Abstract
The coincidence of fluvial and coastal flooding can lead to compound
floods with substantial impacts on human life, property, and
infrastructure. Low-lying coastal areas are particularly vulnerable to
compound flooding because of exposure to multiple drivers such as
extreme coastal high tides, storm surge, and fluvial flooding. In this
study, we develop a bivariate non-stationary flood risk assessment that
accounts for compound flooding from fluvial and coastal events with
consideration of impacts of sea level rise (SLR). Extreme river
discharge values were identified using peak over threshold method and
were paired with the corresponding highest sea-water level within ±1
days of these events across the coastal contiguous United States. The
statistical dependence between the paired data was assessed using
Kendall’s rank correlation coefficient. For the locations with
significant dependence, the best copula fit was used for bivariate
dependence analysis by assuming non-stationarity in the marginal
distribution of sea-water level data. The mixture Normal-Generalized
Pareto Distribution model with SLR as the covariate is used to
incorporate the non-stationary coastal flood frequency. The future risk
was assessed using the notation of failure probability, which refers to
the probability of occurrence of at least one major coastal flooding
(i.e., water level exceed the major coastal flood threshold) or 100-year
fluvial flood for a given design life. Failure probability was
formulated to allow for changing exceeding probabilities over time. The
results indicate that the joint exceedance probability of fluvial or
coastal flooding can be higher when the dependence is considered.
Ignoring the compounding effects may inappropriately underestimate the
flood probability at locations that flood hazard can be influenced by
the interaction of fluvial and coastal events. Moreover, with rising sea
levels, the probability of exceedances of sea-water level over the flood
threshold increases and consequently the compound flood probability
increases as well. In the locations with less dependency between extreme
river discharge and sea-water level, the frequency amplification of
fluvial and major coastal flood events is higher.