I must appologize for not having posted my response to
Dr. Matthews mail on examples of development
impact of agricultural models also in the DSSAT list.
I only realized this when finding out about cross-
discussions between the AGMODELS-L and
DSSAT lists.
Matthias Langensiepen
------- Forwarded Message Follows -------
From: "Dr. Matthias Langensiepen" <[log in to unmask]>
To: [log in to unmask]
Date: Mon, 12 Oct 1998 11:46:21 +0000
Subject: Re: Examples of Development Impact of Agricultural Models
Cc: [log in to unmask], [log in to unmask],
[log in to unmask], [log in to unmask],
[log in to unmask], [log in to unmask],
[log in to unmask], [log in to unmask],
[log in to unmask], [log in to unmask],
[log in to unmask], [log in to unmask],
[log in to unmask], [log in to unmask],
[log in to unmask], [log in to unmask],
[log in to unmask], [log in to unmask]
Priority: normal
Dear Dr. Matthews,
you are speaking out of my heart.
Our group is seeking ways to model yield responses of
a wheat-barley-canola crop rotation to different levels of nitrogen
applications under North-German production conditions. Despite almost
10 years of modeling work the group hasn't succeeded to find a
satisfying solution to the problem so far.
I joined the group about half a year ago. After testing several
home-made and more-widely known models it was my initial task to
continue testing the DSSAT model version 3.1. Due to the large
popularity of the package I was faithful to find a proper solution to
our problem, but quickly found out that DSSAT is unable to predict
yield. The results confirm findings of a paper that has been recently
published in Agricultural and Forest Meteorology (Landau et al.
(1998) Testing wheat simulation models' predictions against observed
UK grain yields. Ag. For. Met. (89): 85-99). We are going to publish
our results next year.
We are now on the search for possible causes for the unsatisfying
model performance: Certainly one source of error is the
experimental setup (number of replications, plant performance under
varying weather conditions (year in, year out), soil heterogenity,
sloping terrain etc.). However, when analyzing the logic structure of
the model (CERES wheat) we found, that the simplicity and empiricism
of the modeling routines are possibly the major causes for its low
performance. After scanning recent literature on different aspects of
crop modeling we came to the conlusion that a higher degree of
mechanisism could possibly lead to a signficant improvment in model
performance.
We will now proceed in a parallel step process, validating
single DSSAT routines (soil, water, nitrogen, growth routines
etc.) and seeking ways to improve modelling towards a more
mechanistical approach .Since DSSAT is a commercialized product it
may be difficult to find latest versions for testing.
Comming to your questions I would like to suggest reading an
interesting edition of the Agronomy Journal on crop modelling
(Agronomy Journal Vol.88 pp. 689-716: Use and abuse of crop
simulations models). I was particularily convinced by Monteiths'
article on the "Quest for balance in crop modeling". Admitting that
crop models are powerful tools he concluded that they often lack
treatment of key physiological processes.
My particular answers on WHY there are no good examples:
1. Simulation models are not yet sufficiently advanced. According to
my experience modeling will be always limited by genetic expression
under different environmental conditions. This suggests that crop
models will only perform well under those conditions for which they
were developed. This means that they are NOT universally applicable.
2. None universal applicability is the reason why simulation models
often fail. However, this should NOT mean that they can be never used
for this purpose anyway. Ever increasing concern on environmental and
resource issues forces modelers to develop simulation tools aiming to
support politics and economy in finding appropiate solutions to these
problems fast enough. We need models !
3. Sometimes I get the feeling that modellers are more interested in
the intellectual challange of building their models rather than using
them. This is indeed very tempting.
I find the question "HOW we can make models more useful in
development issues" as extremly difficult to answer:
First, there must be a greater public awareness for development
issues forcing policy makers (and donor agencies in consequence) to
shift agricultural and environmental research from national to
international interests. Less than 1 percent of the gross domestic
product is commonly spent on international agricultural research.
Polititians must KNOW that they need decision tools (models) to solve
complex problems in international development.
Second, crop models must become HOLISTIC if they want to address
complex questions. I get more and more the feeling that scientists
lost they ability to oversee complex problems, because they are
concerned in solving single issues. Nowadays it is hard to find
a university professors who is able to answer questions on a broad
range of scientific issues, which he was able in the times
of humanistic education. As a consequence we have models that address
single issues, but which are NEVER able to solve intergrated
production issues (A combination of agricultural, environmental and
socio-economic decision support). Increasing complexity,
particularily in agricultural research, requires international
RESEARCH NETWORKS composed of scientist which are WILLING to find
solutions to the burning questions in the less developed world. This
means of course that researchers have to overcome their PERSONAL
EGOISM in the quest for finding solutions of a developing world.
Third, crop models user interfaces must be easy to use for the
applicants. It should be possible to validate them. This means that
their source code must be available, preferrably at no cost.
HOLISM, COMPLEXITY and UNIVERSABILITY will be the comming key issues
in modeling research.
Matthias Langensiepen
--------------------------------------------------------------------------
Dr. Matthias Langensiepen
Environmental and Agricultural Engineer
Department of Crop Science
Faculty of Agriculture
University of Kiel
Olshausenstrasse 40
24118 Kiel
Germany
Tel. +49-431-880-4398
FAX +49-431-880-1396
EMAIL [log in to unmask]
--------------------------------------------------------------------------
--------------------------------------------------------------------------
Dr. Matthias Langensiepen
Environmental and Agricultural Engineer
Department of Crop Science
Faculty of Agriculture
University of Kiel
Olshausenstrasse 40
24118 Kiel
Germany
Tel. +49-431-880-4398
FAX +49-431-880-1396
EMAIL [log in to unmask]
--------------------------------------------------------------------------
|
