Changeset 3876

Timestamp:

Aug 1, 2025, 9:37:44 PM (4 weeks ago)

Author:

emillour

Message:

Common dynamics:
Follow-up of r3874: Stopping the GCM if z2sig.def contains more lines
than expected for given llm is too strict. In some cases (Mars, Venus)
the same z2sig.def is used and "extra" layers for the thermosphere should
just be ignored when running a simulation limited to the troposphere.
While at it converted comments in disvert_noterre.F from French to
English.
EM

File:

• trunk/LMDZ.COMMON/libf/dyn3d_common/disvert_noterre.F (modified) (14 diffs)

trunk/LMDZ.COMMON/libf/dyn3d_common/disvert_noterre.F

@@ -3 +3 @@
 
 c    Auteur :  F. Forget Y. Wanherdrick, P. Levan
-c    Nouvelle version 100% Mars !!
-c    On l'utilise aussi pour Venus et Titan, legerment modifiee.
 
 #ifdef CPP_IOIPSL
@@ -24 +22 @@
 c
 c=======================================================================
-c    Discretisation verticale en coordonn�e hybride (ou sigma)
-c
+c    Vertical discretization for purely sigma or hybrid sigma-pressure 
+c    coordinate depending on user provided flag "hybrid"
 c=======================================================================
 c
@@ -62 +60 @@
       write(lunout,*) trim(modname),': hybrid=',hybrid
 
-! Ouverture possible de fichiers typiquement E.T.
+! Open file esasig.def, or if not there open z2sig.def
 
          open(99,file="esasig.def",status='old',form='formatted',
@@ -73 +71 @@
 
 c-----------------------------------------------------------------------
-c   cas 1 on lit les options dans esasig.def:
+c   case 1 read from esasig.def:
 c   ----------------------------------------
 
@@ -128 +126 @@
 
 c-----------------------------------------------------------------------
-c   cas 2 on lit les options dans z2sig.def:
+c   case 2 read from z2sig.def:
 c   ----------------------------------------
 
@@ -147 +145 @@
          read(99,*,iostat=ierr1)
          if (ierr1 .eq. 0) then
-            write(lunout,*) 'ERROR: Expected nb of levels:', llm
-            call abort_gcm(modname, 
-     &        'z2sig.def has more lines than expected', 1)
+            write(lunout,*) 'WARNING: More lines in z2sig.def'
+     &       //' than expected from nb of levels:', llm
+!            call abort_gcm(modname, 
+!     &        'z2sig.def has more lines than expected', 1)
          endif
          CLOSE(99)
@@ -178 +177 @@
  
 c-----------------------------------------------------------------------
-c    ....  Calculs  de ap(l) et de bp(l)  ....
+c    ....  Compute  ap(l) and bp(l)  ....
 c    .........................................
 c
-c   .....  pa et preff sont lus  sur les fichiers start par dynetat0 .....
+c   .....  note that pa and preff are from start file read by dynetat0 
 c-----------------------------------------------------------------------
 c
@@ -217 +216 @@
       write(lunout,*)  ap
 
-c     Calcul au milieu des couches :
-c     WARNING : le choix de placer le milieu des couches au niveau de
-c     pression interm�diaire est arbitraire et pourrait etre modifi�.
-c     Le calcul du niveau pour la derniere couche 
-c     (on met la meme distance (en log pression)  entre P(llm)
-c     et P(llm -1) qu'entre P(llm-1) et P(llm-2) ) est
-c     Specifique.  Ce choix est sp�cifi� ici ET dans exner_milieu.F
+c     Compute value at the mid-layer :
+c     WARNING : there is an arbitrary decision here to put the
+c     middle of the layer half-way (pressure-wise) between boundaries
+c     In addition for the last layer (the top of which is at zero pressure)
+c     P(llm), we choose to put it with the same interval (in log(pressure)) 
+c     from P(llm-1), than there is between between P(llm-1) and P(llm-2)
+c     This choice is quite specific and is done here AND in exner_milieu.F
+c     Both routines should definitely use the same assumptions.
 
       DO l = 1, llm-1
@@ -282 +282 @@
 
 
-      RETURN
       END
 
@@ -292 +291 @@
 c     esasig.def/z2sig.def lors du passage en coordonnees hybrides
 c     F. Forget 2002
-c     Connaissant sig (niveaux "sigma" ou on veut mettre les couches)
-c     L'objectif est de calculer newsig telle que
+c     Knowing sig (the "sigma" levels where we want to put the layers)
+c     The goal is to compute newsig such that 
 c       (1 -pa/preff)*exp(1-1./newsig**2)+(pa/preff)*newsig = sig
-c     Cela ne se r�soud pas analytiquement: 
-c     => on r�soud par iterration bourrine 
+c     Which cannot be done analyticaly: 
+c     => need for an iterative procedure 
 c     ----------------------------------------------
-c     Information  : where exp(1-1./x**2) become << x
+c     Information  : when exp(1-1./x**2) becomes << x
 c           x      exp(1-1./x**2) /x
 c           1           1
@@ -308 +307 @@
 c           0.269      1.E-5
 c           0.248      1.E-6
-c        => on peut utiliser newsig = sig*preff/pa si sig*preff/pa < 0.25
+c        => one can thus use newsig = sig*preff/pa if sig*preff/pa < 0.25
 
 
@@ -321 +320 @@
             newsig= sig
       else if (sig*preff/pa.ge.0.25) then
-        DO J=1,9999  ! nombre d''iteration max
+        DO J=1,9999  ! overkill maximum number of iterations
           F=((1 -pa/preff)*exp(1-1./newsig**2)+(pa/preff)*newsig)/sig
 c         write(0,*) J, ' newsig =', newsig, ' F= ', F
@@ -331 +330 @@
               newsig=(X2+newsig)*0.5
           end if
-c         Test : on arete lorsque on approxime sig � moins de 0.01 m pr�s 
-c         (en pseudo altitude) :
+c         Convergence is assumed if sig is within 0.01m (in pseudo-altitude)
+c         of the target value.
           IF(abs(10.*log(F)).LT.1.E-5) goto 999
         END DO
@@ -339 +338 @@
        end if
  999   continue
-       Return
+
       END