Design flood estimation in flood hazard studies: a three-decade systematic review of practices in Canada

Design floods for flood hazard studies are often estimated through flood frequency analysis (FFA). However, many decisions involved may lead to inconsistencies. There is also a demand to incorporate climate change into FFA, but guidelines are lacking. Although these challenges are acknowledged, limited literature documents how they manifest in practice. We systematically reviewed design flood estimation practices in Canada based on 75 technical reports from 1990 to 2023 across six provinces (with 65% from Alberta due to availability). We found substantial variations in data pre-processing, assumption validation (distribution homogeneity and stationarity, and regional homogeneity), FFA methods (e.g. candidate distributions, parameter estimation, distribution selection, uncertainty quantification), software, and post-processing. While recent reports have increasingly disclosed key methodological details, several decisions remain subjective. Floods are typically assumed to be stationary and homogeneous, overlooking human-induced impacts and multiple flood-generating mechanisms. Nonstationary and mixture FFA remain seldom applied. Climate change considerations are often limited to commentary sections, and climate-adjusted design floods are only mandatory in some jurisdictions (e.g. British Columbia). Open software is commonly used for FFA, but its limitations have increased reliance on proprietary software. Thus, there is a need for improved consistency, realism, and research-to-practice translation in FFA practices.