Many sensors require algorithms/mathematical functions to translate measurements into practical outcomes. In arrays (sensor groups), the agronomic consequences of variations in individual unit performance, while driving an algorithm, remains uncharacterized. Our objective was to study the performance of individual active canopy reflectance sensors outputting normalized difference vegetative index (NDVI) data, used to prescribe the corrective N fertilization rate for winter wheat (Triticum aestivum L.). We hypothesized that sensor output differences would influence corrective N prescriptions when the NDVI data were interpreted with the usual discontinuous, “stair-step”, algorithms. The NDVI data from an eight-sensor GreenSeeker array (Trimble Navigation Limited, Westminster, CO) were obtained at growth stage Feekes 6 for four wheat fields. Individual sensors were coupled with individual liquid N application nozzle sets, and operational settings make possible the independent fertilization of each 0.56 m2 of field area. Two algorithms (A and B) were evaluated. We observed large differences in individual sensor corrective N prescriptions, despite little short range (<1 m) spatial variation in NDVI. Sensor and algorithm diffrences resulted in a wider range in N prescription rates for algorithm B (CV values between 4.3 and 26.2%) than for algorithm A (CV values between 5.1 and 9.9%). One sensor resulted in 18 to 21% lower N prescriptions when driving algorithm A, and 27 to 56% lower N recommendations with algorithm B. Individual sensor performance, and its interaction with data interpretational algorithms/mathematical functions, should be understood before sensor arrays are used to detect, monitor, characterize and map environmental, plant and soil properties.
|Number of pages||10|
|State||Published - Aug 5 2015|
Bibliographical notePublisher Copyright:
© 2015 by the American Society of Agronomy, 5585 Guilford Road, Madison, WI 53711. All rights reserved.
ASJC Scopus subject areas
- Agronomy and Crop Science