Transcript An update on kansas sensor based N recommendations

An update on Kansas sensor based N
recommendations
Drew Tucker and Dave Mengel
KSU Agronomy
Background:
 Our program was started in 2005 with a visit to OSU.
 Visited Bill Raun and John Soley that summer
 Bill gave us a GreenSeeker red handheld unit to use
 Hired Drew Tucker as a GRA to work on the project
 We began taking measurements with the GreenSeeker
red sensor on an existing sorghum experiment that
summer with the intent of seeing how it worked, and if
it did, develop K-State N rec’s for wheat, and
sorghum.
Background:
 We put out our first field trials in 2006:
 Focus was on sorghum and wheat as those crops were
where we felt we could make the greatest impact.
 Purchased a Crop Circle amber sensor in 2006.
 Setting between OSU and UNL-ARS, we wanted to
understand the differences between the two sensors and
the different approaches the two groups were taking
 We wanted to determine if we needed different algorithms
for different sensors.
Where we’re at today
 Initial downloadable algorithms (GreenSeeker
based) were made available to the public for
wheat and sorghum through the K-State Soil
Testing Lab website in January 2009.
 K-State algorithm options were added to the
menu of the GreenSeeker control unit for wheat
and sorghum for the 2009 crop year.
 Corn algorithms for use at V-8/9 will be available
for 2010.
General Approach
Our General Approach to Algorithms
and Rate Calculators
 Our approach to developing rate calculators is based on:
An in-field reference strip; calculated RI; relationship of RI at sensing to
RI at harvest; expected N uptake; and NUE.
 Yield potential is estimated from reference strip.
 Yield response to additional N is based on the expected RI at
harvest based on RI at sensing and yield potential of reference.
 Additional N need to optimize yield is estimated from N
uptake data across a range of yield levels,
 Estimated NUE is used to covert N uptake need to fertilizer N
need. Base is 50%, but can be adjusted up or down based on
soils, climate or management practice.
K-State Grain Sorghum Nitrogen Rate
Calculator
Farmer Inputs
NDVI Reference Strip
NDVI Farmer Practice
Max Yield for Area bu/ac
Days from planting to sensing where avg. temp > 63 F
Grain Price, $/Bu
Nitrogen Price, $/lb actual N
Application Cost, $/Ac
Expected Nitrogen Efficiency, % Recovery
0.6
0.57
130
35
2.5
0.35
5
50
Outputs
Expected Response Index of Grain Yield
Yield Potential of Reference Strip bu/ac
Yield Potential without N bu/ac
N Rec. lbs N/Ac unadjusted for G:N price ratio
N rec. lbs N/Ac adjusted for G:N price ratio
Gross Return (no Nitrogen) $/ac
Gross Return (using N Rec) $/ac
1.26
100.1
79.6
39.0
40.8
198.94
233.34
Grain Yield Bu/Acre
Grain sorghum yield relationship used to
estimate yield potential of the field.
200
y = 4.9796e175.06x
180
R2 = 0.76
160
140
120
100
80
60
40
20
0
0.0100
0.0120
0.0140
0.0160
0.0180
INSEY (NDVI/DAP)
0.0200
0.0220
Grain Sorghum Yields vs. NDVI 2006-2008
Grain Yield Bu/Acre
200
y = 6.3098e3.923x
180
R2 = 0.49
160
140
120
100
80
60
40
20
0
0.2
0.3
0.4
0.5
0.6
NDVI at GS-3
0.7
0.8
0.9
Grain Sorghum Yields vs. NDVI/GDU’s 2006-2008
Grain Yield Bu/acre
200
y = 7.4016e3341.5x
180
R2 = 0.61
160
140
120
100
80
60
40
20
0
0.0002
0.0004
0.0006
0.0008
INSEY (NDVI/GDU's)
0.001
RI Grain Yield at Harvest
RI at GS-3 vs. RI Grain Yield at Harvest used to
calculate yield response to added N (delta yld)
12
10
y = 0.0331e3.456x
R2 = 0.92
8
6
4
2
0
0.9
1
1.1
1.2
1.3
1.4
RI at GS-3
1.5
1.6
1.7
1.8
Measuring biomass and uptake
Sorghum N Uptake (grain+stover) vs. Yield
160
y = 0.8327x + 11.509
R2 = 0.94
Total N Uptake lbs/ac
140
120
100
80
60
40
20
0
0
20
40
60
80
100
Yield bu/ac
120
140
160
180
Example calculation
 Reference strip NDVI = 0.60 @ 35 DAP, INSEY = 0.0171
 Farmer Practice NDVI = 0.57, INSEY = 0.0163
 RI vegetation = 1.05, RI grain = 1.26
 Yield potential of fertilized reference strip, 100 bu/ac
 For farmer practice without additional fertilizer, 80 bu/ac
 Calculated from reference yield potential and RI grain
 N uptake needed to gain additional 20bu yield to realize
potential
 20 bushels x 0.95 lbs N/bu = 19 lbs N uptake per acre
 Adjusted for NUE or recovery 19/0.5 = 38 lbs N fert/ac
Grain Sorghum Nitrogen Sensor
Calculator Performance 2006-2008
Location Year
Belleville 2006
Manhattan 2006
Partridge 2006
Tribune
2006
Manhattan 2007
Partridge 2007
Tribune
2007
Manhattan 2008
Ottawa
2008
Partridge 2008
Mean difference
Sensor Yield
95
160
48
130
111
77
71
151
58
140
Actual Yield
96
155
32
128
109
70
79
128
64
123
Sensor Rec.
0
33
57
24
98
15
0
45
55
30
Actual N Resp.
0
33
55
15
105
20
0
45
60
15
Sensor Dif
0
0
2
9
-7
-5
0
0
-5
15
4.3
Comparison of Sensor Based N Rec’s to Soil Test
Based Rec’s, 2006-2008
Location Year
Belleville 2006
Manhattan 2006
Partridge 2006
Tribune
2006
Manhattan 2007
Partridge 2007
Tribune
2007
Manhattan 2008
Ottawa
2008
Partridge 2008
Sensor Yield Actual Yield
95
160
48
130
111
77
71
151
58
140
96
155
32
128
109
70
79
128
64
123
Soil Test Rec. Sensor Rec.
40
60
42
30
130
40
54
77
56
41
0
33
57
24
98
15
0
45
55
30
Actual N Resp.
0
33
55
15
105
20
0
45
60
15
Soil Diff. Sensor Diff.
40
0
27
0
-13
2
15
9
25
-7
20
-5
54
0
32
0
-4
-5
26
15
GS NDVI vs. CC NDVI
0.8
0.7
y = 1.3545x - 0.1783
R2 = 0.8986
0.6
GS NDVI
0.5
0.4
0.3
0.2
0.1
0
0
0.1
0.2
0.3
0.4
CC NDVI
0.5
0.6
0.7
0.8
GS RI vs. CC RI
1.3
1.25
y = 1.4428x - 0.4427
R2 = 0.9096
1.2
GS RI
1.15
1.1
1.05
1
0.95
0.9
0.9
0.95
1
1.05
1.1
CC RI
1.15
1.2
1.25
GS and CC Sensor Recommendations at 120 Bushel Yield Potential
140
GS 120 Bushel
Suggested N-Rate lbs/Ac
120
CC 120 Bushel
100
80
60
40
20
0
0.9
0.95
1
1.05
1.1
RI
1.15
1.2
1.25
Winter Wheat N Rate Calculator
Farmer Inputs
NDVI Reference Strip
NDVI Farmer Practice
Max Yield for Area bu/ac
Days from planting to sensing where avg. temp >40 F
Grain Price $/Bu
Nitrogen Price $/lb N
Application Cost $/Ac
Nitrogen Efficiency % recovered
0.6
0.5
70
86
4.85
0.38
5
50
Outputs
Response Index
Yield Potential of Reference Strip bu/ac
Yield Potential without N bu/ac
N Rec. lbs N/Ac unadjusted for G:N price ratio
N rec. lbs N/Ac adjusted for G:N price ratio
Gross Return (no Nitrogen) $/ac
Gross Return (using N Rec) $/ac
1.20
64.6
53.8
35.5
42.6
261.00
297.22
Winter Wheat Yield Prediction 2006-2008
100
y = 12.871e202.23x
R2 = 0.76
90
80
Grain Yield Bu/ac
70
60
50
40
30
20
10
0
0
0.002
0.004
0.006
INSEY (NDVI/DAP) Feekes 4-5
0.008
0.01
Winter Wheat Nitrogen Sensor
Calculator Performance 2006-2009
Location
Year
Manhattan 2006
Manhattan 2007
Johnson
2008
Manhattan 2008
Partridge
2008
Johnson
2009
Manhattan 2009
Partridge
2009
Mean Difference
Sensor Yield Actual Yield Sensor Rec. Optimum N Sensor Diff.
63
45
34
36
66
30
80
47
61
45
31
36
69
21
78
43
78
1
67
47
47
18
35
25
75
0
65
40
44
30
40
60
3
1
2
7
3
-12
-5
-35
-4.5
Issues with sensing wheat
 Planting date and spring vegetation
 In western KS (and I assume OK) more 2nd year and cont wheat
is grown, it’s planted early, more fall-early spring growth, and
the sensor can be used successfully at earlier growth stages
(Feekes 3?)
 With the very early planting common in W Kansas can actually use the
sensor before dormancy in the fall in many fields.
 In eastern Kansas, wheat is planted later (after soybeans), barely
emerges prior to winter and you don’t get adequate levels of
vegetation for sensing until much later in the spring (Feekes
4/5).
 Problem is multiplied with no-till wheat.
Future work on sorghum and wheat
 Testing on farmers fields to confirm the rate calculators work
in the real world.
 Started in 2009, but will be the primary focus of our wheat and
sorghum work in 2010.
 Algorithms seem to work well when little or no N is applied at
or near planting.
 But what about fields where 50% of the expected need was applied
preplant?
 Or a rescue situation where large amounts of preplant N was lost very
early?
Corn work
 In 2007 we initiated a corn study focused on
using the GS and CC at the V-8/9 growth stage.
This study was initiated at Rossville (NE) and
Tribune (WC) under sprinkler irrigation.
 In 2008 added a flood irrigated site at Colby
(NW), and a high yielding dryland site at
Manhattan.
Results from corn work at Rossville, KS 2007&8
Treatment
Total N Applied
Grain Yield
Total N Uptake
2007
2008
2007
2008
2007
2008
Starter Only
20
20
135
82
88
51
120 S+ pre
120
120
219
213
168
146
160
160
160
218
228
154
163
200
200
200
227
234
183
186
120 split
120
120
223
201
182
141
160 split
160
160
224
222
172
152
200 split
200
200
219
226
176
162
120 + GS
125
163
224
224
166
155
120 + CC
125
178
224
222
163
153
120 + Spad
130
143
215
210
167
140
GS + Spad
120
130
214
227
174
148
CC + Spad
120
175
212
230
156
166
Corn Yield Prediction 2006-2008
Corn Grain Yield Bu/Ac
280
y = 18.312e164.78x
R2 = 0.66
240
200
160
120
80
40
0
0
0.005
0.01
INSEY (NDVI/DAP) at V-8,9
0.015
0.02
Corn Sensor Based N Rate Calculator
 Will be finished after this years field work is finished.
 Will be similar to other calculators.
 Hopefully the performance will be as good as the other two.
Additional work on corn
 Late season rescue N applications to corn at V-12 to 16 using
high clearance sprayers.
 Work started in 2009, both on research farms and with
cooperative farmers and dealers/custom applicators
Loss of preplant N, KS 2009
Denitrification N loss in corn, KS 2009
Issues with corn
 Biomass/sensor saturation
 Purchased a Crop Circle ACS470 to look at the “red edge”
 Working in the row middle to reduce biomass in the sensor
footprint.
Red Edge NDVI vs. Red NDVI from
GreenSeeker and Crop Circle Sensors
0.5
0.45
GS
y = 1.3561x - 0.6628
R2 = 0.97
CC
y = 1.3096x - 0.6389
R2 = 0.96
0.4
Red Edge NDVI
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0
0.68
0.7
0.72
0.74
0.76
0.78
Red NDVI
0.8
0.82
0.84
0.86
Estimating NUE
 Can we estimate expected NUE/Recovery based on soil,
climate and crop-fertilizer management?
 Project initiated in 2008
 Includes both a “data mining” and field component
 Intent is to provide quantitative estimates of the expected NUE
from different management practices as impacted by soils and
climate.
 Using standard practices as base and providing/developing
adjustment coefficients up or down.
Questions?