Journal of the NACAA
ISSN 2158-9429
Volume 3, Issue 1 - July, 2010

Spatial Analysis of Precision Agriculture Data: Role for Extension

Griffin, T.W., Assistant Professor - Economics, University of Arkansas - Division of Agriculture
Dobbins, C.L., Professor and Extension Economist, Purdue University
Florax, R.J.G.M., Professor, Purdue University
Lowenberg-DeBoer, J.M., Professor, Associate Dean and Director, Purdue University - International Programs at Purdue
Vyn, T.J., Professor and Extension Agronomist, Purdue University

ABSTRACT

The role of Extension professionals for assisting farmers conducting on-farm trials with precision agriculture technologies were evaluated as part of a three-year case study that followed the decision making process of U.S. and Canadian farmers. The role of Extension was assessed by direct observation of case study subjects, an end-of-project interview, a yield monitor data analysis service, and yield monitor data analysis workshops. Extension may provide spatial analysis services in the short-run. Regardless of who provides spatial analysis services, Extension has a role in assisting farmers when selecting treatments to test, designing experiments, and interpreting statistical results.
Introduction and Background
Farmers using spatial technologies have enhanced ability to conduct on-farm trials, thus providing an opportunity for Extension professionals to work with them. Three sources of evidence were used to provide insights into Extension’s role in analyzing precision agriculture data. These sources included a case study, a data analysis service, and data analysis workshops. A three-year case study intensively followed the decision making process of five farmers who conduct on-farm trials using yield monitor technology. A pilot yield monitor data analysis service was offered over the same three-year period. Two yield monitor data analysis workshops were conducted for farmers and consultants. Topics relating to Extension’s role were included in the final case study interview to evaluate what role Extension may play in the analysis of on-farm data.
 
Precision agricultural technologies such as yield monitors and global positioning systems (GPS) are being used by farmers. More than 20% of corn, rice, and soybean planted acres in the US are harvested with a combine equipped with a yield monitor (Griffin, 2009). Whipker and Akridge (2009) report that 92% of service providers use GPS guidance with manual control and 56% use automated guidance. Spatial analysis is defined as explicitly modeling the spatial heterogeneity and dependence of the data for making statistical inference and farm-management decisions, typically utilizing regression analysis.
 
Data and Methods
Case study evidence was compiled from three sources: (1) direct observation, (2) formal interviews, (3) workshop participation. The main source of evidence was the three-year formal case study. Casual direct observational evidence was collected by the researchers over the project period and formal face-to-face interview data were collected at the end of the project (Griffin et al., 2008). Case study evidence was supplemented by the three-year Yield Monitor Data Analysis Service and Yield Monitor Data Analysis Workshops. Research was funded by a United States Department of Agriculture Sustainable Agriculture Research and Education (USDA-SARE) Graduate Student Research Grant. 
 
Case Study Evidence
Qualitative case study research methods (Yin, 2003) have been used in farm management research in general and farmers’ use of precision agricultural technology in particular (Popp et al., 2002; Griffin et al., 2008; Urcola, 2003). Case study methods were used to evaluate each farmer as a unit of analysis during the three-year project. 
 
Five farmers conducting on-farm trials in Indiana, Illinois, Kentucky and Ontario, Canada initially volunteered to collaborate in yield monitor data analysis during Purdue’s Top Farmer Crop Workshop (see http://www.agecon.purdue.edu/topfarmer/). Farmers were selected based on their expertise in conducting on-farm trials with yield monitors and were identified as innovators who sought appropriate analysis techniques. Three topics pertaining to Extension were discussed with case study farmers. These topics included: 1) expected source of spatial analysis services, 2) farmers’ willingness-to-pay for spatial analysis services, and 3) the expected role of Extension.
 
Case Study Farmers
Case study farmers have at least seven years experience mapping yields. Farmers D, F, and W received a spatial analysis of their on-farm trials. Farmers P and T did not receive a spatial analysis report prior to the final interview; however, these five farmers were not expected to differ from one another otherwise with respect to use of precision agriculture and conducting on-farm trials.  The relevant demographics of all five farmers are summarized in Table 1.
 
The principal investigators worked closely with Farmers D, F, and W, making farm visits and providing spatial analysis reports including production recommendations based upon on-farm trial data over a three-year period.
 
Table 1: Demographics of Case Study Farmers
 
 
 
 
Experience with technology since the year
Farm
 
Region
Attended YMA*
Began Farming
Adopted Computer
Began Internet/ email use
Began Yield map use
Began conducting OFT**
Adopted Lightbar
Adopted Automated guidance
 
Began using GIS analysis mapping
 
Experimental Group***
D
 
Central Illinois
Both
1979
1997
1998
2001
1987
2003
2009
 
2001
 
F
 
Central Indiana
No
1995
1995
1998
2000
1996
1999
2003
 
2000
 
W
 
Western Kentucky
No
1986
1986
2003
1996
2003
1998
2005
 
2003
 
Comparison Group***
P
 
Western Kentucky
2005
1987
1980
1997
1995
1997
2003
2005
 
2000
 
T
 
Ontario Canada
Both
1995
1990
1990
1994
1995
2003
2005
 
1995
 
*2005 = November 2005, 2007= March 2007, Both = 2005 and 2007 Yield Monitor Data Analysis (YMA) Workshops
**OFT = on-farm trial
*** Experimental Group farmers received a spatial analysis report of their on-farm experiments, Comparison Group Farmers did not receive a report on the spatial analysis of their on-farm experiments
 
 
Results
Expected Source of Spatial Analysis Services
Several possible sources exist for a spatial analysis service if and when it becomes common. Farmers D, F, P, and T suggested that farmers conduct their own spatial analysis if interest and time were available and farm size was sufficient (Table 2). Farmers W, P, and T suggested Extension may have a primary role in providing spatial analysis services if the farmer opts to outsource. Other third-party spatial analysts may include dedicated private consultants or a packaged bundle from service providers. 
 
Table 2: Response to advice for other farmers, source of spatial analysis service, and willingness-to-pay questions
Question
Experimental Group
Comparison Group
D
F
W
P
T
What advice would you have for farmers considering on-farm trials for the first time?
Be careful with data, calibrate yield monitor. Tend to little details.
Make sure to devote enough time to trials.
Consult Extension researchers for guidance on experimental designs.
Prepare for extra time commitment, be patient; garbage in garbage out.
Start slow. Do not expect too much. Yield monitors are tools with limitations. Bounce ideas off someone with experience.
Who do you expect to perform the software portion of spatial analysis of on-farm trials?
The farmer.
The farmer, some farmers may outsource.
Extension because unbiased. Do not want to send hybrid trials to seed companies.
The farmer depending upon skill level and interest. Probably consultants or Extension.
Centrally located. Large farmers may have someone in-house. Latest research associated with Extension. 
What would you expect to pay for full-service spatial analysis?
$3 per acre.
$5 per acre or $500 flat fee per experiment. Doubtful to be percentage of expected value.
Percentage of predicted value. Up to several hundred dollars if timely.
$2 per acre. Maybe on a per experiment basis up to $500.
$5 to $10 per acre or 40 to 50% of net.
 
 
Farmers’ Willingness-to-Pay for Spatial Analysis
Farmers’ willingness-to-pay for spatial analysis services may dictate the source of the service. Case study farmers suggested three different fee structures for spatial analysis including a fee per acre or per experiment, and a percentage of net benefits. Farmers suggested per acre fees ranging from $3 to $10 and per experiment fees of $500 (Table 2). Farmer T suggested 40 to 50% of the estimated value in net returns from the spatial analysis.  
                           
The per acre estimates are in the appropriate range for rudimentarily analyses. However, the suggested fees are not expected to entice qualified analysts to offer inferential spatial statistical analysis unless paid across all farm acreage or at least the acreage on which the decision is to be implemented. If suggested per acre fees were intended to cover spatial statistical analysis of field-scale experiments, it can be inferred that public service programs are the likely source of spatial analysis service and not the private sector unless cost-effective automation is developed.
 
The range of suggested spatial analysis service fees are greater than current fees charged for yield monitor data analysis. Nearly 40% of service providers offer yield data analysis at approximately $1 per acre with only 20% of those reporting making a profit (Whipker & Akridge, 2009). Although the exact nature of “yield data analysis” was not defined, it is suspected the survey question was interpreted as printing yield maps rather than any quantitative data evaluation. 
 
The Role of Extension
Case study subjects suggested that Extension may facilitate much of the opportunity that exists in on-farm trials and spatial analysis including: 1) organize farmer peer group meetings similar to marketing clubs, 2) link farmers to applied researchers, 3) advise on experiment designs, and 4) help farmers select treatments to test (Table 3). The benefits of Extension facilitating the spatial analysis service include access to scientific expertise and analysis supported by public funds. All case study subjects valued the direct linkage with applied university researchers. Farmers did not suggest spatial analysis be directly offered as an Extension service.
 
Table 3: Experimental group farmers’ comments on role of extension
Question
D
F
W
What is the role of Extension?
Supporting role like in marketing clubs, maybe develop yield monitor data analysis clubs by facilitating and setting up farmer peer groups.
Doubtful local Extension would have a role or would facilitate spatial analysis or farmer peer groups. Farmers contact individual professors for specific issues.
Recommendations on experimental designs.
 
 
Evidence from Pilot Yield Monitor Data Analysis Service
In 2003, a pilot project was initiated for a Yield Monitor Data Analysis Service to be held in conjunction with the Top Farmer Crop Workshop at no additional charge to workshop participants (Lambert & Griffin, 2004; Lowenberg-DeBoer & Griffin, 2003). Each year, information explaining the service was mailed to registrants and was available on the conference website. Criteria for participation included that datasets comprise a planned on-farm comparison.
 
In the first year of the three-year pilot service, data from four farmers  were analyzed by two agricultural economics graduate students. One of the farmers volunteered to present their results to workshop participants. Three farmers had their data analyzed by four agricultural economics graduate students the second year. One farmer who sent data prior to the workshop was given the  analysis results the first day of the workshop and presented their results to all workshop participants. In the final year of the pilot program, one participant brought yield monitor data to be analyzed by two agricultural economics and one agronomy graduate student.
 
The Yield Monitor Data Analysis Service essentially served as an experiment to determine the demand by farmers for this service. Although there were no additional costs to participate in the service, very few farmers seemed interested or felt they had the appropriate data. Post hoc hypotheses for the low participation were three-fold and include the timing of the workshop which occurred in late July when decisions based on the yield analysis from the previous year’s data would have already been made, the effort required to assemble the complete set of supporting and treatment data for analysis may have been prohibitive, and farmers had a lack of understanding about the benefits of inferential statistics. Although there was a decline in participation in the service over the project period, farmers suggested that they had an interest in learning spatial analysis techniques to perform at least some analysis of their own data. In response, workshops on analyzing yield monitor data were offered.
 
Evidence from Yield Monitor Data Analysis Workshops
Two yield monitor data analysis workshops were offered to farmers, consultants, and university personnel in November 2005 (Erickson, 2005) and March 2007 (Nistor & Florax, 2007). Topics related to on-farm trials were discussed including field-scale experimental designs, hands-on yield monitor data processing and the importance of proper analysis of spatially correlated data. In all, thirty individuals consisting of farmers, consultants, and university personnel from four US states and Canada attended the workshops (Table 4). At least five farmer-participants have conducted spatial analysis of their own data or provided a data analysis service for farmers. The principal investigators have been contacted by university research and Extension faculty in several states, the agricultural software industry, and commodity groups concerning advice regarding designing, conducting, and analyzing field-scale on-farm trial research.
 
 
Table 4. Profile of yield monitor data workshop attendees
 
November 2005
March 2007
Farmers
11
5
Industry
 
11
University
 
3
                                        
 
 
 
 
 
Discussion
Evidence based on the formal case study, yield monitor data analysis service, and yield monitor data analysis workshops indicate Extension has a role in spatial analysis of on-farm trial data although the role may change as developments occur in the industry. In any case, Extension can provide the scientific expertise needed by farmers to design experiments and interpret results. Extension expertise will be relied upon to provide recommendations on designing experiments because farmers are concerned with salesmen designing answers rather than experiments. Rather than training farmers to conduct statistical analysis, Extension’s roles may be to train private providers of services to analyze data and to teach farmers how to interpret the results. Interpretation of statistical results is a key characteristic of using spatial analysis that may best be handled through Extension efforts.
 
Initially, the spatial data analysis service may be facilitated by or offered as an Extension program due to the high costs of specialized human capital and the relatively low farmer willingness-to-pay. Yield monitor data analysis has a public good aspect because it is expected to lead to more efficient crop production using fewer and more carefully targeted agricultural inputs, and also to increased economic activity in rural areas due to more prosperous farmers. Based on the potential public good of yield monitor data analysis, it is reasonable for Extension to invest resources at least until private services are able to enter the market. Extension has a track record of working one-on-one or at least in small groups with interested clientele; whereas private services are likely to 1) target specific clientele based on potential revenue streams or 2) use of computerized automation of the process in conjunction with other services.
 
Although local county Extension offices are not expected to conduct analyses or serve as a dropping point for data, yield monitor data may be analyzed at specialized workshops similar to whole-farm linear programming (LP) services provided by Purdue’s Top Farmer Crop Workshop where farmers have the option to “farm in the computer”. The LP modeling service has been offered at no additional cost to workshop participants for over 30 years (Candler et al., 1970; McCarl et al., 1977). The model is expensive to update and requires specialized skills to help participants input their farm information and interpret the results. Similar services would be cost prohibitive in the private sector. With current estimated costs to conduct spatial analysis based on computational resources and human capital requirements, fees suggested by case study farmers would not entice qualified analysts to offer the service. However, opportunity exists for Extension to offer the service since human capital costs may be considered a public good until computational and software limitations are reduced. 
 
Summary and Conclusions
Case study evidence suggests that Extension should at least offer guidance on selecting treatments to test at the farm-level, designing experiments, and interpreting spatial analysis results for on-farm crop production trials using precision agricultural technology. Extension may offer spatial statistical analysis services until private sector analysis services develop, potentially with the assistance from university Extension programs. Although Extension does not commonly offer computer analysis services to its clientele there are precedents such as the Purdue Top Farmer Crop Workshop, which has provided whole-farm linear programming analysis for farmers.   Given current costs of offering spatial statistical analysis and the farmer willingness-to-pay, no private sector firms are expected to enter the market; Extension therefore has an opportunity to generate a public good by filling that niche until costs drop and farmers realize the value, so that private providers can enter the market. Once private providers of spatial analyses exist, Extension can provide technical training to the private firms in statistical analysis. Regardless of the source of spatial analysis services, Extension is expected to teach farmers to interpret spatial analysis results. 
                                        
Acknowledgements
A United States Department of Agriculture – Sustainable Agriculture Research and Education (USDA-SARE) Graduate Student Grant Program funded this project. The authors thank the farmer-collaborators who have assisted in this research.
 


References
 
Candler W., Boehlje, M. & Saathoff, R. 1970. Computer software for farm management extension. American Journal of Agricultural Economics. 52(1), 71-80.
 
Erickson, B. 2005. Workshop helps farmers utilize one of their key resources:
information. November 2005 SSMC newsletter. Retrieved February 1, 2010 from: http://www.agriculture.purdue.edu/ssmc/Frames/SSMCNewsletter11_05.pdf
 
Griffin, T.W. 2009. Adoption of Yield Monitor Technology for Crop Production. University of Arkansas Division of Agriculture Factsheet FSA37.
 
Griffin, Terry W., Dobbins, C.L., Vyn, T., Florax, R.J.G.M. and Lowenberg-DeBoer, J. 2008. Spatial analysis of yield monitor data: Case studies of on-farm trials and farm management decision-making. Precision Agriculture, 9(5), 269-283.
 
Griffin, T. W. & Lambert, D. M. 2005. Teaching interpretation of yield monitor data analysis: lessons learned from Purdue’s 37th top farmer crop workshop. Journal of Extension [On-line], 43(3) Article 3IAW5. Retrieved February 1, 2010 from: http://www.joe.org/joe/2005june/iw5.shtml
 
Lambert, D. M. & Griffin, T. W. 2004. Lessons learned from the top farmer 2004 yield monitor data analysis sessions. Top Farmer Crop Workshop Monthly Update September 2004. Retrieved February 1, 2010 from: http://www.agecon.purdue.edu/topfarmer/newsletter/YMA_newsletter_TFCW_9_04.pdf
 
Lowenberg-DeBoer, J., & Griffin, T. W. 2003. The 2004 Top Farmer Crop Workshop adds yield map analysis. September 2003 SSMC newsletter. Retrieved February 1, 2010 from: http://www.agriculture.purdue.edu/ssmc/Frames/sept03PrecisionAg_TopFarmeWkshp.htm
 
McCarl, B. A., Candler, W. V., Doster, D. H., & Robbins, P. R. 1977. Experiences with farmer oriented linear programming for crop planning. Canadian Journal of Agricultural Economics. 25(1977), 17-30.
 
Nistor, A. & Florax, R. J. G. M. 2007. Farmers and consultants receive training in spatial analysis of yield monitor data. Site-Specific Management Center website April 2007 Newsletter. Retrieved February 1, 2010 from: http://www.agriculture.purdue.edu/ssmc/Frames/SSMCnewsletter4_2007.pdf
 
Popp, J., Griffin, T. W., & Pendergrass, E. 2002. How cooperation may lead to consensus assessing the realities and perceptions of precision farming in your state. Journal of the American Society of Farm Managers and Rural Appraisers. 65(1) pp.26-31. Retrieved February 1, 2010 from: http://portal.asfmra.org/userfiles/file/journal/popp26_31.pdf
 
Urcola, H. 2003. Economic value added by yield monitor data from the producer’s own farm in choosing hybrids and varieties. Purdue University, M.S. Thesis.
Whipker, L. D, and Akridge, J. T. 2009. 2009 Precision agricultural services dealership survey results. Staff Paper No. 08-09. Department of Agricultural Economics, Purdue University, West Lafayette, Indiana.   Retrieved February 1, 2010 from http://www.agriculture.purdue.edu/ssmc/publications/2009_Precision_Report.pdf
 
Yin, R. K. 2003. Case study research: design and methods (3rd ed.). Beverly Hills, CA: Sage Publishing. 204 pp.