Analysis of Low Streamflow Characteristics: Application to Some Canadian Hydrometric Data

The “deficit-below-threshold” (DBT) method is applied to characterize low flows in terms of volume, duration and intensity at 101 Canadian hydrometric stations. The volume, duration and intensity data series are fitted to six 2-parameter distributions, which are the generalized Pareto (gp), gamma, Weibull, log-logistic (llog), lognormal (lnorm), and kappa. The goodness of fit is assessed using the sample’s approximate transformation to normality followed by application of the Shapiro-Wilk goodness-of-fit (GoF) statistic. The gp, kappa, Weibull, and gamma distributions provide the best fits to intensity, while for volume, the lnorm, llog, and kappa distributions give the best fit. For duration, none of the 2-parameter chosen distributions is flexible enough to provide an adequate fit.