flood damages, this could be one factor resulting in increased flood damages over past decades.

Researches studying Atlanta indicated that the possible mechanisms causing more thunderstorms in urban areas included:

• The urban heat-island effect. A landscape of concrete, asphalt and densely packed buildings can enhance heat. Low pressure forms atop the urban area with higher pressure in surrounding rural areas. The scenario might cause low-level atmospheric convergence, which forces air up into thunderstorms.
• Localized areas of upward moving air resulting from increased surface roughness. A cityscape modifies wind direction and speed, which might lead to enhanced convergence and thunderstorm formation.
• Urban pollution. The science isn’t settled on this issue, but some researchers believe enhanced aerosol production caused by urban pollution can enrich thunderstorms. “We discovered that there may be an aerosol effect with initiation as well, because we had greater storm initiation events during weekdays compared to weekend days.”

(So one possible solution to urban flooding is mandatory 3-day weekends all summer.  Sign me up!)

The Toronto area has been expanding since Governor John Graves Simcoe established the town of York in 1793. 

For comparison, Metropolitan Atlanta had a population of about 5.5 million in 2013 while the Greater Toronto Area population in 2011 was about 6 million.

Is it possible that the urban heat island effects from expanding Mississauga, Brampton, and Vaughan area have increased the incidence of storms over Toronto?

Bloor Street gauge shows no increase in severity of rainfall events
since the 1940's in downtown Toronto (Environment Canada).
Station 6158355.

Environment Canada Engineering Climate Datasets show no increasing trends in the severity of storms downtown at the "Bloor Street" gauge as we call it in the business. This "Toronto City" gauge trend data is shown to the right.

Often the characteristics of severity and frequency are used together. It is reasonable to expect that these move in the same direction when it comes to rainfall events.  That is, if there are ever more frequent events, then there is a higher probability that a more severe event can occur.  The Atlanta research looked into the frequency of thunderstorm 'births' but did not explore the resulting severity of the events.

Toronto Island Airport annual maxima for rainfall over
5 minute to 24 hour durations.  Station 6158665.

Given that the severity of rainfall events has decreased at the Toronto Bloor Street gauge, we could assume that the frequency of events has decreased as well.

Below the Bloor Street data is the Toronto Island Airport and the Buttonville Airport rainfall trends. Both of these locations show decreasing trends in the maximum rainfall depth measured over 5 minute to 24 hour periods each year. Is it possible that the relative stability of the land use around the Island and Bloor Street has meant stable thunderstorm patterns?

'Toronto' Buttonville Airport annual maxima for rainfall
over 5 minute to 24 hour durations.  Station 615HMAK.

Toronto International Airport (Pearson) annual maxima for
rainfall over 5 minute to 24 hour durations.  Station 6158731.

At the Pearson Airport gauge there is a mix in the trends in the maximum rainfall depths. The trend is downward for 5 minute durations and 6, 12 and 24 hour durations, but upward for 10 minute to 2 hour durations.  This data set is extended to include the large storm on July 8, 2013.

Is is possible that the development sprawing to the north and west of Pearson Airport contrbutes to an increase in the severity of annual maximum rainfall depths at Pearson for some durations? Lot likely.

Bear in mind the upward trends some some durations are Peason are minor and that even in a statistically static environment (no change in rainfall severity), trends naturally go up or down as more samples are taken from the underlying 'population' over time - it is just like variable results from a coin toss.  You statistically have a 50% chance of getting heads in the long, long run, but you also have a 6% chance of getting 4 heads in a row on your first tosses.

As we move past 2013 and clock smaller sub-record breaking events, the Pearson Airport trends may also follow the decreasing downtown and uptown Buttonville ones.

The one gauge that has been cited as having experienced many more extreme events over time is the North York gauge.  This is actually a composite data record gauge and so it should probably not be given as much weight as Bloor Street or Pearson which have longer records from more consistent locations.

'Toronto North York annual maxima for rainfall over 5 minute to
24 hour durations.  Station 615S001 (composite gauge).

The verdict? Are storms becoming more severe in the Toronto area and could this be caused by urban heat island effects? The answer is storms are not becoming more severe, and we also conclude no more frequent. This means that neither the macro scale effects of global climate change, nor the local scale effects of heat islands, nor the combination of the two, are having a measurable impact on rainfall patterns in Toronto.

What has certainly changed in the Toronto area to explain increased flooding? Urbanization has increased runoff and that land use change doubles or triples peak flows relative to rural conditions.  The Don River watershed urbanization has increased from 15% in the 1950's to nearly 90% today.  Also, overland flow paths have been enclosed or obstructed due to intensification and infill. Perhaps Ontario needs a "Places to Flow Act" to address deficiencies in the overland flood conveyance capacity in developing cities.

Days with rain include days with a fraction of a millimetre
of rainfall accumulation (anything above a trace amount).
Numbers are up but do not affect flood risks.

Several insurance companies have cited annual precipitation, or number of rainfall events per year as a factor in increasing flood damages, but this is really nonsense.  Understandably, urban flooding is a complex issue and annual precipitation would be a convenient explanation for it.  It is short term rainfall intensities as shown the in Environment Canada charts above that define urban flood risks.  "Days with rain" is  risk factor for mushroom growth on your front lawn.  We explore that in another post.

Looking beyond Toronto and its urban heat island context, Environment Canada's assessment of rainfall trends can be found in the report at the following link:

https://www.dropbox.com/s/pxbuyz7sx3h7ln0/Final%20December%202011%20Draft%20Regional%20IDF%20Technical%20Report.pdf?dl=0

This report is "Methodologies to Improve Rainfall Intensity-Duration-Frequency (IDF) Estimates: A Southern Ontario Pilot Study" by Environment Canada Adaptation and Impacts Research Climate Research Division, December 2011.  See Section
6.1 entitled Trends in Precipitation and its Extremes in Southern Ontario, page 77 for the following:

“Significant increases, as well as decreases, were detected at some stations in a number of the extreme precipitation indicators. However, the majority of station trends were determined to be non-significant and no consistent geographical patterns for increases or decreases were observed across Canada. In most cases, the magnitude of the observed changes was also very small. These results are consistent with the daily Canadian extreme precipitation trend analysis of Zhang et al. (2001) and the Canadian component of global and North American trend analyses of daily precipitation extremes by Alexander et al. (2006) and Peterson et al. (2008), respectively.”

As further information we can refer to the work of Heather Auld, M.Sc. (Meteorology) who joined Risk Sciences International (RSI) as their Principal Climate Scientist in 2011 after spending 32 years with Environment Canada. Please see her presentation on extreme rainfall trends here:

http://projects.upei.ca/climate/files/2012/07/IDFtraining-Auld-final.pdf

Ontario decreases and increases in short term rainfall intensities.

Slide 39 includes a map showing decrease in rainfall intensity (15 minute and 12 hour periods). Slide 40 summarizes overall trends and indicates:

"• Return Period Rainfalls DECREASE in many cases
• High profile extreme rainfall events in recent years lead to
 expectations that return period rainfalls will all show INCREASES"

Raw rain intensity and statistics data is available from Environment Canada here under the IDF files link:

http://climate.weather.gc.ca/prods_servs/engineering_e.html

Lessons for Risk Reduction - Make the Flood Connection (Chuck Woolery not included)

A short running network TV program where water resources
bloggers tried to get the audience to think about flood risk
management as many times as day as they think about sex,
which is 34.2 for men and 18.6 for women according to
Psychology Today. Very short running actually.

Time to make some "Flood Connections". What can we learn from the failure of other systems in other disciplines like health care or energy management?

Well. We can learn that complex systems fail. Sometimes very simple systems fail too, depending on the user.

Here are some examples that can help us explore the best approaches to prevent accidents and reduce risks in systems that require human interaction.  There are some lessons that we can apply to flood risk mitigation. At the end of the post we roll the lessons in these unrelated disciplines into
the flood risk connections of interest to CityFloodMap.Com readers.

Health Systems / Therac-25 Radiation Machine

Therac-25 Radiation Machine - too complex for smart folk.

The Therac-25 was a radiation therapy machine produced by Atomic Energy of Canada Limited (AECL) after the Therac-6 and Therac-20 units.

It was involved in at least six "accidents" between 1985 and 1987, in which patients were given massive overdoses of radiation. Because of concurrent programming errors, it sometimes gave its patients radiation doses that were thousands of times greater than normal, resulting in death or serious injury. These accidents highlighted the dangers of software control of safety-critical systems, and they have become a standard case study in health informatics and software engineering.

Lesson 1: Fail-safe hardware can improve safety.  Earlier versions has both hardware and software controls to prevent high energy doses -  the Therac-25 used only software.

Lesson 2: You can't fix what you don't understand. After the first fatal doses, operators did not fix or even understand the frequent recurring problems. In 1986, now-deceased patient Ray Cox removed himself from the machine after the third painful dose, while the technician continued to apply unsafe doses while ignoring software error messages.

Energy Systems / Compact Fluorescent Light Bulb

Compact fluorescent light bulb - too complex for regular folk.

A compact fluorescent lamp (CFL) is a fluorescent lamp designed to replace an incandescent lamp; some types fit into light fixtures formerly used for incandescent lamps. The lamps use a tube which is curved or folded to fit into the space of an incandescent bulb, and a compact electronic ballast in the base of the lamp.

The general population does not appreciate things like mercury vapour, or safe disposal so the CFL failed too. While rather simple, it demonstrated that the general public does not do well with science.

Lesson 3: Keep it simple, fool-proof and fail-safe. My mom would not use our first microware because she could feel the "rays coming out of it" and was convinced it was unsafe.  She tried to prove this to us kids by holding her hand near the back - we told her that was the fan blowing air out the back exhaust.

Home Safety System / Carbon Monoxide Detector

Even these can be a problem for users.  Recently a Toronto man almost killed his family because he unplugged his CO detector - why? - because it was going off.  Beep Beep Beep.  Meaning his family was being poisoned by carbon monoxide gas.  Thank goodness they found out and are now safe:

http://www.newstalk1010.com/news/2015/12/01/homeowner-charged-after-co-detector-disconnected

But this shows that even the most simple systems (Beep = Danger) can overwhelm the general public.

Lesson 4: See Lesson 3

What are the  "Flood Connections" to the systems described above?

Lesson 1 Flood Connection: Fail-safe hardware can improve safety.

Physical hardware that required no user expertise to ensure safe conditions made pre-Therac-25 devices safer.  In the context of flood risk reduction, this encourages us to rely on "sure things" like fail-safe physical flow conveyance systems, and passive physical flood proofing measures as opposed to actively-managed ones, or mechanical or electrical systems.

Managing risk with fail safe flood proofing retrofits - New York City.

Risk reduction requires prevention, mitigation, and emergency preparedness and response.  Fail-safe physical measures related to the way we build cities can ensure that physical overland flow paths are mapped, managed and maintained by municipalities. Mapping historical flood locations and overland flows show that compromised overland flow systems aggravate flood risk. 

Physical measures related to the way we build homes can ensure that valuable finishes and belonging stay above flood prone levels in sewers and overland - New York has just put out a great document COASTAL CLIMATE RESILIENCY Retrofitting Buildings for Flood Risk available here. 

Sewage pumping stations used to have overflows to adjacent watercourses in case of failure of the pumps or power supply.  To better manage environmental conditions, these fail-safe "hardware" overflow features have sometimes been removed.

Cities should map and manage urban flood risks so that fail-safe, physical controls can be preserved or retrofitted in the landscape.  Calgary had their flood risks mapped but ignored them, building right in the floodway with no physical separation from the risk - that was a formula for disaster. Planning agencies and ministries should take ownership of risk management and promote fail-safe physical controls, even revisiting the benefit of reliable, physical overflow features in pump stations for the most extreme events.

Intact Insurance repeats IBC and ICLR's "Telling the
Weather Story" theoretical statement on bell-curve
probability shifts as 'fact'.

Lesson 2 Flood Connection: You can't fix what you don't understand.

Maybe people think about the weather as much as they think about sex? Is that why it is so easy to mistakenly fixate on rainfall as the explanation of increased flood damages? Is this just Daniel Kahneman's lazy-brain System 1 thinking at work?

Intact Insurance has a web page www.InsuranceIsEvolving.com that states there is a "rising frequency and severity of extreme weather events" and they repeat the discredited theoretical Telling the Weather Story statement as fact. 

Intact Insurance ties this rising frequency to climate change and release of greenhouse gases. While there are many reasons to be concerned about climate and global warming, if we mistakenly point to increasing severe weather (Environment Canada's Engineering Climate Datasets show no increase) as the cause of flooding, we will miss the real opportunities to find effective solutions to flood risk reduction, and will ignore current physical problems and ineffective policies.

New York City - Example Streetscape with Flood-Proofed Bungalows

There is a "fetish" in the water resources community on updating rainfall intensity-duration-frequency, IDF, curves. Countless person-years of effort and hundreds of thousands of dollars have been put toward updating, and massaging these numbers when the historical trends are really nil for this design parameter. If the same effort had been put into urban flood risk mapping as IDF review, we would have maps showing urban risk areas and could apply urban risk reduction policies tomorrow.  If the same effort had been put into updating runoff-coefficients for urban hydrology we would have more conservative and accurate risks identified.  If the same effort had been put into the review of low-intensity design hyetographs for watershed that are misapplied to small flashy urban systems, we would have more conservative flood risks mapped that reflect real, damage-causing storm risks.

Over time, the provincial ministries, their agencies and the insurance industry may straighten themselves out and refocus on true causes of flooding.  In the meantime, many will continue to pucker on the pecunious teat of public and private funding to predict weather patterns.  This despite that fact that due to Chaos, Lies and Butterflies we just can't.

Lesson 3 and 4 Flood Connection: Keep it simple, fool-proof and fail-safe. 

Complex systems will fail.  Systems requiring coordination between agencies and that rely on human judgement will have a tendency to fail.

Sump pump installation.

Recently a robust system of sump pumps has been proposed as a way of limiting basement flood risks and reducing losses. A resilient system includes one or two back-up pumps, and a back-up power supply.  The system must also be regularly inspected and maintained.  The approach is described in the blog "Focus on reducing losses associated with sump pumps":

http://www.insblogs.com/natural-disasters/focus-reducing-losses-associated-sump-pumps/5952

Not simple, not fool proof, sort of fail safe. An in-law of mine bought a small house in Meaford that had a back-up diesel generator.  Cool! Nobody knew what it was for until the spring. It was supposed to help keep groundwater water out of the lower level in a power outage.  Did it work? No. They flooded a few times and then sold the place. That is just one instance but it shows that fail-safe systems are better - a slab on grade and a two storey house instead of a finished basement would have been a smarter, lower-risk build.  Sometimes redundant systems for power supply fail too.

New York City - Floodproofing Illustration.

The Lesson 1 New York City document provides a great perspective on the fail safe approach to building.  That approach could be adapted from storm surges to urban flooding risks.

Make the Flood Connection. We cannot expect the general public who can't manage CFL bulb disposal, who can't understand carbon monoxide warning beeps, who can't tell fan exhaust from escaping microwave waves, or even informed operators who override radiation machine error codes without a second thought, who ignore Environment Canada data, or who send GO Trains out into flooded floodplains to effectively manage complex systems. Flood risk reduction measures should be based on a clear understanding of causes (risk factors) and should be physically fail-safe, minimizing reliance on electrical and mechanical systems, or user intervention.

****

Explore cognitive biases in our thinking that pose barriers to effective flood risk management:

http://www.cityfloodmap.com/2015/11/thinking-fast-and-slow-about-extreme.html

Infographics Replace Rational Thought on Urban Flood Causes and Insurance Risks

I had a favourite Matt Groening cartoon on my grad school cubicle cork board for years.  It was a Life is Hell black and white strip from Now Magazine.  In it Binky is lying in bed awake all night and Mrs. Binky asks him "What is keeping you up?" and Binky replies "Infotainment".

Just as Binky worried about Infotainment then, I lie awake worried about Infographics now. Infographics per Wikipedia are:

"...graphic visual representations of information, data or knowledge intended to present information quickly and clearly. They can improve cognition by utilizing graphics to enhance the human visual system’s ability to see patterns and trends."

RSA links flood damages to more days with rain since the 1950's.
Ummm ... how about urbanization? The Don River watershed was
only 15% urbanized in 1950 and was 80% urbanized by 1994. In 2021
it is predicted to be 91% urbanized. Ahem .. any chance this is
contributing to flood damages? Changing development patterns instead
of changing weather patterns?

Given how busy we are, infographics are wonderful - they can get us up to speed on complex issues in no time flat apparently, distilling a whole bunch of science, statistics and mumbo jumbo into a quick, easily digested glance-friendly format.

Here is the Climate Smart Infographic from RSA Insurance related to their new Waterproof Coverage insurance endorsement.  Hmmm "Protect Property in Extreme Weather", got it. "Weather Changing", sure. "20 more days of rain than in the 1950's", logical explanation for flooding .. uh....wait .. WAIT A MINUTE!  I was almost "Infografected".

What is an Infographection? Its like an infection, easily transitted through exposure to infographics. It affects the weak minded mostly, turning them into a zombie-like herd of followers.  Be afraid be very afraid, because for some reason infographection is becoming the preferred method of communication for insurance companies and provincial agencies.

What is wrong with RSA's infographic? Well, "days with rain" has no bearing on extreme weather or causes of the flooding RSA is selling endorsements for.  If days with rain was relevant, then Vancouver would flood all the time and nobody would get insurance there.  What is forgotten in the RSA "days with rain" statistic is the fact that a tiny amount of rainfall can trigger the counting of a rain day.  How little rain?  Fractions of a millimetre of rain - hardly the events that cause urban flooding.

Inoculate yourself against infographection - this paper explores some the of details we can consider when thinking about number of days with rain by first considering the tiny 'trace' amounts that are recorded:

https://www.researchgate.net/publication/242397703_Adjustments_for_trace_measurements_in_Canada

As noted in the paper, operational changes have increased recording of rainfall events over time (not necessarily a fundamental increase in rainfall):

"The measurement of trace precipitation in Canada went through several phases over time. Not just the definition of trace precipitation (or the lack of definition) has been modified, but the measuring unit was also changed from imperial to metric system. All data in the Climate National Archive was converted from inches to millimetres with the introduction of metric system, which also caused further inconsistency in the data. Station relocations were often accompanied by new set of instruments and a new observer. The training given to the observers often resulted in a jump in the number of daily trace measurement."

Also, the switch to metric since the 1950's meant a smaller measurement unit, and meant its is now easier to record trace amounts and record a 'greater than trace' amount, i.e., a lawn mushroom-forming but not-flood-inducing "day with rain":

"The homogeneity of trace observation record was also seriously affected by switching from Imperial to the Metric system around 1977 - 1978. The units were given in inches before; the minimum measurable level has been decreased by almost 0.1 mm in the metric system. Table 1 gives a summary of the amounts and units used in both systems for rain and snow separately."

Threshold for measuring and recording rain events decreased in the 1970's with conversion to metric system.

So the RSA infographic and simple "days with rain" statistic really fails to explain any cause of flooding or trend in extreme rain characteristics.  As reader of this blog know, Environment Canada Engineering Climate Datasets show no increase in extreme rain characteristics that cause urban flooding:

http://www.cityfloodmap.com/2015/07/storm-intensity-not-increasing-old.html

The unfortunate situation for RSA and flood risks in their portfolio is that by Blaming it on the Rain like Milli Vanilli would, RSA ignores real factors that have caused flows in increase (with no increase in rain intensity), and that have caused infrastructure flow capacity to decrease. Here are some ideas on causes of flooding and flood damage increases:

http://www.cityfloodmap.com/2015/11/why-flood-damages-are-increasing-in.html

It seems RSA has been "infographected" by their own infographics.

*****************

If you got this far, you might also like this article in Psychology Today "Anti-intellectualism Is Killing America, Social dysfunction can be traced to the abandonment of reason".  It highlights "What Americans rarely acknowledge is that many of their social problems are rooted in the rejection of critical thinking or, conversely, the glorification of the emotional and irrational.".  Is this just Daniel Kahneman's lazy-brain System 1 thinking at work?