Refinement of the Missouri Corn Nitrogen
Algorithm Using Canopy Reflectance
Newell Kitchen, Ken Sudduth, and Scott Drummond
USDA-Agricultural Research Service
Peter Scharf, Harlan Palm, and Kent Shannon
University of Missouri
Active-light Reflectance Sensing
Objective: To assess on different Missouri
soils the use of active crop-canopy
reflectance sensors for assessing corn N
need and developing algorithms for
optimizing economic returns with variablerate N fertilizer application.
Methods
•
A total of 16 field-scale experiments were conducted
over four growing seasons (2004-2007)
•
These fields represented three major soil areas of
Missouri: river alluvium, deep loess, and claypan.
•
Multiple blocks of N randomized rate response plots
were arranged end-to-end so that blocks traversed the
length of each field. Each block consisted of 8 N
treatments from 0 to 235 kg N/ha on 34 kg N/ha
increments, top-dressed sometime between vegetative
growth stages V7 and V11
•
For 2006 and 2007, a complete second set of fieldlength blocks were also established where 67 kg
N/ha was uniformly applied over the set of blocks
shortly after corn emergence.
• Adjacent to and on both sides of the response blocks,
N-rich (235 kg N/ha) reference strips were also
established. These ran the full length of the field and
were treated shortly after corn emergence.
•
An AGCO Spra-Coupe (AGCO
Corp., Duluth, GA) high-clearance
applicator equipped with an AGCO
FieldStar Controller was used to topdress between corn rows solution
UAN (28 or 32% N) fertilizer for the N
rate treatments.
•
Crop canopy reflectance sensor
(Crop Circle model ACS-210,
Holland Scientific, Inc., Lincoln,
NE) measurements were
obtained from the corn canopy of
the N response blocks at the
same time the Spra-Coupe was
used to apply N rate treatments.
• On the same day reflectance
sensor measurements were also
obtained from the N-rich
reference strips.
•
Data analysis for these field studies included
four major steps:
1) Determining optimal N with quadratic-plateau
modeling
2) Processing of canopy reflectance senor data from
response plots and the N-rich reference areas
3) Relating modeled optimal N from step 1 with
sensor measurements from step 2
4) Developing optimized-profit algorithms relative to
conventional producer N rates
Results
2006 Ben
100
2005 Lic
Optimal N Rate as a Function of
Canopy Reflectance
Optimal N (kg ha-1)
0
300
300
2006 Rie
20042004
Pet
2004 Sch
Ben
2004 Cop
2004 Wil
200
200
2007 Geb
2007 San
2007 Hac
100
100
0
300
2006 Ben
2005 Lic
2006 Cop
0
0.6
0.8
200
300 00.6
1 0.4
0.8
2004 Pet
g ha-1)
4
2006 Geb
2006 Cop
100
200
0
300
0.5
1 0.4
0.6
Sufficiency Index
1.0
0.8
VIS0.6
0.8
2004 Sch
NIR (N ref)
SI =
VIS
NIR (target)
1 0.4
300
2004 Ben
2004 Cop
2004 Die
2004 Hay
2004 Pet
2004 Sch
2004 Wil
2005 Geb
2005 Lic
2006 Ben
2006 Cop
2006 Geb
2006 Rie
2007 Geb
2007 Hac
2007 San
Optimal N Rate as a Function of
Canopy Reflectance
200
100
Optimal N (kg ha-1)
0
300
200
100
0
300
200
100
0
300
200
100
0
0.4
0.6
0.8
1 0.4
0.6
0.8
1 0.4
0.6
Sufficiency Index
0.8
1 0.4
0.6
0.8
1
Developing Optimized-profit Algorithms
Relative To Conventional Producer N Rates
Inputs:
• Values and quadratic response curves from optimal N rate modeling
• Field-measured SI values for each response block
•
•
A set price of corn grain and N fertilizer
A producer prescribed N rate for each site-year
Variables that were optimized during the iterative phase included:
•
•
•
Slope and intercept values for the N recommendation, based on the
equation: Nrec = a(1/SI) + b (SI = sufficiency index)
The minimum N rate to be applied by the algorithm
The maximum N rate to be applied by the algorithm
The analysis was repeated on 12 subsets of data, based upon factorial
combinations of the following two variables:
• Three soil types & all soils combined
• N applied at planting (0 kg N/ha, 67 kg N/ha, both combined)
Fertilizer To Grain Ratio (FGR) Using SI
Units For Various Combinations Of N
Fertilizer And Corn Grain Prices
N fertilizer cost
---- $ kg-1 ---0.44
0.66
0.88
1.10
1.32
1.54
1.76
1.98
2.21
corn grain price ($ kg-1)
0.08 0.12 0.16 0.20 0.24 0.28 0.32
------------------------FGR ----------------------6
4
4
3
3
2
2
8
7
6
5
4
4
3
11
9
7
6
6
5
4
14
11
9
8
7
6
6
17
13
11
10
8
7
7
20
16
13
11
10
9
8
22
18
15
13
11
10
9
25
20
17
14
13
11
10
28
22
19
16
14
12
11
2.00 3.00 4.00 5.00 6.00 7.00 8.00
corn grain price ($ bushel-1)
N fertilizer cost
---- $ lb-1 ---0.20
0.30
0.40
0.50
0.60
0.70
0.80
0.90
1.00
General Shape of N Algorithm
200
FGR
N Recommendation (kg ha-1)
16
4.7
Maximum N
150
100
Increasing N
50
Minimum N
0
0.4
0.6
0.8
Sufficiency Index
1
Optimal N Rate as a Function of
Canopy Reflectance
Profit Potential Using the Canopy Sensors
And Derived Algorithms Relative to
Fertilizer to Grain Ratio (FGR)
Nitrogen Saved Using the Canopy Sensors
And Derived Algorithms Relative to
Fertilizer to Grain Ratio (FGR)
23 DAP
41 DAP
47
V4
V7
V10
Subtle Differences from the Perspective
of a Canopy Sensor
Days after
planting /
Growth stage
N applied at Planting (kg ha-1)
0.53
NDVI: 0.36
0.64
NDVI:
0.53
0.66
NDVI:
0.36
ISR:
0.31
ISR:
0.22
0.31
0.21
ISR:
0.47
0.47
0
45
246
48.6
SPAD:
49.9
SPAD:
52.8
SPAD:
41.1
SPAD: 39.8
0.57
0.70
0.36
0.28
0.18
0.47
58.8
57.6
43.7
NDVI:
ISR:
SPAD:
SPAD:
0.57
0.70
0.73
0.36
0.28
0.18
0.16
0.47
58.8
57.6
59.8
43.7
23 DAP
41
47
56 DAP
V4
V7
V10
V13
0.53
0.64
0.66
0.36
0.31
0.22
0.21
0.47
48.6
49.9
45.1
39.8
NDVI:
ISR:
SPAD:
0.53
0.66
0.68
0.36
0.31
0.21
0.19
0.47
52.8
52.4
41.1
NDVI:
ISR:
SPAD:
Active light source
crop sensors
Chlorophyll meter
0noatNplanting
40
45 at
kg planting
N ha-1
246 kg
N ha-1
210
at planting
0.40
0.30
SPAD
Sensor (ratio)
0.50
0.20
0.10
0.00
4
5
6
7
10
Growth Stage
75
11
13
15
75
70
65
60
55
50
45
40
35
30
no
no N
N
45
45 kg
kg NNha-1
ha-1
246 kg N ha-1-1
246 kg N ha
2
4
6
8
10
Growth Stage
12
14
16
Summary of In-Field Plant Sensing
for Nitrogen Management
• A significant relationship between canopy sensor
sufficiency index and optimal N rate was observed in
about half of the field studies.
• Combined across all sites meaningful algorithms
were created to give assurance that these sensors
could be used for variable-rate N applications.
• Algorithms should be adjusted as fertilizer and grain
prices vary.
• The primary advantages of sensor-based
measurements is improved accuracy to site-specific
crop need.
Fields and Situations Most Suited
for Sensor-Based Variable Rate Nitrogen Applications
• Fields with extreme variability in soil type
• Fields experiencing a wet spring or early summer (loss
of applied N) and where additional N fertilizer is
needed (i.e., rescue N)
• Fields that have received recent manure applications
• Fields receiving uneven N fertilization because of
application equipment failure
•
Fields coming out of pasture, hay, or CRP
management
• Fields of corn-after-corn, particularly when the field has
previously been cropped in a different rotation
•
Fields following a droughty growing season
An Example of a N Rate Recommendation
Algorithm Shown Relative to Canopy
Sensor-Based Sufficiency Index (SI).
EONR (kg ha-1)
300
corn = 0.24 $ kg-1
nitrogen = 1.32 $ kg-1
slope = 416 kg ha-1
intercept = -348 kg ha-1
min N = 101 kg ha -1
max N = 180 kg ha-1
Mean Profit = 22.98 $ ha-1
200
100
All soil types
Early N = 0
0
0.4
0.6
0.8
1
Sufficiency Index
1.2
1.4
Profit ($ ha-1)
$156.60
$100.00
$50.00
$25.00
$0.00
$-25.00
$-50.00
$-100.00