The lab assistant pours a cup of grain into the top of an instrument, taps the touchscreen to start a test and then barely has time to look-up before the multi-parameter results pop-up on the display.
For anyone working with quality control in areas such as grain handling, the advantages of rapid and convenient analysis with near infrared (NIR) instruments have long been obvious as an alternative to slower reference analysis and the hazardous chemical waste that goes with it. So if NIR is already proven for jobs such as the analysis of grain, feed, dairy powder, soymeal and myriad other applications, why not soil?
On paper, NIR technology is, in fact, well qualified for testing soil. The principle of measuring the vibration of chemical bonds in a sample coincides with the characteristics of key soil parameters. Useful chemical bonds vibrating in the NIR range include C-H, N-H, S-H and O-H bonds. Calibrations can be developed for parameters using direct correlations with chemical bonds such as soil organic carbon or using indirect correlation for parameters such as pH, CEC, sand and silt. Further, the relatively large sample size (at least 20 grams or more) is good for accommodating the inhomogeneous and complex nature of soil samples.
In practice however, considerable groundwork (pun excused) is still required before NIR for soil testing can start to steal laboratory bench space from traditional chemical testing equipment.
The challenges and the rewards
Firstly, the use of NIR for applications such as grain analysis is based on decades of calibration development.
Artificial neural network (ANN) calibrations are based on many thousands of data points covering many harvest seasons and conditions. The calibrations are so well developed it can be argued that the NIR tests are as good as if not more reliable than reference tests.
In comparison, NIR for soil testing is in its infancy. Soil is also complex when it comes to calibrating instruments as Dr. Andreas Zumdick, global business manager at global leading testing company, SGS, explains: “Soil is highly variable, for instance, geographical variations, composition, texture, pH levels and concentration of nutrients.”
Andreas describes how, in 2019, SGS took the initiative to look closer at the specific challenges involved and the potential for NIR to supplement chemical analysis through a collaborative project with analytical solution providers FOSS. “FOSS is already a partner for instance in grain, so it was obvious starting point,” he says. “We wanted to test if we could use NIR as an extension of commercial services. NIR is faster and there’s no hazardous waste.”
Of particular interest was the ability to test texture and organic carbon in soil to meet a growing trend in requests for tests where speed of delivery is more important than slight differences against reference results. “If double digits after the comma are not critical, then the precision of NIR can be as good as the wet chemistry,” says Andreas, but with the caveat: “As long as the calibration is robust. We therefore needed to evaluate if NIR is applicable for a soil matrix.”
The project teams at FOSS and SGS were well qualified to take on the evaluation in a number of key areas. Given the complexity of soil samples, a lot of data from both chemical reference analysis and NIR tests is required to form the basis of a calibration. SGS could provide much of this data through normal laboratory operations generating a wealth of reference test data every day. In concert, FOSS provided a smart web application for calibration development, data collection from different analysers in different locations and administration of a single project for future data collection and calibration.
“Using a very large amount of sample combined with the ANN calibration method, we were able to get NIR predictions with similar performance compared with the lab reference method.”
FOSS senior chemometrician, Clement Peltre
Another factor providing a good platform for the evaluation project was the network of SGS soil-testing laboratories around the world. Among these, a South African laboratory was chosen for its variety of soil types and cropping situations and the high volume of tests handled.
Organic content of South African soil
The laboratory provides soil analysis data on soil texture, composition and nutrients for farmers and crop consultants. Such data is becoming increasingly important for sustainable farming by helping to ensure the proper use of fertilisers to safeguard yield and crop quality.
Manager of the SGS Somerset West laboratory, Stuart Shepherd gives details: “In South Africa we do over 200,000 samples a year in our lab and soil analysis is used extensively by farmers for their fertilizer applications.
Many farmers are now trying to increase the organic content of their soils to obtain better productivity and replenish what they have taken out of the soil and hopefully start to farm in a more sustainable way.”
Foundation for calibration development
Among the parameters investigated in the evaluation, the one that showed significant promise was the analysis of organic carbon. Soil in South Africa is characterized by high acidity which makes the analysis of carbon highly relevant. Any reduction in use of chemicals and time involved for carbon tests could make a significant contribution to the efficiency of laboratory operations.