Spectral reflectance as a basis for in-field sensing of crop canopies for precision husbandry of winter wheat (The SPARTAN project)

Abstract

In general, spectroradiometers measure the spectral characteristics of crop canopies - reflectance, transmittance and absorption of incoming radiation. Historically, simple vegetation indices have been derived from 2 wavebands within the visible and infra-red parts of the spectrum. This 'vegetation index' does correlate with crop parameters but it is hoped that using much larger numbers of wavebands (hyperspectral spectroradiometry) that more accurate determination of crop characteristics could be obtained. Quantification of these hyperspectral characteristics enables a 'spectral signature' of the crop to be obtained. These spectral signatures can then be related to crop canopy size. This report provides an overview of the SPARTAN project (SPectral Analysis Relating To Nitrogen and disease) which aimed to evaluate the use of in-field spectroradiometry to distinguish variation within canopies of winter wheat as an aid to crop management decisions.

Work within the SPARTAN project aimed to evaluate 1) whether crop characteristics can be discriminated with hyperspectral data; 2) whether use of hyperspectral data can improve on the use of vegetation indices such as NDVI. The main emphasis of the project was on the determination of differences in crop canopy size due to nitrogen fertiliser input. Within the SPARTAN project, hyperspectral data were obtained using an in-field, hand-held spectroradiometer measuring in the range 350 - 850 nm.

The project has demonstrated that simple reflectance measurements such as NDVI, although they can be correlated with crop characteristics, are limited in their value. Hyperspectral measurements offer greater scope for determining crop characteristics beyond the range obtainable using simple vegetation indices. The project has demonstrated that good estimates of canopy size can be obtained during the period when major decisions on nitrogen, plant growth regulators and fungicides are made. The relationship between spectral reflectance and canopy size offers the opportunity to develop automatic, routine measurement of crop canopies which can be incorporated into simple models which can be used to generate application maps. Factors such as soil background colour and type, varietal colour and architecture, which can interfere with the data acquisition, were investigated and found generally to be manageable within the range of crop growth stages required for management purposes. Investigation of factors such as the angle of view of the sensor and the effects of shading on crop spectral signatures were carried out and again, no major constraints to using the technology were found. The basis of simple models which could be incorporated into a spatially variable input system for on-farm use is described.

Spectroradiometry technology is rapidly developing and coupled with our scientific understanding of crop growth and development, offers a real opportunity to develop a routine, accurate and inexpensive crop monitoring system. The project has demonstrated that crop characteristics such as canopy size could be determined remotely. The use of spectroradiometry, either as tractor mounted, air-borne or satellite-borne sensors could thus be of real value in crop management in the near future.