Coconut water analysis by NIRS

Significance of the topic

The quality attributes of coconut water (a widely consumed natural beverage) strongly influence its commercial value and consumer acceptance. Rapid, non-destructive, and reagent-free analytical methods are required to support routine quality control across harvesting, processing, and storage stages. Near-infrared spectroscopy (NIRS) offers a fast, high-throughput alternative to conventional wet-chemistry assays, enabling inline or at-line monitoring of multiple quality parameters with minimal sample handling.

Objectives and study overview

This application study (AN-NIR-145) evaluated the feasibility of using an OMNIS NIR Analyzer Liquid with a 1 mm flow-through cell to predict multiple quality parameters of coconut water from a set of commercial samples. The target analytes were Brix, salinity, conductivity, pH, and turbidity. A total of 206 coconut water samples were measured; calibration models were developed from this dataset and externally validated using a 25% independent validation set to estimate routine performance.

Methodology

Measurements were performed with a liquid NIR spectrometer using a flow-through configuration to enable rapid, repeatable analysis and automated rinsing between samples. Spectral data were used to build multivariate calibration models (chemometric predictions) for each reference parameter. Model performance was assessed by standard metrics including coefficient of determination (R2), standard error of calibration (SEC), standard error of cross-validation (SECV), and standard error of prediction (SEP).

Instrumentation used

• OMNIS NIR Analyzer Liquid configured for liquid samples (flow-through analysis).
• Flow-through quartz cell with 1.0 mm optical path length.
• Peristaltic pump for sample introduction; automated rinsing between measurements.
• OMNIS software with Quant Development license for model development; stand-alone operation supported by OMNIS Stand-Alone license.
• Optional OMNIS Sample Robot integration for full automation.

Key results and discussion

The chemometric models showed different degrees of predictive quality across analytes:

• Brix (soluble solids): excellent performance — R2 = 0.997, SEC = 0.04 °Bx, SECV = 0.04 °Bx, SEP = 0.05 °Bx. This indicates high accuracy and precision suitable for routine QC.
• Salinity: excellent performance — R2 = 0.992, SEC = 0.08 %, SECV = 0.10 %, SEP = 0.09 %. NIR reliably quantifies salt content in this matrix.
• Turbidity: strong performance — R2 = 0.956, SEC ≈ 10.05 ppm, SECV ≈ 13.04 ppm, SEP ≈ 14.90 ppm. NIRS can serve as a rapid turbidity indicator.
• Conductivity: good but moderate correlation — R2 = 0.892, SEC = 250 μS/cm, SECV = 286 μS/cm, SEP = 285 μS/cm. Predictive accuracy is acceptable for screening and trend monitoring, though not as precise as Brix/salinity.
• pH: limited correlation — R2 = 0.658, SEC = 0.05 pH units, SECV = 0.05, SEP = 0.06. NIRS here is better suited for classification or trend detection (e.g., distinguishing acidic shifts) rather than replacing potentiometric pH measurements when tight accuracy is required.

Overall, spectral models for compositional parameters (Brix, salinity) performed best, while properties less directly tied to NIR-absorbing features (pH, conductivity) showed reduced quantitative performance. The dataset size (206 samples) and use of an independent 25% validation set provide robust estimates of routine performance.

Benefits and practical applications

• Rapid analysis: liquid measurements in seconds with no chemical reagents or sample prep, enabling high throughput in production or QC labs.
• Automation-friendly: flow-through cell plus automated rinsing and optional sample robot allow inline/at-line deployment and integration with other analytical platforms.
• Multi-parameter monitoring: single spectral acquisition yields simultaneous predictions for several quality attributes, reducing turnaround time and operational costs.
• Non-destructive measurement: preserves sample for additional tests if needed.

Future trends and potential uses

• Expanded calibrations: increasing spectral libraries with diverse origin, harvest time, and storage conditions will extend model robustness across supply chains.
• Hybrid approaches: combining NIRS with targeted wet-chemistry checks or orthogonal sensors (e.g., conductivity probes) can improve reliability for parameters with moderate NIR correlation.
• Inline process control: integration into processing lines with real-time feedback can support sorting, blending, or storage decisions to maintain product consistency.
• Advanced chemometrics and transfer learning: applying domain adaptation, non-linear models, and transfer calibration strategies will facilitate deployment across instruments and locations.

Conclusion

NIRS using the OMNIS NIR Analyzer Liquid and a 1 mm flow-through cell is a practical solution for rapid quality screening of coconut water. It provides excellent quantitative results for Brix and salinity, strong results for turbidity, and useful screening capability for conductivity and pH (classification/trend monitoring). The approach reduces analysis time and sample handling and can be implemented in automated workflows for routine QC. For analytes with limited NIR correlation, complementary measurements should be considered where stringent accuracy is required.

References

• Metrohm Application Note AN-NIR-145: Coconut water analysis by NIRS — Determination of Brix, salinity, conductivity, and pH with one instrument.
• OMNIS NIR Analyzer Liquid product documentation (OMNIS Software, OMNIS Stand-Alone license, Quant Development license; flow-through cell, 1 mm).