Decision -making from on -farm experiments: Spatial analysis of precision agriculture data
Abstract
The goal of this research was to adapt spatial regression methods to on-farm trials in a farm management context. Three essays address this issue. The first essay tested different experimental designs and statistical analysis methods with site-specific data under a range of spatial autocorrelation levels using Monte Carlo simulation techniques. Simulations indicated that data usable for farm management decision making could be gathered from limited replication experimental designs if that data were analyzed with the appropriate spatial statistical model. The second essay adapted spatial analysis methods to a series of on-farm trials conducted by farmers. Farmers conducted on-farm trials using candidate experimental designs identified from the first essay. Available data were assimilated into a spatial dataset, analyzed with traditional and appropriate spatial statistical methods, and results provided to the farmers. Two datasets were created for each field study, one using the yield monitor data exported from farm mapping software with the default parameter settings, and another using yield monitor data subjected to a conscious removal of erroneously measured observations and adjustment for location. Each dataset was evaluated using one aspatial and two spatial analysis techniques to determine if farm management recommendations based on differing levels of yield data quality and statistical method differed from one another. Results indicated that differences in farm management recommendations existed between yield monitor data quality levels unless the most advanced specification of the spatial correlation structure was used in the spatial analysis. The third essay evaluated farmer response to spatial analysis including their perceptions of on-farm trials for farm management decision making and expectations for future spatial analysis services. Case study methods were used to gather evidence and make comparisons across cases. Case study results indicated that farmers receiving spatial analysis reports had more confidence in their farm management decisions based upon on-farm trial data than farmers not receiving spatial analysis reports. Not only did farmers make decisions quicker, but they made more decisions than they would have without spatial analysis. Case study farmers opted to use large treatment block split-field designs, while the control group chose the numerous replications associated with split-planter strip-trial designs.
Degree
Ph.D.
Advisors
Lowenberg-DeBoer, Purdue University.
Subject Area
Agricultural economics
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.