You are on page 1of 43

Dynamical Meteorology

J.J. Smorenburg
Studentnumber: 3982238
Utrecht University
20 oktober 2017

1
Part 1: 1.2-1.7, 1.9-1.12, 1.14-1.22, sessions: 1-8
Chapter 1: Introduction to the atmosphere
1.2 Continuum hypothesis and the continuity equation
The continuum hypothesis regards the mass of a given small volume of air (air parcel)
consisting of many randomly moving molecules which do not interact. This means that the
molecules are spread uniformly over the control volume V, enclosed by a control surface
S. We have a mass flux associated with mass conservation:
∂ρ ~ dρ
J~ = ρ~v , + ∇ · (ρ~v ) = 0, ~ · ~v ) = 0.
+ ρ(∇
∂t dt

1.3 Pressure, hydrostatic balance, equation of state and buoyancy


Since there is a relatively large distance between molecules, we can neglect intermolecular
forces which are not due to elastic collisions. The collective effect of the latter is called
a stress at the boundary of the air parcel. It can act in the normal direction leading to
compression or tension and parallel to the surface leading to deformation (shear stress).
In fluids, shear stresses are associated with molecular viscosity and normal stresses are
pressure. For hydrostatic balance (and initial condition), we have:
 ∂p   ∂p 
0
= −ρg and = −ρ0 g.
∂z x,y,t ∂z x,y,t
If these 2 forces (per unit volume) are not in balance, we get
d2 z dw  ∂p  dw 1  ∂p 
m 2 ≡m = −V − mg, or =− − g.
dt dt ∂z x,y,t dt ρ ∂z x,y,t
We can investigate the dynamics of the vertical acceleration of an air parcel. The air
parcel changes its volume, while its mass is conserved, leading to a change in pressure:
dw ∂p0 ∂p0  = ρ0 gV 0 − V ∂p .
0
m = −( V0 + V 0 ) − (V0 + V 0 ) −mg

dt ∂z ∂z ∂z
The parcel is accelerated vertically by the Archimedes force (buoyancy force) and a
force associated with the pressure perturbation (the source of a sound wave or wave of
expansion). If we neglect the pressure perturbation and use the ideal gas law, we get:
 T T0  mRT 0
pV = mRT and p0 V0 = mRT0 , so: V 0 = V − V0 = mR − ≈ .
p0 +  p0 p0 p0
If we now use p0 = ρ0 RT0 , we get
0 0
dw ∂p
 ρ0 gm R T gmT 0 dw T0
= ρ0 gV 0 −V ≡ g0 .

m ≈ = , so =g
dt  ∂z ρ

 0 R T 0 T0 dt T0

Here g 0 is the buoyancy force per unit mass or reduced gravity. Air can be regarded
as an ideal gas with equation of state:
p = nkB T = ρRT, [n] = molecules/m3 , [R] = J/kg/K, kB = 1.38 · 10−23 J/K.
According to Dalton’s law we can apply the equation of state to each constituent of air:
pd pv p−e e
ρ = ρd + ρv = + = + .
Rd T Rv T Rd T RV T
For the specific gas constant we have
Rd = cp − cv = 1004 − 717 = 287 J/kg/K,
related to specific heat capacity at constant pressure and volume respectively. If air is a
mixture of dry air and water vapor, R is the specific gas constant for this mixture.

2
1.4 Energy conservation and adiabatic/diabatic processes
When we change the volume of a parcel, we need to exert work on it, leading to a change
for the internal energy:

dW = pdAdz = pdV and dU = −dW = −pdV = mcv dT.

This leads to the law of energy conservation for an adiabatic process:


dV
mcv dT + pdV = 0, or cv dT = −p = −pdα.
m
We also have p = ρRT or:
(cp − cv )T = pα, so cp dT + 
pdα
 = αdp +  , or cp dT − αdp = 0.
pdα
Using the hydrostatic balance, we can derive the dry adiabatic lapse rate:
dT g
= − ≈ −10−2 K/m.
dz cp
In an adiabatic process, the potential temperature is conserved:
 p R/cp
ref
θ=T .
p
The energy budget for a diabatic process is the first law of thermodynamics:
cp dT − αdp = Jdt,
where J is the heat added to the air parcel per unit mass per unit time (J/kg/s). So:
dθ ∂θ ∂T θ J J  p R/cp
= = = , with Π = cp .
dt ∂T ∂t T cp Π pref
We can also express J in terms of a change in entropy:
T dS θdS
Jdt = T dS so dθ = = or dS = cp d(ln θ).
Π cp
So the dry adiabatic lapse rate is also referred to as the isentropic lapse rate. An
atmosphere which is completely mixed is said to be in convective equilibrium.

1.5 Static instability, buoyancy oscillations and convection


p0 ), we get from the z-momentum equation:
If we assume no change in pressure (p = p0 + 

dw d2 z T0 θ0 (z, t) θ(z, t) − θ0 (z)


= 2 =g ≈g =g· .
dt dt T0 θ0 (z) θ0 (z)
The vertical acceleration of an air parcel is determined by the potential temperature
difference between the parcel and the environment (at the same height). We suppose that
the air parcel is in equilibrium at a reference height z ∗ with potential temperature
θ(z, t) = θ0 (z ∗ ) We now have:
dθ0
θ0 (z, t) = θ(z, t) − θ0 (z), with θ0 (z) = θ0 (z ∗ ) + δz.
dz
Substituting this into the z-momentum equation gives:
d2 (z − z ∗ )
r
d2 δz g dθ0 g dθ0
2
= =− δz = −N 2 δz, with N=
dt dt2 θ0 dz θ0 dz
dθ0
the Brunt-Väisälä frequency. For static stability we have N 2 > 0, so dz > 0.

3
We can now look at the convective available potential energy (CAPE) for which we
assume a stationary state and u = v = 0:

dw θ0 dw
=g or w ≈ gB.
dt θ0 dz
We integrate from level of free convection (LFC) to the level of no buoyancy (LNB):
Z zLNB
1 2 2 2 2
(wLNB − wLFC )=g Bdz = CAPE or wLFC = wLNB + 2 · CAPE.
2 zLFC

1.6 Summary of equations in a non-rotating reference frame


See problem 1.7.

1.7 Rotating reference frame and the inertial force


The path of an object, which is a straight line to an observer fixed in space, will appear
curved to an observer standing on the rotating Earth. We travel a distance V t and an angle
Ωt, so the object appears to have traveled s = V Ωt2 , with Coriolis acceleration 2V Ω:
~ × ~v = −2 w|Ω| cos φ − v|Ω| sin φ î − 2 u|Ω| sin φ ĵ + 2 u|Ω| cos φ k̂,
  
−2Ω

~ × ~r) in
with u ≈ v ≈ 10 m/s and w ≈ 1 cm/s. For the centrifugal acceleration (with v = Ω
the x-direction, we get:
(u + r|Ω| sin φ)2 u2
= + r(|Ω sin φ)2 + 2u|Ω| sin φ.
r r
Geostrophic balance is a balance between the inertial force and pressure gradient force:
du 1 ∂p dv 1 ∂p
=− + 2Ωv sin φ = 0, =− − 2Ωu sin φ = 0.
dt ρ ∂x dt ρ ∂y
We can also have hydrostatic balance:
dw 1 ∂p
=− − g = 0.
dt ρ ∂z
Here f = 2Ω sin φ is the Coriolis parameter. In high latitudes we must include tan φ terms.

1.9 Atmospheric composition


The concentration of a specific constituent is expressed as mixing ratio (ppb etc.). Air
is a mixture of principally N2 , O2 , Ar, H2 O and CO2 . Trace constituents that play an
important role in the radiation balance of the atmosphere are O3 , O, H2 , N2 and CH4 . If
the particular constituent has a long residence time in the atmosphere, its mixing ratio by
volume is relatively constant in the lowest 80 km of the atmosphere (homosphere). The
mixing ratio of ozone increases with height above the Earth’s surface, reaching a maximum
value at about 30 km, while the mixing ratio of H2 O has a maximum at the Earth’s surface.
Atmospheric water vapor concentration is expressed in the specific humidity and the
water vapor mixing ratio (compare without vapor itself):
mv ρv ρv ρv
q ≡ qv = = , r ≡ rv = = .
ma ρa ρa − ρd ρd

Precipitation is determined by the saturation water vapor pressure (pe ≡ es ) which is


a function of temperature. as will be discussed in the next section.

4
1.10 Clausius-Clapeyron equation
Water vapor enters the atmosphere by evaporation at the Earth’s surface and is transported
upwards by convection (static instability). This leads to the formation of clouds which
remain suspended until they become heavy enough to fall to the ground as rain. The
Clausius-Clapeyron equation for the saturation water vapor pressure pe (es )is given by:

dpe pe Lv d ln es Lv
= , or = ,
dT Rv T 2 d ln T Rv T
with Rv the specific gas constant for water vapor (461.5 J/kg/K) and Lv the
specific latent heat of evaporation (2.5 · 106 J/kg), which depends weakly on T . For
the equation it is assumed that water vapor is an ideal gas:
e ≡ pv = ρv Rv T.

Here pv is the partial pressure exerted by the water vapor molecules, ρv the density of
water vapor in the atmosphere and es the saturated water vapor pressure. If e > es
(RH> 100%) clouds will be created. We use es,0 = 610.78 Pa for the equilibrium or
saturated vapor pressure at T0 = 0◦ C. Integrating the Clausius-Clapeyron equation, gives
the following expression:
Lv  1 1
ln es − ln es,0 = − .
Rv T0 T

1.11 Water vapor distribution


We assume relative humidity:
e
RH ≡
es
to be constant with height. The atmospheric temperature decreases with increasing height
according to

T (z) = Tg − Γz,
with Γ = cgp ≈ 10−2 K/m the lapse rate and Tg = 300 K the ground temperature. Using
the ideal gas law, the relative humidity and the Clausius Clapeyron equation, we can get
an expression for the scale height of the water vapor density:
  
pv e RHes RHes,0 Lv 1 1
ρv = = = = exp − .
Rv T Rv T Rv T Rv (Tg − Γz) Rv T0 Tg − Γz
We can also write this as
   
RHes,0  Γz  Lv Lv  Γz 
ρv ≈ 1+ exp exp − 1+ .
Rv Tg Tg Rv T0 Rv Tg Tg
Γz
So in the lowest part of the atmosphere (z  30000 m, since Tg ≈ 1), we have
 
Lv Γ
ρv ∼ exp − z .
Rv Tg2

From this equation we identify the scale height (Hv ) which characterizes the exponential
decrease of water vapor density
Rv Tg2
Hv = ≈ 1700 m.
Lv Γ

5
1.14 Conditional instability and equivalent potential temperature
According to the Clausius-Clapeyron equation, an air parcel containing water vapor will
become saturated if its temperature decreases during ascent. The saturation mixing
ratio rs also decreases with height. During condensation (latent) heat is released:

drs
J ≈ −Lv ,
dt
If we assume stationary conditions during ascend, we can write
drs drs drs
≈w < 0, for w > 0 and ≈0 for w ≤ 0.
dt dz dt
We assume a potential temperature of θ(z, t) = θ0 (z) + θ0 (z, t), so
dθ dθ0 dθ0 θ0 dθ0 Lv drs
=w + = N 2w + =− w.
dt dz dt g dt Π dz
We use Π(z, t) = Π0 (z) + Π0 (z, t) ≈ Π0 (z), so:
dθ0 θ0 dθ0 θ0 gLv drs
= − N 2 w if w ≤ 0, = − (N 2 + )w if w > 0.
dt g dt g θ0 Π0 dz
So we get the moist Brunt Väisälä frequency:

2 gLv drs
Nm = N2 + .
θ0 Π0 dz
2
If Nm < 0 it means we have conditional instability. Conditional since the air must be
saturated and ascending. By analogy with dry adiabatic motion, where θ is conserved,
we may define a pseudo- or moist adiabatic process in which a so-called equivalent
potential temperature θe is conserved:

2 g dθe0
Nm = .
θe0 dz
We thus get an equation for θe0 :
d d Lv drs
(ln θe0 ) = (ln θ0 ) + .
dz dz θ0 Π0 dz
So we get a relation between equivalent potential temperature and potential
temperature for a saturated air parcel.
L r 
v s
θe = θ exp .
θΠ
We can also write it for an unsaturated air parcel:
 L r   L r 
v v
θe = θ exp = θ exp .
θΠLCL cp TLCL
Here r is the mixing ratio of the air parcel and ΠLCL and TLCL are respectively the Exner
function and the temperature the parcel would have if it expanded adiabatically to
saturation. The air-parcel is saturated at the lifted condensation level for which we
need the static instability or saturated updraught criterion:
∂θe
< 0.
∂z
When the parcel reaches the LCL, condensation occurs and a cloud forms. The formation
of puffy cumulus clouds due to condensation of water vapor in ascending positively
buoyant air is called cumulus convection.

6
1.15 Dew point temperature and Lifted Condensation Level (LCL)
For the Clausius-Clapeyron equation we had
d ln es Lv Lv  1 1
= or ln es − ln es,0 = − .
dT Rv T 2 Rv T0 T
From this it follows that
h L  T − T i
v 0
es = es,0 exp .
Rv T0 T
With es,0 = 610.78 Pa, Lv = 2.5 · 106 J/kg, Rv = 461.5 J/kg/K and T0 = 273.15 K. An
ascending air parcel, which is not saturated, cools by about 1 K for every 100 m ascent,
assuming it does not mix with the environment. The condensation level is reached when
T = Td but we have to take into account that the vapor pressure inside the parcel changes
as it ascends so its dew point temperature also changes with height. Since the vapor
pressure of an unsaturated air parcel is per definition equal to the saturated vapor pressure
at the dew point temperature of the parcel, we can write:
de Lv e
= .
dTd Rv Td2
If we use p = ρd Rd T and e = ρv Rv Td and assume T ≈ Td , we can write
e ρv Rv rv p Rd 287 ρv
= or e = with  = = = 0.622 and rv = .
p ρd Rd  Rv 461.5 ρd
If the water vapor in the parcel does not condense the ratio e/p in the parcel is constant.
So we get an alternative expression of the Clausius Clapeyron equation
de rv dp rv Lv p dp Lv p
= = or = .
dTd  dTd  Rv Td2 dTd Rv Td2
∂p
If we assume T ≈ Td we can write ∂z = − Rpg
d Td
, so we get

dTd gRv Td gTd


Γdew = − = ≈ ≈ 0.0018 K/m.
dz Lv Rd Lv 
We also used cp dT − αdp = 0 or
dT g
Γdry = − = = 0.01 K/m.
dz cp
We use Ts and Tds for the temperature and dew point temperature of the Earth’s surface
respectively. So:
dT dTd
Ts − zLCL = Tds − zLCL .
dz dz
 dT
d dT −1
zLCL = (Ts − Tds ) − ≈ (Ts − Tds )(−0.0018 + 0.0098)−1 = 125(Ts − Tds ).
dz dz

1.16 Thermodynamic diagram


The tephigram displays lines representing isobaric, isothermal and adiabatic (dry, moist)
processes as well as lines of constant rs (T, p). The isotherms (isopleths of T ) slope upwards
to the right at an angle of about 45◦ so the lines of equal pressure are almost horizontal
(skew T-ln p), also making an angle of 90◦ with the isentropes (isopleths of θ or dry
adiabats). The moist adiabats (isopleths of θe ) are more curved. The isopleths of rs are
nearly straight and can be used to find the LCL and LFC if the height, T and Td at the
point of origin are known.

7
1.17 Sea breeze, inertial oscillation and geostophic balance
We have defined convection as being the result of hydrostatic instability. The atmosphere
is hydrostatically stable if the potential temperature increases with height. The first mode
of convection is called Rayleigh-Benard convection and is buoyancy driven. The second
mode occurs due horizontal contrasts in diabatic heating (sea breeze circulation). We
adopt a model in which the coast lies at x = 0, land at x > 0 and sea at x < 0. We assume
that the pressure gradient parallel to the coast is zero:
∂p ∂p
= A cos(Ωt + φ), = 0,
∂x ∂y
with A = 10−3 Pa/m, Ω = 7.3 · 10−5 s−1 and φ = 0 a phase. We neglect frictional forces;

at the coast it is reasonable to assume w = 0, ∂y = 0, and neglect the advection terms:
∂u A 1 ∂u A ∂v
− f v = − cos(Ωt), so v = + cos(Ωt), + f u = 0.
∂t ρ f ∂t ρf ∂t
We can write this as two second order differential equations:
∂2u AΩ ∂2v Af
2
+ f 2u = sin(Ωt), 2
+ f 2v = cos(Ωt).
∂t ρ ∂t ρ
For boundary condtions, we assume u(0) = 0:
AΩ
u(t) = C1 sin(f t) + 
C2 
cos(f t) + sin(Ωt),
ρ(f 2 − Ω2 )
It consists of 2 frequencies: one associated with thermal forcing and the other the eigen.
AΩ2 A Af
v(t) = C1 cos(f t) + cos(Ωt) + cos(Ωt) = C1 cos(f t) + cos(Ωt).
ρf (f 2 − Ω2 ) ρf ρ(f 2 − Ω2 )
Now using v(0) = 0, we get:
A  Af 
u(t) = Ω sin(Ωt) − f sin(f t) , v(t) = cos(Ωt) − cos(f t) .
ρ(f 2 2
−Ω ) ρ(f 2 2
−Ω )
Wind direction is defined as the direction from which the wind is blowing. A direction of
0◦ corresponds to a wind blowing from the north.

1.18 Stability of geostrophic balance


The sea breeze is perhaps the best example of adjustment to (zonal) geostrophic balance:
du 1 ∂p dv 1 ∂p 1 ∂p
− fv = − , + fu = − , so ug = −

.
dt  ρ ∂x dt ρ ∂y ρ ∂y
So we get for the original equations:
du dy d dM dv
= f , or (u − f y) = = 0, = f (ug − u).
dt dt dt dt dt
So we have conservation of linear momentum per unit mass M. We also have:
dug du
ug (y0 + δy) ≈ ug (y0 ) + (y0 )δy, u(y0 + δy) ≈ u(y0 ) + f δy, with = f.
dy dy
We can now substitute this into the y-equation:

d v(y0 + δy) − v(y0 ) d2 δy    dug 
= = f u (y
g 0 +δy)−u (y
g 0 )− u(y0 +δy)−u(y 0 ) = − f f − δy.
dt dt2 dy
| {z }
F2
du
F 2 > 0 denotes inertial oscillation, while F 2 < 0 denotes an unstable flow. If dyg = 0
we have intrinsic stability due to Earth’s rotation. Here f is the planetary vorticity,
dug
dy the relative vorticity and the difference the absolute vorticity.

8
1.19 Thermal wind
We consider geostrophic and hydrostatic balance:
1 ∂p θ ∂Π ∂Π g ∂2Π ∂ 1 g ∂θ
ug = − =− , =− , so = −g = 2 .
ρf ∂y f ∂y ∂z θ ∂y∂z ∂y θ θ ∂y
We can eliminate the Exner function to get the thermal wind equation:
∂ug 1 ∂Π ∂θ θ ∂2Π ug ∂θ
 g ∂θ
=− − =  − .
∂z f ∂y ∂z f ∂y∂z θ ∂z f θ ∂y
The second term dominates the first one by far. As a linear approximation, we can write
∂ug
ug (z + δz) = ug (z) + δz = ug (z) + uT (z, δz) .
∂z | {z }
thermal wind
∂θ
The geostrophic wind increases if < 0 (troposphere) while it decreases with increasing
∂y
height in the stratosphere. So a jet will exist over the zone with the most intense
temperature gradient, the so-called polar front.

1.20 Stability of thermal wind balance


We again assume geostrophic and hydrostatic balance and ignore the pressure variation
in the x-direction. So we get for the momentum equations:
du ∂Π
 du0 dy dM0
= −θ + f v, so −f ≡ = 0.
dt  ∂x dt dt dt
dv ∂Π θ0 ∂Π0
= −θ − f u, so u0 = − .
dt ∂y f ∂y
dw ∂Π ∂Π0 g
= −θ − g, so =− .
dt ∂z ∂z θ0
With M0 = u0 − f y the linear momentum per unit mass. From this we get:
∂u0 1 ∂Π0 ∂θ0 θ0 ∂ 2 Π u0 ∂θ0 g ∂θ0 ∂M0
=− − =  −

= .
∂z f ∂y ∂z f ∂y∂z  θ0 ∂z f θ0 ∂y ∂z
For (a small) displacement of the parcel, we can write the horizontal velocity as:
∂u0 ∂u0
u(y0 + δy, z0 + δz) = u(y0 , z0 ) + f δy, u0 (y0 + δy, z0 + δz) = u0 (y0 , z0 ) + δy + δz.
∂y ∂z
For the y-momentum equation, we get using the geostrophic balance:
dv
= f (u0 − u).
dt
For displacements of the parcel from equilibrium, we write the following
 
d v(y0 + δy, z0 + δz) − v(y0 , z0 )  
= f u0 (y0 +δy, z0 +δz)−u0 (y0 , z0 )− u(y0 +δy, z0 +δz)−u(y0 , z0 ) .
dt
We can also write this as:
d2 δy  ∂u
0 ∂u0   ∂u0 ∂u0 δz 
= f δy + δz − f δy = −f f − − δy.
dt2 ∂y ∂z ∂y ∂z δy
Now we suppose the parcel moves isentropically
∂θ0 f θ0 ∂u0 ∂u0
∂θ ∂θ  δz 
∂y g ∂z ∂z
δθ = δy + δz = 0 or = − ∂θ0 = θ0
=f .
∂y ∂z δy θ
∂z g N
2 N2

9
We can now use this to get:

d2 δy h ∂u0 f  ∂u0 2 i h f − ∂u0 1  ∂u0 2 i


2 ∂y
= −f f − − δy = −f − δy.
dt2 ∂y N 2 ∂z f N 2 ∂z
The thermal wind balance is stable if
∂u0 f  ∂u0 2 ∂M0  f ∂M 2
0
f− > 2 or f < ,
∂y N ∂z ∂y N ∂z
with M0 = u0 − f y. We also get
∂u0 f  g ∂θ0 2  ∂θ 2
0
 θ 2
0
f− > 2 or < F 2N 2,
∂y N f θ0 ∂y ∂y g
∂u 
with F 2 = f f − ∂yg .

1.21 Baroclinic instability and mid-latitude cyclones


We will look at the baroclinic instability from an energetic point of view. If a warm parcel
ascends and a cold one descends this may produce kinetic energy. If the displacement is
adiabatic, we get
cp
p1 V1γ = p2 V2γ , γ= .
cv
If we compare the masses, we now get (see question 6.3):
1/γ
M2 ρ2 V2 ρ2 p 1 θ1
= = 1/γ
= .
M1 ρ1 V1 ρ1 p 2 θ2

The change in potential energy is given by the final value minus the initial value:
M  θ − θ 
2 1 2
∆P E = (M1 gz2 −M1 gz1 )+(M2 gz1 −M2 gz2 ) = M1 g(z1 −z2 ) −1 = M1 g(z1 −z2 ) < 0,
M1 θ2
so in the first case potential energy is reduced and converted into kinetic energy. If we
want to find the maximum decrease in potential energy, we get:
∂θ ∂θ  ∂θ ∂θ 
dθ = dy + dz or θ2 − θ1 = cos E + sin E L.
∂y ∂z ∂y ∂z
Using this for the expression for the potential energy, we get
M1 gL2 sin E  ∂θ ∂θ 
∆P E = · cos E + sin E .
θ2 ∂y ∂z
Minimizing this equation for E leads to:
∂θ
d 1 ∂θ 1 ∂θ  ∂θ ∂θ ∂y dz
sin(2E) − cos(2E) = cos(2E) +sin(2E) =0 or tan(2E) = − ∂θ = .
dE 2 ∂y 2 ∂z ∂y ∂z ∂z
dy

So the optimum slope of parcel trajectories in which there is a maximum conversion


of potential energy into kinetic energy is half the slope of the isentropic surfaces.
Potential energy is not converted into kinetic energy if warm air descends, while cold air
ascends. This derivation does not take the Coriolis force into account and since this
problem is in 2 dimensions, it implies angular momentum conservation. If zonal symmetry
is broken, the Coriolis force will be balanced by the pressure gradient force.

Equatorward motion is downward and poleward motion is upward so the low


pressure region (trough) and high pressure region (ridge) will tilt westward with height.

10
1.22 Circulation, vorticity and potential vorticity
We introduce the three concepts named in the title, with the most fundamental being the
circulation (defined relative to the earth):
I ZZ
∂v ∂u
Γ= ~u · d~l = ~ × ~u) · dS
(∇ ~ and ω
~ · k̂ = ζ = − .
C S
| {z } ∂x ∂y
ω
~

The area S is enclosed by curve C. ω


~ is the relative vorticity and ζ the vertical
component. The vorticity averaged over area S is the circulation around S divided by S.
We can also look at the absolute circulation (inertial frame):
I ZZ
Γa = ~ua · d~l = ~
ω~a · dS.
C S

We can look at the time rate of this:


I I
da Γa da da
= ~ua · d~l = (~ua · d~l).
dt dt C C dt
Now we want to find an expression of the term inside the integral:

da da ~ua ~ da (d~l) da ~ua ~ da ~ua ~ 1


(~ua · d~l) = · dl + ~ua · = · dl + ~ua · d~ua = · dl + da (~ua · ~ua ).
dt dt dt dt dt 2
So we get
I I h i I d ~u
da Γa da da ~ua ~ 1 a a
= (~ua · d~l) = · dl + da (~ua · ~ua ) = · d~l.
dt C dt C dt 2 C dt

Basically we integrate the equation of motion (neglecting friction):


da ~ua ~ − g k̂.
= −θ∇Π
dt
This gives the Kelvin circulation theorem:
I I I
da Γa da ~ua ~ ~ ~ ~ · d~l.
= · dl = − (θ∇Π + g k̂) · dl = − θ∇Π
dt C dt C C

If we now assume we integrate along a circuit on an isentropic surface:


I
da Γa ~ · d~l = 0,
= −θ ∇Π
dt C

meaning absolute circulation on an isentropic surface is conserved. In reality an


parcel is three-dimensional, so we introduce a cylindrical fluid disk. This means that
the vertically averaged circulation of the cylinder
Z θ+∆θ Z θ+∆θ  Z Z
1 da Γ̄a da ζ̄a S 1 
Γ̄a = Γa dθ obeys ≡ = 0 with ζ̄a = ζa dS dθ.
∆θ θ dt dt S∆θ θ S

Assuming hydrostatic balance, we can write


dp dp dp kg
dm = ρdxdydz = −ρdxdy · = − dS = σdθdS, with = −σg, [σ] = .
ρg g dθ Km2
the isentropic density. If there is no cross-isentropic mass transport, we can write
da m da  Γ̄a  da  ζ̄a S  da  ζ̄a  da
= 0, so = = = (PV) = 0.
dt dt m dt σS∆θ dt σ∆θ dt
So the vertically averaged circulation per unit mass of the cylinder is conserved, also called
potential vorticity. It is materially conserved under adiabatic conditions.

11
For a parcel between fixed isentropic surfaces, we may define the isentropic potential
vorticity in terms of the average isentropic density:

ζ̄a ∆θ ζ̄a
Z̄ = = −g ζ̄a = ∆θ .
σ̄ ∆p ρh
So σ̄ represents the density of air when potential temperature is assumed to be the vertical
coordinate, instead of height. So we have a conservation law of potential vorticity:
dZ̄ dθ
=0 if there is no cross-isentropic mass transport: = 0.
dt dt
The absolute vorticity consists of planetary vorticity and relative vorticity (ζ̂). Fluid
parcels have an absolute vorticity v = ΩR sin φ, around a path with radius r = R sin φ.
The average absolute vorticity of a cylinder of air between two isentropic surfaces can be
written as:

ζ̄a = ζ̄ + f.
We also have the isentropic potential vorticity:
ζθ + f  ∂v   ∂u 
Zθ = , with relative isentropic vorticity ζθ = − .
σ ∂x θ ∂y θ

Likewise we have 1PVU = 10−6 Km2 kg−1 s−1 . Isentropic density in the free atmosphere is
determined principally by absorption and emission of radiation and is positive everywhere.
So potential temperature is a monotonic function of height so it can be used as a vertical
coordinate. The stratosphere has a low isentropic density.

12
Part 2: 1.23-1.25, 1.30-1.32, 1.34-1.35, session: 11-12
1.23: Pressure or potential temperature as vertical coordinate
The material derivative of the zonal wind u = u(x, y, t, θ) can be written as

du ∂u ∂u ∂u dθ ∂u
= +u +v +
dt ∂t ∂x ∂y dt ∂θ

We can interpret dθ
dt as the vertical velocity if we choose θ as a vertical coordinate. Under
adiabatic conditions: dθ
dt = 0, so the motion of an air parcel is restricted to an isentropic
surface. Likewise when pressure is the vertical coordinate we can write u = u(x, y, t, p) as:
du ∂u ∂u ∂u ∂u dp
= +u +v +ω , with ω=
dt ∂t ∂x ∂y ∂p dt
the vertical velocity. In the isentropic coordinate system:
∂p ΠC − ΠA ΠB − ΠA ΠC − ΠB δz ΠB − ΠA ΠC − ΠB δz
transforms to = + = +
∂x δx δx δx δx δx δz δx
If we use Π = Π(x, y, z, t, θ), we get:
 ∂Π   ∂Π   ∂Π   ∂z 
= +
∂x yθt ∂x yzt ∂z xyt ∂x yθt
Using the hydrostatic balance equation, we can write this as:
 ∂Π   ∂Π   ∂z   ∂Ψ 
θ =θ +g ≡ with Ψ = θΠ + gz
∂x yzt ∂x yθt ∂x yθt ∂x yθt
the Montgomery or isentropic streamfunction. We can also derive the hydrostatic
equation in θ coordinates:
∂Ψ ∂Π ∂z
=θ +Π+g = Π.
∂θ ∂θ ∂θ
We get for the horizontal equations of motion (with θ as vertical coordinate):
∂u  ∂u   ∂u  dθ ∂u  ∂Ψ  uv tan φ
+u +v + =− + fv + .
∂t ∂x θ ∂y θ dt ∂θ ∂x θ a
∂v  ∂v   ∂v  dθ ∂v  ∂Ψ  u2 tan φ
+u +v + =− − fu + .
∂t ∂x θ ∂y θ dt ∂θ ∂y θ a
We can also look at an isobaric surface (where the Exner function is constant):
ΠC − ΠA ΠB − ΠA ΠC − ΠB δz
0= = + .
δx δx δz δx
Again using the hydrostatic balance
 ∂Π   ∂z   ∂Φ 
θ =g = with Φ = gz
∂x yzt ∂x ypt ∂x ypt
the geopotential height. The hydrostatic equation in pressure coordinates is:
∂Φ ∂z 1 RT
=g =− =− .
∂p ∂p ρ p
We can derive the continuity/mass conservation equation in isobaric coordinates.
We look at a parcel between pressure levels p0 and p0 + δp with mass:
δxδyδp
δm = ρδxδyδz = − = σδxδyδθ.
g

13
Since mass is conserved, we have:
d d 1  1  dδx dδy dδp 
(δm) = − δxδyδp = δyδp + δxδp + δxδy = 0.
dt dt g g dt dt dt
We can divide this by δm to obtain the continuity equation in isobaric coordinates:
1 dδx 1 dδy 1 dδp  ∂u ∂v  ∂ω
+ + = + + = 0.
δx dt δy dt δp dt ∂x ∂y p ∂p
If the flow is two-dimensional, we have
 ∂u ∂v  ∂Ψ ∂Ψ
+ = 0, with streamfunction : u=− , v= .
∂x ∂y p ∂y ∂x
We can also derive the continuity/mass conservation equation in isentropic
coordinates. We look at a parcel between θ = θ0 and θ = θ0 + δθ with mass:
δxδyδp
δm = ρδxδyδz = − = σδxδyδθ.
g
Since mass is conserved, we have:
d d dσ dδx dδy dδθ
(δm) = (σδxδyδθ) = δxδyδθ + σδyδθ + σδxδθ + σδxδy = 0.
dt dt dt dt dt dt
We can divide this by δm to obtain the continuity equation in isentropic coordinates:
1 dσ 1 dδx 1 dδy 1 dδθ 1 dσ  ∂u   ∂v  ∂  dθ 
+ + + = + + + = 0.
σ dt δx dt δy dt δθ dt σ dt ∂x θ ∂y θ ∂θ dt
We can also write this as:
∂σ  ∂σu   ∂σv  ∂  dθ  ∂σ ~ ~
+ + + σ = + ∇ · I = 0, with
∂t ∂x θ ∂y θ ∂θ dt ∂t
~ = ∂ , ∂ , ∂ , and I~ = σ u, v, dθ .
   

∂x ∂y ∂θ dt

the mass flux consisting of an isentropic (adiabatic) component I~a and a


cross-isentropic (diabatic) component
  ID = σ dθ
dt . The associated cross-isentropic

mass flux divergence is ∂θ σ dθ
dt .

1.24 Potential vorticity equation in isentropic coordinates


We neglect the curvature terms (tan φ), to get for the horizontal equations of motion:
∂u  ∂u   ∂u  dθ ∂u  ∂Ψ 
+u +v + =− + f v + Fx ,
∂t ∂x θ ∂y θ dt ∂θ ∂x θ
∂v  ∂v   ∂v  dθ ∂v  ∂Ψ 
+u +v + =− − f u + Fy .
∂t ∂x θ ∂y θ dt ∂θ ∂y θ
Here Fx and Fy represent frictional forces. We define isentropic relative vorticity as:
 ∂v   ∂u 
ζθ = −
∂x θ ∂y θ
Differentiating the y equation with respect to x and vice versa, we get:
∂ ∂v ∂ ∂u ∂u ∂v ∂ ∂v ∂u ∂u ∂ ∂u ∂v ∂v ∂ ∂v ∂v ∂u ∂ ∂u
− + +u − −u + +v − −v +
∂x ∂t ∂y ∂t ∂x ∂x ∂x ∂x ∂y ∂x ∂y ∂x ∂x ∂y ∂x ∂y ∂y ∂y ∂y ∂y
∂  dθ  ∂v dθ ∂  ∂v  ∂  dθ  ∂u dθ ∂  ∂u  ∂u ∂f ∂Fy ∂v ∂f ∂Fx
+ − − = −f −u + −f −v − .
∂x dt ∂θ dt ∂x ∂θ ∂y dt ∂θ dt ∂y ∂θ ∂x ∂x ∂x ∂y ∂y ∂y

14
We can also write this as
∂  ∂v ∂u  ∂  ∂v ∂u  ∂  ∂v ∂u  dθ ∂  ∂v ∂u  ∂f ∂f
− +u − +v − + − +u +v =
∂t ∂x ∂y θ ∂x ∂x ∂y θ ∂y ∂x ∂y θ dt ∂θ ∂x ∂y θ ∂x ∂y
 ∂u   ∂v ∂u   ∂v   ∂v ∂u   ∂u ∂v  ∂u ∂  dθ  ∂v ∂  dθ  ∂F ∂F
y x
− − − − −f + + − + − .
∂x θ ∂x ∂y θ ∂y θ ∂x ∂y θ ∂x ∂y θ ∂θ ∂y dt ∂θ ∂x dt ∂x ∂y
If we use the material derivative and horizontal divergence in isentropic coordinates:
d ∂  ∂  ∂  dθ ∂  ∂u ∂v 
= +u +v + and δθ = + ,
dt ∂t ∂x θ ∂y θ dt ∂θ ∂x ∂y θ
we get the vorticity equation
d ∂u ∂  dθ  ∂v ∂  dθ  ∂Fy ∂Fx
(ζθ + f ) = −(ζθ + f )δθ + − + − .
dt ∂θ ∂y dt ∂θ ∂x dt ∂x ∂y
Here diabatic heating is the vertical velocity. For an adiabatic inviscid flow, the r.h.s
becomes 0, so the absolute vorticity of a two-dimensional air parcel on an isentropic
surface only changes due to isentropic divergence. Using the continuity equation, we get:
1 dσ  ∂u ∂v  ∂  dθ  1 dσ ∂  dθ 
+ + + = 0, or δθ = − − .
σ dt ∂x ∂y θ ∂θ dt σ dt ∂θ dt
If we use this, we can write
1 d 1 dσ  ζθ + f  ∂  dθ  1 ∂u ∂  dθ  1 ∂v ∂  dθ  1  ∂Fy ∂Fx 
(ζθ +f )−(ζθ +f ) 2 = + − + − .
σ dt σ dt σ ∂θ dt σ ∂θ ∂y dt σ ∂θ ∂x dt σ ∂x ∂y
Introducing the potential vorticity, we can write this as:
dZθ ∂  dθ  1 ∂u ∂  dθ  1 ∂v ∂  dθ  1  ∂Fy ∂Fx  ζθ + f
= Zθ + − + − , with Zθ =
dt ∂θ dt σ ∂θ ∂y dt σ ∂θ ∂x dt σ ∂x ∂y σ
the potential vorticity in a hydrostatic atmosphere. The first term on the r.h.s. is
the stretching or shrinking effect of a vertically confined source of heat, affecting the
distance between isentropic surfaces. The second and third term have a tilting effect. It
is non-zero in a baroclinic atmosphere ( ∂u ∂v
∂θ 6= 0, ∂θ 6= 0) and if the heat source is
horizontally confined. We have material conservation of potential vorticity for an
adiabatic, frictionless and hydrostatic flow:
dZθ d ∂  ∂  ∂  dθ ∂
= 0, with = +u +v + .
dt dt ∂t ∂x θ ∂y θ dt ∂θ
In other words, in the absence of heating and (frictional) forces, potential vorticity is
simply advected like a material conserved chemical tracer.

The 315 K and 350 K isentropic surfaces belong to the middleworld. The 415 K
isentrope is an overworld isentrope, which lies completely in the stratosphere. Neglecting
friction and tilting (much smaller than other terms on r.h.s), we get:

∂Zθ  ∂Z 
θ
 ∂Z 
θ dθ ∂Zθ ∂  dθ 
+u +v =− + Zθ .
∂t ∂x θ ∂y θ dt ∂θ ∂θ dt
∂Zθ
In the polar night, we have dθ J
dt = Π < 0 and we deduce that ∂θ > 0 on NH so the first
term on the r.h.s. is positive, implying that diabatic cooling in the polar night (especially
in the stratosphere) establishes a positive potential vorticity anomaly over the North
Pole. The region of uniform potential vorticity is the surf zone, so mixing by large-scale
turbulent motion dominates over diabatic effects.

15
1.25 Isentropic potential vorticity mixing by planetary waves
Middleworld isentropes probe both the stratosphere and troposphere. The transition is
marked by a strong isentropic potential vorticity (PV) gradient at sub-tropical latitudes.
The tropics have low PV values (< 1) and the extra-tropics have much higher values (> 5).
A narrow transition region in between is called the isentropic tropopause.
Meanders reveal the presence of large scale planetary Rossby waves. In a nearly
adiabatic condition, an isentrope is a material surface. Changes in PV-distribution are
determined by two-dimensional isentropic advection on an isentropic surface. The
meridional isentropic mixing and transport of potential vorticity by planetary Rossby
waves is quantified by expressing the adiabatic potential vorticity equation:
dζa ~ · ~u)ζa = 0.
+ (∇
dt
or written in flux-form
∂ζa  ∂ζ 
a
 ∂ζ 
a
 ∂u   ∂v  ∂ ∂ ~ · J.
~
= −u −v − ζa − ζa = − (uζa ) − (vζa ) = −∇
∂t ∂x θ ∂y θ ∂x θ ∂y θ ∂x ∂y
With absolute vorticity ζa = ζθ + f and vorticity flux vector J~ = (uζa , vζa , 0). We
can interpret the absolute vorticity as the concentration (or density) of a fictitious
potential vorticity substance (PVS). So PV can be regarded as the mixing ratio of
PVS, although it is not dimensionless and can be negative. The adiabatic flux PVS has no
component across isentropic surfaces, but this also holds for diabatic circumstances. So
isentropes are impermeable to PVS. We can look at the zonal mean meridional
flux of PVS. With the zonal mean the average around a full latitude circle:
Z 2π Z 2π
1 1
[v] = vdλ, [ζa ] = ζa dλ.
2π 0 2π 0
The deviation from the zonal mean is denoted by an asterisk:
v = [v] + v ∗ , ζa = [ζa ] + ζa∗ , with [v ∗ ] = [ζa∗ ] = 0.
We can now define the zonal mean meridional isentropic flux of PVS:
h i h i
[vζa ] = [v] + v ∗ [ζa ] + ζa∗ = [v][ζa ] + [v]ζa∗ + v ∗ [ζa ] + v ∗ ζa∗ = [v][ζa ] + [v ∗ ζa∗ ].


So the total meridional flux of PVS consists of the mean meridional circulation and zonal
asymmetries associated with eddies and waves. The latter is also referred to as the
contribution due to meridional PVS-mixing. Similarly, the zonal mean meridional
isentropic flux of mass is:
h i
[vσ] = [v] + v ∗ [σ] + σ ∗ = [v][σ] + [v ∗ σ ∗ ].


The time integrated net total fluxes of PVS and mass (after t = t0 ) are defined as:
Z t Z t
∆PVS(t) = [vζa ]dt0 , [m/s2 ] ∆M(t) = [vσ]dt0 [kg/m/K].
t0 t0

2D circulation of mass is the Brewer-Dobson circulation. We can also derive the


impermeability theorem for PVS in presence of diabatic heating/cooling and friction.
A meridional flux of PVS may be interpreted as a zonal force. An equatorward flux of
PVS is equivalent to a negative value of Fx (westward force per unit mass).
d ∂u ∂ θ̇ ∂v ∂ θ̇ ∂Fy ∂Fx
(ζθ + f ) = −(ζθ + f )δθ + − + − .
dt ∂θ ∂y ∂θ ∂x ∂x ∂y
We can also write this as
∂ζa ∂ζa ∂ζa ∂ζa  ∂u ∂v  ∂u ∂ θ̇ ∂v ∂ θ̇ ∂Fy ∂Fx
= −u −v − θ̇ − ζa + + − + − .
∂t ∂x ∂y ∂θ ∂x ∂y ∂θ ∂y ∂θ ∂x ∂x ∂y

16
In flux-form, this equation becomes (since f is not dependent on θ):

∂ζa ∂(uζa ) ∂(vζa ) ∂  ∂v ∂u  ∂ θ̇ ∂u ∂ θ̇ ∂v ∂Fy ∂Fx


=− − − θ̇ − + − + − .
∂t ∂x ∂y ∂θ ∂x ∂y ∂y ∂θ ∂x ∂θ ∂x ∂y
Now collect all x and y-terms together.
∂ζa ∂(uζa ) ∂Fy ∂  ∂v  ∂(vζa ) ∂Fx ∂  ∂u 
=− + − θ̇ − − + θ̇ .
∂t ∂x ∂x ∂x ∂θ ∂y ∂y ∂y ∂θ
We can write this equation in vector notation:
∂ζa ~ · J~ with ∇ ~ = ∂ , ∂ , ∂ , J~ = uζa − Fy + θ̇ ∂v , vζa + Fx − θ̇ ∂u , 0 .
   
= −∇
∂t ∂x ∂y ∂θ ∂θ ∂θ
We can make use of the divergence theorem
Z Z Z
d ~ · JdV
~
ζa dV = − ∇ =− J~ · ~ndA.
dt V V A

It is usually expressed in the PVS:


Z Z Z
d d
PVS(V ) = σZdV so PVS(V ) = ζa dV = − J~ · ~ndA
V dt dt V A

The component of J~ perpendicular to isentropes is zeo. So the time change of PVS within
a given region bounded by two isentropic surfaces is determined by the flux of PVS at the
lateral boundary of this volume.

1.30 Quasi-balance
An atmosphere is in quasi-balance if it is in quasi-hydrostatic and quasi-geostrophic ba-
lance. Circulation associated with thunderstorms is strongly out of balance, whereas the
one associated with planetary waves is close to geostrophic balance. There is a difference
between Eulerian (fixed point) and Lagrangian (moving) timescales and the (total) lifetime
of the circulation system. If we divide the total lifetime by the Lagrangian time-scale, we
obtain a dimensionless time-scale T which gives an impression of coherence in time.
1  ∂Φ  1  ∂Φ 
vg = , ug = − .
f ∂x p f ∂y p
The nonlinear balance equations in the isobaric coordinate system are
1  ∂u ∂u  g  ∂z 
u +v = vgrad − = vgrad − vg .
f ∂x ∂y f ∂x p
1  ∂v ∂v  g  ∂z 
u +v = −ugrad − = −(ugrad − ug ).
f ∂x ∂y f ∂y p
The left hand side represents the stationary inertial-advective wind (component of the
ageostrophic wind). The vertical component of the relative vorticity and the
horizontal divergence on an isobaric surface are respectively:
 ∂v   ∂u   ∂u   ∂v 
ζ= − , δ= + .
∂x t,p,y ∂y t,p,x ∂x t,p,y ∂y t,p,x
The balance between PGF, Coriolis force and the centrifugal force (gradient
wind/cyclo-geostropic balance) in an axisymmetric vortex in isentropic coordinates is:
u2 ∂Ψ
+ fu = .
r ∂r

17
∂Ψ
The associated wind is the gradient wind. Using ∂θ = Π, we get:

∂Π  2u  ∂u ∂ h  p R/cp i R  p R/cp ∂p 1 ∂p
= f+ = cp = = .
∂r r ∂θ ∂r pref p pref ∂r ρθ ∂r
So we get the relation:
∂p  2u  ∂u
= ρθ f + .
∂r r ∂θ
We also have
∂Z ∂ ζ + f  1 ∂ζ Z ∂σ ∂u u 1 ∂(ur)
= = − + =and ζ = .
∂r ∂r σ σ ∂r σ ∂r ∂r r r ∂r
∂σ 1 ∂2p 1 ∂ ∂p 1 ∂ h  2u  ∂u i
=− =− =− ρθ f + .
∂r g ∂r∂θ g ∂θ ∂r g ∂θ r ∂θ
So in the end, we obtain an expression of the so-called invertibility principle for
potential vorticity (given the potential vorticity distribution, we need to invert it to
obtain the balanced wind):
∂Z ∂  1 ∂(ur)  Z ∂ h  2u  ∂u i  2u 
σ = + ρθ f + , if f+ z > 0.
∂r ∂r r ∂r g ∂θ r ∂θ r
If the inequality is satisfied we have an elliptical partial differential equation. A
potential vorticity anomaly induces a wind field in the same way as an electric charge
induces an electric potential field.

1.31 The ageostrophic wind


The horizontal wind consists of a geostrophic part (v~g ) and an ageostrophic part (v~a ):

~v = ~vg + ~va

The horizontal components of the equation of motion in isobaric coordinates are:


du  ∂z  dv  ∂z 
= fv − g = f va , = −f u − g = −f ua .
dt ∂x p dt ∂y p
Using vector notation, we can write:
d~v dv du 1 d~v 1 d~vg 1 d~va
k̂ × = − î + ĵ = f ua î + f va ĵ = f~va , so ~va = k̂ × = k̂ × + k̂ × .
dt dt dt f dt f dt f dt
So the ageostrophic wind vector is always directed perpendicular and to the left of the
parcel acceleration in NH. The geostrophic and ageostrophic wind can be written as:
g ~ p z with ∇ ~p =
h ∂   ∂  i
~vg = k̂ × ∇ , ,0 .
f ∂x p ∂y p
1 hg
~ p ∂z + d~va .
 i
~va = k̂ × k̂ × ∇
f f ∂t dt
So we get if we use height instead of pressure as vertical coordinate:
1  g ∂ ∂z  1 ∂ ∂p 1  g ∂ ∂z  1 ∂ ∂p
ua = − +. . . = − 2 +. . . , va = − +. . . = − 2 +. . . .
f f ∂x ∂t ρf ∂x ∂t f f ∂y ∂t ρf ∂y ∂t
The first term represents the component of the ageostrophic wind due to the gradient of
the pressure-tendency and is called the isallobaric wind. The component of the sea
breeze perpendicular to the coast is an example

18
1.32 Jet streaks
When the cross-front temperature gradient is not uniform along a front, the thermal wind
balance dictates a local wind maximum over the most intense part of the front. This is
called a jet streak an leads to significant departures from geostrophic balance. The streak
travels very slowly with respect to the Earth’s surface, so air parcels accelerate into it:
du
= f va .
dt
This means an ageostrophic flow is necessary to realize acceleration or deceleration. If
du
dt > 0, as is the case in the entrance region, the ageostrophic motion is directed towards
lower geopotential height. After passing through the region of maximum winds, the parcels
decelerate through ageostrophic motion towards higher pressure. In this way secondary
circulations perpendicular to the jet axis are set up at the entrance and exit regions. These
are thermally direct (upward branch is warm and downward branch is cold) in the
entrance or confluence region and thermally indirect in the exit region of the streak.

Moreover, in the exit region, the secondary circulation is frontogenic, which means that
the temperature gradient across the front increases in time. This is because downward
motion on the warm side of the front causes an adiabatic temperature increase and upward
motion on the cold side of the front causes a temperature decrease.

In the entrance region the situation is reversed, so the secondary flow is frontolytic.
Since the intensity of the jet streak is proportional to the vertically integrated cross-frontal
temperature gradient, the jet-streak progresses forward, along the current. Simultaneously
due the secondary circulation, relatively cool air is advected horizontally into the exit
region of the streak at upper levels, while relatively warm air is advected horizontally into
the exit region at lower levels. So the thermal stratification becomes less stable. In the
entrance region it is the opposite.

The left region is the most favorable area for the formation of precipitation systems that
depend upon upward motion and destabilization of the atmosphere. Vertical motion that
is associated with the secondary circulation in a stably stratified atmosphere is strongly
slanted. The slanted upward motion around jets (or fronts) gives rise to the characteristic
layered clouds such as cirrus and stratus. Conditional hydrostatic instability, on the
other hand, gives rise to the puffy cumulus clouds.

1.34 Linear theory of (planetary) Rossby waves


Jets in the extra-tropics exhibit meanders flowing in the eastward direction, consisting of
troughs and ridges, which have been identified with Rossby waves. They exist due
the variation of the Coriolis parameter with latitude. The horizontal components of the
equation of motion in isobaric coordinates are (neglecting the tan φ terms):
∂u  ∂u   ∂u  ∂u  ∂z 
+u +v +ω = fv − g ,
∂t ∂x p ∂y p ∂p ∂x p
∂v  ∂v   ∂v  ∂v  ∂z 
+u +v +ω = −f u − g .
∂t ∂x p ∂y p ∂p ∂y p
The continuity equation is given by:
 ∂u   ∂v  ∂ω
+ +  = 0.
∂x p ∂y p ∂p
Let us assume an eastward flow with Ū the basic state and u0 (x, y, p, t) a perturbation.
u(x, y, p, t) = Ū + u0 (x, y, p, t) such that u0 (x, y, p, t)  Ū .

19
The basic state is in geostrophic balance:
g  ∂ Z̄ 
Ū = − ,
f ∂y p
with z(x, y, t, p) = Z̄(y) + z 0 (x, y, p, t) and |z 0 |  |Z̄|. Using v(x, y, p, t) = v 0 (x, y, p, t),
ω = 0 and substituting the other assumptions into the equations of motion gives:
∂u0  ∂u0   ∂z 0 
+ Ū = f v0 − g ,
∂t ∂x p ∂x p
∂v 0  ∂v 0   ∂ Z̄  ∂z 0 
+ Ū = −fŪ − f u0 −g
 −g .
∂t ∂x p  ∂y p ∂y p
Using the relative vorticity perturbation
 ∂v 0   ∂u0 
ζ0 = − ,
∂x p ∂y p
we can derive an equation for the time-evolution of the relative vorticity:
∂ζ 0  ∂ζ 0   ∂u0 ∂v 0  df
+ Ū = −f + − v0 .
∂t ∂x p ∂x ∂y p dy
We can introduce a steamfunction ψ 0 such that
∂ψ 0 ∂ψ 0
u0 = − , v0 = , ζ 0 = ∇2h ψ 0 .
∂y ∂x
The meridional gradient of the Coriolis parameter is the beta-parameter:
df 2Ω cos(y/a) 2Ω cos φ
β≡ = = .
dy a a
We can now write the time-evolution of the relative vorticity as follows:
∂∇2h ψ 0  ∂∇2 ψ 0 
h ∂ψ 0
+ Ū +β .
∂t ∂x p ∂x
When the equation is applied to mid-latitudes, it is assumed that β is constant. On an
infinite beta-plane, we assume wave-like solutions of the form
  2π 2π 2π
ψ 0 (x, y, t) = Re Ψei(lx+my−ωt) , with ω = , l= , m= .
T Lx Ly
If we substitute this in the vorticity equation, we get the dispersion relation:
βl
−iωΨ(−l2 − m2 ) + Ū Ψil(−l2 − m2 ) + βΨil = 0 or ω = Ū l − .
l2 + m2
This applies to barotropic planetary waves a.k.a. Rossby waves. The existence of
these waves depends on the β-parameter. The zonal component of the phase speed is:
ω β
cx = = Ū − 2 .
l l + m2
Because β > 0, Rossby waves propagate towards the west (stronger for longer waves) with
respect to the basic state flow. Some ridges seems to block the eastward propagation of the
waves, called blocking high’s. There are two types of Rossby wave breaking:
anticylconic and cyclonic. In the upper troposphere (above 500 hPa) the anti-cyclone
possesses a warm core (blocking high), while the cyclone possesses a cold core (cut-off low).
The velocity of the group of waves is called the wave group velocity:
∂ω β 2βl2 2βl2
cgc = = Ū − 2 + = c x + > cx .
∂l l + m2 (l2 + m2 )2 (l2 + m2 )2
This means that a group of Rossby waves travels faster than individual troughs and ridges.

20
1.35 Thermal wind and temperature advection
The equation representing hydrostatic balance can be written as:
dp pg dz RT
=− or =− .
dz RT d ln p g
Now using the geostrophic wind equation, we get the thermal wind equation:
g  ∂z  ∂vg g ∂2z R ∂T
vg = , or = =− .
f ∂x p ∂ ln p f ∂x∂ ln p f ∂x
g  ∂z  ∂ug g ∂2z R ∂T
ug = − , or =− = .
f ∂y p ∂ ln p f ∂y∂ ln p f ∂y
We can also write this as
∂~vg R ~ p T.
= − k̂ × ∇
∂ ln p f
The thermal wind vector is defined as the vector difference between the geostrophic
winds at two levels. Integrating from p0 to p1 (with p1 < p0 ), we get:
R p1 
Z 
~vT = ~vg (p1 ) − ~vg (p0 ) = − ~ p T d(ln p).
k̂ × ∇
f p0
For the x- and y-component, we get:
R p1
R p1 ∂T
Z p1 d ln p 


R∂ T 
Z
R ∂ p0
p 
0
uT = ug (p1 )−ug (p0 ) = ∂ ln p = T d ln p R p1 =− ln .
f p0 ∂y f ∂y p0 p0
d ln p p f ∂y p p1

R p1 R p1 ∂Φ
R∂ T  p 
0
p0
T d ln p p ∂ ln p
d ln p
vT = ln , with T = R p1 = − 0 R p1 .
f ∂x p p1 p
d ln p R p d ln p
0 0

The thermal wind equation can be used to estimate the mean horizontal temperature
advection in a layer. The thermal wind vector and the isotherms on an isobaric surface are
parallel. A geostrophic wind that turns counterclockwise (backs) with increasing height
implies cold air advection. Conversely, clockwise turning (veering) of the geostrophic
wind with increasing height implies warm air advection by the geostrophic wind.

21
Chapter 3: Hydrostatic balance
3.1 Introduction
Diabatic heating or cooling leads to pressure variations
dΠ RΠ ~ RJ
=− ∇ · ~v + ,
dt cv cv θ
bringing the atmosphere out of hydrostatic balance:
∂Π g
=−
∂z θ
For the achievement of hydrostatic balance in the atmosphere we study the dynamics of
small amplitude perturbations, neglecting Earth’s rotation.

3.2 The Boussinesq approximation


We use the following equations:
dθ J d~v ~ − g k̂, dΠ RΠ ~ RJ
= , = −θ∇Π =− ∇ · ~v + .
dt Π dt dt cv cv θ
Any variable can be expressed as
F (x, y, z, t) = F0 (z) + F 0 (x, y, z, t).
The reference state is time-independent, horizontally homogeneous and in hydrostatic
balance. We define Fm as the average of F0 (z):
1 H
Z
Fm = F0 (z)dz,
H 0
with H the total vertical extent of the domain under study. If we assume
Π(x, y, z, t) = Π0 (z) + Π0 (x, y, z, t) and θ(x, y, z, t) = θ0 (z) + θ0 (x, y, z, t),
with Π0  Π0 and θ0  θ0 and that the reference state is in hydrostatic balance
∂Π0 g
=− ,
∂z θ0
we can apply the Boussinesq approximation to the three momentum equations:

du ∂Π ∂Π0 0 ∂Π

0
= −θ + F rx = −θ0 −θ + F rx
dt ∂x ∂x  ∂x
dv ∂Π ∂Π0 0 ∂Π
0
 + F r
= −θ + F ry = −θ0 −θ y
dt ∂y ∂y  ∂y
dw ∂Π ∂Π0 0 ∂Π

0
∂Π
 0 ∂Π0 ∂Π0 θ0
= −θ −g+F rz = −θ0 −θ −θ
0 −θ0 −g
 +F r z = −θ 0 + g+F rz .
dt ∂z ∂z  ∂z  ∂z ∂z  ∂z θ0
This states that time-dependent variations in θ, resulting from changes in T and p, are
only taken into account in the vertical component of the equations of motion. We call:
θ0
B≡
θ0
the Buoyancy. Where we assume that θ0 ≈ θm = constant. This yields:
du ∂Π0 dv ∂Π0 dw ∂Π0 θ0
= −θm + F rx , = −θm + F ry , = −θm + g + F rz .
dt ∂x dt ∂y dt ∂z θm

22
So the first term on the r.h.s. is linear. We can also apply this to the continuity equation.

cv dΠ ~ J cv h ∂Π0 ∂Π0 ∂Π0 ∂Π0 ∂Π0 i ~ J


+ ∇·~v = or 0
+u +v +w +w + ∇·~v =
RΠ dt Πθ ) ∂t
R(Π0 + 
Π ∂x ∂y ∂z ∂z Π0 θ 0
Using the phase speed of sound waves in the atmosphere (c0 = 300 m/s), we can write:
cv cp θ0 θ0 θ0
= = = 2.
RΠ0 Rγcp T0 γRT0 c0
We can now write:
θ0 h ∂Π0 ∂Π0 ∂Π0 i g ~ · ~v = J
2 +u +v − 2w + ∇
c0 ∂t ∂x ∂y c0 Π0 θ 0
We can ignore the terms between the brackets according to:
∂Π uκΠ ∂p
u = ∼ 10−2 ,
∂x p ∂x
leading to a diagnostic (no time derivative) equation. For shallow and adiabatic motion:

~ · ~v = ∂u + ∂v + ∂w = 0.

∂x ∂y ∂z

3.3 Method of normal modes


We assume a height dependent static distribution of potential temperature θ0 (z). We assume
adiabatic conditions, incompressibility and neglect friction. From

d~v ~ − g k̂, dθ ~ · ~v = 0
= −θ∇Π = 0, ∇
dt dt
we get the following equations:

∂u0 ∂u0 ∂u0 0


∂u ∂Π0
+ u0 + v0 + w0 = −(θ0 + θ0 ) ,
∂t ∂x ∂y  ∂z ∂x
∂v 0 ∂v 0 ∂v 0 0
∂v ∂Π0
+ u0 + v0 + w0 = −(θ0 + θ0 ) ,
∂t ∂x ∂y  ∂z ∂y
∂w0 0
∂w ∂w0 0
∂w ∂Π0 ∂Π
 0 ∂Π0 ∂Π0 θ0
+ u0  + v 0 + w0 = −(θ0 + θ0 ) −θ
0 − θ0 −g
 = −θ 0 + g,
∂t  ∂x ∂y  ∂z ∂z  ∂z ∂z  ∂z θ0
∂θ0 ∂θ0
= −w ,
∂t ∂z
∂u0 ∂v 0 ∂w0
+ + = 0.
∂x ∂y ∂z
Introducing the wave-vector, we can write:

3.5 Buoyancy waves

23
Chapter 9: Baroclinic waves, cyclo and frontogenesis
9.1 Introduction
The formation or genesis of cyclones (cyclogenesis) tends to occur in zones that exhibit
a strong low level horizontal temperature gradient (which occurs in baroclinic zones). In
a statically stable atmosphere vertical motion occurs as a response to the destruction of
thermal wind balance by frontogenesis or frontolysis.

9.3 Quasi-geostrophic approximation


With the ideal gas law, we can write the hydrostatic equation as
∂z RT ∂Φ RT
=− or =− , with Φ = gz the geopotential.
∂p pg ∂p p
Integrating this equation gives the hypsometric equation with
R p1
Z
zT = z2 − z1 = T d(ln p)
g p2
the thickness of an atmospheric layer between the pressure surfaces p1 and p2 . Defining
the layer mean temperature as
R p1
p
T d ln p
< T >= R 2p1 ,
p2
d ln p
we obtain:
R p1 p1 R<T >
zT = < T > ln = H ln , with H=
g p2 p2 g
the layer scale height. Thus, the thickness of an atmospheric layer, bounded by isobaric
surfaces, is proportional to the mean temperature of this layer. From
cp dT − αdp = Jdt
we can derive the thermodynamic energy in pressure coordinates:
∂T  ∂T   ∂T  ∂T αω J
+u +v +ω − = .
∂t ∂x p ∂y p ∂p cp cp
or
∂T  ∂T   ∂T  J ∂T RT T ∂θ
+u +v − Sp ω = with Sp = − + =−
∂t ∂x p ∂y p cp ∂p p θ ∂p
the static stability parameter for the isobaric system. If we have large scale motion
systems in mid-latitudes, we can write for the momentum equations:
du ∂Φ dv ∂Φ
− fv = − , + fu = − .
dt ∂x dt ∂y
The continuity equation in the pressure coordinate system is
∂u ∂v ∂ω
+ + = 0,
∂x ∂y ∂p
The continuity equation, the hydrostatic relation, the thermodynamic equation and the
momentum equation form a closed set of equations, called the primitive equations.
These equations can be further simplified by assuming that the horizontal flow is nearly
geostrophic and that the magnitude of the vertical velocity is much smaller than the
magnitude of the horizontal velocity. We may write
1 ~
~v = ~vg + ~va with ~vg = k̂ × ∇Φ.
f0

24
The Coriolis parameter is written as
df 2Ω cos φ0
f (y) = f (y0 ) + (y − y0 ) = f0 + β(y − y0 ) with β= .
dy a
we define the fluid parcel acceleration as being subject to the ”geostrophic constraint”, i.e.
d~v dg ~vg dg ∂ ~ = ∂ + ug ∂ + vg ∂ .
≈ , where = + v~g · ∇
dt dt dt ∂t ∂t ∂x ∂y
We can now write:
dg ~vg ~ = −(f0 + β(y − y0 ))k̂ × (~vg + ~va ) + f0 k̂ × ~vg =
≈ −f k̂ × ~v − ∇Φ
dt
−(f0 + β(y − y0 ))k̂ × ~va − β(y − y0 )k̂ × ~vg ≈ −f0 k̂ × ~va − β(y − y0 )k̂ × v~g .
Here we assumed f0  β(y − y0 ), which implies that the flow is restricted to a relatively
narrow mid-latitude channel.

9.4 Quasi-geostrophic vorticity- and thermodynamic equations


For the geostrophic wind we have
1 ∂Φ 1 ∂Φ
ug = − , vg = .
f0 ∂y f0 ∂x
We can now write for the vertical component of the relative vorticity:
∂vg ∂ug 1
ζg = − = ∇2 Φ.
∂x ∂y f0
The two components of the quasi-geostrophic momentum equation are:
dg ug d g vg
= f0 va + βyvg , = −f0 ua − βyug .
dt dt
This allow us to write:
dg ζg  ∂u
a ∂va  dg ∇2 Φ ∂ω ∂Φ
= −f0 + − βvg or = f02 −β .
dt ∂x ∂y dt ∂p ∂x
The quasi-geostrophic thermodynamic equation can be written, using the hydrostatic
relation, as
dg ∂Φ
∂p RJ
= −σω − .
dt cp p

9.5 Quasi-geostrophic adjustment to thermal wind balance


Thermal wind balance on the f -plane (f is assumed constant) can be written as
∂vg R ∂T ∂ug R ∂T
=− , = .
∂p pf0 ∂x ∂p pf0 ∂y
We neglect ageostrophic motion and set β and J to zero. This gives:
dg  ∂ug  ∂ug ∂ug ∂vg ∂ug R  ∂ug ∂T ∂ug ∂T 
=− − = − +
dt ∂p ∂p ∂x ∂p ∂y pf0 ∂x ∂y ∂y ∂x
dg ∂T
  ∂ug ∂T ∂vg ∂T
=− − .
dt ∂y ∂y ∂x ∂y ∂y
Substracting these equations gives:

25
dg  f0 p ∂ug ∂T   ∂u ∂T
g ∂vg ∂T  dg  ∂T 
− =2 + = −2Qg2 = −2 .
dt R ∂p ∂y ∂y ∂x ∂y ∂y dt ∂y
Here we used:
∂ug ∂vg
+ = 0.
∂x ∂y
Thermal wind balance is destroyed by frontogenetic or frontolytic processes. These
frontogenetic or frontolytic processes are associated with horizontal shear and/or
confluence of the geostrophic wind. Geostrophic motion tends to destroy thermal wind
balance. We will now see that the role of ageostrophic motion is to restore or “conserve”
thermal wind balance. Allowing for ageostrophic motion, we get instead:
dg  f0 p ∂ug ∂T  f 2 p ∂va pσ ∂ω
− = −2Qg2 + 0 − .
dt R ∂p ∂y R ∂p R ∂y
We now assume that thermal wind balance, which is destroyed by frontogenetic processes,
is maintained by ageostrophic motion, i.e.
f02 p ∂va pσ ∂ω
2Qg2 = − .
R ∂p R ∂y
A similar equation can be deduced from the y-component of the momentum equation:
f02 p ∂ua pσ ∂ω dg  dT   ∂u ∂T
g ∂vg ∂T 
− = 2Qg1 = 2 , with Qg1 = − + .
R ∂p R ∂x dt dx ∂x ∂x ∂x ∂y
If we take the y-derivative of the Qg2 and the x of the Qg1 , we get the omega equation

∂2ω 2R ~ ~  ∂~v
g ~ ∂~vg ~ 
σ∇2 ω + f02 2
=− ∇ · Qg with Q~g = (Qg1 , Qg2 ) = − · ∇T, · ∇T
∂p p ∂x ∂y
the geostrophic Q-vector. The omega equation is a diagnostic equation for the field of
vertical motion in terms of the instantaneous fields of geopotential and temperature.

26
Session 1+2 (part 1)
Problem 1.1: Buoyancy
This problem is concerned with the concept of ‘buoyancy’ and Archimedes principle. In
figure 1.1 we see two identical rectangular beakers. The water level is exactly the same in
both beakers. However, a piece of wood is floating in the right beaker.
a) Which of the two systems (beaker+water+piece of wood) is heavier, or do both systems
have exactly the same weight?
b) Which fraction of the total volume of the piece of wood is below the water-level. Assume
that water has a density (ρ1 ) of 1000 kg/m3 , wood has a density (ρ3 ) of 900 kg/m3 and
that air has a density (ρ2 ) of 1 kg/m3 .

Answer problem 1.1


a) Archimedes Principle states that the block of wood experiences an upward force equal
to the weight of the fluid displaced. Because both systems are in rest, this force must be
balanced by the downward gravitational force on the block:

g(ρ1 V1 + ρ2 V2 ) = gρ3 (V1 + V2 ), with the total volume of the woodblock V1 + V2 = V.


ρ − ρ 
3 2
ρ1 V1 + ρ2 (V − V1 ) = ρ3 V or V1 = V.
ρ1 − ρ2

If ρ2 = 0 then ρ1 V1 = ρ3 V both systems would have equal weight, but in this case we
have ρ1 V1 < ρ3 V because air has been displaced by the wood, so the system with wood
is heavier.

899
b) The volume the water displaces is V1 = 999 V , so the fraction that is below the water
level is 1 − 899
999 ≈ 0.10.

Problem 1.2: Law of adiabatic expansion


Show that for an air parcel with volume V in dry adiabatic circumstances: pV γ = constant
where γ = cp /cv .

Answer problem 1.2


We use dU = mcv dT and dW = −dU = pdV , so

mcv dT + pdV = 0.

We also have p = ρRT or pα = RT and V = mα so


p V Rp
pdα + αdp = RdT or dV + dp = − dV
m m mcv
Rp  R
pdV + V dp = − dV or p 1 + dV + V dp = 0
cv cv
dV dp
γpdV + V dp = 0 or γ + =0
V p
ln V γ + ln p = constant, so ln(pV γ ) = constant or pV γ = constant.

27
Problem 1.3: Hydrostatic equation
Show that the hydrostatic equation (1.7) can be expressed in terms of the Exner function
and potential temperature as
 ∂Π  g  p R/cp
=− , with Π = cp .
∂z x,y,t θ pref

Answer problem 1.3


∂p pref R/cp
We use ∂z = −ρg, p = ρRT and θ = T ( p ) :

∂Π ∂Π ∂p R  p R/cp ∂p RT ∂p ρg g
= = = =− =− .
∂z ∂p ∂z p pref ∂z pθ ∂z ρθ θ

Problem 1.4: Reducing pressure to sea-level


A weather station, situated 150 m above sea level, observes p = 985 hPa and T = 15◦ C.
a) The relation, ∂p pg
∂z = − RT , is usually employed to estimate the sea level pressure. On
which assumptions is this relation based?
b) Assuming a constant dry adiabatic lapse rate of about 6 K km−1 in the hypothetical
atmosphere between the Earth’s surface and sea level, estimate the sea level pressure,
using this relation.
c) Would the estimate be significantly different if this hypothetical atmosphere were assu-
med to be isothermal?

Answer problem 1.4


a) We assume a constant lapse rate: T (z) = Tref − (z − zref )Γ and p = ρRT (ideal gas).

b) We get
dp pg pg
= −ρg = − =−
dz RT (z) R(Tref − (z − zref )Γ)
Z z Z z
dp gdz 0
=− 0
0 p 0 R(Tref − (z − zref )Γ)
z g  z
ln Tref − (z 0 − zref )Γ z

ln(p) z =
ref RΓ ref

  g   
ln p(z) − ln p(zref ) = ln Tref − (z − zref )Γ − ln(Tref )

 Γ(z − zref ) g/RΓ
p(z) = p(zref ) 1 −
Tref

In this case we have zref = 150 m, p(zref ) = 985 hPa and Tref = 288 K. We also know
Γ = 6 · 10−3 K/m, R = 287 J/kg/K, and z = 0 m, so we get p(0) = 1002.65 hPa.

c) The difference is 0.02 hPa which is much smaller than the precision of the given air
pressure at the reference level. So if we assume an isothermal atmosphere, we introduce
a neglectable error.

28
Session 3

Problem 1.7: Continuity equation in terms of the Exner function


Show that for an ideal gas (such as the atmosphere) the continuity equation (1.32) can also
be written as eq. 1.36:
1 dρ ~ · ~u, dh  1 i
~ · ~u.
− =∇ ln θ + 1 − ln Π = ∇
ρ dt dt κ

Answer problem 1.7


We use  1
ln θ + 1 − ln Π = ln(θΠ1−1/κ ).
κ
 κ
p
We know that θΠ = cp T and Π = cp pref . We thus get
pref pref
θΠ1−1/κ = cp T Π−1/κ = cp T c−1/κ
p · = cp(1−1/κ) · .
p ρR
If we substitute this back in the original equation, we get
d h  (1−1/κ) pref i ~
ln cp · = ∇ · ~u.
dt ρR
(1−1/κ) pref
If we suppose A = cp R , we get
d h  A i d  d 1 dρ ~ · ~u.
ln = ln(A) − ln(ρ) = (− ln ρ) = − =∇
dt ρ dt dt ρ dt

Problem 1 box 1.5: Cloud formation and vertical motion


A parcel of air with a temperature of 15◦ C (288 K) and a dewpoint temperature of 2◦ C (275
K) is lifted from the ground at 1000 hPa, without mixing with the environment.
a) Determine its LCL and temperature at that level.

b) The air parcel is lifted a further 200 hPa above its LCL. What is its final temperature
and how much liquid water per kg of air is condensed?

Answer problem 1 box 1.5


a) At the tephigram, we locate the initial state as the intersection of the 15◦ isotherm with
the 1000 hPa isobar. Since Td = 2◦ we can deduce rs (2◦ , 1000 hPa) = 4.4 g/kg. This
is the actual mixing ratio of air at 15◦ and 1000 hPa. Since rs (15◦ , 1000 hPa) = 10.7
g/kg, the air is initially unsaturated. So when the air is lifted, it follows the dry adiabat
(constant θ = 288 K), until it intercepts the line where rs = 4.4 g/kg (LCL). The LCL
happens at about 820 hPa and a temperature of −0.7◦ .
b) When lifted further, it follows the moist adiabat (that passes through 820 hPa and −0.7◦ )
up to 620 hPa. The final temperature is about −15◦ and rs (−15◦ , 620 hPa) = 1.9 g/kg.
So 4.4 − 1.9 = 2.5 g of water must have condensed out of each kilogram of air.

29
Extra problem
Demonstrate that the Brunt-Väisälä frequency is constant in an isothermal atmosphere that
is hydrostatically stable. What is the typical time-period of a buoyancy oscillation in the
isothermal lower stratosphere?

Answer extra problem


We have  p −R/cp ∂θ R
θ=T , so =− θ.
pref ∂p pcp
Using this, with p = ρRT , we get:
g dθ g  ∂θ ∂p  g Rθ g2
N2 = = = ρg = = constant.
θ dz θ ∂p ∂z θ pcp cp T

Session 4

Potential temperature
In many publications we find the statement that θe is the potential temperature an air parcel
would have if all the water vapor were condensed by lifting the parcel to zero pressure. This
would imply that it is conserved. On page 70 of the lecture notes we see that according to
our simplified derivation:
L
d(ln θe ) = d(ln θ) + d(rs )
Πθ
Derive a mathematical expression for θe , for an air parcel that is not necessarily saturated,
assuming that θe is indeed conserved.

Answer
We know that the equivalent potential temperature is conserved:

d(ln θe ) = 0.

We integrate from a reference height with potential temperature θ and mixing ratio r to
the height where all water vapor has been condensed:
Z θe Z 0
0 L θ 
e Lr  Lr 
d(ln θ ) = − d(rs ) so ln = so θe = θ exp .
θ r cp T θ cp T cp T

30
Session 5
Sea breeze model
We have the following equations (with A = 1 hpa/100km):
A Af
u(t) = (Ω sin(Ωt) − f sin(f t)), v(t) = (cos(Ωt) − cos(f t)),
ρ(f 2 − Ω2 ) ρ(f 2 − Ω2 )

a) What is the maximum value of u(t) if f = 0 and ρ = 1.16 kg m−3 ? Do you expect higher
or lower absolute windspeeds if f 6= 0.
b) How will the wind vector rotate in time in the northern hemisphere (clockwise or anti-
clockwise), given that u(0) = 0 and v(0) = 0?
c) If u(0) = 0 and v(0) = 0, at what earliest time is u = 0 and v < 0 at the latitude of
IJmuiden (52◦ N)? At what next point in time is this again the case? Does the time
difference correspond to the frequency of the forcing?

Answer sea breeze model


a) If f = 0, we have

A sin(Ωt) A
u(t) = − , so umax = = 11.8 m/s, and v(t) = 0.
ρΩ ρΩ

If f = 2Ω sin φ 6= 0 , we have
A 
u(t) = 2 Ω sin(Ωt) − f sin(f t) .
ρΩ2 (4 sin φ − 1)

At φ = 30◦ (for example) we have 4 sin2 φ − 1 = 0, meaning u(t) → ∞ (resonance).

b) • If Ω < f we have u < 0 and v > 0,


• if Ω > f we also have u < 0 and v > 0,
so the wind vector will rotate in both cases anti-clockwise (in the northern hemisphere).
c) At 52◦ N, we have
π
f = 2Ω sin φ ≈ 1.57Ω ≈ Ω.
2
First we solve u(t) = 0:
π π 
sin(Ωt) = sin Ωt so Ωt = 1.56, 3.76, 6.08, 8.37 (and so on).
2 2

If we substitute these values into v(t), we get (since f 2 > Ω2 ):


 π · 1.56   π · 3.76 
cos(1.56) − cos > 0, cos(3.76) − cos < 0.
2 2
3.76
So we have u(t) = 0 and v(t) < 0 at t = 7.3·10−5 = 14.3 h. Next, we have
 π · 6.08   π · 8.37 
cos(6.08) − cos > 0, cos(8.37) − cos < 0.
2 2
8.37
So for the next point we have t = 7.3·10 −5 = 31.8 h. We see a time difference of 17.5 h,

unequal to the period of forcing of 24 h.

31
Session 6+7
6.1 Zonal velocity and meridional displacements
How far must an air parcel, which is initially at rest at the equator, be displaced pole ward
in the meridional direction in order to acquire a zonal velocity of 10 m/s? Assume a constant
pressure in the environment of the air parcel.

Answer question 6.1


For this we use conservation of linear momentum per unit mass:
M = u − f y = u − (2Ω sin φ)y = 0,
since the parcel is initially at rest. If the displacement from the equator is small we can use:
y2
r
ua
u − (2Ωφ)y ≈ u − 2Ω = 0 or y = = 660 km.
a 2Ω
This neglects the curvature of the earth, so it is better to use conservation of angular
momentum per unit mass
Ma = u × r = (Ωa cos φ + u)a cos φ
So we get (initial state same as final state)
u
Ωa2 = (Ωa cos φ + u)a cos φ or x2 + x−1=0
Ωa
Solving this for x = cos φ gives:
r
u u2
cos φ = − + + 1 = −0.01 + 1, so a sin φ = 896 km.
2Ωa 4Ω2 a2

6.2 Stability of thermal wind


Show that the stability-criterion can be expressed in terms of the slope of potential tempe-
rature surfaces (isentropes) relative to the slope of momentum surfaces.
∂ug 2 ∂θ0
( ∂z ) dz ∂y
F2 − f2 > 0, Sθ = = − ∂θ0 .
N2 dy ∂z

Answer question 6.2


The flow is stable if
∂ug 2 ∂ug 2
( ∂z ) f 2 θ0 ( ∂z )
F2 − f2 >0 or F 2 − ∂θ0
>0
N2 g ∂z

We also know the following (with Mg = ug − f y):


∂Mg ∂ug
∂y f− ∂y F2
SM = − ∂Mg = ∂ug
= ∂ug
.
∂z ∂z f ∂z
∂ug
Now we use the thermal wind balance: ∂z = − fgθ0 ∂θ
∂y :
0

∂θ0
F2 ∂y
∂ug
+ ∂θ0
>0 or SM − Sθ > 0.
f ∂z ∂z

For stability, the slope of the isentropes must be smaller than the slopes of equal M .

32
6.3 Relation between potential temperature, density and pressure
C 1/γ
Show that θ = ρp .

Answer question 6.3


We use
1−1/γ 1−1/γ
p
ref p  pref 1−1/γ p C
θ=T = = ref p1/γ = p1/γ .
p ρR p ρR ρ

Problem 1.17: Conversion of potential into kinetic energy


Assume that all the potential energy is converted into kinetic energy. Derive an approximate
expression for the maximum possible final velocity of each air parcel after the exchange.
What is this velocity (approximately) if L = 200 km? Assume that

∂θ ∂θ
= −5 · 10−5 K/m, = 5 · 10−3 K/m
∂y ∂y

Answer problem 1.17


We use conservation of energy, m1 ≈ m2 = m, v1 ≈ v2 = v, θ2 = 300 K and a small angle:

1 1 mgL2  ∂θ ∂θ 
m1 v12 + m2 v22 = − sin E cos E + sin E
2 2 θ2 ∂y ∂z
We can also write this as
∂θ
mgL2  ∂θ ∂θ  1 ∂y
mv 2 = − E + E , with E=− ∂θ
= 5 · 10−3 .
θ2 ∂y ∂z 2 ∂z

So we get for the velocity


r
9.81(200 · 103 )2 
v= − · 5 · 10−3 − 5 · 10−5 + (5 · 10−3 )2 = 13 m/s.
300

33
Session 8
Problem 1.18: Wind and vorticity in a tropical cyclone
An approximate model of the horizontal distribution of velocity in a tropical cyclone (box
1.6) is the so-called ‘Rankine vortex’. This is an axisymmetric circular vortex with an
azimuthal velocity, vθ , which is a function of the radius, r (the distance from the center of
the vortex), as follows:  v0 r
R if r ≤ R,
vθ = v0 R
r if r > R.
Here v0 is the maximum wind velocity and R is the radius of maximum wind velocity.
a) Calculate and plot the relative vorticity as a function of r for a Rankine vortex with
v0 = 40 m/s and R = 40 km.
b) Estimate the inertial period ‘inside’ the radius of maximum wind speed.
c) Is the inertial stability in the core of hurricane Alicia large or small?

Answer problem 1.18


a) For the average vorticity in the center of the cyclone, we get (for r ≤ R):
I I
1 ~ vθ 2πr v0 r 2v0
ζ̄(r) = 2 ~u · dl = 2 rdθ = 2
= = ζ0 (r ≤ R).
πr πr πr R R
So the vorticity in the core of the cyclone is constant. For r > R, we have
I
vθ 2πr v0 R 2v0 R
ζ̄(r) = 2 rdθ = 2
= (r > R).
πr πr r r2
We also have
Z R Z r
2v0 R 2π  
ζ̄(r) = = ζ0 rdr + ζ(r)rdr .
r2 πr2 0 R

So we get:
Z R Z r Z r
2v0
v0 R = rdr + ζ(r)rdr = v0 R + ζ(r)rdr .
0 R R
| R {z }
=0

We can also use:

~ × ~u) · k̂ = 1 ∂ (rvθ ).
ζ(r) = (∇
r ∂r
So in conclusion, we get
2v0
ζ(r) = 0 for r > R and ζ(r) = for r ≤ R.
R
b) We look at the inertial stability parameter in the core of the cyclone:

r r
p 2v0 2 · 40
F = f (f + ζ) = f (f + ) = f (f + 3
) = 20 · 10−4 s−1 .
R 40 · 10

This is 20 larger than the inertial stability outside the core (f = 10−4 s−1 ).

c) Even larger, since the radius of maximum wind is 30 km.

34
Session 11
Problem 1.29: Gradient wind balance in a circular vortex
Gradient wind balance in a circular symmetric vortex expressed in pressure coordinates is:

u2  ∂Φ 
+ fu = ,
r ∂r p

where u is the gradient (azimuthal) wind (shown in figure 1.49 for hurricane “Alicia”). This
is a nonlinear equation in u. Write down the solution of this equation for fixed distribution
of the radial gradient of the geopotential. Give an interpretation of this solution.

Answer problem 1.29


We can write this as a quadratic equation in u (with r ≥ 0):
r
2 ∂Φ fr 1 ∂Φ
u + f ru − r = 0, with u=− ± f 2 r2 + 4r .
∂r 2 2 ∂r
2
This has no solution if ∂Φ
∂r < − f 4 r . Otherwise there are four solutions possible:

35
Session 12
Problem 1.30
a) Estimate the zonally averaged, time-mean value of Ū in the western hemisphere for the
month of November 1945 from the figure on the right.
b) Given the value of Ū , for which zonal wavelength do we have cx = 0 (assume m = 0)?
∂ω
c) Calculate the group speed in the x-direction: cgx = ∂l .

d) Estimate the theoretical difference between cx and cgx of waves observed in the Ho-
vemöller diagram on the right (assume m = 0). Is this in accord with observed difference?

Answer problem 1.30


a) We use
β ω β
ω = Ū l − and cx = = Ū − 2 .
l l l

We can calculate β and determine cx and l from the diagram to calculate Ū . We have

df 2Ω cos φ 10−4 1
β= = = 3
= 1.6 · 10−11 .
dy a 6371 · 10 ms

The wave travels 270◦ in 28 days with wavelength 100◦ at 45◦ N:

1.5π · 6371 · 103 · cos 45◦ 2π



= 8.1·10−7 m−1 .

cx = = 8.7 m/s, l= 100
28 · 24 · 3600 360 2π · 6371 · 103 · cos 45◦
·


So filling in numbers gives:

β 1.6 · 10−11
Ū = cx + = 8.7 + = 33.6 m/s.
l2 (8.1 · 10−7 )2

b) If cx = 0, we have
s r
β βλ2 Ū 33.6
Ū = 2 = or λ = 2π = 2π = 9.1 · 106 m.
l (2π)2 β 1.6 · 10−11

c) Using the definition of the group speed, we get:

∂ω β 2β
cgx = = Ū + 2 = cx + 2 = 8.7 + 48.8 = 57.5 m/s.
∂l l l

d) Theoretical difference: cgx − cx = l2 = 48.8 m/s.

From the diagram we determine that the group speed travels 200◦ in 9 days, so
400
360 π · 6371 · 103 · cos 45◦
cgx = = 20.2 m/s.
9 · 24 · 3600
Observed difference: cgx − cx = 20.2 − 8.7 = 11.5 m/s.

36
Session 13 (part 2)
Exercise 3.1: The peculiar dispersion of buoyancy waves
The dispersion relation for buoyancy waves in the plane y = 0 (in which case β = 0) is

α2
ω2 = N2
α2 + γ2

a) Show that the vertical component of the group velocity of buoyancy waves is opposite to
the vertical component of the phase velocity of buoyancy waves (see Box 3.1).
b) Show that group velocity is directed parallel to the direction of movement of the oscil-
lating air parcels (or lines of constant phase). HINT: first show, with the equation of
continuity, that the air parcels oscillate perpendicular to the wave vector.

Answer Exercise 3.1


a) We use
αN
ω=p
α2 + γ 2

so we need to show cz and cg,z have an opposite sign:

ω αN ∂ω γαN
cz = = p , cg,z = =− 2 .
γ γ α2 + γ 2 ∂γ (α + γ 2 )3/2

b) Using the wave vector    


u U
= e(αx+γz−ωt)
w W

and the continuity equation, we see that the position and wave vector are perpendicular:

αU + γW = 0, since ~k · U
~ = 0.

Now we need to show that the group speed is perpendicular to the wave vector. For
this we also need the group speed in the x-direction:

∂ω N α2 N γ2N
cg,x = =p − 2 = .
∂α α2 + γ 2 (α + γ 2 )3/2 (α2 + γ 2 )3/2

So we see that the group speed is perpendicular to the wave vector:

γ 2 αN γ 2 αN
~cg · ~k = − 2 = 0.
(α2+γ )2 3/2 (α + γ 2 )3/2

Since the group speed is perpendicular to the wave vector and the wave vector is
perpendicular to the position vector, we conclude the group velocity is parallel to the
movement of the oscillating air parcel.

37
Exercise 3.2: Two-dimensional buoyancy- and acoustic- waves
In this exercise we repeat the analysis of section 3.3 without assuming incompressibility.
Assume that the atmosphere is homogeneous in the y-direction (two-dimensional motion)
and we have an adiabatic, frictionless and inertial motion.

a) Write down the x- and z-components of the equation of motion with u, w, Π and θ as
unknown variables and t, x and z as independent variables. Linearise this system of
equations around the steady hydrostatic state of rest by assuming
• ∂Π0
∂z = − θg0
• Π(x, z, t) = Π0 (z) + Π0 (x, z, t), with Π0 << Π0
• θ(x, z, t) = θ0 (z) + θ0 (x, z, t), with θ0 << θ0
0 0
• u(x, z, t) = 
u
0 + u (x, z, t) and w(x, z, t) = 
w
0 + w (x, z, t)

and neglecting products of perturbation quantities. Assume that θ0 ≈ θm = constant in


the pressure gradient and buoyancy term (shallow Boussinesq approximation).

dθ0
b) Write down the linearised potential temperature equation assuming dz = Γ = constant.
Close the set of equations with the linearised continuity equation.
c) Assume Π0 ≈ Πm = constant and θ0 ≈ θm = constant in the continuity equation.
Remember Γ = constant in the linearised potential temperature equation. Assume an
infinite domain and substitute a solution of the form
 0  
u U (z)
w0  W (z) i(αx−ωt)
 0 = 
 θ   θ(z)  e ,

Π0 P (z)

with α the wavenumber and ω the frequency. Find a differential equation of the form

d2 W dW
2
+ s(z) + r(z)W = 0.
dz dz
d) The above equation can be transformed to an equation of the form

d2 W 2
h1 Z i 1 1 ds
2
+[m(z)] W = 0, with W = W exp s(z)dz and m2 (z) = r(z)− s(z)2 − .
dz 2 4 2 dz

Write down the expression for m(z).



e) Now, assume W∼Re exp(iγz) (with γ the vertical wave-number). Find the dispersion
relation of the wave-like solution to the linear system. Investigate the case Γ = 0 (acoustic
waves) and the case cm → ∞ (buoyancy-waves) where c2m = RΠcmv θm .
f) Compare your results with the analysis of sections 3.3 to 3.6.

38
Answer exercise 3.2
a) Using an adiabatic, frictionless and inertial motion, we use three equations for the three
variables ( ~v , θ and Π) as in the lecture notes:

d~v ~ − g k̂ , dθ J dΠ RΠ ~ RJ
= −θ∇Π =  = 0, =− ∇ · ~v +  .
|dt {z } dt
| Π
{z } |
dt cv
{z
cv θ
 }
momentum equation potential temperature continuity equation
equation

Using u0 = 0 and w0 = 0, we substitute


0
• u(x, z, t) = 
u
0 + u (x, z, t)
0
• w(x, z, t) = 
w
0 + w (x, z, t)

• θ(x, z, t) = θ0 (z) + θ0 (x, z, t), with θ0 << θ0


• Π(x, z, t) = Π0 (z) + Π0 (x, z, t), with Π0 << Π0

into the x and z momentum equation. This gives, by neglecting y-derivatives and
products of perturbations, the following expressions:

∂u0 ∂u0 0
∂u ∂Π0 ∂Π0
+ u0 + w0 = −θ0 − θ0 ,
∂t ∂x  ∂z ∂x ∂x
∂w0 0
∂w 0
∂w ∂Π0 ∂Π0 ∂Π0 ∂Π0
+ u0  + w0 = −θ0 − θ0 − θ0 − θ0 − g.
∂t  ∂x  ∂z ∂z ∂z ∂z ∂z

Implementing ∂Π0
∂z = − θg0 , we get for the momentum equations:

∂u0 ∂Π0
= −θ0 , (1)
∂t ∂x
∂w0 ∂Π0 θ0
= −θ0 + g. (2)
∂t ∂z θ0


b) Using dt = 0 and the definition of the material derivative, we get:

∂θ0 ∂θ0 ∂θ0 ∂θ0


+ u0 + w0 + w0 = 0.
∂t ∂x ∂z ∂z
∂θ0
Products of perturbations have been neglected. Using ∂z = Γ, we get:

∂θ0
= −w0 Γ. (3)
∂t
Looking at the continuity equation and using the material derivative, we get:

∂Π0 0
∂Π 0
∂Π ∂Π0 RΠ0  ∂u0 ∂w0  RΠ0  ∂u
0
∂w
 0

+ u0  + w0 + w0 =− + −  + .
∂t  ∂x  ∂z ∂z cv ∂x ∂z cv ∂x ∂z

Where we neglected products of perturbations and used Π0 << Π0 . Again using


∂Π0 g
∂z = − θ0 , we get for the continuity equation:

∂Π0 g RΠ0  ∂u0 ∂w0 


= w0 − + . (4)
∂t θ0 cv ∂x ∂z

39
c) Substituting the wave-like solution in (1), (2), (3) and (4), gives respectively:
αθm P
−iωU = −θm iαP or U= , (5)
ω
dP Θ
−iωW = −θm + g, (6)
dz θm
ΓW
−iωΘ = −W Γ or ,Θ= (7)

g RΠm  dW 
−iωP = W− iαU + . (8)
θm cv dz

Taking the z-derivative of (8), we get:

dP g dW RΠm  dU d2 W 
−iω = − iα + .
dz θm dz cv dz dz 2
RΠm
Now we divide the equation by cv and take all terms to the left-hand side:

d2 W dU iωcv dP cv g dW
+ iα − − = 0.
dz 2 dz RΠm dz RΠm θm dz
dU αθm dP
Now we take the z derivative of (5) to substitute dz = ω dz :

d2 W iα2 θm dP iωcv dP cv g dW
2
+ − − = 0.
dz ω dz RΠm dz RΠm θm dz
dP
Next using (6) and (7), we can find an expression for dz :

dP iω Γg
= W+ 2
W.
dz θm iωθm

Using this, and collecting terms from second to zeroth order, we get:

d2 W cv g dW iα2 θm  iω Γg  iωcv  iω Γg 
− + + W − + W = 0.
dz 2 RΠm θm dz ω θm 2
iωθm RΠm θm 2
iωθm

This can also be written as follows:

d2 W cv g dW  2 α2 Γg cv ω 2 cv Γg 
− + −α + + − W = 0.
dz 2 RΠ θ dz ω 2 θm RΠm θm RΠm θm2
| {zm m} | {z }
s(z) r(z)

So in conclusion we have

cv g α2 Γg cv ω 2 cv Γg
s(z) = − = constant, r(z) = −α2 + 2
+ − 2
. (9)
RΠm θm ω θm RΠm θm RΠm θm

d) Looking at the expression of m2 (z), we need (9):

1 1 ds α2 Γg cv ω 2 cv Γg c2v g 2
m2 (z) = r(z) − s(z)2 − = −α2 + 2 + − 2
− . (10)
4 2 dz ω θm RΠm θm RΠm θm 4R2 Π2m θm2

The question asked for an expression of m(z), which is just the square root of (10), but
since we will use m2 (z) in part e) I wrote out the latter.

40
e) Using m2 (z) from part d) and the information in the question:
d2 W
+ [m(z)]2 W = 0

and W = Re exp(iγz) ,
dz 2
RΠm θm
we substitute cv = c2m in m2 (z) to obtain:

α2 Γg ω2 Γg g2
−γ 2 − α2 + 2
+ 2 − 2 − 4 = 0.
ω θm cm cm θm 4cm

We can multiply this equation with c2m ω 2 to obtain:

α2 c2m gΓ gΓ 2 g2
−γ 2 c2m ω 2 − α2 c2m ω 2 + + ω4 − ω − 2 ω 2 = 0.
θm θm 4cm

Collecting ω 4 , ω 2 and constant terms, we get:


 gΓ g2  α2 c2m gΓ
ω4 + − γ 2 c2m − α2 c2m − − 2 ω2 + = 0.
θm 4cm θm

Using N 2 = θm , we get the dispersion relation
 g2 
ω 4 − (γ 2 + α2 )c2m + N 2 + 2 ω 2 + N 2 α2 c2m = 0. (11)
4cm

We first look at Γ = 0 (or N = 0):


 g2 
ω 4 = (γ 2 + α2 )c2m + 2 ω 2 .
4cm

Dividing by ω 2 gives the dispersion relation for acoustic waves:

g 2
ω 2 = (γ 2 + α2 )c2m + 2 . (12)
4cm

With scale analysis we can ignore the last term (just like in the book). Looking at the
group speed (cg,x , cg,z ) and phase speed (cx , cz ) we can find an expression for c2m .
p
ω cm γ 2 + α2 ∂ω ∂ p cm α c2
= m.

cx = = , cg,x = = cm γ 2 + α2 = p
α α ∂α ∂α γ 2 + α2 cx
p
ω cm γ 2 + α 2 ∂ω ∂ p cm γ c2
= m.

cz = = , cg,z = = cm γ 2 + α2 = p
γ γ ∂γ ∂γ γ 2 + α2 cz

Looking at these expressions, we conclude:

c2m = cg,x cx = cg,z cz .

Next we look at cm → ∞. We divide (11) by c2m to obtain:

ω4  2 2 N2 g  2
− (α + γ ) + + ω + N 2 α2 = 0.
c2m c2m 4c4m

Now we can apply cm → ∞ to get the dispersion relation for buoyancy waves:

α2
−(α2 + γ 2 )ω 2 + N 2 α2 = 0 or ω2 = N 2. (13)
α2 + γ 2

Since s(z) = − RΠcm


vg g
θm = − c2 → 0, we get W = W .
m

41

f) • We ignore ∂y terms and movement in that direction (v = 0), which is not done in
section 3.3 and 3.5.

• We have one extra term in the continuity equation, since we do not assume
incompressibility (as in the lecture notes).

• Looking at r(z) and s(z), we notice r(z) has two the same and two different terms
while s(z) is totally different, because we do not assume incompressibility.

• Dispersion relation for acoustic has one extra term since in the lecture notes there
is only vertical motion (α = 0).

• Dispersion relation for buoyancy waves lacks the β 2 term compared to the lecture

notes since we ignore ∂y terms.

42
Session 14
Exercise 9.3: Divergence, vertical motion and phase speed
Given the following expression for the geopotential field:
n h  πp  i 1 o
Φ = Φ0 (p) + U f0 − y cos + 1 + sin l(x − ct) ,
pref l

where Φ0 = Φ0 (p), c is the phase speed, U has the dimensions of velocity, l is a zonal wave
number, pref = 1000 hPa and f = f0 + βy is the Coriolis parameter (f0 and β constants).

a) Obtain the horizontal divergence field, which is consistent with this Φ-field. Use the
quasi-geostrophic vorticity equation in pressure coordinates:
dg ζg  ∂u
a ∂va 
= −f0 + − βvg ,
dt ∂x ∂y

where ua and va are the ageostrophic horizontal wind components and

dg ∂ ∂ ∂ ∂vg ∂ug 1 ∂Φ 1 ∂Φ
= + ug + vg , ζg = − , ug = − and vg = .
dt ∂t ∂x ∂y ∂x ∂y f0 ∂y f0 ∂x

b) Derive an expression for ω(x, y, p, t) by integrating the continuity equation,

∂u ∂v ∂ω
+ + = 0,
∂x ∂y ∂p

upwards from the earth’s surface. Assume that the vertical velocity at the earth’s
surface (i.e. at p = pref ), ω(x, y, pref )=0.

c) Under which condition is the expression for ω(x, y, p, t), derived in 1(b), consistent with
the boundary condition ω = 0 at p = 0? Give an interpretation of this expression.

Answer exercise 9.3


a)
b)
c)

43

You might also like