INTRODUCTION
Single
station Intensity-Duration-Frequency (IDF) statistics are commonly used to
characterize point observations of extreme rainfall and derive synthetic design
storms. Such point data, applied in
flood vulnerability assessments of urban infrastructure and natural drainage
systems, is distributed across small and large sewersheds and watersheds,
sometimes with an areal reduction in peak rainfall. Areal Reduction Factors
(ARFs) are available based on world and North American data. Over eighty years
of storm event data has been compiled by Environment and Climate Change Canada
(ECCC) in “An Index To Storm Rainfall In Canada”, documenting how peak rainfall
diminishes over larger areas.
This
paper reviews historical extreme storm characteristics in Ontario including
peak rainfall volume and areal extent of moderate rainfall, defined by ECCC as
a 50 mm volume. This 50 mm threshold is
deemed suitable to assess areas that may experience nuisance flooding and may
contribute to significant downstream flooding. Available Ontario ECCC data
spans from 1919 to 1984 and includes 118 extreme
storms. Ontario technical guidance for
ARFs has been compared to long term Ontario data to assess the applicability of
literature values in local system assessments.
The ARF is a common factor in large natural drainage system assessments
and can be more widely considered in urban systems where areal-dependent
reductions in RDII are apparent in large municipal wastewater systems.
Trends in storm areal extent are compared to trends in urban development
in the Greater Toronto Area, illustrating the “Expanding Bull’s Eye Effect”
coined by Stradler and Ashley (2015).
The effect has been used to help explain increasing extreme weather
damages over time whereby storms are the ‘arrow’ and urban areas are the
‘target / bull’s eye’. The effect shows
that damages may increase under stationary hazards (i.e., storm sizes remain
the same) when the size of exposed assets increase (i.e., urban areas
expand). Trends
in rainfall intensities across southern Ontario are presented to explore any
correlation between intensities of various durations and return periods. Any correlation may be used to infer project
changes in extreme (e.g., 100-year) short-duration intensities related to flood
risks and changes in frequent (e.g., 2-year) long duration intensities such as
those predicted in climate models as “heavy precipitation”. Frequent, annual intensity trends assessed
across Canadian regions by ECCC (Cannon et. al 2024) and correlated with
yearly temperatures are reviewed.
ECCC 2024 annual intensity review assessed trends across large regions
such as the Great Lakes/St. Lawrence River Basin. This paper compares the size of this research
region for trend analysis, the size of storms in ECCC’s historical storm index,
and the size of typical infrastructure drainage catchments and sewershed. The merits and limitations of assessing
rainfall trends for large areas and the applicability to smaller catchment
design storms are discussed. In essence,
regionalization of rainfall station trends can be viewed as an “Expanding Arrow
Effect” where, opposite to the Bull’s Eye Effect the arrow (i.e., the size of
the storm data characterizing the hazard) expands while the target system it
affects remains the same.
Lastly, this paper explores trends in IDF curves across a small region,
the City of Markham. Small return period
statistics are compared for 18 gauges with periods of record of up to 28 years. The comparison illustrates how
regionalization, or aggregation, of single station data can result in
observation of higher localized extremes.
The paper compares the localized extremes from Markham gauges with the long-term
IDF statistics and 95% confidence limits for the Buttonville Airport climate
station.
100 YEARS OF STORM SIZE VS RECENT STORMS
Assessing
extreme rainfall impacts on drainage and infrastructure systems to guide risk
reduction strategies requires characterization of rainfall events. Intensity Duration Frequency (IDF) statistics
represent one key characteristic that is updated regularly with new rainfall
data. While such point data on peak
rainfall can characterize conditions over a small local area, the distribution
of rainfall or a larger area is needed to assess impacts over watersheds and
urban drainage catchments. As summarized
in the authors’ review of 2024 storms in their 2025 WEAO Conference paper
(Karney et al.) real storms, peak intensities are observed to decrease
over larger areas. A review of
historical storm sizes, and the areal extent of high and moderate rainfall can
guide hydrologic modelling such that distributed intensities over large
catchments are not realistically represented.
The authors’ related 2026 Conference paper on rainfall derived inflow
and infiltration in sanitary sewer systems presents extensive data
demonstrating the significant decrease in rainfall effects on collection
systems as the area of the system increases.
Clearly areal reduction of storm impacts is an important design consideration.
The
Atmospheric Environment Branch of Environment Canada published An Index to
Storm Rainfall in Canada (“AES Storm Index”) in 1961 following Hurricane
Hazel. The 1988 edition involved the
review of almost 1000 storms from 1900 onward, identifying 118 significant
Ontario storms between 1919 to 1984.
Characterization included storm centre location, maximum rainfall depth,
duration, and area of 50 mm rainfall (previously 2 inches). Figure 1 shows AES trends in maximum
rainfall depth with a sample of depths added from southern Ontario storms over
the next 40 years (Harrow 1989, Peterborough 2004, Toronto 2005 and 2013,
Burlington 2014, and Toronto July and August 2024 shown as red triangles) and
in the preceding 76 years (Toronto 1843, 1878, 1897 and 1905 shown as black
diamonds; CBC 2024). It is noted that
the proliferation of rainfall gauges over recent decades as outlined in the
authors’ 2025 paper can help identify peaks that were previously not monitored
in less-dense rain gauge networks.
FIGURE 1 – SIGNIFICANT ONTARIO RAINFALL TOTAL
DEPTH 1843-2024
Many
readers will be familiar with some labelled significant events, especially
those recent events that had devastating impacts on highly populated
communities. The 1966 Sandusky event may
be unfamiliar as it was centred in Ohio.
With 118 events from 1919 to 1984 in the AES series there have been 1.8
events per year on average demonstrates that extreme events, most over 4 inches
(100 mm) are not rare.
On event frequency, prior to 1951 the average duration between
significant events was 282 days while from 1951 onward the duration was slightly
longer at 314 days. This suggests that
significant events are not occurring more frequently, despite the expected increase
in rainfall gauges and reporting channels over the past 100 plus years.
Figure 2 illustrates the size of significant storms in the AES index including
some recent storms.
FIGURE 2 – AREA OF RAINFALL OVER 2 INCHES
(50 mm) 1919-2024
Based on AES data, the median 2-inch (50
mm) rain depth area was 23,900 square miles before 1951 and 14,500 square miles
after (i.e., 61,900 to 37,500 square kilometres). This suggests that the size of significant storms
is not increasing. Five of the largest
area storms are labelled and were in central and northern Ontario. Given their remote locations, these events
are not associated with significant damages and are therefore never referenced
in the context of significant Ontario storms.
To put the storm areas above into context, a large municipal wastewater
collection system such as Toronto’s Ashbridges Bay wastewater treatment plant
has a service area of less than 100 square miles (250 square kilometres). A large GTA watershed like the Rouge River
watershed has an area of 130 square miles (336 square kilometres). Figure 2 illustrates that significant storms
have covered vastly larger areas over the past 100 plus years – from 1951
onward the median storm size was two orders of magnitude greater than the Rouge
River watershed area.
AREAL REDUCTION FACTORS – DESIGN
GUIDANCE VS DATA
The authors’ 2025 paper compared the observed Areal Reduction Factors
(ARF’s) of several 2024 Toronto-area storms with design guidance used in
watershed hydrologic modelling. Figure 3
reproduces the observed June, July and August 2024 storm ARF curves and the WMO
24 hours design curve and adds the observed AES index significant storms
reported over 1919-1984. The x-axis
scale has been changed to a logarithmic scale given that the WMO guidance cited
by MNRF extends up to an area of 1000 km2 while the AES index storm
sizes were extended to much greater areas, even over 100,000 km2.
FIGURE 3 – AREAL REDUCTION FACTORS FOR
SIGNIFICANT ONTARIO STORMS, 2024 GTA STORMS AND WMO GUIDANCE
Figure 3 illustrates that areal reduction of storms continues for areas
beyond the WMO 1000 sq.km guidance, as would be expected. There is considerable overlap in the short,
moderate and long duration AES index storm ARFs with very similar median values
of 70% for each of the three duration groups.
This characteristic of large area AES storms is contrary to WMO guidance
that recommends smaller ARFs for smaller area, shorter duration storms.
The AES index shows that significant storms can be sustained over large
distances. Designers may presume such a
characteristic could simplify hydrologic analyses, assuming consistent
responses in drainage and infrastructure systems when catchment areas
increase. Figure 4 explores that
response in the context of wastewater collections systems and rainfall derived
inflow and infiltration (RDII) rates in the Region of York. While the authors’ parallel paper on RDII
provides full background to the analyses, Figure 4 shows how 100-Year RDII flow
rates, normalized relative to the highest rates observed, decline as the
catchment area increases. In comparison
to 2024 storm ARF decline, the decline in RDII is more accelerated with
expanding area. While it is beyond the
scope of this paper to explore this phenomenon further, factors such as the
spatial and temporal variability of rainfall across the catchment, and
hydraulic routing in the system could explain some of the decline. Another explanation could be that the
100-year RDII values are derived based on a data series including small,
localized storms with higher areal decline than represented in the large 2024
storms or the WMO guidance.
FIGURE 4 – AREAL REDUCTION FACTORS FOR 2024
GTA STORMS AND RDII REDUCTION (100-YEAR STORM RESPONSE)
EXPANDING BULL’S EYE EFFECT - GTA
EXAMPLE
The previous sections illustrate that based on Environment Canada’s data
significant rainfall events with large maximum depths and extensive areas have
been occurring regularly in Ontario for over 100 years. Over the past 25 years the impact of
significant rainfall events on urban systems resulting in extensive property
damage has been highlighted by municipalities and by the insurance industry
more recently. The House of Commons Standing
Committee on Environment and Sustainable Development (2025) initiated a study
in recognition of high 2024 catastrophic weather insurance losses and the need
to protect Canadians.
The increase in flood damages since the early 1980’s has been reviewed
by Natural Resources Canada (NRCan) in the report Canada in a Changing
Climate: National Issues (2021). NRCan
reported that:
“Costs associated with damage from extreme weather
events in Canada are significant and rising, largely due to growing exposure
and increasing asset values. The scale of costs suggests that households,
communities, businesses and infrastructure are not sufficiently adapted to
current climate conditions and variability.”
National Research
Council’s flood cost-benefit guideline (2021) supported the above and
demonstrated that observed losses across Canadian provinces was proportional to
the proportion to growth and asset (e.g., GDP, population, dwelling counts and
sewer length).
Growing
exposure has been explained by the Expanding Bull’s Eye Effect (Stradler and
Walker, 2015) where weather hazards are equivalent to arrows and assets are the
target with a bull’s eye in the middle.
Stradler and Walker illustrate how growth in the size of the bull’s eye,
e.g., represented by expanded urbanized land use, increases the exposure to
hazards such as tornados or hurricanes.
They echoed NRCan’s observation stating:
“… disaster frequency and magnitude increases may be,
at least in part, attributable to the surge in people and their assets exposed
to hazards, not necessarily due to changes in the climatology of hazard events
themselves. That is, growing population, developed landscapes, and wealth are
likely important factors in the increasing trends in disaster counts and
impacts.”
The increase in
the size of an asset “target” can be illustrated in the growth of the City of
Toronto over the past 160 years.
Historical land use maps and building construction dates were used to estimate
development limits in 1865, 1905, and 1945 relative to today’s development. Figure 5 maps the August 17, 2024 storm
‘arrow’ relative to the historical and current development bull’s eyes. As the asset bull’s eye has clearly expanded
over time, the 2024 storm is expected to cause greater damage in 2024 than in
1945 or 1965 when exposure was limited with the much smaller bull’s eyes.
FIGURE 5 –
EXPANDING BULL’S EYE EFFECT EXAMPLE – AUGUST 2024 STORM EXPOSURE OVER 1865,
1945 AND CURRENT GROWTH
CORRELATION OF INTENSITY TRENDS ACROSS
VARIOUS DURATIONS, RETURN PERIODS AND AREAS
Real
storms have unique characteristics that make the creation of synthetic design
storms challenging; only a select number of a natural storm characteristics can
be represented in the design event. This
is due to significant variability across characteristics as shown in the AES
storm charts above, all of which have unique return periods. While there can be weak correlations across
parameters (i.e., larger area storms generally have longer durations) there are
more likely to be ‘bands’ of characteristics.
For Example, Figure 6 below shows the high variation in size of short
and long duration 2-inch (50 mm storms), and the duration variation to 2-inch
storms of the similar size.
FIGURE 6 - ONTARIO SIGNIFICANT STORM AREA VS. STORM DURATION - 1919 to 1984
Beyond the variability across significant storms shown above,
Environment Canada data has shown that rainfall characteristics such as design
intensity change over time as new climate data is considered. A review of Engineering Climate Datasets
shows that trends in IDF values are not consistent across different storm
durations or return periods. Figure 7
illustrates shifts in IDF values at 21 long-record southern Ontario when 10
more years of data were added to the v2.0 dataset.
Chart illustrates the average ratio of Engineering
Climate Dataset intensities in v3.1 dataset (records up to 2017) with
intensities in the v2.0 dataset (records up to 2007) for 21 climate stations.
FIGURE 7 – AVERAGE SOUTHERN ONTARIO IDF
TRENDS ACROSS DURATIONS AND RETURN PERIODS
As shown above, 2-year rainfall intensities (green symbols) for various
durations may increase or decrease while 100-year intensities for almost all
durations decrease. This indicates that
the ‘extremes’ are not changing in step with the ‘means’ - this has important
implications when extrapolating trends in annual maximum series (AMS) to
expected trends in rare severe rainfall responsible for flood impacts (see next
section). It also has important
implications when interpreting climate model projections that assess daily
‘heavy precipitation’ statistics but cannot accurately assess short-duration
extreme precipitation linked to urban flooding.
Similar analysis for 29 northern Ontario climate stations in Figure 8
below shows increases in long-duration extremes (50-year and 100-year), unlike
southern Ontario where those intensities decreased. Northern Ontario 1-hour extremes decreased
while southern Ontario value increased.
One consistent trend is that the shortest duration 5-minute to 15-minute
extremes decreased in both regions.
Other regions in Canada can show completely oppose trends to Ontario,
e.g., Alberta showing increases in short duration extremes and decreases in all
moderate and long duration intensities (i.e., 2-year to 100-year, 1-hour to
24-hour) (Muir, 2020).
FIGURE 8 – AVERAGE NORTHERN ONTARIO IDF
TRENDS ACROSS DURATIONS AND RETURN PERIODS
Broaden to extremes across all of Canada, Figure 8 shows how short and
moderate duration 100-year IDF intensities changed across Canada for 226
long-record climate stations. The two
Ontario IDF charts above average station IDF data, obscuring the variability in
the underlying data. In contrast the
Canada-wide Figure below shows individual station data to highlight
variability. Figure 9 clearly shows that
changes in short and moderate duration intensities are not correlated, with any
combination of increases and decreases in the data.
Chart
illustrates the ratio of Engineering Climate Dataset intensities in v3.1
dataset (records up to 2017) with intensities in the v2.0 dataset (records up
to 2007) for 229 climate stations.
FIGURE 9 – DIVERGING SHORT AND MODERATE
DURATION 100-YEAR IDF TRENDS ACROSS CANADA
ECCC ANNUAL SERIES AND REGIONAL TRENDS
ECCC has completed research on 5-minute to 24-hour annual maximum
rainfall in Canada using data from 1950 to 2021 (Cannon et al. 2024). While only annual maxima were analyzed, the
introduction references intense rainfall related to flood impacts:
“Intense short-duration rain can cause flooding and
the failure of buildings, roads, bridges, and other infrastructure, which, in
turn, threatens public health and safety. Furthermore, flooding, primarily in
residential basements, accounted for more than CAD 1 billion in average annual
insurable property damage claims in Canada for the period 2009–21, which is
substantially more than CAD 125–250 million in inflation and wealth adjusted
losses recorded for 1983–2008 (Bakos et al. 2022).”
The
introduction also notes that “engineers use statistics that summarize the
intensity, duration, and frequency of sub hourly to daily rainfall extremes to
design buildings and infrastructure” however the paper does not analyze any of
the statistics that engineers use to characterize intense design rainfall. Annual maximum rainfall observations in the
research can characterize “heavy precipitation” from a broad climate
perspective. For example, the
Intergovernmental Panel on Climate Change (IPCC, 2021) Sixth Assessment Report characterizes
heavy precipitation based on “average annual maximum precipitation amount in a
day (Rx1day)”. The expected annual
maximum daily rainfall amount in Toronto is 47 mm based on ECCC’s v3.3 dataset
which is not sufficient to cause flooding in any modern municipal
infrastructure system, nor widespread flooding in most historical systems.
To
offer clarity on the distinction between changes in extremes and averages, IPCC
has published answers to Frequently Asked Questions for Assessment Report 6
(AR6) specifically “FAQ 11.1 | How Do Changes In Climate Extremes Compare With
Changes In Climate Averages?” such as (our bold):
“Climate change will also manifest differently for
different weather regimes and can lead to contrasting changes in average and
extreme conditions.”
“At the local scale, average and extreme surface
temperature changes are strongly related, while average and extreme
precipitation changes are often weakly related. For both variables, the changes
in average and extreme conditions vary strongly across different places due
to the effect of local and regional processes.”
The
ECCC 2024 research found the following for the Great Lakes/St Lawrence Basin
region that would encompass the majority of urban centres in Ontario:
i) there are statistically
significant increasing trends in annual maximum rainfall in 5 of 11 regions of
Canada, for one or more durations,
ii) 5 of 11 regions including Great
Lakes/St Lawrence Basin do not have statistically significant increases in
annual maximum rainfall for any durations;
iii) with the exception of southern
Alberta, provincial and territorial regions showed statistically significant
positive scaling for one or more durations rates (i.e., rainfall increased with
higher summer dewpoint temperatures).
While
some may interpret heavy precipitation trends as indicative of changes that
could explain increased flooding, ECCC clarifies that:
“Results here apply only to the time and temperature
dependence of average annual maximum rainfall. Observational evidence of
changes in longer return period rainfall events would require a different
statistical methodology …”
EXPANDING ARROW EFFECT – A BULL’S EYE
EVERY TIME
ECCC’s
2024 paper notes that the detectability of significant trends can increase with
wither the aggregation of regional data, or the use of longer records.
“… detectability of significant trends at individual
stations was poor but improved as data were aggregated to larger regions. In an
update to the analysis of daily extreme precipitation by Westra et al. (2013),
Sun et al. (2021) found that the use of longer records led to improvements in
the ability to detect the intensification of extreme precipitation.”
In
the authors’ 2025 WEA conference paper adopted the longer record approach to
assess Toronto-area trends. It used ECCC daily rainfall records dating back to
1840 to assess changes in extreme rainfall statistics such as 100-year design
intensities. This record was 100 years
longer than the data used in ECCC’s Toronto City gauge (Station ID 6158355) IDF
analysis.
ECCC’s 2024 paper adopted both the longer
record and regionalization approaches, considering 15 more years in the
rainfall data record and assessing trends in large regions. The Great Lakes/St Lawrence Basin Region area
adopted is approximately 1000 km long (Windsor, Ontario to Quebec City) and
over 150 km wide, representing an area of about 150,000 km2. The AES index data presented earlier
indicates that significant storms generally cover only a fraction of that very
large region. Consequently, regional
data may not be representative of local conditions.
Regionalization of rainfall data did not
result in increases in annual maximum rainfall in the Great Lakes/St Lawrence
Basin Region. However, there is a risk
that pooling data over a large area could result in an “Expanding Arrow Effect”
whereby localized increases are extrapolated across areas without such
changes. While there is merit in a
precautionary approach and the incorporation of safety factors in design to
account for future changes and uncertainties, the Expanding Arrow Effect could
overstate risks and result in costly infrastructure overdesign.
MICRO-REGION IDF VARIATIONS COMPARED TO
LONG-TERM IDF
Regions with multiple rain gauges will intrinsically have variability in
observed rainfall and derived IDF statistics at each gauge. This is due to the localized nature of
convective storm peak intensities, as illustrated in the review of 2024
storms. Across the City of Markham, 18
local gauges have been collecting local rainfall data as early as 2008. Figure 10 shows the maximum and minimum
2-year IDF values across these gauges, compared to the Markham 2-year standard,
the 5-year standard and the 95% confidence bands at the long-term Buttonville
Airport gauge (Station ID 615HMAK). The
local IDF data up to the end of 2024 is included.
FIGURE 10 – MARKHAM LOCAL RAIN GAUGE 2-YEAR
IDF VS LONG-TERM STANDARDS
The Markham 2-year standard, based on long-term City of Toronto data,
falls within the maximum and minimum values of all the local gauges. The long-term Buttonville Airport 95%
confidence bands generally fall within the maximum and minimum local
values. It should be noted that the
period of record for the Buttonville Airport gauge in ECCC’s v3.3 Engineering
Climate Dataset extends from 1986 to 2016.
Given that some significant local storms occurred in 2017 and 2024 and
would not have been considered in the Buttonville data, it is not unexpected
that the minimum local Markham gauge values may be slightly above the
Buttonville lower confidence limit for some durations.
Figure 10 shows that the maximum 2-year
statistics approached and slightly exceeded the Markham 5-year standard design
intensity (i.e., the 6-hour duration).
The maximum short duration intensities for periods of 2 hours and lower
fell below the 5-year standard. For the shortest durations, the maximum local
values approached the city standard 2-year value. The Markham local IDF data
illustrates that additional data introduces additional variability even across
a relatively small ‘region’. While
maximum values exceed standard values at specific gauge points, the average
values of the longest-period gauges were below the Markham standard for 5-to-30-minute
durations, and close to the Markham standard for 1-to-6-hour durations. Recognizing that the maximum rainfall depths
are not sustained over wide catchments, but instead lessen over wider areas,
the Markham standard IDF values are considered to be reasonable. That is the Markham standard is similar to
the average of the long-term local gauge values.
Over time local rainfall data may be used to review low return period
values considered in design, e.g., beyond the 2-year values above. However, given that the Markham standard is
based on Toronto data extending back to 1940, it may be several decades before
local data can be used to assess infrequent storm design intensities. The above example of variability of IDF
values across a small region should be considered when conducting regionalized
rainfall analysis. For example, any
increase in values observed locally across a region may not be applicable
across large catchments that will experience averaged values as opposed to
observed peaks at gauge points.
CONCLUSIONS
This paper demonstrates that extreme
rainfall events in Ontario exhibit substantial variability across depth,
duration, and spatial extent, with wide scatter in storm characteristics
observed in both historical and recent data. The long-term AES storm index
shows that large, high-depth storms with extensive areal coverage have occurred
regularly for more than a century, and that neither the frequency nor the
spatial scale of these significant storms has increased in a systematic way
over time. Recent damaging storms in southern Ontario fall well within the
historical envelope of observed storm sizes and depths, reinforcing the
conclusions that extreme rainfall is not a new phenomenon in the region.
The
analysis highlights that spatial scale is a dominant factor in interpreting
rainfall data and translating it into infrastructure response. Point-based IDF
statistics characterize local peak intensities but do not represent the
spatially average rainfall that governs watershed and urban drainage system
performance. Historical storm data and recent observed events consistently show
that peak rainfall intensities attenuate with increasing area, often well
beyond the ranges addressed in commonly cited design guidance. Comparison of
observed Ontario storm ARFs with WMO design curves indicates that real storms
can sustain high rainfall over large areas in ways not fully captured by
generalized guidance, showing the importance of using locally relevant data
when assessing areal effects.
The comparison between rainfall and areal reduction and observed RDII
response further illustrates that system behaviour attenuates more rapidly than
rainfall alone as contributing area increases. This divergence reflects the
combined influence of spatial rainfall variability, temporal distribution,
hydraulic routing, and system storage, and confirms that infrastructure
response cannot be inferred directly from rainfall statistics without explicit
consideration of scale and system characteristics. These findings show the need
to avoid extrapolating small scale extremes to large catchments or systems
without appropriate spatial adjustment.
The paper also demonstrates that trends in rainfall intensity are not
consistent across durations, return periods, regions, or spatial extents.
Updates to Engineering Climate Datasets show that changes in frequent,
long-duration rainfall intensities are often weakly correlated with changes in
rare, short-duration extremes that drive urban flooding. Regional analyses
across Canada further reveal that increases and decreases in extreme
intensities occur in various combinations, even within the same region. This
lack of coherence cautions against assuming that trends in “heavy
precipitation,” as characterized by annual maximum series or daily extremes,
are indicative of changes in flood-causing design storms. The increase in
reported flood damages over recent decades is more consistently explained by
growing exposure than by demonstrable changes in storm characteristics. The
Expanding Bull’s Eye Effect provides a useful framework to interpret this
pattern, whereby urban growth and increased asset density expand the target
exposed to historically recurring storms. Conversely, the aggregation of
rainfall data over very large regions introduces the risk of an Expanding Arrow
Effect, in which localized changes are generalized across areas that do not
experience the same conditions. While regionalization and long records improve
statistical detectability of trends, they may also obscure the localized
variability that governs infrastructure response at the scale of sewersheds,
watersheds, and urban catchments.
Local gauge analysis within the City of Markham illustrates how
additional data increases observed variability rather than converging toward a
single design value. While localized peaks can exceed standard design
intensities at individual gauges, spatial averaging across catchments results
in responses that are more consistent with long-term regional standards. This
reinforces the appropriateness of design values derived from long records,
while highlighting the need for caution when interpreting short record local
extremes or regionalized trend results.
Overall, the findings demonstrate that no single storm, statistic, or
regional trend can adequately represent the range of rainfall conditions
relevant to infrastructure and watershed flood response. Deterministic design
storms remain a practical and necessary engineering tool, but their limitations
must be understood, particularly with respect to spatial representation and
scale dependence. As rainfall monitoring networks expand and datasets lengthen,
there is an increasing opportunity to supplement deterministic approaches with
probabilistic frameworks that explicitly account for variability, uncertainty,
and spatial averaging. Continued reliance on observed data, careful attention
to scale, and clear distinction between exposure-driven risk and hazard change
are essential to support robust flood resilience assessments and informed
updates to design practice.
BIBLIOGRAPHY
Cannon, A.J, Jeong, D-I, Yau, K-H.
(2024). Updated Observations Provide
Stronger Evidence for Increases in Subhourly to Hourly Extreme Rainfall in
Canada. Journal of Climate, Volume 37. DOI: 10.1175/JCLI-D-23-0501.1 https://journals.ametsoc.org/view/journals/clim/37/12/JCLI-D-23-0501.1.xml
CBC News (2024). Toronto just saw record rainfall. Why wasn't it more prepared? https://www.cbc.ca/news/canada/toronto/rainfall-flood-toronto-record-1.7266064
Conference Board of Canada (2024).
Bright Future, Ep. 40: Blair Feltmate on Adapting to our Changed Climate https://www.conferenceboard.ca/insights/blair-feltmate-on-adapting-to-our-changed-climate/
Environment Canada, Atmospheric
Environment Branch (1988). An Index to Storm Rainfall in Canada by B.Routledge,
D.Carr and W.Hogg. CLI-1-88 https://publications.gc.ca/site/eng/9.880154/publication.html
House of Commons Canada, Standing
Committee on Environment and Sustainable Development (2025). Minutes of
Proceedings, 45th Parliament, 1st Session https://www.ourcommons.ca/DocumentViewer/en/45-1/ENVI/meeting-6/minutes
Intergovernmental Panel on Climate Change (2021). Sixth Assessment
Report, Working Group 1: The Physical Science Basis, Chapter 11: Weather and
Climate Extreme Events in a Changing Climate, 11.4 Heavy Precipitation/ 11.4.2 Observed Trends. https://www.ipcc.ch/report/ar6/wg1/chapter/chapter-11/
Karney, B., Zuccaro, C., Muir, R.,
Parhizagari, Z., Zi, J. (2025) How Big is a 100-Year Storm? A
Simple Definition with a Surprisingly Complex and Elusive Reality, Considering
Both Climate Change Trends and the 2024 Storm.
2025 WEAO Technical Symposium, London, Ontario, April, 2025. Paper link: https://drive.google.com/file/d/1-CIOu4NEHgo2EnTvKbJHPqeDl5My7XCG/view?usp=drive_link;
presentation link: https://drive.google.com/file/d/1oTttyYKVqbj7f4xPeYzep2774EIbasge/view?usp=sharing
Marsalek, J, W.E. Watt (1983),
Environment Canada, National Water Research Institute, Design Storms for Urban
Drainage Design. https://drive.google.com/file/d/1N4xc7qJAJv0ZhY24KDVOcSrLV0NbMn72/view?usp=sharing
Muir, R. (2020). Can We Use Daily Rainfall Models To Predict
Short Duration Trends? Not Always - Observed Daily and Short Duration Trends
Can Diverge https://www.cityfloodmap.com/2020/07/can-we-use-daily-rainfall-models-to.html
National
Research Council of Canada (NRC) (2021) Guidelines on Undertaking a
Comprehensive Analysis of Benefits, Costs and Uncertainties of Storm Drainage
Infrastructure and Flood Control Infrastructure in a Changing Climate. Appendix
D | Direct & Indirect Long Time Horizon Damages, Section 3.1 Insurance Bureau
of Canada Catastrophic National and Regional Losses, Figure 4: Proportion of
GDP, Population, Dwellings and Sewer Infrastructure as Indicators of Expected
Annual Damages (Water Perils), by Province https://nrc-publications.canada.ca/eng/view/object/?id=27058e87-e928-4151-8946-b9e08b44d8f7
Strader,
Stephen & Ashley, Walker. (2015). The Expanding Bull's Eye Effect.
Weatherwise. 68. 10.1080/00431672.2015.1067108. https://www.researchgate.net/publication/281294791_The_Expanding_Bull's_Eye_Effect
