Using GEOS-5 Forecast Products to Represent Aerosol Characteristics in Operational Day-Ahead Solar Irradiance Forecasts, for the Southwest United States

Utility companies need to meet the public energy demand. They need an accurate forecast of expected energy supply from solar panels to maximize the amount of renewable energy that goes into meeting the total demand. Active research in solar forecasting includes cloud location, cloud optical depth, and clear-sky aerosol optical depth. Aerosol Optical Depth (AOD) describes the transparency of the atmosphere. An accurate representation of AOD is important to the success of the radiation scheme within an atmospheric model to predict surface solar variables. This work aims to improve the way AOD is represented and then used in the WRF model, in Arizona, specifically to improve forecasts of surface solar irradiance. We find that using gridded GEOS-5 forecasts of AOD reduces mean forecast error for DNI in clear-sky conditions and reduces the bias. GEOS-5 aerosol forecasts better capture the spatiotemporal characteristics of aerosols compared to a point-measurement-based AOD forecast. To incorporate these findings into an operational forecasting configuration, we describe how we account for operational considerations such as GEOS-5 forecast latency and data outages.