Geomorphic study
In order to obtain a large set of data points, active channel width was measured using current orthophotos with GIS software (ArcMap 10.5, 2017). Active channel width was measured between the channel banks, perpendicular to the downstream direction, every 50 m. Several other geomorphic factors were also recorded, including the process domain as identified from the orthophoto, the surficial geology, and proximity to glacial features (e.g., eskers, moraines). Surficial geological deposits were recorded in four different categories (bedrock, peat, subglacial and deltaic fine sediment (including postglacial clay, silt and sand and subglacial sands), coarse till). Using a 2-m digital elevation model, we determined the channel slope and the drainage area at each measured transect, the latter of which was used as a proxy for bankfull discharge. We then plotted the power relationships between drainage area and channel width for all transects and each process domain separately to determine whether the stream networks conform to expected downstream hydraulic geometry relationships.
To predict channel width we used multiple regression models, based on drainage area distance downstream from lake, channel slope, proximity to glacial feature, and type of surficial geology deposits (categorical). Selection was based on lowest AIC with backward-step selection. The relationship between channel width and drainage area were analysed using power regressions. Multiple linear regressions were used to determine logged values of power regression. All statistical analyses were performed in R studio version 3.6.0 (R Core Team, 2021).