Use of field-based infrared spectroscopy in crop sensing equipment has the potential to significantly improve efficiencies in identifying plant nitrogen (N) content and water-soluble carbohydrates (WSC) across wheat and barley crops.
A two-year SAGIT-funded University of Adelaide project, New field-based tools (near infra-red) to rapidly assess crop nitrogen and stress status, aims to predict total N content from field and laboratory-based near infra-red (NIR) devices and provide a single ‘accurate and robust’ calibration for crops.
Agronomist Michael Zerner says testing this year from more than 400 plots and about 1000 plant samples of wheat and barley – with multiple sowings in varying environments at Roseworthy, Mintaro and Loxton – is expected to confirm promising results from 2016 trials that will build confidence in the data and lay the foundations for cost-efficient use of hand-held technology to speed up assessment of plant nutrition across paddocks.
In the first year, a range of wheat and barley varieties were grown under varying fertilizer regimes to establish contrasting N and WSC concentrations and these traits were measured in conjunction with non-destructive field and lab-based NIR spectroscopy of tissue samples prepared for analysis.
Plots were scanned using a hand-held NIR instrument at a height of about 0.5 metres – similar to the method that GreenseekerTM uses to scan vegetative growth – at early and late reproductive growth stages.
Samples of the crops were returned to the laboratory and, after being dried and ground, and subjected to NIR and traditional benchtop analysis. Initial cross-validation methods for in-field assessment of wheat N content, using portable NIR, were positive, with data across all sowing times and growth stages combined and high coefficients of determination obtained between measured and predicted total N content.
“In the initial analysis of NIR data, the results are very encouraging for use of NIR spectral data to predict N content in the field from non-destructive scans and in the lab using prepared ground tissue samples,” Mr Zerner said.
“The calibration regression of dry, ground plant tissue samples for total N content was very accurate across types and growth stages, with most R2 values consistently greater than 0.95 with a predictive error of between ±0.22 and 0.27%.” Accuracy would improve as more data was included in the model.
Mr Zerner says use of the hand-held FieldSpec NIR device to provide a predicted N content percentage ‘real time’ in the field provided some promising results: wheat and barley were found to have good levels of accuracy with R2 values of 0.92 and 0.85 respectively.
“There are some small gaps in all the initial regression calibration plots because of a lack of samples tested in specific N concentration ranges, but this will be overcome as more data is included in prediction models obtained from 2017 trials to capture the full range in N content calibrations,” he said.
Mr Zerner said the field-based NIR technology – a rapidly expanding field – was “the most exciting component of the project”.
“Most current NIR crop sensors only measure two to three specific wavelengths for use in various indices, such as Normalized Difference Vegetation Index (NDVI), but the use of whole spectra from 300 nanometers (nm) to 1100nm is much more powerful and enables more accurate determination of the composition and chemistry of the plant material,” he said.
“After this initial phase of the project, the preliminary results indicate that it will be possible to predict total N% in the field on-the-go (and) once all the lab analysis of WSC is completed, it will be analysed following the same procedure to test the predictive ability of field and lab NIR spectroscopy.” Final results of analysis will be available in June next year.
Michael Zerner is an agronomist with Landmark Pfitzner & Kleinig and remains an affiliate with University of Adelaide for the ongoing project.