TJ10.6 Subseasonal predictability of regional wind power generation

Monday, 7 January 2013: 5:15 PM
Room 6A (Austin Convention Center)
Jake Mittelman, Georgia Institute of Techonology, Atlanta, GA; and J. A. Curry, K. Shrestha, and J. Belanger

At intermediate time scales of weeks to months, energy trading (oil and gas) is increasingly influenced by anticipation of future wind generation in regions such as Texas where there is a large penetration of wind energy into the overall supply. Wind power forecast information on subseasonal time scales of weeks to months can help stabilize energy costs and supply from oil and gas. The predictability of regional wind power generation on subseasonal time scales is investigated for the period 8/11 – 7/12 using the ECMWF Monthly Forecasting System, which produces ensemble forecasts out to 32 days (resolution ~50 km) of 100 m winds. We analyze wind forecasts in weather zones affecting the Electric Reliability Council of Texas (ERCOT) grid. Uncalibrated forecasts of 100 m winds and wind power generation are assessed against the ECMWF operational analyses to assess the predictability of the weather variability that contributes to wind power anomalies. The monthly forecast products are calibrated using higher resolution products from the ECMWF daily forecasts that have been calibrated using SODAR data and regional power generation records. Predictability of 5 day integral wind power generation is assessed using bounding boxes, Brier Skill Score, Reliability Diagrams, and Relative Operating Curves, and Rank Histograms. Additionally, a covarying analysis of wind power generation with heat wave and cold outbreak periods assesses the wind power predictability during periods of peak need.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner