Urban drainage design methods in Québec and Canada must evolve to address the increasing pressures associated with climate change. Current approaches, often based on synthetic design storms, assume spatial and temporal stationarity and fail to capture real rainfall variability or observed storm dynamics.
This poster presents a new methodology that aims to revise the design storm extensively used in sizing and assessing the performance of drainage networks through two complementary objectives.
The first objective focuses on redefining the historical design storm, using the province of Québec as a case study. Rainfall events from 198 rain gauge stations will be extracted and classified by duration and return period for each station. A functional principal component analysis (FPCA) will be used to identify the dominant characteristics of each event, which will then be normalized and clustered using a projected k-means algorithm to establish median hyetographs for each category. These representative hyetographs will be made available through an interactive map, allowing designers to select the most critical storm for their project. The method will be validated in several Québec cities equipped with calibrated SWMM models and will be adaptable for any rain gauge in Canada.
The second objective is to adapt this historical design storm to future climate conditions. By combining the Clausius-Clapeyron relationship with projections from global climate models (CMIP6) and convection-permitting regional models (CP-WRF, CP-CRCM6), rainfall intensity can be adjusted for multiple time horizons (2011-2040, 2041-2070, 2071- 2100) and emission scenarios (SSP1-2.6 to SSP5-8.5). The evolution of historical and future hyetographs will also be compared using FPCA to assess potential structural variations in rainfall patterns.
This new design storm methodology, with enhanced spatial and temporal definition, will improve the accuracy of hydraulic design and strengthen the capacity to anticipate future climate impacts on urban drainage infrastructure.