RAND Statistics Seminar Series

A Multivariate Nongaussian Bayesian Spatial Modeling Framework for Hurricane Surface Wind Fields

Presented by Montserrat Fuentes, North Carolina State University
Thursday, April 12, 2007
RAND Corporation, Santa Monica, CA
Please contact Nancee Inouye if you would like to attend this seminar.

Abstract

Modelling surface wind fields is essential for hurricane forecasting. A wind field gives the direction and velocity of the wind at any location in the vicinity of the hurricane. The numerical ocean models used to predict the storm surge for coastal areas rely heavily on wind field inputs. Currently, deterministic formulas are used to generate the wind fields for the storm surge model based on a few meteorological inputs such as the radius and central pressure of the storm. This paper develops a statistical framework that uses both the deterministic wind model and observed satellite and buoy wind data to improve estimates of the wind field. Since wind fields often display erratic behavior, such as sudden changes in time or space, the multivariate spatial model for the true wind field avoids assuming the wind fields are normally distributed. Instead, the multivariate spatial prior for the underlying wind field is nonGaussian, and an extension of the nonparametric Bayesian stick-breaking prior. The spatial stick-breaking prior represents the distribution at each location as a potentially infinite mixture of normals, with the mixture probabilities varying spatially according kernel functions. This nonparametric spatial model is shown to improve prediction compared the deterministic model alone and to usual Bayesian Kriging methods for the wind field of Hurricane Ivan in September, 2004.



Attending a Seminar

RAND Visitors are welcome to attend the statistics seminars, but must RSVP at least one day prior to the seminar. To ensure your attendance, contact Nancee Inouye at inouye@rand.org with your name, company or university affiliation, and national citizenship (for security purposes). Light refreshments will be served.

For directions to RAND see: http://www.rand.org/about/locations/santa-monica.html

Visitors must enter through the north-parking garage, which is accessible from Main Street in our new office building. Inform the attendant that you are there for the Statistics Seminar Series and you will be directed to the appropriate parking area. If there is no attendant present, use the intercom and tell them that you are here for the Statistics Seminar Series. After parking, follow the instructions to the appropriate conference area. (1776 Main Street)

Reminder: the old RAND surface parking lots have been permanently closed.

For further information and to be added to the mailing list contact Nancee Inouye at inouye@rand.org.