source: trunk/LMDZ.MARS/README @ 357

== 17/09/08 ==
>>> Build a version with new soil but old radiative transfer,
    but keeping possibility of switching back to new radiative transfer),
    which incorporates changes & improvements currently included in the
    'reference version' GCM (see /u/emlmd/LMDZ.MARS.mixdyn)

>>> start by modifying makegcm as in /u/emlmd/LMDZ.MARS.mixdyn, so that it runs
    without environment variables and set LIBOGCM to /tmp15/emlmd/libo

>>> directory contents of 'aeronomars', 'grid' and 'filtrez' are simillar
    to those in /u/emlmd/LMDZ.MARS.mixdyn

>>> in bibio , only file mxva.F needed be upgraded

>>> get phymars and dyn3d contents from /u/emlmd/LMDZ.MARS.mixdyn
    (and remove all *old files)

>>> check differences between dyn3d and /u/emlmd/LMDZ.MARS.170908/libf/dyn3d
    and upgrade when necessary:
    - removed 'netcdf.inc' file (has nothing to do there!)
    - comgeom.h and comgeom.h : made fortran90 compliant
    - control.h : made fortran90 compliant
    - dynredem.F : more read/write controls + comments in english
    - ini_archive.F : new soil/thermal inertia changes
    - integrd.F : added additional information to output when crashing
    - lect_start_archive.F : new soil/thermal inertia changes
    - newstart.F : new soil/thermal inertia changes + comments in english
    - start2archive.F : new soil/thermal inertia changes
    - vanleer.F : removed inapropriate 'external' statement
    - write_archive.F : enable writting a subterranean field

>>> check differences between phymars and /u/emlmd/LMDZ.MARS.170908/libf/phymars
    and upgrade when necessary:
    New soil stuff:
    - added comsoil.h
    - iniwrite.F : new soil changes
    - added iniwritesoil.F90 and writediagsoil.F90 for subterranean fields
    - added interp_line.F (for subterranean grid interpolation)
    - adapted phyetat0.F for subterranean temperature & inertia
    - adapted physdem1.F to include new soil stuff
    - physiq.F : added calls to writediagsoil 
    - soil.F : new routine (fixed vertical grid + variable thermal inertia)
    - added soil_settings.F (to read/initialize/interpolate soil properties)
    - updated surfdat.h (since thermal inertia is now in comsoil.h)
    - updated tabfi.F : include new soil properties
    - updated testphys1d.F
    - updated dimphys.h (set nsoil=18 as default)

== 18/09/08 ==
>>> add the possibility of easily switching to Tran radiative transfert
    - updated aerdust.h.ocke97 (changed some variables name) so it can
      replace aerdust.h (which is currently the same as aerdust.h.clan91).
    - imported Tran's 'gfluxv.F' routine
    - imported Trans' version of 'swr.F' routine, saved it as 'swr.F.toon'

>>> Backup of 'old' Morcrette swr.F is 'swr.F.morc'
    NB: to switch from one radiative transfer to the other, just copy
    swr.F.morc or swr.F.toon to swr.F (and eventually 'touch swr.F' so that
    makegcm recompiles swr.F)
    No other dependencies (swr.F.toon uses 'gfluxv.F' and swr.F.morc uses
    'dedd.F').

>>> Changed the latter, so that users can switch from one to the other
    - modified swr.F.toon to become swr_toon.F (and to include gfluxv.F)
    - modified swr.F.morc to become swr_fouquart.F (and to include dedd.F)
    - added a flag in callkeys.h, swrtype (parameter to be set/changed by
      the user 1=Fouquart and 2=Toon)
    - update readtesassim so that the coefficient by which opacity
      is multiplied is set according to the 'swrtype' parameter

== 25/09/08 ==
>>> Implement the use of tracer-by-name in physics
    - in phymars/tracer.h set 'noms' length to 20 (instead of 10)
    - in phymars/callsedim2q.F and phymars/callsedim.F, use tracers by name
    - in phymars/dustopacity.F, use tracers by name
    - in phymars/vdifc.F, use tracers by name

== 26/09/08 ==
>>> Change implementation strategy (for now); don't move surface tracer around
    i.e.: surface ice remains equivalent to qsurf(nqmx)=qsurf(i_h2o_vap)
          and likewise for surface tendencies ...
    - modified vdifc.F and callsedim.F back
    - modified initracer.F (so that water names are h2o_vap & h2o_ice)
    
== 29/09/08 ==
    - modified aeronomars/init_chimie_B (cosmetics)
    - corrected aeronomars/moldiff.F internal routine tridag; changed
      "pause" error messages to 'stop' messages
    - modified phymars/watercloud.F to use tracers by name
    - corrected aeronomars/molvis.F (undefined 'fac' and 'Akk' written to 
       output at first call)
    
== 30/09/08 ==
    - modified aeronomars/calchim.F to use tracers by name
    - adapted aeronomars/photochemist_B.F to use tracers by name
    - adapted aeronomars/chemtermos.F to use tracers by name
    - adapted aeronomars/concentrations.F to use tracers by name
    - corrected aeronomars/conduction.F (undefined 'Akk' written to output
      at first call)
    - adapted aeronomars/euvheat.F to use tracers by name
    - adapted aeronomars/moldiff.F and moldiffcoeff.F to use tracers by name

== 01/10/08 ==
    - For more compatibility with LMDZ4; mimic reading a 'traceur.def' file
      in the dynamics via a call to a routine 'iniadvtrac.F' and saving
      tracers names in 'advtrac.h'
     -> created 'iniadvtrac.F', 'advtrac.h' and modified gcm.F
    - modified 'dynetat0.F' so that tracers are loaded from 'start.nc' by name
    - modified 'dynredem.F' so that tracers are written to 'restart.nc'
      by name
    - modified 'initracer.F' tu use tracers by name

== 02/10/08 ==
    - removed use of 'nqchem_min' everywhere:
      adapted 'euvheat.F','inifis.F','physiq.F'
      (leave 'inichim.F' for later)
    - updated 'phyetat0' and 'physdem1.F' to read/write surface tracers by name
    - modify things so that surface water ice index is the same as
      atmospheric water ice (except when running without water ice; then
      simply set i_h2o_ice=i_h2o_vap).
      NB: the easiest is to have global storage of tracer names/indexes in
      tracer.h
      => changed initracer.F & tracer.h to have global igcm_something indexes

== 03/10/08 ==
    - adaptations for surface ice index, modified files:
      phyetat0.F : if there is a dynamical tracer 'h2o_vap' then load
                   surface tracer called 'h2o_ice' instead
      initracer.F : in 'old' tracer name case: move qsurf(nqmx)->qsurf(nqmx-1)
                    and set i_h2o_ice=i_h2o_vap if iceparty=.false.
      physdem1.F : if old tracer names: move qsurf(nqmx-1)->qsurf(nqmx)
                   if iceparty=.false., write surface tracer 'h2o_ice'
                   (and not 'h2o_vap') to file.
      adapted vdifc.F, callsedim.F & watercloud.F & physiq.F so that surface
      ice is now identified as qsurf(i_h2o_ice)
    - updated aeronomars/perosat.F (cosmetics)

== 06/10/08 ==
    - modify newstart.F and lect_start_archive.F to use tracers by name
== 07/10/08 ==
    - adapted inichim_newstart.F (added qsurf to arguments)
      and inichim_readcallphys.F
== 08/10/08 ==
    - implement reading traceur.def in dyn3d/iniadvtrac.F

== 16/10/08==
    -small change in inifis.F (only warn if too many tracers, compared
     to the expected number, not stop).
     - corrected bug in initracer.F
== 21/10/08 ==
    - modified newstart.F to load B.Diez subsurface ice maps.
    - corrected small bug (uninitialized variable) in interp_horiz.F
== 22/10/08 ==
    - updated iniwritediagsoil.F so that thermal inertia is written to
      diagsoil.nc
== 31/10/08 ==
    - changed xvik.F program so it works even if we don't have atmospheric
     temperature at hand (then it uses a 10km reference scale height) and so
     that it does surface pressure interpolation log-wise.
== 03/11/08 ==
    - modified physiq to compute (and output) co2 column.
    - added improvement by Francois in newcondens.F about computing CO2
      partial pressure. This behavior is turned on by setting internal logical
      flag 'improved_ztcond' to '.true.' (and running with a co2 tracer)
    - updated 'start2archive' to work with 'new' gcm output (soil, tracers ...)
== 04/11/08 ==
    - upgraded xvik program to look for temperature in 7th layer variable if
     there is no global atmospheric temperature field at hand.
== 05/11/08 ==
    - more modifs to newcondens.F: added another internal flag 'bound_qco2' to
     enforce (if set to .true.) that co2 mass mixing ratio remains bounded.
== 07/11/08 ==
    - corrected 'writediagfi' & 'writediagsoil' so that an error message is 
      issued if called with a variable name which is too long.
== 18/12/08 ==
    - corrected bug in dyn3d 'addfi.F', (dimensions of local array p())
== 23/02/09 ==
    - modified "aeronomars/param_read.F" to do strictly fortran data
     initialization (otherwise xlf compiler complains)
    - changed a few '1.e-30' to '1.d-30' in aeronomars/photochemist_B.F
     so that max functions has 2 doubles as arguments (otherwise xlf
     compiler complains)

==07/04/09 ==
    -cosmetic changes/minor improvements in the handling of tracers in:
     aeronomars/photochemist_B.F
     aeronomars/perosat.F
     aeronomoars/euvheat.F
     aeronomars/moldiffcoeff.F
     aeronomars/moldiff.F
     aeronomars/cocentrations.F
     aeronomars/chemtermos.F
     aeronomars/calchim.F

--> NB: still there are differences in outputs when order of tracers is changed

== 09/04/09 ==
>>> fixed problem in 'vdifc.F' which lead to different results when moving
    tracers around.

== 10/04/09 ==
>>> corrected small bug in diagnostic outputs of 'watercloud.F' (tendencies were
    not added to tracer values).

== 21/04/09 ==
>>> corrected small bug in "physdem1.F" about writing water ice surface tracer
    to file

== 07/05/09 ==
>>> very minor correction (firstcall not set to true after first call
    if no tracers) in convadj.F

== 30/06/09 ==
>>> Implement reading *def files with IOIPSL ersatz 'getin' function
   - import "ioipsl_errioipsl.F90","ioipsl_getincom.F90","ioipsl_stringop.F90"
     in bibio
   - adapted 'dyn3d/defrun_new.F' to use "getin" function
   - adapted 'phymars/inifis.F' to use "getin" function

== 01/07/09 ==
>>> Adapted 'create_make_gcm' so that the "use" in *.F files is identified and
    corresponding dependencies included in the makefile rules.

>>> Added the 3D scattering from aerosols by JB Madeleine:
    - minor changes in aerave.F
    - added the calls to aeropacity.F, and aeroptproperties.F in callradite.F
    - changed the calls to lwu.F and swr.F in lwmain.F and swmain.F, respectively
    - added 3D scattering properties in lwu.F and swr.F
    - added the new aeroptproperties.F, aeropacity.F and suaer.F90 routines
        (removed dustopacity.F)
    - updated aeropacity.F with new tracer names
    - changed the call to callradite.F in physiq.F, added the initialization 
        of reffrad and nueffrad (aerosol effective radius and variance)
    - removed all the lines relative to the old "activice" option, including
        temperature variation due to latent heat release (now in comments)
    - renamed nsize into naersize in watercloud.F, watersat.F and newsedim.F,
        to avoid conflicts with another "nsize" variable in the radiative transfer
    - added the statement of nuice in watercloud.F, which is the effective variance
        of the log-normal distribution for ice
    - updated yomaer.h and removed aerice.h (and corresponding "includes")

== 02/07/09 ==
>>> Adapted 'aeronomars/inichim_readcallphys.F' (called by newstart)
     to use "getin" routine.
    + minor correction in 'inifis.F' (close 'iradia.def' file)
>>> Minor correction in 'dyn3d/dynetat0.F' and 'phymars/phyetat0.F'; do not
    attempt to reindex tracers if none were found.

>>> in 'deftank' added examples of 'traceur.def' files (traceur.def.co2 : 1
    co2 tracer; traceur.def.watercycle : 2 traceurs, water vapour and water ice
    tracer.def.chemistry : all 15 species)

== 06/07/09 ==
>>> Modified 'makegcm' and makegcm_g95' so that modules files are put 
    with libraries (and not in current directory)
    
== 21/07/09 ==
>>> Added in "testphys1d.F" a check that the (required) 'run.def' file is
    around (that file should contain the "INCLUDEDEF=callphys.def" instruction
    otherwise getin() calls won't work.
    Also added reding of 'traceur.def' (or initialisation of tracer names
    to dummy values q01,q02, ...) in testphys1d.F

== 22-24/07/09 == JBM
>>> Removed "iceparty" everywhere (calchim.F, inichim_readcallphys.F,
        chimie_data.h, inichim_newstart.F, photochemist_B.F, callkeys.h,
        callradite.F, callsedim.F, aeropacity.F, inifis.F, initracer.F, physdem1.F,
        physiq.F, watercloud.F). Water = .true. now implies the use of two
        tracers, i.e. water vapor and water ice.

>>> Removed fisice.h and stated the corresponding variables in the right
        places (physiq.F, callsedim.F, watercloud.F, newcondens.F, calchim.F);
        also removed unused cltop and clsurf variables in physiq.F.

>>> Removed the variable's splitting, which is now obsolete, in callradite.F
        (and its subroutines lwi.F, lwxb.F, lwxn.F, lwflux.F, lwmain.F, lwxd.F).
        Also removed the variable's splitting in calldrag_noro.F. Finally removed
        ndomain from dimradmars.h.

>>> Removed useless tests in aeropacity.F.

== 27-30/07/09 == JBM
>>> Cleaned the WRITEDIAGFI section in physiq.F, and moved the "mtot",
        "icetot" and "tauTES" variables from watercloud.F to physiq.F. Also
        cleaned the albedo change due to water ice deposition.

>>> Renamed rdust into rnuclei in callsedim.F, physiq.F and watercloud.F.

>>> Added a logical test for (water.and..not.tracer) in inifis.F.

>>> Removed "qsurf","zls" and "icount" from the list of inputs in
        watercloud.F (these variables were not used by the subroutine).

>>> Added a call to watercloud.F at firstcall, using typical dust optical
        depth (taufirstcall), in order to correctly initialize the ice particle
        size distribution for the radiative transfer scheme. Also created a new
        subroutine to load the effective radius and variance of the aerosols used
        by the radiative transfer scheme. Its name is updatereffrad.F, and it is
        called before aeroptproperties.F in callradite.F.

== 05/08/09 == JBM
>>> TES water-ice opacity is now fully computed using the radiatively active
        aerosol scheme. Absorption coefficient is calculated using Qext and
        omega at the IR reference wavelength. Omega was not computed before; it
        was only computed in the GCM channels, not at the reference wavelength.
        Thus it has been added in suaer.F90, aerave.F, yomaer.h, callradite.F,
        aeropacity.F and aeroptproperties.F. If "activice" is false, the
        TES opacity is computed using the old method (fit of the Qabs as a
        function of reff curve).

== 07/08/09 == EM
>>> Removed "iceparty" option from callphys.def

>>> modified physiq.F and initracer.F so that building of array niqchem() which
    contains the indexes of all chemistry tracers + water vapour and water
    ice is done in initracer.F (array niqchem(:) is now a common
    in tracer.h ).
    Also had to adapt inichim_newstart.F to behave similarly.

== 24/08/09 == EM included corrections by JBM
>>> Added declaration of nqchem(nqmx) local array in
    aeronomars/inichim_newstart.F

>>> Encapsulated calls to writediagfi & wstats in if (ngrid.ne.1) clause
    in phymars/aeropacity.F (otherwise it crashes in 1D).

== 26/08/09 == EM
>>> modified tracer.h, initracer.F, inichim_newstart.F and physiq.F to not use
    an niqchem() array (added in 07/08/09 changes)

>>> modified phymars/readtesassim.F90 and phymars/aeropacity.F so that
    assimilated dust for MY24 or MY25 or MY26 may be used (with iaervar=
    24,25 or 26); we keep iaervar=4 to also read MY24 dust for compatibility
    with older versions of code.
    Modified deftank/callphys.def : added comments about new iaervar values

== 25/09/09 == EM
>>> modified phymars/testphys1D.F : added incrementation of tracer values
    after call to physiq().

== 27/11/09 == EM
>>> updated comments in makegcm (translated help to english)

>> shifted to reading file traceur.def
   (dyn3d/iniadvtrac.F) in an Earth-LMDZ4 like way:
   first line == number of tracers
   and then tracer name (1 per line; later we'll add advection scheme type)
>> also updated example 'traceur.def' files in deftank accordingly

== 02/12/09 == EM
>>> upgraded testphys1d.F and profile.F to load run parameters with getin()
    function (from run.def ; no need for a "testphys1d.def" any more)
    added an example 'run.def.1d' file to 'deftank'

==10/12/09 == EM
>>> minor correction in testphys1d.F (was still checking if there is a
    testphys1d.def file around ; which is not used anymore)

==15/01/10 == JBM
>>> aeropacity.F: implemented a weighting of the dust opacity profile
  by using the dust size vertical profile defined in updatereffrad.F.
>>> aeroptproperties.F: changed the integration scheme (Gauss-Legendre)
  of the scattering parameters.
>>> suaer.F90: removed the use of an ad-hoc "solsir" factor. It is now
  directly computed from the scattering properties read in the ASCII
  files. Consequently, the IR extinction coefficient has been divided
  by solsir=2 in the ASCII file (called optprop_dustir_x0.5.dat instead
  of optprop_dustir.dat to allow compatibility with the older version
  and avoid chaos).
>>> updatereffrad.F: changed the dust size vertical profile.
>>> yomaer.h: the particle radius variable is now in simple precision,
  because the new scattering property integration scheme has changed.
>>> aeronomars/inichim_readcallphys.F (small) bug correction load value of
    'water' before testing its value...

== 18/01/10 == EM
>> added possibility to read (in inifis.F and aeropacity.F) the value of
   dust opacity tauvis from callphys.def file

== 01/02/10 == EM
>> added JBM updates of "callradite.F" (coments) "aeroptproperties.F" (bug fix
   of bad array bounds) and "aeropacity.F" (encapsulate wstats calls in
   if (callstats) )
>> added implementation of TES Cap albedos: "albedocaps.F90" and adapted
   "newcondens.F"  (and physiq.F, because added 'zls' argument to newcondens)
   and surfdat.h
>> changed default settings for dust : set a 1.3 factor in readtesassim.F90
   when using Toon radiative transfer, use M.Wolff-T-Matrix files in suaer.F90
 
== 03/02/10 == EM
>> Updated newstart.F in dyn3d, so that sub-surface thermal inertia values may
   be different in North and South hemispheres.

>> Updated "makegcm" and "makegcm_g95" scripts (cosmetic + default compilation
   option changes)
   
>> Minor changes in aeronomars/init_chimie_B.F (do not use lnblnk(); F90
   trim() intrinsic is much safer and better), and in initracer.F (better
   control over a possible array bound underflow).
   Also, in dyn3d/iniadvtrac.F, close input file properly,
   and in infis.F, more verbose message to output.

== 26/02/10 == EM
>> Updated makegcm and makegcm_g95 : default usage is now to set everything
   ("environment variables") in the script. Changed some default compilation
   options.
>>> removed 'float()' instructions in tabfi.F and iniwrite.F
    use "real()" to be compliant with standards.
>> Corrected small bug in testphys1d.F (look for file traceur.def)
   also added initialisation of tracers
>> Cleaned up inifis.F and initracer.F (some sanity checks were obsolete
   and/or wrong)
>> Improved writediagfi.F so that 1D (individual column in the GCM, or
   fields in testphys1d) data can be written in the diagfi.nc file
>> Minor changes/improvements in calls to writediagfi from physiq.F for dust

== 08/03/10 == EM
>> Minor update of "makegcm" and "makegcm_g95": use instruction "./makdim"
   (instead of "makdim"; in case "." is not in user's path)
>> put "real()" instructions instead of "float()" in dyn3d routines:
   disvert.F , dynredem.F , fluxstoke.F , fxhyp.F , fxy.F ,
   fxysinus.F , fyhyp.F , gcm.F , grid_atob.F , grid_noro.F , grid_noro1.F ,
   ini_archive.F , inigeom.F , newstart.F , ran1.F , sortvarc.F ,
   sortvarc0.F
>> Minor update of aeropacity.F (added if (callstats) around call to wstats)

== 28/04/10 == EM
>> Put the splitting in radiative transfer back in the model (JB):
   updated calldrag_noro.F  callradite.F  dimradmars.h  lwflux.F 
   lwi.F lwmain.F lwxb.F lwxd.F lwxn.F swmain.F
>> Fix bug (AS) in callradite.F (wrong loop boundaries line 332)
>> Fix bug (AS+JB) in "swr_toon.F" to enable running with more than
   100 levels...
>> Fix bug (JBM) in callsedim2q.F about setting pdqs_sed(:niq(iq)) to zero

== 25/08/10 == EM
>> Add a 'makegcm_gfortran' for compiling with gfortran
   and a 'makegcm_ifort' for compiling with ifort (on Gnome)
   
== 03/09/10 == EM
>> Modifications to enable running in double precision (using starts in r4
   or r8); just add options '-r8 -DNC_DOUBLE' to compile GCM in double precision
   -> adapted dyn3d/dynetat0.F, physmars/physdem1.F, phymars/soil_settings.F,
      phymars/readtesassim.F90, phymars/writediagfi.F, phymars/def_var.F90,
      phymars/writediagsoil.F90, phymars/wstats.F90, phymars/inistats.F
>> Added dyn3d/writediagdyn.F90 routine (to output scalar dynamical fields),
   adapted 'comconst.h' and 'comvert.h' to be Fortran77/Fortran 90 compatible.

== 14/12/10 == EM
>> Add -f option to #!/bin/csh in makegcm* scripts (to make sure that it is
   the bash environment compiler that is used as a default)
>> Update convadj.F with RW's version (fixes bug of non conservation of tracers
   in cases where convection stops at one level and starts at the next).

== 13/12/10 == EM
>> Update testphys1d.F so that initial tracer profiles may be loaded at
   initialization 

== 24/01/11 == JBM(+ some cleanup by EM)
>> Reactivated the "doubleq" method (two-moment scheme for dust
   transport) and connected it with the radiative transfer code. The
   opacity is set constant below a level indicated by the variable
   cstdustlevel in aeropacity.F to remove the thick layer of
   dust near the surface created by the constant lifting rate.
   The "density scaled opacity" used by the MCS team is computed
   and saved in dsodust.
   Updated routines: callradite.F, aeropacity.F, updatereffrad.F,
   callsedim.F, newsedim.F, initracer.F, vdifc.F, suaer.F90.
>> Added the use of named scatterers (same method as tracer-by-name)
   in the radiative transfer code. Scatterers are declared at
   firstcall in callradite.F (which is the equivalent of traceur.def)
   and the corresponding indices are saved in the common called
   aerkind.h (the equivalent of tracer.h).
   Updated routines: callradite.F, updatereffrad.F, aeropacity.F,
   suaer.F90, aerkind.h.
   EM: aerdust.h, aerdata.h and aerice.h are not used any more 
>> Merged callsedim2q.F and callsedim.F in one single routine
   callsedim.F to allow doubleq to be used with other tracers.
>> Added an input parameter called beta to newsedim.F, that allows
   to account for the shape of the particles in the computation of
   the sedimentation velocity.
>> Added the ability to transport a radiatively active population of
   submicron dust particles (flag "submicron" in callphys.def).
   Updated routines: callradite.F, updatereffrad.F, aeropacity.F,
   initracer.F, vdifc.F, suaer.F90, inifis.F, callkeys.h.
>> Connected the predicted size and amount of dust to the water
   cycle. The size of the particles is now given by rdust in
   updatereffrad.F, and the amount of cloud condensation nuclei
   (CCN) is given by ccn in aeropacity.F. The calculation of ccn is
   done in aeropacity.F because it's deduced from the opacity when
   doubleq is not used. rnuclei and dustcores are removed from
   watercloud.F and replaced by rdust and ccn.
   Updated routines: physiq.F, callradite.F, updatereffrad.F,
   aeropacity.F, watercloud.F.
>> Removed the "fake" call to watercloud.F in physiq.F which was used
   to give the size of the ice particles to the radiative transfer
   code at firstcall. Instead, rice is computed in updatereffrad.F
   using a simple equation and a typical amount of dust nuclei (ccn0).
>> Increased the number of Gauss integration points in
   aeroptproperties.F.
>> Added the ability to write the 3D scattering parameters of a
   given aerosol in the outputs using the out_qwg flag in
   aeroptproperties.
>> Changed "DO iir=1,4" into "DO iir=1,nir-1" in suaer.F90 (in case
   the number of infrared channels is changed).
>> Added nuice_ref in tracer.h and initracer.F, which is the
   effective variance of the log-normal distribution used for water-ice
   particles in the radiative transfer code.
>> Updated the computation of rice, reffice and rsedice in
   updatereffrad.F and callsedim.F.
>> Added nuice_sed in callsedim.F, which is the effective variance
   of the lognormal distribution used for the sedimentation of
   water-ice particles.
>> Added ccn_factor in watercloud.F, which is the ratio of the total
   number of dust particles over the number of condensation nuclei.
>> The variable beta is not saved anymore in newsedim.F.
>> Corrected the BIG bug in lwu.F that was responsible for unstabilities
   when clouds were radiatively active (FF+EM+JBM!)
>> Turned ilwd, ilwn and ilwb to 1 in inifis.F.
>> Added dust and ice visible opacities in the outputs.
   Modified routine: aeropacity.F.
>> Named water cycle tuneable parameters (nuice_sed, nuice_ref,
   alb_surfice, ccn_factor) which are mentioned at firstcall by the GCM
   (flag "water_param"). Modified routines: callsedim.F, physiq.F,
   watercloud.F.

== 25/01/2011 == EM
>> updated testphys1d.F: removed #include "aerdust.h"
>> cleanup in suaer.F90: no more calls to zerophys; added test to check
   there is no overflow of isize index (in visible domain averaged properties
   case)
>> minor cleanup in newsedim.F (make it more F90 oriented)
>> added flag 'TESicealbedo' (set to .true. to impose polar surface
   albedos as observed by TES) in callphys.def 
   
== 28/01/2011 == EM
>> added additional tests (to check correct reading of input files) in suaer.F90
>> updated inifis.F  so that the directory where external data files are to
   be found (e.g. TES opacities, dust properties, etc.) can be specified in
   run.def (or callphys.def) as " datadir = /path/to/the/directory "

== 29/01/2011 == AS
>> added updated mesoscale-related routines in phymars:
   ----------------------------------------------------------------------------
   NAME                       CHANGES compared to GCM counterpart 
   ----------------------------------------------------------------------------
   meso_callkeys.h        --> one variable is added [consider merging w/ GCM?] 
   meso_dustlift.F        --> stress + alpha default, or read in a file 
                               stress.def if here [consider merging w/ GCM?]
   meso_newcondens.F      --> correction on U V T tendencies is switched 
                               off (unstable in mesoscale) 
   meso_physiq.F          --> major modifications mainly related to I/O
   meso_slope.h           --> additional common for slope scheme
   meso_dimphys.h_ref     --> reference common serving as a basis for a 
                               compilation script (makemeso)
   meso_inifis.F          --> major modifications mainly related to I/O
   meso_param_slope.F90   --> slope scheme by Spiga and Forget GRL 2008 
                               [consider adding to GCM?]
   meso_readtesassim.F90  --> an old version because the new F90-compliant 
                               version needs the new makegcm scripts [TBD]
   meso_testphys1d.F      --> similar to GCM except for routine names
   ----------------------------------------------------------------------------
    NB: in meso_dustlift.F and meso_readtesassim.F90, the subroutines 
         have the same name as in the GCM.
         this is because those files are supposed to be copied 
         in specific temporary folders for compilation
>> any future change in the following GCM routines in phymars:
      - callkeys.h
      - dustlift.F
      - newcondens.F
      - physiq.F
      - inifis.F
      - readtesassim.F90
      - testphys1d.F
   will be in need to be impacted to the corresponding meso_ routines 
   [hence it is important to document this README file]
>> any change in any other GCM routines than the ones listed will
   have an effect in mesoscale simulations as well:
     -- the two models are being kept updated at the same time :)
     -- the two models would be possibly broken at the same time :(

== 15/02/2011 == EM
>> updated dissipation coefficients in indissip.F

== 25/02/2011 == EM
>> corrected bug in 'inifis.F' and 'datafile.h' to really be able to
   specify (in callphys.def; using datadir=/whatever/path/to/use ) the path
   to external datafiles (topography, surface properties, etc.)

== 28/02/2011 == JBM + AS
>> used settings reached by JBM to obtain his PhD results
      alb_surfice = 0.45  ---  in physiq.F and meso_physiq.F
      ccn_factor = 4.5    ---  in watercloud.F
      nuice_sed = 0.45    ---  in callsedim.F
>> NB: this is supposed to be further refined in the future

== 01/03/2011 == AS + JBM
>> nasty bug in the water cycle when iradia != 1  [no problem when iradia = 1]
   --> mesoscale runs w/ water cycle had strange 5-hour fluctuations in RICE, from 80mic to 5mic
>> PB: calculation of ccn [condensation nuclei] is done in callradite.F
      * ccn must be saved 
        --> corrected in physiq.F and meso_physiq.F
      * ccn must not be modified elsewhere [e.g. in watercloud, when divided by ccn_factor] 
        --> all calculations on ccn are now moved in callradite

== 03/03/2011 == AS
>> Added a pre-compilation flag MESOSCALE so that the LMDZ.MARS GCM
will compile without stating errors because of mesoscale routines.
[meso_physiq.F, meso_inifis.F]
>> Now, this MESOSCALE precompilation flag can be used to lower
the number of meso_* routines when adaptations for mesoscale
applications are not very extended.
   --> meso_testphys1d.F, meso_testphys1d.F, meso_dustlift.F
       routines were deleted and changes are now moved under 
       the MESOSCALE flag in the original GCM routines
   --> Completely transparent for GCM compilation 
       since it is devoid of the -DMESOSCALE option
   --> Very good for syncing because changes in dustlift, newcondens
       will be directly available in the mesoscale model

== 04/03/2011 == AS + JBM
>> new version version for aeroptproperties.F in phymars to limit uncertainties and be able to play with ngau
>> this was coded by JBM in his personal reference version but not transmitted to the team reference version

== 10/05/2011 == AS + JF
>> in newsedim.F used for mesoscale computations, spurious values close to the surface
   --> this was related to 1 - exp(-x) calculated as zero in w(ig,l) if x is very small
   --> fix: when this happens, replace exp(-x) by 1 - x since x ~ 0
>> in newsedim.F, "if (dztop.gt.epaisseur(ig,l)) then" was closed too soon by an "endif"
   --> hence basically the simple method was never used 
       and useless calculations with the complex method were carried out
   --> fix by moving the closing "endif" [in addition to corrections mentioned in the previous point]

== 17/05/2011 == EM
>> set internal computations using double precision in growthrate.F and
   watercloud.F (otherwise we sometimes end up with Nans).
>> add extra checks in newcondens.F to avoid possibility of out of bounds
   evaluation of array masse()
  
== 25/05/2011 == AS
>> found that the 10/05/2011 bug fix in newsedim.F is also useful for GCM runs.
>> no more need to modify callradite.F prior to compilation [but still dimradmars.h must be modified] 
   --> in callradite.F we now have
       -- DEFAULT name_iaer(1) is "dust_conrath"
       -- IF (doubleq.AND.active) name_iaer(1) = "dust_doubleq"
       -- IF (water.AND.activice) name_iaer(2) = "h2o_ice"  

== 27/05/2011 == EM
>> minor bug correction in writeg1d (JYC)
>> add "-check" to debug option in makegcm_ifort

== 08/06/2011 == EM
>> minor bug fix in lect_start_archive.F (using wrong surface temp. array).
   + swiched output messages to english and added that tracers not found
   in file must be initialized by user.
>> minor bug fix in datareadnc.F : 'datafile' path must be initialized.

== 08/06/2011 == EM
>> Significant update on how the number of scatterers is managed:
   Instead of having to manualy change 'nearkind' in dimradmars.h, the
   number of scatterers must now be set when compiling, using makegcm
   "makegcm -s 1" for one scatterer (dust) or "makegcm -s 2" (e.g. dust
   and water ice), default behaviour (ie not specifying -s #) is -s 1
   Modified phymars/dimradmars.h , added directory phymars/scatterers
   with script make_scatterers , and adapted makegcm* scripts.

== 17/06/2011 == AC
================================================
======== IMPLEMENTATION OF THERMALS ============
================================================
The main goal of this revision is to start including the thermals into the model
for development purposes. Users should not use the thermals yet, as
several major configuration changes still need to be done.

This version includes :
- updraft and downdraft parametrizations
- velocity in the thermal, including drag
- plume height analysis
- closure equation
- updraft transport of heat, tracers and momentum
- downdraft transport of heat

This model should not be used without upcoming developments, namely :
- downdraft transport of tracers and momentum
- updraft & downdraft transport of q2 (tke)
- revision of vdif_kc to compute q2 for non-stratified cases

Thermals could also include in a later revision :
- momentum loss during transport (horizontal drag)

Compilation of the thermals has been successfully tested on ifort, gfortran and pgf90

================================================
================================================

M      libf/phymars/callkeys.h
M      libf/phymars/inifis.F

Added new control flags to call the thermals :
- calltherm (false by default) <- to call thermals
- outptherm (false by default) <- to output thermal-related diagnostics (for dev purposes)
================================================
M      libf/phymars/vdifc.F
^------> added a temporary output for thermal-related diagnostics
M      libf/phymars/testphys1d.F
^------> added treatment for a initialization from a profile of neutral gas (ar)
      -> will be transformed in a decaying tracer for thermal diagnostics
M      libf/phymars/physiq.F
^------> added a section to call the thermals
      -> changed the call to convadj
      -> added thermal-related outputs for diagnostics
M      libf/phymars/convadj.F
^------> takes now into account the height of thermals to execute convective adjustment
      => note : convective adjustment needs to be activated when using thermals, in case of a 
         second instable layer above the thermals
================================================
A      libf/phymars/calltherm_interface.F90
^------> Interface between physiq.F and the thermals
A      libf/phymars/calltherm_mars.F90
^------> Routine running the sub-timestep of the thermals
A      libf/phymars/thermcell_main_mars.F90
^------> Main thermals routine specific to Martian physics
A      libf/phymars/thermcell_dqupdown.F90
^------> Thermals subroutine computing transport of quantities by updrafts and downdrafts
A      libf/phymars/thermcell.F90
^------> Module including parameters from the Earth to Mars importation. Will disappear in future dev
================================================
================================================

== 17/06/2011 == EM
>>> Updates and corrections (to enable compiling/running in debug mode with
    ifort)
- removed option "-free-form" from makegcm_ifort and set mod_loc_dir="."
  so that module files (produced in local directory by ifort)
  are moved to LIBO
- updated initracer.F, physdem1.F, physiq.F, inichim_newstart.F
  to avoid referencing out-of-bound array indexes (even if unused)
- cosmetic updates on inwrite.F, datareadnc.F
- updated newstart.F to initialize and use 'datadir' when looking for files
- corrected bug on interpolation of sub-surface temperatures in
  lect_start_archive.F

== 17/06/2011 == AC
>>> Important updates to thermals parameters
- Tuned aspect ratio of thermals to suit Buoyancy estimations from LES in CLOSURE relation
- Renormalization of alim_star after plume
- Removed alimentation mixing of estimated Teta in plume
>>> Minor change in makegcm_ifort

== 22/06/2011 == EM
- added modifications (from JYC & FGG) to tracer.h & initracer.F for ions
- minor improvement to newstart.F (q=x option, check that tracer index 
  provided by user is valid).
- minor correction to callradite.F (to enable compilation in debug mode
  with ifort when there is only one tracer).

== 17/06/2011 == AC
- Added maximum vertical velocity and heat flux output from thermals
- Added buoyancy diagnostics
- Minor modifications in thermals routines 

== 23/06/2011 == EM
- correct bug (introduced previously) in lect_start_archive.F on loop
  boundaries for soil temperature.

== 01/07/2011 == AC
- Added new settings for the Martian thermals from new LES observations
- Revamped thermcell's module variables to allow it's removal
- Minor changes in physiq and meso_physiq for the call to thermals
- Switched from dynamic to static memory allocation for all thermals variable 
to gain computation speed

== 04/07/2011 == AC
- Minor setting modification to thermals
- Added new flux optimization in thermcell_main_mars.F90, to run properly in 3D

== 14/07/2011 == JBM
- Tidying up dust properties in DATAFILE for better consistency (cf. suaer.F90)
- Cosmetic changes in aeropacity.F (changed comment and put a print inside a water flag)

== 15/07/2011 == EM
>> Implemented using 'z0' roughness length map (important: 'z0' reference
   field is in datafile surface.nc, which has also been updated).
- made z0 a z0(ngridmx) array and moved 'z0' from 'planete.h' to 'surfdat.h';
  added a 'z0_default' (common in surfdat.h) corresponding to the 'control'
  array value (contole(19) in startfi.nc).
- adapted 'tabfi.F' to use 'z0_default'.
- adapted 'phyetat0.F' to look for a 'z0' field in startfi.nc. If 'z0' is not
  found in the startfi.nc file, then the uniform default value (z0_default)
  is used.
- modified 'physdem1.F' to write 'z0' field to restart.nc
- adapted use of z0() in 'physiq.F' (diagnostic computation of surface stress),
  'vdifc.F' and 'vdif_cd.F'.
- adapted 'dustdevil.F' to use 'z0_default'.
- 'testphys1d.F' now uses 'z0_default', and the value to use can be set
  in run.def (with "z0=TheValueYouWant").
- modified 'datareadnc.F' to load reference map of 'z0' from surface.nc, and
  added a 'z0' option in 'newstart.F' to force a uniform value of z0. Note
  that the use of the z0 map is automatic when using newstart, but only when
  it loads a start_archive.nc file.

== 15/07/2011 == AS
- Modified the mesoscale part so that the previous change by EM does not
  imply an error in the mesoscale case. More development is needed though
  to get the "varying z0" capability in the mesoscale model.
- Worked on versions of meso_physiq and meso_inifis as close as possible
  to physiq and inifis for more continuity in the process of impacting changes 
  (and even possibly to reach a common version of physiq and inifis).
  >>> The main point is to make the mesoscale significant specific parts 
  coded into include files in meso_inc so that meso_physiq and meso_inifis
  looks very close to physiq and inifis.
  >>> This is completely transparent for GCM users who does not need the
  contents of meso_inc.
- Slight cosmetic changes to physiq.f and inifis.F
  --- some of them e.g. to prepare convergence between meso_physiq and physiq
- Dropped the meso_ prefix from the slope routines
  (currently those are only used in the mesoscale model,
   but one could imagine using those in hi-res GCM runs)
- Added a MESOSCALE flag to writediagfi so that mesoscale simulations
  are not bothered with calls put in routines by GCM people

== 19/07/2011 == AS
- Finished converging meso_physiq.F and meso_inifis.F towards physiq.F and inifis.F
  --> see previous point 15/07/2011
  --> meso_ routines no longer exist (everything is in meso_inc and transparent to GCM users)
  --> GCM routines include mesoscale parts within MESOSCALE precompiler commands
  --> MESOSCALE parts are as hidden as possible not to mess up with GCM users/developers
- Cleaned inelegant or useless #ifdef [or] #ifndef MESOSCALE in physiq and inifis so that
  a minimum amount of such precompiler commands is now reached [mainly related to I/O]
- Added the SF08 slope insolation model in the general physics parameterizations.
  Added a callslope keyword in inifis.F and callkeys.h
  --> This keyword is False by default / True if you use -DMESOSCALE
- Removed the obsolete call to Viking Lander 1 diagnostic
  Replaced it with a diagnostic for opacity at the domain center [valid for GCM and mesoscale]

== 01/08/2011 == EM
- Fixed bug in readtesassim (timeflag was not always set before being used)

== 03/08/2011 == AC
- Modified physiq.F to interface new SL parametrization
- New SL parametrization based on a bulk Richardson Monin-Obukhov theory formulation
   Stability functions are taken from D.E. England et al. (95)
   Similarity functions coefficients based on Dyer and Hicks (70).
   Includes thermal roughness length computation, heat and momentum drag coefficient computation
   Can be used to output hydrodynamic-related SL quantities (bulk Richardson, turbulent Prandtl number estimation, Reynolds number)
- Minor modification in thermcell_dqupdown.F to suit picky compilers
- vdifc.F Now takes into account sub-grid gustiness, evaluated from thermals activity (it's proxy being the maximum vertical velocity)
- Minor modification in meso_inc_les.F: u* is now taken from the new vdifc and not recomputed from a simple law
== 08/08/2011 == AC
- Formula correction for the Reynolds in vdif_cd : will affect the thermal roughness length. Results obtained with the precedent formula (previous commit) probably underestimate by about 1K the temperature in the mixing layer.
- Added a new subroutine to libf/phymars called surflayer_interpol.F. This routine can be called as a diagnostic for interpolation of horizontal velocity norm and temperature in the surface layer, in 1D or 3D. The output height is controlled by a parameter z_out, to be set in physiq.F. This might be later added to a .def file. z_out is set to 0. by default, resulting in no call to this subroutine.

== 16/08/2011 == AC
- Minor modification in physiq.F to pass Ch from vdifc to meso_inc_les
- Major modification to the formulation of integrals in surflayer_interpol.F
  Now stable for most cases. Some cases with highly negative Monin Obukhov length
  remain to be explored.
- Added gustiness to the Richardson computation in vdif_cd.F. Gustiness factor is for now of beta=1., after
  several comparisons with LES aerodynamic conductances. May be subject to a minor change (+/- 0.1)
  in the near future. (almost transparent for the user)
- Minor modifications relative to variables in vdifc.F.
- Esthetic change to calltherm_mars.F
- Changed the definition for HFX computation in the LES in meso_inc/meso_inc_les.F. New definition yields
  very similar results to old one and follows a strict definition of what HFX should be.

== 24/08/11 == TN
Attempts to tune the water cycle by adding outliers 
       + A few structural changes !!
* watercap.h is now obsolete and removed -- all is in surfdat.h
* watercaptag initialized in surfini.F (up to 7 areas defined) instead of initracer.F
  - settings proposed by AS commented
  - experiments by TN decommented. use with caution.
* water ice albedo and thermal inertia in callphys.def and inifis.F
* water ice albedo in surfini.F
* water ice albedo computation in albedocaps.F90
* alb_surfice is now obsolete in physiq.F, albedo_h2o_ice is used instead
* frost_albedo_threshold defined in surfdat.h
* water ice thermal inertia in soil.F
TODO: * calibrate thermal inertia and ice albedo
      * have a look at subgrid-scale ice with dryness ?

== 07/09/2011 == AC

- Added new flag for the Richardson-based surface layer :

callrichsl, can be changed in callphys.def

One should always use the thermals model when using this surface layer model.
Somes cases (weakly unstable with low winds), when not using thermals, won't be well represented by the
Richardson surface layer. This stands for Mesoscale and Gcm but not for LES model.

Correct configs :

callrichsl = .true.
calltherm = .true.

callrichsl = .false.
calltherm = .false.

callrichsl = .false.
calltherm = .true.

Previously unstable config :

callrichsl = .true.
calltherm = .false.

- To be able to run without thermals and with the new surface layer, a modification has been made to
physiq.F to account for gustiness in GCM and MESOSCALE for negative Richardson, so that :

callrichsl = .true.
calltherm = .false.

can now be used without problems, but is not recommended.

- Consequently, callrichsl = .false. is now the default configuration for thermals.

We recall the available options in callphys.def for thermals :

outptherm = BOOLEAN (.false. by default) : outputs thermals related quantities (lots of diagfi)
nsplit_thermals = INTEGER (50 by default in gcm, 2 in mesoscale) : subtimestep for thermals model.
         It is recommended to use at least 40 in the gcm, and at least 2 in the mesoscale.
         The user can lower these values but should check it's log for anomalies or errors regarding
         tracer transport in the thermals, or "granulosity" in the outputs for wmax, lmax and hfmax.

== 08/09/11 == AS
LMDZ.MARS + MESOSCALE. 
---> Setting up a more realistic water ice source at the poles (notably outliers)

[[ surfini.F ]]
Main changes and bug fixes
* reference to comcstfi.h was wrong. big problem because e.g. pi was not known.
* commented about a problem to be fixed, due to surfini being called before initracer.
* MESOSCALE: put the mesoscale north cap definition into a precompiling flag #MESOSCALE
             for the moment: if [alb_mean_TES > 0.26 and lat > 70] then outliers
             (previously done in meso_inc_caps.F)
   
[[ inifis.F ]]
Just changed a comment with wrong formatting

--> below, only MESOSCALE

[[ soil.F ]]
if somewhere IT > IT_outliers, then makes it = IT_outliers

[[ physiq.F ]]
[[ meso_inc/meso_inc_caps.F ]]
[[ meso_inc/meso_inc_ini.F ]]
meso_inc_caps no longer called. keep for reference for the moment.

[[ meso_inc/meso_inc_var.F ]]
deleted lines with *_lim variables, now useless

== 08/09/11 == EM
-Fixed problem about undefined tracer names in 'surfini.F' by calling
 'initracer' before 'surfini' in physiq.F

== 09/09/11 == AS
--> Fixed a problem with .eq. used with booleans in physiq.F
[some good picky compilers complain about this]
--> Added a warning in inifis about using
callrichsl = .false.
calltherm = .true.
which is not recommended. The new surface layer has been built to go
with the new mixing layer scheme (thermals). And anyway it is a much
better scheme than the previous one.

== 09/09/11 == AC

 This is a major update for thermals and richardson layer parametrization. This update stabilizes thermals (further
 studies might show that we can reduce the value of nsplit in gcm. To be tested.), and prevent downdrafts from descending into
 the surface layer, which was acting as a cold feedback on the thermals. The Richardson surface layer now features
 different gustiness coefficients for Richardson, heat and momentum so that u* and t* are correctly represented.
 
 Upcoming updates will change surflayer_interpol.F90 to implement those changes in the interpolation scheme as well.
 
 ***************************
 IMPORTANT : several parts of the vdifc code might want to use these new definitions for gustiness, u* and t*. exemple : dust devil routines
 that recompute u* ? lifting routines ? TODO !
 **************************
 
 
 M             289   libf/phymars/thermcell_main_mars.F90
 ^-----------------> Added iterations to the velocity / buoyancy / entrainment / detrainment
                     computation to ensure correct convergence. Iteration number is for now set to
                     4, which is probably too much. This will be changed once several cases are tested.
                     The minimum is probably 2 iterations.
 
 M             289   libf/phymars/vdifc.F
 ^-----------------> Subgrid gustiness parametrization now uses different gustiness beta coefficients
                     for heat and momentum. Comparisons with LES at Exomars landing site, matching u*
                     and teta* suggests values of beta=0.7 for momentum and beta=1.2 for heat, where
                     the formula for large scale horizontal winds in the first layer is :
                          U0^2 = pu^2 + pv^2 + (beta*wmax_th)^2
 
 M             289   libf/phymars/vdif_cd.F
 ^-----------------> LES data suggests that Richardson number distribution during daytime has a very large
                     standart deviation due to highly negative Richardsons on several gridpoints. As a consequence,
                     the mean Richardson in daytime in the LES is much more negative than in LES. In the gcm,
                     parametrized subgrid turbulence prevents such cases, which can be dangerous in nearly unstable conditions.
                     Hence, we use beta=0.5 instead of one, so that we keep the anti-crazy-hfx function of beta and we increase the
                     norm of negative Richardsons in general for daytime conditions. This is set in conjonction with
                     beta settings for heat and momentum in vdifc.
 
 M             289   libf/phymars/meso_inc/meso_inc_les.F
 ^-----------------> HFX and USTM computations now uses the different betas for heat and momentum.

== 21/09/11 == AS
--> Added MESOINI precompiler flag so that all fields needed to initialize the mesoscale model
     are being output in the diagfi.nc file. Previously it was made through a separate physiq.F
     which needed to be updated every time...
--> This is completely transparent to the casual GCM users and only appears in the WRITEDIAGFI section of physiq.F

== 21/09/11 == AC

Revision on several settings for the thermals model. This version relies on fits done for an article to be published, and is more precise.

 M             299   libf/phymars/thermcell_main_mars.F90
 ^-----------------> Changed downdraft to updraft mass flux ratio from -1.8 to -1.9
                     Changed first level for downdraft mixing from k=3 to k=2. Only level 1 is non-mixed now.
                     Changed coefficients for downdraft to updraft thermal buoyancy ratios.

 M             299   libf/phymars/calltherm_mars.F90
 ^-----------------> Changed r_aspect_thermals from 2. to 1.5 for the GCM version to better match buoyancy profiles.

== 29/09/11 == AS
--> To easily explore sensitivity to lifting thresholds: in dustlift.F, ustar_seuil=sqrt(stress/rho) 
    and alpha_lift[dust_mass] can be prescribed through an external stress.def parameter file.
    --- alpha_lift[dust_number] is computed from alpha_lift[dust_mass] as in initracer.F
    --- ustar_seuil is more user-friendly than stress because direct comparison with ustar from model 
--> For the moment this is MESOSCALE only, but potentially useful to everyone

== 30/09/11 == TN
>> Bug correction in lect_start_archive.F ; in some cases layer(:) was not
   initialized.


== 10/10/2011 == AC

***********
This commit aims at increasing the thermals speed. Using these corrections, gcm performances in 64x48x32 using 1 tracer goes from 27.9% elapsed time in thermals to 18.76%.

***********
Additional work needs to be done in tracer advection to gain speed in high tracer number configuration. (tracer advection (but not momentum nor temperature) could be decoupled from sub-timestep, as they do not act on the thermals scheme (water vapor is neglected as we use theta and not theta_v, and radiative effect of dust is not computed in the thermals.))

***********
=> TOP 5 of routine contributions to gcm runtime :

Each sample counts as 0.01 seconds.
  %   cumulative   self              self     total
 time   seconds   seconds    calls   s/call   s/call  name
 18.76      6.33     6.33      960     0.01     0.01  thermcell_main_mars_
 17.19     12.13     5.80                             __svml_powf4.A
 13.72     16.76     4.63    10369     0.00     0.00  filtreg_
  3.94     18.09     1.33                             __intel_new_memset
  3.73     19.35     1.26     2880     0.00     0.00  thermcell_dqupdown_

note: thermcell_main_mars_ does call quite a lot power computations (see __svml_powf4.A), but this number will not increase with tracer numbers.

***********
=> LOG:

M             312   libf/phymars/thermcell_main_mars.F90
^------------------- removed (commented) computations on buoyancy which is purely diagnostic
                     tuned internal convergence loop and added convergence criterion

M             312   libf/phymars/thermcell_dqupdown.F90
^------------------- removed (commented) downdraft-related if-loops (as we do not advect tracers and momentum in downdrafts for now)

M             312   libf/phymars/calltherm_mars.F90
^------------------- removed (commented) diagnostic-related computations
                     changed default thermals spliting and aspect ratio
                     corrected a bug where maximum height was not correctly computed and could result in convective adjustment used in place of thermals
                     when using certains sets of nsplit and r_aspect (was not happening with the baseline version, so that this correction is transparent to
                     users)
********************

== 14/10/2011 == AS + AC
- Monin-Obukhov length is now output from surflay_interpol and written in diagfi if z_out not 0.
- in calltherm_interface, defaut settings for qtransport_thermals and dtke_thermals

== 20/10/2011 == EM
- added FF's upgrade of writediagfi. Now, if at runtime there is a diagfi.def
  file, it should contain the list of variables (1 per line) than will be put
  in the diagfi.nc file. If there is no diagfi.def file, then all variables
  are put in the diagfi.nc file (as was the case before).
 
== 21/10/2011 == EM
- Corrected small bug in newstart: initracer was not always used and thus
  some tracer indexes (igm_co2, igcm_h2o_vap,...) were not set. This
  however means that we now also call inifis from newstart and that we read
  in flags set in 'callphys.def' (required for sanity checks in initracer).
  Also adapted 'inichim_readcallphys': removed some obsolescent tests on
  number of tracers for given combinations of options.

== 21/10/2011 == AS
- Added possibility for CH4 tracer in tracer.h and initracer.F

== 27/10/11 == AS
--> Added geticecover.F90 which computes seasonal ice cover given ls, lati(ngrid), long(ngrid)
    as proposed by T. Titus from TES observations [fitting functions for crocus line] 
    ... output is icecover(ngrid) which value is 0 [no ice cover] or 1 [ice cover]
    ... no calculations are done for latitudes between -40 and +40 [ice cover is directly set to 0]
--> In physiq.F, co2ice is set to a dummy high value to simulate a CO2 cap 
    wherever icecover(ngrid) is 1. This is done at each timestep before newcondens is called.
--> For the moment this is MESOSCALE only, but potentially useful to everyone.

== 28/10/11 == FL + AS
ADDED THE NEW FRAMEWORK FOR PHOTOCHEMISTRY
This is not the last version. A new rewritten version of calchim.F [using LAPACK] is yet to be committed.
--> A new version for photochemistry routines : *_B.F no longer exists, new routines in aeronomars
D              333   libf/aeronomars/photochemist_B.F
D              333   libf/aeronomars/init_chimie_B.F
A                0   libf/aeronomars/read_phototable.F
M              333   libf/aeronomars/calchim.F
A                0   libf/aeronomars/photochemistry.F
M              333   libf/aeronomars/chimiedata.h
--> Changed calls to calchim and watercloud [surfdust and surfice needed] in physiq.F
--> Left commented code for outputs in physiq.F [search for FL]
--> Added settings which works for 35 levels in inidissip.F according to FL. Commented for the moment.
--> Checked compilation and run, looks fine. Note that 'jmars.20101220' is needed.

== 02/11/2011 == AC

***********
This commit aims at increasing the thermals speed, especially for large tracer number configurations. The idea behind this commit is to advect non-active conserved variables outside of the sub-timestep of the thermals. Because these variables are not used in thermals computation, we can decouple them:

momentum: can be decoupled because we assume a constant ratio between horizontal velocity in alimentation layer and maximum vertical velocity in the thermal
s.

tracer: can be decoupled because we do not take condensation of any tracer into account and hence do not liberate latent heat nor form clouds in the thermal
s.

temperature: cannot be decoupled (of course)
***********

D             336   libf/phymars/thermcell_dqupdown.F90
^----------------   Deleted and replaced by a simpler version. Notes about downdraft advection are still available from revision 336 of SVN in thermcell_dqu
pdown.

A               0   libf/phymars/thermcell_dqup.F90
^----------------   New upward advection for tracers and momentum in thermals. Several changes are done compared to the old approach:
                          - Updraft quantities are not longer computed by making hypothesis on the amount of advected air.
                          - In general, the formalism for updraft computation is much simpler and clearer.
                          - Tracer tendancies are no longer computed using the conservation equation. Instead, we use the divergence
                          of an approximated turbulent flux of the concerned quantity, where downdraft are also neglected.

M             336   libf/phymars/thermcell_main_mars.F90
^----------------   The Main does not call anymore thermcell_dqupdown, which it was doing 2+tracer_number times per subtimestep (140 times per physical step for a 2 tracer config)

M             336   libf/phymars/calltherm_mars.F90
^----------------   Entrainment, detrainment and mass-fluxes are recomputed in  the sub-timestep loop. Their final value after iterations is used by the new
 advection routine to compute tracer and momentum fluxes.

*********** Results

- Conservation of tracers has been assessed over 1 yr in 1D and found to be comparable to that obtained with the simple convective adjustment. (it actually
seems to be better by a factor of 10%!)
- GCM speed-up is of about 20% compared to the old thermal configuration, for a 2 tracer case.
- Advection of sharp tracer profiles has been successfully observed, similar to the old method.