Urban Ontario River Regimes: An Analysis of Four Major Watersheds

Abstract

The effects of urbanization on river systems are not well understood as appropriate field parameters for representing such changes require years of consistent monitoring data for an accurate comparison. Furthermore, due to their varying degrees of urbanization and management efforts, conditions are not consistent even within the same watersheds so representative sites are difficult to distinguish. This level of data is not typically available for watersheds in Ontario due to budget and personnel constraints of organizations undertaking such monitoring activities. To direct monitoring and management efforts more effectively, a comparative aerial imagery analysis was undertaken for a ~50-year timeline for twenty study sites across the Mimico Creek, Etobicoke Creek, Highland Creek, and Duffins Creek watersheds as well as a comprehensive field analysis to characterize current conditions. A statistical analysis that included a Pearson correlation coefficient analysis and stepwise regression utilizing values obtained through the Stream Power Index for Networks (SPIN) tool and manual calculations including three ratios from Wood-Smith & Buffington (1996) was also completed. These ratios included: bankfull width (Wbf)/bankfull depth (hbf), D50/bankfull depth (hbf), and critical shear stress (τc50)/bankfull shear stress (τbf). Results indicated a strongly negative correlation between entrenchment and the Wbf⁄hbf ratio and positive correlations of varying strengths between the τc50⁄τbf ratio and both the Wbf⁄hbf and D50⁄hbf ratios. Though weak, percent imperviousness and specific stream power were able to predict the τc50⁄τbf ratio. Finally, t-tests between sites categorized as “rural” (≤30% imperviousness) and “urban” (>30% imperviousness) revealed when a control is placed on drainage area, increases in specific stream power, D50, bankfull width, bankfull depth, and slope is observed in “urban” areas. It is believed that such statistical parameters could be considered applicable as a first order estimate of further stream pattern analyses but are not strong enough correlations to be utilized for design purposes.