Dear all,

I have followed this discussion with great interest. I have a paper in
press (once I have proof read it) in Agricultural and Forest Meteorology
which has a discussion section on this topic. My one-line contribution
is to say that the level of complexity of a crop model should be
dependent on what question it is being used to answer. This idea is not
new and goes back at least to Sinclair and Seligman (2000; Field Crops
Research 68 165-172).
Process-based approaches of intermediate complexity do exist (e.g.
Challinor et al., 2004, Ag. For. Met. 124 99-120; Challinor et al.,
2007, Agric. Ecosys. Env. 119 190-204) - I can send pdfs to any
interested individuals.

Best wishes,

Andy

-------- Original Message --------
Subject:        Re: weakness of CERES-wheat
Date:   Wed, 12 Mar 2008 11:40:28 -0500
From:   Albert Weiss <[log in to unmask]>
Reply-To:       Albert Weiss <[log in to unmask]>
To:     [log in to unmask]



Currently in order to model the crop, we need to grow the crop to
determine develop and growth responses. If we had a thorough understanding
of the crop genetics, this process would not be necessary. One could
simulate crop responses to the environment from first principles, genetic
knowledge. So rather than call crop related inputs into models “genetic
coefficients”, a better term may be “genotype trait coefficients”. (See
Baenziger et al. (2004), Field Crops Research 90:133-143.)
Since about 1735, thermal time has been used to simulate plant
development. Its popularity is due to its simplicity, one need only know
how to add, subtract, and divide, which may be a blessing for those who
have trouble with multiplication. Unfortunately, there is no standard way
to compute thermal time. (See McMaster and Wilhelm (1997), Agricultural
and Forest Meteorology, 87:291-300.)
 One important assumption associated with thermal time is that the
developmental response is linear with accumulated thermal time. This
approach works well under some environmental conditions, but not all
conditions. A more general approach to simulating crop development is to
use a non-linear approach as typified by the beta function. (See Yin et
al. (1995), Agricultural and Forest Meteorology 77:1-16.) Also see Streck
et al. (2003), (Agricultural and Forest Meteorology 115:139-150); they
simulated winter wheat phenology using a form of the beta function where
the three cardinal temperatures changed for the three developmental
phases, emergence to terminal spikelet, terminal spikelet to anthesis, and
anthesis to physiological maturity. While using the beta function may have
advantages, like many things in life, there are some disadvantages. One
disadvantage is determining the optimum development rate for each phase,
which requires growing the crop and making detailed developmental
observations.
A way to make progress in crop simulation modeling is to never get too
comfortable and constantly challenge ourselves about current knowledge. It
will also be necessary to work with crop breeders, crop physiologists, and
plant geneticists to address the challenges that lie ahead. This idea is
not new. It was easy to type this last sentence; but it is very hard to
implement this type of joint effort.

Albert Weiss, Professor
School of Natural Resources
University of Nebraska-Lincoln
703 Hardin Hall
3310 Holdrege Street
Lincoln, NE 685830-0987

Phone: 402.472.6761
Fax: 402.472.2946
Email: [log in to unmask]
SNR web site: http://snr.unl.edu




--
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Dr. Andrew Challinor
Lecturer: Climate change and impacts modelling
Institute for Climate and Atmospheric Science,
School of Earth and Environment,
University of Leeds, Leeds, LS2 9JT, UK

Tel.  +44 (0)113 3433194
http://info.see.leeds.ac.uk/people/a.challinor
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~