Iterative assimilation of geostationary satellite observations in retrospective meteorological modeling for air quality studies

Clouds impact many aspects of both the physical and chemical atmosphere at many different spatial and temporal scales. Because of this, their accurate representation within numerical weather prediction (NWP) models is vital. In this study, a previously developed assimilation technique for assimilating Geostationary Operational Environmental Satellite (GOES) derived cloud fields into the Weather Research and Forecasting (WRF) meteorological model was tested over the August-September 2013 time period, on a 12-km domain covering the contiguous United States (CONUS). At the same time, additional improvements to the assimilation technique were introduced to account for the model cloud time tendency. The improvements resulted in more consistent cloud fields, while also improving the model surface statistics when compared to the original technique. The results indicate that both implementations of the assimilation technique improve the agreement between the model-predicted and GOES-derived cloud fields, but the additional refinements significantly improve the overall model performance. The daily average percentage increase in the cloud agreement was 6.53% for the original technique, compared to 11.27% for the refined technique. With the improvement in the model cloud fields, the average error in the predicted solar irradiance across the CONUS domain was reduced by 4.4 W m(-2) and 24.6 W m(-2) for the original and revised assimilation techniques, respectively. The revised assimilation technique also reduced the slight degradation in the surface statistics of wind speed, temperature, and mixing ratio that was created with the original technique. This resulted in surface error statistics that were nearly the same as a control simulation, but with improved model cloud performance.