Abstract:
The objective analysis of the frequency and risk of heavy rainfall can provide scientific basis for disaster prevention and mitigation planning and disaster warning. In this study, the Copula methodology is applied to establish a bivariate Copula distribution model for heavy rainfall variables, i.e. the duration and accumulated precipitation, based on hourly rainfall data collected at 18 rain-gauge stations in Beijing during the period from 2005 to 2014. The conditional return period is then calculated and the risk of heavy rainfall is analyzed. These analyses demonstrate that the duration of heavy rainfall is less than 24 h in Beijing, which mainly follows the generalized extreme value (GEV) or logarithmic normal (LN) distributions, while accumulated precipitation is fitted to the GEV distribution in most stations. The Gumbel Copula is identified to be appropriate for describing the relationship between rainfall duration and accumulated precipitation for most gauges. The return period of short-duration rainfall is significantly influenced by the factor of duration in Beijing, which will result in the underestimation of heavy rainfall risk when only the factor of precipitation is considered. However, it is found that the conditional return period based on the bivariate Copula can comprehensively describe the property of heavy rainfall frequency and corresponding hazard risk. The heavy rainfall events with a duration less than 12 h and an accumulated precipitation of more than 50 mm often occur in an area extending from the northeast to southwest of Beijing, while the heavy rainfall events with duration less than 6 h and an accumulated precipitation of more than 50 mm occur more frequently in urban area and the northeastern area of Beijing.