Successful assimilation of satellite retrieved cloud properties is directly related to the cloud microphysical schemes used within a convection permitting model. To determine the sensitivity of assimilating satellite retrievals of cloud water path (CWP) as a function of cloud microphysics scheme, several experiments are conducted using both idealized and real-data cases. Both sets of experiments use the Advanced Research Weather and Research Forecast (WRF-ARW) model and a 40-member ensemble with the Data Assimilation Research Testbed (DART) ensemble Kalman filter. For the idealized simulation, a Truth run is generated using a deterministic WRF run combined with an homogeneous environment defined using a modified Weisman - Klemp sounding. Simulated CWP observations are generated from the Truth data for assimilation into the idealized experiments. Four cloud microphysics schemes are tested and include LIN, Thompson, WDM6, and Milbrandt and Yau.
For the idealized experiments, assimilating simulated CWP generates a mature supercell after approximately one hour for all microphysics schemes. However, the characteristics of the idealized convection differ. Each produces a different hydrometeor profile generally consistent with the design of that scheme. In particular, Thompson characterizes the storm-top as mostly snow whereas Milbrandt and Yau characteristics the same feature as ice.
The real data experiments use a severe weather event from 10 May 2010 in the Southern Plains, with the focus placed on early convection in northern Oklahoma and southern Kansas. One set of experiments assimilates conventional observations, while the second assimilates the identical conventional observations and the satellite retrievals. The non-CWP assimilating experiments all fail to properly capture a developing supercell, often generating spurious convection in the wrong place, especially WDM6. The CWP assimilating experiments performed much better, generally capturing the location and evolution of the storm. As with the idealized experiments, the distribution of the cloud hydrometeor variables differs significantly depending on the microphysics scheme used. The differences observed offer insights into the importance of selecting cloud microphysics schemes when assimilating satellite observations.