Agriculture Reference
In-Depth Information
by illustrating the flow of information back from the receiver to the information
source. The part of the KT cycle which we have not shown is often concerned with
defining a problem which is to be solved and thus defining what the message is to be
in the part which we have shown. For example, the problem may be a practical
difficulty in knowing when to take action to control disease and the desired message
is information/advice that resolves the problem. Clearly, it is possible for uncertainty
(or noise) to enter the cycle of problem definition, message formulation and delivery
at several points and from several sources. We have indicated some of these in Fig.
12.2, particularly the role that those acting as the Information Source and
Destination play. As will become clear in the following discussion, the human
element and uncertainty are important factors in the success of any IT-based
forecasting system.
Figure 12.2. An adaptation of Shannon's (1948) schematic diagram of the process of communi-
cation illustrating the steps during which uncertainty might arise.
For practical purposes, a forecasting system must do two jobs. First, it must make a
useful forecast of disease outbreaks. Secondly, it must present the results of the
forecast in a useful format. Early in the development of IT-based forecasters the
primary objective was most often to develop tools that would give growers warnings
of disease outbreaks so that preventative action could be taken. More recently,
concern over the damaging effects of overuse of pesticides on the environment, and
a steady reduction in the profitability of crop production, have seen the avoidance of
unnecessary pesticide applications become an important objective for disease
forecasters. Irrespective of whether the aim of a forecasting system is to provide
warnings of disease outbreaks (what might be called positive predictions) or
assurances that outbreaks are unlikely (what might, correspondingly, be called
negative predictions) the basic performance criterion that any system should meet
was spelled out by Campbell and Madden (1990):
