An Evaluation of Nine ARW-WRF Microphysics Schemes for Solar Power Forecast in Arizona

Our work aims to improve the accuracy of the University of Arizona operational solar energy forecast by evaluating the performance of nine microphysics schemes, using a 1.8 km-horizontal grid spacing and convective-permitting WRF-ARW model configuration.  Surface solar irradiance forecast is largely influenced by cloud structure, dynamics, and radiative transfer that are modulated by cloud microphysical processes in the model.  We studied 20 days with subjectively good initializations under deep convection, high, middle, and low cloud scenarios.  We assessed the accuracy of the WRF irradiance forecasts by comparing them to irradiance derived from GOES visible and infrared channels, as well as a network of surface observations.  The Critical Success Index was computed as an objective measure to access the forecast skill of the individual microphysics schemes and the ensemble mean.  The microphysical parameterizations evaluated are Goddard, Milbrandt-Yau, Morrison, CAM v5.1, SBU-YLin, WRF Double-Moment 5- and 6- class, Thompson, and Thompson aerosol-aware.  The results suggest a unique, superior double-moment bulk microphysics scheme does not exist for the Southwest United States, and most parameterizations tend to overpredict global horizontal irradiance under all scenarios.  The evaluation will provide guidance to operational solar energy forecast service on identifying the optimal microphysics option.