Monitoring sugarcane growth response to varying nitrogen application rates: A comparison of UAV SLAM LiDAR and photogrammetry

Abstract

The capabilities and utility of UAV LiDAR and surface from motion photogrammetry have been of wide discussion in the remote sensing community and assumptions made, often speculative, about the potential strengths and limitations of these systems. Here, we employ a side-by-side test of the CSIRO Hovermap LiDAR and Micasense RedEdge multispectral camera simultaneously mounted to a single UAV platform to acquire a time-series data set from both sensors over the growth cycle of a sugarcane crop in northeast Queensland, Australia. The primary aim was to compare the ability of each system to accurately measure crop height, over a single growing cycle. A secondary aim examined the correlation between these measures and the sugarcane biophysical parameters of stalk population (stalks·m −2 ), total fresh biomass (TFB, t·ha -1 ) and cane yield ( Yield , t cane·ha -1 ). The experimental design included a randomised complete block design of four nitrogen fertiliser treatments (0, 70, 110, 150 and 190 Nkg·ha -1 ) with four replications to assess if either optical measure could detect significant effects of nitrogen application on crop growth. Both systems demonstrated similar capabilities for accurately measuring crop heights throughout the growth period with statistically significant coefficients of determination observed when comparing the maximum ( Adj R 2 = .885, F(1,118) = 910.806, p ≤ .001) and mean ( Adj R 2 = .929, F(1,118) = 1548.404, p ≤ .001) crop height estimations of both instruments. In addition, both systems responded similarly in the detection of differences in crop structural properties response to nitrogen treatments. Only Hovermap, however, demonstrated the capacity to obtain sufficient ground returns over the course of the growth period to enable comparisons to the biophysical samples using a ground-to-non-ground return ratio ( Stalk Population: Adj R 2 = .788, F(1,18) = 70.688, p ≤ .001, TFB: Adj R 2 = .713, F(1,18) = 48.198, p ≤ .001, Yield: Adj R 2 = .707, F(1,18) = 46.921, p ≤ .001) with the best results of RedEdge obtained when comparing mean height measures ( Stalk Population Adj R 2 = .502, F(1,18) = 20.172, p ≤ .001, TFB: Adj R 2 = .309, F(1,18) = 9.481, p = 0.006, Yield: Adj R 2 = .322, F(1,18) = 10.028, p ≤ .005). The results suggest that although both systems are comparable for accurate crop height measurements, and as such, provide early detection of potential problems, UAV LiDAR provided more consistent and significant correlations with optical remotely sensed data to the biophysical parameters of sugarcane.

Plant phenotyping relevance

UAV LiDARとフォトグラメトリを用いたサトウキビ草高測定の比較・精度評価が中心であり、バイオマスや収量との関連も検証しているため、植物表現型計測手法研究に該当する。

Code and data availability

公開論文であることは確認できましたが、現在の公式API・許可済み取得経路では本文を自動取得できませんでした。

No evidence-backed public reproduction asset is currently recorded.