Using Multi-Year Yield Analysis to Create Management Zones for Variable-Rate Seeding
Crop Insights written by Bob Gunzenhauser, Pioneer Technical Applications Manager, and John Shanahan, Pioneer Agronomy Research Manager
Crop Insights written by Bob Gunzenhauser, Pioneer Technical Applications Manager, and John Shanahan, Pioneer Agronomy Research Manager
Before developing a variable-rate seeding (VRS) prescription, it is first necessary to delineate a field into different management zones, or regions of distinct crop yield potential (Butzen and Gunzenhauser, 2009). In the context of precision agriculture, management zones are field areas possessing homogenous features for landscape and soil properties, which lead to similar crop yield potential and input-use efficiency for seed, nutrients, and water (Doerge, 1998; Schepers et al. 2004).
Yield mapping is one approach that has been suggested as a means to delineate management zones (Pierce, and Nowak, 1999). Yield mapping refers to the process of collecting georeferenced (with location coordinates) yield and other crop data (e.g., moisture content) while the crop is being harvested, using a combine equipped with a yield monitor and a differentially-corrected global positioning (DGPS) receiver (Adamchuk, 2004). The georeferenced yield data for a given field can then be entered into a geographical information software (GIS) package, where it can be displayed as a yield map and examined for spatial patterns or subjected to further processing. Many pioneers of precision agriculture have already generated multiple years of yield data, and are now interested in additional ways these data can be processed, interpreted and used. This Crop Insights provides a discussion of the steps involved in conducting Multi-Year Yield Analysis (MYYA) as well as the process for using MYYA to develop management zones for variable-rate seeding (VRS) prescriptions.
Briefly, Multi-Year Yield Analysis is a process whereby multiple years of yield data for a field are combined into a single composite layer in order to view trends in spatial yield patterns across many years and to generate management zones.
Performing Multi-Year Yield Analysis: Within the GIS software, yield values for each year are normalized to 100% of the field average, generating higher relative values for above-average yields and lower values for below-average yields. Relative yield values for each year are then placed into a uniform grid, using the exact same size and location for each year. Finally, a composite Multi-Year Yield (MYY) layer is created by combining all relative yield values from the same grid location (cell) and calculating an average relative yield value and coefficient of variation across relative yield values.
Based upon each cell's average relative yield and level of variation, a label can be assigned to the cell (e.g., low-, medium-, or high-yielding), thereby grouping it with other cells of similar characteristics. Usually, these grouped cells are located in distinct regions, creating spatial patterns in the field that can be used as management zones for site specific management.
The primary advantage of using MYYA to develop management zones is the zones are based on actual yield history. While not a guarantee, previous yield history can provide guidance regarding future productivity. MYYA-derived management zones reflect actual productivity levels versus soil information, which may or may not always reflect productivity potential.
Disadvantages of using MYYA include the requirement of having enough years of collected yield map data to include in the analysis, the potential of poor quality data from prior years, and the inherent limitation of the available data sets not covering a wide enough range of environmental conditions. The user should understand these considerations when generating a MYYA layer for use in management zone development.
The creation of a high quality management zone layer derived from Multi-Year Yield Analysis depends on:
Pioneer research has found that using more years of data can improve the accuracy of zone selection (Figure 2). For example, in a study involving 20 fields in north-central Iowa with four consecutive years of yield data, the most recent year's (Year 4) relative yield levels were grouped into Low, Medium, and High range levels. Then, combinations of previous years (Year 3, Years 2 and 3, or all 3 years) were grouped by the same relative yield levels. Then the area represented by each group and year were compared.
In areas considered "High" in Year 4 only 40% of Year 3's area matched. However, when Years 1, 2, and 3 are combined for the same High group, 57% of that combination's area matches that of Year 4. This implies that having multiple years of data improves the odds of properly locating management zones compared to individual years.
Utilizing quality data in the MYY analysis is also important. One aspect of data quality is completeness. Figure 3 below shows a field with missing yield data. Using data from this example would provide incomplete information when generating a MYY layer. To be considered useful for analysis, a field's yield data should be at least 85% complete.
Additionally, raw yield monitor data files should be examined or filtered to remove yield measurement errors. These errors may be due to incorrect start-of-pass, end-of-pass, operation flow delay, or swath width settings (see Figure 4), or improper yield calibration, especially when two or more harvesters are present in the same field (see Figure 5).
Adjustments to correct for these errors (i.e., start-of-pass, end-of-pass, flow delay, swath width and yield calibration settings) may be made with software supplied with yield mapping systems (Adamchuk, 2004).
To create a management zone layer based on MYY data, various settings or parameters may be adjusted to define the output to the user's needs. These are some common settings found in software capable of creating management zones based on MYYA:
Below Average Cutoff - Percent value of the multi-year normalized value or lower at which the cell is labeled "Low."
Above Average Cutoff - Percent value of the multi-year normalized value or higher at which the cell is labeled "High."
Stability Cutoff - Coefficient of variation percent value at which a cell is labeled "Unstable;" used only when three or more years of data are present.
Cell Size - Size of each cell within the grid.
A primary goal of using MYYA is to generate management zones which represent distinct differences in yield productivity that may warrant management changes. To accomplish this, the user of MYYA must take into consideration the amount of potential deviation from the average required for a management change to be useful and profitable.
Since the Below and Above Average Cutoff values are in terms of percent of average yield, one can consider the ranges they create as potential yield goal levels.
For example, if the average yield goal for a field was 200 bushels per acre, and the Above Average Cutoff was 110%, it could be suggested that areas labeled "High" would have a potential yield goal of 220 bushels per acre or more.
If the user believed that when yields changed 10 bushels per acre or more from this field's average that a management change was necessary, the Above Average Cutoff would be set to 105% (200 bushels per acre * 105% = 210 bushels per acre).
Another example would be a field that had a yield goal of 150 bushels per acre. If the user wished to label any areas above 170 bushels per acre yield potential as "High", the Above Average Cutoff would be set to 113% (170/150 = 113%).
If the user would like to consider areas of unstable yield, the Stability Cutoff function can be used. Lowering the Stability Cutoff will increase the area considered "Unstable", while increasing the number will reduce the area. Setting a Stability Cutoff value to less than 100% will delineate areas of the field that have greater yield variability from one year to the next (Figure 6). This may be useful in determining the level of risk involved in planting these areas, and therefore, seeding rates deemed appropriate by the operator. Setting the Stability Offset to 100% will remove any areas labeled Unstable (Figure 7).
Some GIS software may allow the user to change the Cell Size, while in others the value is fixed. A smaller Cell Size will increase file size but provide for finer spatial resolution. A larger Cell Size will reduce file size but can cause some features to be lost. The user should consider the level of resolution possible from the planter width when selecting Cell Size for the MYY layer. There is no advantage in using a Cell Size resolution of less than 30 feet if the seeder/planter is greater than 30 feet in width.
Multi-Year Yield Analysis tools offer a great amount of flexibility to the user, but careful planning and decision making about one's own yield goals and potential for each field should be undertaken when planning management zones using this technique.
Note - It should be understood that a field might not have much natural variation in yield potential. If it is appears from multiple years of yield data that there is little variation, it is not recommended to raise the Below Average Cutoff and lower the Above Average Cutoff to produce apparent variation that may not be significant enough to justify different seeding rates.
Adamchuk, V., A. Dobermann, and J. Ping. 2008. Listening to the story told by yield maps. University of Nebraska Cooperative Extension Publication EC704.
Doerge, T. 1998. Defining management zones for precision farming. Crop Insights Vol. 8, No. 21. Pioneer Hi-Bred, Johnston, IA.
Butzen, S and B. Gunzenhauser, B. 2009. Putting variable-rate seeding to work on your farm. Crop Insights Vol. 19, No. 15. Pioneer Hi-Bred, Johnston, IA.
Pierce, F.J., and P. Nowak. 1999. Aspects of precision agriculture. Adv. Agron. 67:1-85.
Schepers, A.R., J.F. Shanahan, M.A. Liebig, J.S. Schepers, S.H. Johnson, A. Luchiari. 2004. Appropriateness of management zones for characterizing spatial variability of soil properties and irrigated corn yields across years. Agron. J. 96, 195- 203.