Site-Specific Farming — Number 2
Soil Sampling and Variable-Rate Fertilizer Application
SF-1176 (2), June 1999
Dr. Dave Franzen, NDSU Extension Soil Specialist
One of the first available pieces of site-specific application equipment has been the
variable-rate fertilizer applicator. The idea of varying the rate of fertilizer across a
field is not new and has been conducted in a rough manner by growers since at least 1929.
These early attempts at site-specific application meant that areas in a field were spot
treated and the boundaries used to separate rates were not easily seen by applicator
operators. Flagging, guidance using swath widths, counting rows and other difficult to
administer methods have been used to try to apply different rates of fertilizers to
different areas of the field. Equipment that allows variable-rate application now performs
its duties automatically along with the application data in the on-board computer.
Together with GPS, fertilizer is applied at the recommended rates with much more accuracy
and assurance than in the past.
To determine whether a variable-rate application is needed, soil sampling is needed to
determine what rates should be used and where the material is to be applied. There are two
types of sampling used to direct site-specific application- grid and zone. Grid sampling
uses a systematic approach and assumes that patterns are more random in a field, while the
zone approach uses a more subjective and intuitive approach and assumes that patterns are
present because of some logical basis. Research in many states, including North Dakota,
concluded that a good sampling grid density to use is about one sample per acre (200-220
feet in a regular grid). This density was selected because it consistently recognized
fertility pattern boundaries and reproduced soil levels similar to greater density grids.
However, many growers are reluctant to use this dense a grid because of cost and choose
instead to use a less dense grid, of 300 feet to 500 feet (3-5 acre grids). Doing this
gives up much of the detail of a recommended grid sampling for economic reasons. Although
some features of the variability of the field are retained, much of the boundary
definition and correctness of values is lost when less dense grids are used, compared to a
200-220 foot grid.
Soil sampling in North Dakota is more difficult and expensive than many parts of the
country because of use of the nitrate soil test. The nitrate test recommended for every
nitrogen responsive crop requires at least a 0-2 foot deep sample, often divided into a
0-6 inch and 6-24 inch depth. The depth, the multiple depths per site, and the annual
nature of testing make soil testing in North Dakota much more expensive than in states
where P, K and soil pH are analyzed on a 0-6 inch core every three to four years.
Fortunately, research in North Dakota has shown that grid sampling may not be necessary
for most fields to describe fertility boundaries or reproduce soil test levels to direct
site-specific application of nutrients. Figure 1 shows that at Valley City, nitrate
patterns in the soil were similar between years. Figure 2 shows that the nitrate levels
were related to topography, so the field might be sampled in zones rather than grids,
using topography as the basis for sampling. There was a strong relationship between
topography and soil test N levels at each of the North Dakota sites at Mandan, Colfax,
Gardner and Hunter when soil variability was present.
Figure 1. Valley City NO3-N levels, 1994 and
1995. 1994 crop was wheat, 1995 was sunflower. Patterns persisted through 1998, when the
study was concluded. (17KB color image)
Figure 2. Valley City NO3-N levels, 1995, over
topography. (13KB color diagram)
Nutrients are related to topography because of the way that nutrients move in and over
the soil. Organic matter is usually higher in depressions than on hilltops, and water
moves away from hilltops towards depressions both from surface movement and subsurface
movement. Nutrients like nitrate, chloride and sulfate tend to accumulate in depressional
areas. In eastern North Dakota where rainfall is higher than in the west, microorganisms
may transform nitrate to unusable nitrogen gases in depressions in the late spring and
summer through a process called denitrification, decreasing the amount of N available in
wet years, nitrates may be highest on upland landscapes and lowest in depressions. Both
situations have been seen in North Dakota research.
The following are criteria for choosing grid sampling over a zone approach:
- The field history is unknown.
- Fertility levels are high due to high rates of fertilizer application.
- There is a history of manure application.
- Small fields have been merged into large fields.
- Non-mobile nutrient levels are of primary importance (P,K, Zn).
The following are criteria for choosing zone sampling methods over grid
sampling:
- Yield monitor data or remote imaging show a relationship with topography.
- There is no history of manure application.
- Relatively low fertility levels are present, or low fertilizer rates of non-mobile
nutrients (less than maintenance) have been applied over the most recent years.
- Mobile nutrients, especially N, are important to map.
It is not easy determining where to draw the line between one topography position and
another. To improve the decision, more than one zone development tool should be used.
Aerial photography and/or satellite imagery can be used. Figure 3 shows the Valley City
field using a July SPOT satellite image. This image reveals many of the zones seen in the
intensive soil sample nitrate results. The differences are caused by water and nutrient
movement within the landscape. Figure 4 shows the electrical conductivity maps from two
different sensors — an EM-38 unit and the surface readings from the Veris
conductivity detector. Both sensors gave similar patterns of conductivity and closely
resembled patterns from both the SPOT image and nitrate testing. Between the imaging,
conductivity and topography, four to five nutrient management zones could be established
that describe a large amount of nutrient variation within the field. The least dense grid
that comes consistently close to the same information as the zone approach on this and
other research farms is one sample per acre.
Figure 3. SPOT satellite image of Valley City, 1998 wheat.
(14KB color image)
Figure 4. EM-38 and Veris conductivity readings, 1997.
(13KB color image)
Using a zone approach to nutrient testing results in a large amount of information
about field nutrient levels and patterns with minimal costs. Once topography and
conductivity are measured in a field, it is unlikely that these measurements would need to
be repeated unless there was an unusual event such as field leveling, or in the case of
conductivity, a large manure application.
Regardless of the sampling method, zone or grid, multiple cores should be taken from
each sampling area. In the grid approach, 8 to12 cores are necessary, especially when soil
test levels are high, to reveal the nutrient value of each grid point. Research suggests
that these samples be taken in a point-sampling technique, with all cores obtained from an
area 10-20 feet in radius. In the zone approach, also 8 to 12 cores should be taken, with
the cores randomly obtained from the entire zone area, not from a small location within
the zone.
Variable rate application is possible today using not only large floater-style
equipment, but also with smaller, grower-owned equipment such as sprayers, air-seeders,
ammonia applicators, granular applicators and planters. Variable-rate application can be
profitable if fields are variable in some factor that is controllable and makes a
difference in yield, if the variability is in a range where varying the rate will result
in an increase in yield or a savings in product cost, and the variability is revealed
through sampling and sensing. Variable-rate application will not pay if there is no
variability, if the variability is all in the high range, or poor sampling techniques have
not revealed the true nature of fertility levels in the field.
SF-1176 (2), June 1999
|
