Dear Emmanuel,
what you say is all nice and probably 80% true in theory, but unfortunately
real life tells us something different:
Read for instance the comment of Walter Bowen in the header of the
MZCER980.CUL file of DSSAT:
MAIZE GENOTYPE COEFFICIENTS - GECER980 MODEL
!
!The P1 values for the varieties used in experiments IBWA8301 and !UFGA8201
were recalibrated to obtain a better fit for version 3 !of the model. After
converting from 2.1 to 3.0 the varieties !IB0035, IB0060, and IB0063 showed
an earlier simulated flowering !date. To correct this, the P1 values were
recalibrated. !The reason for this is that there was an error in PHASEI in
!version 2.1 that had TLNO=IFIX(CUMDTT/21.+6.) rather than
!TLNO=IFIX(SUMDTT/21.+6.); see p. 74 of Jones & Kiniry. !-Walter Bowen, 22
DEC 1994.
Thus, adjustments to the genotype coefficients are made to obtain better
model fit!
(Btw, for those who don't know: P1 is the thermal time from emergence to the
end of juvenile phase.)
As I have said before, APSIM and DSSAT have some overlap regarding the
notation and use of genetic coefficients. BUT, for instance, APSIM uses a
slightly different approach to calculate thermal degree days/thermal time.
See:
http://www.apsim.info/apsim/Documentation/Modules/maize_science.pdf
page 4.
As a result, thermal times calculated by APSIM and DSSAT (which uses the
"classical" approach) only crudely match. So if you use the same genotype
coefficients for both models, e.g. the P1 value, results will be different.
However, coming back to the original question, I fully agree that once
genetic coefficients are optimized, they should be kept the same throughout
the modeling exercises. Everything else I would simply call "cheating"!
(Regarding the latter, if you ask me, publications about model applications
should always contain the full set of variables used and adjustments made.
The author's should probably even ask to provide the respective files for
people interested in "re-running" the experiment.)
Best regards,
Rolf
> -----Original Message-----
> From: Emmanuel Xevi [mailto:[log in to unmask]]
> Sent: Tuesday, February 03, 2004 16:28
> To: [log in to unmask]
> Subject: Genetic coefficients
>
> Dear all
>
> I found it prudent to enter this debate about genetic coefficient. I
> find it very alarming that some people think genetic coefficients are
> dependent on the environment. More disturbing is the notion that they
> are even model dependent. I agree with Ajeet Singh Nain that this
> coefficients are cultivar and species dependent only. These
> coefficients are as good as the experimental data from which they are
> derived in the first place. The degree days to Emergence, flowering
> and maturity for a maize plant are dependent only on cultivar and are
> fairly constant but for some random error. Environmental factors are
> anly modifiers for things like biomass and grain yield.
>
> I am open to further debate on this issue.
>
> Dr. Emmanuel Xevi
> CSIRO Land and Water
> PMB 3
> Griffith
> Australia
----------------------------------------------------------------------------
--
Rolf Sommer
CIMMYT (International Maize and Wheat Improvement Center)
Apdo. Postal 6-641, 06600 Mexico, D.F., Mexico
Tel. +52 (55) 5804 2004 Ext. 2128
Fax +52 (55) 5804 7558/59
www.cimmyt.org
----------------------------------------------------------------------------
--
|
