Meteo 465
Middle atmospheric dynamics
You can augment your reading with Holton, Chapters 10 and 12.
While it is correct to state that the general atmospheric circulation is driven by the differential absorption of solar energy at the surface, this statement may be only partially correct for some regions, such as the stratosphere. For the stratosphere, eddies are as fundamental to the circulation as the differential solar heating.
In discussing the general circulation, we will be looking for those processes that maintain the mean zonal, and meridonal circulations. We will see that these are coupled and that waves have a lot to do with it.
Without eddies,
· the zonal mean temperature would match, with a ~10-20 day lag, the radiative equilibrium temperature and would vary annually;
· the flow would be only the zonal mean flow, which is determined by the meridional temperature gradient and the thermal wind balance;
· there would be no stratosphere troposphere exchange;
Heating and cooling patterns result from eddies driving the stratosphere away from radiative equilibrium.
Important quantities:
1. Potential temperature (units: K).
q = T (po/p)R/cp = T
(1000/p)0.286
Potential temperature is conserved when there are no diabatic effects. An air parcel will tend to move on an isentropic surface, or surface of constant potential temperature.
2. (Ertel’s) Potential vorticity (units: K kg-1 m2 s-1).
PV = (z + f) (-g ¶ q /¶p)
where z = k·(Ñ x u) is the relative vorticity, which can be either shear vorticity or curvature vorticity.
PV is conserved in an adiabatic, frictionless flow.
The figure give the PV in the stratosphere relative to the troposphere.
3. Geopotential height:
F = ò g dz
We need also consider geostrophic winds, which are the balance of Coriolis and pressure gradient forces.
vg = - (1/f) ¶ F/¶x
ug = - (1/f) ¶ F/¶y
where F = ò g dz, the geopotential height, R = gas constant, H = scale height, and f = 2W sinj ~ 10-4 s-1 @ 45oN, the Coreolis parameter.
In the analysis to follow, we will assume that the motions are approximately geostrophic, or quasi-geostrophic.
Now, ![]()
From these equations, we can derive the thermal wind equations:
and 
where the “p” subscript implies that the differentiation occurs while the pressure is constant.
Let’s look at the zonal mean circulation first and then we will move on to the mean meridional circulation. We will use the log-pressure coordinate:
z = -H ln(p/po),
where po is taken to be 1000 hPa.
1. Start with the equations of motion:
conservation of momentum:
![]()
![]()
where the first term is the total
derivative of u and v,
, the second term is the Coriolis forcing, the third
term is the pressure gradient forcing, and X and Y are zonal and meridional
components of drag forcing due to small eddies.
hydrostatic equation:
![]()
where H is the scale height.
conservation of mass:
![]()
and thermodynamic energy equation (conservation of energy):
![]()
where
, and J is the
diabatic heating rate.
2. Find the zonal means, which will be needed to examine the vertical and meridional flows. To do that, we must first write all the variables as the sum of a mean and a perturbed (or eddy) part. Thus,
u = ū + u’.
The Eulearian means are found by taking the zonal averages for a fixed latitude, altitude, and time
![]()
Assume that the flow does wander too far in the meridional direction, so that we can use the beat plane approximation.
, where fo
= 2W
sinfo
and b
= 2 W
a-1 cosfo
and A is Earth’s radius = 6400 km.
The result of all these assumptions is the equations:
zonal mean momentum equation:
![]()
change in the mean zonal momentum with time Coriolis forcing mean eddy momentum flux divergence mean zonal eddy drag
thermodynamic energy equation:

T change with time adiabatic cooling mean eddy heat flux divergence diabatic effects
![]()
where is N is the buoyancy frequency, the atmosphere’s natural resonance frequency for gravity-forced oscillations. We have neglected advection by ageostrophic mean meridional circulation and by vertical eddy flux divergences
The meridional momentum equation
can be found by assuming geostrophic balance:
![]()
which when combined with the
hydrostatic relationship, gives the thermal wind relation:
![]()
The ageostrophic mean meridional circulation is constrained by the thermal wind equation: the relationship between the zonal mean wind and the potential temperature distribution.
Without a mean meridional
circulation,
, the eddy momentum flux divergence and the eddy heat
flux divergence would tend to change the mean zonal wind and temperature fields
so that thermal wind balance would be destroyed.
But small departures of the mean zonal wind from geostrophic balance cause a mean meridional circulation, thus maintaining the thermal wind balance. So a balance is established in the meridional direction so that
the Coriolis force
~ the
divergence of eddy momentum fluxes
adiabatic cooling ~ diabatic heating and convergence of eddy heat fluxes
Look at the figure on the atmosphere in radiative equilibrium versus reality. The temperature difference between the summer and winter poles in the 30-60 km region is less than expected from radiative equilibrium. Above, 60 km, the gradient is even backwards. To understand these differences, the Eulerian Mean Circulation does not account for the tendancy of eddy heat flux convergence and adiabatic cooling to cancel each other out, with the diabatic heating being a small factor.
Thus, an air parcel can rise only if its potential temperature is increased by diabatic heating. It is this small diabatic heating gives rise to a residual meridional circulation, which in turn determines the mean meridional mass flow.
We can define
the residual circulation as follows:
![]()
![]()
Now adiabatic motions and eddy thermal flux divergence and convergence are accounted for without assuming that they drive the circulation.
So, the zonal mean momentum equation and the thermodynamic energy equations have the form:
![]()
![]()
where F is the Eliassen-Palm
(EP) flux, which arises from large-scale eddies, and X is the forcing from the
small-scale eddies, like gravity wave drag.
G is the total zonal drag force. F
= jFy + kFz , where the two components are
given by:
![]()
We
can best understand how this all works by starting with a simple model and
progressing to more complex models.
How
big is this forcing? We can get an idea
from our knowledge of the magnitude of fov*. fo ~ 10-4 s-1. Air moves from the tropics to the poles,
about ¼ of Earth’s circumference, in ~2 years.
Thus, in steady-state, G ~ 10-4 s-1 x .2 m s-1
= 2x10-5 m s-2.
First assume no seasonal cycle. In this case, all time derivatives = 0, and
the equations of motion become:
![]()
The Coriolis force due to the residual meridional
velocity just balances the eddies due to large and small eddies. The residual adiabatic cooling is just
balanced by diabatic heating. If the
eddy forcing didn’t exist, then the residual meridional velocity would be 0 and
the meridional drift would stop.
The two relationships are connected by the continuity
equation, which gives the relationship:
If the eddy forcing does not exist, then the diabatic
heating and the residual vertical velocity must also be 0. The simplest model of the atmosphere is thus
one that is in radiative equilibrium, with the zonal mean temperature being
equal to the steady-state radiative balance.
Next, we assume a diabatic heating that mimics the
annual solar cycle. In this model, we
parameterize the diabatic heating as the departure of the stratospheric
temperatures from the radiative equilibrium value:
![]()
where αr
is the Newtonian cooling rate.
The stratosphere’s temperature response will lag
the temperature forcing because of the air’s thermal capacity. This lag is small compared to the annual
cycle because the thermal relaxation time is 5-20 days. Otherwise, this model produces an annually
varying temperature distribution that looks very much mike radiative
equilibrium.
If we substitute this annually varying diabatic heating
rate into the continuity equation, we find the equation:
![]()
where we assume that the density
change is the most important change with altitude.
![]()
where tr is the inverse of ar.
So, we see that the eddies drive the stratosphere away
from thermal equilibrium and the radiation tries to drive it back.
The largest departures from radiative equilibrium occur
in the:
How does all of
this determine the mean meridional circulation?
In the wintertime, mesosphere, internal gravity waves propagate from the
troposphere into the mesosphere, where they break. These breaking waves deposit energy,
resulting in a strong zonal force that acts like friction, slowing down the
zonal velocity. Stationary planetary
waves play a similar role in the wintertime stratosphere.
Consider the
equation:
![]()
In the Northern Hemisphere winter, fo > 0 and G = wave
drag is a westward force and thus <0.
Thus the residual meridional velocity must be >0. By mass continuity, the residual vertical
velocity must be negative, and air descends and warms.
In the Southern
Hemisphere summer, fo < 0 (the zonal wind is westward,
easterlies). As a result, G > 0
(eastward force to opposing the wind)
Thus –(-|fo|)v*
= G implies that v* > 0 , or northward drift.
Linear Waves
Wave
classifications
Introduction to wave theory.
Assume
a shallow fluid wave.
(diagram)
h(x,t) = H + h’(x,t)
momentum equation:
![]()
continuity equation:
![]()
We assume a harmonic solution, as
always occurs for linear wave theory:
![]()
![]()
where
![]()
To
stay on the wave’s crest, which gives us an idea of the wave’s phase speed, we
choose f so that this
happens. We substitute these wave
equations into the momentum and continuity equations and get the dispersion
relations:
![]()
To satisfy these two simultaneously, we must
have:
![]()
The phase speed, c, is given by:
![]()
This relationship is non-dospersive, since c
does not depend on k.
Buoyancy waves (Gravity waves).
These are vertically propagating waves.
Assume first that the mean zonal wind is zero.
(diagram)
In this case,
![]()
If you displace the
parcel along a slantwise path, it will stay along that path because the gravity
force is down, but the pressure force acts to move it back along the path.
![]()
The vertical force,
component of the vertical buoyancy force along the path are:
![]()
Thus, the resulting
description of the motion is the simple harmonic oscillator:
![]()
which has the
solution:
![]()
where k is the wave
number in the zonal direction and m is the wavenumber in the vertical
direction.
![]()
where Lx
~ 10-100 km, Lz ~ 5-15 km, and the period is ~ minutes to an hour.
The phase speed in
this case is dependent on both k and m, making it dispersive, meaning that the
wave packet changes shapes as it proceeds:
![]()
The group
velocities, which tell us about energy propagation, is found by taking the
derivation of the frequency w with respect to k and m:

When the phase is
being propagated downward and to the east, the group velocity and hence the
energy is propagated upward and to the east.
We can see this by taking certain signs of the solutions:
cx >
0 leads to cgx > 0, but cz < 0 leads to cgz
> 0
The group velocity
is perpendicular to the phase velocity.
What is the source
of gravity waves?
The atmosphere can
be excited at frequencies only up to the Brunt-Vasaila frequency, N. If the frequency is less than N, then we will
get waves that are excited on the slant paths.
Assume now that
there is a mean zonal wind.
![]()
Now the group
velocities are:

The group velocity
is increased to the east by the addition of the mean zonal wind.
It should be
pointed out that the actual air molecules are not moving very far from the
position that they would have with the mean zonal wind. In the stratosphere, this motion is no more
than a 100 meters or so.
Inertio-gravity waves. Inertio-gravity waves are gravity waves that
have wavelengths long enough (~100’s of km) that they are significantly
affected by the Coriolis force. In the
Northern Hemisphere, the Coriolis force deviates the waves so that they turn
clockwise with height.
(diagram)
In the momentum and
continuity equations, we must introduce terms for the Coriolis force. When we do this, the dispersion relation that
we achieve is:
![]()
where l is the
wavenumber in the meridional (y) direction.
For the waves to
exist, |f| < |w| << N.
Inertio gravity
waves tend to propagate more horizontally than do internal gravity waves.
Planetary waves (Rossby waves).
![]()
where b = df/dy = the planetary vorticity gradient.
For dy < 0, the rotation is cyclonic
(counterclockwise), znew > 0, because