Just How Good Is Yield Data?
Since its introduction, yield monitor data has gone from a novelty to a requirement for everything from variable rate planting and nutrient application to assigning values to productivity zones in fields. Many producers now have 10 to 15 years or more of yield data available for input in decision making. However, as in the case of any data, the first question that needs to be asked is how good is it?
"Early yield monitor data is as good as the operator who collected it," said Scott Shearer, professor and chair, Food, Agricultural and Biological Engineering, Ohio State University. "The way the monitor was calibrated and how the machine was run were the major factors in the quality of data."
RESOLVING TECHNICAL PROBLEMS
A 1997 side-by-side swath comparison of several makes and models of combines and yield monitors conducted by Shearer found differences in measured yield, attributable to machine and operator variability. He recalled one operator who started the wheat harvest season with wheat at 30 percent moisture and never recalibrated again as moisture levels dropped. Shearer suggested that even with the improvements that have been made in ease of calibration, calibration remains more of a concern than innate monitor accuracy. However, there were problems with early monitors.
"I think the manufacturers have done an excellent job of resolving problems such as sensor temperature drift," said Shearer. "They also recognized the importance of placement of impact and moisture sensors for improved accuracy, ease of servicing and better coordination of moisture and mass flow sensing data."
Shearer has visited and revisited yield monitor technology throughout much of his career, working his way through academic ranks to professor and later chair of the Department of Biosystems and Agricultural Engineering, University of Kentucky. Working with a grain elevator gimble in his University of Kentucky lab, he simulated a wide variety of field conditions, including grain flow rate profiles and variations in topography. Research conducted in that lab did much to encourage improvements in yield monitor accuracy.
Yet for all the improvements in the technology, one problem that can't be resolved is unrealistic operator expectations. One of those expectations arises from the name "yield monitor," when in fact, what is monitored is the impact of grain striking the sensor plate.
PHOTO COURTESY OF CLAASCalibration remains the biggest challenge to capturing accurate yield data from yield monitors. "People tend to judge the accuracy of the yield monitor by how well the accumulated mass flow rate compares to the weight of the crop across the scale," said Shearer. "In fact, the yield monitor was intended to measure the variation of yield across the field. What they need to know is how accurate the estimate of yield is for a given geographic location in the field."
This leads to the problem of assumed position for mapping purposes. Although yield monitors gather constant data points for the grain mass flow rate striking the sensor, they don't necessarily reflect the actual mass per unit area produced at corresponding GPS coordinates.
"There is an averaging effect that takes place as the grain moves from the header through the combine threshing and cleaning mechanisms, in particular when the cylinder is full, and finally to the clean grain elevator," said Shearer. "As combines continue to get bigger, there is more averaging over greater areas."
The time delay from where the grain was harvested and where the mass flow data point was collected when the combine is at full field speed can be misleading. The impact of this delay is multiplied with adjacent passes in a field when pass-to-pass yield comparisons are made. This difference can be amplified when making adjacent harvest passes in opposite direction on sloping fields. Shearer emphasized the importance of making adjacent passes in the same direction with all operating parameters as similar as possible when using yield monitors to compare test plot results or treatment difference.
Any correction of these problems is going to come with a cost, suggested the researcher. "Looking ahead, reflectance-based biomass sensors mounted on the header may have some potential for improving accuracy and resolution," said Shearer. "We may even see hybrid systems where mass flow sensors are augmented with relative reflectance sensors."
Since the first yield monitors were introduced, they and their data have come a long way. Where sensor technology and yield monitor accuracy go in the future is an economic decision, suggested Shearer. One thing is certain, the more accurate the yield estimates, the more real value these data will carry.