Solar energy forecasting

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner
Wednesday, 26 January 2011: 8:45 AM
Solar energy forecasting
309 (Washington State Convention Center)
Betsy Weatherhead, Univ. of Colorado, Boulder, CO; and C. S. Long

The renewable energy communities are rapidly increasing the amount of installations for solar renewable energy with an increase of at least 50% each year since 2003. Even more solar installations are planned for the coming years. As the amount of energy derived from the sun increases, the demand for a more quantitative understanding of solar renewable energy production increases. Climatological understanding of solar radiation levels around the country have been produced by a number of sources, these are summarized and an updated approach will be presented. The uncertainties are assessed, including difficulties with the sparseness of ground-based measurements, and the difficulties with assessing cloud transmission. A better understanding of long-term climatologies will be useful for planning new installations and estimating future energy production from different installation configurations. Short-term (0-5 day) forecasts can also be produced that will have direct impact on successfully incorporating solar radiation into grids for energy use and distribution. For forecasting, significant lessons can be learned from the NWS development of the UV index to forecast 24-36 hour solar UV (Ultra-Violet radiation) levels. As with the UV forecasts, the uncertainties of solar renewable energy forecasts will depend on available data and on the accuracy of the forecasting model. Model sensitivities will be presented and a discussion of their impact on solar forecasting will be illustrated with a five day example from the southeastern US. Identifying better models for solar forecasting will rely on comparing small differences from a suite of different models. Choosing a model that will work well for solar energy may be different from choosing a model that will work well for more standard meteorological forecasts including temperature, winds and relative humidity. Data from the Weather Research and Forecasting (WRF) model, the Regional Atmospheric Modeling System (RAMS) and the Non-hydrostatic Mesoscale Model (NMM) will be examined.