Thermodynamics Lectures, MIT, 2015 65

 
  The   adiabatic   lapse   rate   is   (a)   proportional   to   the   thermal   expansion  
coefficient   and   (b)   is   independent   of   the   fluid’s   compressibility.     Indeed,   the  
adiabatic  lapse  rate  changes  sign  at  the  temperature  of  maximum  density  (where  
α t , α θ  and   α Θ  all  change  sign)  whereas  the  compressibility  is  always  positive.    
This  change  in  sign  of  the  adiabatic  lapse  rate   Γ  occurs  even  though  the  work  
( )
done  by  compression,   p + P0 dv ,  is  always  positive  (for  a  increase  in  pressure).      
  Hence,  in  cold  lakes  where  the  thermal  expansion  coefficient  is  negative,  the  
adiabatic  lapse  rate  is  negative,  so  that  as  the  pressure  is  increased  adiabatically,  
the  in  situ  temperature  actually  decreases!    The  adiabatic  lapse  rate   Γ  represents  
that  change  in  temperature  that  is  required  to  keep  the  entropy  (and  also   θ  and  
Θ )   of   a   seawater   parcel   constant   when   its   pressure   is   changed   in   an   adiabatic  
and  isohaline  manner.    
  The  traditional  explanation  has  found  its  way  into  our  textbooks  because  it  
works   perfectly   for   a   perfect   gas;   the   missing   term   that   we   identified   just  
happens  to  be  zero  for  a  perfect  gas,  but  it  is  the  dominant  term  for  a  liquid.      
  Remember,  the  adiabatic  lapse  rate  has  nothing  whatsoever  to  do  with  the  
( 0 ) dv   work   done   in   changing   the   internal   energy   of   a   fluid   parcel.     This  
p + P
explanation  is  wrong  even  for  a  perfect  gas  (where  you  get  the  right  answer  for  
the  wrong  reason);  for  a  liquid  it  is  wrong  by  orders  of  magnitude.      
 

65
Thermodynamics Lectures, MIT, 2015 66

The  adiabatic  lapse  rate  and  the  potential  temperature  of  ice  Ih  
 
Ice  Ih  is  the  form  of  ice  with  hexagonal  packing  of  the  water  molecules.    This  is  
the  form  of  ice  that  is  found  in  the  range  of  temperatures  and  pressures  found  on  
planet  earth.  
The  adiabatic  lapse  rate  is  equal  to  the  change  of  in  situ  temperature  experienced  
when   pressure   is   changed   while   keeping   entropy   (and   salinity)   constant.     This  
definition  applies  separately  to  both  ice  and  seawater  (where  one  needs  to  keep  
not   only   entropy   but   also   Absolute   Salinity   constant   during   the   pressure  
change).    In  terms  of  the  Gibbs  functions  of  seawater  and  of  ice  Ih  the  adiabatic  
lapse  rates  of  seawater   Γ  and  of  ice   Γ Ih  are  expressed  respectively  as    

Γ =
∂t
=
∂t
=
∂t
= −
gTP T + t αt
= 0
(,  
) (Ice_1)  
∂P S ∂P S ∂P S gTT ρ cp
A ,η A ,Θ A ,θ

and    

Γ Ih
=
∂t
=
∂t g Ih
= − TP =
(T 0 )
+ t Ih α tIh
,   (Ice_2)  
∂P η ∂P θ Ih Ih
gTT ρ Ih
c Ih
p

where   α t  and   α tIh  are  the  thermal  expansion  coefficients  of  seawater  and  ice  Ih  
respectively  with  respect  to  in  situ  temperature.        
The  adiabatic  lapse  rates  of  seawater  and  of  ice  are  numerically  substantially  
different   from   each   other.     The   thermal   expansion   coefficient   of   ice   does   not  
change   sign   as   does   that   of   seawater   when   it   is   cooler   than   the   temperature   of  
maximum  density,  and  the  specific  heat  capacity  of  ice   c Ih p  is  only  approximately  
52%  that  of  seawater   c p .      
Figure   Ice_1(a)   below   shows   the   ratio   Γ Γ Ih   of   the   adiabatic   lapse   rates   of  
seawater   and   ice   at   the   freezing   temperature,   as   a   function   of   the   Absolute  
Salinity   of   seawater   and   pressure.     For   salinities   typical   of   the   open   ocean,   the  
ratio   Γ Γ Ih   is   about   0.1   indicating   that   the   in   situ   temperature   of   ice   varies   ten  
times  as  strongly  with  pressure  when  both  seawater  and  ice  Ih  are  subjected  to  
the  same  isentropic  pressure  variations.    This  must  be  taken  into  account  when  
considering  the  vertical  motion  of  frazil  ice  and  the  vertical  motion  of  seawater  
and  frazil  ice  mixtures.      

 
Figure  Ice_1.    (a)  The  ratio  of  the  adiabatic  lapse  rates  of  seawater  and  of  ice  Ih,  
Γ Γ Ih ,   at   the   freezing   temperature.     (b)   The   difference   (in   °C )   between   the  
potential  temperatures  of  seawater   θ  and  of  ice   θ Ih  for  parcels  of  seawater  and  
ice  whose  in  situ  temperature  is  the  in  situ  freezing  temperature.      
 

66
Thermodynamics Lectures, MIT, 2015 67

The  freezing  temperature  of  ice  in  contact  with  seawater    

The   freezing   of   seawater   occurs   at   the   temperature   tfreezing   at   which   the  
chemical  potential  of  water  in  seawater   µ W  equals  the  chemical  potential  of  ice  
µ Ih .     Hence   the   freezing   temperature   tfreezing   is   found   by   solving   the   implicit  
equation    
( ) (
µ W SA , tfreezing , p = µ Ih tfreezing , p ,   ) (Ice_3)  

or  equivalently,  in  terms  of  the  two  Gibbs  functions,    

( )
g SA , tfreezing , p − SA g S
A
(S A , tfreezing , p ) ( )
= g Ih tfreezing , p .   (Ice_4)  

The   freezing   in   situ   temperatures   derived   from   Eqn.   (Ice_4)   were   converted   to  
the  Conservative  Temperature  at  which  air-­‐‑free  seawater  freezes  and  are  shown  
in  Figure  Ice_2(a)  as  a  function  of  pressure  and  Absolute  Salinity.    You  can  see  
that   whether   a   water   molecule   prefers   to   remain   in   seawater   or   prefers   to   join  
the   solid   matrix   of   water   molecules   called   “ice”   depends   on   the   salinity   of   the  
seawater  and  on  pressure.      
 

Figure   Ice_2.     (a)   The   Conservative   Temperature   (in   °C )   at   which   air-­‐‑free  

seawater   freezes   as   a   function   of   pressure   and   Absolute   Salinity.     (b)   The  
difference  between  the  freezing  Conservative  Temperature  derived  from  EOS-­‐‑80  
and  that  of  TEOS-­‐‑10,  with  the  contours  being  in  mK.      
 
 

67
Thermodynamics Lectures, MIT, 2015 68
 
The  figure  below  is  another  way  of  plotting  the  freezing  temperature  of  seawater.    

 
 
Figure   Ice_3.       The   in-­‐‑situ   freezing   temperature   (in   °C )   of   air-­‐‑free  
seawater   as   a   function   of   pressure   (in   dbar )   and   Absolute   Salinity,  
determined  from  the  equilibrium  freezing  condition  Eqn.  (Ice_4).    In  
the   context   of   sea   ice,   the   in   situ   temperature   is   the   temperature   of  
both  the  pure  ice  Ih  phase   t Ih  and  of  the  trapped  pockets  of  brine.      
 
When   discussing   the   thermodynamic   equilibrium   between   seawater   and   ice  
in   the   oceanographic   context   there   are   two   common   situations.     One   is   called  
“sea   ice”   where   there   are   trapped   pockets   of   seawater   inside   a   matrix   of   ice  
crystals.    This  trapped  seawater  is  commonly  called  “brine”  as  its  salinity  can  be  
very  large  when  the  temperature  is  cold.      
The  other  situation  is  where  there  are  small  ice  crystals  (frazil)  are  suspended  
in  a  much  larger  volume  of  seawater  so  that  the  mass  fraction  of  ice  is  small.      
In   both   situations   the   ice   and   the   seawater   exist   in   thermodynamic  
equilibrium,   so   that   their   in   situ   temperatures   are   the   same.     However,   as   we  
have   seen,   the   potential   temperatures   of   the   ice   and   seawater   phases   are  
different  (unless  the  sea  pressure  is  zero).      
 
 

68
Thermodynamics Lectures, MIT, 2015 69

Melting  of  ice  into  seawater    

The   First   Law   of   Thermodynamics   says   that   when   a   process   occurs   at  
constant   pressure,   and   without   any   external   input   of   energy   with   the  
environment,  then  total  enthalpy  is  conserved.      
So  the  conservation  equations  for  mass,  salt  and  enthalpy  during  an  adiabatic  
melting  event  at  constant  pressure  are    
f i
mSW = mSW + mIh ,   (Ice_5)  
f
  mSW SAf = mSW
i
SAi ,   (Ice_6)  
f
  mSW hf = mSW
i
hi + mIh h Ih .   (Ice_7)  
The  superscripts  i  and  f  stand  for  the  “initial”  and  “final”  values,  that  is,  the  
values  before  and  after  the  melting  event,  while  the  subscripts  SW  and  Ih  stand  
for  “seawater”  and  “ice  Ih”.    The  mass  of  ice   mIh  is  assumed  to  melt  completely,  
so  in  the  final  state  there  is  no  ice  as  all;  it  is  all  seawater.      
The   mass,   salinity   and   enthalpy   conservation   equations   (Ice_5)   –   (Ice_7)   can  
be  combined  to  give  the  following  expressions  for  the  differences  in  the  Absolute  
Salinity  and  the  specific  enthalpy  of  the  seawater  phase  due  to  the  melting  of  the  
ice,    

( m
)
  SAf − SAi = − fIh SAi = − wIh SAi ,  
mSW
(Ice_8)  

)=( ) ( h − h ) ,  
f
S − SAi
(h f
−h i
) = −w Ih
(h − h
i Ih A

SAi
i Ih
(Ice_9)  

where  we  have  defined  the  mass  fraction  of  ice  Ih   wIh  as   mIh mSW f
.    The  initial  
and   final   values   of   the   specific   enthalpy   of   seawater   are   given   by  
( ) ( ) ( ) (
hi = h SAi ,t i , p = ĥ SAi ,Θ i , p  and   hf = h SAf ,t f , p = ĥ SAf ,Θ f , p .    These  equations  )
are  illustrated  in  the  following  diagram    

 
Figure   Ice_4(a).     This   Absolute   Salinity   –   enthalpy   diagram   illustrates   Eqns.  
(Ice_8)   and   (Ice_9)   which   embody   the   conservation   of   Absolute   Salinity   and  
enthalpy  when  ice  Ih  melts  into  seawater  at  fixed  pressure.    The  initial  values  of  
the   Absolute   Salinity   and   enthalpy   of   seawater   and   of   ice   Ih   are   shown   by   the  
two   solid   dots,   and   the   final   values   of   Absolute   Salinity   and   enthalpy   of   the  
seawater   after   the   ice   has   melted   are   shown   by   the   four   open   circles   (for   four  
different   values   of   the   ice   mass   fraction   wIh ).     These   final   values   lie   on   the  
straight  line  on  this  diagram  that  connects  the  initial  values  (the  solid  dots).      
 

69
Thermodynamics Lectures, MIT, 2015 70
 

 
 
Figure   Ice_4(b).     The   same   initial   and   final   data   are   shown   on   the   Absolute  
Salinity   –   in   situ   temperature   diagram.     Note   that   the   final   points   (the   open  
circles)  do  not  lie  on  the  straight  line  connecting  the  initial  points  (the  solid  dots)  
on  this  diagram.      
 
The   final   values   of   Absolute   Salinity,   SAf ,   and   enthalpy,   hf ,   given   by   Eqns.  
(8)  and  (9)  are  illustrated  in  Fig.  Ice_4(a)  for  four  different  values  of  the  ice  mass  
fraction   wIh   (the   four   open   circles).     These   final   values,   SAf , hf ,   lie   on   the   ( )
straight  line  on  the  Absolute  Salinity  -­‐‑  enthalpy  diagram  connecting   SAi , hi  and   ( )
( )
0, h Ih .     The   fact   that   the   same   data   does   not   fall   on   a   straight   line   on   the  
Absolute  Salinity  –  in  situ  temperature  diagram  in  Fig.  Icw_4(b)  nicely  illustrates  
that  temperature  is  not  conserved  when  melting  occurs.      
 
 

The  linearized  expression  for  the   SA − Θ  ratio  when  melting  occurs    

Here   we   linearize   equations   (Ice_8)   and   (Ice_9)   to   find   the   expressions  
(Ice_16)  –  (Ice_18)  for  the  ratio  of  the  changes  in  salinity  and  temperature  when  a  
vanishingly  small  mass  fraction  of  ice  melts  into  seawater  at  a  given  pressure.      

The  enthalpy  difference   hf − hi  in  Eqn.  (Ice_9)  is  expanded  as  a  Taylor  series  
in  the  differences  in  Absolute  Salinity  and  temperature,  and  the  first  order  terms  
in  these  differences  are  retained,  leading  to    

(t f i
)
− t cp + ( SAf − SAi )h ≈
(S f
A − SAi ) (h − h )
i Ih
(
= − wIh hi − h Ih ,   ) (Ice_10)  
SA
SAi
where   c p   is   the   specific   heat   capacity   of   seawater,   c p = ∂h ∂T ,   and  
SA , p
hS = ∂h ∂SA  is  the  derivative  of  the  seawater  specific  enthalpy  with  respect  
A T ,p
to  Absolute  Salinity  at  constant  in  situ  temperature  and  constant  pressure.      

By  regarding  specific  enthalpy  to  be  a  function  of  Conservative  Temperature  

( )
in   the   functional   form   ĥ SA ,Θ, p   the   Taylor   series   expansion   of   Eqn.   (Ice_9)  
yields    

(Θ f
− Θ ĥΘ + i
) ( SAf − SAi ) ĥ ≈
(S f
A − SAi ) ( h − h ) = − w ( h − h ) ,  
i Ih Ih i Ih
(Ice_11)  
SA
SAi
where   ĥΘ = ∂h ∂Θ  is  the  partial  derivative  of  the  seawater  specific  enthalpy  
SA , p
with   respect   to   Conservative   Temperature   Θ   at   fixed   Absolute   Salinity,   and  
ĥS = ∂h ∂SA  is  the  partial  derivative  of  the  seawater  specific  enthalpy  with  
A Θ, p
respect   to   Absolute   Salinity   at   fixed   Conservative   Temperature   Θ .     These  
equations  can  be  rewritten  as      

(S f
− SAi ) h−h
δ T cp ≡ ( )
tf − ti cp ≈
A

SAi
( Ih
− SA hS
A
) (
= − wIh h − h Ih − SA hS
A
) .      (Ice_12)  
70
Thermodynamics Lectures, MIT, 2015 71
( SAf − SAi ) h− h
(
δΘ ĥΘ ≡ Θ f − Θ i ĥΘ ≈ ) SAi
( Ih
− SA ĥS
A
) = − w (h − h Ih Ih
− SA ĥS
A
) .    (Ice_13)  
The  bracket  on  the  right-­‐‑hand  side  of  Eqn.  (Ice_12),   h− h Ih − SA hS ,  if  evaluated  
( )
A
at   the   freezing   temperature   tfreezing SA , p ,   is   the   latent   heat   of   melting   (that   is,  
the  isobaric  melting  enthalpy)  of  ice  into  seawater.    Note  that  at   p = 0 dbar   ĥS is  
A
zero  while   hS  is  nonzero.      
A

The  derivation  of  the  isobaric  melting  enthalpy  in  Feistel  et  al.  (2010)  and  IOC  
et  al.  (2010)  considered  the  seawater  and  ice  to  be  in  thermodynamic  equilibrium  
during   a   slow   processes   in   which   heat   was   supplied   to   melt   the   ice   while  
maintaining   a   state   of   thermodynamic   equilibrium   during   which   the  
temperature   of   the   combined   system   changed   only   because   the   freezing  
temperature  is  a  function  of  the  seawater  salinity.    During  this  reversible  process  
the   enthalpy   of   the   combined   system   increased   due   to   the   heat   externally  
applied.    The  latent  heat  of  melting  is  defined  to  be  (from  Eqn.  (3.34.6)  of  IOC  et  
al.  (2010))    

( ) ( )
LSIp SA , p = h SA ,tfreezing , p − h Ih tfreezing , p − SA hS ( ) A
(S A ,tfreezing , p ) .   (Ice_14)  

In  contrast,  the  present  derivation  (that  is,  Eqns.  (Ice_12)  and  (Ice_13))  applies  
to   the   common   situation   when   the   seawater   is   warmer   than   the   ice   which   is  
melting   into   it,   so   that   the   two   phases   are   not   in   thermodynamic   equilibrium   with  
each   other   during   the   irreversible   melting   process.     That   is,   the   seawater  
temperature  may  be  larger  than  its  freezing  temperature  and  the  ice  temperature  
may   or   may   not   be   less   than   its   freezing   temperature.     The   guiding  
thermodynamic   principle   is   that   there   is   no   change   in   the   enthalpy   of   the  
combined  seawater  and  ice  system  during  the  irreversible  melting  process,  since  
this  process  occurs  adiabatically  at  constant  pressure.      
When   freezing   (as   opposed   to   melting)   is   considered,   the   Second   Law   of  
Thermodynamics   implies   that   spontaneous   freezing   cannot   occur   except   when  
the  seawater  is  at  the  freezing  temperature,  and  there  must  be  some  incremental  
external   change   (for   example   a   decrease   in   pressure   in   the   case   of   frazil  
formation,  or  a  loss  of  heat  from  the  system)  in  order  to  induce  the  freezing.    

Taking   the   limit   of   melting   a   small   amount   of   ice   into   a   seawater   parcel   so  
that   the   changes   in   the   seawater   temperature   and   salinity   are   small,   we   find  
from   Eqn.   (Ice_12)   that   the   ratio   of   the   changes   in   in   situ   temperature   and  
Absolute  Salinity  are  given  by  (using  Eqn.  (Ice_8)  for  the  salinity  increment)    

δt h − h Ih − SA hS
SA = A
δ SA melting at constant p
cp
 .   (Ice_15)  

=
( ) (
h SA ,t, p − h Ih t Ih , p − SA hS SA ,t, p ) A
( )
(
c p SA ,t, p )
while  the  corresponding  ratio  of  the  changes  in  Conservative  Temperature  and  
Absolute  Salinity  are  (from  Eqn.  (Ice_13))    

δΘ h − h Ih − SA ĥS
SA = A
δ SA melting at constant p
ĥΘ
    (Ice_16)  

=
( ) (
ĥ SA ,Θ, p − h Ih t Ih , p − SA ĥS SA ,Θ, p ) A
( )
,
(
ĥΘ SA ,Θ, p )

71
Thermodynamics Lectures, MIT, 2015 72
where   the   second   lines   of   these   equations   have   been   included   to   be   very   clear  
about  how  these  quantities  are  evaluated.    At   p = 0 dbar  these  equations  become    

δθ h0 − h0Ih − SA hS SA ,θ ,0( )
SA = A
δ SA melting at p=0
c p ( SA ,θ ,0 )
    (Ice_17)  
h ( SA ,θ ,0 ) − h Ih (θ Ih , 0 ) − SA hS ( SA ,θ ,0 )
= A
,
c p ( SA ,θ ,0 )
and    

SA
δΘ
=
h0 − h0Ih
= Θ −
(
h Ih θ Ih , 0 ) ,   (Ice_18)  
δ SA melting at p = 0
c0p c0p

where   the   potential   temperatures   of   seawater   θ   and   of   ice   θ Ih   are   both  

referenced   to   p = 0 dbar .     Note   that   the   potential   enthalpy   of   seawater  
( ) (
referenced   to   p = 0 dbar ,   h0 = h SA ,θ ,0 = ĥ SA ,Θ,0   is   simply   c0p   times   )
Conservative   Temperature   Θ   where   c0p   is   the   constant   “specific   heat”  
c0p ≡ 3991.867 957 119 63 J kg −1 K −1 .    

The   use   of   Conservative   Temperature   rather   than   potential   temperature  

means   that   the   slope   of   the   melting   process   on   the   SA − Θ   diagram,   δΘ δ SA ,  
involves   a   simpler   expression,   especially   when   the   melting   occurs   at   the   sea  
( )
surface   at   p = 0 dbar ,   Eqn.   (Ice_18),   where   (i)   ĥS SA ,Θ,0   is   zero,   and   (ii),   the  
( )( )
A
relevant   “specific   heat   capacity”   of   seawater,   ĥΘ = c0p T0 + t T0 + θ ,   reduces   to  
the  constant   c0p ,  so  that  the  specific  enthalpy  of  seawater  is  simply   c0p  multiplied  
by   the   Conservative   Temperature   Θ .     Note   that   the   numerator   of   the   middle  
expression   of   Eqn.   (Ice_18)   is   simply   the   difference   between   the   potential  
enthalpies  of  seawater  and  of  ice.      

Note  that  the  right-­‐‑hand  side  of  Eqn.  (Ice_18)  is  independent  of  the  Absolute  
Salinity  of  the  seawater  into  which  the  ice  melts.      

We  first  illustrate  these  equations  for  the  ratio  of  the  changes  of  Conservative  
Temperature   to   those   of   Absolute   Salinity   by   considering   the   melting   to   occur  
very   close   to   thermodynamic   equilibrium   conditions.     If   both   the   seawater   and  
the  ice  were  exactly  at  the  freezing  temperature  at  the  given  values  of  Absolute  
Salinity  and  pressure,  then  no  melting  or  freezing  would  occur.    In  Fig.  Ice_5  we  
consider  the  limit  as  the  temperatures  of  both  the  seawater  and  the  ice  approach  
the  freezing  temperature.      The  ratio   δΘ δ SA  from  Eqn.  (Ice_16)  is  shown  
equilibrium
in   Fig.   Ice_5(a)   with   the   seawater   enthalpy   evaluated   at   the   freezing  
Conservative   Temperature   and   with   the   ice   enthalpy   evaluated   at   the   in   situ  
freezing  temperature,  at  each  value  of  pressure  and  Absolute  Salinity.    This  ratio  
is  proportional  to  the  reciprocal  of  Absolute  Salinity,  so  it  is  more  informative  to  
simply   multiply   δΘ δ SA   by   Absolute   Salinity   SA   and   this   is   shown   in  
equilibrium
Fig.  Ice_5(b).    It  is  seen  that  the  melting  of  a  given  mass  of  ice  into  seawater  near  
equilibrium  conditions  requires  between  approximately  81  and  83  times  as  much  
heat  as  would  be  required  to  raise  the  temperature  of  the  same  mass  of  seawater  
by   1°C .      

The   corresponding   result   for   the   ratio   of   the   changes   of   in   situ   temperature  
and   Absolute   Salinity   near   equilibrium   conditions  
SI
( ) ( )
SA δ t δ SA equilibrium = Lp SA , p c p SA ,tfreezing , p   can   be   calculated   from   Eqn.  
(Ice_15),  and  the  difference  between   SA δ t δ SA  and   SA δΘ δ SA  is  
equilibrium equilibrium
shown  in  Fig.  Ice_5(c).    The  largest  contributor  to  this  difference  between  Eqns.  
72
Thermodynamics Lectures, MIT, 2015 73
(Ice_15)   and   (Ice_16)   is   due   to   the   dependence   of   the   specific   heat   capacity  
( )
c p SA ,tfreezing , p  on  (i)  Absolute  Salinity,  involving  a  6.8%  variation  over  this  full  
range  of  salinity,  and  (ii)  on  pressure,  involving  a  change  of  2.2%  between   0 dbar  
to   3000 dbar .      

 
 
Figure  Ice_5.      (a)  The  ratio  of  the  change  of  Conservative  Temperature  to  that  of  
Absolute   Salinity   when   the   melting   occurs   very   near   thermodynamic   equilibrium  
conditions,   δΘ δ SA ,   from   Eqn.   (Ice_16)   with   the   seawater   enthalpy  
equilibrium
evaluated   at   the   freezing   Conservative   Temperature   and   with   the   ice   enthalpy  
evaluated   at   the   in   situ   freezing   temperature,   at   each   value   of   pressure   and  
( )
−1
Absolute  Salinity.    The  values  contoured  have  units  of   K g kg −1 .      
  (b)  This  panel  is  simply  Absolute  Salinity   SA  times  the  values  of  panel  (a),  that  
is,  it  is  the  right-­‐‑hand  side  of  Eqn.  (Ice_16),  evaluated  at  equilibrium  conditions.      
  (c)   The   right-­‐‑hand   side   of   Eqn.   (Ice_15)   minus   the   right-­‐‑hand   side   of   Eqn.  
(Ice_16),   both   evaluated   at   equilibrium   conditions,   illustrating   the   difference  
between   using   in   situ   temperature   versus   Conservative   Temperature.     The  
quantities  contoured  in  panels  (b)  and  (c)  have  units  of  temperature,   K .      
 

Equation   (Ice_16)   for   SA δΘ δ SA   is   now   illustrated   when   the  

melting at constant p
seawater  and  the  ice  Ih  phases  are  not  at  the  same  temperature  and  they  are  not  in  
thermodynamic  equilibrium  at  the  freezing  temperature.    We  begin  by  considering  
melting  of  ice  Ih  at  the  sea  surface,  specifically  at   p = 0 dbar ,  when  Eqn.  (Ice_16)  
reduces   to   Eqn.   (Ice_18),   and   this   equation   is   illustrated   in   Fig.   Ice_6(a)   which  
applies  at  all  values  of  Absolute  Salinity.    The  contoured  values    of  Fig.  Ice_6(a),  
( ) ( )
h0 − h0Ih c0p = Θ − h Ih θ Ih c0p ,   increase   as   1.0   times   changes   in   Θ   and  

73
Thermodynamics Lectures, MIT, 2015 74
Ih 0
decrease  approximately  as   c p c p ≈ 0.52  times  changes  in  the  temperature  of  the  
ice.        

δΘ h0Ih δΘ δΘ
SA = Θ −                                                                   SA − SA    
δ SA melting at p = 0
c0p δ SA melting at p = 500
δ SA melting at p = 0

           
 
 
Figure  Ice_6.          (a)  Contours  of  Eqn.  (Ice_18),    
  SA δΘ δ SA
melting at p = 0( ) ( )
= h0 − h0Ih c0p = Θ − h Ih θ Ih c0p ,   for   the   melting   of   ice   Ih  
into   seawater   at   p = 0 dbar .     The   six   stars   are   at   the   freezing   temperatures    
( t  and   Θ )  for  Absolute  Salinity  values  starting  at   5 g kg −1  with  increments  of   5 g kg −1  
up   to   30 g kg −1 .     (b)   Difference   between   contours   of   Eqn.   (16)   at   p = 500 dbar ,  
SA δΘ δ SA ,  and  the  corresponding  ratio  of  panel  (a)  (where  the  pressure  
melting at p = 500
was   0 dbar )  at   SA = SSO = 35.16504 g kg −1 .    The  double-­‐‑starred  point  is  at  the  freezing  
i

temperatures  ( t  and   Θ )  at   p = 500 dbar  and   SAi = SSO = 35.16504 g kg −1 .      

 
 

74
Thermodynamics Lectures, MIT, 2015 75

Taking  the  potential  enthalpy  of  ice  Ih  to  be  a  conservative  variable    
By   comparing   panels   (a)   and   (b)   of   Fig.   Ice_6   we   are   able   to   deduce   a   very  
important   approximation   that   will   prove   invaluable   to   coupled   ocean/ice  
modelling.     Panel   (b)   shows   the   error   in   assuming   that   it   is   the   potential  
enthalpy  of  ice  that  is  conserved  when  ice  melts  into  seawater,  rather  than  taking  
the   enthalpy   of   ice   to   be   conserved,   which   is   the   correct   thing   to   do.     At   a  
pressure   of   500   dbar   the   assumption   that   the   sum   of   the   potential   enthalpies   of  
ice   and   seawater   are   conserved   leads   to   an   error   of   0.15%   in   the   change   in  
Conservative   Temperature   of   the   seawater   as   a   result   of   melting.     Most   of   this  
error   is   due   to   the   assumption   regarding   ice,   not   seawater,   since   the   error  
involved   with   assuming   that   the   Conservative   Temperature   of   seawater   is  
totally  conservative  reaches  a  maximum  of  0.15%  only  at  a  much  larger  pressure  
of  4000 dbar  (Graham  and  McDougall,  2013).      
The  ratio  of  Eqns.  (Ice_16)  to  (Ice_18)  is    

δΘ
δ SA melting at constant p
=
(T0 + θ ) ⎡⎢ h − hIh − SA ĥS ⎤⎥
A

δΘ (T0 + t ) ⎢⎣ h0 − h0Ih ⎥⎦
δ SA (Ice_19)  
melting at p = 0

( θ − t) (T0 + θ ) ⎡⎣( h − h0 ) − ( h − h0 ) − SA ĥS ⎤

Ih Ih
A⎦
= 1+ + ,
(T0 + t ) (T0 + t ) ( h0 − h0Ih )
and   the   combination   of   enthalpy   differences   in   the   numerator   of   the   last   term  
can  be  expressed  as    

⎣ 0 ( 0 A SA⎦ )
⎡ h − h − h Ih − h Ih − S ĥ ⎤ =
P P   (Ice_20)  
( ) ( ) (
∫ ⎡⎣ v̂ SA ,Θ, p′ − v! θ , p′ ⎤⎦ dP′ − SA ∫ v̂SA SA ,Θ, p′ dP′.
P0
Ih Ih

P0
)
The  last  term  here  is  small,  showing  that  the  dominant  contribution  is  simply  the  
pressure  integral  of  the  difference  in  the  specific  volumes  of  seawater  and  of  ice.      

( )(
In   Eqn.   (Ice_19)   the   second   term   on   the   right-­‐‑hand   side,   θ − t T0 + t ,   is   )
small  compared  with  the  third,  so  that  the  non-­‐‑unity  nature  of  Eqn.  (Ice_19)  can  
be   understood   as   being   due   to   this   third   term,   evaluated   with   the   aid   of   Eqn.  
(Ice_20),  and  this  evaluation  agrees  with  the  plot  of  Fig.  Ice_6(b).      
We  will  take  advantage  of  the  smallness  of  panel  (b)  versus  panel  (a)  of  Fig.  
Ice_6,   or   equivalently,   the   fact   that   Eqn.   (Ice_19)   is   quite   close   to   unity,   to   treat  
the   potential   enthalpy   of   ice   as   conserved   during   not   only   advection   but   also  
during  melting  and  freezing  events.    This  will  greatly  reduce  the  complexity  of  
coupled   ocean/ice   numerical   models.     This   approximation   brings   the   same  
simplicity  to  ice  as  the  introduction  of  Conservative  Temperature  has  brought  to  
physical   oceanography,   in   that   the   only   variables   that   now   need   to   be  
considered   when   discussing   “heat”   budgets   of   seawater   and   of   ice   are   the  
potential  enthalpies  of  seawater  and  of  ice.      
 

75
Thermodynamics Lectures, MIT, 2015 76

An  illustration  from  the  Amery  Ice  Shelf    

Figure   Ice_7   shows   oceanographic   data   obtained   under   the   Amery   Ice   Shelf  
that   illustrates   the   ratio   of   the   changes   in   Absolute   Salinity   and   Conservative  
Temperature,  as  given  by  Eqn.  (Ice_16),  when  melting  of  ice  occurs.    The  vertical  
profile   named   AM06   begins   under   the   ice   at   a   pressure   of   546   dbar   and   the  
uppermost   175   of   the   vertical   profile   is   shown.     The   data   in   the   uppermost   50-­‐‑
100  dbar  is  closely  aligned  with  the  ratio  given  by  Eqn.  (Ice_16)  (as  shown  by  the  
dashed   line)   evaluated   at   this   pressure   and   with   the   ice   temperature   being   the  
freezing  temperature  at  this  salinity  and  pressure.    Two  freezing  lines  are  shown  
in  Fig.  Ice_7(b),  for  pressures  of   0 dbar  and   578 dbar .      

Any  observations  cooler  than  the  freezing  temperature  appropriate  to   0 dbar  
is   evidence   of   the   influence   of   melting   of   ice   or   of   heat   lost   by   conduction  
through  the  ice.    AM06  is  located  on  the  eastern  side  of  the  ice  shelf  in  an  area  
that  is  melting,  as  can  be  inferred  by  the  presence  of  ocean  water  at  AM06  that  is  
well   above   the   in   situ   freezing   temperature   at   the   base   of   the   ice   shelf.     This  
water   is   thought   to   be   flowing   in   a   primarily   southwards   direction   from   the  
open   ocean   as   it   enters   the   under-­‐‑ice   cavity.     The   other   CTD   profile   was   taken  
from  borehole  AM05,  located  on  the  western  side  of  the  ice  shelf  in  an  area  that  
is  refreezing  (as  is  drawn  in  panel  (a))  and  represents  flow  that  has  likely  come  
from  deeper  in  the  sub-­‐‑ice-­‐‑shelf  cavity,  than  at  AM06  (Post  et  al.,  2013)  and  hence  
has  been  in  contact  with  the  ice  for  longer.    The  upper  50m  or  so  of  this  cast  is  at  
the   freezing   temperature   of   seawater   at   this   pressure.     For   both   casts   the   data  
near  the  upper  part  of  the  water  column  has  the  ratio  of  the  changes  of   SA  and  
Θ  in  close  agreement  to  the  ratio  given  by  Eqn.  (Ice_16),  the  ratio  predicted  from  
melting  ice  into  seawater  (dashed  lines).    The  ice  temperature  that  is  needed  to  
calculate   this   SA − Θ   ratio   for   each   location   has   been   taken   to   be   the   in   situ  
freezing   temperature   of   ice   in   contact   with   the   seawater   at   the   pressure   at   the  
base  of  the  ice  shelf.    Moreover,  on  this  figure  the  uppermost  100m  of  data  of  the  
AM05   data   is   approximately   related   to   that   of   the   AM06   data   through   the  
SA − Θ  ratio  of  Eqn.  (Ice_16).    This  would  be  consistent  with  the  notion  that  the  
same   fluid   is   proceeding   from   AM06   to   AM05   without   being   exposed   to  
significant  heat  loss   Q  to  the  ice  (see  panel  (a)).    The  vertical  profiles  shown  in  
panel  (b)  are  the  average  of  several  vertical  profiles  taken  over  the  course  of  two  
days,  and  the  two  locations  were  drilled  within  two  weeks  of  each  other.          
 

76
Thermodynamics Lectures, MIT, 2015 77
 

     

   

 
Figure   Ice_7.   (a)   Sketch   of   the   flow   under   an   ice   shelf.     An   inflow   of   relatively  
warm   water   from   the   open   ocean   provides   heat   to   melt   the   ice   shelf.     Buoyant  
freshwater  that  is  released  during  the  melting  process  rises  along  the  underside  of  
the  ice  shelf  and  can  become  locally  supercooled  at  a  shallower  depth,  leading  to  
the  formation  of  frazil  and  basal  accretion  of  marine  ice.      
               (b)   The   top   175m   of   two   CTD   profiles   taken   below   the   Amery   Ice  
Shelf  in  East  Antarctica  at  a  melt  site  and  at  a  refreeze  site  are  shown.    The  warmer  
and  saltier  of  the  two  casts  is  AM06  (see  Fig  1  of  Galton-­‐‑Fenzi  et  al.  (2012))  starting  
at  a  pressure  of  546  dbar.    The  large  round  dot  is  ocean  data  very  near  the  ice  at  
546  dbar,  the  triangle  is  50  dbar  deeper,  the  diamond  100  dbar  deeper  and  the  star  
is   150   dbar   below   the   bottom   of   the   ice   shelf   at   this   location,   indicated   by   the  
circle.     The   other   vertical   cast,   AM05,   is   typical   of   re-­‐‑freezing   locations.     The  
uppermost  50  dbar  of  this  cast  is  all  at  the  freezing  temperature  at  this  pressure.      
 

77
Thermodynamics Lectures, MIT, 2015 78

Melting  of  sea  ice  into  seawater    

Sea  ice  contains  a  certain  mass  fraction  of  brine  trapped  inside  the  ice  matrix.    
Sea  ice  is  produced  when  the  surface  of  the  ocean  is  cooled  rapidly  by  very  cool  
air.    The  ice  crystals  then  form  so  fast  that  some  of  the  seawater  is  trapped  in  
small  “pockets”  inside  the  matrix  of  ice  crystals.      

We  can  quantify  the  melting  of  sea  ice  into  seawater  by  conserving  the  same  
three  quantities,  namely  (i)  mass,  (ii)  salt,  and  (iii)  enthalpy,  leading  to    
(see  McDougall,  T.  J.,  P.  M.  Barker,  R.  Feistel  and  B.  K.  Galton-­‐‑Fenzi,  2014:  
Melting  of  ice  and  sea  ice  into  seawater,  and  frazil  ice  formation.    Journal  of  
Physical  Oceanography,  44,  1751-­‐‑1775.  for  details)    

(
        SAf − SAi ) = −
mseaice
f
mSW
(S i
A − SAseaice )= (
− wseaice SAi − SAseaice     )

           
(h f
− hi ) (
= − wseaice hi − h Ih )+w seaice
mseaice
(
mbrine brine
h − h Ih )
   
= − wseaice (h − h ) .
i seaice

where  the  specific  enthalpy  of  the  composite  material  “seaice”  is  defined  as  the  
mass-­‐‑weighted  sum  of  the  specific  enthalpies  of  the  two  phases,    

( ) (
hseaice = mIh mseaice h Ih + mbrine mseaice h brine ,   )  

and  the  Absolute  Salinity  –  enthalpy  mixing  diagram  looks  like    

 
 
 

78
Thermodynamics Lectures, MIT, 2015 79

For  sea  ice  melting  into  seawater  at   p = 0 dbar with  initial  properties  
SAi = SSO = 35.16504 g kg −1 ,   Θ i = 4°C ,  and  with  the  sea  ice  salinity  taken  to  be  
SAseaice = 5 g kg −1 ,  the  change  in  Conservative  Temperature  is  shown  in  the  left-­‐‑
hand  figure  below.    The  right-­‐‑hand  figure  shows  the  corresponding  plot  when  
SAseaice = 0 g kg −1 .      

! !
 
What  is  the  ratio  of  the  changes  in  Conservative  Temperature  and  Absolute  
Salinity  of  seawater  when  a  vanishingly  small  mass  of  sea  ice  melts  into  it?      
To  find  this  ratio  we  again  linearize  the  above  expressions  for  a  vanishingly  
small  mass  fraction  of  sea  ice  that  melts,  giving    

(S A
− SAseaice ) δδSΘ
A melting at constant p
=

         
( ) ( )
⎛ SAseaice ⎞ SAseaice
⎜ 1 − ⎟ h− h Ih
− S ĥ + h − h brine − ⎡ SA − SAbrine ⎤ ĥS
⎝ SAbrine ⎠
A SA
SAbrine ⎣ ⎦ A
ĥΘ

and  this  is  illustrated  below  at   p = 0 dbar  and  at   SAi = SSO = 35.16504 g kg −1  and  
Θ i = 1°C .      

(S A − SAseaice ) δδSΘ
A melting at p = 0
   

!  
 

79
Thermodynamics Lectures, MIT, 2015 80

Frazil  ice  formation    

When  seawater  at  the  freezing  temperature  undergoes  upwards  vertical  motion  
so   that   its   pressure   decreases,   frazil   forms,   primarily   due   to   the   increase   in   the  
freezing  temperature  as  a  result  of  the  reduction  in  pressure.    When  this  mixture  
of   seawater   and   frazil   continues   to   rise   to   lower   pressures   (assisted   by   the  
buoyancy  provided  by  the  presence  of  the  ice),  the  frazil  crystals  will  experience  
a   larger   change   in   in   situ   temperature   than   does   the   seawater,   simply   because  
the   adiabatic   lapse   rate   of   ice   is   much   larger   (ten   times   as   large)   than   that   of  
seawater  (as  we  have  found  above).      

We  will  here  consider  this  situation  under  the  assumption  that  the  frazil  and  
the  seawater  moves  together,  so  ignoring  the  tendency  of  the  frazil  to  rise  faster  
than   the   seawater,   driven   by   the   buoyancy   of   the   individual   ice   crystals.     We  
further   assume   that   the   uplift   rate   is   sufficiently   small   that   the   in   situ  
temperature   of   the   ice   and   the   seawater   are   the   same   at   each   pressure,   this  
temperature  being  the  freezing  temperature.    Under  these  conditions  no  entropy  
is   produced   during   the   freezing   process,   i.e.,   this   freezing   process   is   reversible  
and  can  be  reversed  by  increasing  the  pressure,  leading  to  the  related  reversible  
ice  melt.      

We   will   study   the   thermodynamics   of   this   process   of   adiabatic   uplift   of   a  

seawater-­‐‑ice  mixture  via  a  thought  process  composed  of  two  separate  steps  (Fig.  
Ice_13).     First   we   imagine   the   mixture   of   pre-­‐‑existing   ice   and   seawater   to  
undergo   a   reduction   in   pressure   but   without   any   exchange   of   heat,   water   or   salt  
between  the  two  phases.    That  is,  during  this  first  part  of  the  process  the  mass  of  
ice  and  the  mass  of  seawater  remain  constant,  and  the  change  in  the  enthalpy  of  
the   ice   and   the   change   in   the   enthalpy   of   the   seawater   are   only   due   to   the  
pressure   change.     During   this   adiabatic   process   an   (infinitesimal)   contrast   in   in  
situ   temperature   will   develop   between   the   ice   phase   and   the   seawater   phase  
because   the   adiabatic   lapse   rate   of   ice   is   much   larger   (by   about   an   order   of  
magnitude)  than  that  of  seawater.      

During   the   second   part   of   our   thought   experiment   the   ice   and   seawater  
phases   will   be   allowed   to   equilibrate   their   temperatures   and   further   frazil   ice  
will   form   so   that   the   temperature   of   both   the   ice   and   seawater   phases   and   the  
final  Absolute  Salinity  of  the  seawater  phase  will  be  consistent  with  the  freezing  
temperature   at   this   pressure.     This   part   of   our   thought   experiment   occurs   at  
constant  pressure  and  so,  from  the  First  Law  of  Thermodynamics,  we  know  that  
enthalpy  is  conserved.      

80
Thermodynamics Lectures, MIT, 2015 81
 
 

 
Figure   13.     Sketch   showing   the   two-­‐‑step   thought   process   involved   with  
quantifying   the   formation   of   frazil   ice   Ih   by   the   adiabatic   uplift   of   a   seawater  
parcel  which  may  contain  pre-­‐‑existing  frazil  ice.    The  step  from  stage  1  to  stage  2  
is  undertaken  without  any  exchange  of  heat  or  mass  between  the  seawater  and  
ice  Ih  phases.    While  the  in  situ  temperatures  of  the  seawater  and  ice  phases  are  
assumed   to   be   identical   at   stage   1,   at   stage   2   they   are   unequal   because   the  
adiabatic  lapse  rate  of  ice  Ih  is  much  larger  than  that  of  seawater.    The  step  from  
stage   2   to   stage   3   is   undertaken   at   constant   pressure.     In   this   step   further   ice  
forms  (as  shown  by  the  increase  in  number  of  the  frazil  ice  crystals)  and  at  the  
end  of  this  step,  the  seawater  and  ice  phases  have  the  same  in  situ  temperature,  
namely  the  freezing  temperature  appropriate  to  (i)  that  pressure  and  (ii)  the  final  
value  of  seawater  salinity.      
 

Let   the   mass   fraction   of   ice   be   wIh ;   the   mass   fraction   of   seawater   in   the   ice-­‐‑
( )
seawater  mixture  is  then   1 − wIh .    The  total  enthalpy  per  unit  mass  of  the  ice-­‐‑
seawater   mixture   at   stage   1   of   Fig.   13   is   the   weighted   sum   of   the   specific  
enthalpies  of  the  two  phases,  namely    

(1 − w ) ĥ( S
Ih
1 A1 ,Θ1 , p1 ) ( )
+ w1Ih h Ih θ1Ih , p1 ,   (Ice_34)  

where  we  have  chosen  to  write  the  specific  enthalpy  of  ice  in  the  functional  form  
( )
h Ih θ Ih , p  where  the  temperature  variable  is  the  potential  temperature  of  ice   θ Ih  
with   reference   pressure   0 dbar   ( θ Ih   is   not   to   be   confused   with   the   potential  
temperature   of   seawater   θ ,   since   these   two   potential   temperatures   are   not  
equal).      

In  going  from  stage  1  to  stage  2,  both  the  seawater  and  ice  phases  undergo  an  
adiabatic  change  of  pressure   δ P  which  changes  their  specific  enthalpies  by   v δ P  
and   v Ih δ P   respectively   (here   v   and   v Ih   are   the   specific   volumes).     Hence   at  
stage   2   the   total   enthalpy   per   unit   mass   of   the   ice-­‐‑seawater   mixture   is   (noting  
that   w2Ih = w1Ih   and   that   at   leading   order   in   the   perturbation   quantities,   it   is  
immaterial  whether   v  and   v Ih  are  evaluated  at  the  properties  of  stage  1  or  those  
of  stage  2)    

81
Thermodynamics Lectures, MIT, 2015 82
(1 − ) ⎡⎣ ĥ( S
w1Ih A1 ,Θ1 , p1 ) + v δ P ⎤⎦ + w1Ih ⎡h
⎣
 Ih (
θ1Ih , p1 ) +v Ih
δ P ⎤ .  
⎦
(Ice_35)  

In   going   from   stage   2   to   stage   3,   the   total   enthalpy   of   the   mixture   is  

conserved.    Hence  we  equate  the  total  enthalpies  at  these  two  stages,  giving    

(1 − w ) ⎡⎣ ĥ( S
Ih
1 A1 ⎣ (
,Θ1, p1 ) + v δ P ⎤⎦ + w1Ih ⎡ h Ih θ1Ih , p1 + v Ih δ P ⎤ =
⎦ )     (Ice_36)  
(1 − w ) ĥ( S
Ih
3 A3 ,Θ3 , p3 )+ (
w3Ih h Ih θ 3Ih , p3 . )
For  an  externally-­‐‑imposed  change  in  pressure  this  equation  may  be  regarded  as  
giving  the  amount  of  new  ice  formed   w3Ih − w1Ih  due  to  the  adiabatic  uplifting  of  
the   ice-­‐‑seawater   mixture.     The   other   important   constraint   that   we   know   is   that  
the   ice-­‐‑seawater   mixture   is   at   the   freezing   temperature   at   both   stages   1   and   3.    
This  turns  out  to  enough  information  to  solve  the  problem.      

( ) ( )
The   enthalpies   ĥ SA3 ,Θ3 , p3   and   h Ih θ 3Ih , p3   on   the   right-­‐‑hand   side   of   Eq.  
(Ice_36)  are  now  expanded  in  a  Taylor  series  about  the  values  at  stage  1,  keeping  
the  leading  order  terms.    The  pressure  derivatives  of  these  enthalpies,  being  the  
specific   volumes   of   seawater   and   of   ice,   give   terms   that   cancel   with   the  
corresponding  terms  on  the  left-­‐‑hand  side  of  the  equation  to  leading  order.    The  
remaining  leading-­‐‑order  terms  are    

(h − h ) δ w
Ih Ih
( )(
− 1 − wIh ĥS δ SA + ĥΘ δΘ
A
) − wIh h IhIh δθ Ih = 0    
θ
(Ice_37)  

where   δ wIh = w3Ih − w1Ih .    Since  the  salt  always  resides  in  the  seawater  phase,  the  
( )
product   1 − wIh SA  is  constant  so  that      

(
SA δ wIh = 1 − wIh δ SA ,     ) (Ice_38)  

which  reduces  Eqn.  (Ice_37)  to    

(h − h Ih
− SA ĥS
A
)δS A
− SA ĥΘ δΘ − SA
(1 − w )
wIh
Ih
h IhIh δθ Ih = 0 .  
θ
(Ice_39)  

One   of   our   key   results   for   frazil   ice   is   already   apparent   from   this   equation,  
namely   that   as   the   mass   fraction   of   frazil   ice   wIh   tends   to   zero,   Eqn.   (Ice_39)  
tends  to  our  existing  result  Eqn.  (Ice_16)  for  the  ratio   δΘ δ SA  for  the  melting  of  
ice  Ih  into  seawater,  repeated  here,  

δΘ h − h Ih − SA ĥS
SA = A
δ SA melting at constant p
ĥΘ
    (Ice_16)  

=
( ) (
ĥ SA ,Θ, p − h Ih t Ih , p − SA ĥS SA ,Θ, p ) A
( )
.
(
ĥΘ SA ,Θ, p )
However,  there  is  an  important  difference  as  well,  namely  that  the  present  frazil  
ice   relation   Eqn.   (Ice_39)   for   the   ratio   δΘ δ SA   is   actually   simpler   (or   more  
restrictive)   than   Eqn.   (Ice_16)   because   the   temperatures   of   both   the   ice   and  
seawater   components   are   constrained   to   be   at   the   freezing   temperature;   the   ice  
temperature   cannot   be   lower   that   the   freezing   temperature   nor   can   the  
Conservative   Temperature   of   the   seawater   exceed   its   freezing   temperature.    
Hence   in   the   limit   as   the   mass   fraction   of   frazil   ice   wIh   tends   to   zero,   as   the  
pressure  of  a  seawater-­‐‑frazil  mixture  is  changed,  the  ratio   δΘ δ SA  is  illustrated  
by   the   equilibrium   situation   of   our   existing   Figures   Ice_5(a)   and   (b).     We   will  
return   to   this;   for   now   this   paragraph   is   just   a   heads   up   on   the   comparison  
between  what  we  have  derived  already  (Eqn.  (Ice_16))  and  where  we  are  headed  
with  the  equations  for  frazil.      

82
Thermodynamics Lectures, MIT, 2015 83
Ih
Returning   to   the   more   general   situation   in   which   w   is   not   vanishingly  
small,  we  need  to  evaluate   h IhIh δθ Ih  in  terms  of  differentials  of  Absolute  Salinity  
θ
and  pressure.    The  partial  differential   h IhIh  can  be  written  as    
θ

∂h Ih ∂h Ih ∂t Ih ∂t Ih
h IhIh ≡ = = c Ih
p
.   (Ice_40)  
θ ∂θ Ih p
∂t Ih p
∂θ Ih p
∂θ Ih p

The   in   situ   temperature   of   ice   Ih   can   be   expressed   as   a   function   of   the  

potential   temperature   of   ice   Ih   and   pressure   as     t Ih = t Ih θ Ih , p   so   that   the   total   ( )
differential  of  the  in  situ  temperature  of  ice  is    
∂t Ih
dt Ih = dθ Ih + Γ Ih dP .   (Ice_41)  
∂θ Ih p

This   equation   applies   to   any   material   differentials   dt Ih , dθ Ih   and   dP ,   and   in  

particular   will   apply   to   the   differences   between   these   properties   at   stage   1   and  
stage  3  of  our  thought  process.    Hence  we  can  write    
∂t Ih
δ t Ih = δθ Ih + Γ Ihδ P .   (Ice_42)  
∂θ Ih p

But   the   ice   at   both   stages   1   and   3   is   at   the   freezing   temperature  

( )
tfreezing = tfreezing SA , p  so  that   δ t Ih  can  also  be  expressed  as    
∂tfreezing ∂tfreezing
δ t Ih = δ SA + δ P ,   (Ice_43)  
∂SA ∂P
p SA

and  the  partial  derivatives  here  are  known  functions  of  the  Gibbs  functions  of  ice  
Ih  and  seawater.      

Combining   Eqns.   (Ice_42)   and   (Ice_43)   and   using   the   result   in   Eqn.   (Ice_40)  
gives  our  desired  result  for   h IhIh δθ Ih ,  namely    
θ
⎡ ∂t ⎛ ⎞ ⎤
h IhIh δθ Ih = c Ih ⎢ freezing δ S + ⎜ ∂tfreezing − Γ Ih ⎟
δ P ⎥ .   (Ice_44)  
p ⎢ A ⎥
θ ∂SA ⎜ ∂P ⎟
⎢⎣ p ⎝ SA ⎠ ⎥⎦
Substituting   this   equation   into   Eq.   (39)   gives   a   relationship   between   only  
δ SA , δΘ ,  and   δ P ,  namely    
⎛ ∂tfreezing ⎞
wIh
⎜ h − h Ih − SA ĥS − SA c Ih
⎟ δ SA − SA ĥΘ δΘ
⎜⎝ A
(
1 − wIh
p
∂SA ) p⎠
⎟
(Ice_45)  
Ih ⎛ ∂t ⎞
w
− SA c Ih ⎜ freezing − Γ Ih ⎟ δ P = 0.
1− w Ih p
⎜ ∂P S
⎝ (A
⎟
⎠ )
Another   relationship   between   δ SA , δΘ ,   and   δ P   can   be   found   from   the  
knowledge   that   in   both   stages   1   and   3   the   seawater   is   at   the   freezing  
Conservative   Temperature,   and   since     Θ freezing   is   a   function   of   only   SA   and   P ,  
the  differences   δ SA , δΘ ,  and   δ P  are  related  by    
∂Θ freezing ∂Θ freezing
δΘ = δ SA + δ P ,   (Ice_46)  
∂SA ∂P
p SA

and   expressions   for   these   partial   derivatives   are   known   in   terms   of   the   Gibbs  
functions  of  seawater  and  of  ice  (we  do  not  derive  them  here).      

Eqns.   (Ice_45)   and   (Ice_46)   are   two   equations   in     δ SA , δΘ ,   and   δ P   from  

which   we   can   find   our   desired   relations   for   the   ratios   of   changes   in   our  
seawater-­‐‑frazil  ice  mixture  due  to  adiabatic  uplift,  namely   δΘ δ SA ,   δΘ δ P  and  

83
Thermodynamics Lectures, MIT, 2015 84
δ SA δ P .      By  eliminating  the  pressure  difference  from  these  two  equations  we  
find  that      

⎛ ⎡ ⎛ ∂t ⎞ ⎤⎞
⎜ ⎢ ⎜ freezing − Γ Ih ⎟ ⎥⎟
⎜ ⎢ ⎜⎝ ∂P S ⎟⎠ ∂Θ ⎥⎟
wIh ∂t
Ih ⎢ freezing ⎥⎟
⎜ h − h Ih − S ĥ − S c − A freezing
⎜
⎜
A SA A
1− w(Ih p ⎢
⎢) ∂S A p
∂Θ freezing
∂S A
⎥⎟
p⎥⎟
⎜ ⎢ ∂P ⎥⎟
SA
δΘ ⎝ ⎢⎣ ⎥⎦ ⎠
SA = .    (Ice_47)  
δ SA frazil ⎛ ⎛ ∂t ⎞⎞
⎜ ⎜ freezing − Γ ⎟ Ih ⎟
⎜ Ih ⎜ ∂P ⎟⎠ ⎟
⎜ ĥ + w Ih ⎝ SA
⎟
c
⎜ Θ
⎜
(
1 − wIh )
p
∂Θ freezing ⎟
⎟
⎜ ∂P ⎟
SA
⎝ ⎠

The   leading   terms   in   both   the   numerator   and   denominator,   namely  

h − h Ih − SA ĥS   and   ĥΘ   are   the   same   as   in   Eqn.   (Ice_16)   which   applies   to   the  
A
melting  of  ice  Ih  into  seawater  at  fixed  pressure,  the  only  difference  being  that  in  
the  present  case  both  the  ice  and  the  seawater  are  at  the  freezing  temperature.      

So,   as   the   mass   fraction   of   ice   tends   to   zero,   Eqn.   (Ice_47)   tends   to   Eqn.  
(Ice_16),  so  that  at   wIh = 0  Eqn.  (Ice_47)  can  be  illustrated  by  Fig.  Ice_5(a),  which  
is  repeated  below.      

 
 
Figure   Ice_5.     (repeat   of   this   figure)     (a)   The   ratio   of   the   change   of   Conservative  
Temperature   to   that   of   Absolute   Salinity   when   the   melting   occurs   very   near  
thermodynamic  equilibrium  conditions,   δΘ δ SA ,  from  Eqn.  (Ice_16)  with  the  
equilibrium
seawater  enthalpy  evaluated  at  the  freezing  Conservative  Temperature  and  with  
the   ice   enthalpy   evaluated   at   the   in   situ   freezing   temperature,   at   each   value   of  
( )
−1
pressure  and  Absolute  Salinity.    The  values  contoured  have  units  of   K g kg −1 .      
  (b)  This  panel  is  simply  Absolute  Salinity   SA  times  the  values  of  panel  (a),  that  
is,  it  is  the  right-­‐‑hand  side  of  Eqn.  (Ice_16),  evaluated  at  equilibrium  conditions.    
 
For   non-­‐‑zero   ice   mass   fraction   Eqn.   (Ice_47)   is   plotted   in   Fig.   Ice_14(a)   at  
SA = SSO = 35.16504 g kg −1   (actually   what   is   plotted   is   δΘ δ SA frazil ).     The  
dependence   on   the   mass   fraction   of   sea   ice   can   be   illustrated   with   the   case  
wIh = 0.1   when   δΘ δ SA frazil   is   different   to   the   value   at   wIh = 0   by   about   7.4%.    
Most   of   this   sensitivity   to   wIh   comes   from   the   denominator   in   Eqn.   (Ice_47).    
Eqn.   (Ice_47)   is   again   illustrated   in   Fig.   Ice_15(a)   where   we   show   contours   of  
SA δΘ δ SA frazil   at   the   fixed   salinity   SA = SSO = 35.16504 g kg −1 .     That   is,   Fig.  
84
Thermodynamics Lectures, MIT, 2015 85
−1
Ice_15(a)   is   simply   35.16504 g kg   times   Fig.   Ice_14(a),   so   that   the   quantity  
contoured  in  Fig.  Ice_15(a)  is  in  temperature  units.      
 

 
Figure  Ice_14.    (a)  Plot  of   δΘ δ SA frazil  from  Eqn.  (Ice_47)  as  a  function  of  the  ice  
mass   fraction   wIh   and   pressure.     (b)   Plot   of   δΘ δ P frazil   from   Eqn.   (Ice_48)   as   a  
function  of  the  ice  mass  fraction   wIh  and  pressure.    (c)  Plot  of   δ SA δ P frazil  from  
Eqn.  (Ice_49)  as  a  function  of  the  ice  mass  fraction   wIh  and  pressure.    All  three  
panels  have  the  seawater  salinity   SA = SSO = 35.16504 g kg −1 .    Panel  (a)  has  units  
( ) ( )( )
−1
( )−1
of   K g kg −1 ,  panel  (b)  is  in   K Pa ,  while  panel  (c)  is  in   g kg −1 Pa .    The  
−1

values   contoured   in   this   figure   were   evaluated   from   the   GSW   algorithm  
gsw_frazil_ratios_adiabatic   of   the   GSW   Oceanographic   Toolbox   (www.TEOS-­‐‑
10.org).      
 

85
Thermodynamics Lectures, MIT, 2015 86
 

Similarly,  by  eliminating   δ SA  from  Eqns.  (Ice_45)  and  (Ice_46)  we  find    

⎛ ⎡ ⎛ ∂t ⎞ ⎤⎞
⎜ ⎢ ⎜ freezing − Γ Ih ⎟ ⎥⎟
⎜ ⎢ ⎜⎝ ∂P S ⎟⎠ ∂Θ ⎥⎟
wIh ∂t
Ih ⎢ freezing ⎥⎟
⎜ h − h Ih − S ĥ − S c − A freezing
⎜
⎜
A SA A
(
1− w Ih
) p ⎢
⎢
∂S A p
∂Θ freezing
∂S A
⎥⎟
p⎥⎟
⎜ ⎢ ∂P ⎥⎟
δΘ ∂Θ freezing ⎝ ⎢⎣
SA
⎥⎦ ⎠
= ,  (Ice_48)  
δ P frazil ∂P ⎛ ∂Θ freezing ∂tfreezing ⎞
SA wIh
⎜ h − h Ih − SA ĥS − SA ĥΘ − SA c Ih
⎟
⎜⎝ A ∂SA
p
1 −(w Ih p
)∂SA ⎟
p ⎠

and  when   δΘ  is  eliminated  from  these  same  two  equations  we  find    

⎛ ⎛ ∂t ⎞⎞
⎜ ⎜ freezing − Γ Ih ⎟ ⎟
⎜ ⎜⎝ ∂P S ⎟⎠ ⎟
⎜ ĥ + wIh ⎟
c Ih A

⎜
⎜
Θ
1− w(Ih p
) ∂Θ freezing ⎟
⎟
⎜ ∂P ⎟
1 δ SA ∂Θ freezing ⎝ SA
⎠
= .      (Ice_49)  
SA δ P ∂P ⎛ ∂Θ freezing Ih ∂tfreezing ⎞
frazil SA w
⎜ h − h Ih − SA ĥS − SA ĥΘ − SA Ih
cp ⎟
⎜⎝ A ∂SA
p (1 − w ) Ih ∂SA ⎟
p ⎠

The   variation   of   Conservative   Temperature   with   pressure   under   frazil   ice  

conditions,   δΘ δ P frazil ,   from   Eqn.   (Ice_48)   is   plotted   in   Fig.   Ice_14(b)   at  
SA = SSO = 35.16504 g kg −1 .     It   is   seen   that   δΘ δ P frazil   is   quite   insensitive   to   the  
frazil  ice  mass  fraction   wIh .    This  is  confirmed  in  Fig.  Ice_15(c)  where  we  show  
the   difference   between   δΘ δ P frazil   and   the   corresponding   derivative   of   Θ freezing  
with  pressure  at  constant  Absolute  Salinity,   ∂Θ freezing ∂P .      
SA

The   variation   of   Absolute   Salinity   with   pressure   under   frazil   ice   conditions,  
δ SA δ P frazil ,   from   Eqn.   (Ice_49)   is   plotted   in   Figure   Ice_14(c)   at  
SA = SSO = 35.16504 g kg −1 .    This  figure  follows,  of  course,  as  simply  the  ratio  of  
the  figures  of  panels  (a)  and  (b).      

86
Thermodynamics Lectures, MIT, 2015 87

 
Figure  Ice_15.    (a)  Plot  of   SA δΘ δ SA frazil  from  Eqn.  (Ice_47)  as  a  function  of  the  
ice   mass   fraction   wIh   and   pressure,   for   SA = SSO = 35.16504 g kg −1 .     This   is  
simply   35.16504 g kg −1  times  Figure  Ice_14(a).      
  (b)  Contour  plot  of  panel  (a)  with  the  values  of   SA δΘ δ SA frazil  evaluated  
at  ice  mass  fraction   wIh = 0  subtracted  at  each  pressure.      
  (c)  The  difference  between   δΘ δ P frazil  and  the  corresponding  derivative  
of   Θ freezing   with   pressure   at   constant   Absolute   Salinity,   ∂Θ freezing ∂P (obtained  
SA
from   gsw_CT_freezing_first_derivatives).     The   contoured   values   of   panel   (c)  
( ) −1
are   in   K Pa   and   the   seawater   salinity   was   taken   to   be  
SA = SSO = 35.16504 g kg −1 .     Notice   that   the   numbers   contoured   here   are   only   a  
few  percent  of  those  of   δΘ δ P frazil  shown  in  Fig.  Ice_14(b).      
 
The  dependence  of   SA δΘ δ SA frazil  on  the  mass  fraction  of  ice  is  illustrated  in  
Fig.  Ice_15(b)  which  shows  the  difference  relative  to  the  case  when   wIh = 0 .  
 
 
 
 
 

87
Thermodynamics Lectures, MIT, 2015 88
When   no   frazil   is   present   in   seawater,   its   Conservative   Temperature   is  
unaffected   by   adiabatic   and   isohaline   changes   in   pressure,   but   its   in   situ  
temperature   changes   with   pressure   according   to   the   adiabatic   lapse   rate   Γ  
which   is   usually   positive.     When   frazil   is   present   in   seawater,   an   increase   in  
pressure   results   in   changes   in   Conservative   Temperature   as   contoured   in    
Fig.   Ice_14(b).     This   dependence   of   the   temperature   (both   Conservative  
Temperature  and  in  situ  temperature)  of  the  frazil-­‐‑seawater  mixture  to  changes  
in   pressure   is   rather   large   and   negative   compared   with   the   (usually   positive)  
adiabatic  lapse  rate  of  seawater  which  is  typically  less  than  one  twentieth  of  the  
values  shown  in  Fig.  Ice_14(b)  for   δΘ δ P frazil ,  and  is  usually  of  the  opposite  sign.    
Another  way  of  stating  this  is  that  the  adiabatic  lapse  rate  of  the  frazil-­‐‑seawater  
mixture   is   large   and   negative   when   frazil   is   present,   compared   with   the   small  
and  positive  adiabatic  lapse  rate  of  seawater  in  the  absence  of  frazil.      

Note   that   the   rate   at   which   the   freezing   Conservative   Temperature   changes  
with   Absolute   Salinity   at   fixed   pressure,   ∂Θ freezing ∂SA ,   is   quite   different   (even  
p
different  signs)  to  the  corresponding  change  involving  frazil  ice  as  the  pressure  
varies,   δΘ δ SA frazil .    A  typical  value  of   ∂Θ freezing ∂SA  is   − 0.0583 K g −1 kg  while  
p
a  typical  value  of   δΘ δ SA frazil  is   2.3 K g −1 kg .    By  contrast,  we  have  seen  that  the  
variation  of  Conservative  Temperature  with  pressure  for  frazil  ice,   δΘ δ P frazil ,  is  
only   a   few   percent   different   to   the   corresponding   change   at   constant   Absolute  
Salinity,   ∂Θ freezing ∂P .      
SA
 
All   of   the   material   above   in   these   lectures   concerning   ice,   sea   ice,   and   frazil  
ice  can  be  found  in  the  paper      
McDougall, T. J., P. M. Barker, R. Feistel and B. K. Galton-Fenzi, 2014:
Melting of ice and sea ice into seawater, and frazil ice formation. Journal of
Physical Oceanography, 44, 1751-1775.  
The   properties   of   ice   and   its   equilibrium   properties   with   seawater   can   be  
evaluated   using   the   GSW   Oceanographic   Toolbox,   available   from   www.TEOS-­‐‑
10.org.    
 

88
Thermodynamics Lectures, MIT, 2015 89

The  interaction  between  ice  and  seawater  in  ocean  models    

In   a   model,   the   mixture   of   seawater   and   frazil   ice   must   be   advected   and  
diffused,   and   the   frazil   crystals   must   be   allowed   to   migrate   vertically   (Stokes  
drift),  but  two  questions  arise.    
1 What  variables  should  the  model  carry  to  conserve  salt  and  “heat”?    
2 How  should  thermodynamic  equilibrium  be  re-­‐‑established  after  the  
advection,  diffusion  and  frazil-­‐‑Stokes-­‐‑drift  part  of  each  time  step?    
 
 
Bulk  Absolute  Salinity  and  Bulk  potential  enthalpy  
We  have  shown  that  conserving  the  potential  enthalpy  of  ice  Ih  is  sufficiently  
accurate  (rather  than  having  to  conserve  the  enthalpy  of  ice  Ih  during  
melting/freezing  and  then  worry  about  how  the  enthalpy  of  ice  Ih  varies  with  
pressure).      

This  greatly  simplifies  our  task  because  the  First  Law  of  Thermodynamics  can  
be  simplified  to  be  the  conservation  of  the  potential  enthalpy  of  the  seawater-­‐‑ice  
mixture.      

Along  with  the  ice  mass  fraction,   wIh ,  the  conserved  model  variables  during  
the  advection  and  diffusion  part  of  the  time  step  should  then  be  the  “Bulk  
Absolute  Salinity”   SAB  and  the  “Bulk  potential  enthalpy”,   ! B ,    

( ) ( )
SAB ≡ 1 − wIh SA                      and                      B ≡ 1 − wIh c0pΘ + wIh  Ih .      

After  the  advection,  diffusion  and  frazil-­‐‑Stokes-­‐‑drift  part  of  the  time  step,  we  
B
have  values  of   SA1  and   ! 1B ,  for  the  model  box,  but  these  values  will  not  be  in  
thermodynamic  equilibrium  with  each  other.      

How  do  we  re-­‐‑establish  thermodynamic  equilibrium  in  the  second  half  of  the  
time  step?      
 
 
The  thermodynamic  equilibrium  condition  between  seawater  and  frazil  ice  
During  the  equilibration  process,  there  is  no  exchange  of  mass,  salt  or  heat  with  
neighbouring  boxes,  so  the  Bulk  Absolute  Salinity  and  the  Bulk  potential  
enthalpy  are  conserved,  so  that  when  thermodynamic  equilibrium  is  reached  at  
the  end  of  the  full  time  step  we  must  have      

(1 − w ) S
Ih
A
B
= SA1 ( )
                   and               1 − wIh c0pΘ + wIh ! Ih = ! 1B ,    

and  in  addition,  the  values  of  all  these  thermodynamic  variables  mutually  adjust  
so  that  they  satisfy  the  freezing  condition  at  the  end  of  the  full  time  step.      

Hence  we  seek  the  zero  of  the  function  of  the  ice  mass  fraction   wIh        

( )
0 = f wIh ( )
= ! 1B − 1 − wIh c0pΘ freezing ( SA , p ) − wIh ! Ih
freezing ( S A , p ) .    

where   SA  is  related  to   wIh  by   SA = SA1

B

Ih
( ) ( )
1 − wIh .    Note  that   f wIh  is  indeed  a  
function  of  only  the  ice  mass  fraction   w .    
 

89
Thermodynamics Lectures, MIT, 2015 90
To  solve f w ( ) Ih
= 0  we  use  an  improved  version  of  Newton’s  Method  which  
converges  at  the  rate   1 + 2    and  is  described  in  McDougall T. J. and S. J. Wotherspoon,
2014: A simple modification of Newton’s method to achieve convergence of order 1 + 2 . Applied
Mathematics Letters, 29, 20-25. http://dx.doi.org/10.1016/j.aml.2013.10.008 .      
 
( )
The  derivative   f ′ wIh  is  given  by  

∂Θ freezing
( )
f ′ wIh = c0pΘ freezing ( SA , p ) − ! Ih
freezing ( S A , p ) − S A c p
0
∂SA
p
   
SA wIh ∂! Ih
freezing
− .
(1 − w ) Ih ∂SA
p

 
 
An  efficient  way  of  determining  when  there  is  no  frazil  ice  component      
If,  at  the  end  of  the  first  part  of  the  time  step   ! 1B  is  sufficiently  “warm”,  there  
will  be  no  frazil  ice.    In  this  case  the  solution  is    

wIh = 0 ,                     SA = SA1
B
               and                         Θ = ! 1B c0p .    
What  is  an  efficient  way  of  detecting  when   ! 1B  is  “too  warm”  for  frazil  ice  to  
be  present?    Based  on  the  definition  of  the  equilibrium  condition  (repeated  from  
above),    

0 = f wIh ( ) ( )
= ! 1B − 1 − wIh c0pΘ freezing ( SA , p ) − wIh ! Ih
freezing ( S A , p ) ,    
we  evaluate  this  function  when  the  ice  mass  fraction  is  zero,  that  is,    

(
  f wIh = 0 ) = ! 1B − c0pΘ freezing SA1
B
(
, p ,   )  

and  if  this  is  positive  then  the  answer  is  simply  seawater  (and  no  frazil  ice),  so  
that  we  can  set,   wIh = 0 ,   SA = SA1 B
 and   Θ = ! 1B c0p .    The  computer  time  
involved  with  making  this  decision  is  simply  the  time  it  takes  to  evaluate  the  
freezing  Conservative  Temperature   Θ freezing SA1 , p .       ( )
The  computer  code  needed  to  restore  the  seawater/ice  properties  at  the  end  of  
B
the  first  half  of  the  time  step   SA1 ( )
, ! 1B , p  to  thermodynamic  equilibrium  is  
contained  in  the  computationally  efficient  code      
gsw_frazil_properties_potential_poly(SA_bulk,  h_pot_bulk,  p)    

(
which  has  outputs  of   SA , Θ, wIh .     )
 
 

90
Thermodynamics Lectures, MIT, 2015 91
Buoyancy  frequency  N  

 
  First  consider  the  incompressible  situation  as  illustrated  in  the  figure.    The  
figure  shows  that  when  a  parcel  is  displaced  upwards  from  its  resting  position  
in  a  stably  stratified  fluid,  it  experiences  a  downwards  buoyant  force  because  it  
is  denser  than  the  fluid  of  the  environment  that  surrounds  it  at  its  new  location.    
This  force  is  indicated  by  the  downwards-­‐‑directed  arrow  in  the  figure.      
  When  the  fluid  is  compressible  there  is  a  vertical  gradient  of  in  situ  density  
ρ ,  given  by    
∂ρ
P = ρκ Pz ,     (vertical  isentropic  density  gradient)  
∂P S ,Θ z
A

even   when   a   fluid   layer   is   completely   well   mixed   so   that   Absolute   Salinity,  
entropy   and   Conservative   Temperature   are   all   independent   of   height.     In   this  
compressible  well-­‐‑mixed  case,  the  fluid  parcel  illustrated  above  would  decrease  
its  in  situ  density  in  moving  upwards  by  the  distance   h ,  but  at  its  new  location,  
its  density  would  be  the  same  as  that  of  the  fluid  around  it  at  this  height.    So  in  
order   to   quantify   the   vertical   stability,   that   is,   in   order   to   quantify   the   vertical  
buoyant   force   that   the   parcel   experiences   at   its   new   location,   we   need   to   take  
into   account   this   vertical   gradient   of   in   situ   density   ρ   due   to   the   fluid’s  
isentropic  (and  isohaline)  compressibility   κ .    
  The  square  of  the  buoyancy  frequency  (sometimes  called  the  Brunt-­‐‑Väisälä  
frequency),   N 2 ,   is   given   in   terms   of   the   vertical   gradients   of   density   and  
pressure,  or  in  terms  of  the  vertical  gradients  of  Conservative  Temperature  and  
Absolute   Salinity   by   (the   g   on   the   left-­‐‑hand   side   is   the   gravitational  
acceleration,  and  x,  y  and  z  are  the  spatial  Cartesian  coordinates)    

(
g −1 N 2 = − ρ −1ρ z + κ Pz = − ρ −1 ρ z − Pz / c 2 )   (3.10.1)  
= α Θ Θz − β Θ ∂SA ∂z .
x, y x, y

The  buoyancy  frequency   N  has  units  of  radians  per  second,  and  since  a  radian  
is  unitless,   N  has  dimensions  of   s −1 .    The  buoyancy  frequency   N  is  the  highest  
frequency  of  internal  gravity  waves  in  a  density-­‐‑stratified  fluid  like  the  ocean  or  
atmosphere.     The   corresponding   shortest   period   of   internal   gravity   waves   is  
2π N  which  varies  from  about  20  minutes  in  the  upper  ocean  to  a  few  hours  in  
the   deep   ocean.     (This   is   to   be   compared   with   2π f ≥ 12 hours   where  
f = 2 Ω sin φ = 1.458 423 00 x 10−4 sin φ s−1 ,  is  the  Coriolis  parameter    where   φ  is  
latitude  and   Ω  is  the  rotation  rate  of  the  earth  [in  radians  per  second]).      
  For   two   seawater   parcels   separated   by   a   small   distance   Δz   in   the   vertical,  
an   equally   accurate   method   of   calculating   the   buoyancy   frequency   is   to   bring  
both   seawater   parcels   adiabatically   and   without   exchange   of   matter   to   the  
average   pressure   and   to   calculate   the   difference   in   density   of   the   two   parcels  
after   this   change   in   pressure.     In   this   way   the   potential   densities,   defined   with  
reference   pressure   being   the   mean   pressure   of   the   two   fluid   parcels,   are   being  
compared   at   the   same   pressure.     This   common   procedure   calculates   the  
buoyancy  frequency   N  according  to    

91
Thermodynamics Lectures, MIT, 2015 92
Θ
(
N 2 = g α Θ Θ z − β Θ SA z ) ≈ − ρg ΔΔρz ,   (3.10.2a)  

or    
g 2 Δρ Θ
(
    N 2 = g 2 ρ β Θ SA − α Θ ΘP ≈
P ) ΔP
,   (3.10.2b)  

where   Δρ Θ  is  the  difference  between  the  potential  densities  of  the  two  seawater  
parcels   with   the   reference   pressure   being   the   average   of   the   two   original  
pressures  of  the  seawater  parcels.    Eqn.  (3.10.2b)  has  made  use  of  the  hydrostatic  
relation   Pz = − g ρ ,  and   ΔP  is  the  difference  in  the  pressures  of  the  two  parcels,  
in   Pa .      
  This  difference  in  potential  density,   Δρ Θ ,  between  two  seawater  parcels  can  
be  evaluated  more  easily  when  density  is  expressed  in  the  form   ρ = ρ̂ SA , Θ, p   ( )
(
than  when  it  is  expressed  in  the  form   ρ = ρ SA , t, p ;  witness     )
( ) (
Δρ Θ = ρ̂ SAdeep , Θdeep , p − ρ̂ SAshallow , Θshallow , p )  
( ( ) ) ( ( ) )
= ρ SAdeep , θ SAdeep , t deep , pdeep , p , p − ρ SAshallow , θ SAshallow , t shallow , pshallow , p , p

( )
where   p = 12 p deep + p shallow .    Compared  with  the  first  line  of  the  above  equation,  
the  second  line  requires  more  calculations,  and  the  expression  is  unnecessarily  
convoluted.      
  The  “Stability  Ratio”   Rρ  of  a  vertical  water  column  is  defined  as    
α ΘΘ z
Rρ = .   (3.15.1)  
β Θ ( SA )z
Rρ   is   the   ratio   of   the   vertical   contribution   from   Conservative   Temperature   to  
that  from  Absolute  Salinity  to  the  static  stability   N 2  of  the  water  column.    
 
 
The neutral tangent plane
The   “neutral   tangent   plane”   is   that   plane   in   physical   space   in   which   the   local  
parcel  of  seawater  can  be  moved  an  infinitesimal  distance  without  being  subject  
to  a  vertical  buoyant  restoring  force;  it  is  the  plane  of  neutral-­‐‑  or  zero-­‐‑  buoyancy.      

 
Take   the   seawater   parcel   at   the   central   point   and   enclose   it   in   an   insulating  
plastic   bag,   then   move   it   to   a   new   location   a   small   distance   away.     Its   density  
will  change  by   δρ = ρκδ P .    At  the  same  location  the  seawater  environment  has  a  
( )
density   difference   of   δρ = ρ κδ P + β Θδ SA − α ΘδΘ .     If   the   seawater   parcel   is  
happy   to   sit   still   at   its   new   location,   it   must   not   be   feeling   a   vertical   buoyant  
(Archimedean)   force,   and   this   requires   that   its   density   is   equal   to   that   of   the  
environment  at  its  new  location.    That  is,  we  must  have    

(
ρκδ P = ρ κδ P + β Θδ SA − α ΘδΘ .   ) (Neutral_1)  
Hence,  along  a  neutral  trajectory  the  variations  of   SA  and   Θ  of  the  ocean  obey    
92
Thermodynamics Lectures, MIT, 2015 93
Θ Θ
β δ SA = α δΘ .   (Neutral_2)  
  Incidentally,  this  thought  experiment  involving  the  adiabatic  and  isohaline  
displacement   of   fluid   parcels   is   typical   of   our   thinking   about   turbulent   fluxes.    
We  imagine  the  adiabatic  and  isohaline  movement  of  fluid  parcels,  and  then  we  
let  these  parcels  mix  molecularly  with  their  surroundings.    Central  to  this  way  of  
thinking  about  turbulent  fluxes  are  the  following  two  desirable  properties  of  the  
tracer  that  is  being  mixed.      
  (1)   it   must   be   a   “potential”   property,   for   otherwise   its   value   will   change  
during  the  adiabatic  and  isohaline  displacement  so  it  is  difficult  to  define  a  flux  
of  the  quantity,  and    
  (2)   it   should   preferably   be   a   “conservative”   fluid   property   so   that   when   it  
does   mix   intimately   (that   is,   molecularly)   with   its   surrounding,   we   can   be   sure  
that   no   funny   business   is   going   on;   no   magic,   undesirable   production   or  
destruction  of  the  property.      
Expressing  this  definition  of  a  neutral  tangent  plane   β Θδ SA = α ΘδΘ  in  terms  
of   the   two-­‐‑dimensional   gradient   of   properties   in   the   neutral   tangent   plane,   we  
have  that      

(
− ρ −1∇ n ρ + κ ∇ n P = − ρ −1 ∇ n ρ − ∇ n P / c 2 )=α Θ
∇ nΘ − β Θ ∇ n SA = 0 ,   (3.11.2)  

where,   by   way   of   reminder   the   relevant   thermal   expansion   coefficient   α Θ   and  

saline  contraction  coefficient   β Θ  are  defined  by    
1 ∂ρ 1 ∂ρ
αΘ = −              and             β Θ = .   (2.18.3),  (2.19.3)  
ρ ∂Θ S ρ ∂SA
A, p Θ, p

  Here   ∇n  is  an  example  of  a  projected  non-­‐‑orthogonal  gradient    

∂τ ∂τ
∇rτ ≡ ∂x r
i + ∂y r
j + 0 k ,   (3.11.3)  

that   is   widely   used   in   oceanic   and   atmospheric   theory   and   modelling.    

Horizontal   distances   are   measured   between   the   vertical   planes   of   constant  
latitude  x  and  longitude  y  while  the  values  of  the  property   τ  are  evaluated  on  
the   r  surface  (e.  g.  an  isopycnal  surface,  or  in  the  case  of   ∇n ,  a  neutral  tangent  
plane).    Note  that   ∇rτ  has  no  vertical  component;  it  is  not  directed  along  the   r  
surface,  but  rather  it  points  in  exactly  the  horizontal  direction.      

93
Thermodynamics Lectures, MIT, 2015 94
  A   very   accurate   finite   amplitude   version   of   achieving   β Θδ SA = α ΘδΘ   is   to  
equate   the   potential   densities   of   the   two   fluid   parcels,   each   referenced   to   the  
( )
average  pressure   p = 0.5 p a + p b .    In  this  way,  when  two  parcels,  parcels  a  and  
( ) ( )
b,   are   on   a   neutral   tangent   plane   then   ρ̂ SAa ,Θa , p = ρ̂ SAb ,Θ b , p ;   see   the   figure  
below  which  involves  the  thought  process  of  moving  both  parcels  to  pressure   p .      

 
  The   (three   dimensional)   normal   vector   to   the   neutral   tangent   plane   n   is  
given  by    

(
g −1 N 2 n = − ρ −1∇ρ + κ∇P = − ρ −1 ∇ρ − ∇P / c 2 )   (3.11.1)  
Θ Θ
= α ∇Θ − β ∇SA .
As  defined,   n  is  not  quite  a  unit  normal  vector,  rather  its  vertical  component  is  
exactly   k ,  that  is,  its  vertical  component  is  unity  ( k ⋅ n = 1 ).      
 

94