|Sudduth, Kenneth - Ken|
Submitted to: International Conference on Precision Agriculture Abstracts & Proceedings
Publication Type: Proceedings
Publication Acceptance Date: 7/19/2000
Publication Date: N/A
Citation: N/A Interpretive Summary: Precision agriculture is a crop management strategy which seeks to address within-field variability. Crop yield maps are required for evaluating the economic efficiency of precision agriculture systems and are an important part of the decision making process. However, plant population has a significant effect, and often in corn the largest effect other than climate eon crop yield. Therefore, it would be useful to measure within-field plan population efficiently. A combine-mounted sensor would be an ideal way to collect corn population data and integrate it with yield data. In earlier work, we designed and fabricated such a sensor that counted corn plants as they were harvested. In this study, we evaluated the performance of this corn population sensor over multiple years and locations. When compared to hand counts obtained at harvest, the sensors tended to underestimate actual population, especially at closer plant spacings and faster combine speeds. However, it was possible to compensate for these errors, allowing the sensor to provide accurate population information. This research will benefit other researchers by documenting the accuracy of population sensors used to obtain research data. It will also benefit ag equipment makers who may be considering the integration of population sensors into their equipment. Finally, if population sensors such as those evaluated here become commercially available, farmers using precision agriculture will benefit by being able to efficiently collect population data, which is important for optimizing precision management.
Technical Abstract: The performance of a previously designed experimental corn population sensor was evaluated over multiple years and locations. Performance was good under most conditions, although population was underestimated at high populations. When compared to hand counts obtained at harvest under all operating conditions, the sensors underestimated population on average by 4.4% (r2=0.93) with a standard error of 3830 plants/ha. However, when operating in test blocks without weak stalks (and/or doubles) at speeds less than 2.5 m/s, the average underestimation was reduced to 0.08% (r2=0.96) with a standard error of 2720 plants/ha. This represents an error of less than two plants in a 10 m transect. In test plots seeded at varying rates, the underestimation and standard error of the predicted population were directly related to the corn stalk feed rate into the sensor. After compensating for this underestimation, sensed population was san excellent predictor of actual, hand-counted population (r2=0.93, zero mean error). When the feed rate was restricted to less than 9 plants/s, the standard error was 1800 plants/ha. This threshold would represent a combine operating speed of 2.0 m/s at a population of 60,000 plants/ha.