Adjusting Mid-Season Nitrogen Fertilizer Using a Sensor-Based
Optimization Algorithm to Increase Use Efficiency in Corn
B. Tubana, R. Teal, K. Freeman, B. Arnall, B. Chung, O. Walsh, K. Lawles, C. Mack and W. Raun
Annual ASA Meeting, Indianapolis9:30 am, Nov. 15, 2006
Presentation Outline
•Technology Developed by OSU
•Background of the Study
•Components of the Algorithm
•Methodology
•Results
•Conclusion
Need to Improve NUE
•Cereal grain NUE averages only 33% worldwide
•Rise in the price of fuel and N fertilizer
•Increase environmental risk
Applications
Success of the Technology
A 15 % increase in wheat NUE was achieved compared with conventional methods (OSU 2002, Agronomy Journal 94:815).
Yield Potential Equation
800-1000 GDD
y = 0.8855e2802.3x
R2 = 0.75P < 0.001
YP0 = 1.161e2802.3x
0
2
4
6
8
10
12
14
16
18
20
0 0.0002 0.0004 0.0006 0.0008 0.001 0.0012
GDD INSEY
Gra
in y
ield
(Mg
ha
-1)
Efaw , OK OFIT 05
LCB, OK OFIT 05
Perkins, OK OFIT 05
Efaw , OK OFIT 04
LCB, OK Catchup 05
Efaw , OK Catchup 05
Haskell, OK Catchup 05
LCB, OK Nrate 05
Haskell, OK Nrate 05
LCB, OK Regional 05
Haskell, OK YP0 03
LCB, OK YP0 03
Haskell, OK YP0 04
LCB, OK YP0 04
By row 04
LCB, OK YP0 02
(Teal et al., 2006)
Algorithm Components
•YP0 Estimates of corn grain yield potential using NDVI and cumulative GDD
•RI N Responsiveness estimated using NDVI in the N Rich Strip and NDVI in the farmer practice or check
•CV Coefficient of variation determined from NDVI sensor readings collected in each plot
Nitrogen Fertilization Optimization Algorithm (NFOA)
Components of Algorithm
•YPN = (YP0*RI)
•N Rate =
FactorEfficiency
NuptakeNuptake YPYPN )( 0
•YP0 does not rely on historical data but rather is a simple predictive model. This approach uses seasonally dependent data capable of predicting differing yield potentials and adjusting N rates accordingly.
•YP0 changes every year as does RI.
Capability of Algorithm
YP0 and RI are independent of one another (on-farm trials 2002-2005)
0
10
20
30
40
50
60
70
1 1.2 1.4 1.6 1.8 2 2.2 2.4
Response Index
N R
eco
mm
end
ed
On Farm Trials
C Mack y = -0.0572x + 30.13
R2 = 0.0078
0
5
10
15
20
25
30
35
40
45
50
0 10 20 30 40 50 60 70
N Recommended
Yie
ld o
f C
hec
k, b
u/a
c
Spatial variability can be masked bylarger plants
NDVI= 0.60NDVI= 0.60
Do these areas have the same yield potential?
CV= 23 CV= 10
RICV-NFOA
Gra
in Y
ield
INSEY INSEY
YP0 YP0
RI-NFOA RICV-NFOA
YPmax
RI-NFOA and RICV-NFOA
YPN YPN
RI = 2.0
RI = 2.0
CV
RI = 2
.0
RI = 2
.0
RI = 1
.5
RI = 1
.5
CV
Description of NFOA
•RI-NFOA – consists of YP0 and RI
YPN = YP0*RI
•RICV-NFOA – consists of YP0, RI and CV
)(*)*0(CRITICALCAP
CAPMAX CVCV
CVCVRIYPYPNY
New CV Algorithm, docking for CV>20
0
500
1000
1500
2000
2500
3000
3500
4000
4500
5000
0 0.002 0.004 0.006 0.008 0.01
INSEY
Gra
in y
ield
, kg
/ha
0
10
20
30
40
50
60
70
80
90
100
N R
ate,
kg
/ha
YPN-CV
YP0
YPN old
N Rate CV
N Rate-RI
Objectives
•To evaluate different nitrogen fertilization optimization algorithms for prescribing mid-season fertilizer N.
•To determine the optimum resolution for treating spatial variability in corn.
Methodology
•Established in 2004 at 3 sites (1-irrigated, 2-rainfed system) in Oklahoma.
•Employed RCB Design with 3 replications
Treatment StructureTRT
Preplant N kg ha-1
Mid-Season Topdress Rate kg ha-1
Resolution m2
1 0 0 -
2 0 67 -
3 0 134 -
4 67 67 -
5 67 0 -
6 134 0 -
7 0 RICV- NFOA 0.34
8 67 RICV-NFOA 0.34
9 0 Flat RICV-NFOA -
10 67 Flat RICV-NFOA -
11 67 RICV-NFOA 2.32
12 0 RI-NFOA 0.34
13 67 RI-NFOA 0.34
Results
Treatment Preplant
kg ha-1
Topdresskg ha-1
Grain YieldMg ha-1
Nitrogen Use Efficiency
%
2004 2005 2006 2004 2005 2006 2004 2005 2006
Check 0 0 0 0 9.5 6.2 5.6 - - -
Common Flat Rate
67 67 67 67 13.4 10.3 9.6 48 57 38
67-RICV 67 25 127 52 13.9 12.0 11.1 31 52 57
67-RICV flat 67 25 127 52 13.3 11.5 10.3 35 49 48
67-RI 67 13 66 24 14.0 12.4 11.9 77 74 79
Common Flat Rate versus Algorithms at Efaw site from 2004-2006
With Preplant Nitrogen
Results
Treatment Topdresskg ha-1
Grain YieldMg ha-1
Nitrogen Use Efficiency
%
2004 2005 2006 2004 2005 2006 2004 2005 2006
Check 0 0 0 9.5 6.2 5.6 - - -
Common Flat Rate
67 67 67 13.1 9.9 9.4 71 69 73
Common Flat Rate
134 134 134 11.8 10.2 8.8 35 44 34
0-RICV 59 100 58 11.2 8.6 6.9 32 50 37
0-RICV flat 59 100 58 13.5 9.4 9.1 50 51 67
0-RI 17 66 48 12.9 10.1 9.2 79 73 83
Common Flat Rates versus Algorithms at Efaw site from 2004-2006
Without Preplant Nitrogen
Results
Algorithm Resolution
m2Total N Applied
Kg ha-1Grain Yield
Mg ha-1
Nitrogen Use Efficiency
%
2004 2005 2006 2004 2005 2006 2004 2005 2006
Check - 0 0 0 9.5 6.2 5.6 - - -
RICV-NFOA 0.34 25 127 52 13.9 12.0 11.1 31 52 57
RICV-NFOA flat 25 127 52 13.3 11.5 10.3 35 49 48
RICV-NFOA 2.32 25 132 56 13.3 11.4 10.1 35 46 54
RI-NFOA 0.34 13 66 24 14.0 12.4 11.9 77 74 79
RICV- versus RI-NFOA at Efaw from 2004-2006.With Preplant N
Results
TRT
Description Total N Appliedkg ha-1
Grain YieldMg ha-1
NUE, %
1 Check 0 5.5 -
2 *CFR-topdress 67 8.2 59
3 *CFR-topdress 134 8.5 40
4 *CFR-split 134 9.3 48
5 *CFR-preplant 67 8.0 56
6 *CFR-preplant 134 9.1 44
7 RICV-NFOA 66 7.5 49
8 RICV-NFOA 131 9.1 43
9 Flat RICV-NFOA 66 7.8 51
10 Flat RICV-NFOA 131 8.9 42
11 RICV-NFOA-2.32 133 8.8 46
12 RI-NFOA 61 8.3 63
13 RI-NFOA 119 9.6 56
On-average
* CFR : Common Flat Rate
Summary
•NUE was generally higher when mid-season N rates were generated by NFOA compared with flat farmer rates.
•Increased NUE was attributed to the lower N rates applied.
Summary
•Use of RI NFOA resulted in a higher increase in NUE than RICV NFOA.
•There was limited benefit of treating spatial variability at the high resolution (0.34 m2, RICV algorithm).
•NFOA approaches didn’t project high N rates that did not affect increased yields.
Conclusions
•Functional N rate algorithm developed for corn can increase NUE.
•Applications- Sensor Based N Rate
Calculator- Variable Rate Technology
(0.4m2)
THANK YOU!
www.nue.okstate.eduwww.nue.okstate.edu