ALADIN 2D version of ALADIN and associated tools

(more details vivoda++at++synoptik.shmu.sk)

There was need to introduce 2d-plane version of ALADIN. There were a few objectives for this:

• 1. to test existing dynamics in idealized conditions where analytical solutions are known
• 2. to prepare a tool for model development where dynamics can be easily controlled

The (y,z) 2d plane model has been developed in Toulouse during the summer 1999 and model changes have been phased into CY21 and according to references the 2d plane model is working properly now.

To run 2d plane model it is necessary to do:

• 1. to use CY21T1/AL11 or later
• 2. to prepare an initial field for the idealized test
• 3. to prepare LBCs if needed
• 4. to set the model namelist consistently with 2d model and with your idealized test
• 5. to run diagnostics

Model cycles

Correct cycles of model can be fetched from Toulouse whenever it is necessary and we will not discuss this point here.

Initial conditions and LBCs

To prepare initial conditions and LBCs you need to run the program ACADFA(ACADemical FA file).

This program allows you to prepare idealized flow regimes characterized by constant wind, dry atmosphere, temperature profile defined by constant Brunt Vaisala frequency (BVF), constant pressure at mean sea level and Agnesi shape obstacle. Height and half width of obstacle together with BVF and wind speed define the character of the regime (hydrostatic or nonhydrostatic regime, linear or nonlinear regime). The initial filed is hydrostatically balanced.

Namelist settings

There is not too much that has to be set in the namelist to run 2d model once you have initial conditions and LBCs. You have to set the following variables in the namelist:

• LMESSP=3D.FALSE. (2d model is running in SM but it can be run in more than one task)
• NPROMA=3D5 (this is a recommendation)
• NGRSP=3D0 (no spectral subpacking)
• NBZONL=3D0
• LMAP=3D.FALSE.

Previous switches have to be set to get reasonable results from technical point of view. Next switches are more scientific and they are from new namelist NAMPONG and they define characteristics of sponge. This sponge was introduced to solve the problem of waves reflecting from model top.

• LEPONGE=3D.TRUE. (turn on sponge)
• REPONCO=3D10.0, (dimensionless sponge intensity)
• REPONBT =3D 4000., (bottom of the sponge layer in meters)
• REPONTP =3D 6200., (top of the sponge layer in meters)
• RV00 =3D4.0, (background wind speed)
• RT00 =3D285., (background temperature for LIZOT=3D.TRUE. or surface temperature for LIZOT=3D.FALSE.)
• RAGNESI=3D400., (height of the Agnesi shape obstacle)
• RWIDTH =3D400., (half width of the Agnesi shape obstacle)
• LIZOT =3D.FALSE., (isothermal or non isothermal atmosphere)
• RBRVAF =3D0.01, (Bruint Vaisala frequency to define temperature profile in the case of LIZOT=3D.FALSE.)
• RTB00(1:NFLEVG), (background temperature profile at full levels for LIZOT=3D.FALSE., since temperature is not horizontally constant; in this case RTB00 represent averages of temperature at each model level)

The model variables are coupled with the background variables (RT00,RV00,RTB00) during the model execution exactly in the same way as it is done in the lateral coupling case. The background variables are usually equal to the initial conditions. The weight function is fully defined by REPONCO,REPONTP and REPONBT. Vertical velocity is coupled with zero in the case of nonhydrostatic model run rather than pseudovertical divergence. Pressure departure is coupled with zero also for nonhydrostatic run.

Diagnostics

To perform diagnostics, a new tool has been developed. This tool allows you to calculate true vertical velocity (in m/s) on model levels, vertical momentum flux at constant height levels and surface drag.

Input to this program is the ICMSH file directly. Any new calculation could be easily added into this program.

Experiments using 2d plane model

We tried to repeat idealized simulations which can be found in (Bubnova, 1995). We have focused on two regimes:

• 1. Linear hydrostatic
• 2. Nonlinear non-hydrostatic

Both regimes have been tested with eulerian, semi-lagrangian 3-time level and semi lagrangian two-time level. Nonhydrostatic version of the model has been initialized using (LNHDYN=3D.TRUE.) for all experiments.

Linear hydrostatic regime

Model setting was as follows:

• Time step 120s
• Mesh size 3200m, number of points in the horizontal 192, truncation 63, coupling zone width 8 points
• The obstacle is Agnesi shape with height 1m and half horizontal width 16000m
• Background and initial state is isothermal with temperature 239.5K (equivalent to 0.02s-1 BV frequency), wind is constant at each level 8m/s
• 201 vertical levels with approximate distance 100m

Result after 2200 time steps are presented on figure 1.

Figure.1. Vertical velocity perturbation after 2200 time step for linear hydrostatic flow regime. Test has been performed with eulerian advection scheme and nonhydrostatic model version.

Nonlinear nonhydrostatic regime

Model setting was following:

• Time step 2s
• Mesh size 80m, number of points in the horizontal 180, truncation 59, coupling zone width 8 points
• The obstacle is Agnesi shape with height 400m and half horizontal width 400m
• Background and initial state is constant Bruint Vaisala frequency 0.01s-1 ,wind is constant at each level 4m/s
• 101 vertical levels with approximate distance 100m

Result after 4000 time steps are presented on figure 2.

Figure 2. Vertical velocity perturbation after 4000 time step for nonlinear nonhydrostatic flow regime. Test has been performed with semilagrangian three time level advection scheme and nonhydrostatic model version.

Changes to enhance number of model levels

If someone wishes to run ALADIN with more than 100 or even more than 200 vertical levels, he (she) has to carry out the following steps (valid for CY21T1/AL11). These steps are a redefinition of parameters used to define the size of statically allocated variables (variables are set to 300 for example):

• 1. Set JPXNIV=3D300 in facomp.h
• 2. Set JPMXXLEV=3D300 in echien.F90
• 3. Set JPMXLE=3D300 in pardim.F90
• 4. Set JPMXLEV=3D300 in yomarg.F90
• 5. Set JPOPLEV=3D300 and JPOGGLEV=3D300 in yomppc.F90

This is a general list. Default values in the operational model are 100 for facomp.h and 200 for the rest of changed parameters.

Then all routines dependent on previous modules and includes have to be recompiled etc...

Future plans

There are troubles to get good initial condition for nonlinear nonhydrostatic regime using ACADFA. The solution we currently get from file prepared by ACADFA is not analytical. Results presented here for the nonlinear non-hydrostatic regime have been reached using initial conditions prepared by R.Bubnova in 1995. So, first thing we have to solve is correction of this ACADFA shortcoming.

Next plan is to use 2d-plane version to stabilize 2-time level semi-lagrangian scheme for the non-hydrostatic version of ALADIN. This is pretty important in order to prepare conditions for operational running of non-hydrostatic ALADIN.

All programs mentioned here are available. It is enough to contact vivoda++at++synoptik.shmu.sk

Home