New scanner predicts THC levels weeks before harvesting

Scientists from’University of Adelaide (Australia), in collaboration with the German technology company Compolytics, have developed a non-destructive scanning technique capable of predicting the profile of the cannabinoid of a plant several weeks before harvesting.

A predictive tool for cannabinoid profiles

The method, known as hyperspectral fan-leaf reflectance (FLHR), involves scanning intact leaves to capture their light reflection patterns over hundreds of wavelength bands. These «spectral signatures» are then interpreted using machine learning models to estimate the final cannabinoid concentrations in mature flowers.

This diagram shows how the FLHR method developed by the researchers can facilitate cannabis cultivation processes.

Traditional cannabinoid testing methods, such as liquid or gas chromatography, These require sample destruction, chemical reagents, expensive equipment and specialized laboratories. In contrast, FLHR provides instant readings directly in the culture room or in the field.

«This measurement technique was superimposed on machine learning models trained on FLHR spectra and achieved high predictive accuracy that outperformed previous approaches,» explains Dr Aaron Phillips of the University of Adelaide, who led the study published in Industrial Crops & Products. The technique is based on a portable hyperspectral device, enabling rapid, in situ assessments to be made without cutting plants or sending samples to a laboratory.

«The ability to predict cannabinoid profiles weeks before harvest has important implications for cannabis production, enabling growers and breeders to improve product quality, reduce costs and ensure regulatory compliance,» explains Dr Phillips.

Growers can thus identify plants likely to exceed legal THC limits, avoiding costly crop destruction, and give priority to plants with high potential. The tool also speeds up selection cycles, enabling growers to optimize harvest timing according to expected potency, rather than relying on guesswork or repeated laboratory analysis.

The benefits go beyond compliance. As independent reports cited by the study explain, this technology helps growers avoid allocating resources to plants with low cannabinoid potential. It also offers researchers a powerful means of classifying cultivars and selecting optimal breeding plants earlier in the growth cycle, reducing the need for time-consuming phenotyping.

Although today's scanning equipment is cumbersome, Compolytics is working on a compact version similar to a supermarket barcode scanner, designed for everyday use by growers.

The future scanner imagined by Compolytics

The research team also aims to push back the boundaries of how accurately FLHR can predict cannabinoid levels early in the growth cycle. Surprisingly, early results suggest that scans taken at the start of flowering can sometimes perform better than measurements taken later.

For the future, Dr. Phillips envisions larger-scale applications: «We'd also like to test our approach with drones capable of scanning hemp fields to spot plants that exceed legal THC thresholds.»

Such a system could automate compliance monitoring for large outdoor farms, representing an unprecedented level of agricultural precision in the Cannabis industry.