Editor's note: The following item was provided by Dairyland Laboratories in Arcadia, Wis. It is the first of a three-part series.

The last five years have illustrated a renewed interest in bringing NIR feed testing technology to the farm. While the first attempts at bringing NIR to the farm date back over 20 years, advancements in instrument manufacturing and computing power have made this application more realistic than ever before.

In the following three-part series we hope to shed some light on the current efforts to bring NIR to the farm, illustrate where some limitations are, and propose practical use cases.

Without delving too deeply into how NIR spectroscopy works, the basic concept is that we teach a computer to recognize nutrients in a feedstuff based on the way the sample reflects light. This is a less sophisticated version of how our brains learn to recognize colors, objects, and faces based on the light reflected through our eyes. In either case, all of the information we have to work with is contained in the patterns of light that are reflected from the object we are trying to interpret.

The accompanying chart (shown below) illustrates the amount and quality of information that can be collected from different instrument types and sample presentations. To create this chart we scanned a sample in its “undried and unground” state just as it would be when the sample is taken from the field or storage structure. Then, we dried the sample to remove the moisture and rescanned it. Finally, we ground the sample to a uniform particle size and scanned it a third time. The purpose of this illustration is to show the effect of moisture, particle size, and instrument type on the information available in NIR measurements.

The most important idea that this chart should demonstrate is that moisture has an enormous influence on how much light is reflected from a sample. Because the effect of moisture is so dominant, it is very easy to quantify by NIR. Unfortunately, this moisture peak also covers up a lot of information about the other nutrients in a sample. The difference is visible in the lines on the chart as additional smaller peaks appear when the sample is dried.  This is the principle reason that reputable laboratories dry every sample prior to scanning by NIR. Removal of the moisture is crucial to our ability to measure the 25+ other nutrients included on a complete NIR forage report.

The second idea this chart illustrates is that the particle size of a sample changes how light is reflected. By grinding a sample to a fine powder, a laboratory can remove this variation and gain an additional level of accuracy in NIR measurements. 

Finally, the bottom section of this chart depicts the light range measured by varying quality instruments. Top-end laboratory NIR instruments can cost upwards of $100,000 and cover roughly 2,100 different light wavelengths. Of the instruments currently being promoted for on farm use, the price range runs from roughly $5,000-$30,000 and the number of wavelengths measured range from 400-1,000. To go back to our eyesight analogy, the difference between these instruments is similar to the difference between trying to recognize a face in broad daylight versus trying to recognize that same face through dense fog at daybreak. The more light we have to work with; the more accurate our perception is of the object we are looking at.

Hopefully, we’ve now established the basic concepts of how moisture, particle size, and instrument quality can change the amount of information available in an NIR analysis. In the next part of this series, we’ll take a look at how this information affects the ability to measure moisture and other nutrients. Then, the third and final part of this series will focus on some practical applications for on-farm NIR.

Is there a role for NIR on the farm?