Lentil grain quality and segregation opportunities in‐field using remote sensing

Abstract

AbstractFor broad‐acre crops grown in Mediterranean‐type environments, variation in lentil (Lens culinaris) yield and quality occurs due to seasonal abiotic and biotic stresses. Because grain quality affects the price paid to growers, in‐season assessment of likely final quality using remote sensing technologies could limit economic losses by informing spatial management at harvest. For a survey of lentil crops grown in southern Australia, in 2019 and 2020, Moran's I analysis identified significant field spatial autocorrelation for the grain quality traits of grain protein concentration (GPC), grain size, and grain brightness (CIE L*, where CIE is International Commission on Illumination), indicating an opportunity for zoning at harvest. Partial least squares calibration models of observed grain quality and proximal reflectance spectra were successfully derived for grain weight (R2 = 0.80), GPC (R2 = 0.80), and CIE L* (R2 = 0.86). For late senescence, Sentinel‐2 satellite canopy reflectance, grain size was best predicted (R2 = 0.79) and GPC was poorer (R2 = 0.42). Spatial maps of fields for grain size, informed by models, could be derived and determined that for the market critical threshold (38 mg), field area that exceeded this threshold ranged between 30% and 94%. Overall, we determined that sensing technologies had utility for mapping lentil grain quality across fields, providing a potential tool for growers to selectively harvest to achieve best aggregate price based on grain quality targets. Further calibration and validation with multiple years and locations is also needed to test model stability and application to varying environments.