Evaluation of partial duration and annual maxima series rainfall models using the low impact development treatment train tool

Sandra Vrban, Yi Wang, Ed McBean and Bahram Gharabaghi


The design of stormwater management infrastructure relies upon the use of rainfall data that is characteristic of the return period being modelled. As the design of traditional stormwater infrastructure has transitioned towards the use of Low Impact Development (LID) practices, it is pertinent that accurate rainfall data is used to minimize costs and maximize use. Traditionally, historic rainfall data has been used for this purpose as it was considered to be constant and representative of present rainfall events. The use of historic rainfall data has been questioned in recent years due to the increasing intensity and frequency of storm events associated with climate change. Various rainfall frequency models exist which can more accurately represent current day storm events. The objective of this study is to compare Partial Duration Series (PDS) and Annual Maximum Series (AMS) rainfall frequency models using the Low Impact Development Treatment Train Tool (LID TTT). The LID TTT was used to model three case-study sites located within the Greater Toronto Area. All sites incorporate a number of LID practices, and were modelled using both PDS and AMS derived rainfall data for 6 and 2-hour storm durations with 2 and 5-year return periods. The study found that the use of PDS derived rainfall data resulted in significantly higher values of total site rainfall, site infiltration, and total external outflow, when compared to the results obtained using AMS derived rainfall data. PDS derived rainfall data is therefore more characteristic of 2 and 5-year storm events, and is better suited for the development of stormwater management infrastructure in anticipatio

Permanent link: