Pesticide application coverage training (PACT) tool: development and evaluation of a sprayer performance diagnostic tool

Current operator feedback from in-field pesticide application operations conveys limited information and often does not allow the operator to visualize a true representation of their performance. Farm management information systems (FMIS) typically do not account for overlap, varying application rates across the width of the spray boom during turns, or off-rate errors due to controller response. The pesticide application coverage training (PACT) tool was developed to deploy data analytics methodologies to sprayer operational data collected during field applications. The goal was to compare enhanced feedback via the PACT tool versus data generated from commercially available FMIS software today. Data were collected for multiple Nebraska fields and processed by the PACT program which consisted of a novel MATLAB program developed for this project. The PACT program successfully generated high-resolution as-applied maps and the automated application report further quantified the contributions of these errors to the total error to illustrate how overlap, turning errors or controller response issues may have individually affected application accuracy. PACT program output metrics were compared with current data provided by FMIS software. Field-average metrics were not found to be significantly different when comparing the PACT program to the FMIS output; however, when examining how in-field errors were distributed amongst various application rate ranges, significant differences were noted in comparison to the FMIS output. Thus, the PACT program was able to quantify and illustrate application rate variation due to boom section overlap and turning movements unaccounted for in traditional FMIS software.

Automated summary

The study found that the PACT program could quantify and illustrate the application rate variation caused by overlap and turning movements that were not considered by traditional FMIS software. While field-average metrics did not show significant differences, significant variations were found when examining errors distributed among different application rate ranges.