Constructing A Universal General

1
Constructing A Universal General Circulation
Model
Isentropic Vertical Coordinates A significant
source of error in current atmospheric models is
known to come from inaccuracies in the
vertical-velocity terms. This problem affects all
models that use geometric height or pressure as
their vertical coordinate. Because large-scale
atmospheric motions are nearly adiabatic, models
that use entropy as their vertical coordinate,
called isentropic-coordinate models, do not
suffer this problem. Instead, the coordinate
surfaces follow the motion such that there is no
effective vertical velocity except where there is
diabatic heating. Recent studies have shown
that even when moist convection is a significant
source of diabatic heating, as in the case of
thunderstorms and cumulus convection,
isentropic-coordinate models produce more
accurate results because they better represent
the long-range, subsaturated transport of the
fuel that drives the storms, water vapor. The
Explicit Planetary Isentropic-Coordinate (EPIC)
atmospheric model, developed at UofLs
Comparative Planetology Laboratory (CPL) takes
full advantage of isentropic coordinates and is
the first general circulation model (GCM)
designed to simulate all one dozen atmospheres in
the solar system. It uses standard Message
Passing Interface (MPI) library routines to run
on any parallel computer system. The EPIC model
is freely downloadable from NASAs Planetary Data
System (PDS) Atmospheres Node (http//atmos.nmsu.e
du). The EPIC model is currently designed to
run on PC clusters, but through the collaboration
with the KAOS Lab, the model software and
hardware will be optimized to significantly
improve its performance for all applications.
EPIC model prediction of atmospheric buoyancy
waves (gravity waves) generated from the
collision of Comet Shoemaker-Levy 9 with Jupiter.
The governing length scale in the atmospheres of
both Earth and Jupiter is approximately the same,
about 1000 km, but because Jupiter has 120 times
more surface area than Earth, the use of parallel
computers is required to simulate the nonlinear
physics, and the same holds for accurate computer
simulations of Saturn, Uranus, and Neptune.
Hybrid Vertical Coordinates The one region where
pure isentropic coordinates are not preferred is
at the bottom of an atmosphere. For
terrestrial-class atmospheres (Venus, Earth,
Mars, Io, Titan, Triton, Pluto) the problem is a
technical one, isentropes intersect the surface
at large angles. For gas-giant-class atmospheres
(Jupiter, Saturn, Uranus, Neptune, the Sun) the
problem is a physical one, their deep fluid
interiors have nearly uniform entropy as a result
of convection driven by escaping internal heat,
and some other coordinate beside entropy must be
used to resolve vertical structure. The best type
of coordinate near a solid surface is a
terrain-following one, the most common being the
sigma coordinate, where sigma equals the pressure
divided by surface pressure. Unfortunately,
sigma-coordinate models suffer the same vertical
truncation errors aloft that affect all
non-isentropic coordinate models. At the CPL we
are developing a hybrid EPIC model that smoothly
combines isentropic coordinates aloft with sigma
coordinates near the bottom, using ideas
presented by modeling groups from around the
nation at the 1st Hybrid Modeling Workshop, held
in Madison, WI, Aug. 15-16, 2000. The 2nd Hybrid
Modeling Workshop will be hosted by the CPL in
Louisville, KY, March 7-9, 2002.
Three-dimensional view of the distribution of
ammonia (thin upper cloud) and water vapor (thick
lower cloud) in an EPIC model of Jupiters Great
Red Spot. The current development of a hybrid
isentropic-terrain following vertical coordinate
allows the EPIC model to penetrate deep into the
convecting interior of gas giants, in addition to
increasing the accuracy of simulating surface
processes for terrestrial-class planets.
The Venus EPIC model is the highest-resolution
GCM for Venus developed to date. Shown is the
topography of Venus as accurately mapped by the
Magellan Spacecraft (vertically exaggerated),
with color indicating the model surface
temperature variations. The slices indicate the
domain decomposition used by the EPIC model in
the case of a 4-processor simulation. The
collaboration with the KAOS Lab will optimize the
message passing for arbitrarily large numbers of
processors, greatly increasing the speed of the
model.
Planetary Boundary Layers Another challenge
presented by terrestrial planets is the effect of
the turbulence generated by flow over the surface
topography, a region known as the planetary
boundary layer (PBL). The physics of the PBL is a
critical factor in many phenomena like the
ocean-air carbon transfer process on Earth, the
Martian Dust storms, and surface ice sublimation
on Io and Triton. Yet, useful measurements of
the Earths PBL are rare measurements of other
planetary PBLs are largely non-existent. The
collaboration between the CPL and UK CFD will
apply the knowledge and techniques developed for
simulating engineering boundary layers in order
to construct an improved model for planetary
boundary layers.
Hybrid EPIC Venus model showing the smooth
transition of layers from isentropic aloft to
terrain-following (sigma-coordinate) near the
surface.