Contact: B. Wyman
Reviewers:
Change history: WebCVS Log for bgrid_core.f90
The basic driver module for the B-grid atmospheric dynamical core.
This version can be run on distributed memory machines.
The B-grid dynamical core updates the tendencies of the
prognostic variables, the variables themselves are not modified.
The prognostic variables are surface pressure, the zonal and
meridional momentum components, temperature, and optional
tracers. Specific humidity is considered a tracer.
This version of the B-grid dynamics use a two time level scheme.
The gravity waves are integrated using a forward-backward scheme
with a relatively short time step and the advection terms are
integrated using a modified Euler backward scheme and longer
time step. Horizontal diffusion is done on the advective time step.
Derived-type variables for the horizontal and vertical constants
must be initialized before initializing the B-grid dynamical core.
The initialization routine of the B-grid dynamical core returns
a derived-type variable that is needed when updating the tendencies
of the prognostic dynamical variables. These prognostic variables
are also contained as a derived-type variable.
Here is a detailed write-up of the finite differencing used in the
B-grid atmospheric dynamical core.
bgrid_prog_var_mod bgrid_horiz_mod bgrid_vert_mod bgrid_masks_mod bgrid_advection_mod bgrid_horiz_diff_mod bgrid_horiz_adjust_mod bgrid_vert_adjust_mod bgrid_polar_filter_mod bgrid_boundary_mod bgrid_sponge_mod fms_mod constants_mod
use bgrid_core_mod [ ,only: bgrid_dynam_type,
bgrid_core_init,
update_bgrid_core,
bgrid_core_end ]
bgrid_dynam_type
A derived-type variable that contains grid constants, time step information,
and other pre-computed terms that are needed by the dynamical core.
This includes other derived-type variables and pointers to the physical
grids of the model, but does not include the prognostic variables.
bgrid_core_init
Initializes the derived-type variable, bgrid_dynam_type, and initializes
other modules used. There is no namelist for bgrid_core_mod, all runtime
options are controlled with optional arguments to this subroutine.
This subroutine must be called before update_bgrid_core.
update_bgrid_core
Called at every (atmospheric) time step to update the tendencies of the
prognostic variables. No time differencing is done.
bgrid_core_end
Called at the end of the model run to terminate the module.
Currently this call does nothing.
type bgrid_dynam_type
-- derived data types contained in bgrid_dynam_type --
(see the appropriate module for details)
Hgrid = horizontal grid constants, initialized by horiz_grid_mod
[horiz_grid_type]
Vgrid = vertical grid constants, initialized by vert_grid_mod
[vert_grid_type]
Masks = eta coordinate topography masks and indices,
initialized by grid_masks_mod [grid_mask_type]
Pfilt = polar filter constants, initialized by polar_filter_mod
[pfilt_control_type]
Advec = advection constants, initialized by advection_mod
[advec_control_type]
Hdiff = horizontal diffusion constants, initialized by horiz_diff_mod
[hdiff_control_type]
-- real, dimension(:,:) --
fis = geopotential height of the surface (m2/s2)
fisl = geopotential height at eta=1. (for eta coord = 0.0,
res = reciprocal of eta at the surface
-- scalars --
nt_adj = no. of adjustment time steps per advection step [integer]
nt_adv = no. of advection time steps per atmospheric step [integer]
dt_adj = adjustment time step in seconds [real]
dry_model = if true then water vapor will not be considered in
the equation of state [logical]
verbose = verbose flag [integer]
wcorr = coefficient for the grid separation smoothing operator [real]
fopt = filtering option [integer]
NOTE
Hgrid, Vgrid, fis, and res are pointers to the arguments passed
via the initialization call (bgrid_core_init).
Dynam = bgrid_core_init ( Hgrid, Vgrid, fis, res, dt, ntadj, ntadv
[, damp_order_vel, damp_order_tmp, damp_order_trs,
damp_coeff_vel, damp_coeff_tmp, damp_coeff_trs,
damp_scheme, damp_slope_coeff_vel, damp_slope_coeff_tmp,
num_horiz_fill, num_vert_fill,
advec_order_vel, advec_order_tmp, advec_order_trs,
advec_coeff_vel, advec_coeff_tmp, advec_coeff_trs,
grid_sep_coeff, filter_option, filter_weight,
ref_lat_filter, num_sponge_levels,
sponge_coeff_vel, sponge_coeff_tmp, sponge_coeff_trs,
dry_model, verbose ] )
INPUT
Hgrid Derived-type variable containing horizontal grid constants (see horiz_grid_mod).
[type(horiz_grid_type)]
Vgrid Derived-type variable containing vertical grid constants (see vert_grid_mod).
[type(vert_grid_type)]
fis Geopotential height of the surface (m2/s2).
Should have horizontal indexing consistent with the B-grid core.
[real, dimension(:,:)]
res Reciprocal of eta at the surface. Used at a switch for sigma vs. eta.
For the sigma coordinate, all res = 0.
Should have horizontal indexing consistent with the B-grid core.
[real, dimension(:,:)]
dtadv Time step in seconds for each call to update_bgrid_core.
This should be the atmospheric time step.
[real]
ntadj The number of adjustment time steps for each advective time step.
[integer]
ntadv The number of advection time steps for each call to update bgrid_core.
[integer]
RETURNS
Dynam Derived-type variable that contains quantities needed
by the dynamical core. [type(bgrid_dynam_type)]
OPTIONAL INPUT (Note: These argument can be modified through a
namelist interface in the bgrid_core_driver_mod)
damp_order_vel The horizontal damping order for momentum,
damp_order_tmp temperature, and default order for all
damp_order_trs prognostic tracers. The damping order must be
an even number; damp_order = 0 turns off damping.
[integer, default: damp_order = 4]
damp_coeff_vel The horizontal damping coefficients for
damp_coeff_tmp momentum, temperature, and default value for
damp_coeff_trs all prognostic tracers. The coefficients are
expressed as non-dimensional values for the
second-order diffusion operator (range = 0,1).
[real, default: damp_coeff = 0.50]
damp_scheme Determines how horizontal damping coefficients
vary with latitude.
= 1, constant
= 2, varies as inverse of diagonal grid distance
= 3, varies as inverse of x-axis grid distance
Note: damp_scheme = 1 is recommended,
damp_scheme = 2,3 is experimental.
[integer, default: damp_scheme = 1]
damp_slope_corr_vel The topography slope correction applied to horizontal
damp_slope_corr_tmp damping of momentum and temperature (including all
prognostic tracers). The coefficients (with range = 0,1)
are expressed as arrays of size 4. The first 3 values are
coefficients for the lowest 3 model layers, the last value
represents the remaining uppermost layers. A NON-ZERO
value turns the correction ON. Typical values might be
(/ .25, .50, .75, .95 /).
[real, dimension(4), default: damp_slope_corr = 0.,0.,0.,0.]
advec_order_vel The advection order for momentum, temperature,
advec_order_tmp and default order for all prognostic tracers.
advec_order_trs The advection order must be an even number.
[integer, default: advec_order = 2]
advec_coeff_vel Coefficients for modified Euler-backward advection
advec_coeff_tmp scheme for momentum, temperature, and all
advec_coeff_trs prognostic tracers.
NOTE: advec_coeff=0 is the Euler-forward scheme which
is unstable, advec_coeff=1 is the Euler-backward scheme
which is highly dissipative.
[real, default: advec_coeff = 0.7]
num_horiz_fill The number of successive horizontal and vertical passes
num_vert_fill applied in the tracer borrowing/filling scheme. This
conservative scheme is used to fill negative tracer values.
It is applied in either the vertical and horizontal directions.
Each successive pass should remove more negative values,
however an optimum number of passes is probably between 1-3.
This is applied after advection to all prognostic tracers.
[integer, default: num_fill = 1]
grid_sep_coeff Coefficient to suppress grid-separation problem
associated with the B-grid. Currently, this option has been
disabled within the model, so that this coefficient does nothing.
[real, default: grid_sep_coeff = 0.00]
filter_option Determines how polar filtering is performed.
filter_option = 0, NO filtering
= 1, not implemented
= 2, filter horiz OMG/DIV,
advec mass tendencies,
and momentum
[integer, default: filter_option = 2]
filter_weight Weight applied to the polar filter that will
increase (or decrease) the strength of the standard
polar filter response function.
SS(new) = SS(std)**filter_weight,
where SS(std) is the Arakawa and Lamb response function.
[integer, default: filter_weight = 1 ]
ref_lat_filter The reference latitude at which polar filtering
(in each hemisphere) will begin to be applied.
Setting this argument >= 90. will turn off
polar filtering.
[real, default: ref_lat_filter = 60.]
num_sponge_levels Number of uppermost model level where a band-pass
filter is applied to damp undesirable waves.
Currently num_sponge_levels > 1 is not allowed.
If num_sponge_levels = 0, no damping is done.
[integer, default: num_sponge_levels = 0 ]
sponge_coeff_vel Damping coefficients for the sponge layer(s) in
sponge_coeff_tmp the uppermost model levels. Coefficients have been
sponge_coeff_trs normalized and must be in the range [0,1].
If num_sponge_levels = 0, the value of the coefficients
is ignored. There is no option to specify coefficients
that vary with level, although currently
num_sponge_levels > 1 is not allowed.
[real, default: sponge_coeff = 0.]
dry_model Flag that determines whether or not water vapor effects
are included in the hydrostatic equation.
[logical, default: dry_model = true]
verbose Flag that control additional printed output.
Currently, this option is not being used.
[integer, default: verbose = 0]
NOTES
fis and res should be dimension by the size of the global grid,
the number of longitude points by number of latitude points.
The declaration of fis and res must have the target attribute
and the storage must be static.
---------------------------------------------------------------
call update_bgrid_core ( Var, Var_dt, Dynam, omega )
INPUT
Var A derived-type variable that contains the B-grid's
prognostic variables.
[prog_var_type, see prog_var_mod]
INPUT/OUTPUT
Var_dt A derived-type variable that contains the TENDENCIES
for the B-grid's prognostic variables.
[prog_var_type, see prog_var_mod]
Dynam The derived-type variable returned by a previous call
to bgrid_core_init (see above).
[type(bgrid_dynam_type)]
OUTPUT
omega The omega diagnostic (from the thermodynamic equation) in
pascals per second. The array should have horizontal dimensions
that are consistent with the B-grid dynamical core.
[real, dimension(ilb:,jlb:,:)]
---------------------------------------------------------------
call bgrid_core_end ( Dynam )
INPUT
Dynam Derived-type variable returned by a previous call to bgrid_core_init.
[type(bgrid_dynam_type)]
FATAL ERRORS in bgrid_core_init
input argument ntadj must be >= 1
If you were using the namelist interface &bgrid_core_driver_nml
then check the namelist variable corresponding to this variable.
input argument ntadv must be >= 1
If you were using the namelist interface &bgrid_core_driver_nml
then check the namelist variable corresponding to this variable.
input argument dtadv must be > 0.
The model time step usually is set at the highest program level.
Check the namelist for the main program.
Here is a detailed write-up of the finite differencing used in the
B-grid atmospheric dynamical core.
There are no known bugs.
internal options:
alpha_implicit determines how the coriolis and press grad force
terms are solved
= 0.5 trapezoidal implicit
= 1.0 fully implicit
[real, default: alpha_implicit = 0.5]
1) Addition polar filter options
(e.g., filter prognostic variables or tendencies)