**************************************************************************************************
V200511par-rc5 wrt. V200511par-rc4
**************************************************************************************************

1 - Merging POPs version
========================

A development version taking in account Persistent Organic Pollutants has been merged.
V200511par-rc5 is now the basis for developing the POP version.

2 - cosmetic
============

Some modifications in Makefile headers and in netCDF documentation


**************************************************************************************************
V200511par-rc4 wrt. V200511par-rc3
**************************************************************************************************

1 - chm2grads.f90 failed when compiled with pgf90 and g95
========================================================

Corrected chm2grads.f90 and Makefiles.

2 - Reading WRF files with exachim.sh
=====================================

Now exachim.sh can read WRF output files.

3 - Palettes
============

Added PM10 palette to exachim

4 - Added GEMS domain
=====================



**************************************************************************************************
V200511par-rc3 wrt. V200511par-rc2
**************************************************************************************************

CHIMERE is now a SPMD application, which no longer requires an MPI-2 compatible parallel
environment. It only uses MPI-1 function calls.

Corrected two bugs in exachim, which precluded the use of the difference display exadiff.sh.



**************************************************************************************************
V200511par-rc2 wrt. V200511par-rc1
**************************************************************************************************

1 - Scripts
===========

Perform the test of existence of coarse domain coordinates file only if nested run.

2 - Toolbox
===========

New processing in exachim : we no longer use a COARDS intermediate file. Instead, we build
directly GrADS compliant files.
A new utility, chm2grads.f90, replaces chm2coards.f90. All kind of CHIMERE grids may be used,
not only rectilinear ones.
A more sophisticated command line options set allows for displaying only selected variables and
time slots.
The udunits library is no longer used.

Added a new tool to transform old CHIMERE binary files into netCDF files

3 - Makefiles
=============

Corrected a bug of endianness flag in Makefile.hdr.g95.
Deleted references to udunits, which is no longer used

**************************************************************************************************
V200511par-rc1 wrt. V200510B
**************************************************************************************************

No changes in the scientific core of the model.

CHIMERE is now parallelized using the distributed memory scheme, and the LAM-MPI library.

There is one master program, which spawns several worker programs, each worker been
in charge of the processing of a subdomain.
Subdomains are the result of a cartesian division of the main geographical domain by the
number of processors.

The master program is in charge of the initializations and file I/O.
The workers are in charge of the scientific processing and inter-subdomains
boundary conditions updates.

Despite the tremendous architectural change in the code structure, most scientific subroutines,
now called by the worker programs, have remained unchanged.

As far as the user interface is concerned, the only apparent modifications are :
- definition of subdomains in chimere-mm5 script : nzdoms and nmdoms parameters.
- use of the LAM-MPI library : LAM-MPI must be installed for running CHIMERE,
  even if you run on only one processor with only one subdomain.



**************************************************************************************************
V200510B wrt. V200501H
**************************************************************************************************

There are no changes in the scientific core of the model.

But this version represents a major change in the format of internal and output files, since 
most of the files are now written in the netCDF format.
The rationale was :
- to allow for a better portability among different computer architectures,
- to allow for the use of widespread post-processing and analysis tools,
- to increase I/O performance through direct record addressing.

Another change is the generalization of Fortran 95 as a programming language.

However, the format of the external interface input files has not changed.
There are no modification in the database format and structure.


We will not walk in detail through all the changes, since most subroutines where file I/O are
involved have been completely rewritten.

To summarize, from the user point of view, the main changes are :

- modification in the chimere scripts. The previous monolithic script has been splitted into:
  * a main script "chimere-mm5" which launches the jobs
  * one or more configuration scripts "chimere-*-.conf" which deal with model parameters and
    file locations
  * an execution script "chimere-mm5.exec which does all the low-level job setup and execution.

- more error management included in the scripts.

- error management in the netCDF I/O, which reports the line number in the source code where
  the error occured.

- use of "Makefiles" to drive the program compilation.

- rewriting of the meteo interface script.

- use of configuration files or namelists (*.nml) passed as arguments to scripts or executables. 
  The previous way was to pass arguments through Unix pipes. 
  The new way makes debugging easier since configuration files are static and can be edited
  to run separately some parts of the model.

- Model internal emission and boundary concentrations files are now in the netCDF format, 
  but the files in the  data base remain in their original binary format. 
  The preparation routines have been modified accordingly. 
  For those who use only the emission and boundary concentrations data provided
  with CHIMERE distribution, there is nothing to change. Those who use their own database may
  have to check the preparation programs : prepemis_std.f90, preplat_lmdzinca2.f90, 
  preptop_gocart.f90, preplat_mozart2.f90, preptop_lmdzinca2.f90, preplat_gocart.f90, 
  preptop_mozart2.f90 for compatibility with their data.

- Model internal meteo files are now in the netCDF format. But the external format is still
  the well known MM5 binary format. Users who connect CHIMERE to MM5 will not have to modify
  the meteo interface. In other cases, users will have to rewrite their own meteo interface, taking
  prepexmm5.f90 as a template.

- In the MM5 meteo interface, prepmm5.f and the interpolation routine exdom-mm5.f have been 
  merged into a single program prepexmm5.f90.

- The meteo diagnostic program has been split into two programs :
  * diagmet for pure meteo diagnostics
  * diagbio for bioemissions calculations

- diagmet has been deeply rewritten, but care has been taken to separate the netCDF 
  overhead mainly contained in diagmet.f90, from the scientific core of the parametrizations
  contained in diagmet_science.f90.

- The processing of meteo information is now :
  * input : MM5 (raw binary)
  * processed by prepexmm5.e to produce : 
  * intermediate : exdomout.nc (netCDF)
  * processed by diagmet.e to produce :
  * final : METEO.nc (netCDF)
  * used by chimere.e

- The output concentration files created by chimere.e are now in the netCDF format. Users will
  have to modify their post-processing programs. A template program to read CHIMERE netCDF files
  is provided in toolbox/readcdf directory.

- the visualization tool "mapit" has been adapted to netCDF and renamed "ncmapit".

- another visualization tool, "exachim", has been added.

- Most fundamental physical constants are now defined in a single file. This may lead to
  subtle numerical differences in the results, compared to V200501H.

