Radar Reflectivity-based Initialization of Precipitation Systems Using A Diabatic Digital Filter Within The Rapid Update Cycle

Abstract

Despite a significant research effort over the past two decades, the prediction of convective storms and the associated warm season precipitation prediction problem remains a formidable modeling and assimilation challenge. The large forecast uncertainty associated with convective situations, even at very short lead times, coupled with the severity of weather often associated with convective storms, makes this perhaps the most significant short-range forecast challenge confronting the operational numerical weather prediction community. As an example, the commercial aviation industry is particularly vulnerable to convective storms, with resulting flight delays and diversions spiking every summer. The challenge of accurately predicting convective storms is complex and includes both the convective initiation problem and the convective evolution problem. For both of these problems, a very accurate prediction of mesoscale environmental fields, including temperature, moisture, and winds near the surface and at midlevels is needed. For midday summer conditions, in which positive convective available potential energy exists over a large portion of the United States, the accurate analysis and prediction of small capping inversions and weak forcing mechanisms is crucial for the convective initiation problem. Surface observations, wind-profilers and increasingly aircraft observations all play a key role in providing the crucial asynoptic information needed to improve short-range forecasts of mesoscale convective environments. At night, the problem is more difficult (convective forecast skill is generally even lower), as convection is often rooted above the surface layer, decreasing the utility of surface observations. Once convection is ongoing, the national network of WSR-8DD radars provides an invaluable set of observations. While these observations have greatly improved operational thunderstorm warning and nowcasting and a great deal of research on how to use them in model initialization procedures has occurred, the operational modeling community has been slow to utilize radar observations in operational models. Factors that have played a role in this are difficulties in providing a real-time feed of the voluminous radar data to operational centers (now largely solved), difficulties in using the two primary radar fields within numerical models, and a mismatch in scales between the highly detailed radar fields and the scales analyzed in operational models. With respect to the problem of the Dopplerradar observed fields, reflectivity has a complex underdetermined relationship with various precipitation hydrometers. Worse yet, many operational models do not include even simple 5- class prognostic microphysics schemes. The main other field from the Doppler radar, radial velocity, is but a single component of the wind. Furthermore, the radial velocity observations typically only occur in small patches and represent a scale of air motion that the operational models often cannot resolve. Because most operational models are still run at horizontal grid resolutions that preclude the explicit representation of individual convective elements, a cumulus parameterization scheme is required. While much of the current convective modeling research effort has focused on highresolution limited domain experimental forecasts that explicitly resolve convective storms, it is important to remember that for at least the next few years, operational convective modeling improvement will necessarily depend on improvements to data assimilation and modeling systems that employ cumulus parameterizations. With that reality in mind, a new approach for using a national mosaic radar reflectivity data to initialize the Rapid Update Cycle (RUC) model has been developed, and is currently being testing in real-time at NOAA/ESRL/GSD. In this paper, we briefly describe the RUC system and its new radar assimilation procedure (section 2), illustrate its application for a simple test case (section 3), present some preliminary results from our ongoing real-time tests (section 4), and describe ongoing and planned work (section 5).