Young green seedlings growing from dark soil mixed with biochar
Enriched Soil

Which Biochar Best Stops Herbicide Leaching? Nut Shells vs Tyre Char

Metamitron is a herbicide widely used on crops such as sugar beet. Like many agricultural chemicals, it can leach down through soil and reach groundwater and surface water. Biochar mixed into soil can slow that movement by holding the herbicide near the surface. But testing every biochar and every dose with laboratory chemistry is slow and expensive.

A 2026 study in the International Journal of Environmental Science and Technology by Alperay Altıkat, Mehmet Hakkı Alma, Sefa Altıkat and R. Gürbüz did two things: compared how well three biochars reduced metamitron leaching, and tested whether a faster, light-based measurement combined with machine learning could predict the results.

Study at a glance

  • Type: controlled laboratory experiment with soil–biochar mixtures
  • Biochars: from hazelnut shells, apricot kernel shells and waste car tyres, at 5%, 15% and 25% by weight
  • Methods: GC–MS chemical analysis, hyperspectral reflectance (400–1000 nm), and four machine-learning models
  • Source: Altıkat et al. (2026), International Journal of Environmental Science and Technology 23(4), 340

Which biochar worked best?

Chemical analysis by gas chromatography–mass spectrometry (GC–MS) showed statistically significant differences between the biochars. The two plant-based (lignocellulosic) biochars left substantially less metamitron in the leachate than the tyre-derived biochar. The lowest residual concentration, 5.18 micrograms per litre, came from apricot kernel shell biochar at the highest application rate of 25%. Tyre biochar treatments left higher residues overall.

Can light reveal the herbicide?

Hyperspectral sensing measures how much light a sample reflects at hundreds of narrow wavelengths. The researchers found that the 600–690 nm range, in the red part of the spectrum, was the most informative, with reflectance at 670 nm correlating strongly with metamitron concentration (r = 0.81–0.82).

They then compared four models for predicting metamitron levels:

  • Support vector regression performed poorly (R² = 0.016).
  • Random forest was moderately accurate (R² = 0.753).
  • An artificial neural network (ANN) was the most accurate (R² = 0.965) and performed consistently across training, validation and test data.

R² measures how much of the variation the model explains; 1 would be a perfect fit. An analysis of what drove the ANN’s predictions found that biochar type and application rate mattered most, while BET surface area contributed very little. That echoes a pattern seen in other biochar studies on this site: surface area alone is a poor guide to performance.

What to keep in mind

  • Laboratory only. The authors state that external validation across different soils and field settings is still needed before the model can be used more widely.
  • High application rates. 15–25% biochar by weight is far more than would typically be applied to farmland, so the strongest results may not translate directly to fields.
  • Tyre-derived char. Material made from tyres can contain metals and other contaminants, an additional reason to favour plant-based biochars for soil.

Why it matters

The study confirms that the source of a biochar matters for keeping herbicides out of water, with nut-shell biochars outperforming tyre char. Its more novel contribution is methodological: if hyperspectral measurements and neural networks can reliably predict herbicide levels, testing biochar treatments could become faster and cheaper than running full chemical analyses on every sample.

Reference

Altıkat, A., Alma, M. H., Altıkat, S., & Gürbüz, R. (2026). Integrating biochar characterization, hyperspectral signatures, and artificial neural networks for predictive modeling of metamitron leachate attenuation. International Journal of Environmental Science and Technology, 23(4), 340. https://doi.org/10.1007/s13762-026-07122-3 (open access)

Related reading

Comments

No comments yet. Why don’t you start the discussion?

Leave a Reply

Your email address will not be published. Required fields are marked *