Risks Are Where You Map Them - The Truth in WYSIATI (What You See Is All There Is) When Defining Urban Riverine Flood Plain Risks

"To Map or Not To Map, That Is The Question"

If buildings along a channel are subject to frequent flooding but there is no regulatory mapping, is it really a 'flood risk zone'?

Risks are sometimes equated to the presence of regulatory mapping that defines risk. But risks exist whether you map and manage them or not. This post explores the "Where?" and the "How?" of floodplain mapping to best define risks.

First some background - regulatory mapping, such as under Ontario's Conservation Authorities Act regulations, delineates risk areas where it has been decided to map risks. It is a choice - to map or not to map. Regulation and risk are therefore quite different, as risk can extend upstream of the last flood plain mapping sheet, extending further up the river system, or even up through the urban landscape and infrastructure systems. Obviously, our goal in terms of risk management and urban flood damage mitigation would be to map all the important risks - those contributing to frequent or extensive damages - and to then develop a strategy to address those risks.

While Ontario is quite advanced in terms of the identification and management of river flood risks, there are opportunities for improvement where risks on smaller urban watercourses are defined. Traditionally, a catchment size limit of half a square mile, or about 125 hectares, was used to define rivers reaches where hydrologic and hydraulic modelling would be completed to delineate the extent of the regulatory flood plain. Reaches for smaller reaches were not typically mapped. In new development areas, a smaller threshold is often applied during land use planning and development servicing studies which ensures that local risk are managed. However in historical development areas, it is not uncommon that some river reaches with catchment sizes of several hundred hectares are not mapped, meaning that risks in small flashy systems (those that respond to high intensity rainfall convective summer storms) are not identified, regulated, or mitigated.

Given today's focus on Best Practices to mitigate existing community flood risks, the mapping of riverine flood risks is a subject of attention. The federal government, Natural Resources Canada and Public Safety Canada, have recently published a Federal Floodplain Mapping Framework Version 1.0, 2017 that forms part of the Federal Floodplain Mapping Guidelines Series - a link to that document is here :

http://publications.gc.ca/collections/collection_2017/rncan-nrcan/M113-1-112-eng.pdf

The following figure summarizes the framework steps.

The report notes priority setting including "where to conduct floodplain mapping"  as a future activity:
We encourage that Priority Setting be robust to answer the "where" and "how far" mapping will be pursued. Why? There is a considerable amount of focus on next steps, getting lost in the weeds with LiDAR data acquisition and climate change assessments in the next delineation step in the framework. So the question is "Is there enough prioritization relative to other technical considerations?", as suggested in the follow graphic:


How about an example showing why "Where to Map?" is more important than "How to Map?".

The Don Mills Channel is a small urban drainage channel, with a historically realigned watercourse called Cummer Creek in the Don River Watershed. Floodplan mapping has been pursued since the 1960's in Ontario and a great summary of floodplain mapping history is available here: The State of Floodplain Mapping in Ontario Presented to the Institute for Catastrophic Loss Reduction, June 15, 2007, Don Pearson, General Manager, Conservation Ontario.

The City of Markham is conducting a flood reduction Class Environmental Assessment study to understand the causes of flooding in the Don Mills Channel study area and to develop a range of alternative solutions to reduce flooding and flood damages. The Class EA flood risk analysis modelling (not the regulatory mapping) shows the extent of flooding during a 100-year event:


While the watercourse is an area of focus for flood control, regulatory floodplain mapping was completed for the channel only in 2011. Prior to that, estimation mapping was completed (by this post's author) to support generic regulation updates in 2006.

So how has the characterization of risks changed with this 'new' mapping of the Don Mills Channel?The following figures illustrate how building flood risks in the newly mapped Don Mills Channel compare with city-wide Markham building flood risks identified as part of its Flood Emergency Response Plan.

Extending floodplain mapping a one tributary can dramatically change the characterization of overall riverine flood risk for the 100-year flood event. 

Riverine flood risks defined by depth of flooding at building structures changes significantly with the extension of floodplan mapping to previously unmapped watercourse reaches.
Frequent flooding during low return period events increases significantly with the extension of floodplan mapping to the Don Mills Channel, a tributary of the Don River also called Cummer Creek.

The charts above clearly show that the accounting of at-risk buildings within the extension of regulatory floodplain mapping in a single tributary dramatically changes the characterization of city-wide risks.  For example, the number of high flood depth structures (depth greater than 0.6 metres) during a 100 year event in the new reach is more than double the entire previous city-wide number. For frequent flood events like the 10 year storm in the last chart, the number of high-depth structures increases by 10 times or more - these new, frequent, high-depth structures can be expected to represent a high proportion of average annual flood damages.

While development occurred in the Don Mills Channel area in the 1960's (channelization and realignment was in fact to support development at the time according to the original 1966 design report), floodplain mapping was not initiated until 40 years later with the generic regulation update estimation mapping (i.e., HEC GeoRAS considering only overland components and not culvert enclosures - see this post for our perspective on riverine flood vulnerability methods). Advanced mapping to support regulation was completed later in 2011.

So what was the critical question to defining risk? Was it "How?" the floodplain should be mapped, like using various hydraulic models:

1) basic 1-dimensional HEC-RAS with no culverts (the 2006 HEC-GeoRAS estimation) or
2) more advanced 1-dimensional HEC-RAS with culverts (2011 mapping) or
3) even more advanced 2-dimensional PCSWMM integrated with the municipal storm pipe network (2018 Class EA)?

.. or using various design storms to determine design flows like:

1) TRCA's low-intensity watershed storm, or
2) Markham's high-intensity urban storm

... or considering various climate conditions like:

1) today's climate and IDF curves (which are not reflected in the watershed storms at all) or today's regulatory storm (Hurricane Hazel) or
2) tomorrow's estimated climate and IDF curves (take your pick on methods ... none of which converge) or tomorrow's regulatory storm (Hurricane Hazel ... exactly the same as today's Hurricane Hazel)

.. or refining the digital elevation model using LiDAR to get those flood depths to millimetre accuracy around every metre of he perimeter of each building, instead of typical mapping?

A recent presentation describes the modelling uncertainties in "How" various analysis methods have been applied from the 1960's to today:




Or is the most important question in defining risk the "Where?" Yes, "Where?" is the important part and the other considerations of "How?" are secondary or tertiary at best.

***

Nobel laureate Daniel Kahneman, author of Thinking Fast and Slow, referred to WYSIATI, an acronym for "What You See Is All There Is", which is a way of saying that we often miss things by acknowledging only the things we know (the known knowns) while being oblivious to the unknown unknowns. We recognize only what we see - or what we map and acknowledge. In the case of flood plain mapping, the risks we map are often considered to be all there is, since that is all that is being regulated or managed. Obviously then, "Where?" and "If" we map riverine flood risks can be of critical importance. And although our technical methods for delineating flood risks must be carefully considered, "How?" we define risks is relatively less important.

Climate Models Predict Decreasing Extreme Rainfall Intensities and No Change for Moderate Storms In Southern Ontario

Abstract
A study by University of McMaster researchers entitled "Assessment of Future Changes in Intensity-Duration-Frequency Curves for Southern Ontario using North American (NA)-CORDEX Models with Nonstationary Methods" predicts that extreme 50-year rainfall amounts will decrease in Southern Ontario by 2050, that moderate 25-year rainfall amounts will remain flat, and frequent 10-year rainfall amounts will increase overall for long durations but are mixed for short durations.

The paper is available at this link.

Ganguli and Coulibali state in the abstract "Our results showed that extreme precipitation intensity driven by future climate forcing shows a significant increase in intensity for 10-year events in 2050s (2030-2070) relative to 1970-2010 baseline period across most of the locations. However, for longer return periods, an opposite trend is noted."

The following tables illustrate how rainfall intensities are predicted to change over periods of 1 hour to 24 hour at Southern Ontario Locations including Toronto, Hamilton, Ottawa, Windsor, London Trenton, Stratford and Shand (Fergus Shand Dam).

Southern Ontario Extreme Rainfall Predicted to Decrease With Climate Change 

Southern Ontario Moderate Rainfall Predicted to Not Change Overall With Climate Change
Southern Ontario Frequent Rainfall Predicted to Increase With Climate Change For Long Durations
Specifically, the Toronto-Hamilton 50-year rainfall amounts over 1-2 hours are predicted to drop by up to 5% or increase by 1% assuming non-stationary distributions - such changes are considered insignificant in the realm of infrastructure design given uncertainties with other factors and analysis methods. Across Ontario, the largest predicted decrease is 44% at Shand and the largest increase of 14% is in Windsor. These short duration amounts are most relevant to peak flow affecting urban flooding.

Meanwhile, the 25-year rainfall amounts are predicted to increase or decrease by 7% and 4% respectively, again an insignificant amount in design. Across Ontario, the largest predicted decrease is 35% at Shand and the largest increase of 10 % is in Windsor.

In contrast, the 10-year rainfall is predicted to increase overall, especially for long durations. For short durations of 1-2 hours the maximum increase is 22% in Hamilton and the maximum decrease is 32% in London.  For the 1-hour duration 5 stations show a decrease in 10-year rainfall (-5% to -22%), while only 3 stations show increases (+5% to + 22%).

The authors conclude that "The findings, which are specific to regional precipitation extremes, suggest no immediate reason for alarm, but the need for progressive updating of the design standards in light of global warming."

It is interesting that authors refer to 'global warming' as opposed to climate change, perhaps since extreme rainfall if not predicted to change with future temperatures.

***

Previously, we analyzed the trends in the Engineering Climate Datasets for long term Southern Ontario gauges:

http://www.cityfloodmap.com/2018/01/short-duration-frequent-rainfall-show.html

The review showed for a 2-hour duration the 10-year intensities decreased on average 0.8% from 1990 to the most current Version 2.3 dataset. The McMaster research predicts an average increase of 4.3% for 2-hour 10-year rainfall (non-stationary vs non-stationary). A greater increase is predicted for stationary vs stationary. Question: when will the real data observations start to show an increase like the model suggests? Maybe it won't. Reminds us of this quote:

"It doesn't matter how beautiful your theory is, it doesn't matter how smart you are. If it doesn't agree with experiment, it's wrong."
Richard P. Feynman

Given rainfall design intensities are decreasing in many Southern Ontario cities based on past observations it is refreshing to see climate modelling that predicts trends that are consistent with real data - yeah !

New Version 3.0 data for southern Ontario shows a further decrease in design intensities since 1990. This data shows a greater decrease for the lower return periods, contrary to the model predictions indicating low return period intensities will increase.