The development and adoption of precision sampling and variable rate fertilizer application technology has been of great benefit to producers. This technology has allowed for more efficient utilization of nutrients, allowing nutrients to be placed where they are needed and not applied to areas where they are not. However, like many technological advancements, understanding how to implement these technologies is not always cut and dried. Currently in precision ag, we have the ability to collect more data at a greater resolution than ever before. However, how to effectively utilize that ability in a practical way can present challenges.
The concept might be easier understood with a real world scenario. The names have been changed to protect the guilty and any similarities to actual occurrences are completely intentional. An agronomist is looking to utilize yield map data, along with soil test data to create a VR fertilizer application prescription while taking both data layers into account. A dry box will then be used to spread the fertilizer. With today’s technology at our disposal, it is very simple to utilize these two sets of data to make a very detailed, high resolution application map. However, just because the technology to generate the prescription exists does not mean that it is the best approach for this scenario. Let’s look into the practical application of this prescription.
Let’s first examine the soil test data layer. Soil samples were collected on a 2.5 acre grid; therefore our hypothetical 40 acre field is represented by 16 samples. Using mathematical modeling and techniques such as kriging, we can generate a smooth, almost seamless map to represent our soil test data, and can theoretically have tens of thousands of sets of data points for this 40-acre field. However, remember that while the color map generated from this data is filled in based on mathematical models, there are only 16 actual data points.
Now consider the yield data in this example. Depending on the monitor and configuration of the equipment, data can be collected by the combine every second. This leads to about 18,000 to 21,000 yield data points in the same field. If the raw yield data and modeled soil test data are utilized for creating prescriptions, unless a lager grid/cell size is used, the resulting prescription map for will contain 18,000 to 21,000 unique fertilizer rates within this 40- acre field.
While many computers with robust GIS software can handle this data volume today, a larger question remains: is the controller and application equipment up to the task? Many GPS controllers in the cab of application equipment cannot handle more than a few hundred unique rates within a given prescription file without performance issues. Beyond the controller, the other question is whether or not the fertilizer application equipment can meet the demands generated by this level of detail.
If, in our scenario, the combine used to collect the data was only 20 to 35 ft wide and the dry box spreader used to apply the variable rate fertilizer is capable of spreading 60 to 110 ft wide, it could be covering a number of unique fertilizer rates across the spread width. In addition, the spreader is most likely traveling 4 to 10 times faster than the combine, meaning that the spreader could be prompted a new fertilizer application rate as often as 10 times a second. When this occurs, spatial accuracy in the as applied map is usually poor. The rate controller cannot change rates fast enough to keep up with the prescription map change, and the application equipment cannot physically change quickly enough to match these rates. This usually leads to spatial inaccuracies and often the total product applied to the field will have a greater than desired level of error.
While the addiction to high resolution data grows, it is important to be mindful of the intended outcome for that data. While contouring data can reduce the special resolution of the data layer, controller files must be contoured to the practical restrictions and abilities of the application equipment in order to improve the accuracy of the final application.