Hallo Narasinha.

We in South Africa have the same contriants in the use DSSAT as
you experience.

To run the DSSAT models within a data scare environment is one
off the challenges facing researchers in the developing world. It is
one of the main reasons why modelling projects are abandon.

As part of the Highveld Ecoregion project in South Africa (Funded
by the Ecoregion Fund- ISNAR) we are busy  developing
methodolgy to solve the problem in this regard.

Regarding  genetic coefficients, their are no arguing that high quality
data is needed to determine these coefficients. But as you stated
these data are not availble and it is to expensive to do spesial
genetic coefficient trial (although preferable).  We are working on
software that will determine genetic coeffisient(GC) from at least
yield, in a worst cause senario. Thus rather have something than
nothing approach and one must always treat this GC with caution.

The norm is that you compare yields to validate the model. But it
have little value with lack of enough input and observed data.  If you
run the long term trials with the input data availlable and you then
compare the rankings of managment practises with the rankings
from field trials, from this you could determine if you would have
made the same managment dissusion using the simulation results
as would be the case from the field trials.

We have developed a Model evalution statistical package for
DSSAT models and will send you copy if you whant it.

Regards
Andre


**************************************
* ANDRE DU TOIT                      *
* AGRICULTURAL RESEARCH COUNCIL      *
* SUMMER GRAIN CENTRE                *
* PRIVATE BAG X1251                  *
* POTCHEFSTROOM 2520                 *
* TEL: (018) 2996249                *
* FAX: (018) 2947146                *
* INTERNET: [log in to unmask]      *
**************************************