Risk-based hydrology: baysesian flood-frequency analyses using paleoflood information and data uncertainties

摘要

It is essential, and not particularly difficult, to rigorously incorporate observed data uncertainties in flood-rrequency analyses, "Best fitting" (median, method of monents, maximum likelihood, etc.) flood-frequency distributions, particularly from analyses that do not explicitly include data uncertainties. have infinitestimal total probability and are irrelevant for risk-based hydrology. In contrast, a Bayesian approach incorporates data and parameter uncertainities. Peak discharge probability regions are calcualted as a function of annual probability to quantify hydrologic hazards. Paleohydrologic bounds demonstrate that peak discharges of about 70,000 fr~3/S and 90,000 FT~3/S had not been exceeded on the Santa Ynez River in 700 years and 2900 years, respectively. These paleohydrologic bounds are combined with gage and historical data to estiamte flood-frequency probabilities. If only gage and/or historical data are available, an inescapable conculsion is that a flow exceeding spillway capacity (160,000 ft~3/s) is likely to occur in a 10,000 year period.