Changes between Version 1 and Version 2 of Tags/196/ExternalisationParameters/TechnicalModifications

Timestamp:

2011-11-29T09:29:23+01:00 (12 years ago)

Author:

dsolyga

Comment:

--

Tags/196/ExternalisationParameters/TechnicalModifications

@@ -40 +40 @@
 
 We add a new module called '''qsat_moisture.f90'''. This module regroups the routines for the calculation of qsat (previously in '''constantes_veg.f90'''). [[BR]]
-'''intersurf.f90''' reads the number of PFTs and the parameters values chosen by the user. 
+'''intersurf.f90''' reads the number of PFTs and the parameters values chosen by the user. [[BR]]
+We put the routines get_vegcorr and get_soilcorr in slowproc.f90. We correct the following bug in the Olson classification (subroutine get_vegcorr) :
+
+{{{
+    ! 78 warm C3 woody savanna
+    vegcorr(78,:) = &
+         & (/0.0, 0.0, 0.4, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.6, 0.0, 0.0, 0.0/)
+    ! 79 warm C4 woody savanna
+    vegcorr(79,:) = &
+         & (/0.0, 0.0, 0.4, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.6, 0.0, 0.0, 0.0/)
+
+}}}
+ 
+The good formulation is :
+
+{{{
+    ! 78 warm C3 woody savanna
+    vegcorr(78,:) = &
+         & (/0.0, 0.0, 0.4, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.6, 0.0, 0.0, 0.0/)
+    ! 79 warm C4 woody savanna
+    vegcorr(79,:) = &
+         & (/0.0, 0.0, 0.4, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.6, 0.0, 0.0/)
+}}}
+
 
 
@@ -46 +69 @@
 
 We delete '''stomate_constants.f90''' in order to move its parameters in ''src_parameters''. The too-specific parameters were moved to '''stomate_data.f90'''. [[BR]]
-Due to Fortran language, we have to delete the following structures in ORCHIDEE (previously in '''stomate_constants.f90''':
+Due to Fortran language restrictions (we can't allocate a field of a structure), we have to delete the following structures in ORCHIDEE (previously in '''stomate_constants.f90''':
 
 {{{
@@ -162 +185 @@
 
 The most part of the externalized parameters which were still in src_sechiba are in src_parameters now.
+
+
+=== SUMMARY OF THE CHANGES ===
+
+I summarize the modifications and the features of the externalized version of ORCHIDEE :
+ * 3 bugs corrected : black_carbon(lpj_fire), herbivores(season) and routine get_vegcorr(slowproc)
+
+ * 3 new modules : pft_parameters, constantes_mtc and qsat_moisture
+
+ * 4 modules deleted :  constantes_soil, constantes_co2, constantes_veg and stomate_constants
+
+ * ext_coeff replace in all the code ext_coef and ext_coeff
+
+ * type phenology deleted : because we can't use the attribute ALLOCATABLE for a field in a structure
+
+ * Rewrite the explicit referenced to PFTs :(IF j==12 .OR. j==13) replaced by IF(.NOT. natural(j)) => adapte code for a generic number of PFTs
+
+ * Add the function getin_p for characters array
+
+ * Delete multi definitions for Earth radius, pi, min_sechiba, min_stomate
+
+ * Unify parameters definitions for weather and ORCHIDEE : the values used by weather are the same used by ORCHIDEE.
+
+ * Create Labels for all new externalized parameters and add Config Units for the previous parameters
+ 
+ * Configuration of the number of PFTs : from 2 to X
+ 
+ * Configuration of the values of some 400 parameters without having to recompile the code.
+
+ * SCRIPTS : design new default files
+
+Delete unused parameters : 
+
+
+
+=== Choices made ===
+
+
+
+I try to regroup all the externalized parameters in src_parameters.[[BR]]
+
+For the pft_parameters : they have to be allocated and some are used by many modules. Moreover, I want to have
+a flag called IMPOSE_PARAM which controls the reading of the parameters. [[BR]]
+
+If Someone wants to go back to the standard values, this flag has to be set to TRUE. [[BR]]
+I think also the module pft_parameters as the symetric module of constantes_mtc. 
+With constantes_mtc.f90, we have only one file to modify to add a new metaclass and we can track more easily the changes made to the values.
+
+
+
+
+For the other parameters, I want to avoid multiple definitions as some parameters are used by different modules.[[BR]]
+
+I regroup also these parameters because of the flag IMPOSE_PARAM. [[BR]]
+
+
+The other advantages is that we can distinguish variables from parameters in the code and that we have a
+
+better readibility of the code (personal point of view). But this is incomfort for some people, we can dispatch
+
+again the parameters (except the pft-parameters ones).[[BR]]
+
+
+
+=== NEED TO BE DISCUSSED BEFORE THE TAG ===