Agriculture Reference
In-Depth Information
conductivity sensing for site-specifi c characterizing and delineating of soils in humid
areas probably is due to the fact that this method at least for sandy and loamy conditions
indicates the ability of providing water for crops. Yet a fl aw of electrical conductivity
sensing is that the results to some extent vary like those of a snap-shot. This is since in
humid areas the electrical conductivity senses texture as well as water of soils. And
because of the water, incidental rain affects the results. So the electrical conductivity
sensing aims at signals that indicate the ability of soils to provide long-term plant avail-
able water capacity. But this objective is infl uenced by short-term weather variations.
Yet if the site-specifi c water content is sensed in addition and separately, it might
be possible to remove this transient feature from the signals of electrical conductiv-
ity. Hence basically this method of sensor fusion makes very well sense. But
sensor
fusion
alone is not suffi cient. It must include
data fusion
. Because the control of
farming operations needs singular signals. But how should the data from electrical
conductivity sensing on the one hand and from the separate water sensing on the
other hand be fused to a common signal? The best mathematical combination of the
data from both sensors is not known yet and deserves efforts.
The fact that sensor fusion alone does not help much but that it should end up in
data fusion applies to many of the approaches that are listed in Table
13.1
. And here
too
logical mathematical concepts
for the respective data fusion are needed.
13.3
From Properties to Treatment - Map Overlay
Any control via mapping allows the farmer to inspect the sensed results, to com-
pare them with those from other records and to decide about the application of
diverse processing methods. Whereas with on-the-go control via sensor fusion the
processing of the signals is determined
a priori
, with mapping this decision can be
made
ex post
. The expertise of the farmer or of his adviser can still be included in
the control after the sensing of the data has occurred. Hence a more fl exible control
is possible.
With temporally stable soil properties, another feature is that maps can spare
sensing operations. Over the years, it might be advisable to create “maps of the
mean” in order to eliminate stochastic or incidental errors (Sects.
5.1
and
5.2.2.4
)
.
But once reliable “maps of the mean” have been obtained, sensing for this soil prop-
erty can be omitted. Instead, “
perennial maps
” can take over.
Any
data fusion in mapping
or any creation of maps of the mean needs a pre-
cise procedure in processing. The objective is to combine the information from
several maps and thus to improve the site-specifi c control of farming operations.
Visually, this objective is defi ned as “
map overlay
”. The aim of it is a single inte-
grating or treatment map for the control of farming operations (Fig.
13.1
).
The term “map overlay” makes it easy to grasp the objective. However, this
should not hide the fact that several logical steps must be considered in order to get
reliable results:
• Before any further processing of signals takes place,
data cleaning
might be
needed. Artefacts such as extremely low or exceedingly high values, which
