Accounting for the Effects of Unresolved Clouds in the Shortwave Irradiance Forecast of the WRF-Solar Model to Improve Solar Power Forecasts

Numerical weather prediction (NWP) is a critical component of solar power forecasts; yet until recently, NWP was not optimized for forecasting for solar power, specifically for solar irradiance. We demonstrate a specific upgrade to an NWP model targeted at improving solar power forecasts.

WRF-Solar is an upgraded version of the Weather Research and Forecasting (WRF) model aimed at improving solar power forecasting that provides a better representation of the aerosol-cloud-radiation feedbacks. An important upgrade simulates the effects that the unresolved clouds generated by both deep and shallow convection produce on the shortwave irradiance. This presentation will examine the importance of accounting for the effects of the unresolved clouds using a set of WRF-Solar experiments with different representation of the sub-grid scale clouds. Given the high spatio-temporal variability of the clouds, each experiment consists of a 10 member WRF-Solar ensemble run at 9 km of horizontal resolution over the contiguous U.S. (CONUS) using the Stochastic Kinetic-Energy Backscatter Scheme (SKEBS). The numerical experiments span the year of 2014 and are compared against high quality observations of the global horizontal irradiance (GHI). Results indicate the necessity of including the effects of the unresolved clouds to largely suppress a positive bias (~ 50 W/m2) in the GHI simulation. The shallow cumulus feedbacks to radiation are responsible for the largest bias reduction. Correcting this bias using improved physics is shown to provide a better solar irradiance forecast at solar plants, which in turn produces a better power forecast