Sensitivity of solar irradiance to model parameters in cloud and aerosol treatments of WRF-solar

We investigate the parametric sensitivity of solar irradiance to a set of parameters within the sub-grid cloud scheme (CLD3) and an upgraded aerosol-aware Thompson-Eidhammer scheme (TE14) in an upcoming enhanced version of the Weather Research and Forecasting-Solar model. We conduct ensemble simulations over the Southern Great Plains and Hanford, California, focusing on the parametric sensitivity under cloudy conditions with different aerosol loading. We adopt the Quasi-Monte Carlo sampling approach to explore the high-dimensional parameter space and apply the generalized linear model to quantify the relative contribution of individual parameters to the total variance. We find that CLD3 parameters related to entrainment and the cloud condensation threshold contribute to most of the variance in the ensemble simulations under less-polluted conditions. Larger entrainment and a higher condensation threshold produce more solar irradiance via decreasing cloud fraction and cloud optical depth. As the sensitivities of cloud fraction and cloud optical properties to perturbed parameters vary with cloud cover, the relative contributions of parameters to irradiance variance also change with cloud cover. As a contrast, under the heavily-polluted wildfire conditions, the model sensitivities to the parameters within the enhanced TE14 scheme become pronounced, especially those related to water-friendly aerosol emission rate and the modal radius of black carbon. The variations of those aerosol parameters cause significant changes in simulated irradiance, especially over the clear sky, suggesting the importance of accurate representation of emission source for aerosols and their precursors.