Figures & data
Figure 1. Study area showing the upper Bow River watershed, and the rain gauge location on which the analysis in this paper is based.
Figure 2. Plots for selection of thresholds above which the approximation by a generalized Pareto distribution is reasonable. Top panel shows the mean excess plot; bottom panel displays the parameter stability plots. The plots are computed for data restricted to the high-level period from May to August.
Figure 3. Cluster maxima in excess of 2.95-m threshold based on the interval de-clustering method and blocking by years.
Figure 4. Sensitivity of the 200-year return level estimate 200 and upper bounds of 95% confidence intervals computed using the delta method (DM), parametric bootstrap and profile likelihood (PL) to the length of data record. Estimation is repeated for expanding record length starting in year 1915 until 2012; the threshold is set to 2.95 m.
Figure 5. Sensitivity of the 200-year return level estimate 200 and upper bounds of 95% confidence intervals computed using the delta method (DM) and profile likelihood (PL) to threshold used for declustering and model estimation. The complete data record from 1915 to 2012 is used in estimation.
