Scope: 

In this work an efficient computational framework is developed for evaluation of hurricane risk for the Hawaiian Islands with particular emphasis on online risk estimation. The foundation of this framework is the parameterization of each hurricane scenario by five model parameters, (i) the location of landfall x_{o}; (ii) the angle of impact at landfall θ; (iii) the central pressure c_{p}; (iv) the forward speed during final approach to shore v_{f} ; and (v) the radius of maximum winds R_{m}. These variables ultimately constitute the model parameters vector, x, describing each hurricane scenario. Description of the uncertainty in these parameters, through appropriate probability models, leads then to quantification of hurricane risk as a probabilistic integral. Output quantities for interest for the risk quantification include the (i) the still water level (SWL), defined as the sea level in absence of wind waves, the (ii) wave breakup level (WBL), defined as the sea level including breakup of wind waves on shore, or (iii) the significant wave height H_{s} along with the corresponding period of wave oscillation T_{s}.
One of the greater advances in this field has been the development and adoption of high fidelity numerical simulation models for reliable and accurate prediction of hurricane impact for a specific event. This approach significantly increases, though, the computational cost for estimating risk since this estimation requires evaluation of the surge response for a large number of hurricane scenarios, representing the entire space for possible hazard uncertainties. For efficiently performing this task an approach based on response surface methodologies is developed: using information from a small amount of precomputed highfidelity numerical simulation, response surfaces are built as a surrogate model for efficiently predicting surge responses and these surfaces are then used for efficient risk assessment. A moving least squares response surface methodology is considered in particular for this purpose.


Fast online estimation of hurricane risk: 

During an approaching hurricane event, the anticipated hurricane track and intensity as well as the uncertainty in these prediction (probabilistic description) is characteristics is ultimately provided by the National Weather Service (NWS). This leads to a probabilistic integral quantifying hurricane risk, which can then be online estimated at a small computational cost using the developed surrogate model.
For implementation of such a risk estimation, a standalone executable is developed as a risk assessment tool. The graphical user interface of version of this tool is shown below. The tool accepts as input the parametric configuration for the most probable hurricane track as well as the estimate for time till landfall, used in the current version to select the probability models for the hurricane characteristics (based on standard estimation errors for the NWS predictions). Based on this input and the precomputed information from the high fidelity simulations, the surrogate response surface approximation is used to predict either the output for the most probable hurricane or the hurricane risk, estimated as the threshold with a prespecified probability of exceedance. The outputs from the risk estimation are graphically presented as contours for the SWL and WBL around the island. The total time needed for the tool to provide the required output (analyzing 2000 potential hurricane scenarios) is less than 2 min on a 3.2 GHz single core processor with 2 GB of RAM, which illustrates that it can be efficiently used for online risk evaluation.




Publications: 


This material is based in part upon work supported by the Army Corps of Engineers under grant number W912HZ09C0086. This support is greatly appreciated. Any opinions, findings and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the Army Corps of Engineers.