5.2
Short Term Load Forecasting with a Meteorologist

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner
Tuesday, 4 February 2014: 8:45 AM
Room C114 (The Georgia World Congress Center )
W. Jeff House, Tennessee Valley Authority, Chattanooga, TN

Short term load forecast accuracy can improve with a meteorologist delivering the load forecast. Industry average short term load forecasting error is between 3-5%. Utilities with a meteorologist delivering the short term load forecast average 2-3% error. Often utility meteorologists deliver temperature and other meteorological inputs for the short term load forecast, improving load forecast accuracy. However the in house meteorologist can add even more value delivering the final load forecast. Vendor and even proprietary load forecasts weigh temperatures most, but other weather variables can have disproportionately large impacts on days with unseasonable weather. The operational meteorologist can discern these events before models adjust. Precipitation is challenging for both temperature models (MOS) and the load forecasting models. A meteorologist using pattern recognition can often improve the precipitation and cloud cover forecast for use in load forecasting. Also when MOS struggles, correcting MOS and other variable may still not be enough to yield an accurate load forecast. After meteorological input is adjusted, the meteorologist can add the final improvements to the load forecast. Examples requiring a higher load forecast include arctic fronts under cloudy skies in winter, cold air damming, and quickly recovering hot air masses in summer. Examples requiring a lower load forecast include MCS thunderstorm complexes in summer and return flow or black stratus at night in winter. Improvements in the load forecast are most significant on the outlier days. Significant improvement in the short term load forecast is accomplished using a meteorologist to deliver the load forecast.