Crop Circle Experiences

Info Ag 2007
Jim Schepers

USDA-Agricultural Research Service Lincoln, Nebraska

Crop Circle Specifications

Low power requirement (~100 ma) Light weight (~3/4 lb) Sample rate (10 per second) Polychromatic light source Amber (595 nm) Near infrared (880 nm)

ACS - 210

Important Information
Foot-print ~4 x 20 @ 36 uniform effective intensity size increases with distance Operational distance stable from ~16 to >96 Output individual bands programmable indices

Polychromatic LED Emission Graph Visible

Chlorophyll
LED Output Intensity

Biomass vigor

greenness

N status
How fast the Factory is Operating
400nm 400nm CWL1 700nm

NIR
Size of the Factory
CWL2 900nm

Wavelength (nm)

Modulation/Demodulation Using Polychromatic LEDs

TARGET
Identical Foot-print Simultaneous readings

SENSOR
LED PD1, <700nm PD2, >700nm

Small plot research attempts to develop the science

Field experiences are where we must learn to deal with reality

What Does the ACS-210 Sensor See ?

Foot-print reflectance is measured 40,000 times per second

At 6 mph (~9 ft / sec)

and

10 outputs / sec

Foot-print advances ~11 per output (1/10 sec) and contains the average of 4,000 individual readings

Plants absorb RED and BLUE light during photosynthesis. They use ORANGE, YELLOW, and GREEN light less effectively. This is why plants containing chlorophyll appear GREEN > YELLOW > ORANGE

vegetation

Implications
Once there is enough vegetation to absorb most of the RED radiation, sensors that measure RED light lose their sensitivity.

The literature shows that leaf area index (LAI) > ~2.0 are more than enough to absorb a majority of the RED light (~V8-V10 growth stage for corn).

What Happens if the Sensor in NOT Centered ?

Adjust sensor height so that foot-print is equal to row spacing

Misaligned Crop Circle sensor is not a significant problem because sensing is uniform across the foot-print and what is lost from one row is gained by the adjacent row

If illumination is not uniform across the foot-print - careful driving is essential and wide or narrow guess rows can be problematic with large sprayers.

Time of Day
Leaf Moisture

Day

versus

Night

No preference - can be confused with water stress

Dry

versus

Wet

No significant difference - related to water status

H2O

Well-watered

versus

Water Stressed

Water stress affects leaf turgor and NIR reflectance Extended periods of water stress reduce leaf chlorophyll concentration Differentiating between water and nitrogen stresses requires additional wavebands

Sensor Configuration
Number of Sensors
1 2 3 4 6 - - - more

Boom Segment Rows

4 6 8 12 16 24 36 sensor * rows *** **** ****** ********
Etc.

Evaluation of Row-to-Row Variability

8-row ridge-till continuous corn center-pivot irrigation
Chlorophyll Index Averages
5.5 5.3 5.1
Chlorophyll Index

4.9 4.7 4.5 4.3 4.1 3.9 3.7 3.5 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24

Row

Courtesy: Darrin Roberts, University of Nebraska

Field Calibration
Reference Strips Potential Problems Spatial variability in soil properties Ramped Calibration Strips Potential Problems Spatial variability in soil properties

Other Ideas Imagery statistically identify upper 10-20% of the

pixels and use as field reference value

Sensor Mapping
generate virtual reference for the field by krigging data from several reference strips select less vigorous portions of reference strip as the field reference value statistical non-reference strip approach using sensor data from several random strips

Calibration of Sensors N-Rich area was developed for small plots using Minolta
Chlorophyll Meters (SPAD 502)
Replication 1 2 3 4

= Sufficiency Index (Relative SPAD) For Example:

< 95% means needs more N

How to extrapolate the small-plot calibration concept to whole fields ?

Current year reference strips

Last years reference strips

Vegetation Index
low vigor high vigor

Corn @ ~V10

Area used as reference when making variablerate N applications

Three adequately fertilized reference strips. Yellow lines depict soil type differences. Note differences in sensor vegetation index values even though each strip is considered to have abundant N availability.

N rate application map for research strips. Dark blue area was used as the reference (no N application area).

Implementing In-Season N Technologies

Sensors and GPS Imagery
Beep .. Beep .. Beep ..

Available - with limitations

Flow Rate Controller

Pressure / Flow Regulation Speed monitor

Nozzles
Ideal
rate
Low High Medium High

Variable-Rate Delivery Approach

Orifice Sizes 1x, 2x, 4x
OFF plus 7 rates

Cost ~$100 per unit Concerns Requires special controller / driver Power requirement can be high when many nozzles are involved
Rate control at nozzle

Capstan Nozzle
Orifice Sizes User selected Delivery Rate

0 to 2.0 gal / min

example

Cost Depends on number of nozzles Concerns Uniformity at low application rates at 3 Hz ??? Highest application rate may not be enough ???
Rate control at nozzle

VariRate Nozzle Approach

by SprayTarget

Output Rate 0.1 to 3.0 gal / min

gal / min

3.5 3 2.5 2 1.5 1 0.5 0 0 10 20 30 PSI 40 50 60 70

Cost ~$60 per unit

Concerns Three phase delivery function could be difficult to obtain accurate rates in the sensitive range

According to the reference pressure table, a 2 psi change in pressure can increase the output by 50% at some pressures. What is the pressure drop across a long boom ?
Flow rate / pressure control

Linear Variable-Rate Nozzle

Orifice Sizes Not available Delivery Rate 0 to ~1.5 gal / min
Other options all linear
gal UAN / min

1.60 1.40 1.20 1.00 0.80 0.60 0.40 0.20

Cost To be determined

0.00 0

?
20 40

y = 0.0118x + 0.4419 R 2 = 0.9945

60 PSI

80

100

No free-drainage

Concerns Uniformity across the boom at low pressures ???

Flow rate / pressure control

Experiences with High-Clearance Sprayers

equipped with Crop Circle sensors
Sensor readings are highly correlated with SPAD meter readings and visual observations when driving through fields. Fertilizer delivery rate adjustments are nearly instantaneous with multi-orifice and Capstan nozzle packages but delayed several seconds with standard controller / nozzle systems. (2 sec @ 6 mph ~18 ft) Changes in speed to accommodate the topography need to be directly linked to the controller to make rapid pressure adjustments. Visual display of individual sensor values on GeoSCOUT controller provides assurance of uniformity and/or signals problems.

Flow rate / pressure adjustments need to be made in <1 sec and nozzle delivery needs to have a wide dynamic range and preferably with a simple pressure versus flow-rate relationship.

For information about active sensors, leave your name, take my card, or contact me at: Jim.Schepers@ars.usda.gov Information Available Active Sensor Tidbits (5 pages) Real-Time Variable Rate --- OR NOT (6 pages)
For more information about active sensors see InfoAg 2005
presentation at www.hollandscientific.com

USDA ARS Fort Collins, CO

Walter Bausch

Crop Circle

GreenSeeker

Nitrogen Response Index Normalized Difference Vegetation Index Normalized Chlorophyll Index

Chlorophyll Index

Courtesy: Walter Bausch

Nitrogen Sufficiency Index

 The farmer is sure that the Crop Circle was indicating what was actually there. The sandy hill is definitely noticeable.

Walter Bausch

Courtesy: Walter Bausch

Genetics, Weather, and N Rate

Fred Below and Matias Ruffo University of Illinois

0.70

0.65

NDVI

0.60

2005 2006

Even under good growing conditions (2006), corn hybrids have different reflectance characteristics.

0.55

0.50

50

100

150

200

250

Fertilizer N (lbs/A)

Weather can have a significant effect on sensor readings but in a relative sense the data are consistent across N rates.

Hybrid A Hybrid B

Columbia, Missouri USDA ARS

Newell Kitchen Ken Sudduth Scott Drummond

Courtesy: Newell Kitchen

Example of a GeoSCOUT GLS - 420 V-Rate System

DGPS Receiver

Raven or Rawson Variable-Rate Compatible Controller

Pre-processed base map and application algorithm

Yield Management zone Soil type Elevation Soil color / OM Electrical conductivity Soil testing

Real-time crop data

Up to 8 sensors with a minimum of 2 sensors

Courtesy: Newell Kitchen

Sensor-Based, Variable-Rate Nitrogen on Corn Missouri Field-Scale Studies

Application Equipment USDA-ARS Spra-Coupe Producer Rigs Dealer Rigs 2004 2005 7 5 7 2006 5 8 4

Courtesy: Newell Kitchen

Dry

Fluid

Anhydrous Ammonia

Courtesy: Newell Kitchen

Average Nitrogen Rates and Corn Yield for Five ARS Studies in 2006

Averages Treatments lb N / A bu / A

Producer Rate @ Emergence Producer Rate Late Side Dress Sensor N at Planting + Sensor
Courtesy: Newell Kitchen

162 162 128 131

157 161 161 160

Wheat Fertilization
Jim Wilson England
Monitor for individual sensor index readings (4 of 8 channels)

Special Problem
N-Rich strips are not allowed for environmental reasons Multiple N applications are made using a tram-line system

N fertilizer application to winter wheat (England)

Courtesy: Jim Wilson

Scanner device to evaluate row-to-row variability

Courtesy: Jim Wilson

Sensor / SPAD Comparison

Jack Meisinger USDA ARS (Beltsville)

Sensor height adjustment mechanism

Amber Sensor & LCM vs. Corn TN Uptake on JD203 in MD 0.80 0.75 0.70 0.65 0.60 0.55 0.50 4 5 6 7 8 9 10 11 12 13 14 Corn TN Uptake, g N / m2
Amber NDVI, r = 0.86 Chlor Meter, r = 0.90

65 60 55 50 45 40 35

Courtesy: Jack Meisinger

Leaf N concentration is highly correlated with SPAD readings and sensor index values, but sensor sensitivity diminishes at high N uptake values. N is subject to luxury consumption, chlorophyll is not.

LCM Reading

NDVI

Crop sensor tools with potential to estimate plant Health

Light pattern

Ability to substitute for SPAD meter ?

R.L. Nielsen, NitrogenMgmt2007.ppt

Sensors relate well to each other

2006 Purdue N Trials, Camberato et al.; R.L. Nielsen, NitrogenMgmt2007.ppt

Crop Circle sufficiency index values

- relate well to relative yield

2006 Purdue N Trials, Camberato et al.; R.L. Nielsen, NitrogenMgmt2007.ppt

A 5% relative stress at V11 resulted in a ~10% yield loss

Vineyard Management

Greece
Goal - Improve wine quality
by correlating with soil properties, landscape position, plant vigor, and grape yield

Significant correlation between amber NDVI at fruit set and pruning weight after harvest Implication - Where and how much

in-season pruning is needed to optimize harvest quality and quantity

Spatial Vineyard Management

We are initially using this assembly in the vineyard with a DGPS/data logger to characterize vine biomass. It works so much better than a single-side-looking unit. (David Lamb)

Crop Circle Sensed Imagery

CCSI

Scan vineyard canopy

Uniformity Irrigation Disease Nutrients Harvest decisions Pruning

New Zealand, Australia, Greece

Conceptual Depiction

Bare Soil Image

After Planting

Moist soil accentuates color differences

Soil color mapping behind seed corn planting (male
rows only)

Soybean Mapping

Source: Nyle Wollenhaupt / Aaron Schepers

Crop Circle is a convenient method of calibrating airborne multi-spectral imagery for generating maps
David Lamb
Precision Agriculture Research Group University of New England, Australia

Evaluating a Triticale trial

Active Red / NIR reflectance sensor
define Optical Crop Height Index (OHPI) = NiR/R (ie based on Simple Ratio)

Crop Circle response

12

CropCircle-derived OCHI

10 R = 0.91 8 6 4 2 0 10 15 20 25
Optical crop height index (OCHI)
2

Triticale (late tillering)

30

35

40

Actual crop height (cm)

UNEBird
Colour-infrared image

Pre-graze (late tillering)

Conversion to OPHI

Conversion to crop height

Hay Market Park

Lincoln, NE

ANDVI NIR/amber

Crop Circle Mapping

amber

NIR
Each waveband contains unique information

For information about active sensors, leave your name, take my card, or contact me at: Jim.Schepers@ars.usda.gov Information Available Active Sensor Tidbits (5 pages) Real-Time Variable Rate --- OR NOT (6 pages)
For more information about active sensors see InfoAg 2005
presentation at www.hollandscientific.com