Agriculture Reference
In-Depth Information
patches or areas than fungal infections in their initial stages do. Hence for sensing
fungal infections adequately, much smaller optical fi elds of view are needed. This
will be diffi cult to obtain with sensing from satellites. But for proximal sensing
from fi eld based machines, principally it is possible to provide the small fi eld of
view by a narrow angle of vision and by a short distance to the canopy. In experiments
with infections of yellow rust ( Puccinia striiformis ) and leaf blotch ( Septoria ) in
winter-wheat, Moshou et al. ( 2011 ) used a fi eld of view with 20 cm diameter on the
canopy level for sensing from a sprayer-boom. However, small fi elds of view must
be accompanied by many records in order to scan a fi eld adequately. Already
because of this, sensing for a discrete-spot effect will be more expensive than the
conventional refl ectance sensing for nitrogen fertilization.
Another point to consider is the fact that the top leaf of a plant in crops generally
is not infected because it has only recently unfolded and incubation periods for the
fungi hold (West et al. 2003 ). Hence in fi eld operations, the viewing should be such
that the top leaf is omitted as far as possible. Since with most crops - especially with
small grains - the top leaf initially has a vertical position, oblique view directions
towards the canopy should be avoided. With vertical viewing directions, primarily
the horizontally oriented leaves get into the fi eld of view, and the recently unfolded
new vertical leaves affect the results less.
11.3.2
Spectra and Indices of Refl ectance
There are many different types of fungal infections that can occur and these may
affect the refl ectance in different ways. Even within the same crop, different fungi
can cause different refl ectance spectra . This can be seen when the spectra of three
different fungal infections of sugar beet leaves - cercospora leaf spot, powdery mil-
dew and rust - are viewed (Fig. 11.3 ). The refl ectances of the artifi cially infected
leaves were recorded in a laboratory with increasing severities of the diseases
(Mahlein et al. 2010 ).
The courses of the spectra for the non-infected leaves correspond to the known
pattern within the visible and near-infrared wavelengths (Fig. 6.3 ) . However, the
deviations from these courses that result from the infections are different for each
fungus. The powdery mildew infections cause rather uniform increases in refl ec-
tance in the visible as well as in the near-infrared range. The deviations for the
infections of the cercospora and rust fungi do not comply with this in the range from
about 720 to 900 nm wavelength. Within this range, the infections by these fungi
result in decreases of the refl ectance. And an unusually steep rise in the red refl ec-
tance evolves from cercospora fungi for highly infected leaves. But it has to be
pointed out that these are results from one trial and its course of the refl ectance
could also be interpreted as the fact that e . g . 20 % of cercospora severity damages
the crop the same as 100 % rust.
For the sensing of many crop properties, the use of refl ectance indices has led to
success. Generally indices allow to limit the sensing to just a few, narrow
Search WWH ::




Custom Search