Teaching Interpretation of Yield Monitor Data Analysis: Lessons Learned from Purdue's 37th Top Farmer Crop Workshop

Introduction

The percentage of grain combines with yield monitors has increased at a dramatic pace. The USDA-ARMS surveys indicated that 29% of 2002 U.S. soybean acres and 37% of 2001 U.S. corn acres were harvested with a combine equipped with a yield monitor. 2004 estimates are close to 50% (Griffin et al., 2004). Now that farmers are collecting site-specific yield data, demand for spatial yield monitor data analysis and on-farm planned comparison trials has increased. Because yield monitor data is easily collected "on-the-go," on-farm planned comparisons are more readily implemented compared to weighing individual loads. On-farm planned comparisons by farmers usually test varieties, tillage alternatives, input timing, or herbicide options, but rate trials could be analyzed with similar methods.

Lessons learned from teaching yield monitor analysis interpretation of planned comparisons, which was introduced in 2004 at one of the longest running Extension programs in the United States, Purdue Extension's Top Farmer Crop Workshop, are summarized in this article. While individual Extension staff members have worked with some farmers to analyze yield monitor data, this was the first publicly announced yield monitor analysis program by any Extension system. Yield monitor analysis service and interpretation was offered to farmers wanting information about their planned comparisons. This service is an example of Extension making use of a new technology.

Because data collected with yield monitors is spatially autocorrelated, specialized statistical methods are needed to make correct inference. The complexity and time requirements of most yield monitor analyses pose challenges with respect to teaching and interpretation of on-farm planned comparison results at Extension events. Techniques and statistical methods adapted by Griffin, Lambert, and Lowenberg-DeBoer (2005) were used to analyze the data. Inferences about treatment effects are compromised when traditional statistical analyses are used in the presence of spatial dependence. Spatial statistics is a tool that produces more reliable results when data are spatially dependent.

Yield Monitor Data Analysis Service

Analysis had to be adapted for each farmer because of differences in the questions asked and on-farm comparison design. Some farmers use classic on-farm comparison designs (e.g., strip trials or split planter trials), but yield monitor analysis can be used with a broader range of designs, including split field and paired field layouts. Therefore, the methodology had to be flexible enough to cater to a broad array of on-farm designs.

Requirements for participation in yield monitor analysis were 1) an on-farm planned comparison was made; 2) the farmer had a specific input management-crop production question to be addressed; and 3) combine yield monitor data was accompanied by other geographical information system (GIS) data layers (Lambert & Griffin, 2004). Data from four farmer-fields met the Top Farmer criteria and were analyzed to determine which treatments reliably provided higher yields. Results were useful for farmers wanting information about variety yields across soils and topography. In the process of analysis and interpretation, one satisfied farmer volunteered to present his interpretations to Top Farmer participants. As of January 2005, five farmers meeting these criteria have contacted Top Farmer analysts for the July workshop.

Teaching Interpretation of Yield Monitor Data Analysis

The future of Extension is not disseminating university research facts to clientele but assisting clientele in understanding and making appropriate decisions with available data. For instance, soil test analysis is conducted commercially, but Extension provides interpretation education. One-size-fits-all standardized answers to clientele questions are no longer suitable. Norland (1990) warns of Extension being more service-oriented than teaching. Extension can have a viable place if distinctive services are provided, most likely in the form of interpretive education.

Yield monitor analysis requires specialized software, skills, and human resources. In light of yield monitor adoption rates and farmer concerns of lack of complementary services such as dealer support and data analysis services, spatial analysis techniques were adapted to provide yield monitor analysis. Extension traditionally disseminates university research, but yield monitor analysis service is different. One attribute of yield monitor analysis is the focus on teaching interpretation skills needed to draw inference from on-farm planned comparison results instead of teaching analysis per se. This is consistent with Astroth and Robbins' (1987) assertion that Extension should not only disseminate information, but also "help interpret and synthesize it in practicable ways."

Future of Yield Monitor Data Analysis

Most U.S. farmers have become expert at assimilating new information, but yield monitor analysis requires new skills and tools. Currently available spatial statistical software is hard to use because it is designed with researchers in mind. Some fertilizer dealerships and crop consultants have started to offer rudimentary yield monitor analysis, mostly focusing on yield averages by soil type, variety, or other categories, with no statistical tests to indicate the reliability of comparisons. No commercial service is currently offering inferential statistical analysis of yield monitor data. Because of the time and skills required, U.S. farmers are typically reluctant to analyze the data themselves.

It will be several years before yield monitor analysis methods are developed and integrated into commercially available farm-level GIS software. In addition, current computer processor speed is a limiting factor in processing the vast amounts of data. Commercial yield monitor analysis is possible when these software and computational constraints are relaxed, but will require continued interpretation education by Extension.

Conclusions and Summary

Spatial analysis is the most difficult step in the yield monitor analysis process, but care must be taken in the interpretation step to make correct management decisions. As with any new information, interpretation is key with respect to incorporating farm management decision-making tools such as yield monitors. At the same time, the lack of interpretive training and services is the bottleneck to yield monitor analysis complementary services.

If Extension is to survive in the 21st century, Extension professionals must stay on the cutting edge of technology and applied research to continue teaching farmers skills they desire. Providing specialized services is one way to develop an audience hungry for learning new skills, such as yield monitor analysis interpretation. The land-grant laboratories used to provide soil test analyses but now mainly provide soil test interpretation education. The value of soil test analysis was quickly realized by the private sector, which offered competitive prices. It remains to be determined if yield monitor analysis will be commercially provided or continue to be an Extension service. Yield monitor analysis will be offered again at the 2005 Top Farmer Crop Workshop.

References

Astroth, K. A., & Robbins, B. S. (1987). Recess is over. Journal of Extension [On-line], 25(3). Available at: http://www.joe.org/joe/1987fall/a2.html

Griffin, T. W., Lowenberg-DeBoer, J., Lambert, D. M., Peone, J., Payne, T., & Daberkow, S. G. (2004). Adoption, profitability, and making better use of precision farming data. Staff Paper #04-06. Department of Agricultural Economics, Purdue University, West Lafayette, Indiana. Available at: http://agecon.lib.umn.edu/cgi-bin/pdf_view.pl?paperid=14656&ftype=.pdf

Griffin, T. W., Lambert, D. M., & Lowenberg-DeBoer, J. (2005). Testing for appropriate on-farm trial designs and statistical methods for precision farming: A simulation approach. In D. J. Mulla & J. A. Swenson (Eds.), Proceedings of the 7th International Conference on Precision Agriculture and Other Precision Resources Management, ASA/SSSA/CSSA, Madison, Wisconsin.

Lambert, D. M., & Griffin, T. W. (2004, September). Some suggestion for producers considering yield monitor data analysis, SSMC Newsletter. Available at: http://www.agriculture.purdue.edu/ssmc/

Norland, E. V. T. (1990). Extension is not just service. Journal of Extension [On-line], 28(4). Available at: http://www.joe.org/joe/1990winter/tp1.html