|
Activity Number:
|
113
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Monday, August 4, 2008 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Section on Bayesian Statistical Science
|
| Abstract - #301977 |
|
Title:
|
Bayesian Modeling of Wind Fields Using Surface Data Collected Over Land
|
|
Author(s):
|
Margaret Short*+
|
|
Companies:
|
University of Alaska Fairbanks
|
|
Address:
|
P.O. Box 750125, Fairbanks, AK, 99775,
|
|
Keywords:
|
wind field ; process convolution ; Markov chain Monte Carlo
|
|
Abstract:
|
We propose an approach to modeling wind field in which the error structure includes the type of instrumentation used to collect such data over land, namely anemometers (for wind speed) and vanes (for wind direction). Thus the model can handle both the periodicity of the wind direction and the non-negativity of the wind speed. The measurement error depends in part on the wind speed; this characteristic is incorporated in the model. We use a Bayesian process convolution approach, implemented via Markov chain Monte Carlo. We illustrate the model performance with surface wind data set from Fairbanks, Alaska. We present initial results incorporating topography in our model.
|
