An Alternative Approach to Nonhydrostatic Modeling, Part I: Cold Bubble Test
Z.I. Janjic1), J.P. Gerrity, Jr.1) and S. Nickovic2)
As reported previously by the authors, a nonhydrostatic model has been developed using an alternative approach. The model uses the vertical coordinate based on hydrostatic pressure (mass), and the model equations are consistent with those discussed by Laprise (1992). The basic idea is to split the system of nonhydrostatic equations into two parts: (a) the part that corresponds basically to the hydrostatic system, except for higher order corrections due to the vertical acceleration and (b) the system of equations that allows computation of the corrections appearing in the first system due to the vertical acceleration. The procedure does not require linearization or approximation of any kind.
The nonhydrostatic model designed using this approach can be considered as a natural, evolutionary extension of a hydrostatic model. The nonhydrostatic effects are introduced in a transparent way through an add-on module. This module can be turned on or off depending on resolution, so that the same model can be run in the hydrostatic mode at lower resolution with no extra cost. This feature appears attractive for models designed for a wide range of horizontal resolutions, and in particular for unified global and regional forecasting systems. The proposed concept has been applied within a three-dimensional model with full physics and appears to be computationally robust at all resolutions, and efficient for NWP applications.
In order to demonstrate the soundness of the formulation, the nonhydrostatic model should be able to reproduce important nonhydrostatic motions at very high resolutions even though such resolutions may not be affordable for NWP applications in the near future. These motions include small-scale buoyancy driven flows, as well as the disturbances of uniform flow induced by small orographic obstacles. In order to study such problems, a two-dimensional version of the model has been developed. This version is restricted to the vertical plane and ignores the Earth rotation.
The cold bubble test was designed following the set-up of Straka et al. (1993) who used this set-up in order to produce a well converged solution that would serve as a reference for solutions obtained using lower resolution nonhydrostatic models. In a neutrally stratified atmosphere with the potential temperature of 300, an initial cold disturbance was introduced in the temperature field of the form
m, m, m, m.
The integration domain extended 40 km in the x direction, and the free surface was located at 442 hPa, i.e., at about 6400 m. The center of the initial disturbance was in the middle of the domain in the x direction, i.e., 20 km away from either of the lateral boundaries. As in the main tests in the Straka et al. (1993) study, the horizontal resolution was 100 m, and the vertical resolution was 100 m on the average. The time step was 0.3 s, which was proportionally considerably longer than that used by Straka et al. There was no divergence damping or Rayleigh damping. However, lateral diffusion was applied in both horizontal and vertical directions. The diffusion coefficient was somewhat enhanced compared to that used in the Straka et al. (1993) experiments in order to reach about the same amount of dissipation. Namely, in contrast to the Straka et al. (1993) experiments, in the tests reported here there was no time filtering at all.
The initial potential temperature and the potential temperatures after 300 s, 600 s and 900 s in the right hand part of the integration domain extending from the center to 19200 m, and from the surface to 4600 m are shown in Fig. 1. The u component of the wind and the w component of the wind after 900 s are shown in Fig. 2 in the same region. The dashed contours indicate negative values. Comparison of these figures with the Straka et al. (1993) converged reference solution using the 25 m grid size reveals a remarkable, both quantitative and qualitative similarity. As can be seen from the results summarized in Straka et al. (1993), the present model did equally well or better than other established models specifically designed for studying processes on these scales.
1) NCEP/EMC, 5200 Auth Rd., Camp Springs, MD 20746
2) University of Athens, Greece and ICoD, University of Malta, Valetta
Fig. 1. The cold bubble test: initial potential temperature and the potential temperatures after 300 s, 600 s and 900 s in the right hand part of the integration domain extending from the center to 19200 m, and from the surface to 4600 m. The grid size is and .
Fig. 2. The cold bubble test: the u component of wind and the w component of wind after 900 s in the right hand part of the integration domain extending from the center to 19200 m, and from the surface to 4600 m. The grid size is and . The dashed contours indicate negative values.
Laprise, R., 1992: The Euler equations of motion with hydrostatic pressure asan independent variable. Mon. Wea. Rev.,120, 197‑207.
Straka, J.M., R.B. Wilhelmson, L.J. Wicker, J.R. Anderson and K.K. Droegemeier, 1993: Numerical solutions of a non-linear density current: a benchmark solution and comparisons. Intl. J. Numerical Methods in Fluids, 17, 1-22.