AN INTEGRATED GIS AND KNOWLEDGE-BASED DECISION SUPPORT SYSTEM IN
ASSISTING FARM-LEVEL AGRONOMIC DECISION-MAKING
Zongyao Sha* *, Fuling Bian*
* Center of Spatial Information and Digital Engineering, Wuhan University, Hubei Province, China, 430079 -
(zongyaosha, fulingbian)@eyou.com
Commission VI, WG 1 1/2
KEY WORDS: GIS; Knowledge; Representation; Digital; Agriculture; Decision Support
ABSTRACT:
A main task in agriculture production is field management of water and fertilizer. Excessive fertilizer and water application not only
waste resource but also pollute the environment. The traditional principle of digital agriculture can only apply in flat or plain place
where a large field usually should be partitioned evenly into many little grids and thus decisions on grid-specific agronomic
operation can be recommended. This mode of digital agriculture is called grid-level decision-making. But this seldom happens in
farming system in mountainous region due to the complex landform where even grids partition cannot be implemented and so this
mode is called farm-level decision-making. In this paper, we develop a web-based decision support system that integrates the expert
knowledge, analysis model and GIS to assist farm-level agronomic decision-making and that is fit for any circumstances in
agriculture production region with efficient knowledge support. The approach adopted involves general GIS spatial data
management (geo-referenced digital map, spatial agriculture decision unit, etc.), agronomic diagnosis and decision-making with
integration of expert knowledge and analysis model, so that the variable-rate application of water and fertilizer to any regular or
irregular cultivated field can be addressed. The core technology involved includes expert knowledge representation, model
organization, software data exchange standard and integration of GIS, expert knowledge and analysis model. With this approach and
the basic principle of the traditional digital agriculture, it is possible to tap the variable-rate water and fertilizer application to
agronomic fields even in the mountainous and remote region and gain maximum benefit with minimum purchased input, which is
very useful in mountainous countries with scattered and small-scale agriculture production. The framework of the developed system
is a hybrid structure model composed of B/S (Browser/Server) and C/S (Client/Server), which not only extends the capability of
decision support service space but also makes the system easy to maintain. A case study is done in Guangzhou city, located in inter-
tropical belt in the South of China and covered by mountainous landform, and shows an exciting result.
*
the actual need even GIS are capable of spatial data
management and displaying. Spatial models, as an import tool
to enhance spatial analysis ability, can be integrated with GIS to
solve more sophisticated and special problems, thus spatial
model integrated system, also known as decision support
system (SDSS), is widely researched and used in practice.
Many domain models are developed by specialists and used
widely in varied fields, including agriculture, environment
management. Digital agriculture uses irrigation or fertilizer
application model to give decision-makers how and how much
water should be irrigated or fertilizer should be applied and uses
GIS to express information about where and how much those
elements lack. In some cases, model should also be connected
with expert knowledge to solve semi- or ill-structured problems.
Those semi- or ill-structured problems are known as complexity
of decision tasks, difficulty in prediction of decision process,
multi stages and intersection of those stages in decision process.
Knowledge representation. is the key factor in knowledge
utilization and is an important focus in knowledge engineering.
Even Chinese farmers have accumulated rich experience and
knowledge in agriculture production (Guo, 1988; Chinese
Academy of Agricultural Science, 1993), they still lacks
scientific guide. With the development of economy, it is very
urgent to make Chinese farmers free from onerous agriculture
operation. Building a usable, widely suitable expert knowledge-
1. INTRODUCTION
A main task in agriculture production is water and fertilizer
management. Excessive fertilizer application not only wastes
limited resource but also pollutes the environment. Deficit
application, however, may limit the growth of crop. Due to the
large spatial variation of agriculture field environment (e.g. soil,
climate, terrain, etc.), location specific farm-level management
is critical in crop growth management. The choice of cropping
pattern, when and how much to apply fertilizers, water,
pesticides etc., the choice of the period for planting, plant
population density, etc., are all factors that should be considered,
for which the appropriate choice (associated with maximum
production or minimum risk) depends upon the soil feature,
crop feature, climate, besides social requirement. As farm field
is location-varied, different field should take different
management measures. Runquist et al. (2001) proposed a
compact, raster-based geographic information system designed
specifically field-level management, realizing numerical data
representing either circular (center-pivot irrigated) or irregularly
shaped crop fields. Agricultural operation is closely connected
with natural resources that have an obvious spatial character
which is considered essential character of Geographic
Information Systems (GIS) (Josef B., et al. 2002), thus GIS has
an important function to play in agriculture production, esp. in
field irrigation and fertilizer application. But it is still far from
* Corresponding author.
186
Internationa
feret
based and G
a great step |
We will use
spatial data,
integrated ai
agriculture |
problem of
give strateg
enhance pro
approach w
sufficient. e
casy-to-unde
executable 1
based decis
information
China, aidi
integrate de
functions in
system inte
Browser/Ser
2.1 Overvi
Expert syste
the knowled
problems or
fulfill a func
may play the
some functi
are then be 1
secure result
preexistence
calculation. I
commercial
scheduling o
The first DS.
Since the ear
the PC revo
hardware anc
upon definiti
based systen
models to sol
problems suc
and resource
definition of
agreed that S
which comt
algorithms a
and objective
vivid graph «
assess the in
clearly. Its pr
for interactiv
spatial. data;
relationships
including anc
analysis and 1
forms; and |
effectiveness
explicitly de:
decision-mak
analysis and
and flexible n
