The Effects of Digital Filter Initialization on Convection Initiation in a WRF-DART Mesoscale Ensemble Analysis

The digital filter initialization (DFI) has been shown to reduce high-order imbalances and inconsistencies among model variables. Using DFI to remove noise has been shown to improve the accuracy of forecasts in deterministic simulations. Imbalance is inherent to EnKF analysis states following each assimilation window. Various numerical filters available within WRF largely damp resultant gravity and sound wave noise during the first few hours of the simulation, however remnant noise may still degrade the subsequent analysis. Applying DFI is expected to reduce noise and potential negative impacts on subsequent analyses, with greater positive impact as the cycling frequency is increased. To test the impact of DFI on mesoscale cycling applications, continuously cycled EnKF data assimilation simulations using WRF-DART will be conducted over a CONUS domain during the period from 11-18 May 2010. Data assimilation update frequency will be varied from one, three, and six hours. Convection initiation seems to occur at a greater rate in the hours just after an assimilation period in continuously cycled EnKF data assimilation simulations. DFI enabled simulations will be compared to a no-DFI control simulation and to the different assimilation windows. Differences in analysis, convection initiation vigor after an assimilation cycle, and forecast skill of the two EnKF simulations will be presented.