Agriculture Reference
In-Depth Information
red band (660 nm, minimum) and near-infrared (750 nm, maximum) (Weis and
Sökefeld
2010
).
Vrindts and de Baerdemaeker (
1997
) as well as Biller (
1998
) used spectrometers
to detect weeds between the crop rows or before sowing and after harvesting the
crop by measuring the refl ectance in the green, red and near-infrared light wave
bands. Green leafs were characterised by a high refl ectance in the green and near-
infrared and a low refl ectance in the red spectrum compared with the refl ectance
curve of bare soil. A few commercial products for weed control with optoelectronic
equipment exist that use this spectral information;
e.g.
DetectSpray® (evaluated by
Biller
1998
) and WeedSeeker® (used by Sui et al.
2008
).
10.2.2
Fluorescence Sensors
After exposing green plants with radiation for a specifi c amount of time, leafs emit
radiation of a longer wavelength as the excitation light. The intensity of this radia-
tion named fl uorescence highly depends on the leaf properties and on the physiolog-
ical state of plants (Cerovic et al.
1999
).
UV-induced
chlorophyll fl uorescence
has also been applied to discriminate
plant species based on the characteristic leaf structure. Longchamps et al. (
2010
)
measured a range of fl uorescence spectra of maize, grass-weeds and broadleaved
weeds under greenhouse conditions with natural illumination. They classifi ed the
three plant species groups based on their distinct spectral signatures with a recogni-
tion rate above 90 %. Tyystjärvi et al. (
1999
) developed a method called fl uores-
cence fi ngerprinting with which leafs are exposed to a series of different spectra and
intensities of light to record changes in the fl uorescence of chlorophyll a. The emit-
ted light curve could be used to identify plant species with an accuracy of more than
90 % under laboratory conditions. Later, Tyystjärvi et al. (
2011
) applied a similar
approach under fi eld condition and achieved 90 % recognition in maize and weeds
when plants were shaded for 1 s before measuring.
In our working group, the MiniVeg® sensor (Fritzmeier Umwelttechnik) has
been used in fi eld and greenhouse studies to map the spatial distribution of weed
species in arable crops. Red- plus far red fl uorescence was induced by a red laser.
When the laser hit plants, fl uorescence was induced and recorded in the processor
of the sensor. Due to the high frequency of measurements (500 s
−1
),
plant density
highly correlated with the number of hits. As crop density was rather homogeneous
within the fi elds at early growth stages, variations of the hit-number correlated with
the weed density. Therefore, weed distributions maps could be derived from the
sensor measurements when a GPS-receiver was mounted on the sensor vehicle
(Fig.
10.3
). Blackgrass (
Alopecurus myosuroides
Huds.) was the dominant weed
species in the winter wheat fi eld sampled in 2008. Forty-four percent of the area
remained untreated when site-specifi c weed control was applied in this fi eld using a
weed control threshold. The MiniVeg®-sensor provided 75 % correct decisions
compared to manual weed countings.
