Farms.com Home   Expert Commentary

Yield Data Quality For Post-Harvest Analyses

Nov 21, 2017
 
By Elizabeth Hawkins
Field Specialist, Agronomic Systems
 
Yield Monitor
 
As the number of tools and services utilizing precision ag data to aide in decision making continues to increase, the importance of having quality data is also increasing. Most producers understand the importance of yield monitor calibration for generating accurate yield estimates, but there are other errors that can impact both the accuracy and the spatial integrity of yield data. Spatial integrity of yield data becomes very important when being used to generating prescriptions for fertilizer and seeding. Spatial inaccuracies in yield data become a problem when using yield maps to create management zones and subsequent input decisions by zone within a field. Taking the time to evaluate quality and removing erroneous data ensures prescriptions and other maps based off yield data are correct.
 
When processing yield data this winter, some errors to be mindful of include: header height setting, quick stop-start errors, flow delay setting, and header/ platform width setting. Each of these errors will result in inaccurate yield estimates impacting maps created from yield maps. The following outlines some of the potential errors:
 
Incorrect header height setting
 
An incorrect header height setting can result in data being collected when the combine is not harvesting the crop. This error often results in low or zero yield values being collected in headland or point-row areas while the combine is turning. If this error is not recognized and addressed, the yield estimates for zones that include headland areas will be underestimated due to additional acreage being measured with little or no additional crop sensed. Figure provides an illustration where the addition of the turning data would underestimate yield in these areas. It is suggested the turning data be removed before any zone or spatial analyses be conducted.
 
Figure 1. Incorrect header height setting resulting in data collection in headland turning.
 
Quick Start-Stop Errors
 
The accuracy of the yield monitor calibration depends on consistent flow of material through the combine. Sudden changes ground speed (accelerating or deaccelerating) result in areas where yield estimates are incorrect. Often yield is overestimated where the sudden stop occurred and then underestimated until the material flow through the combine has time to ramp back up. Figure 2 illustrates three areas where the combine was stopped quickly during soybean harvest. These data do represent the actual yield in this area and should be removed.
 
Figure 2. Red yield points in the yield map indicate areas where the combine stopped and then restarted during soybean harvest. Removing the stop-start error in center of this map and then interpolating to a grid changes the 50x50 ft grid value for this spot by 23 bu/ac (soybean yield).
 
Flow Delay setting
 
When the flow delay setting is incorrect, it will result in each mass flow estimate being attributed to the wrong location in the field. This causes not only a spatial error, but also a miscalculation of yield when the combine’s speed is changing. This error is often easiest to spot around the perimeter of the field on the yield map, as shown in Figure 3. It is important to note that the data is offset for the entire length of the field and not only at the edges.
 
Figure 3. The characteristic zig-zag pattern created when the flow delay setting is incorrect.
 
Header width error
 
Having an accurate estimate of the area being harvested is key to having accurate yield estimates. A common problem occurs when the corn or platform header is not being operated at full capacity and the header width setting is not updated to reflect that in the yield monitor. When this happens, the amount of crop estimated is divided over a larger area than actually harvested resulting in an underestimation of yield. These errors most commonly appear in yield maps as straight lines down a pass or in fields with point-rows. One needs to either adjust the harvest width for these points or remove from the map before analyses. Figure 4 presents a partial pass (e.g. partial header width) in soybeans where the header width was set to full-width producing low yield estimates for this pass.
 
Figure 4. Area of a field where yield was underestimated while harvesting a partial platform width because of an incorrect header width setting.
 
These are just a few of the common errors that can cause problems within yield map data. Identifying and fixing these problems can help ensure that you are using high quality data to drive your decisions. We recommend fixing or removing these erroneous data if using yield maps to create the following:
  • Yield zones
  • Yield stability maps
  • Profit maps
  • Nutrient removal maps
  • Fertility recommendations
  • Seeding zones and recommendations
  • On-farm study analyses