Agriculture Reference
In-Depth Information
-
eloquently put, in general terms, by Bourke (1953)
“Even the finely evolved model
can scarcely avoid the defects of over-simplification and over-rigidity, over-
simplification because the complexity of the phenomena involved cannot be
precisely reflected in any easily handled formula; over-rigidity because, even if the
criteria are to be in a form capable of objective application by a number of workers,
they must introduce abrupt, and to a certain extent, arbitrary discontinuities which
appear in nature only as gradual transitions. This does not mean that the evolution of
a good working model would not be of considerable value...”. The inclusion of
husbandry factors, as proposed for the Danish system of sclerotinia risk forecasting,
or of eyespot, by weighting each factor empirically, gives some confidence that the
system might work as it does involve the crop and the field history, plus an
assessment of potential inoculum pressure. Likewise, testing petals of oilseed rape
for the presence of spores and the percentage that are affected gives an indication of
potential infectivity (Morrall and Thomson, 1995). Computer models, such as
NEGFRY (Hansen
et al
., 1995), which can be modified to take account of local
factors are an improvement but still include a level of subjectivity. A summary of
some of the constraints are covered by Zadoks (1984).
9.9.2 Equipment
The move away from the synoptic network to the use of in-crop monitors to bring
more precision to forecasting imposes its own problems. What is the accuracy of the
data capture equipment, how regularly it is calibrated, what is the seasonal drift and
where is it sited? All these have a bearing on the likely success of the model.
Sensors for determining relative humidity and leaf wetness are perhaps the two
that are most likely to be subjected to inaccurate readings. Manufacturers generally
only guarantee RH sensors to + or - 2 % accuracy. What chance is there of triggering
the 90% RH required for many of the potato late blight forecasts? Likewise, leaf
wetness sensors are prone to false readings when exposed to the detritus blown onto
them from the crop. Regular checking of the sensors is important but difficult if the
equipment is remote from the user. The problem of continuous reading when
failures occur due to battery life, solar panels becoming obscured and problems with
the transmitters or cellphone networks all impose potential problems and poor
delivery from single sites (Hims
et al
., 1995).
Where should the equipment be sited in the field, near the farm building for ease
of access or in an area thought to be more representative of the local climate? How
valid will the forecast be with only one set of sensors? With a disease like potato
blight, where precision is important, where in the field should the equipment be sited
to give the most reliable reading if at one end there is a wood and the other a lake or
river? The questions begs the answer that we may be looking for a precision that is
not attainable. It is possible that Large (1956) had the answer, that it is the intelligent
interpretation of a network of stations that is important. A network obviously has the
advantage in that failure, or inaccurate readings, from one or two sites will not fail to
trigger a general warning (Smith, 1956).
