Extreme value analysis in typhoon prone areas: case study of the Pearl River estuary

Emiel Moerman, Reimer de Graaff, Joao de Lima Rego, Deepak Vatvani

Thursday 2 july 2015

16:15 - 16:30h at North America (level 0)

Themes: (T) Extreme events, natural variability and climate change, (ST) Learning from disasters

Parallel session: 13I. Extreme events - Lessons Disaster

Extreme events such as tropical storms and typhoons are often the determining factor for the extreme values of wind, wave and water level conditions. The storm track, its propagation speed, the air pressure drop and the wind speed intensity of a typhoon determine the maximum occurring wave heights, water levels and currents. The stochastic behaviour of typhoons and tropical storms, however, lead to uncertainty in the extreme value analysis, because a slight variation of the typhoon track, propagation speed or wind speed intensity can have a significant impact on these local extreme hydrodynamic conditions. To determine the significance of the stochastic behaviour of typhoons a model assessment is performed comparing standard extreme value analysis values of measured water levels (e.g. values of 1/10, 1/50, and 1/100 year return periods) against model results of artificial typhoons. In the model assessment, making use of Delft3D, various artificial typhoons are modelled in which the typhoon tracks, propagation speeds and wind speed intensities are varied within realistic ranges (based on observed historical typhoons). The study focusses on the Pearl River estuary (China) where typically about 5 to 10 tropical storms or typhoons are observed every year. Once every few years an extreme typhoon hits the area. By quantifying the potential impact of artificial typhoons the uncertainties in the extreme water level values in such a typhoon prone area are better assessed. The model is validated simulating several historic typhoons. Subsequently the typhoons tracks, their propagation speeds and wind speed intensities are varied. The extreme water level values (extreme surge height + mean high water value) that follow from the artificial typhoon modelling are compared against values from a standard extreme value analysis, making use of the central limit theorem for the extreme values in a sample. A Peaks over Threshold approach is applied and the extremes are fitted and extrapolated according to a Generalized Pareto Distribution. One of our main conclusions is that while the peak surge heights and related total water levels resulting from the historical and synthetic typhoon simulations can exceed the once per 100 year extreme total water level estimates, they are generally within the 95% confidence interval of the estimate.