A DEVICE which would enable real-time information about a crop's nutrient status and generate actionable insights is in the works by CSBP Fertilisers.
The project was originally initiated to investigate a range of technology options including drone and satellite based remote sensing platforms, handheld sensors that take images at a canopy level and other handheld devices such as mid-infrared spectrometers and x-ray fluorescence meters.
Since then it has evolved to focus on the use of the combination of miniaturised handheld near infrared (NIR) spectrometers and machine learning algorithms.
CSBP senior digital agricultural specialist Doug Hamilton said the initial desktop work on various technologies started in early 2015 where they identified near infrared spectroscopy as the best option for further investigation.
"The project then moved into field work during the 2015 growing season in WA, with the use of expensive laboratory grade instruments to validate the technology can produce useable results from data collected from the field," Mr Hamilton said.
"The next step was to evaluate the range of options in the market for NIR spectrometer hardware that provided a balance between pricing point and reliability.
"The project has leveraged a range of internal and external people and skillsets including agronomists, data scientists and chemists, university researchers, spectroscopy specialists, electronic engineers, electrical manufacturing consultants and growers."
Spectroscopy is a branch of science that is concerned with the investigation and measurement of how light, also known as spectra, interacts with matter.
With NIR spectroscopy focusing on light in the near-infrared region, which are the light wavelengths beyond visible light.
A near-infrared spectrometer will shine a broad spectrum of NIR light onto a sample and then register how various individual wavelengths within the broad range of NIR light have interacted with the sample.
Mr Hamilton said the degree to which each of the individual wavelengths of light were absorbed or reflected can be analysed to understand the chemical composition of that sample.
"The most common form of NIR spectroscopy that growers in WA will be familiar with are grain protein analysers," he said.
"To build an understanding of the relationship between light and chemical content, the first step is to collect NIR spectra from individual plants and send those plants to the CSBP Soil and Plant laboratory to be analysed for nutrient content.
"Machine learning algorithms are then used to build correlations between the NIR spectra and the various nutrients analysed in the plant material."
The solution that CSBP has implemented is a NIR spectrometer that connects via bluetooth to a smartphone application - this sends the NIR light measurements, plus user inputs, to a cloud-based processing engine.
This processing engine then returns the result back to the user's smartphone application as well as a user interface hosted with CSBP Fertilisers' Decipher AgTech mapping platform.
To date, CSBP's key focus has been on the nitrogen percentage of cereal crops, as nitrogen is the biggest single nutrient cost to growers and has the greatest potential to increase yield and therefore profit to growers.
"Over applying nitrogen can lead to poor return on investment in that nutrient along with off-site environmental impacts," Mr Hamilton said.
"Growers will also commonly 'play the season' with nitrogen, which means they withhold committing too much nitrogen at the start of the growing season in order to tailor the nitrogen applications as best they can to match the yield potential of the crop given the unfolding season conditions.
"This allows them to decrease costs in low rainfall, poor yielding years and maximise profit in high rainfall, high yielding years."
The device and machine learning algorithms, created by CSBP, are the enablers for measuring the nitrogen percentage in the plant in real-time.
However that is not necessarily an actionable output for everyone as the nitrogen percentage also needs to be interpreted in the context of the size of the plant and the yield potential of the crop to generate the nitrogen status of the crop.
Mr Hamilton said from that nitrogen status they were able to calculate a nitrogen recommendation in kilograms per hectare and present that information in a report to growers.
"Making this output available in real-time can allow growers to apply, or withhold, nitrogen with greater confidence just prior to a finishing rainfall event, this will enable the grower to have the best chance of maximising returns on their investment in nitrogen," he said.
"Most of the time spent to date has focused on collecting data from the field and building models that can interpret spectra, with CSBP able to collect data from plants with a broad range of nutrients and environmental conditions from the field trials conducted each year across WA.
"We also collected samples from trial sites in Victoria, South Australia and New South Wales, with the diversity of plant material and the environments the plants were exposed to enabling us to build more robust models."
CSBP has only recently moved towards working through in greater detail how this concept will be applied at scale, and what additional information is required in order to maximise the value of the outputs to growers.
The company is also still investigating a range of commercial options for the device, bearing in mind that while the device is part of the cost of the solution, the biggest cost, and also the highest value, sits in the modelling that interprets the NIR spectra collected by the device.
Mr Hamilton said CSBP deployed several devices into the field this year as part of a pilot stage of the project to test the nitrogen modelling results across farmers' paddocks and to seek feedback on an early stage solution.
"The modelling and hardware have generally performed well and the feedback has been positive," he said.
"There is still further work to be done on fine-tuning the outputs of the solution before we can go down the path of bringing it to market."
"CSBP is committed to building a solution that growers can have confidence and see value in and is proven to be reliable."
