A climate-dependent statistical model of tropical cyclone hazard and risk

Tropical cyclones have a complex relationship with their environment and large-scale climate. As much uncertainty remains in answering how tropical cyclones will behave with climate change, creating tools to examine the tropical cyclone-climate relationship has become very important. Statistical models of tropical cyclones in the western North Pacific and North Atlantic basins are constructed based on observations for the 1970-2010 “present day” time period with the ultimate goal of landfall risk assessment. The genesis, track propagation, and intensity components of the model use local environmental variables as statistical predictors. The genesis component, which simulates annual storm count, initial location, and initial date depend on monthly sea surface temperature, potential intensity, vertical wind shear, and depth of the ocean mixed layer. The track propagation component, which determines the location of a storm every six hours, depends on mid-level daily winds and sea surface temperature. The intensity component models both minimum central pressure and maximum sustained wind speeds jointly, and it depends on persistence and the ventilation index. The models are used to simulate the 1970-2010 period in order to evaluate the model and establish the landfall rates along U.S. and East Asia coastlines for the present climate. The model is also used to simulate different El Nino/Southern Oscillation states using environmental conditions taken from historical composites. This method of statistical tropical cyclone track modeling is effective in capturing the predominant effects of climate and local environment on tropical cyclones, especially pertaining to landfall risk. Along with careful comparison to physics-based numerical models, it can be used as an observation-based tool to assess climate-related tropical cyclone risk.