UAV-based prediction of ryegrass dry matter yield

ABSTRACT

Forage yield is traditionally measured by manual harvesting, drying, and weighing and has multiple uses, including plant breeding and pasture management. The goal of this paper was to determine the accuracy of unmanned aerial vehicle (UAV)-based prediction of ryegrass percentage cover, vegetation volume, and dry matter (DM) yield in Autumn from 300 rectangular 1.5 m² plots at a height of 20 m above ground level, compared to the current manual method. The secondary goal was to evaluate the UAV-based method for the determination of dry matter yield from five different ryegrass cultivars. A photogrammetry-based technique combined with a spectral method to determine the soil level was used to determine the percentage cover and vegetation volume of ryegrass plots, which were then used to obtain calibration curves to predict DM yield per plot. Calibration curves were obtained for five different ryegrass cultivars, with concordance between calibration curves for four of the five cultivar populations. The relationship between predicted forage volume (m³) and measured DM (g per 1.5 m² plot) for ryegrass Populations 1,2,4,5 had an R² = 0.61. Population 3 was different to Populations 1,2,4,5, with a two-fold difference in DM yield for the same forage volume. This demonstrated that 61% of the variance in DM yield can be explained by forage volume determined by a UAV-based photogrammetry method. We also further tested the methodology from 70 rectangular 2.4 m² ryegrass plots in a Spring trial. The relationship between the predicted DM yield per 2.4  m² plot (based on the average predictions from forage volume and forage area models) and measured DM yield per plot had an R² = 0.66. UAVs can therefore increase the acquisition of field data for research studies and the management of pasture in grazed farm systems.

Acknowledgements

The field trials used in this study were supported by Pastoral Genomics (PSTG1501), a research consortium funded by the Ministry of Business, Innovation and Employment (MBIE), DairyNZ, Beef + Lamb New Zealand, Dairy Australia, AgResearch Ltd., Barenbrug Ltd., and Grasslands Innovation Ltd. We also thank Marty Faville, Jana Schmidt and Peter Moran, who set up the field trials and provided access to samples.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was funded by AgResearch, New Zealand.