6.4 Variability
in Grain Crop Yields Based on Landscape Position and Other Attributes
Investigators
Richard
I Barnhisel,Agronomy, rbaarnhis@ca.uky.edu
Tom Mueller, Agronomy, mueller@pop.uky.edu
Scott Shearer, Biosystems and Agricultural Engineering, shearer@bae.uky.edu
Burak Ferhatoglu, Laboratory Technician, bferhato@ca.uky.edu
Cooperators
Tom
Luck and Joe Dan Luck, Luck Farms, Hopkins County, Tom_Luck@hotmail.com
George Kelley, Hopkins County Extension, gkelley@ca.uky.edu
Bernard Peterson, Nelson County, blpete@bardstown.com
Ronald Bowman, Nelson County Extension, rbowman@ca.uky.edu
Mike Peters, Woodford County Research Farm Manager, mpeters@ca.uky.edu
Gifford Turner, Turner Farms, Todd County, gifturn@vci.net
Curtis Judy, Todd County Extension, cjudy@ca.uky.edu
Mike Ellis, Worth and Dee Ellis Farms, Shelby County, wdemike@iglou.com
Introduction
It
is has been established from earlier and current research projects that corn
yields are affected by topographic variations and the associated changes with
respect to variable seeding and nitrogen rates.
These terrain attributes interacted with these variable input parameters
and certainly topography will affect grain yields.
This is likely the result of creating differences in available water as
well as it has affected soil development over several millenniums.
Changes in yield potentials are also influenced by differences in past
erosion as a result of management. This
effect due to the management factor will vary from field to field since the
applied management will vary from grower to grower as well as over time by
changing of practices such as the result of using no-till practices.
Furthermore the effect of past management as well as terrain attributes
are likely to vary over time as well as between grain crops.
With the development of
precision agriculture technology and the availability supplied by yield monitors
and software, there is the potential to allow through statistical models the
evaluation of the effect of variable seeding and nitrogen rates over time.
This project is designed to test the effect of these conditions first on
variations in corn yield and subsequently between grain crops.
This project should have a significant impact on economic benefit of
precision agriculture. We plan to
“dove tail” this project with the existing study at a few of the existing
locations as well as to seek data from other growers of selected fields.
It will be proposed to look at locations across the state.
Cooperators for the new locations will be selected in the second year of
the project once the model to evaluate the yield data has been established.
Objectives
1.
To correlate topographic and other soil attributes such as topsoil and
solum depth or thickness, and soil fertility with corn yield.
2.
To develop a model to set variable seeding and nitrogen rate parameters
based on these attributes that have affected grain yields over time.
Background
As
stated in the introduction, seeding and nitrogen rates have been varied along
strips in several regions in Kentucky (Barnhisel et al., 1997, Barnhisel et al.,
1999). White and Blackmer, 1999,
presented the reliability of corn yield potential as a guide to variable-rate
applications of nitrogen. Nolan et
al., presented data in 1998 on the effect of landscape classes on yield response
to nitrogen. We have had strips
either 24 or 48 rows wide where the effect of these parameters on corn grain
yield have been measured. Plots
have been located in Woodford, Shelby, Nelson-Marion, Hardin, Hopkins, and
Todd-Christian counties. In
general, varying the seeding rate based on topographic position or topsoil
thickness has increased yields up to 30 bu/a.
Over a 7-year average, this has resulted in an increased net return
$13/acre. Nitrogen rates have been
varied at many of these locations over the past two years.
This has resulted in an additional increase in net return of $17/acre.
The increase in net return was based on a constant value of corn of
$2.25/bu and a cost of seed of $1/1000 kernels.
In this case, the economic advantage was derived from differences in
yield between the variable seeding strip and a constant seeded strip of 27,000
kernels/acre. The increase in net
return for sites where nitrogen was also varied was based on the adjacent
variable seeding rate strip.
We
selected the location for these strips in the various counties where the changes
in topographic attributes were uniform for at least the 48 30-inch rows wide
where the study was located. We
also selected the strip to have two or more such changes in topography. These multiple locations of similar terrains were used as
replications. However, this strip
did not represent the average conditions for the entire field.
We tried to find fields that the total area of the test strip encompassed
at least 4 acres. In some cases, the total area was as large as 5 or 6 acres,
but usually this represented only 10% of the entire field.
Simply multiplication of the increase in net return per acre from our
test strip by the total acreage of the field would not have been correct.
Therefore, this study is needed to determine the “real” increase in
net return on a whole field basis. The
advantages we realized are likely the maximum a particular grower could
experience in that particular field. What
growers are more interested in is the economic advantage for the entire field so
they can determine the cost benefit ratio for their entire operation.
A
model has been developed in the past to predict a soil productivity index in the
evaluation of reclaimed prime farmland on surface mines, Barnhisel et al., 1992.
Corn yield was used as the crop in this model.
Since this model also included many of the parameters that we are likely
to use in the proposed model, it should serve as a starting point thereby
reducing the time needed to new model. The
major change required is equating the results to seeding and nitrogen rates
based on the various soil properties as well as including past corn yield from
landscape positions.
Procedures
Sites in Hardin,
Hopkins, Nelson, Shelby, and Woodford Counties have been evaluated over the past
three years as a part of other investigations.
In addition, data from the Peterson Dairy Farms, Turner Farms, Luck
Farms, and Woodford Co. Research
Farm and possibly Worth and Dee Ellis Farms will be used to test the model on a
larger whole farm approach. This
data collection also includes sites from currently funded projects as sites
occurring on these farming operations. However,
not all of these cooperators have the same capabilities of varying seed and
nitrogen rates.
During the first year,
data evaluation and model development will be the focus of this project. The selected producers have yield monitors and archived yield
data that will be made available prior to our selecting them as a cooperator.
Data sets for multiple years will be merged into one composite or layered
map to illustrate the effect of terrain or other attributes have on corn yield. In some cases, data from other grain crops will be pro-rated
and merged into this data set. We
anticipate a model can be developed that will have applicability for all parts
of Kentucky and not site specific. It
is anticipated that this model will have some parameters that will allow for
adjustments from region to region if in fact they are required.
In addition, it has also
not been determined if the entire operation from the Petersons Dairy Farm and
Luck Farms data set will be used in this initial model development phase, but
only fields where multiple years of corn yields will be used initially. As the project progresses, additional fields will be added as
multiple years of data become available. Although
the initial focus will be on fields with rolling topographies data will also be
analyzed on fields that occur in river or creek bottoms as well as
topographically uniform soils in upland positions.
This project will also
require a significant amount of additional grid soil samplings that will lend
support to why yields variations may be occurring within the various fields.
These intensive soil sampling required to establish which soil parameters
are controlling factors for evaluating the effect of the terrain attributes on
grain yield. Soil maps will be generated based on these parameters for the
selected fields. A digitizing board
is available to generate these soil maps if not available from the NRCS.
We hope to obtain a copy of the individual field sheet for this project.
Some of these fields have already been grid sampled, and near the end of
this project they may be re-sampled to determine stability of soil test
parameters. We plan to use GPS
technology to establish sampling locations to be used in following years. Most of the GPS equipment needed is already available, but a
smaller computer to interface with the GPS instrument will be important for
location of soil sampling positions. Re-sampling
will occur at the same grid points previously used in an attempt to eliminate
micro variability if this actually occurs.
A tractor, or “mule”
will be furnished for soil sampling. A
yield monitor on each combine, data loggers, DGPS and GPS receivers, and a weigh
wagon (if needed for calibration), will be required to harvest all fields by the
various cooperating farmers. The
brand of combine or yield monitor does not have to be identical on the
cooperating growers equipment nor does it need to be the same over the time
period data has and will be collected. Variations
in equipment will be noted and their effect evaluated as much as possible.
During the second phase
of the project, the model will be tested on whole field or farm conditions.
Based on the model developed from these data, a whole field or farm
approach to varying both the seeding and nitrogen rates will be used on paired
fields. If capabilities in varying
seeding and nitrogen rates change and are no longer possible, other cooperators
will be sought to implement the approach on a five regions in Kentucky
represented by those currently listed. This
will be done in time to allow us to grid soil sample and process existing yield
data for model improvement or modification.
It is also unlikely that experiments will be conducted on all of these
locations the second year of the project even if capabilities of the cooperating
growers did not change. This is
especially true because funds are not usually made available until after the
planting season first year of the project, which makes it difficult to grid soil
sample all the “test” fields on all the four farms listed.
These other areas will be used to test the model the third year. This will also allow us to adjust the model to set the
various seeding and nitrogen rates based on the correlations obtained the second
year of the project. That is where
the Worth and Dee Ellis farms comes into the project.
We plan to use this as a final verification of the model.
Even though yield data are available, we will use only their soil sample
data to determine if the model can predict the variations that have occurred in
the past.
The software that will
be used is that from SSToolbox, which was purchased on a previous grant,
however, up-grade and the service contract will be charged to this project. The university will furnish labor and equipment for soil
sample that are to be analyzed for available nutrients, texture, organic matter,
and moisture holding capacity, regardless of their being sampled in previous
years. A rigorous topographic
survey will be conducted using a survey grade GPS instrument at many of these
sites. Soil survey data will be
linked to the yield and topographic maps. These
surveys will be geographically referenced.
Soil cores will be collected as the time and cost of sampling permits at
selected locations the second and third year at as many sites as time will
allow. Most of this will be done
prior to planting as obviously once the crop reaches a certain stage soil
sampling becomes difficult.
Expected
Benefits
Benefits assume that
both of our objectives have positive responses.
Firstly, a model can be developed to set seeding and nitrogen rates for
whole fields or farms as we have observed from the test strips conducted in the
past 3 to 5 years. This model will
allow growers to adapt this technology to their operations.
Secondly, that varying these two components will remain a significant
economic benefit when applied to a whole field or farm approach.
This is particularly true since seed and nitrogen are major input cost
factors. This approach assumes that
for rolling terrains, net returns will be increased $10/acre if such fields or
farms have only 1/3 the response we have found in the past when we selected the
most probable region in the field where yield responses would occur.
Deliverables
A model to set variable seeding and nitrogen
rates that can be used by growers with such equipment as yield monitor data over
a few years. Develop programs for
controllers to operate seeding and nitrogen applicators to implement the model
on fields or farms. Prepare at
least two scientific and extension publications.
Demonstrate the model and its benefits at field days.