|
Activity Number:
|
263
|
|
Type:
|
Invited
|
|
Date/Time:
|
Tuesday, July 31, 2007 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
IMS
|
| Abstract - #307761 |
|
Title:
|
Models for Short-Term Wind Speed Prediction
|
|
Author(s):
|
Amanda S. Hering*+ and Marc Genton
|
|
Companies:
|
Texas A&M University and University of Geneva
|
|
Address:
|
3143 TAMU , College Station, TX, 77843-3143,
|
|
Keywords:
|
Spatial time series ; Prediction ; Circular variables ; Space-time simulation
|
|
Abstract:
|
The regime-switching space-time model proposed by Gneiting et al. (2006) used to predict average hourly wind speed at a wind energy farm works efficiently but is specific to its geographic location. The model is regime-dependent, changing based on whether the wind is blowing to the east or west. We eliminate the regimes and include wind direction as a covariate in the model. The two hour forecasts improve upon Gneiting et al.'s (2006) best model. Comparing the two models leads us to simulate both wind speed and wind direction over space and time. Wind direction must be treated as a circular variable, whose values are close when observations are near the endpoints of its range, and this introduces additional complexity to the simulation. Of particular interest are characteristics of the data that would make Gneiting et al.'s (2006) model superior to our and vice versa.
|
