6.4 A High Resolution Numerical Simulation of a Midwestern Quasi-Linear Convective System

Tuesday, 5 October 2004: 2:15 PM
Jason T. Martinelli, Creighton University, Omaha, NE; and R. W. Pasken and Y. J. Lin

The Fifth-Generation Penn State/NCAR Mesoscale Model Four Dimensional Variational Data Assimilation (MM5-4DVAR) system is employed to investigate the structure, evolution, and internal circulation of a Midwestern linear mesoscale convective system (MCS) that occurred on 15 April 1994. The model employs four domains with 27, 9, 3, and 1 km grid spacing, respectively. The largest domain covers the entire central U. S., while the smaller domains are optimally placed to capture pertinent topographical or environmental features. The model simulation is initialized using the National Center for Environmental Prediction (NCEP) Global Optimum Interpolation (GOI) 2.5 degrees analyses at 0000 UTC 15 April 1994, and is integrated forward for 24 hours.

In order to quantify any improvement or degradation in results through use of the MM5-4DVAR system, a control experiment was performed prior assimilation of Doppler radar data. Initial results show that MM5 is capable of simulating the synoptic conditions favorable for MCS initiation. Additionally, the model successfully simulates the MCS evolution from its incipient stage through the mature stage similar to that observed by the Saint Louis, Missouri WSR-88D Doppler radar (KLSX). Specifically, the model is able to show the detailed storm evolution including cell-growth and dissipation, local acceleration along portions of the line, numerous transient storm-scale circulations along the leading edge of the system, and secondary development. Additionally, the model was able to accurately reproduce the internal airflow patterns associated with the storm-relative front-to-rear and rear-to-front flow branches as ascertained through detailed examination of the reflectivity and Doppler velocity fields observed by KLSX. The primary inaccuracy noted in the control case is a considerable time lag between the observed squall line and the modeled system. This is likely attributed to the resolution of the data with which the model is initialized.

The MM5-4DVAR system determines the optimal initial (OI) and lateral boundary conditions (LBC) for the MM5 through the fitting of model to the observed data over a specified time. The optimal condition is measured by a cost function that expresses the discrepancy between the model forecast and the observed data. The incorporation of information at multiple time periods (as opposed to one time period) is used to improve the initial analysis though the model's internal physics and dynamics. The MM5-4DVAR system allows assimilation of data in their raw or nearly raw format, thus averting the loss of information often encountered when a retrieval method is applied. Three advantages arise when employing the 4DVAR method: 1) the ability to assimilate observational data into the model in a form resembling their original measurements, 2) a more accurate description of the error characteristics of the measurements is provided, and 3) the full non-linear model is employed as a dynamic constraint in the analysis of the observations.

WSR-88D data is assimilated using the MM5-4DVAR in order to achieve the OI conditions for the numerical simulations. This is accomplished first by critically evaluating experiments that assimilate radial velocity data only, followed by experiments that assimilate reflectivity data only, and finally with simultaneous assimilation of radial velocity and reflectivity data.

The MM5-4DVAR system's ability to accurately reproduce convective-scale features associated with linear MCSs will be presented. Model results will be compared with the observed radar reflectivity and velocity fields and accepted conceptual models in order to identify any gains made through the use of the MM5-4DVAR system.

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner