Forecast evaluation of an observing system simulation experiment assimilating both radar and satellite data

In the first part of this study, Jones et al. compared the relative skill of assimilating simulated radar reflectivity and radial velocity observations and satellite 6.95-μm brightness temperatures T_B and found that both improved analyses of water vapor and cloud hydrometeor variables for a cool-season, high-impact weather event across the central United States. In this study, the authors examine the impact of the observations on 1-3-h forecasts and provide additional analysis of the relationship between simulated satellite and radar data observations to various water vapor and cloud hydrometeor variables. Correlation statistics showed that the radar and satellite observations are sensitive to different variables. Assimilating 6.95-μm T_B primarily improved the atmospheric water vapor and frozen cloud hydrometeor variables such as ice and snow. Radar reflectivity proved more effective in both the lower and midtroposphere with the best results observed for rainwater, graupel, and snow. The impacts of assimilating both datasets decrease rapidly as a function of forecast time. By 1 h, the effects of satellite data become small on forecast cloud hydrometeor values, though it remains useful for atmospheric water vapor. The impacts of radar data last somewhat longer, sometimes up to 3 h, but also display a large decrease in effectiveness by 1 h. Generally, assimilating both satellite and radar data simultaneously generates the best analysis and forecast for most cloud hydrometeor variables.