State of Engineering in Precision Agriculture, Boundaries
Download
Report
Transcript State of Engineering in Precision Agriculture, Boundaries
State of Engineering in
Precision Agriculture,
Boundaries and Limits for
Agronomy
Topics
• Spectra
• Sensors
–
–
–
–
–
–
No. of Wave Lengths
Measurement units
Lighting Source
Field of View
Sampling Time
Range of Height
• Mechanisms for Variable Rate Application
– Effect of treatment resolution
– Available Mechanisms
Spectra
• Remote sensed electromagnetic radiation
in the visible and near infrared bands
– Sensors that work outside of that range are not
useful for sensing and variable rate application
because of cost, difficulty in manufacturing
and operating in a production agriculture
environment.
• There are other sensors that may be of
value, but the vast majority of the research
to date has been with optical sensors.
% Reflectance
0.50
Visible
Light
Near Infrared
0.25
450
550
650
750
850
Wavelength nm
950
1050
Interfering Inputs:
Soil Reflectances - Oklahoma
1
Reflectance (Fraction)
0.8
0.6
Tipton
Stillwater
Perkins
Mangum
Lahoma
Haskell
Goodwell
Ft. Cobb
Chickasha
Altus
Agron. Stwr.
0.4
0.2
0
350
450
400
450
550
500
650
550
600
Wavelength (nm)
650
700
750
800
Reflectance and Vegetative
Indices
• We normally do not use absolute
measurements of the reflected light.
• To minimize the effects of variation in the
sources of illumination all data must be
normalized by dividing the magnitude of
the reflected light by the incident light
Reflectance.
• Measurements are generally presented in
the form of vegetative indices, which are
associated with plant properties
NDVI
• Normalized Difference Vegetative Index
NIR RED
NDVI
NIR RED
• Developed as an irradiance Index for remote sensing
• Varies from -1 to 1
• Soil NDVI = -0.05 to .05
• Plant NDVI = 0.6 to 0.9
• Typical plants with
soil background NDVI=0.3-0.8
• NDVI from different sources vary
– Bandwidths for Red, NIR vary
– Irradiance vs. reflectance based
Illumination Source
• Natural (Sunlight) Lighting – Passive
Sensor
Or
Self Contained Lighting – Active Sensor
Passive Sensors
• Multi or Hyper Spectral
• Multi- Spectral Sense two or more wave lengths
• Measure both incident and reflected light
Or
Corrects be regularly measuring light reflected
from a “White” plate
• Generally uses Band Pass or Interference optical
filters
• Commercially available as a sensor/applicator
from Hydro
• 4 Band sensor being built by Holland Scientific
Passive Sensors
• Satellites such as IKONIS, LandSat, and
Quickbird all provide multispectral
imagery
• There a number of sources of aircraft
based imaging cameras
• There are digital cameras (e.g.
DuncanTech) the can be mounted on
aircraft or ground-based vehicles at a
“reasonable” cost
Hyper-Spectral Sensor
• Measures many (100’s) of band
• Spectrometer (with integrating sphere for
corn)
• Prototype sensor for scanning by Holland
Scientific
Active Lighting Sensors
• Uses self-contained illumination, usually high
intensity LED’s
• Wave lengths are limited to available LED’s
• Uses high frequency (approximately 40,000
cycles per second) light which enables the
electronics to filter out background (sunlight).
Measurements are independent of the
background illumination.
• Currently limited to 2 bands, e.g. 670 and 780 nm
or 550 and 780 nm
Sensing Technique
Detection of
Reflected
NIR and RED
+Sun
Direction
NIR and RED
Modulated
Illumination
Crop Target
Height
• The size of the field of view of all sensors
is affected by the height above the target
• Within a limited range it is possible to
minimize the effect of the height.
– GreenSeeker uses masks to maintain a
constant field of view from a height of 32 to 48
inches above the target.
• Field of View size ranges from 30 by 30 m
(LandSat TM) to 0.5 by 24 inches
(GreenSeeker)
Sensor Sampling or
Conversion Time
• All digital optical sensors require a finite
amount of time to convert measurements
to a digital format and to store the data.
• This may range for one second or more
for a digital camera to less than 0.1 s for
the GreenSeeker sensor.
• The combination of the sampling time and
the vehicle speed dictate the “length” of
the field of view.
Variable Rate Applicators
• I am not aware of any granular applicators
that can provide the distribution patterns
and response time needed to do high
speed variable rate application of nitrogen
or other fertilizers.
• Anhydrous ammonia – Kansas State
University/Capstan Corporation
• Liquid fertilizers
Tri-valve Set for Varying Rate
• 8 Stepped rates
– OFF to 7x
• Treats 24”x 24”
area applying the
target rate on 85%
of that area
Valve
Rate
1x 2x 4x Total
0
0
0
0
1
0
0
1
0
2
0
2
1
2
0
3
0
0
4
4
1
0
4
5
0
2
4
6
1
2
4
7
1x
2x
4x
Sensor Function
Light signal
Light
detection
Light
generation
Valve settings
Calculate NDVI
Lookup valve setting
Apply valve setting
Send data to UI
“Sensor”
Valves
and
Nozzles