Sample-Level Validation and Applicability of Near-Infrared Spectroscopy for Rapid Screening of N, P2O5, and K2O in Compound Fertilizers

Summary

This study evaluated the utility of compact near-infrared spectroscopy for rapid, non-destructive screening of major nutrients (N, P2O5, K2O) in unground granular compound fertilisers using 143 samples (286 spectra). Ensemble models achieved good internal calibration performance (R2 0.80–0.90), with substantial RMSE improvements over simple reference models. However, external validation on subsequent production batches revealed mixed transferability: nitrogen and phosphorus pentoxide retained screening potential, but potassium oxide failed to transfer reliably, suggesting that broader calibration across production batches and formulations would be needed for operational deployment.

Regional applicability

This methodology has direct applicability to fertiliser quality control in the United Kingdom and across Europe, where rapid, cost-effective nutrient verification at point-of-use or in supply chains is valuable. The finding that model transferability is formulation-dependent suggests that UK fertiliser producers would need batch-specific or broad multi-source calibration datasets to implement NIR screening reliably.

Key measures

Internal coefficients of determination (R2) for N (0.9005), P2O5 (0.8279), and K2O (0.8034); root-mean-square error (RMSE) values in percentage points; RMSE reduction relative to nutrient-only linear reference models; external validation R2 values on 29 subsequent production-batch samples

Outcomes reported

Evaluated the predictive performance and applicability of compact near-infrared spectroscopy for rapid screening of nitrogen, phosphorus pentoxide, and potassium oxide in granular compound fertilisers. The study assessed model stability across repeated measurements and examined transferability to subsequent production-batch samples.

Topic tags