Agriculture Reference
In-Depth Information
on-the-go control of field operations for which just one soil property is very
significant. Examples for this are controlling the sowing depth according to soil
moisture or the application of pre-emergence herbicides in proportion to organic
matter in the soil. All first attempts of online and on-the-go controls for site-specific
operations have been based on single soil property sensing concepts.
However, the better the knowledge about the agronomic effects of soil properties
is and the more the technology of spectroscopic sensing advances, the more favor-
able the terms get for multiple soil property sensing. This concept corresponds to
the notion that some field operations probably require site-specific adjustments that
simultaneously are based on several factors. For example: controlling the sowing
depth could be reasonable not only according to soil moisture in the vertical soil
cross-section, but also proportional to texture. And for the application of pre-
emergence herbicides, the control could be based not only on organic matter, but
again on texture as well.
There is another point that supports the multiple soil property concept. It is the
fact that several soil properties do not exist independently from each other. In many
cases, the organic matter content too depends on the texture. For instance, the water
content firstly depends on the weather. Yet following precipitation, what remains in
the topsoil and is available to a crop depends on texture and organic matter as well.
The cation-exchange-capacity relies heavily on texture and organic matter too. So
there are many interdependencies among soil properties.
An important question is, on which wavelengths the sensing should be based on.
There are two general approaches for this: the full spectrum approach or the discrete
waveband approach.
A
full spectrum approach
means that within the spectrum chosen (
e.g.
visible
and near-infrared range) practically all wavelengths are included in the sensing pro-
cess by recording in steps of for instance 5 nm or even less. Modern spectro-
radiometers can do this within a fraction of a second. Hence at every site-specific
spot, many signals are sensed. These signals are subjected to sophisticated statisti-
cal evaluation processes like partial least squares regressions or others in order to
obtain information about soil properties.
Since a full spectrum can include the important ranges or wavebands of several
soil characteristics with their specific “fingerprints”, this method is principally
suited for
multiple soil property sensing
. Such multiple soil property sensing and
mapping is close to becoming a reality for a variety of site-specific farming opera-
tions (Lee et al.
2009
; Viscarra Rossel et al.
2006
).
The
discrete waveband approach
dispenses with signals from a wide spectral
range and just is confined to using narrow key wavelength bands. Much effort has
been and still is devoted to detecting these
key narrow bands
. This sensing method
can be reasonable for estimating just one soil property,
e.g.
water or organic matter. In
many cases, interdependencies among soil properties are not taken into account. This
might not be necessary, if
a priori
the correlations between the respective properties
and the key reflectances are high, as it is the case with water and organic matter.
Experience will show, which properties can successfully be sensed via full
spectrum approaches or when discrete waveband approaches are reasonable. But
