For locations that could be influenced by multiple flood sources, considering the probability of flooding from each source individually does not provide a full understanding of the likelihood of a given flood depth occurring. Current methods of calculating the probabilities of flooding from multiple flood sources typically make the assumption that the flood sources are independent of each other. This research developed a method of calculating the probabilities of flooding for a location of interest based on multiple flood sources when a dependent relationship existed between those flood sources. Joint distributions were developed to describe the relationship between the flood sources using Archimedean copulas. Then, probabilities that correspond to total flood depths were calculated by taking integrals under the joint pdfs determined based on the copula equations. This process was carried out for both two flood sources and three flood sources, using both simulated and observed data. Zero-flood years were found to significantly impact the fitting of marginal and joint distributions in this process.

A common assumption is that the flood sources are independent of each other. When a relationship between the flood sources is evident, the assumption of dependence or independence was observed to significantly impact the calculated exceedance probabilities. Flood risk estimates were made based on probabilities calculated while considering all flood sources, to determine the impact that this new flood hazard assessment would have on flood risk assessments. Finally, the expected error in the results of the flood hazard assessment was assessed through an uncertainty analysis. This analysis focused on the parameters used to calculate the flood depths from the observed gage measurements, which were primarily related to the physical characteristics of the location of interest, and on the distributions used to represent the total flood depths. A high level of uncertainty was found to exist in the results of the analyses, indicating the importance of using the most accurate information possible to determine the parameter values and of using care in selecting the best distribution to represent the total flood depths.