Southern Ontario Extreme Rainfall Intensity Trends - Update From Environment Canada Engineering Climate Datasets

Environment and Climate Change Canada has updated and extended the Engineering Climate Datasets as noted in the last post. This post shows the updated trends in extreme rainfall intensities across long-term southern Ontario climate stations - the good news is that intensities have not increased. This means that infrastructure built in the last few decades is not undersized considering current rainfall design intensities.

Previously trends in some southern Ontario intensity-duration-frequency (IDF) values and annual maximum series were evaluated in my paper "Evidence Based Policy Gaps in Water Resources: Thinking Fast and Slow on Floods and Flow" in the Journal of Water Management Modeling

Further analysis of trends in long-term stations has been presented in this blog and in the National Research Council of Canada's "National Guidelines on Undertaking a Comprehensive Analysis of Benefits, Costs and Uncertainties of Storm Drainage and Flood Control Infrastructure in a Changing Climate" - that guideline included trends in southern Ontario up to Version 3.10, as presented in the earlier post:

The southern Ontario IDF trend data has now been updated based on the Version 3.3 dataset released in May 2023 and includes some station updates to 2021- 9 of the 21 stations were updated. The following table and charts show the trends in 2-year to 100-year design rainfall intensities.

The table below shows changes in average intensity - decreases since 1990 are shaded in green and increases are in red. Note the the trends are weighted by record length. Across all durations and return periods the average decrease is - 0.33 %.  That is a slight decrease from the Version 3.20 datasets, meaning less intense rainfall when more recent data has been included. On average 30 statistics decreased while 20 statistics increased.

It is noteworthy that none of the 2-year intensities increased and the largest increase was 0.8% for 100-year intensities at one station for durations of 30 minutes and 1 hour. Overall for 21 stations 100-year intensities were virtually unchanged with the average intensities decreasing 0.1% after 30+ years, and the median increasing 0.2%. Skewed data statistics should increase over time with longer records - check out this post for more on that:

The following chart shows the range of changes for each return period as well as the average change. The decreases are greater than any increases for the 5 to 100-year events.

The following chart provides more of a breakdown by duration. One can see the red 5 minute intensities decreased on average for all return periods. The 2 hour to 24 hour intensities decreased for most return periods and where there were increase they were minor compared to other decreases. For the 5-year to 100-year return periods the 15 minutes to 1 hour intensities increased, but by no greater than 0.8%. These increases and decreases are basically insignificant in terms of impacts on infrastructure design.

This last table is annotated to show how various statistics are used in design. Infrastructure that has been designed considering short duration intensities like local sewer systems are now subject to virtually the same 2 to 10-year design intensities that existed over 30 years ago. Ponds designed for long duration higher return periods (e.g., 100-year events) are now subject to virtually the same intensities, or design event volumes, they were subject to decades ago as well. 

Environment Canada IDF Curve Update - Version 3.30 Release Adds Stations and Extends Data Records

Environment and Climate Change Canada has released the Version 3.30 of the Engineering Climate Datasets:

The number of stations and the length of average station record has increased with 276 updated stations, 38 new stations and 9 joined stations. This brings the total number of stations to 714, a significant increase relative to the time I started working.

The following table and charts illustrate how the number of stations and station-years of record have increased over time.

The number of "station-years" of data in the current data set (some data up to 2021) is 66% greater than the approximate number of station years in earlier data up to 1990. That is good to see.

Average record length is now up to 26.2 years, and the average 'last year' of data is just past 2007 (there are 3 stations that stopped in the 1940's, which brings the average last year down).


This chart shows how the number of stations and station-years of data have increased over the past several decades. Note that the years on the x-axis for 3.1, 3.2, and 3.3 are the respective release years (2019, 2021 and 2023), and not the newest data year included.

There have been suggestions that the number of stations has declined, adversely affecting the ability to assess changes in extreme rainfall intensities - see discussion in an earlier post

Mekis et. al noted that the number "Manual" stations has declined as shown in the following chart - that can reflect the change in technology to automated stations that replaced manual stations (remember those analog strip charts?). The number of stations with IDF data has increased though, from only 532 stations in the 1990 data set to 714 now - a 34% increase. Those increases in station numbers are shown against the decrease in manual stations below. 

The average record length has been increasing overall as well (see table above), resulting in more reliable trend data today. Note that the addition of many new stations tends to pull the average record length down as shorter record length stations are factored in. Overall, more data is better and the new station records can extend over time.