Dina 1991 - 1999

brewerobstructionIA et Robotique

7 nov. 2013 (il y a 7 années et 11 mois)

542 vue(s)


13. januar 2000


Dina 1991


Edited by Mogens Flensted
Poul Einer Hansen and Iver Th




Initial principles and their implementation

The network

Research education and

Dina Research School

Research and development


Evaluation and visions


Table 1

Member institutions • History

Table 2

Funding of Dina

Table 3

Overall budget for Dina

Table 4

Doctor’s theses • PhD theses • PhD students

Table 5

People involved in Dina 1991

Table 6

Project matrix from Dina Status 1994

Table 7

Publication list







Dina was initiated in 1991 when the Internet was still an exotic phenomenon. But IT technologies
were developing rapidly, and there was a growing need for advancin
g computer science methods in
the agricultural sector. The purpose of Dina was to form a network connecting universities, research
institutions, advisory services and companies in order to promote all types of collabora
tion to apply

that is,

computer science, statistics, mathematics

and information technology in agr
culture, taken in a broad sense.

It is our conviction that the initial vision behind Dina has proved valid and durable. Starting
with 3 member institutions (two of which later
merged into one), Dina today includes a total of 7
members. In the first years, funding came from the Ministry of Agriculture and the National Danish
Research Councils (through the PIFT programme). Already in 1993
94 there was an interna
of the PIFT programme, which gave Dina a most favourable mention. The member instit
tions have found Dina so useful that they have taken over the funding responsability. In Oct. 1999 the
board of Dina recommended to the members that the network be kept up
at least until 2002

The success of Dina is based on the collaboration spanning from basic science over applied
research to advisory services; on the research school activities where new ground has been broken in
the area of Danish PhD courses; and on th
e local benefits with respect to scientific development and

Dina will still be needed far into the 21’st century, even if its method of work may be modified. It is
true that tangible results from applying IT and informatics will gradually b
ecome integrated as part of
other sciences and activities, but these sciences will also repeatedly find themselves in need of new
advanced methods, of developing specific IT facilities, and of interdisciplinary c

Agricultural production is base
d on living organisms

on plants and animals. Therefore, to
apply IT in agriculture meaningfully one must interpret the biological phenomena and take into r
gard the highly varied way in which they manifest themselves. Most often agricultural problems are

poorly structured and cannot once and for all be stated and clearly specified to the IT developer. Good
agricultural IT applications are the result from combining up to date knowledge in both informatics
and agriculture, that is, a joint effort of differe
nt scientific disciplines has to be made.

Dina can procure this for agriculture. When new achievements in informatics and new demands
in agriculture are combined in cooperation, they present a vast potential which has been, and shall be
through Dina in th
e future, detected and explored in time. Agriculture will benefit directly, and so
will the basic informatics sciences in that they are taken into an area of challenging applica

Present efforts in Dina fall into three main types of activity: networ
k function, training of r
searchers, and work with specific projects. The recommendations in the report pertaining to each of
these three can be summarised as follows.


function should be kept up at the same level as now, or be slightly expande
d. Vi
ible initiatives include: Workshops and thematic days; participation in EFITA; and other intern
cooperation and exchange. But equally important is the more invisible activity and results stemming
from the very fact that there exists a forum in

Denmark for cooperation between agricu
ture and i
formatics, a place where people can get in contact with experts from other fields whose help or co
laboration they need.

In order to fix an order of priority and to select the tasks where chances are
hest for a profitable outcome of the effort, Dina has in later years centered most of its network
activities around a number of
priority research areas

and has appointed respective

look after these areas, as described in more detail belo


Dina’s contribution in the area of
research education

should be expanded as far as the avai
ity of professional expertise and economical resources will allow. The two main aspects of this
activity are the work in individual projects at the member i
nstitutions, the ‘Dina nodes’ (frequently
involving a cooperation between two or more nodes); and the workshops and summer schools a
ranged by

Dina Research School
. It is difficult to estimate the value

with respect to research educ

of the body o
f project work, but we maintain that in a large majority of the Dina related pr
jects, students have been working with enthusiasm and have acquired a considerable professional
outcome. There is also evidence that the Research School functions well. Partici
pants have given
enthusiastic evaluations of the workshops and summer schools.

The third role of Dina, as administrator of
specific research projects
, has during later years
been toned down to apply merely to large projects financed by EU or similar sourc
es. This was
judged sensible since, as a rule, individual projects are based most naturally at one or a few of the
member institutions. We suggest that this practice is maintained so that in the future Dina will only in
exceptional cases act as project adm


vision for the future of Dina

is to focus on the further development of the Dina Research
School based on a simple but effective network organisation; to develop the Nordic, European and
international dimension; and to maintain the yearly
general workshops / seminars as an attractive
meeting point for scientists from the involved disciplines. The basic funding principle

that all activ
ties be visible in the budget and that each member institution commit itself to co
finance Dina

be preserved.

As a supplement to research and research education, Dina might consider taking up new tasks
related to education at master and bachelor level, and to presenting IT to a broader agricul
tural aud
ence. Such activities are much needed and in a
ny case they must be based on research collabor
between the Dina member institutions.

Throughout its short history, Dina has met some difficulties from the fact that we deal with
‘Informatics in agriculture’

which is not a scientific discipline, ju
st a meeting point for collaboration
or transfer of knowledge between disciplines. As such it is important, but the quality of research d
pends heavily on the participants being active and recognized in their respective disciplines, be they
agriculture or
informatics; thus in the first years rather few Dina related papers were published in
international journals. But the situation is fast improving; and Dina seniors in general have a fine
record of refereed publications.

This report first presents Dina’s
nitial principles and their implementation
; then focuses on the three
main activities:

PhD programme/Research School;

research and application projects
A concluding section presents our
evaluation of the past and the visions for the future

One of the objectives of the report is to introduce people not familiar with Dina to the basic
ideas, achievements and future plans for the network. If you wish to see detailed lists of activities,
publications etc., look in the web
version of the report



Initial principles and their implementation

Dina, Danish Informatics Network in the Agricultural Sciences, was established in 1991 with the
objective to ensure intensified interdisciplinary re
search in applications of informatics in agricultural
research. This was supposed to embrace the use of mathematical, statistical and computer science

The challenge

The background for the creation of Dina was the development in the 1980’s: New
demands were
facing agriculture and agricultural research, and at the same time the IT technologies were
developing rapidly, thus offering new potentials. These reasons in combination called for the
application of professionel computer science methods in

There had been a long tradition for interaction between statistics and agriculture under labels
such as biometry and biostatistics, but a similar tradition did not exist with computer science or with
mathematics, at least not to an appropriat
e extent. Mainfraimes and central databases related to
statistical calculations and to animal breeding and production were already in use. But the
development of personel computers and computer networks opened up a whole range of new
possibilities that dem
anded new expertise on models, on algorithms and on programming principles.

Ib Skovgaard, a leading figure of the The Danish Farmers’ Association, was instrumental
within the Ministry of Agriculture in drawing the attention to these matters. A model organ
was the Biometric Research Unit attached to the medical faculty at Copenhagen University and
headed by Prof. Niels Kei
ding. Another inspiration came in the early 1980’s when Prof. Dines
Bjørner, DTU formulated a proposal for a national strategy fo
r informatics.

The following citation from one of the original documents leading to the formation of Dina,
The National research council SJVF’s preparatory paper, Nov. 2 1989, indicates a clear
understanding that new expertise and new collaboration were

”It is necessary to concentrate the informatics effort within agricultural science … Major
tasks for the center will be the training of researchers and making contact to the
informatics environments at the universities. In addition the center shou
ld establish an
integrated cooperation with the agricultural profession, the advisory system and the
attached industry. ... In areas such as sensors, communication and process control we
suggest that agricultural research cooperates with the technological
environments... To
reach significant results it is necessary to make a large and well
targeted effort over
several years … To ensure a smooth cooperation between the local centers, high
electronic connections must be established in order to transfer
ta, including images …”

At that time we doubted that computer science expertise could be built up in the center or at
KVL, cf. this citation from our first letter of intent, Oct. 16 1990:

”It could not be regarded a realistic possibility to create a s
olid basis of expertise in data
processing at the center. Yet it is imperative to ensure the professionel environment


through a close contact to the university environments in data processing.”

However, as shall be apparent later, it still turned out to
be possible to build up such expertise
at KVL, and to attract computer scientists to be employed at DJF, Foulum.

On April 11 1991 the final project description was submitted.

During the first 3 year period in Dina, the agricultural focus changed

from e
production and optimal economic outcome, to obtaining a sustainable agriculture. This challenge is
summarised as follows (Mikkelsen 1994):

• Agriculture and environment

• Overproduction concerning certain products

• Reduction of production


• Product diversification and product quality to meet consumer preferences

• Land use and rural development

The agricultural and related sciences are reformulating their priorities to meet these
challenges in their research programmes. However
the results tend to become
increasingly complex and hence difficult to apply. It is important to develop and use
tion theory and technology in order to overcome this difficulty. One way is to
incorporate the new knowledge in efficient
decision supp
ort systems
. Also the
technological development in communication chains, electronic sensors and actuators
offer brand new possibili
ties to agricultural production and management, possibilities
that can lead to new ways of meeting the above
mentioned chall
enges. Finally, the
increasing use of informatics on the farm level, on the equipment and machinery level
and on the agro
industry level will increase the need for data communication drastically,
and there is therefor a need to improve the information infr
astucture significantly.”

In the SJVF 1989 paper cited above, the challenges from IT were seen in the following way:

”Informatics should enable us of applying the achievements of IT to carry out tasks
concerning: Measurement and data registration; Stor
ing of information and data; Data
processing; Planning, steering and control of processes, production and logistics as well
as of communication, diss
mination and decision.”

Around 1990 PC
s and other computers were still expensive, and over 400.000 DKR

spent for the first server to Dina’s UNIX network. Today we buy far stronger PC
s at, order of
magnitude, 30.000 kr. But equipment was already starting to become faster and cheaper. At the same
time, old and well
known but previously unviable algorit
hms were implemented and new methods
developed that could take advantage of the new computing potentials. New methods in genetics,
statistics and decision support became practicable due to fast calculation and communication. For
example, weather forecasts
using radar images via the internet was unimaginable in 1990 but is today
a natural facility, among other in Pl@nteInfo.

Challenges continue to change. Emphasis in agricultural research now also includes ethics and
animal welfare; new items such as bioinf
ormatics, DNA mapping, biodiversity, GMO safety,
environmetrics and genomics turn up in the media and the public agenda; and so do IT innovations
such as mobile phones and microchips. The need for informatics and IT is larger and more apparent


than ever.

Starting point and initial achevements

for Dina was in fact two pre
Dina projects

one on image processing (M. Rudemo / SJVF) and one
on databases in agriculture (M. Flensted
Jensen /Ministry of Agriculture; see case box ”The SFD
project”). Out of the lat
ter grew a project on so
called Expert Systems; it was financed by The Mini
try of Agriculture and attempted to use AI, Artificial Intelligence principles but since this turned out
to be rather unrealistic, focus soon changed to what is now called DSS, De
cision support systems,
and for a long time DSS was the buzz
word. In fact many Dina projects still are on the borderline
between modelling of agricultural systems and using models and other information sources to create

From the start of Dina muc
h emphasis was on PhD education; 7 PhD scholarships were quickly
given out and the students were offered specially designed courses in computer science, based on an
exchange agreement between KVL and DTU (see later). Thus
research education

with a
ciplinary PhD training programme was established. Many finished the education, their PhD
projects being carefully chosen so as to both being concerned with important agricultural applications
and dealing with new and interesting informatics topics.

The t
ables in the Annex illustrate the
success of this PhD programme.

As part of the PhD programme and in relation to the senior scientists in Dina, several

were carried out. Though, in the first period the main results were obtained through
the PhD

Some of the Dina related research projects are further elucidated as

Initially we handled the research projects related to Dina in a strict structure of intertwined
informatics and agricul
tural domains, and for each project t
here would be a scientific committee with
one scientist from informatics and one from agriculture. This practice has later been somewhat
relaxed, but to be considered a good Dina project it must still satisfy the double claim of being
interesting from an i
nformatics point of view but also deal with important agricultural issues. See the
table of first
generation Dina related PhD projects.

Organisation and Networking

Dina soon had a significant effect within the participating institutions, but also on the i
and personal relations between agricultural sciences and informatics disciplines at the other Danish
universities. A seemingly minor but not unimportant point was that we took great care in choosing
design and had the Dina logo registrered.

initial participants

were SP, SH, AAU and KVL, cf. table. Primary

came from the
Research Councils through the PIFT
programme, and from The Ministry of Agricultutre through the
projects described above, cf. table. The organisation included an adv
isory board, the PhD
arrangement, and regular workshops / thematic days organised for Dina’s scientists but open to
others. Much emphasis was put on creating an
electronic communication infrastructure
, based first
on a UNIX network, next also on the Intern
et, and soon after even on web
technology which was just

On the initiative of Dina a video
link was established between KVL and Foulum. Being early
starters with this technology we ran into some beginner’s trouble, but the link quickly became an

indispensable tool in the communication within the network. Later video
link facilities were
established at several other universities and institutions.


The network cooperation soon extended internationally through an EU
concerted action
EUNITA, European

Network for Information Technology in Agriculture, initiated mainly through the
work of Dina. This network led to common research projects and later also to the establishment of
EFITA and DSIJ, see below.

Collaboration grew with other universities, rese
arch institutions and with DAAC, Danish
Agricultural Advisory Center, and it became natural to invite new members into Dina which has
today 7
, that is, units at 7 different member institutions; see Chapter 4 for an introduction to the

nodes. All alon
g this expansion proces, the networking principles were carefully considered. Dina is
not an institution but a network / a consortium of institutions, so it was made clear that all Dina
activities shall take place within, or in collaboration between, exist
ing institutions, and that all
activities must be visible in the budget.

There was some initial confusion about the concept of ‘associated projects’ which made it
unclear what Dina could take responsibility for, or the honor for. In 1996
97 Dina was reor
under a carefully designed set of bylaws, such that today Dina has a legal and formalised status, but
still the essential among the original ideas have been kept up.

Dina was evaluated favorably in the international mid
term evaluation of the PIFT

Some recommenda
tions were given, in particular to strenghen the involvement of professional
computer scienti
sts. This was taken seriously and was successfull, see later.

Apart from its specified tasks, that is, working with research educati
on and specific research projects
and performing network functions, Dina also feels obliged to be an analysing, debating, inspiring and
initiating organism rooted in existing research environments. It is Dina’s ambition to have a
catalyzing effect on the d
evelopment within all its member in

”The fathers of Dina had providence.
What we now consider in the nature of
things was not so in those days.”

The network

Network is a keyword to the way Dina promotes its general objectives o
f making contact between
disciplines and institutions in agricultural research, IT and informatics.

The basic network activities within the organisation itself are supervised and coordinated by the
Network Committee

which consists of 12
15 senior scientis
ts from the member instit

Information about Dina and its various activities is provided mainly by Dina’s
Internet hom

Supervised by Dina’s leadership, the homepage is maintained on a day
day basis.

The network committee organizes yearly
general workshops or thematic days
, with the purpose
of providing a meeting point across the disciplines and institutions.

It was an EU initiative from Dina which led to the creation of
, European Network for
Information Technology in Agriculture an
d Environment, in 1996. Dina arranged the first EFITA


conference at KVL in 1997 and still plays an active role in the EFITA activities as well as in those of
, the Danish Society for Informatics in Agriculture.

Priority Research Areas and IT Coordina

Dina could not be expected to deal with all the sub
disciplines and research areas that are relevant to
its general objectives. To focus its activities, a number of
priority research areas

have been chosen

the choice to be seen as a dynamic process.

For each priority area one or several

been appointed, to act as mediators. The specific tasks of the IT
coordinators include: To write and
maintain a home page for the priority area in question; to arrange yearly professional workshop
s in
the area; to assist in coordinating Dina applications to large research programmes; and to partic
in the activities of Dina’s Research School whenever appropriate.

Among the areas where Dina has so far not been active (but may be so in the futur
e), let us
mention: Econometrics, epidemiology, systems analysis, and sub
disciplines in food science such as
chemometrics and food production documentation.

Below find some details on the five research areas that have status as Dina priority areas as o
November 1999.

Priority research area: Spatial statistics in agriculture

IT Coordinators: Morten Larsen, Dina KVL, and Rasmus Waagepetersen, Dina DJF.

(The web page

gives a description of the research area within
ina; provides links to people working in the area as well as links for download of publications; and
contains various news and announcements.)

Present activities mostly fall in five subareas.
Stochastic geometry

deals with random geometric
objects and is
related to

which has been used e.g. to count and measure cells in onions; to
quantify spraing in potatoes; and to describe the structure in grass swards.

Point processes

are mo
els for the spatial distribution of the points in point patterns.
Such data occur frequently in spatial
statistics and in agriculture. Examples of application include: Positioning of trees in a forest; and d
scribing the development of weed plants.

Image analysis

extracts information from images, for e
ample to disting
uish between weeds, crop and soil in a field.


mainly deals with spatial
prediction. Examples: Estimating, on basis of measured clay content in a number of locations, the
clay content at unobserved locations; describing how weed count data
depend on organic matter, clay,
silt, etc.


(Markov chain Monte Carlo) is a general computational technique which is useful
when statistical models become analytically intractable.

The IT coordinators maintain the webpage 'Spatial Statistics in Agric
ulture'. They organised
two workshops, 'Geostatistics in the Agricultural Sciences' (Sep. 18 1998) and 'Image Analysis and
Spatial Statistics in Forestry' (Nov. 2 1999, a Nordic activity partly funded by NorFA). Further R.
Waagepetersen organized a sessio
n at the IBS Nordic Regional Conference, June 1999. Both IT c
ordinators lectured in Dina Research School at the workshop in April 1998 and the summer school in
August 1998, and they produced an introductory lecture note on MCMC for the workshop while
per Møller produced a note on MCMC for the summer school.

Priority research area: Information Technology and Agricultural Engineering

IT Coordinator (until 1998): Hans Rischel, Dina DTU


The basic idea for creating this area was that the development of
IT intensive equipment for the
primary agricultural sector is hardly possible unless engineering experts are involved. In order to
approach the main goal DTU and DJF (Bygholm) initiated a work on a preliminary project proposal
CompuWeed addressing develop
ment of solutions for non
chemical weed control.

To throw light on CompuWeed, and the expertise within information technology and agricu
tural engineering available from the four DTU Departments, the IT Coordinator organised a seminar
for Dina at DTU on
March 11, 1998. Results from the seminar were used to continue the work on a
research proposal. In 1999 a development plan, “CompuWeed 2”, has been presented; it involves
Flakkebjerg and a number of the DTU experts (see later section Dina DTU).

ority research area: Internet
based Advisory Systems in Agriculture

IT Coordinator: Henrik Kjær Nielsen, Dina Skejby

(The web page

provides a general intr
tion to the priority area; refers to I
nternet related Dina publications and people; and gives

Danish language

a more thorough but easily read description of the area.)

The Internet is rapidly gaining momentum, both in terms of use and of the number of applic
tions based on Internet tec
hnology. In agriculture and forestry there is a large potential for applying
based applications to answer questions and to help solve everyday problems regarding pr
duction and management.

The establishment of this Dina priority area aims at stre
ngthening the cooperation between
nodes in Dina that are actively involved in research and development with respect to the Internet. The
intention is to support the transfer of technology and know
how between parties within Dina or with
close relations to

On March 10 1999, a workshop was held at the Danish Agricultural Advisory Centre, Skejby
where topics such as XML
technology, Internet as a medium for communication, and Testing your
application were touched upon, and various Internet applications w
ere presented. In connection with
the workshop the Dina Thematic Day 99, Danish Agriculture in the Communica
tion and Information
Society, took place on March 11. For more information on the specific projects, see the section “R
search and development” in
this report, and also the case box on

Priority Research area:

Based Decision Support Systems in Agriculture

coordinators: Erik Jørgensen, Dina DJF and Allan Leck Jensen, Dina DJF


based decision support systems (DSSs) have a well
established tradition within agr
culture. The DSSs range from simple accounting
based systems to systems based on detailed dete
ministic or stochastic models. Usin
g different methods to model the same domain will produce di
ferent results, so if we want to compare the strengths and shortcomings of different DSSs, the unde
lying models of the DSSs must be well documented.

The purpose of this priority area is to foc
us on the use of models in agriculture; to follow the
development of new modelling methods; to promote good modelling principles; and to inspire and
coordinate joint applications for funding. Underlying assumptions and model aspects of DSSs in a
are: Observations and measurements; Recognition of uncertainty in the system; Ma
ment of uncertainty; Decision Support; Model Evaluation; From prototype to end

In relation to the priority area, a Dina Workshop was held in Foulum on Oct. 22
3 1998, with


54 participants. The workshop had four main sessions: Modelling of growth processes; Bioinfo
ics; Decision support; and Graphical models and model selection

A Dina Workshop, Probabilistic Modelling of Potato Late Blight, is scheduled f
or the spring of
2000. With such a narrow subject, it is intended that the workshop go into detail with the specific
problems concerning the ongoing development of the DSS.

For information on specific projects, see below in “Research and development”.

riority research area: Bioinformatics and Quantitative Genetics

IT Coordinators: Gunter Backes, Dina Risø and Mogens Sandø Lund, Dina DJF

Initiated in the fall of 1999, this priority area aims at promoting the application of bioinforma
ics and quantitat
ive genetics in agricultural research

by participating in the drafting of new projects
concerning respective topics, by offering educational opportunities, and by building up an interface
for the contact to and between experts in these research fields.


applies computer algorithms to sort through and to analyse biological data. It

, that is: methods to analyse information from DNA and from pr
tein, respectively. Because of the explosive growth in biological d
ata there is a need for more and
more advanced search and comparison algorithms, but this growth has also made an extended fun
tional analysis possible.

Quantitative genetics
takes the trait expression as starting point and aims at
identifying genetic va
riation in the trait of interest. One important topic in this field is the mapping of
quantitative trait loci (QTL), to identify chromosomal regions with genes that affect quantitative
traits. To this purpose, genetic markers with known positions are ident
ified and assoc
ated with ph
notypic records.

There is a large gap between bioinformatics and quantitative genetics, yet also signs that the
two fields are moving towards each other: In bioinformatics, the change from pure data collection to
functional an

in quantitative genetics, the evolution of QTL analysis from roughly defined
traits to more sophisticated characters, revealed more precisely and under well
defined conditions.
Therefore we decided not to divide the area in ‘bioinformatics’ and ‘q
tive genetics’, but rather
in animal related topics (M.S. Lund, DJF

QTL mapping and marker assisted selection in cattle and
pigs) and plant related topics (G. Backes, Risø

QTL analysis, localisation of resistances and marker
construction in bar
ley and wheat).

The first output from this priority area was a contribution to a project application, “Mini
for Statistical Bioinformatics”.

The Dina nodes

Dina’s activities at the member institutions are integrated in the local departments as pa
rt of their
regular activities. However all essential obligations to the Dina network should be clearly expressed
in the budgets of Dina and of the institution.

Below we briefly update on the seven
Dina nodes
, a term that refers to Dina’s activities at ea
single member institution. The text is based on views of the past and the future, as seen

and from varying angles

by local scientists at the nodes.

Dina DJF

Chief responsible: Iver Thysen, Deputy Head of Dina

Dina Foulum
, as was the
name of Dina DJF until 1997, was from the beginning a main node in


Dina, located at Research Center Foulum as a joint project of the Danish Institute for Plant and Soil
Science and the Danish Institute for Animal Science which in 1997 joined to form the Da
nish Inst
tute of Agricultural Sciences = DIAS, in Danish:

Initially, Dina Foulum was staffed with 5 PhD
students, 1 Masters student, 1 senior researcher (from 1993: 2), 1 computer technician and 1 secr

In 1992 a
Department for Informatics and
was established at the research center, and
Dina Foulum became a part of the new department, constituting its informatics section but also pla
ing a crucial role in the development of the biometry section and for its reputation among Danish
ticians. In 1997 the two sections became research groups in a new
ment for Agricultural

While Dina Foulum was quite well
defined, Dina DJF is less clearly demarcated; many activ
ties are co
financed by other sources than Dina, and Dina pe
ople work in separate research groups at
DJF. Main research activities at Dina DJF are: 1. Internet
based information / decision support, man
festing itself in
; 2. Decision support systems based on Bayesian methods; 3. Spatial st
tistics. The tw
o latter are described above under the relevant ‘priority research area’ labels while
Pl@nteInfo, the most broadly
aiming of Dina’s activities

and stunningly successful as such
, is
elaborated in a case box.

Dina has had a major importance for the appl
ication of informatics at DJF. Dina Aalborg is a
main partner, and there has been a constant flow of knowledge from AAU to Foulum, notably on
Bayesian networks and applied statistics. Cooperation on PhD education and projects has been exte
sive and several

AAU candidates are now employed as researchers at DJF.

The rapid development of the biometry group at DJF, increasing from 2 to10 scientists during
99, would not have been possible without Dina. The extraordinary funding through Dina of
projects and the presence of qualified researchers were a necessary basis for the relatively
early uptake of the emerging Internet technologies resulting in Pl@nteInfo.

Further activities of Dina DJF include IT controlled equipment.

To conclude, the main

benefit of Dina for DJF is that the institute has obtained research cap
bilities on an international level within the fields of biometry and informatics. Future plans include
utilizing these capabilities in a broad range of activities within the scope of
the research at DJF. A
specific aim is to develop the area of decision support under uncertainty.

Dina KVL

Chief responsibles: M. Flensted
Jensen, Head of Dina, and A.K. Kristensen,

Head of Dina Research

Dina KVL played a leading role in promotin
g computer science and IT methods in Danish agr
cultural science in the early 1990’s, and in establishing IT based communication systems within the
Dina network (Unix/Internet, VideoLink).

The main objective for Dina KVL is to make KVL the focal point for

Dina in expanding the
research relations in informatics/IT between the universities and agricultural research. The hope for
the future is also to support the role of informatics/IT also in relation to education, particularly at PhD
level as it is already
done, but also at master level, in disciplines such as bioinformatics, genetics,
economics, agricultural engineering, and mathematical modelling.

The Dina activities at KVL, many of which concern collaborative work described in the pr
jects section and
in case boxes, are concentrated at three departments.

The Department of Mathematics and Physics

houses since 1996 the management and secreta


at of Dina; provides an IT coordinator (M. Larsen) and a research school expert (P. Sestoft); and co
tributes to
various projects, the most important being: Forest inventory by image analysis of aereal
photographs and possibly from satellites (M.Rudemo, M. Larsen, J. Lund, M.J. Tarp
Johansen; co
laboration with Dina FSL), Pl@nteInfo (P. Sestoft, M. Larsen; collaborat
ion with Dina DJF and Dina
Skejby), and the education project Biomodelling (M. Flensted
Jensen and other teachers) which may
later be followed up by other undergraduate education project proposals.

The Department of Agricultural Science /Laboratory for

Agrohydrology and Bioclima
, two projects in 1997
99 have financed efforts concerning the reprogramming of the model
Daisy (S. Hansen and P. Abrahamsen). See the case description.

The Department of Animal Science and Animal Health

houses Dina Resea
rch School, the head
of the school, A. Ringgaard Kristensen, being relieved from part of his ordinary duties by an assistant
professor funded by Dina. The department also collaborates with Dina DJF on IT related research
projects in animal husbandry.


KVL has taken an active interest in the establishment of the IT University of Copenhagen
which was initiated in 1999 by the government. P. Sestoft, who is associate professor at KVL and one
of the experts of Dina Research School, teaches at present half
ime at the new IT University. It
seems natural that Dina contributes to developing and coordinating IT education in applications rela
ed to agriculture.

Dina KVL has been involved in the coordination of large project proposals and in applying for
from national research initiatives such as the IT Programme, the SUE programme and the
Biotechnology Programme.

Dina Aalborg

Chief responsible: Steffen L. Lauritzen

Dina has been of great value to some of the research environments at AAU, in later years
ly in statistics. The significance has been both direct since Dina has provided funding for research
appointments, and indirect by creating cooperation and personal contacts. Dina Aalborg provided
expertise to agricultural applications, in particular
n probabilistic decision support systems.
cooperation, in the true sense of the word, between experts from AAU and other Dina nodes has ta
en place, but to a limited extent only. An important outcome has been that several experts trained at
AAU ar
e now holding permanent positions at DJF or KVL (while a few have moved the other way,
from KVL to Aalborg), and also the supervision by AAU experts of PhD students at DJF. Dina has
contributed markedly to keeping up an active and variated PhD enviro
at AAU and has in add
tion inspired several interesting master thesis projects.

In the future Dina Aalborg expects to develop the cooperation in Dina in new ways, viz. by way
of the newly established, Aalborg based ‘Center of Agricultural Engineering and
IT’. Besides Dina
Aalborg strongly supports the continuation of the Dina Research School and intends to contribute to
the activities of the research school.

Dina DTU

Chief Responsible: Kaj Madsen.

In the beginning of the 1990’s DTU researchers like A.P.

Ravn and J. Steensgaard
were instrumental in basing the Dina activities on sound computer science principles. J. Steensgaard
Madsen worked at Dina KVL for about 3 years and contributed valuably to the development of data
processing as a discipline

at the university. There was a collaboration with Dina KVL on the SFD


project, and other projects at DTU were financed by Dina in 1995

In later years the activity has been concerned with the research school (Prof. P.C. Hansen) and
with the attempt to

organise a research effort in IT oriented, non
chemical weed control. A develo
ment plan, “CompuWeed 2”, has been presented; it involves a pilot project for developing a self
conducting vehicle for mechanical weed removal. Researchers at DJF
will cooperate
with a number of DTU experts covering the following areas: image processing; optimization; aut
mation and control engineering; construction and product development. The DTU experts can have
decisive importance in the future realisation of pr
ojects like CompuWeed 2 which will be of value
not only scientifically but also in education and industrial product development of machines and

DTU shall in the future be involved in Dina with the following four departments: Dept. of I
on Technology (IT), Dept. of Control and Engineering Design (IKS), Dept. of Mathematical
Modelling (IMM) and Dept. of Automation (IAU). The main goal was and still is to participate in
new research projects within Information Technology and Agricultural En
gineering and contribute to
the Dina Research School.

Dina FSL

Chief responsible: J.P. Skovsgaard

The Dina node at the Danish Forest and Landscape Research Institute focuses on digital image
analysis for forest inventory purposes. Research, carried out

in close cooperation with mathema
ticians and statisticians at Dina KVL, has been concerned with stem mapping and mensuration of tree
and stand dimensions, and with variables based on ordinary pan
chromatic aerial photographs for
pure even
aged forest sta
nds. Similar projects take place at the post
doctoral level for mixed multi
storeyed stands.

The joint effort, a smooth and well
targeted cooperation between the two Dina
nodes, has resulted in an internationally high position in digital image analysis f
or forest inventory. In
the future FSL and KVL intend to maintain and enhance efforts in digital image analysis for forestry
and lan
scape use.

The Dina membership has been an advantage for FSL in other matters as well, such as the co
tinuing development
of an IT
based information and extension service, regarding PhD courses; and
besides for ensuring a generally high standard in mathematical and statistical methods for scientific
analyses. This cooperation too should be enhanced in the future.

Dina FSL is reinforcing research in Forest operations, in particular by establishi
ng a
research professor chair. Activities in this area will benefit greatly from an intensified cooperation
within Dina, for example on the development and use of IT
based h
igh precision technology in ten
ing and harvesting operations in forestry.

Dina Risø

Chief Responsible: Hanne Østergaard

Established in 1998, this node covers the Dina activities at the Risø National Laboratory, i
volving mainly the Plant Biology and Bi
ogeochemistry Department which is engaged in research to
establish the scientific basis for new methods in industrial and agricultural production, notably met
ods that exert less stress and strain on the environment. The different programmes of the departm
reflect the various approaches to achieve this aim: Biogeochemistry; Plant genetics and epidemiol
gy; Plant ecosystems and nutrient cycling; Plant products and recycling of biomass; Plant
symbioses; and DLF
Risø Biotechnology.


The main

of Dina Risø is to enhance the use of advanced computer based mathemat
cal and statistical techniques within the Risø research areas. In co
operation with the other Dina
nodes, the improvement of knowledge of the present staff as well as the education o
f young scientists
shall be carried out.

PhD students and young scientists at Risø have already profited from workshops and courses
arranged by Dina Research School. In the near future (year 2000) the Dina relevant
research projects

will focus on the plan
t related parts of the priority research area Bioinformatics and Quantitative G
netics (see above), in particular on methodological analyses of QTL (quanti
tative trait loci) in relation
to marker assisted breeding of crop plants; IT in relation to sensor
data from precision farming; and
mathematical and statistical models for population dynamics in space and time. But also collabor
tion on bioinformatics is foreseen.

Dina Skejby

Chief responsible: Henrik Kjær Nielsen

The Dina node at the Danish Agricult
ural Advisory Center was established in Jan. 1996, with
the overall objective to encourage the use of informatics in primary agriculture, agricultural extension
service and the agro
stries. Following the establishment of Dina Skejby, focus was directe
d t
wards two main areas: Information infrastructure and decision support systems.

information infrastructure
, work has had an impact on the heavily growing interest in
applying Internet technology in the agricultural extension service. Dina Skejby

is also a key co
tor to the ongoing development of Internet
based services from the Danish Agricultural Advisory

Although Dina Skejby (as opposed to other nodes) is not directly involved in research activities
and PhD training, the work cond
ucted at Skejby still benefits from the collaboration with the other
nodes, primarily with Dina DJF and Dina Aalborg.

Future work of Dina Skejby will focus on preparing and evaluating new technology to be a
plied in Internet
based advisory services. More
specifically, the possibilities for exchanging, integra
ing and presenting information and components from several contributors over the Internet will be
examined, because such possibilities form the basis of future systems to distribu
te, to modify and to

illustrate knowledge in useful ways.

Research Education and Dina Research School

Background and objectives

From its start, Dina placed emphasis on PhD education

Dina initially funded 8 PhD students and
another 6 w
ere associated to Dina. All PhD students funded by Dina must include a major course in
computer science in their curricula. In order to provide this teaching an agreement was made between
IT; Dina payed an assistant professor at DTU, and DT
U payed a senior r
er in informatics to spend three years at KVL.

Much experience was earned from computer science courses for PhD students with a different
main scientific background. The importance of these activities led to establishing the




in 1997, in order to strengthen the training on applying informatics in agricultural research.
The objective of the research school is “to support and inspire the research training of new academic
staff members in the agricultural research,

the advisory service and the agricultural sector in general”.
The support is mostly rendered as courses and workshops, open to all PhD students whose projects
combine agriculture and informatics. Mostly students with agricultural background attend the cou
and workshops which are announced through an open mail list, but also some with an informatics

The Dina Research School is headed by Assoc.Prof. A.R. Kristensen, KVL
IHH. The head is
responsible for the administration, and together with a

number of experts from the informatics disc
plines he is responsible for the scientific contents of the activities. As of December 1999, the experts
connected to the research school are: Prof. P.C. Hansen (Numerical analysis), DTU
soc.Prof. P. Sest
oft (Computer science), KVL
IMF, Senior Researcher H. Stryhn (Statistics), Danish
Veterinary Labor

Each year the research school organizes two regular
; they are held at appropriate
conference centers and the duration is two half days from

lunch the first day to lunch the next day.
Subjects covered until now have been: Introduction course, Dec. 1997; Stochastic simulation, April
1998; Numerical problems and mathematical software, Dec. 1998; Bayesian networks, April 1999;
Differential equati
ons, Dec. 1999. The number of participants (PhD students) has varied between 16
and 20. As a rule, the research school experts and some of Dina’s IT Coordinators teach at the wor
shops, together with external lecturers.

The research school organizes an an
summer school
, with a duration of two weeks (10
working days). From 1999 these courses form part of a Nordic collaboration, and there are partic
pants from all five Nordic countries. Up to now, two summer schools have taken place, both at Tune
kole. Titles were: Software construction in Java for agricultural applications, Aug. 1998, and
Computer Intensive Statistical Methods

with Biological Applications, Aug. 1999. The number of
participants in 1998 was 18, of which 12 were PhD students paid
by Dina while the remaining 6 paid
themselves; and in 1999 there were 21, of which 7 were from Denmark, 5 from Finland, 5 from No
way and 4 from Sweden. More than 40 PhD students applied in 1999 but the number of partic
was limited, both for economic
al and practical reasons. The course was financed by NOVA.

On the
initiative of the Dina Research School a
Nordic research training network

financed by NorFA has
been established. The objectives of the Nordic network is basically the same as for the Dina R
School, in particular to organize common summer courses. Many PhD students need informatics
training but the specific needs of the individual vary from project to project, and from person to pe
son, so the number of potential participants in a spec
ific course can be rather limited. A Nordic ne
work for research training is a significantly better basis for such courses, and three have already been
planned: 1. Differential equations and dynamic systems in agriculture. Organized by the Dina R
search S
chool, Denmark, 2000; 2. Geograhical informa
tion systems and spatial statistics. Organized
by Gerhard Skagestein, Norway, 2001; 3. Design of data generation

experimental design. Organized
by Hannu Rita, Finland, 2002.

Plans for the future
. The courses
and workshops will be continued, and it is intented to create a
‘common market’ for workshops in the Nordic countries. PhD students financed by Dina will be a
sociated with the research school. Through project applications, it has been tried to acquire suc
ing but up to now unsuccesfully.

. The organisational structure of the research school, with a head from agricultural
research and associated experts from the informatics disciplines, has proved to be an efficient co


struction in practice.

The structure has made it possible to offer workshops and summer schools at an
advanced level, but still relevant and understandable to PhD students from the agricultural disc
plines. Workshops and schools have been evaluated very positively by the partic
ipants, and it has
been demonstrated that it is possible to offer arrangements that are simultaneously relevant to PhD
students from the agricultural and the informatics disciplines. Even though the main emphasis in on
presenting informatics which can be u
seful in agricultural applications. The Nordic extension of the
research school has been accepted in all five countries; collaboration in the contact group is smooth
and the first common summer school had applicants in excess in all countries except Icelan

Research and development

This section briefly reviews recent and ongoing research projects in Dina. The review is organized
according to the five Dina priority research areas, supplemented by a special section for projects
which do not fit into this

scheme. We also give some indications of the number and character of pu
lications produced within the projects, or in relation to Dina in general. For details the reader is r
ferred to the web (
) and to the yea
rly reports from Dina’s various member institutions,
since all Dina activities are carried out at one of these institutions or in collaboration between them.
For projects from the starting phase, 1991
94, we further refer to the list of projects in Table 6

and to
a number of Dina status reports mentioned below.

A number of the most successful and interesting Dina projects are presented in separate ‘case
boxes’, written with the purpose to tell the non
specialist what we consider essential results. A more

professional presentation has to be looked for in the relevant publications.

Projects related to Spatial statistics in agriculture

There has been, and still are, several projects related to image processing and statistical methods, cf.
the description o
f the priority research area and the ‘case box’ on
Spatial tree pattern analysis from
aerial photographs

Other current projects concern geostatistical methods and their applications in agricultural field
experiments. Researchers involved are O.F. Christe
nsen, J. Møller, F.Aa. Tøgersen and R. Waagep
tersen. For example, one ongoing project concerns "Analysis of spatial data using linear mixed mo
els and Lange
type Markov chain Monte Carlo", where various methods for Bayesian, likelihood
and non
tric inference procedures for spatial generalized linear mixed models are developed
and computational aspects are studied in detail. The methods are applied to a data set of counts of
weed plants in a field.

Projects related to Information Technology and
Agricultural Eng

Only a few projects have been established in Dina under this heading, that is, they should involve not
only IT but also engineering research in a true sense. Under the priority research area we have me
tioned OPTIMAL, a collaborati
on between DJF, AAU and KVL; and the project proposals Co
puWeed and The Farm Animal Information Society. The two latter have not yet been realised, but it
is felt that Dina should develop the area further. Partly in relation with Dina, Finn Conrad obtain
funding from the Danish Research Councils to the project IMCIA

Intelligent Motion Control and


Intelligent Actuators (1992

1997, collaboration with AAU and three companies) and DIMEC

Danish InfoMechatronic Centre (1997). Four PhD students have fini
sh with the PhD Degree. Fu
thermore, partly in relation with Dina, the Research Frame Programme OPTIMAL in agricultural
engineering obtained funding from the Danish Research Councils. OPTIMAL (1996

1999) in co
laboration between KVL, DTU, DJF and AAU, Pr
ogramme Coordinator: Bent S. Bennedsen, KVL.
The budget was limited to employ two research assistants in two years and one PhD student

Projects related to Internet
based Advisory Systems in Agriculture



) is an information system for crop production where the I
ternet is used to supply farmers and agricultural advisers with just
time information about risks of
diseases, pests, drought, etc. The information can be selected on a

national or a regional level, and in
some cases it can be tailored to fit the particular situation of the farm by allowing the user to submit
specific data. Pl@nteInfo is constantly developing since new features are added every year.

, a new

feature of Pl@nteInfo, offers the possibility to query a centrally located dat
base about results from field trials for crop varieties. As results are collected from several different
sources, they are processed and instantly made available to the public
on the web.

The concept as well as the technology underlying Pl@nteInfo are planned to be exported to
countries in the Baltic Sea area. The goal is that researchers will be able to share decision models,
technology and know
how across borders.



) is the name of a centralized, web
based information sy
tem run by the Danish Agricultural Advisory Centre. The system contains papers, reports and han
books published by the Centre and targets on far
mers, advisers, researchers and teachers in the agr
cultural sector. In addition to the large number of publications, L@ndbrugs
Info also serves as a pla
form for various on
line services, subscriptions etc. Future plans involve the preparation and the di
tribution of information to regional advisory centers, in order to be integrated in their web
based o
fers. Other plans focus on improved visualization and presentation of inform

Projects related to Model
Based Decision Support Systems in Agricultur



/nippp) aims at achieving added value of Internet
based information
services for plant protection from cooperative efforts in the Nordic countries. An application for
ng of these activities has been submitted to EU’s Fifth Framework Programme.

Development of Decision Support Systems for Integrated Pest Management in the Baltic


) is a project about transfer of knowledge and technology for decision
support. An infrastructure for capture and dissemination of weather data is being developed, and e
ng Danish DSSs (PC
Plant Protection, NegFry and Pl@nteInfo) are modified, exported and val
dated for the Baltic countries, and to some extent also for Poland. Basic principles for the develo
ment of future DSSs are established to ensure far more flexible,
maintainable and exportable applic

Probabilistic models in Pl@nteInfo


). Pl@nteInfo has reached the level
of development where a framework fo
r management of Internet applications exists, and where most
existing models of relevance to internet
based decision support within crop production have been
implemented. The models are usually quite simple, for example based on accumulated daily te
res. This project investigates how more advanced modelling techniques can be applied to handle
uncertainty in predictions, reasoning, decision optimisation etc. The project has focused on a specific
disease: potato late blight. This fungal disease is the t
arget of the largest consumption of fungicide in


Danish agriculture. As its first objective, the project aims at having daily probabilistic forecasts of
sporulation level and weather conditions for spraying implemented in Pl@nteInfo for the growing
of 2000.

Model based decision support in animal production



This project title covers a range of research initiat
ives, with researchers at Institute for Animal Health
and Animal production, at KVL and Biometry research group at DIAS as the main partners. A d
tailed overview (in Danish) can be found at the home page. A key activity is optimisation methods,
where meth
ods for simultaneous optimisation of decisions with varying time horizon have been d
veloped. The concept behind Hierarchic Markov Processes for decision support has thus been e
hanced to include multiple levels. A PhD student is currently studying further

aspects of the tec
niques, such as the similarities with robot navigation. The combination with Influence Diagrams is an
important topic and research at Dina Aalborg concerning the so
called Limited Memory Influence
Diagram is a promising part of this pro
cess. Using these techniques approximate solutions may be
found for complex decision problems, that currently is unsolvable. This research area is a very active
partner in international collaboration.

Another key area is methods for monitoring in pig
production. A PhD project is in progress in
collaboration with IHH and National Committee for Pig Breeding and Production (NC), with impl
mentations concerning water and feed consumption in piggeries. In addition, a current prototype d
veloped at DIAS for
monitoring pregnancy rate using Bayesian networks, will be implemented in
selected sow units, in collaboration with NC.

An object oriented simulation model of a pig herd has been developed. This model includes
aspects concerning the information processing

i.e. registration, data processing (Belief Manage
and decision support, and are used in various contexts, such as project collaboration, undergraduate
and graduate studies etc.

The group participates in several projects financed by the Danish Mini
stry for Food and Agr
culture, such as Respiratory diseases in pigs, Animal Health Economy (CEPROS). Objective Health
in Pig Production (CEPROS).

Recently, the research group has formulated a research project
The Farm Animal Information
The aim

is to implement at farm level, the various elements of information processing in a
coordinated framework.

Funding for this project has not yet been obtained.

Projects related to
Bioinformatics and Quantitative Genetics

Being a new priority research are
a no research project have been defined in Dina, except for an appl
cation to the research councils: Statistical bioinformatics (mini
centre) in collaboration between
Foulum and Risø. The centre is not likely to receive funding, but there may
be other
ways to start smaller project activities in this direction, which is important for agricultural applic
tions of modern DNA

Other projects

Several project related to the general objectives of Dina, but not fitting into the present p
riority r
search areas, could be mentioned. From the ‘cases’ we have: Daisy, Biomodels and the SFD project
as well as the EU
project EFITA. Other EU projects have existed which had some relation to Dina:



The senior scientists in D
ina have in many cases also participated in research with no immed
ate relation to the interest sphere of Dina. These projects are not mentioned here unless Dina in one
way or the other has contributed with means or resources. But such activity is importan
t, in order to
keep up good contact with the respective research disciplines.


Only key reports related to the Dina activity and organization are listed below. For a more extensive
list of scientific and other literature, consult Dina’s web

where you can get
to the

compiled for this report. See also Table 4 which lists all doctoral theses and PhD
theses obtained in relation to Dina. The database on the web concerns the period

99 and has the
following number of entries:


dissertations and monographs


refereed scientific articles


refereed proceedings papers


other publications (preprint, internal report, un
refereed proceeding paper etc.)

in these numbers are the publications and preprints published in the “Dina report”, “Dina
Notat” and “Dina Publication” series.

It should be noted that since Dina is a network of people from several disciplines and instit
tions, the borderline between Di
na and non
Dina publications is not very sharp. However we have
tried to include only publications produced in full or partially with Dina funds, or which has a clear
relation to the collaboration initiated by Dina.

The Danish Research Councils, 1993.
term Evaluation. Danish Research Programme on Info
. 52 pp.

The Danish Research Councils, 1993.
Danish Research Programme on Informatics 91

38 pp.

Dindorp, U; Thysen, I; Flensted
Jensen, M (eds.), 1993
. Dina 1 May 1991

31 January 1993, Stat
with appendices.

Dina Notat no. 12. Dina KVL, The Royal Veterinary and Agricultural University,
Copenhagen. 104 pp.

Thysen, I (ed.), 1993.
Dina 1 February 1993

31 August 1993. Supplement to Dina Status 1993
Dina Notat no. 15. Dina Foulum, Research C
entre Foulum. 18 pp.

Mikkelsen, S A; Flensted
Jensen, M, 1994.
The Strategy for Dina to Year 2000
. Dina Notat no. 20
(17), Dina Foulum, Research Centre Foulum. 12 pp.


Jensen, M, 1994.
Dina 1 September 1993

31 August 1994, Status and Plans
. Din
a Notat
No. 24. Dina KVL, The Royal Veterinary and Agricultural University, Copenhagen. 101 pp.

Jensen, M; Thysen, I; Mikkelsen, S A, 1994.

Strategy and Plans for the Period 1995

Dina Notat No. 25. Dina KVL, The Royal Veterinary and
Agricultural University, Copenhagen,

53 pp.

Jensen, Finn V. May, 1995.
Final report on the activities in the research programme

ODIN (01.03.1991
Aalborg University, Department of Computer Science, Internal R


Evaluation and visions

In an attempt to evaluate the effort of Dina during its 8
9 years of existence it can be concluded right
away that the story of Dina is essentially a success story. The original objectives have been met to a
large extent, and
the initial vision behind th
e network has proved valid and durable. This is reflected
in several ways. 1. The number of members has expanded: Starting with three member institutions
(two of which later merged into one), Dina today includes a total of seven ‘nodes’. 2. An
nal evaluation of the Danish national PIFT research programme, which formed part of the
funding basis for Dina in its first years, gave Dina a most favourable mention. 3. After Dina had
initially received direct national support, the member institutions f
rom the mid
1990’s took on the
basic funding respon
sability. 4. In October 1999 the board of Dina recommended to the members
that the network be kept up at least until 2002.

The success of Dina is based firstly on the

to some extent invi
sible, but not less
real for that reason

established from basic science over applied research to advisory services;
secondly on
local benefits

with respect to scientific development and infrastucture; thirdly on the

that has been formed,

a.o. expressing itself in the organisation of general thematic
days, workshops and courses; and last, but not least on
the research school

activities where a new
standard for PhD courses has been set.

From the start of Dina, the task of
PhD training

been in focus

The first generation of Dina
related PhD projects were especially important; not only did they lead to good research but they also
produced strong specialists who are now holding key positions for the implementation of modern IT
in Danish ag
riculture and land use. The experiences from this first brood of PhD’s formed the basis
for establishing
Dina Reseach School

which quickly became one of the most successful activities of

including the Nordic collaboration it has initiated. Danish co
mputer science has no tradition,
like there is in for example statistics, for creating courses for PhD students in the biological sciences,
such as agriculture. Dina Research School has by now broken new grounds with this type of courses.

When enumerating

local benefits and efforts

at the seven member institutions, it is natural
to state first that Dina has been no less than decisive for the creation of a strong research group at

(Foulum) in the fields of biometrics and informatics. The process pro
fited from fruitful links to
research groups in informatics at the universities, primarily at KVL and AAU but also at AU. Much


work was done in Decision Support Systems where, at an early stage, the particularly succes
ful pr
ject Pl@nteInfo demonstrated t
he potentials of the Internet for agricultural advising.


(Copenhagen), computer science research hardly existed before 1991. Dina helped fil
ing in this gap, in close collaboration with DTU. Also at the educational level, KVL has profited su
ally from this cooperation. Another educational initiative taken by Dina is the project Biomo
els which has received support from the Danish Ministry of Education and has been carried out at
KVL 1997
99, with quite some success. With regard to specific re
search projects, Dina KVL has
cooperated with FSL in statistical image processing, with results that reach the international leading
edge in forest inventory; the plant
atmosphere model

was thorough
ly repr
grammed, thus
creating a platform for
professional applications; and decision support in animal production was d

The research groups in statistics and computer science at

(Aalborg University) provided
strong expertise in probabilistic decision support systems to agricultural app
lications. The AAU
groups also created the successful system / company Hugin A/S; Dina can hardly take any credit for
that but has benefited greatly from Hugin. Several scientists trained at AAU are now working at other
Dina nodes; and one of the first Din
a PhDs is employed at AAU.

Dina, in particular the creation of Pl@nteInfo, helped in giving

(Skejby) an early start in
ing smoothly to the rapid development of IT, the Internet and the web. Dina projects formed
basis for improvements of model
s behind DAAC’s advisory systems.

From the start of Dina, computer scientists at

(Lyngby), in particular A.P. Ravn and J.
Madsen, were instrumental in basing the Dina activities on sound computer science
principles. It is fair to say that
Dina DTU has up to now yielded more to cooperation in Dina than
benefited from it but it is intended to change this situation in the near future.


(Hørsholm), the main impact of Dina relates to forest inventory using statistical image
processing, cf
. the above para on KVL. However also the collaboration on Internet advisory systems
is promising.


(Roskilde) is a newcomer in Dina, and it is too early to evaluate the effect of the cooper
tion, but it seems most promising in view if the importance
of agricultural applications of bioinfo
matics and quantitative genetics.

Let us mention some aspects of the general
network and collaboration

activities which have
resulted from Dina.

The organisation’s working practice has been an example of early pro
motion of
electronic infrastructure
, video
conferencing and extensive use of the WorldWideWeb.

Dina has
been active at the

level, among other by giving a vital contribution to the creation of the
European network organisation EFITA.

By se
lecting a number of
priority research areas

and pr
curing them with associated
IT coordinators
, Dina has from 1997 contributed even more to tighte
ing and targeting the over
all Danish effort in IT in agricultural research and development.

Dina has
a platform for establishing
research project proposals

and for applying for funding for such
projects. Some of them were obtained by Dina directly, others were helped along. In later years se
eral major applications were submitted by Dina or in relation to


To fill actual needs but also
to enhance Dina’s function as a meeting point for scientists from different backgrounds, Dina has
organised regular

thematic days and workshops

as well as occisional courses; many people have
taken part in these arran
gements and it is the general impression that they have been appreciated.

One reason for Dina to claim the status of a pioneer organisation is that ‘Informatics in agr
ture’ does not exist as a recognized scientific discipline

the expression is merely

a label, notably
used by Dina, to denote a meeting point for collaboration or transfer of knowledge between disc


plines. As such it is important, but the quality of research depends heavily on the participants being
scientifically active and recognized in

their respective disciplines, be they agriculture or informatics.
This is the reason why relatively few Dina related papers in the area were published in international
journals in the first years. But the situation is improving; besides the Dina seniors i
n general have a
good record of refereed publications. The number of Dina based PhD dissertations should also be
noted; see Tables 4 and 7 below.

Dina should continue

All along the period 1991
1999 it has been understood that Dina might be only a temporar
y ph

that after a certain period, when IT and agriculture had reached a really good level of cooper
tion with mutual benefit, then it could be time to decide that Dina has had its day and can now close
down its activities. When drawing up the bal
sheet, like we do in this report, it is appropriate to
ask quite bluntly: Should Dina continue into the 21


Since there is a general agreement that Dina Research School is a valuable activity, a more rea
istic phrasing of the question would

be: Should Dina close down its activities as a general network
and continue only the research school, with a minimal background organ

Such a decision would have a number of
negative consequences

. When a network and a forum for organized inter

such as regular workshops and
thematic days

between scientists do not exist anymore, it is less likely that new achievements in
informatics and new demands in agriculture which, when combined in cooperation, could have a vast
potential, are dete
cted and explored in time.

. Dina Research School, including its newly established Nordic network funded by NorFA,
will suffer serious damage. Even if the research school were given a more permanent existence within

NOVA University, it will hardly be abl
e to survive without the backing from Dina, including the m
tivation and the direct scientific support from university informatics environments. It is possible that
the new IT Universities in Copenhagen and in Western Denmark will some day develop courses
informatics for biology oriented PhD students but they will hardly place a strong emphasis on agricu
tural and agro
environmental subjects.

. Locally, at the seven nodes, various losses can be foreseen.

At DJF, the lack of strong links
to KVL, AAU a
nd other universities will make it harder to recruit scientists to the biometry and i
formatics research groups in Foulum, and probably also to develop IT applications in Flakkebjerg,
Bygholm and Aarslev, even if the impact of Dina these three places has b
een limited.

research group in computer science may have to close down and it can become necessary to ask the
IT University of Copenhagen to take over teaching and PhD education in IT related subjects for KVL
students. New projects and the produ
ction planning / decision support group in animal science at
KVL will get less chances to draw on computer science expertise at other universities.

other sectors can provide applications for the statistics and computer science groups and challeng
new theoretical developments, but AAU would still benefit from taking agricultural applications and
challenges seriously, and with Dina this is more likely to happen.

Whether Dina exists or not,
DAAC must spend much money and manpower on developing IT

and Internet
based advisory sy
tems, and DAAC can buy the expertise they need. But without Dina it becomes less easy to provide a
smooth link to IT and informatics environments at the universities.

The question for DTU will be
whether a closing of Dina
will reduce chances of developing agricultural applications and of suppl
ing the engineering sciences with agricutural and environmental challenges.

Without Dina, FSL


will be forced to find new ways to provide expertise and collaboration in IT and modell
ing, in parti
ular to support PhD projects.

Since Risø has just joined Dina, stopping Dina would hardly be a di
aster. But promising poss
bilities would be cut off, just before they were to be tried out.

The overwhelming communication capabilities in th
e information society is an example of a new
technology which makes it possible to interpret biology in new ways. This is what Pl@nteInfo has
achieved in the area of plant pest and disease forecasts

it has applied existing knowledge exhau
ively and has
initiated new research, and the results are tailored to serve the farmer smoothly online.
Another example is bioinformatics, a new discipline created to meet the call for specialised info
ics in gene technology. Such challenges will continue to arise an
d to create a permament need for
collaboration between biologists and informatics experts.

To conclude, there is ample evidence that Dina will still be needed well into the next century. The
Dina board came to the same conclusion in October 1999 and decid
ed to recommend to the member
institutions that Dina be continued at about the same level as now for at least three more years, that is,

to 2002. In the following we outline the conclusions for Dina’s future activities.

The future Dina

Dina should contin
ue. And there is no need for any drastic changes in the organisation nor in the f
nancial basis of the Dina Network

which is certainly not the same as to say that nothing new will
happen at the profe
sional level.

research school activities

are pra
ised from all sides; they should continue essentially as
now, with workshops and summer courses, and might even be further developed. One development,
which would give a still better basis for Dina Research School, would be at hand if a basic group of
students were funded directly by the school and were involved in Dina projects. Ideally the r
search school activities

workshops and summer schools

should have 50% such Dina PhD students
and 50% other PhD students as attendants.

Apart from contributin
g to the Biomodels project, Dina has up to now been only little engaged

at a pre
PhD level. Dina might in the future consider the possibility of involving more
master and bachelor students in its activities. Among the possible areas for joint
educational action,
the following could be mentioned: Decision support systems; Biomodelling (mathematical models);
Stochastic modelling in agriculture; Bioinformatics in agriculture; Engineering

and IT
technology in
agriculture; Internet based advisory s

general network

of Dina works smoothly, and there is no reason to change its basic fram
work including the system of priority research areas and IT coordinators (though the set of specified
areas can of course change, coordinators can stop an
d new can begin). Neither should Dina leave the
practice of arranging workshops and thematic days from time to time. It is the ambition that the yea
ly general workshop or seminar is established as an attractive meeting point for scientists from the
ed disciplines and for everybody else who is seriously interested in the use of info
matics in

Among the
local visions

at the Dina nodes, let us mention a few.

Dina DJF

the develo
ment phase of the successful outward
looking activity Pl
@nteInfo can now be considered carried
through, and it is time to focus more on identifying scientific project results and data sets that can be
enlightened by using Pl@nteInfo. Also Dina’s interaction with other DJF departments than those at


Research Cent
er Foulum are to be developed, to make scientific collaboration more frequent.

Dina KVL

there are several areas of possible new collaboration that could be looked into.

students at KVL

are generally well aware of the importance of statistics,
the visibility of informatics
and mathematical modelling in the curriculae should be enhanced.

It is intended to involve the
ly competent experts in
Dina Aalborg

even better and more directly in research interplay with
agricultural research. The same

goes for experts in computer science at KVL and in mathematical
modelling at DTU; however these experts are already of crucial value to Dina Research School.

It is to be expected that Dina will in the future place less emphasis on leader
induced project
applications. Also there are no plans of expanding the role of the Dina website which will rather be
used mostly as a means of communication within Dina itself. And it is intended to reduce the admi
istrative costs of running Dina as much as possible, with
out compromiszing the principle that all a
tivities be visible in the budget.

Finally Dina should continue its contact and exchange of ideas with organisations like DSIJ and
EFITA and should generally strengthen the international dimension, notably in a N
ordic and a Eur
pean context.

Budget implications

The basic principles of Dina, as expressed in the bylaws par. 8, state among other that all common
Dina activities must be funded in a visible way; that each member institution commits itself to co
e Dina; and that the contributions from the 7 member institutions are to each other as follows:
AAU: 1, DJF: 4, DTU: 1, FSL: 1, KVL: 4, LR: 1, Risø: 1.

If Dina continues with essentially the same activity, a total budget of roughly 700 000 DKR
can be for
eseen, corresponding to a member contribution unit of about 50 000 DKR.



ODINA: Probability at work

How scientists created a new company

Fig.: Simpel HUGIN
skærm med model

When Dina started in 1991, a marriage took place between the AAU
based ODIN project and
Dina. The result was the ODINA cooperation. ODIN was about Bayesian networks, a formalism for
the construction of decision systems to treat problems characterised by reasoning under uncertainty.
The uncertainty expresses itself clear
ly in agricultural connections where results of actions and
decisions are frequently of a stochastic nature.

The ODIN group’s work was, and still is, focusing on methodology, and the cooperation in
Dina offered an evident opportunity to try out the method
s that had been developed, in new areas of
application. The cooperation proved most fertile and resulted among other in a system for
determining the blood type of Jersey cows, BOBLO; a system for supporting mildow control in
winter wheat, MIDAS; and a syst
em to assist pig production planning (See

From Artificial Intell
gence to Decision Support Systems
). These prototypes have given a considerable insight into the


problems connected with the construction of applicable systems, and they have also contributed
understanding better the problem areas in question. The experiences that have been made form basis
for the project proposal ”The Farm Animal Information Society” which is hoped to result in an
integrated practical system for support in the different pha
ses of pig production.

A common trait in these various applications is that they are based on HUGIN

environment for developing and executing Bayesian networks. HUGIN was created by the ODIN
group, and the continuing development of HUGIN benefits from

the inspiration that came from the
cooperation in ODINA. It has led to several publications written in cooperation, among these a
popular account of Bayesian networks in the Danish magazine Naturens Verden (vol. 7, 1998).


The European


Fig.: Underskrifter på EFITA stiftelsesdokument

In 1995
97 Dina coordinated the concerted action EUNITA (European Network for Information
Technology in Agriculture) funded by the European Union. The objective of EUNITA was to
enhance the use of

informatics throughout Europe by the efforts of ten working groups who gathered
information and experiences on various subjects. The results of the working groups are published in a
number of reports by the participating institutes and organisations.

e of the working groups was concerned with the creation of a European association for IT
in agriculture. The association was founded in Wageningen in 1996 and was named EFITA,
European Federation for Information Technology in Agriculture, Foods and the Env
ironment. The
members of EFITA are national organisations for IT in agriculture. In Denmark DSIJ, The Danish
Society for Information Technology in Agriculture, was created for this purpose closely related to
Dina. By Sept. 1999, 11 countries are members: D
enmark, France, Ireland, Italy, Georgia, Germany,
Greece, Great Britain, Spain, Sweden and The Netherlands. EFITA held its first conference in
gen in 1997 and the second one in Bonn in 1999. Iver Thysen was president of EFITA from
1997 to 1999.


Daisy: Modelling the plant
atmosphere system

Fig.: Foto af Charlottes forsøg, eller standard Daisy

The purpose of the original Daisy model was to simulate water and nitrogen dynamics in
agricultural fields as influenced by agricultural
practices. That Daisy is a well
functioning model has
been confirmed in a number of successful comparative tests.

In later years, Daisy has been transferred into a new code which has several advantage to the
original one: It allows for in
built alternativ
e process descriptions so that for example plant
competition can be taken into regard; it gives room for new water flow models which makes it easy to

add new processes to the model, notably to simulate the fate of pesticides and other agro
chemicals in

atmosphere system; and it provides an interface to other model systems, in particular
the hydrological catchment model MIKE/SHE, a most interesting feature since MIKE/SHE forms the
basis of the ”DK Model” developed by GEUS to estimate Danish wa
ter resources better. In addition,
the new code supports distributed use of the Daisy model.

The new features have proven valuable in ongoing research, e.g. the projects ”Pesticides and


Groundwater” under The Danish Environmental Research Programme, and ”
Remote Sensing based
Crop Simulation and Soil
Transport Modelling” under The Earth
Observation Programme. The code architecture efficiently supports the development of new process
ons and adds new functionality to the model.

The project has shown the value of using new and sound computer science principles. The
implementation of the generic framework for adding alternate process descriptions has only been
possible thanks to the financial support from Dina. The framework is an

early adoption of software
engineering research and is as such risky and inappropriate for funding through the more
oriented projects. However such funding was provided for the actual implementation of
alternate process descriptions, and this
helped in testing the Dina
financed framework.



Interactive extension services

Fig.: Skærmbillede med radar mv

If a plant grower wants to cultivate his land in an environmentally friendly way, he needs l
ots of
date information about such factors as the weather, pests, and pesticides. He needs to adapt this
information to local conditions and utilise it to make the right decisions as to whether or when to
spray, and the doses to apply.

With Internet

access he can obtain help from the Pl@nteInfo (

) system,
which provides information and decision
making support to farmers and consultants. Pl@nteIn
collects, collates, and communicates up
date data on plant growing; it enables a
farmer to
download local climate data from the Danish Meteorological Institute and to use it along with
information from his own fields to calculate the risk of attacks by a particular pest at any given time.

Pl@nteInfo processes raw data from various so
urces and presents the results in graphic form.
Via a subscription system help can be tailored to the individual farmer, because the system knows
where he lives, and it has details about his land. He can even add assessment by a local expert. The
result of

all this is that the farmer receives good, useful help so he can make the right decisions.

Pl@nteInfo was initiated in 1996 when the Internet and WorldWideWeb were still exotic
phenomena for most people in agriculture. The visions for use of the Interne
t in agriculture that
initiated Pl@nteInfo have now been realised. These visions were, however, so apparent in the first
version of Pl@nteInfo that the agricultural advisory system immediately responded

and started to
move towards the information society
. It took a great effort to change structures and working
methods in this large organisation, but the Danish advisory system is now at the leading edge of using
the Internet in agricultural extension.

At present Pl@nteInfo has more or less incorporated al
l available knowledge on plant pest and
disease forecasting; therefore new research is being initiated to improve decision support in the area.
At the same time other suitable application areas for online advice are being examined.


Spatial tree patt
ern analysis from aerial photographs

Fig.: Skov
luftfoto med billedbehandlingsspor

High resolution aerial photographs of a forest can, in the hands of foresters, statisticians and
computer specialists, give surprisingly precise answers as to the number o
f stems in the forest stand,


the positions of the single trees, health status, the estimated volume of wood and other properties of
interest for forest and landscape monitoring. One key project in Dina has been to develop such
methods, building on the theo
ry of spatial point processes which has been growing vigorously in later
years. A joint effort between KVL and FSL, the project has surpassed even optimistic expectations
and brought Denmark into a front position in the area. The visible expression of this

consists of two PhD theses and a series of papers in international journals by Dina researchers: K.
Dralle, M. Larsen, J. Lund and M. Rudemo.

The number of stems can be estimated by smoothing the digitized aerial photograph by a so
called Ga
ussian kernel which can furthermore be used to find tree positions. Another method based
on an optical single
reflection model allows for an accurate analysis of images obtained from varying
viewing angles. Observational errors and simulations to support t
hese forestry applications have also
been studied in detail.

The methods developed so far mostly pertain to homogeneous forest stands where all trees are
of the same species and the same age. To turn the methods into a useful tool for monitoring forests
a regular basis, say from aerial photographs every fifth year, or from the new generation high spatial
resolution satellites, further improvements are needed. In the near future the Dina group will
investigate the problem of mixed forests and also focus

on spatial analysis of multiple photos.


From paper to the Internet

Fig.: skærmbillede fra LR's systemer

"The Web Changes Everything" has become a popular term these days. Albeit true only to some
extent, facts are that the web has changed and will

change a lot within the agricultural extension

Back in 1996 the Danish Agricultural Advisory Centre decided to begin publishing its
production of papers and handbooks online. Since then, more than 20,000 documents in addition to a
number of onli
ne applications have been published

and much more is on the schedule. Dina helped
jumpstart this process and has been an important player all the way up to now.

As the possibilities and complexities of the web are exploding, the need for collaboration o
content as well as technology is as large as ever. In the future, a key issue will be to serve the adviser
or farmer with information and decision support, precisely tailored to his needs. To fulfill this goal,
data and knowledge from many sources need t
o be collected and combined in new ways. A
challenge? Sure, and we are ready to face it!


KultuNaut.dk: A Danish culturel network and event calendar

Fig.: Skærmbillede fra web & citat Lone Dybkjær: ros til KultuNaut

At a very early stage, before In
ternet and Web were really known, a student assistant in Dina
(M. Gudmand
Høyer) through his work became aware of the technology and the opportunities of
collaboration over the Internet, and he used this experience to develop a web calendar:


Over the years 1995
99 the calendar has become the largest event database in Denmark. Based
on software technology from Dina, the calendar allows for a collaboration and exchange of data with


other cultural prom
oters. Local authorities, counties and tourist bureaus support the system, and today
the calendar system is used in adjusted versions that may easily be integrated into other home pages,
among these several of the largest Danish web

KultuNaut com
bines the Internet database with other media, and for example the newspaper
Politiken's weekly “In
Town” section is printed from the KultuNaut database. Besides of being a
successful web site, KultuNaut is distinguished by exploiting the Internet for a new

type of
collaboration to maintain and distribute data between many kind of organizations in different parts of
Denmark that were not able to collaborate before.

(Figure caption:

More than 50 websites make ther own calendar from KultuNaut, e.g. with ev
from some local area. The figure shows two examples of integration of the database into other
websites and other layout. DR
P3 (top) presents a music calendar while Uland (bottom) focuses on
events related to Third World countries.


From Artific
ial Intelligence to Decision Support Systems

Fig.: figur fra BOBLO fælles med ODINA.

Optimal agricultural production management relies on close observations of animals and fields
and, if problems arise, on a timely reaction based on expert knowledge. To
assist this process, ma
agement information and decision support systems have been used for some time

but not always
with sufficient quality and capacity.

Artificial intelligence seemed an interesting option to improve the dissemination of expert
dge. From the start the so
called rule based expert systems were promising. Here the domain
expert is modelled by elicitation of the rules he or she uses to reach a solution. However, it soon b
came evident that such systems were of limited use, because th
ey could not adequately handle the
uncertainty in agricultural production.

Then techniques related to the Bayesian network methodology came into focus. Instead of
modelling the expert, his knowledge of the domain (e.g., animal production) is used in the m
The uncertainty is an integrated part of the model, and observations can be directly incorporated to
improve the decisions. Promising prototypes included bovine paternity testing, optimal winter wheat
management, and mating management of sows. These

prototypes were met with a certain amount of
scepticism, but indicated the potential of the techniques.

Currently, new agricultural applications are on the verge of being implemented for decision
support within sow and slaughter pig production. Furthermo
re the techniques have become an int
grated part of several research projects related to production management. The agricultural applic
tion areas have indicated methodological limitations. New developments to solve these problems are
sought in the Dina co
llaboration merging the expertise of researchers with different theoret
cal bac


Biomodels: Teaching biologists how to model

Fig.: Klip fra MAPLE
udskrift med figur

In 1995 the Danish ministry of education launched a programme to back new in
itiatives in
supported education. Dina suggested that KVL’s Dept. of Mathematics and Physics


(IMF) should formulate a project and apply for funding. A three
year project, Biomodels, was drawn
up, and IMF received a grant of about 1 mio DKR, w
hich was supplemented by KVL and Dina with
roughly ½ mio DKR each to cover the essential expense: the salary of 2 assistant professors (one in
mathematics and one in data processing) and ½ computer technician in 3 years.

As the project period (1997
99) is

about to run out, we can look back on a successful and i
spiring educational experiment. Two new courses were designed, Biomodels G at bachelor level and
Biomodels O at master level; they have run 3 and 2 times, respectively, with 6

15 participants. M
le V was chosen as computer system and the courses have made extensive use of IT facilities like
WorldWideWeb and e
mail. Teaching has been problem oriented, with students working in small
groups or individually on mini projects connected with their agricu
ltural interest; reports were handed
over as a Maple worksheet and made accessible on www. Course evaluations have been enthusiastic,
and there is no doubt that the courses have opened the eyes of the participants for the possibilities of
IT and modelling.

Another ambition in the project was to create a number of hands
on modules for use in co
supported supplementary teaching. A good start was made on this task and it will be followed
up after the project period has ended. Also the course Biomodels O

will prevail as a permanent activ
ty at IMF.

A variant of Biomodels O was arranged in NOVABA setting as an intensive one
course for Baltic agricultural students in Kaunas, Lithuania in Sep. 1998.



SFD project

A successful failure


Klip fra SFD
model og dens repræsentation på web'en

One of the starting points for Dina was a database project. In the late 1980’s, databases had
become a buzz
word in agricultural research project applications, and the Ministry of Agriculture
wanted pr
ofessional computer scientists to be involved, in order to secure the future. However it is
not quite easy to secure anything in the IT world; all the same a project was established in collabor
tion between DJF (SP+SH), KVL and A.P. Ravn at DTU, in order t
o create a formalism that allowed
for a system to handle datasets along with the specification of the data structure. Input and output
could then be automatically organized, in turn allowing for individual datasets to be stored in ASCII
format in a self
plicatory way.

Such a system was in fact constructed and, in connection with the EU project ADDA, tested in
relation to a study of field trial datasets. Also, facilities to automatically present the system and the
data on the internet were developed.

fortunately it turned out that nobody wanted to use the system in practise. But the many di
cussions about the project built up awareness of data discipline and created good and fruitful contacts
between agricultural research and computer science environme
nts, in particular at DTU. The ministry
was ready to further develop these contacts when AI

artificial intelligence

became the new buzz
word. This initiated the Expert system project, and before long Dina was established by the Danish
Research Councils

Thus, seemingly a failure the project has had a quite useful outcome.


Moscow ML

A worldwide success


One outcome of Dina’s computer science activities which has no relation to agriculture is the
creation of
Moscow ML
, a light
weight computer imp
lementation of the data processing system Core
Standard ML. Initiated in 1993, Moscow ML has been developed continuously by P. Sestoft, A
soc.Prof. in data processing at KVL and expert in Dina Research School, in cooperation with S.
Romanenko, The Russian
Academy of Science. The system has been distributed from

since Sept. 1994 and over 10,000 copies have been downloaded from more than 40 countries. Mor
over P. Sestoft has participated in the design, imp
lementation and testing of the Standard ML Basis
Library which is a joint effort of the three main Standard ML computer implementations.