Use of Local Soil and Vegetation Classifications to Improve Regional Downstream Hydraulic Geometry Relations

Abstract

Regional downstream hydraulic geometry equations are developed using flow with a 1.5-year recurrence interval, and corresponding widths, depths, and velocities. The study includes 634 gauge stations located across Canada and are grouped by characteristics that affect or are correlated with bank and bed stability. Jack-knife analysis was done to test the predictive capabilities of each grouping. For width prediction, grouping the data by geology and vegetative characteristics improved predictions in 14 of the 15 categories, with drainage characteristics and mode of deposition groupings showing the best results. Grouping the data by vegetative characteristics showed the best improvement among noncontiguous categories when considering all of the hydraulic geometry variables (width, depth, and velocity). These groupings are investigated further. The study demonstrates that classifying downstream hydraulic geometry equations by local geology and vegetation improves the predictive capabilities of the equations. It also shows that though regionalization does not need to occur in a contiguous region, regionalization still provides better results due to the limited resolution of nationally available GIS datasets. As demonstrated in previous studies, it is also shown that downstream hydraulic equations can be applied to semialluvial rivers. Finally, multiple regression was tested with width prediction as a function of flow and up to three grouping variables. The interaction of the grouping variables showed no improvement in the results.