Two time-splitting methods for integrating the elastic equations are presented. The methods are based on a third-order Runge-Kutta time scheme and the Crowley advection schemes. The schemes are combined with a forward-backward scheme for integrating high-frequency acoustic and gravity modes to create stable splitexplicit schemes for integrating the compressible Navier-Stokes equations. The time-split methods facilitate the use of both centered and upwind-biased discretizations for the advection terms, allow for larger time steps, and produce more accurate solutions than existing approaches. The time-split Crowley scheme illustrates a methodology for combining any pure forward-in-time advection schemes with an explicit time-splitting method. Based on both linear and nonlinear tests, the third-order Runge-Kutta-based time-splitting scheme appears to offer the best combination of efficiency and simplicity for integrating compressible nonhydrostatic atmospheric models.

SUMMARYA comparison between solutions from simulations of a non-linear density current test problem was made in order to study the behaviour of a variety of numerical methods. The test problem was diffusion-limited so that a grid-converged reference solution could be generated using high spatial resolution. Solutions of the test problem using several different resolutions were computed by the participants of the 'Workshop on Numerical Methods for Solving Nonlinear Flow Problems', which was held on 11-13 September 1990 at the National Center for Supercomputing Applications (NCSA). In general, it was found that when the flow was adequately resolved, all of the numerical schemes produced solutions that contained the basic physics as well as most of the flow detail of the reference solution. However, when the flow was marginally resolved, there were significant differences between the solutions produced by the various models. Finally, when the flow was poorly resolved, none of the models performed very well. While higher-order and spectral-type schemes performed best for adequately and marginally resolved flow, solutions made with these schemes were virtually unusable for poorly resolved flow. In contrast, the monotonic schemes provided the most coherent and smooth solutions for poorly resolved flow, however with noticeable amplitude and phase speed errors, even at finer resolutions.

No abstract

Warnings about convective-scale hazards are currently based on observations, but the time has come to develop warning methods in which numerical model forecasts play a much larger role.

An ''additive noise'' method for initializing ensemble forecasts of convective storms and maintaining ensemble spread during data assimilation is developed and tested for a simplified numerical cloud model (no radiation, terrain, or surface fluxes) and radar observations of the 8 May 2003 Oklahoma City supercell. Every 5 min during a 90-min data-assimilation window, local perturbations in the wind, temperature, and water-vapor fields are added to each ensemble member where the reflectivity observations indicate precipitation. These perturbations are random but have been smoothed so that they have correlation length scales of a few kilometers. An ensemble Kalman filter technique is used to assimilate Doppler velocity observations into the cloud model. The supercell and other nearby cells that develop in the model are qualitatively similar to those that were observed. Relative to previous storm-scale ensemble methods, the additive-noise technique reduces the number of spurious cells and their negative consequences during the data assimilation. The additive-noise method is designed to maintain ensemble spread within convective storms during long periods of data assimilation, and it adapts to changing storm configurations. It would be straightforward to use this method in a mesoscale model with explicit convection and inhomogeneous storm environments.