Comparison of regression relations of bankfull discharge and channel geometry for the glaciated and nonglaciated settings of Pennsylvania and southern New York

Streambank erosion in areas of past glacial deposition has been shown to be a dominant source of sediment to streams. Water resource managers are faced with the challenge of developing long and short term (emergency) stream restoration efforts that rely on the most suitable channel geometry for project design. A geomorphic dataset of new (2016, n=5) and previous (1999–2006, n=96) estimates of bankfull discharge and channel dimensions at U.S. Geological Survey streamflow-gaging stations was compiled to present and contrast the glaciated and unglaciated noncarbonate settings of southern New York and Pennsylvania that included selected areas of Maryland. Empirical models were developed by using simple linear regressions that relate bankfull discharge and channel geometry to drainage area (regional curves). Significant relations (p<0.05) were able to explain variability with coefficient of determination (R2 ) values of 0.89 for bankfull discharge, 0.94 for cross-sectional area, 0.87 for bankfull width, and 0.83 for bankfull depth. These regression relations for the glaciated noncarbonate settings of northern Pennsylvania and southern New York were able to provide a slightly better fit than regional curve models developed previously for the entire noncarbonate region of Pennsylvania. Although, the analysis of covariance (ANCOVA) results for comparison between regression equations for the glaciated and unglaciated settings showed that except for the significant intercept of bankfull discharge versus drainage area (F=8.26, p-value<0.005), the regression equations are not significantly different between the glaciated and unglaciated setting of Pennsylvania and southern New York. Therefore, data stratification by glaciation does not improve regional curves relations developed previously for the noncarbonate (glaciated and unglaciated) and carbonate settings of Pennsylvania and Maryland. Further analysis that incorporates data stratification or multivariate approaches based on mean annual runoff, precipitation, slope, stream classification, or other relevant parameters may optimize the accuracy and utility of statewide models. The new estimates of bankfull discharge and channel dimensions at streamflowgaging sites and updated drainage areas from StreamStats were incorporated into previously developed regional curves to produce an updated set of regression relations of bankfull discharge and channel geometry for the noncarbonate and carbonate settings of Pennsylvania and Maryland.