Agriculture Reference
In-Depth Information
actuators. The sensors are used to measure the variables of interest, such as crop can-
opy reflectance, plant height, and soil properties. The controllers collect and process
signals from the sensors and make decisions based on predetermined algorithms.
According to the decisions from the controller, actuators apply the input plants need
in real time
in situ
. Sensor-based systems are used in cotton to apply nitrogen, plant
growth regulator, and defoliant chemicals.
The sensor is the key component in a sensor-based VRA system. Several sensors are
commercially available for VRA use. Commonly used sensors for plant canopy reflec-
tance measurement are Crop Circle and GreenSeeker crop canopy sensors. The Crop
Circle sensor is made by Holland Scientific Inc. (Lincoln, NE). It uses modulated LEDs
as a light source and is able to measure the reflectance in three bands (one red band at
670 nm and two NIR bands at 730 and 780 nm; Model ACS-430). The GreenSeeker sen-
sor, manufactured by NTech Industries Inc. (Ukiah, CA), also uses modulated LEDs, at
656 nm (red) and 774 nm (NIR) (Model RT102), as light sources and measures the light
reflected from the plant canopy. NTech released a sensor-based N applicator that consists
of 30 GreenSeeker optical sensors and a variable-rate controller. Each sensor controls
three solenoid valves equipped with sprayer nozzles (Solie et al., 2002).
For successful use of a sensor-based VRA system, an understanding of how the
sensor works and what it actually measures is required, as well as knowledge of
relationships between sensor measurements and plant needs. The Crop Circle and
GreenSeeker sensors are positioned above a crop canopy and measure reflectance in
specific spectral bands. The reflectance data can be processed and used for various
other applications based on the relationship between reflectance and other variables
of interest. For example, Khalilian et al. (2008) conducted field studies to develop an
algorithm for variable-rate N application in cotton by using plant NDVI (based on the
GreenSeeker) and soil EC data. Results indicated potential for using mid-season spe-
cific plant NDVI data for VRA of N for cotton. Similar studies and results have been
reported by Taylor et al. (2007) and Sharma et al. (2008). Carrillo et al. (2006) evalu-
ated multispectral reflectance, N, water, and insect interactions on cotton in New
Mexico. They observed that treatments with higher NDVI readings corresponded
to higher levels of N fertilizer applied. Scharf et al. (2008) found good potential for
using the Crop Circle, CropScan (Rochester, MN), and GreenSeeker sensors to accu-
rately apply N on the go to cotton on various types of soil. Measurements with all
three sensor types appeared to be useful for indicating optimum N rate at mid square
and early flower growth stages (Khalilian et al., 2011; Scharf et al., 2008).
Sui et al. (1989, 2005) and Sui and Thomasson (2006) reported development of a
ground-based sensing system for determining N status in cotton plants. The system
consists of a multispectral optical sensor, an ultrasonic sensor, and a data acquisi-
tion unit (DAQ). The optical sensor uses modulated LEDs to provide panchromatic
illumination of the plant canopy and measures plant reflectance in four wavebands
(400-500, 520-570, 610-710, and 750-100 nm). The ultrasonic sensor determines
plant height. The DAQ simultaneously collects and processes data from the optical
sensor, ultrasonic sensor, and spatial information from a GPS receiver on the go.
Spectral-reflectance and plant-height data were compared to laboratory measure-
ments of plant leaf N content and used to train an artificial neural network (ANN) for
predicting N status in cotton plants. The trained ANN was able to predict N status
