New technique accurately predicts cannabis crop potency

The technique, known as fan leaf hyperspectral reflectance (FLHR), involves taking measurements across a plant’s canopy during early and late flowering periods and was found to reliably predict the final cannabinoid content in the mature plant.

“This measurement technique was overlayed with machine learning models trained on FLHR spectra and achieved high predictive accuracy that outperformed previous approaches,” says the University of Adelaide’s Dr Aaron Phillips.

“Importantly, our method utilises a hand-held, non-destructive hyperspectral device that enables rapid, in situ assessment of intact fan leaves without the need for sacrificial sampling or laboratory analysis.”

FLHR uses specialised lighting and detectors to capture subtle differences in light reflected from a plant’s leaves.

“Every chemical and structural feature in a leaf – such as pigments, water, and sugars – affects how it reflects light at different wavelengths,” says Dr Phillips, whose study was published in Industrial Crops & Products.

“By scanning across hundreds of narrow bands of light, we can record a detailed “spectral signature” for each leaf. These spectral signatures can then be used to develop mathematical models that can predict all sorts of plant traits.”

Dr Phillips says there are a range of industry applications for the technology.

“The capacity to predict cannabinoid profiles weeks before harvest has significant implications for cannabis production, enabling growers and breeders to enhance product quality, reduce costs, and ensure regulatory compliance,” says Dr Phillips.

“This is particularly important for industrial hemp growers, whose crops are subject to strict THC limits, and medicinal cannabis operations, which can now track and predict their yields.”

The development of an accurately predictive tool will also help growers avoid expensive regulatory penalties.

“This tool can be used by growers to eliminate plants that violate THC regulations, potentially saving entire crops from destruction,” Dr Phillips says.

“Growers can also use the measurements to inform their decisions in the field, avoiding investment in low-quality and therefore less profitable individuals or to speed up grow cycles or breeding procedures.”

Dr Phillips says future work on the technology will include more genotypes and test the earliest timepoint that FLHR profiles can accurately predict cannabinoid content of flowers at harvest.

“We want to know whether we can, for example, measure FLHR on four-week-old plants and obtain reliable estimates of cannabinoids at final harvest,” he says.

“We were surprised that for some cannabinoids the early measurements in our study did a better job of predicting cannabinoids than the later measurements. So, we just don’t know how far back we can push the measurements.

“The current tech is quite bulky and cumbersome to use, so Compolytics are also working on the deployment of hand-held devices, similar in size and shape to a barcode scanner you might find at the supermarket.

“We would also like to test our approach with drones that can scan fields of hemp to find plants that exceed legal THC thresholds.”