Crop spectral reflectance to support decision making on crop nutrition

Keywords: nitrogen fertiliser management, spectral reflectance, crop nitrogen status, remote sensing.

This paper reviews the different spectral sensors and associated platforms supplying crop reflectance and enabling decision support on crop nutrition. The focus is on nitrogen (N), an essential element in photosynthesis and several plant-growth processes.

N recommendation systems for crops are generally based on an N-balance approach at the field level. Inputs to the balance can be assessed through crop biomass reflectance measurements. For example, winter cover crop biomass relates significantly to N supply. Crop nitrogen status (CNS) assessment is useful for refining N recommendations during the growing season.

Conventional methods for determining CNS have used laboratory techniques such as Kjehldahl digestion or Dumas combustion as invasive methods. At the leaf level, chlorophyll and fluorescence have been measured using specifically designed field sensors for point-based readings. At the field scale, crop canopy N concentration can be approximated by spectral sensors measurements in visible and near infra-red (NIR) bands. While laboratory techniques are destructive for the crops, ground based sensors suffer from soil interference and illumination conditions dependency, and optical aerial and space-borne sensors from additional cloud contamination. Visible and NIR remote sensing offer the potential to detect the CNS from a few to several hundreds of fields at the same time but may fail to detect N surplus in crop canopies. However, a new generation of satellite constellations such as the Sentinel suite offers unprecedented higher spectral, spatial and temporal resolutions as well as a systematic acquisition scheme to overcome most of these drawbacks. First results highlight their potential to improve CNS monitoring.

Optical remote sensing techniques dedicated to crop N management respond mainly to crop biophysical variables, such as the leaf (or green) area index and the leaf/canopy chlorophyll content, providing information at the canopy level. Several remote sensing techniques are proposed, based either on (semi-) empirical methods (e.g. spectral vegetation indices) or on physically based models (e.g. radiative transfer model inversion). Research projects developed in Belgium illustrate the potential and results of decision support systems (DSS) based on the use of crop reflectance measurements for N fertilisation management before sowing/planting and during the growing season. Perspectives for the use of crop reflectance are discussed in relation to crop N management.

Jean-Pierre Goffart, Walloon Agricultural Research Centre (CRA-W), rue de Liroux 9, 5030 Gembloux, Belgium.

Anne Gobin, Flemish Institute for Technological Research (VITO), Boerentang 200, 2400 Mol, Belgium.

Cindy Delloye, Catholic University of Louvain, Earth and Life Institute (UCL/ELI), Place Croix du Sud 2, 1348 Louvain-la-Neuve, Belgium.

Yannick Curnel, Walloon Agricultural Research Centre (CRA-W), rue de Liroux 9, 5030 Gembloux, Belgium.

28 pages, 3 figures, 3 tables, 74 references