Optimizing a Diode Array VIS/NIR Spectrometer System to Detect Plant Stress in the Field

Abstract

AbstractSeveral spectrometer types for the detection of plant stress symptoms by measuring reflectance of crop canopies have been described in the literature. Only a few of them were field‐portable and combine both the visual (VIS) and near‐infrared (NIR) spectral ranges. In particular, NIR spectra enable considerable improvements in assessing crop properties. In the current study, a diode array two‐channel VIS/NIR spectrometer combination was adapted from laboratory to field application and optimized with regard to measuring configuration and procedure. Two plant canopies representing two extremes of morphological differentiation were studied. They consisted of a winter oilseed rape and a grass turf canopy of high vs. low morphological complexity. Derived from grass turf and winter oilseed rape canopy reflectance in the VIS and NIR spectral range vegetation indices (REIP, R850 and SRWI) were calculated. It was shown that the choice of sensor head configuration mainly affected sensitivity and accuracy of measurements, whereas the vegetation indices were mainly influenced by sensor‐target angle, sensor‐target distance, daytime of measurement and cloud coverage. However, most impairments were substantially reduced by using a continuous referencation. Best results were obtained using an open fibre optic cable, sensor‐target distance of 1 m in nadir direction and a measurement interval between 12.00 and 16.00 h at clear sky conditions. To show the applicability of the proposed configuration and measurement protocol, growth traits of a winter oilseed rape canopy were investigated at an early stage of development. Based upon vegetation indices, reductions in shoot dry matter, shoot‐N concentration and shoot water concentration were reliably detected. Conclusively, this paper studies the effect of sensor head configuration, sensor‐target angle, sensor‐target distance and daytime of measurement on observed canopy reflectance in the VIS and NIR spectral range. Additionally, it presents, for the first time, a comprehensive and integrated analysis of the way in which these factors affect vegetation indices (REIP, R850 and SRWI) derived from VIS and NIR spectra. The optimized procedure reduced interferences, expanded operation time and enabled high‐resolution reflectance measurements in the field.