Monte Carlo simulation of paleofloods: information content of paleoflood data in flood-frequency analysis

Abstract

Paleoflood hydrology is a technique for assessing the magnitude and frequency of floods in bedrock canyons. I calculate effects of adding paleoflood data to gaging data on estimating low return-period quantiles using peak discharge sample statistics from a bedrock-constrained river, the Salt River near Chrysotile, Arizona, as representative of hydrologic conditions in the southwestern United States since the late-Holocene. A Monte Carlo method was used to randomly generate 5,000 realizations of a 4000-year annual flood series from a log-Pearson type III distribution. Using a censored model of paleoflood deposition, I fit the combined paleoflood and gaging data with the expected moments algorithm and calculated model bias and precision. Compared to gaging data alone, paleoflood information enhances low-frequency quantile estimation even when incorporating bias representing differences between peak flood stage and the associated deposit height. Even if only the largest flood in 4,000 years is known, paleoflood data significantly improves flood-frequency analysis.