A blended technique utilizing observations and retrievals of cloud properties from geostationary satellite platforms combined with high-resolution mesoscale model data is applied to short-term (0-6hr) forecasts of surface insolation for solar power generation.

GOES observations of the continental United States (CONUS), combined with cloud property retrievals including cloud type and cloud top height are used to create a database of cloud cover by cloud steering height. Combined with forecast model wind guidance, satellite-observed clouds can be advected forward in time by model winds to give a short-term forecast of cloud location. This information, when combined with a radiative transfer model, and accounting for variables such as solar geometry, can be used to provide an improved surface insolation forecast over current empirical methods.

Additionally, model output from the High-Resolution Rapid Refresh (HRRR) model is being evaluated for potential use as an insolation forecast model. Mesoscale models utilizing real-time platforms such as the WSR-88D for initialization show promise in forecasting cloud position and therefore insolation. Blending these model forecasts with satellite-derived cloud advection forecasts promises to greatly improve the accuracy and timeliness of insolation forecasts for the growing solar power industry.