[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: co2calc.f and ta_iter_1.f

OCMIP modelers:

Thanks Rick for identifying this problem and warning others of it.
Thanks also Keith for the fixed routine "co2calc.f".  I've now changed
some of Keith's comment lines and have added a few. It is that new
fixed version of "co2calc.f" (version 1.8) which is now available via
the Abiotic HOWTO on the OCMIP Web page.  Versions of "ta_iter_1.f" and
"drtsafe.f" remain unchanged. All OCMIP modelers should
recuperate the version 1.8 of "co2calc.f".

REMINDER: these routines (co2calc.f, ta_iter_1.f, and drtsafe.f) are NOT
optimized for model runs.  Our experience at IPSL is that when used
without any modification, they are very expensive indeed.  Modelers can of
course decrease the range between "phlo" and "phhi", as suggested by Chris
Sabine and Bob Key (see README.Cchem).  However, even after doing that
there are still substantial gains possible.  For example, Olivier Aumont
tells me that Ernst Maier-Reimer has cut the calculation time for the
carbon chemistry, by using his standard iterative technique with the same
constants. After a short initial period, Ernst only needs about 1
iteration. Compare this with the 4 iterations mentioned in "README.Cchem". 
There is also the issue of vectorization.

Finally, Rick, the links you mention on the Abiotic HOWTO will be
operational shortly.

Best regards,

   James Orr                       Laboratoire des Sciences du Climat 
   LSCE, CEA Saclay                             et de l'Environnement    
   Bat. 709 - Orme                 Unite Mixte de Recherche CEA-CNRS    
   F-91191 Gif-sur-Yvette Cedex    Phone:   (33) (0)1 69 08 77 23      
   FRANCE                          Fax:     (33) (0)1 69 08 77 16
                                             |    |     
   mailto:  orr@cea.fr            (country code)  (when in France)