By Laura Thompson and Joe Luck

During the course of a growing season, farmers often experiment with different products or practices on their operations. Tools like variable-rate controllers and GPS product mapping can make it simple to try different practices, and document what you've tried.

For example, you might try raising or lowering your seeding rate, turning on and off a product such as starter fertilizer, or increasing or decreasing your nitrogen rate. Your test can easily be documented with as-applied maps, such as the ones shown in Figure 1 from a soybean seeding rate study conducted in 2016. If you’ve put such trials in place already, great! You are well on your way to collecting valuable information. As harvest rolls around, it is now time to think about collecting quality yield data and planning how you will evaluate your on-farm trial.

Here are a few points to help make the most of your on-farm experiments.

1. Yield monitor calibration. Before harvesting your experiment, it’s important to first calibrate and set up your yield monitor. This is a critical step, as uncalibrated yield monitors can dramatically alter the reliability of your results. For example, if your lag time or flow-delay setting is not properly set, the yield data points you are recording will not be logged in the correct location in the field. In the experiment in Figure 2, blocks testing five different nitrogen rates were placed in the field in five locations. Each block is 300 feet long. In the zoomed-in inset, the black data points are the yield data points. If the lag time or flow-delay setting is incorrect, yield data points will be shifted, as shown with the red arrows. In this case, if you evaluate the yield data points within the blocks, you would actually be evaluating some corn yield from outside the block area due to the misplaced yield data points. If this happens, there are ways we can address this; however, it is best if the monitor is properly set up in the first place. You can learn more about the importance of yield monitor calibration in this Extension yield map publication.

2. Harvesting the trial. Once you have a well calibrated yield monitor, it is time to harvest your test strips. Many times, test strips may be the length of the field, such as in Figure 1. Other times, a test strip may only be a few hundred feet such as in Figure 2. A good rule of thumb is that test strips should be at least 300 feet long. When harvesting, make sure your header is fully within or outside the test strip. You will want to avoid harvesting a mixed pass, for example — one which was partially with starter and partially without. This can often be an issue with soybean seeding rate studies, where the combine header width may not match the planter width. It’s very important that you aim to drive a consistent speed while harvesting the experiment area, and avoid any sudden increases or decreases in speed.

3. Evaluate data quality. Now that you’ve successfully harvested your experiment, you should carefully examine your as-applied data and yield monitor data.

As-applied data evaluation: When changing rates, such as with the nitrogen rate experiment in Figure 2, the rate controller may not change to hit the target rate instantly. For example, in Figure 3, you can see the target rate (a), how nitrogen was actually applied (b) and an overlay comparison showing the target rate boxes in black overlaying the as applied data (c). We see that there is a shift, and a “ramp-up” and “ramp-down” period. We will need to take this into account when analyzing our experiment (Step 4).

Yield monitor data evaluation: It is a good idea to conduct an initial visual examination of your yield monitor data in order to assess the quality of the data, and identify any issues which may impact your analysis. Data can be visualized in your field data management software of choice (Climate FieldView, AgLeader SMS, Apex, MyJohnDeere, etc.) You may find that there are errors in the data due to incorrect flow delay, places where you sped up or slowed down, or other issues. For example, in Figure 4, we see the result of data errors: In the before image, the pivot road appears jagged or out of focus; and in the after image, we have corrected this by properly shifting data points using yield editing software. If you would like to learn more about yield monitor errors, how to spot them and how to correct them, Nebraska Extension has a new digital agriculture online course available at this Extension webpage.  For a limited time, this course is free to access and allows you to watch videos on these topics and try hands-on activities.

4. Buffer data and determine average yield. There are numerous ways to account for inaccuracies in as-applied data and yield monitor errors. While more sophisticated approaches may be taken, a simple starting point is to buffer the data. In our previous example with 300-foot-long nitrogen rate blocks, we eliminated 50 feet of yield data points from each end of the block, leaving only the center 200 feet for evaluation. This removes yield data points from the areas where the nitrogen rate was ramping up or ramping down, and was not at the target rate.

After buffering data, the yield average for the remaining yield data points can be determined. Having two combine header passes within each block improves the data averaging for yield values. Ideally, you will have several replications of your tests in a field. Evaluating repeated tests (or replications) in a field will give you a lot more confidence in your results. In our nitrogen rate example, we had five replications of our five nitrogen rates.

5. Collect additional data and evaluate economically. While understanding yield impact is a good starting point, the economics of a given treatment should also be considered. The highest-yielding treatment may not always be the most economical. Take into account the yield difference and expected selling price, as well as the products or rates you varied. In our example, we would look at both the yield differences as well as the nitrogen fertilizer rate and cost of nitrogen fertilizer.

Finally, while seeking to understand your results, there may be additional data that would help you answer your questions. For example, collecting soil samples may help you understand treatment differences. The additional data that you collect will depend on what your research question is.

Nebraska Extension has many resources to help evaluate your on-farm experiments. Visit the Nebraska On-Farm Research network website to get started.

Thompson is a Nebraska On-Farm Research Network coordinator and ag technology Extension educator. Luck is a Nebraska Extension precision agriculture engineer.