Gunz And Rows – Soil Sampling – Checking Your Fertility Account
Periodic soil testing is the best way to give your fields’ soil a check-up and get an understanding of where the soil test levels are in relation to known sufficiency levels. I think of soil tests as like an account statement from a bank – if you put money into your account (fertilizer) and took money out (harvested crop), you should have a periodic statement (soil tests) to determine the condition of your account. And while the bank-soil analogy doesn’t work completely (not all applied fertilizer will show up immediately in a soil test), it does help us to understand why soil testing is important to understand our fields’ current status.
Here are a number of questions that are typically raised around soil sampling:
- How often should samples be taken?
- What time of the year should samples be taken?
- At what density (samples per acre) should the samples be taken, and how should the samples be spatially arranged?
- Which interpolation method should be used to estimate soil values between samples?
- What soil test analyses should be used?
The number of years between samples typically ranges from 1 to 4, and some of this is driven by crop diversity and level of farm management. In a general Midwest corn-soybean crop rotation, samples taken every 4 years are common, with the samples being taken before a planned corn crop. 3 years is also used, perhaps when there is a corn-corn-soybean rotation. In higher management and crop diversity scenarios, samples every 1 or 2 years may be called for. There are also some approaches where samples are taken every year on a field but at different locations, providing a “rolling estimate” of results every year.
Timing of Sampling
In some farming areas, soil sampling starts immediately after the combine harvesters roll in the fall. This may be done in order to capture the soil test levels to in turn drive fertility applications prior to the next crop, and often there’s a desire to have this applied before fall tillage takes place.
For others, spring or even in-season sampling can take place. While there are some analyses that are more timing specific than others (ex. nitrate-nitrogen), most soil sample analyses are less timing specific. This can allow soil sampling to be performed when labor is more readily available.
However, due to residue tie-up and breakdown, it is important to take periodic samples during the same general time from one event to the next. If samples have been taken on a field in the fall 4 years ago, taken them again in the fall this year, and the same with the spring. This way, comparisons between sampling events are on equal footing with similar residue-based nutrient availability.
Density and Spacing of Samples
While the first two points can find a fair amount of agreement amongst soil samplers and agronomists, density and spacing of samples can have a range of answers.
Before variable-rate application equipment became available about 25 years ago, soil sampling was performed typically in zones or regions of the field, oftentimes at a lower sampling density (1 sample per 10 or 20 acres). With the ability to change application rates based on a map, VR application equipment has driven the opportunity and need to develop higher resolution soil fertility maps. Soon, grid-based approaches of 1, 2.5, 3.3, 4.4 and 5 acres per samples were generated using mapping software, and samplers were guided to the grid points using mobile GPS-enabled software.
However, some agronomists pointed out that the soils in fields are not perfectly square and grid-like, so why should the sampling patterns be the same? If given other spatial guidance, say aerial or satellite images, yield data from on-the-go yield monitors, digital elevation models, or even USDA soil maps, shouldn’t samples be taken using any number of these external inputs? This lead to the development of zone sampling, which at times may entail fewer overall samples but directed placement of the sub-sample cores which make up the samples.
Which method is better? Like all good agronomy questions, the answer is “It Depends”. Zone sampling appears to have the greatest uptake in areas where soil properties and terrain and/or color link closely, such as in glacial till-derived soils. Sampling by these terrain and color features will often delineate strong differences in organic matter, CEC and pH, which may be key drivers for yield. Phosphorus and potassium may not have as strong of ties to these soil type/terrain properties, however.
In fields that have more homogenous soils and terrain, grid sampling at 2.5 acres per sample is more likely to be found. There may not be as great of terrain and soil color differences in these fields, and therefore zone sampling may not be as good of a fit, and therefore grid sampling is a way to capture a fair amount of spatial variability through the sample results.
I like to use the term “intensive sampling,” regardless of sample spacing. In my home area of southern Iowa, we have fields that may be on ridges with downward slopes towards ditches and waterways. A true grid approach is problematic because sometimes the computer-generated grid sample points may fall on or near the edge of a ditch, waterway or terrace, and doesn’t necessarily represent the soil of that part of the field. I will take liberties to move the sample points to where they can better capture that area of the field’s specific characteristics. The ridge tops may still have a grid or semi-grid arrangement, but specific samples will capture the slopes and “fingers.”
Regardless of interval, timing, and placement, good sampling includes taking a number of sub-samples (cores) that are representative for the area the sample is meant to represent. 6-8 cores or more are typically sufficient. Likewise, having a consistent sampling depth is key, especially if there’s a change that nutrient stratification may be taking place. State recommendations range from 6” to 9”, but regardless of the recommended depth, be consistent about taking samples to that depth.
Because samples are spread apart from each other, there still is a need to estimate the soil test result between samples in order to drive fertility applications. Interpolation is a method of constructing new data points within a range of known, discrete set of data points. The proper interpolation method to be used is another area where agronomists and soil scientists can disagree.
Common methods of interpolation include Nearest Neighbor, Kriging and Inverse Distance Weighting.
There are various settings and approaches to these interpolation methods that may affect the outcome. With interpolation, the estimated value between points will most likely be incorrect, but there are ways to make the estimates “less incorrect”.
As a way to determine which interpolation method (and settings) best minimized the error between interpolated and measured soil test values, Granular Agronomy Science worked with data sets from 5 fields that had soil samples taken at the same time but at different spacing arrangements, for a total of 12 comparisons. We interpolated one set of samples using 38 different combinations of interpolation methods and settings (including the methods listed above), then used a second set of samples to compare its measured soil test values with the interpolation-based estimates at the same location. Error was also determined between the interpolating soil sample point measured values and the estimated values from interpolation. This was to gauge any error due to over-smoothing or fitting by the interpolation method. This analysis was performed on P, K, OM, pH, Buffer pH and CEC soil test values.
Granular Agronomy Science found that the Inverse Distance Weighting method, with an inverse distance power of either 2 or 3 and a maximum search of 10 points generated the least error of the 38 methods across the 12 scenarios. This was found across most of the nutrients as well. Here’s an example for phosphorus:
This map shows the interpolated 2.5 acre sample points (blue) along with the 1 acre sample check points (red), with the interpolated surface based on the 2.5 acre sample points.
Soil Test Analyses
Oftentimes soil analysis labs will provide packages of analyses, perhaps basic versions with the most-desired measurements, and more detailed versions covering macro- and micro-nutrients. These will range in costs and sometimes in turnaround time, but generally a soil test analysis will run $5 to $13 per sample
It is worthy to note that soil testing labs may have varying methods of extraction and units of measure for output (ppm or lbs/acre), so beware when comparing test results from two or more labs. For phosphorus, typical extraction methods include Bray, Olsen (Bicarbonate), Mehlich-1 and Mehlich-3. For potassium, Mehlich-3, Bray, and Ammonium Acetate are typical methods. Each of these may produce different values for the same soil test, and will have different sufficiency levels. Labs adopt these methods based on a) applicability for the regions they serve (ex. Olsen P method works best in neutral to high pH soils) b) historical preferences and c) cost and time of analyses and prep. In recent years, many labs have moved to the Mehlich-3 extraction method due to speed and cost while still maintaining reasonable accuracy.
A basic soil test should include phosphorus, potassium, and pH as bare minimum; most basic packages will also include buffer pH, CEC and Organic Matter. Micronutrients of interest may be zinc, sulfur, and manganese, but others may accompany as well. Find the best balance between cost and results for your situation.
Like sampling timing, it is generally a good idea to stick with a soil testing lab from one sampling even to another to ensure consistent results. If a change of labs is made, look for the next lab to use the same or similar methods as the prior to reduce questions around results comparisons.
There are a number of considerations for proper soil sampling, each of which is an important contributor to the overall accuracy of the tests. Because soil testing is, in a way, a science experiment, consistency is key. Consistent depth, number of core sub-samples, timing, density, laboratory and lab analyses are all important factors. These are taken every 2 to 4 years, so it is worth it to do this right.