Simultaneous Multiscale Data Assimilation Using Scale‐ and Variable‐Dependent Localization in EnVar for Convection Allowing Analyses and Forecasts: Methodology and Experiments for a Tornadic Supercell

Abstract

AbstractThis study introduces a simultaneous multiscale data assimilation method by implementing model space spatial scale‐dependent localization (SDL) and variable‐dependent localization (VDL) within an ensemble variational system. This method updates all resolved scales by assimilating all observations at once. The impact of such an approach is examined by a series of radar data assimilation experiments. Single‐observation experiments show that SDL concurrently and more properly updates the storm and its ambient environments compared to a traditional single scale localization (SSL) for radar data assimilation. Including VDL on top of SDL (SDLVDL) realistically decreases the spatial coverage and intensity of moisture increments compared to SDL. Comparisons are then performed on the analyses and forecasts of the 8 May 2003 Oklahoma City supercell storm. Results show that SDL improves the analyses and forecasts during the data assimilation cycling by producing more realistic enhanced low‐level convergences than SSL. SDLVDL obtains more accurate analyses and subsequent forecasts for moisture than SDL. SDLVDL yields the best performance in reflectivity forecasts and storm maintenance. Compared to SSL, SDL has higher forecast skills before 2230 UTC and produces degraded forecasts in the later lead time.