Estimation of the compound hazard severity of tropical cyclones over coastal China during 1949-2011 with copula function

In China, the intensity of a tropical cyclone (TC) is officially classified into six scales based on the 2-min sustained winds. However, the destructive potential of a TC is determined not only by winds, but also by other hazards such as rainfall. Therefore, the integration of the effects of all the various hazards to reflect the compound severity of a TC is of great importance. In this paper, the maximum wind speed (MWS) at landfall and total precipitation volume (TP) over land in China are obtained as two representative hazards. Copulas, having the capability to construct joint probability distributions of two or more random variables with an unidentified dependence structure among the variables, are utilized to construct the dependence of TC wind and rainfall. Firstly, the cumulative probability distribution functions (CDFs) are fitted separately based on four probability models with 218 historical TC records from 1985 to 2011 over China. Three copula functions are employed to construct the joint probability using the best-fitted marginal CDFs. The best probability functions are determined according to the OLS value. Secondly, the univariate return periods of MWS and TP, and two types of joint return period ( and ) are calculated. We found that the univariate return periods are discrepant for individual hazards and neither is a proper indicator for TC compound hazard severity. The joint return period is found to be a better indicator of the compound hazard severity of TCs. Finally, the economic losses of 178 historical TCs from 1985 to 2010 are normalized by eliminating the impacts of inflation, increased assets, and population. We find that the joint return period has a better capacity for representing TC-induced loss. This study highlights the improved capacity of multi-hazard joint return period over univariate return periods for representing the compound hazard severity and damage.