Determining the Scalability and Investigating Limitations of Yield Monitor Data for On-farm Research

Download or read book Determining the Scalability and Investigating Limitations of Yield Monitor Data for On-farm Research written by Alysa Annette Gauci and published by . This book was released on 2022 with total page 0 pages. Available in PDF, EPUB and Kindle. Book excerpt: There is limited knowledge regarding the size of on-farm research (OFR) treatment areas needed for yield monitoring technology to accurately estimate yield (i.e., what treatment sub-plot length is needed within an experimental unit). Small OFR designs typically do not account for limitations of the yield monitoring technology such as the dynamics of grain flow through a large combine. The goal of this research was to better understand the limitations of yield monitor data to support on-farm research and provide recommendations to utilizing yield data accurately. The research objectives were to (1) investigate the ability of grain yield monitoring technologies to accurately inform strip trials when frequent yield variability exists within an experimental unit, (2) provide a recommended minimum treatment resolution (TR) length to utilize yield monitoring technology in on-farm research based on combine mass flow dynamics, (3) determine if combine ground speed influences the ability of yield monitoring technology to report yield differences, and (4) assess if shifting yield monitor data impacts yield estimates for different yield monitoring technologies. A combination of six sub-plot treatment resolutions (TR) that differed in length of imposed yield variation (7.6, 15.2, 30.5, 61.0, 121.9, and 243.8 m) were harvested at combine ground speeds of 3.2, 6.4, 7.2, and 8.1 kph, depending on study site (three study sites total). Intentional yield differences in maize (Zea mays L.) were created for each sub-plot by alternating the amount N applied: 0 or 202 kg N/ha. Yield was measured by four commercially available yield monitoring technologies. A plot combine was used to ground truth yield estimates every 7.6 m, and a weigh wagon measured accumulated weight per plot. Data analysis for evaluating strip trials consisted of calculating percent differences between the yield monitoring (YM) technology and weigh wagon, testing the significance of the plotted relationship between YM and weigh wagon (REG procedure in SAS v.9.4), and a two-tailed t-test (alpha = 0.05) to compare accumulated weight estimates across yield monitoring technologies. The ability of the yield monitoring technologies to detect yield variability across varying TR lengths and also the impact of combine ground speed compared was analyzed using the GLIMMIX procedure and LSMEANS statement in SAS v.9.4. Exponential regression curves (PROC NLIN) of the delta yield (average difference between high and low yields) values were used to estimate a minimum recommended TR length. Finally, yield data from the yield monitoring technologies were shifted, and the shift resulting in the highest correlation between YM and plot combine data was used in analyses. Results indicated that yield monitoring technology can be used to evaluate strip trial performance regardless of yield variability within an experimental unit when operating within the calibrated range of the mass flow sensor. Operating outside of the calibrated range of the mass flow sensor resulted in >15% error in estimating accumulated weight and an average overestimation of yield by 23%. The recommended minimum TR length varied from 43 m to 107 m depending on the yield monitoring technology. At small TR lengths (≤15.2 m), there was not sufficient length for mass flow rate to be changed in the combine and ultimately to be measured at the mass flow sensor. Differences in mass flow rate at 30.5 m was detectable by the mass flow sensor, but the magnitude of yield difference was not accurate across all the yield monitoring technologies until longer TR lengths. Combine ground speed did not significantly impact yield estimates (p ≥ 0.31 for all speed interactions, except speed x method due to lack of calibration) or the differences between yields (all speed interactions p ≥ 0.40). Shifting yield monitor data resulted in improved statistical yield comparisons for only one yield monitor while also reducing the minimum TR length recommendation from 107 m to 43 m: a similar TR length compared to the other three yield monitor technologies. A single shift could not be applied per yield monitoring technology, though the need for shifting yield data was minimized at longer TR lengths. In conclusion, yield monitor data can accurately characterize yield for an OFR treatment area using a minimum treatment length of 107 m and when the mass flow sensor is calibrated over a full range of clean grain flow rates.