Relationship of Drone‐Based Vegetation Indices with Corn and Sugarbeet Yields

Abstract

Successful adoption of drone‐based remote sensing depends on changes in sensitivity over vegetation indices (VIs) and growth stage(s). During 2017–2018, experiments were conducted to relate between vegetation indices and corn (Zea mays L.) and sugarbeet (Beta vulgaris L.) yields in western Minnesota. Aerial images were collected using an unmanned aerial vehicle (UAV) equipped with a passive light optical sensor (Micasense RedEdge). Using Pix4D software, spectral reflectance data were derived from flights at V6 and VT growth stages of corn, and V10 and V15 growth stages of sugarbeet in 2017. In 2018, images were collected every week from the V4 to R2 growth stages in corn, and from the V4 to V15 stages in sugarbeet. In addition to red normalized vegetation index (RNDVI) and red edge normalized vegetation index (RENDVI), crop height was determined from UAV based digital terrain and digital surface models. Yield prediction (YP) model was derived from the linear regression between crop yield and vegetation indices. For corn‐YP model, R² value increased over the growing period and optimized at the R1 growth stage. Considering 3 site‐years, RENDVI was the best predictor for corn YP than other VIs based on the maximum R² value. For sugarbeet YP, model R² value declined over the growing season and optimized at V7 or V10 growth stages. Considering 4 site‐years, RNDVI was best related to root yield and recoverable sugar yield. Drone‐based remote sensing can be successfully used for corn and sugarbeet YP. Drone‐based remote sensing has potential in corn and sugarbeet YP, but it varied over growing seasons. Core Ideas Drone‐based passive optical sensor can be used to predict crop’s yield. Red‐ and red edge‐normalized vegetation index and crop height are potential indices. Red edge normalized vegetation index best correlated with corn yield. Red normalized vegetation index best correlated with root yield prediction. Over growing season predictability improved for corn but declined for sugarbeet.

Plant phenotyping relevance

UAV画像・光学センサー・DSMから植生指数と作物高を抽出し、収量を予測・比較する手法の実質的な適用と評価が中心である。

Code and data availability

公開本文の所在を確認できませんでした。非公開または購読が必要な可能性があります。

No evidence-backed public reproduction asset is currently recorded.