precision SENSING

1
precision SENSING
Oklahoma State University
2
Approaches

• VRT applicator (temporal and spatial)
• Hand-held Sensor SBNRC (temporal)
• Calibration Stamp (temporal)

3
GreenSeekerTM For Wheat

• Sense and treat every 0.4m2
• Application rates of 0-88 kg/ha in 12 kg
  increments

4
GreenSeekerTM For Corn
5
GreenSeeker RT200 6 Sensors/20m
GreenSeeker RT500 30 sensors/20m
6
OSU/Ntech Algorithm

• Recognize small-scale spatial variability
• Based on mid-season Predicted Yield
• Independent use of the Response Index
• Yield Potential, RI, Estimates of Removal
• Coefficient of Variation

7
Potash Phosphate Institute
Predicting Nutrient Removal
8
1. Recognizing spatial variability
0.4 m2 in wheat 0.4 m2 pastures By-plant in corn
9
Distance, cm 35 45 84 99 137
10
Height, cm 189 190 204 197 185 Yield,
g 1038 866 1351 1324 943 Ear wt,g 258 222 344 349
245 Distance 15 24 26 26 28 Yld.
g/l-cm 16.8 9.2 12.9 13.1 8.5 2X
11
Yield per plant based on area occupied
Biomass, g 100 20 120
5X
Distance, cm 30 40
70 2.5X
Yield. g/l-cm 3.3 0.5
1.7 6.6X
12
NDVI at F5

INSEY
Days from planting to sensing, GDDgt0
Winter Wheat
Units N uptake, kg ha-1 day-1 where GDDgt0
13
Spring Wheat (Mexico)
NDVI a F5

INSEY
Days from planting to sensing
Units N uptake, kg ha-1 day-1
14
Corn (Oklahoma Nebraska, Mexico)
15
Yield PotentialPrediction Equations
16
Response Index
17
(No Transcript)
18
Response Indexvariable from year to year in the
same field
2004
1995
Long term N-P-K Experiments(1969-2004) Dr. Billy
Tucker
19
Response Index
Robert Mullen
Gordon JohnsonKyle FreemanPaul Hodgen
20
Response Index NDVI, N Rich Strip NDVI,
Farmer Practice Sensor GreenSeeker Growth Stage
F5-F7 V8-V12
21
Procedure to Calculate N Rate
Predict Yield Potential, YP0, with no additional
N Predict Response Index, RI, with additional N
Fertilizer using N Rich strip Calculate
Potential Yield with additional fertilizer, YPN
based on RI YP0 Fertilizer N f (GNUPYPN
,GNUPYP0,?)
3
2
1
3
4
2
4
22
CalibrationStamp
23
December
CalibrationStamp
March
24
CV and NDVI
25
NDVI 0.4CV ?
NDVI 0.4CV ?
26
RICV-NFOA
RI2.0
YP0
YPN
YPMAX
CV
Grain yield
CV
INSEY (NDVI/days from planting to sensing)
Nf ((YP0RI)(100-CV/100-CVlim)) YP0/EfCVlim
40 (wheat)CVgt20, YP0RI cannot increase
27
(No Transcript)
28
RI-NFOAYPNYP0 RI
YP0
YPN
YPN
YPMAX
RI1.5
RI2.0
Grain yield
INSEY (NDVI/days from planting to sensing)
Nf (YP0RI) YP0))/Ef
29
Max Yield-NFOA
YP0
YPMAX
Grain yield
INSEY (NDVI/days from planting to sensing)
Nf (YPMAX-YP0)/Ef
30
RI-CV Demonstration
31
Change in INSEY (Corn) Required to effect a
Change in N Rate
32
Change in INSEY (Winter Wheat) Required to
effect a Change in N Rate
33
http//www.soiltesting.okstate.edu/SBNRC/SBNRC.php
http//www.dasnr.okstate.edu/nitrogen_use
34
OSUCIMMYTUSDA-ARSCollaborative
Research Planning Matrix
35
CORN
36
Effect of Sensor Height on NDVI in Corn
37
Maiz, Oklahoma
38
Medida NDVI (FRN) Sensor GreenSeeker Estado
V6 - V8
39
Indice de Respuesta (RI) y NDVI con tiempo,
Perkins, OK 2004
40
3. Integración de componentes
Rendimiento Predicho con fertilizacion(YPN)
RI
Indice de Respuesta 0.75/0.57 1.31
N-no limitado
Practica del agricultor
Fecha de siembra
Rendimiento predicho sin fertilizacion (YP0)
Dias desde la siembra
NDVI (fecha de medida)
INSEY (NDVI/dias desde la siembra, GDDgt0)YPN
YP0 RIDosis de Fertilizacion (N asimilado
YPN N asimilado YP0)/0.7
41
Coeficiente de variacion
42
Mas variabilidad sin N
Menos variabilidad con N
43
Utilizacion de informacion estadisticaCoeficiente
de variacion desviacion estandar/media
NDVI 0.2CV ?
NDVI 0.4CV ?
NDVI 0.8CV ?
44
NDVI 0.4CV ?
NDVI 0.4CV ?
45
Brian Arnall, estudiante de Maestria
46
Variabilidad (CV) por Tratamiento a través del
tiempo, 3 localidades, Oklahoma 2004.
47
Rendimiento Predicho por Planta, Maiz, 2003
48
Maiz, Oklahoma
49
Rendimiento Predicho por Planta, Maiz, 2004 (pre
VT)
50
Ecuaciónes para Rendimiento Predicho, Maiz, Trigo
51
Brian Arnall, estudiante de Maestria
52
Relacion entre NDVI medido por surco versus
contra surco
Brian Arnall, estudiante de Maestria
53

• NDVI por planta (cada 0.5 cm) y medicion de
  altura utilizando ultra sonido
• Por planta NDVI
• Por planta CV

36
20
24
Kent Martin, estudiante de Maestria
54
Grain yields could be predicted from mid-season
NDVI readings (Agron J. 93131), divided by the
number of days from planting to sensing where GDD
gt0 Nitrogen Fertilization based on mid-season
estimates of yield potential and a response index
increased NUE by gt15, when fertilizing each 1m2
area independently (Agron J. 94815)
55
Response Index vs. Sufficiency ?
56
Integrating Componentsa. Predicted Yield b.
Response Index
57
Nitrogen Fertilizer Optimization Algorithm
Aug 16, 2002
Aug 28, 2002
Nov 7, 2002
1. Establish preplant N Rich Strip(NRS) 4.
Determine Response Index (RI)2. NDVI (biomass)

NDVINRS/NDVIFarmer3. Predict potential yield
5. Predict potential
grain yield need of days from planting to
(YP0) with added N, YPN (YP0RI)
sensing (INSEY biomass prod./day) 6.
Fertilizer Rec (grain N uptake INSEY vs.
Yld eqn. YPN
grain N uptake YP0/0.7)
CORN
YPN
RI
May
June
August
NDVI (sensing date)
YP0
Planting date
days from planting to sensing (GDDgt0)
WHEAT
November
March
June
58
Wheat grain yield response to fertilizer N
applied using variable N rate strategies and
fixed preplant rates, Chickasha, OK,
2004. Trt Preplant Topdress Total
Method Grain NUE Gross N Rate N Rate N
Rate Yield Revenue kg/ha kg/ha kg/ha kg/ha
/ha 1 0 0 0 - 1510 - 194 2 135 0 135 All N
pre 2631 20 264 3 0 27 27 Var. NFOA 2152 57 262 4
50 12 62 Var. NFOA 2377 32 271 5 0 29 29 Flat
NFOA 1855 28 222 6 50 13 63 Flat NFOA 2647 42 305
41/ha 7 0 100 100 Flat Topdress 1781 6 174 Se
nsor CV at F5 12 (NDVI 0.3) __________________
_________________________________________ NFOA
nitrogen fertilization optimization algorithm
Flat rate fertilizer applied uniformly to the
entire treatmentGross Revenue using 52 USD /ton
of wheat and 0.62 per kg N.
59
Wheat grain yield response to fertilizer N
applied using variable N rate strategies and
fixed preplant rates, Efaw, OK,
2004. Trt Preplant Topdress Total
Method Grain NUE Gross N Rate N Rate N
Rate Yield Revenue kg/ha kg/ha kg/ha kg/ha
/ha 1 0 0 0 - 2105 - 270 2 135 0 135 All N
pre 2712 11 274 3 0 25 25 Var. NFOA 2724 60 337
63/ha 4 50 9 59 Var. NFOA 2693 23 313 5 0 34 34
Flat NFOA 2587 34 314 6 50 16 66 Flat
NFOA 2724 22 314Sensor CV at F5 16 (NDVI
0.4) _____________________________________________
_______ NFOA nitrogen fertilization
optimization algorithm Flat rate fertilizer
applied uniformly to the entire treatment Gross
Revenue using 52 USD per ton of wheat and 0.62
per kg N.
60
Wheat grain yield response to fertilizer N
applied using variable N rate strategies and
fixed preplant rates, Lake Carl Blackwell, OK,
2004. Trt Preplant Topdress Total
Method Grain NUE Gross N Rate N Rate N
Rate Yield Revenue kg/ha kg/ha kg/ha kg/ha
/ha 1 0 0 0 - 3297 - 424 2 135 0 135 All N
pre 4065 13 448 3 0 14 14 Var. NFOA 3650 57 461 4
50 11 61 Var. NFOA 4155 33 500 5 0 22 22 Flat
NFOA 3519 26 440 6 50 14 64 Flat NFOA 4215 33 506
58/ha 7 50 34 84 Flat pre/top 4099 16 481Sen
sor CV at F5 22 (NDVI 0.5) ___________________
_______________________________________ NFOA
nitrogen fertilization optimization algorithm
Flat rate fertilizer applied uniformly to the
entire treatment Gross Revenue using 52 USD per
ton of wheat and 0.62 per kg N. obtained from
long-term trial 601 at LCB.
61
(No Transcript)
62
By-plant-yield (entire plants, stalkear
harvested at the base) over 6 m of row, and
corresponding NDVI sensor readings collected from
this exact same area at the 7 leaf and 10 leaf
growth stages, 87 and 70 days earlier, Texcoco,
Mexico, 2002.
63
(No Transcript)
64
Procedimiento
Resultados Rend. Potencial sin N, Rend.
Potencial con N, dosis de N requerido
Entrar Rendimiento maximo para el area, fecha de
siembra sensor
65
2002 Field trials

• 10 trials using 60-ft boom sensor-applicator.

• Web site.
• http//www.dasnr.okstate.edu/nitrogen_use/

66
Summary of Results - 10 Field Trials
GreenSeekerTM Field Sprayer (OSU Wheat)
67
1. Basing Algorithm on Predicted Yield

• Grain N
• Wheat 2.13 N (Tkachuk, 1977)
• Corn 1.23 N (Tkachuk, 1977)
• Wheat
• 33 kg N, 1Mg/ha (2 lb/bu) (Johnson 2000)
• Corn
• 20 kg N, 1Mg/ha (1.2 lb/bu) (Schmitt et al.,
  1998)

68
Catch-Up-2003
69
Catch-Up-2004
70
GreenSeekerTM for Corn

• Sense and treat each 0.4m
• 0-120 kg N/ha in increments of 12 kg

71
Mobility of N
72
(No Transcript)
73
NDVI, 4 Weeks later
NDVI, March 1, 2004Hennessey, OK
20 hectares
74
(No Transcript)
75
Bermuda
y 133.85e1846x R2 0.64
76
OSU GreenSeekerMechanical Error