Guidelines for Idealized Simulations Using MM5
Two choices:
- Create a pressure-level file (DATAGRID or RAWINS output type)
for input to INTERP
- Create a sigma-level MM5 input file
1. Pressure-level file
Advantages: Do not have to worry about terrain-following
coordinate. Only need to specify variables on constant p surfaces
(e.g constant, simple gradients). INTERP will take care of vertical
interpolation to model levels, and initialize p' in hydrostatic
balance and w. Also creates boundary file.
Disadvantages: Difficult to use reference state other than
standard log p profile unless also modifying INTERP and MM5.
2. Sigma-level file
Advantages: More control over reference state (have to
change MM5 slightly if not using a log p profile). Able to set initial
state exactly equal to reference state (removes pressure-gradient
force errors because initial p'=0).
Disadvantages: Harder to set initial fields in terrain-following
coordinate. Need to calculate heights of grid-points to set initial
values. If reference profile is not log p, need to set reference
state T0(i,j,k) and PS0(i,j) with PS0 being a surface pressure consistent
with T0 profile. If initial state virtual temp not equal to reference
state T0 need to calculate p' for hydrostatic balance [see method
in INTERP which uses virtual temp profile in each column and surface
actual (not reference) pressure to integrate p' upwards.] Surface
actual pressure needs to be consistent with virtual temp profile
in terrain regions. Need to estimate initial w (may be OK to use
zero everywhere). Have to create a boundary file consistent with
initial state.
Notes on running model with idealized soundings
-
In no-wind situations with terrain, motion may develop if
initial state is not identical to reference state due to pressure-gradient
force errors.
-
Even if ref state equals initial state motion may develop
due to horizontal diffusion on sigma-surfaces (Using ITPDIF=1
helps).
-
Both the above effects become less important, and even negligible,
when a flow is imposed.
-
When there is an initial wind that is not in geostrophic or
thermal wind balance with the temperature and pressure gradient,
Coriolis effects will lead to rotation. This can be avoided
by subtracting off the initial wind in the model's Coriolis
calculation, assuming it is geostrophic and there is no pressure
gradient in the model input. With a frictional layer, this geostrophic
wind may not necessarily equal the initial wind, or if it does
the initial model wind will change with time due to friction.
-
There are no open radiative or periodic lateral boundary options
in the standard model. Values have to be specified there. Frictional
turning or radiative cooling or PBL processes may lead to development
of gradients at the boundary. Also waves may be partially reflected
there. Options are to move the boundaries far away, or nest,
or develop your own boundary conditions (mods for periodic conditions
exist, and possibly radiative conditions in the future, but
these are not supported pieces of code).
-
If using periodic conditions, you should set the map-scale
factors to 1, Coriolis parameter to constant, and latitude and
longitude to constant in the model input.
|