Nowcasting Case Studies with SunCast

As more utility-scale and distributed solar energy installations come online, incorporating the highly variable power output from these solar farms into the grid is a growing challenge for energy utilities. In order to obtain reliable predictions of the power output from a given solar farm, accurate predictions of the irradiance are essential. In addition to day-ahead forecasts for energy trading and unit commitment, grid operators require nowcasts of irradiance and power, from minutes to hours ahead, to maintain a balanced electrical grid.

The SunCast solar energy forecasting system is designed to meet these challenges. Several forecasting components predict 15-minute average global horizontal irradiance (GHI) over a range of nowcasting time horizons. These components include site-based statistical learning methods (StatCast), satellite cloud advection forecasting (CIRACast), multi-sensor advective diffusion forecasting (MADCast), and the WRF-Solar numerical weather prediction model. The goal of SunCast is to blend these systems into a single solar energy forecasting product.

In this study we compare the performance of SunCast component forecasts for different sky cover regimes near Sacramento, including marine stratocumulus, which is a notoriously difficult forecasting problem for utilities on the West Coast. We also show better GHI forecasts for these case studies as the component systems were continually improved over the past year.