Sensitivity of Assimilating Cloud Water Path as a Function of Horizontal Localization Radius

Assimilating satellite data into numerical weather prediction (NWP) models is a potentially important step towards higher resolution data assimilation. Compared to other assimilation sources such as radar reflectivity, assimilation of satellite retrievals is not as thoroughly studied or understood. The focus of this research is to examine how the horizontal localization radius (LR) used for assimilation of GOES-13 cloud water path (CWP) retrievals affects model performance. Using an Advanced Research Weather Research and Forecasting Model (ARW-WRF) 40-member ensemble with 3km grid spacing, several experiments were run with different LRs from 5-50km used for data assimilation. The models were run using a 5 minute time step, with a 60 minute assimilation window and a 60 minute forecast window. Model output was compared against an idealized case for radar reflectivity, CWP, and water vapor mixing ratio. The idealized case spun up a storm by 30 minutes into the assimilation window, and maintained it through the end of the run. All of the experiments had the storm represented well initially, but translated it eastward too quickly and dissipated it too early, resulting in negative biases in the three variables by the end of the forecast cycle. In general, increasing LR resulted in a slight increase in the dry bias of the experiments, however this effect was small and was quickly outweighed by differences in the ensemble spread. Also examined was the effect of using different LRs for assimilating zero versus non-zero CWP retrievals, both for the idealized case as well as an actual severe weather event occurring in Oklahoma on May 24, 2011. Using smaller LR for assimilating zeros appears to cause the model to generate a larger area of low reflectivity precipitation.