Analysis of Risk due to Compound Extreme Events in the Coastal Urban Environment

High-impact weather events frequently depend on a number of physical factors and interactions among processes that may vary over space scales that range from global to mesoscale and time scales that vary from subseasonal to hours. Often, individual physical factors that are not extreme individually can lead to an extreme event when combined. Such compound extreme events owe their existence to the multivariate nature of physical processes and a statistical dependence among variables. An example of compound, connected events is the occurrence of large-scale rain over a coastal region via a predecessor rain event (PRE) days ahead of the passage of a weakening tropical cyclone (TC). Cases of combined PRE- and TC-induced rainfall that led to extreme flooding include Hurricane Floyd (1999) and Hurricane Irene (2011) over the northeastern United States. Although the PRE rainfall alone or the TC-induced storm surge alone may not lead to an extreme flood event, the connection of the surge that prevents efficient discharge of the PRE rainfall may lead to widespread flooding. During Hurricane Harvey (2017) an extreme event triggered by a period of prolonged intense rainfall over the Houston metropolitan area was made more extreme due to the compound influence of a large geographic region of coastal storm surge, which was not extreme in itself but contributed to reduced discharge of water over the inland area.

Often hazard risk assessment efforts focus on one driving factor rather than the compound nature of multiple forcing mechanisms. It is certainly possible that extreme event analysis in a univariate context may underestimate the probabilities of extreme events associated with impacts to lives, properties, and economies. In this study, a framework is defined that addresses multivariate analysis of risk due to compound hazards. A combination of observations and model simulations is used to identify the changes in expected frequencies of annual chance flood events due to storm surge and overland precipitation often forced via connected mesoscale events. The framework emphasizes the statistical dependence of multiple physical variables that contribute to the extreme compound event when individual events are not extreme.

Finally, an amplification factor is defined as a joint function of frequency distributions of individual events. Amplification is defined as change in the expected frequency of a chance flood over an annual time period due to compound events. Additionally, a simple assessment of amplification due to changes in factors expected from standard climate change scenarios is provided.