Agriculture Reference
In-Depth Information
It should be noted that reliable semivariograms cannot be obtained from fewer
than 100 data. The nugget to sill ratios of the standardized default semivariograms
should match those of the respective fields (Kerry and Oliver
2008
).
The term “standardized semivariogram” should not conceal the fact that in real-
ity the curves for the semivariances are rather unique and different for every field.
Yet precise sampling and site-specific management relies on the semivariograms for
information about the cell-size that should be used. If the cell-size is oriented at less
than half the range (Kerry and Oliver
2004
,
2008
) as outlined above, what ranges for
soil- and crop properties do actually occur?
Actually, the
ranges
that have shown up in research with semivariograms vary
immensely - as do the local conditions around the globe. In extreme cases, ranges
on the one hand as short as 1 m (Solie et al.
1992
) and on the other hand as long as
26 km (Cemek et al.
2007
) have been recorded. However, the vast majority of
ranges for soil- and crop properties is between 20 and 110 m (McBratney and
Pringle
1999
). So for most cell-sizes, the upper limits of side lengths should be
between 10 and 55 m. This is still a rather wide span. Therefore, to define the
actual need more closely, a careful deduction of required cell-sizes from suitable
standardized semivariograms is necessary.
2.4
Processing and Adjusting the Resolution
Finding the appropriate upper limit of cell-sizes deserves some effort. This is
because knowing about it is important on several stages of site-specific farming,
first when sampling, then for mapping and finally when the machinery operates in
the field. It is obvious that the sampling must occur within the distance limits defined
by about half the range. But the same holds for techniques used to make the maps
and finally for the cell-sizes, on which the farm machinery works in a site-specific
way. If on any of these stages the distance limits that are defined by less than half
the range of the respective semivariogram are exceeded, the precision of site-speciic
farming is impaired.
The question is, at which stage - sampling, mapping or machine operations -
striving for small cell-sizes is most difficult. As long as sampling of soil properties
and of nutrient contents is done in a manual way, this will be the sampling stage. In
the long run, however, manual sampling will be more and more replaced by online
and on-the-go sensing methods. Many of these methods will allow for sensing of
small cell-sizes. As a consequence, then the cell-sizes that can be realized with wide
farm machinery become important.
It is not recommended to directly combine fine grid spacings for sampling or
sensing on the one hand with a much coarser resolution for the machinery operations
on the other hand without any signal corrections. This is because the control of the
machinery is less erratic and is more stable if
averages
of highly resolved signals are
used. If the machinery is controlled via online and on-the-go sensing, this averaging
step can easily be implemented into the processing computer program. In case a field
