Assessment of crop traits retrieved from airborne hyperspectral and thermal remote sensing imagery to predict wheat grain protein content

Abstract

Wheat (Triticum spp.) is among the world’s most widely grown crops and receives large quantities of nitrogen (N) fertiliser. Grain protein content (GPC) is influenced by genetic, agronomic and weather conditions affecting crops’ physiological status and stress levels; accurate GPC prediction has potential to reduce N losses and improve profit. Success in GPC estimation from remotely sensed plant traits has been limited. For progress to be made, it is necessary to robustly identify imaging spectroscopy-based physiological traits most closely associated with GPC in both experimental and commercial contexts. We present results from piloted hyperspectral flights and ground campaigns at two dryland field experiments with divergent water supply and wide-ranging N treatments, and from two years’ flights over 17 commercial fields planted to either bread (T. aestivum) or durum (T. durum) wheat, in the southern Australian wheat belt. Imagery was acquired with airborne hyperspectral and thermal sensors, with spatial resolutions of approx. 0.3 m and 0.5 m for experimental plots and 1 m/1.7 m in commercial fields. Leaf clip measurements, leaf and grain samples were collected and, in commercial fields, ∼40,000 records obtained from harvester-mounted protein monitors. Crop Water Stress Index (CWSI), solar-induced fluorescence (SIF), reflectance indices and PRO4SAIL radiative transfer model inverted parameters were retrieved for each plot and GPC record location. The photochemical reflectance index (PRI) related to xanthophyll pigments was consistently associated with GPC at both leaf and canopy scale in the plots and transect. In the commercial crops, a gradient boosted machine learning algorithm (GBM) ranked CWSI as the strongest indicator of GPC under severe water stress, while SIF, PRI and inverted biochemical constituents anthocyanins and carotenoids were consistently important under more benign conditions. Structural parameters inverted from the hyperspectral reflectance imagery were not prominent except under severe drought. We attained statistically significant results estimating GPC in unseen samples, with best relationships between predicted and observed GPC of r2 = 0.80 in a model built with thermal and physiological traits obtained from the hyperspectral and thermal imagery.