1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
DETERMINATION OF OPTIMAL NUMBER OF MANAGEMENT
ZONES
A.Melnitchouck
Decisive Farming Corp.
Beiseker, Alberta, Canada
ABSTRACT
There were numerous research carried out to determine the optimal number of
management zones for variable rate application of fertilizer. It was shown that the
number of zones is usually determined by field size and spatial variability of soil
properties or crop conditions. However, in real industry it is necessary to consider
characteristics of field equipment that will be used to apply fertilizer, and also the
economics of variable rate technology. In this study, we investigated the
relationship between within-field variability, number of management zones and
feasibility of product application using variable rate maps with the different
number of zones ranging from 3 to 12. It was found that in many cases the
increase of the number of zones did not improve the accuracy of variable rate
application of fertilizer. In most cases, the optimal number of management zones
varied from 3 to 7. Based on the natural variability, creation of 10-12 zones was
not economically efficient and caused major difficulties for the application of
fertilizer in the field.
Keywords:
Precision agriculture, satellite imagery, variable rate technology,
management zones.
INTRODUCTION
There are many different concepts and methods of delineating management
zones for variable rate application of fertilizer. They are based on utilizing grid
soil sampling, soil electrical conductivity (EC) measurements, yield mapping,
analysis of aerial and satellite imagery, field topography etc. Regardless to the
base method used, there are always two conceptual questions: how to classify the
data, and how many zones to delineate.
The question about the number of management zones was studied by different
research groups. In most cases, they tried to find the best solution for explaining
within-field variability. In particular, Fridgen et al. (2000, 2004) developed an
automated procedure for creating zones and testing the number of zones to create.
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
This procedure was implemented in the software program called Management
Zone Analyst. Another aspect of the problem was studied by Chang et al. (2004).
They examined four management zone delineation approaches: (i) sample areas
impacted by old homesteads separately from the rest of the field; (ii) separate the
field into grid cells; (iii) use geographic information systems or cluster analysis of
apparent electrical conductivity, elevation, aspect, and connectedness to identify
zones; and (iv) use the Order 1 soil survey. Based on a six-year study, they
suggested using both nutrient and yield variability for delineation of zones. Pelcat
et al. (2004) analyzed the optimum number of management zones created from
normalize difference vegetation index (NDVI) using the Fuzziness Partition Index
(FPI) and Modified Partition Entropy (MPE). They found that based on the FPI
and MPE plots, the greatest improvement in classification and organization
between classes occurred at three (3) zones, with additional improvement up to 910 zones, depending on within-field variability.
All these results take into account various layers of information and their
spatial variability. However, when it comes to the practical application of
precision Ag technologies on real farms, there are many other factors coming into
play, such as accuracy of GPS, swath of field equipment, speed of field equipment
etc. In precision agriculture industry, there are various opinions about the optimal
number of zones to create, and suggest number often ranges from 3 to 12 for the
same field.
MATERIALS AND METHODS
Our experiments were carried out in a real farmer’s fields near Calgary, AB,
Canada. We tried to analyze the number of management zones in comparison to
the natural within-field variability and quality of fertilizer application. The main
aim of our study was to determine, what information must be taken into account
to make decisions about the optimal number of management zones, and how
feasible is the accurate application of fertilizer depending on the number of
management zones and field equipment.
For delineation of management zones, we used over 25 years of multispectral
satellite imagery, yield data and field relief. Based on the concept that variable
rate application of fertilizers is aimed to optimize the balance of nutrients in soil
for best development of field crops, the natural within-field variability was
estimated as variability of green biomass and yield within field boundaries.
For each field used in the analysis, we calculated variable fertility index, i.e. an
average potential of land to produce yield. Based on the fertility index, in each
field we delineated from 3 to 12 management zones. Composite soil samples were
collected from each zone and analyzed for the main characteristics including soil
organic matter, pH, concentrations of nitrogen, phosphorus, potassium, calcium,
magnesium, and micronutrients. For each zone, site-specific fertilizer rates were
created. Fertilizers were applied using grower’s field equipment. After the
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
application, as-applied files were collected from the equipment and overlaid with
fertilizer recommendations for further analysis.
RESULTS AND DISCUSSION
Our results obtained from several years of field tests indicate that the number
of five management zones is the most realistic number that fits both spatial
variability of most fields, and technical characteristics of field equipment.
Visual comparison of our zone maps with as-applied data indicated relative
accuracy of the field operation (Fig. 1). However, on the edges of each zone we
can clearly see dents related to the errors of field application. The average length
of the dents was approximately 30 m. In this particular field, carbamide was
applied by an air seeder, and look-ahead time was set up to 2 seconds. Even in
this case, the seeder was unable to ensure 100%-accurate application of the
product. If we compare the size of the dents and area covered by them with the
total size of management zones and whole field, we can state that five
management zones in this case was a realistic number fitting both within-field
variability and functionality of field equipment.
Rx N4600, lbs per acre
500
0
500
1000
1500
2000
As-applied N4600, lbs per acre
2500
Rx N4600, lbs per acre
109
(4.7 ac.)
133
(12.4 ac.)
170
(30.8 ac.)
220
(89.8 ac.)
228
(125.0 ac.)
112
113
114
115
116
117
118
119
120
121
3000
3500
4000
4500
5000
5500
6000 Feet
As-applied N4600, lbs per acre
109
(3.5 ac.)
133
(12.0 ac.)
170
(31.4 ac.)
220
(91.9 ac.)
228
(124.7 ac.)
Fig. 1. Comparison of fertilizer recommendations and as-applied data.
Increase of the number of zones to from 5 to 10 creates serious difficulties for
the application of fertilizer in this field. It is easy to see that the size of dents on
the edges of management zones exceeds the width of the zones (Fig. 2).
Analysis of fitted variogram for fertility index in this field indicated it reached
the sill at approximately 30 m (Fig. 3). It gives a good idea about the smallest
details that we need to take into account when managing this field on a sitespecific basis.
Rx N4600
500
0
500
As-applied N4600, lbs per acre
1000
Rx N4 6 0 0
10 0
10 5
11 0
12 0
16 0
18 0
21 0
23 0
24 0
25 0
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
1500
( 4 .5 ac .)
( 1 7 .8 a c .)
( 1 7 .9 a c .)
( 5 4 .3 a c .)
( 4 3 .3 a c .)
( 1 6 .3 a c .)
( 3 8 .2 a c .)
( 3 7 .3 a c .)
( 2 1 .3 a c .)
( 1 1 .9 a c .)
2000
2500
3000
3500
4000
4500
5000
5500
6000 Feet
As-applied N4600, lbs per acre
109
(3.5 ac.)
133
(12.0 ac.)
170
(31.4 ac.)
220
(91.9 ac.)
228
(124.7 ac.)
Fig. 2. Comparison of fertilizer recommendations with the increased number of
zones and as-applied data.
Fig. 3. Fertility index variogram analysis.
There is a point of view in the industry that contemporary equipment can
manage over ten zones within one field. However, our analysis demonstrates that
it can be done in large fields with low variability. Fig. 4 shows the field, where
the equipment has to pass through eight zones within 70-80 meters. In practice,
field equipment cannot ensure such precision, especially without RTK and
sectional control. In addition, increasing the number of management zones in the
fields from 5 to 10 will double cost of soil analysis and considerably increase
labor expenses for soil sampling, if zone method is used. This is particularly
important in Canadian Prairies with high percent of low value crops, such as
wheat and barley.
(132.2ac.)Field Boundary
Fertility Index
72.169 - 80.906 (1.0 ac.)
80.906 - 87.418 (2.4 ac.)
87.418 - 92.197 (7.5 ac.)
92.197 - 95.218 (11.5 ac.)
95.218 - 97.716 (17.1 ac.)
97.716 - 100.063 (20.7 ac.)
100.063 - 102.39 (21.4 ac.)
102.39 - 104.68 (20.2 ac.)
104.68 - 107.23 (19.7 ac.)
107.23 - 112.114 (10.8 ac.)
N
W
500
0
500
E
1000 Feet
S
140
141
142
143
144
145
146
147
Fig. 4. Increased number of zones.
In some cases, operators of field equipment try to increase speed in order to
increase the total area applied per day. In practice, it leads to unpredictable results
(Fig. 5). Excessive speed coupled with incorrect look-ahead time increases the
lengths of dents on the edges of management zones to 50-70 m, and it makes the
whole idea of variable rate application useless.
500
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
0
500
1000 Feet
Fig. 5. Example of very low accuracy of variable rate fertilizer application.
Among the most important characteristics influencing the optimal number of
zones are the factors considered for the analysis and method of zone delineation.
For example, Fridgen et al. (2000, 2004) using principal component analysis
showed that 6-7 principal components explained over 90% of the cumulative
variance. In their experiments, principal component 1 was dominated by soil EC,
organic matter, Mg and K. The second one was related to elevation, slope, pH and
P. The optimal number of management zones delineated from yield data in their
experiments was 4-6, depending on the field variability. These numbers were
based on fuzziness performance index (FPI) and modified partition entropy
(MPE). However, it is important to remember that the goal of management zones
is not to find areas within the field with different properties. The main aim is to
explain, why some areas in the field produce above or below average, and provide
solutions for the most efficient way to manage soil fertility and crop production,
i.e. the most important factor for delineation of zones is variability of grain yield
or green biomass.
Our experiments and practical experience show that when using yield data or
satellite imagery, the average of five management zones give good results for
168
169
both explaining within-field variability, cost efficiency and applying fertilizer in
the field.
(1166.6ac.)Field Boundary
Fertility Index 5
53.822 - 88.856
88.856 - 97.826
97.826 - 105.014
105.014 - 111.826
111.826 - 132.152
N
W
E
500 0 5001000 Feet
S
170
171
172
173
174
175
176
177
178
179
180
181
Fig. 6. Variability of green biomass, five zones.
Fig. 6 illustrates variability of green biomass in 472-hectare field in Southern
Alberta. Delineation of five zones gives a good picture of within-field variability
and helps to adjust fertilizer rates according to the yield goals and soil
characteristics. At the same time, it makes both soil sampling and application of
fertilizer in this field feasible from the technical point of view.
Increasing the number of zones from 5 to 6, and then to 8, add some additional
details to the map, but does not make the application map more manageable (Fig.
7 and 8).
(1166.6ac.)Field Boundary
Fertility Index 6
53.822 - 86.878
86.878 - 95.135
95.135 - 101.596
101.596 - 107.121
107.121 - 113.265
113.265 - 132.152
N
W
E
500 0 5001000 Feet
S
182
183
Fig. 7. Variability of green biomass, six zones.
(1166.6ac.) Field Boundary
Fertility Index 8
53.822 - 84.094
84.094 - 91.409
91.409 - 96.8
96.8 - 101.596
101.596 - 105.804
105.804 - 109.943
109.943 - 115.195
115.195 - 132.152
N
W
E
500 0 5001000 Feet
S
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
Fig. 8. Variability of green biomass, eight zones.
The areas with the medium yielding potential become very narrow and,
therefore, very problematic for successful application of fertilizer in the field.
Width of the zones in transitional areas decreases to 5-10 m. At the same time,
analysis of variogram for this field indicates that spatial variability of this large
field is very similar to the one shown in Fig. 1, with the range just over 30 m (Fig.
9).
Fig. 9. Fertility index variogram analysis.
It supports the statement that in the conditions of Southern Alberta details of
within-field variability smaller than approximately 30 m in size are not important
for delineation of management zones for variable rate application of fertilizer.
Increasing number of management zones often leads to creation of maps, which
cannot be applied in real conditions using existing field equipment, increased cost
of zone soil sampling and reduction of net profit from the application of variable
rate technology.
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
CONCLUSION
The optimal number of management zones for variable application of fertilizer
depends on both within-field variability and technical characteristics of field
equipment. Another important factor is related to the economics of variable rate
technology, i.e. comparison of expenses related to delineation of zones and soil
analysis, and net profit obtained as a result of variable rate technology application.
REFERENCES
Fridgen, Jon J., Newell R. Kitchen, and Kenneth A. Sudduth. 2000. Variability of
Soil and Landscape Attributes within Sub-field Management Zones. Proceedings
of the 5th International Conference on Precision Agriculture, Bloomington,
Minnesota, USA, 16-19 July, 2000, publ. 2001, pp. 1-16.
Fridgen, Jon J., Newell R. Kitchen, Kenneth A. Sudduth, Scott T. Drummond,
William J. Wiebold, and Clyde W. Fraisse. 2004. Management Zone Analyst
(MZA): Software for Subfield Management Zone Delineation. Agron. J. 96:100–
108.
Jiyul Chang, David E. Clay, Charles G. Carlson, Cheryl L. Reese, Sharon A.
Clay, and Mike M. Ellsbury. 2004. Defining Yield Goals and Management Zones
to Minimize Yield and Nitrogen and Phosphorus Fertilizer Recommendation
Errors. Agron. J. 96:825–831.
Yann Pelcat, Brian McConkey, Prakash Basnyat, Guy P. Lafond, Alan Moulin.
2004. Direct Seeding Conference: "The Key to Sustainable Management".
Regina, SK, pp. 42-53.