9537 Identification and Quantification of Cotton Yield Monitor Errors

Wednesday, January 7, 2009: 1:45 PM
Salons E/F (Marriott Riverwalk Hotel)
Jason C. Head, John Wilkerson, William E. Hart and Philip Allen, University of Tennessee, Knoxville, TN
A Cotton Yield Monitor is a valuable tool for collecting yield map data and making precision farming decisions on a production scale.  Researchers have shown an interest in using cotton yield monitors rather than a weighing boll buggy for collecting production scale variety trial data.  Yield monitors can simplify and increase efficiency of harvesting large-scale varietal test plots; however, current recommendations suggest that the cotton yield monitor should be recalibrated when harvesting a new variety.  In a variety trial there are typically several varieties within a field which would require numerous, time consuming calibrations that make data collection with a cotton yield monitor less appealing.  Additionally, the new breed of pickers with module building capabilities significantly increases the size of the smallest measureable unit of cotton; essentially, one module will be the smallest unit that can be weighed.  Once these pickers become popular among producers production scale research will rely solely on the cotton yield monitor.

This research targeted many environmental and varietal variables to identify potential sources of error in the Ag Leader® cotton yield monitor.  Data were collected in 2007 from two fields planted with two cotton varieties in each field and two fields with a single cotton variety.  The abnormally dry year resulted in an overall poor crop and an ideal environment for harvest with no rain interruptions.  These conditions produced little variation in yield monitor error (absolute monitor error ranged from 0.1% to 6.8% with a mean of 2.0% across 44 loads). The 2008 experimental design includes one large field planted with a total of seven different cotton varieties, more closely representing typical varietal test plots.  2008 yields and harvest season more closely represented a typical year.  The experimental process along with an evaluation of 2007 and 2008 data will be reported.