Determining an Ideal Vertical Localization for Cloud Water Path Assimilation within WRF-DART

Assimilating satellite-retrieved cloud water path (CWP) data representing the column-integrated cloud water content into a numerical weather prediction (NWP) model has already shown to improve the model's precipitation analysis and forecast accuracy. As satellite technology advances, it becomes crucial to reform current methods of satellite data assimilation to maximize model accuracy and performance during convective events. This study attempts to further improve convective scale modeling by defining a vertical coordinate within the NWP model as the satellite-derived height of the center of a cloud to test how changing the vertical localization radius of influence impacts the CWP data assimilation and subsequent model accuracy. Several ensemble data assimilation and prediction trials of varying vertical localizations are conducted to assimilate CWP from Geostationary Operational Environmental Satellite (GOES) observations using the Advanced Research Weather Research and Forecasting (WRF-ARW) model with the Data Assimilation Research Testbed (DART) Ensemble Adjusted Kalman filter (EAKF) system. A “nature” run from which synthetic satellite data generated from an idealized convection is considered the “Truth” run and is used to verify the success of the other “Trial” assimilation runs that have vertical localizations of 3, 6, or 12 kilometers or an undefined value. The outputs of these vertically-defined WRF-DART trials, along with an additional trial left undefined, are compared against Truth to test the accuracy of assimilation of the respective vertical localization. Through model output comparisons of each Trial against Truth, assessments of the performance of each localization trial are made. Graphical and statistical analyses strongly indicate that WRF-DART assimilation trials with a defined vertical localization perform better than a vertically undefined model, with the smallest, 3-kilometer localization being the most accurate. These results support hopes that refining satellite data assimilation techniques can continue to advance storm convection modeling and forecasting.