Statistical-deterministic approach to natural disaster prediction

• US 20070168155A1
• Filed: 03/23/2006
• Published: 07/19/2007
• Est. Priority Date: 01/13/2006
• Status: Active Grant

• Current Assignee: Kerry Emanuel, Sai Ravela
• Original Assignee: Kerry Emanuel, Sai Ravela
• Inventors: Ravela, Sai, Emanuel, Kerry
• Granted Patent: US 7,734,245 B2
• Time in Patent Office: Days
• US Class Current: 702/179
• CPC Class Codes: G01W 1/10 Devices for predicting weather conditions computers per se G06; display devices G09
• CPC Class Codes: G06Q 10/04 Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem" market predictions or forecasting for commercial activities G06Q30/0202
• CPC Class Codes: G06Q 50/26 Government or public services business processes related to the transportation industry G06Q50/40
• CPC Class Codes: Y10S 706/93 Environment
• CPC Class Codes: Y10S 706/931 Weather

Abstract

A combined statistical-deterministic approach to methods and systems for assessing risk associated with natural disasters, in particular, hurricane wind risk. One example of a method of predicting wind speed distribution within a predetermined distance from a point of interest includes steps of statistically synthesizing a large plurality of wind storm tracks that pass within a predetermined radius of the point of interest, running a deterministic simulation of wind intensity along each one of the large plurality of wind storm tracks to produce an output representative of wind speed distribution along each track, and using the output to estimate an overall wind speed probability distribution from a combination of the wind speed distributions along each track within the predetermined distance from the point of interest.

Claims

1. A method of predicting wind speed distribution within a predetermined distance from a point of interest, the method comprising steps of:

statistically synthesizing a large plurality of wind storm tracks that pass within a predetermined radius of the point of interest;

running a deterministic simulation of wind intensity along each one of the large plurality of wind storm tracks to produce an output representative of wind speed distribution along each track; and

using the output to estimate an overall wind speed probability distribution from a combination of the wind speed distributions along each track within the predetermined distance from the point of interest.

2. The method of claim 1, wherein the step of synthesizing the large plurality of wind storm tracks includes:

originating the large plurality of storm tracks by randomly selecting a corresponding large plurality of genesis points from a space-time probability density function of genesis locations based on at least one of historical wind storm data and simulated climate data.

3. The method of claim 2, wherein the step of originating includes:

constructing a space-time grid over an area of interest; and

counting a number of wind storms originating within each cell of the space-time grid based on at least one of the historical wind storm data and the simulated climate data to provide estimated distribution of genesis points for the large plurality of wind storm tracks within the area of interest.

4. The method of claim 3, wherein the step or originating further includes smoothing the estimated distribution of genesis points using a spatially decaying window function to create the space-time probability density function of genesis locations.

5. The method of claim 4, wherein the spatially decaying window function includes a three-dimensional Gaussian kernel.

6. The method of claim 1, wherein the step of statistically synthesizing the large plurality of wind storm tracks further includes:

generating each of the large plurality of wind storm tracks using a Markov chain constructed with statistical features of historical or simulated wind storm tracks.

7. The method of claim 6, wherein state variables of the Markov chain include speed and direction of the large plurality of wind storm tracks.

8. The method of claim 6, wherein the state variables of the Markov chain include zonal and meridianal components of the large plurality of wind storm tracks.

9. The method of claim 1, wherein the step of statistically synthesizing the large plurality of wind storm tracks includes:

synthesizing a time series of wind at at least two pressure levels including a first pressure level and a second pressure level; and

modeling wind storm motion as a weighted average of air flow at the at least two pressure levels plus a beta drift correction.

10. The method of claim 9, wherein the at least two pressure levels include a 250 hPa pressure level and an 850 hPa pressure level.

11. The method of claim 9, wherein synthesizing the time series includes modeled the time series of wind as a Fourier series with randomly varying phase, with means and variances constrained by one of observed and modeled climate data, and with a preset kinetic energy spectrum.

12. The method of claim 1, wherein the step of statistically synthesizing the large plurality of wind storm tracks, each wind storm track having characteristics that conform to statistics derived from at least one of historical wind storm data and simulated wind storm data.

13. The method of claim 1, wherein the step of running a deterministic simulation of wind intensity along each one of the large plurality of wind storm track includes:

modeling the wind storm intensity using a model phrased in angular momentum coordinates and initialized using environmental thermodynamic parameters and wind shear.

14. A computer readable medium encoded with a plurality for instructions adapted to be executed on at least one processor to perform the method of claim 1.

15. A data set comprising a probabilistic wind speed distribution within the predetermined radius of the point of interest, the data set being generated according to the method of claim 1.

16. A method of generating a data set corresponding to a probabilistic wind speed distribution within a predetermined radius of a point of interest, the method comprising:

statistically generating a large plurality of synthetic wind storm tracks that pass within the predetermined radius of the point of interest;

modeling an intensity of a synthetic wind storm at a plurality of points along each one of the large plurality of synthetic wind storm tracks;

based on the intensity modeling step, generating, for each one of the large plurality of synthetic wind storm tracks, a probabilistic wind speed distribution at selected points along the synthetic wind storm track;

producing the data set by combining the wind speed distributions for each of the large plurality of synthetic wind storm tracks within the predetermined radius of the point of interest.

17. A method of estimating long-term risk associated with wind speed within an area of interest, the method comprising:

modeling an intensity of a synthetic wind storm at a plurality of points along each one of a large plurality of statistically generated synthetic wind storm tracks, together with a parametric or modeled wind distribution to produce an estimate of wind speed distribution associated with each track;

generating a probabilistically weighted combination of the estimated wind speed distributions along each track to create a probabilistic wind speed distribution within the area of interest; and

estimating a risk of wind speed within the are of interest exceeding a predetermined threshold value based on the probabilistic wind speed distribution.