Monitoring quality of intact olives with near-infrared spectroscopy

Importance of the topic

Near-infrared spectroscopy (NIRS) offers a rapid, non-destructive and reagent-free approach to assess key quality attributes of intact olives relevant for olive oil production. Determining oil content, moisture and an empirical maturity index prior to processing supports optimal harvest timing, improves mill throughput planning and enables early estimation of oil yield and economic returns. Replacing or reducing laborious reference methods (Soxhlet extraction, oven drying) with NIR can shorten decision cycles and reduce solvent use in routine QC and production environments.

Study objectives and overview

The application study evaluated the feasibility of using an OMNIS NIR Analyzer Solid to predict oil content, moisture and a maturity index from intact olives. The work aimed to develop and validate calibration models that enable rapid screening of intact fruit without sample preparation and to demonstrate robustness across two common cultivars (Picual and Arbequina) and varying ripeness levels.

Methodology and instrumentation

• Sample set: 800 intact olive samples comprising two varieties (Picual, Arbequina) spanning a range of oil and moisture contents and maturity stages.
• Spectral acquisition: OMNIS NIR Analyzer Solid operating in diffuse reflection mode over 1000–2250 nm. To address sample heterogeneity the protocol used a large 100 mm cup/holder with automated rotation and multipoint measurements.
• Hardware/software: Large holder OMNIS NIR (100 mm), Large cup OMNIS NIR (100 mm), OMNIS Stand-Alone and Quant Development software license for model building.
• Reference methods: Oil content by Soxhlet extraction; moisture by loss-on-drying (oven); maturity index determined following International Olive Council guidelines.
• Calibration strategy: 75% of samples used for calibration, 25% reserved for validation. Leave-one-out cross-validation was applied during model development and figures of merit (FOM) were reported for model performance.

Instrumentation used

• OMNIS NIR Analyzer Solid (Article No. 2.1071.0010) — near-infrared spectrometer for solids and viscous samples.
• Large holder OMNIS NIR, 100 mm (6.07402.100) — ensures reproducible positioning and rotation of large sample vessels.
• Large cup OMNIS NIR, 100 mm (6.07402.110) — sample vessel for powders/granulates and large intact fruit measurements.
• OMNIS Stand-Alone license (6.06003.010) and Quant Development license (6.06008.002) — software for instrument control and quantification model development.

Main results and discussion

• Oil content prediction: Calibration showed R2 = 0.811 with SEC = 1.39% (SECV 1.43%, SEP 1.44%). This indicates a strong relationship between NIR spectra and Soxhlet-derived oil content, suitable for screening and process-control decisions where sub-percent precision is not critical.
• Moisture prediction: R2 = 0.868 with SEC = 1.70% (SECV 1.75%, SEP 1.81%). Moisture was predicted with the best performance of the three parameters, reflecting moisture-sensitive NIR absorption features and suitability for rapid moisture monitoring.
• Maturity index prediction: R2 = 0.706 with SEC = 0.48 (SECV 0.49, SEP 0.51). The maturity index model is less precise than oil/moisture models but still provides useful relative information about ripeness trends across batches.

Interpretation: Models demonstrate feasibility of non-destructive prediction directly on intact fruit. Moisture and oil content predictions are robust enough for routine screening; maturity index predictions are moderately predictive and benefit from careful calibration against IOC reference scoring. Use of rotating, multipoint sampling reduced effects of fruit heterogeneity. Reported SEP values on the independent validation set confirm expected routine precision during operational use.

Benefits and practical applications

• Non-destructive and fast measurement (<10 seconds per measurement on the OMNIS NIR Analyzer Solid) eliminates need for grinding or solvent extraction for routine checks.
• Reduced laboratory workload, no chemical reagents and lower turnaround time enable more frequent monitoring of orchard ripeness and mill intake quality.
• Portable workflow for mills and QC labs: models can inform harvest timing, sorting and acceptance criteria and estimate oil yield before processing.
• Automatable and integrable into production IT systems, supporting real-time decisions and traceability.

Limitations and considerations

• Calibration dependence: Models are specific to the sample set, cultivar range and conditions used for calibration; transferability requires additional calibration transfer or augmentation across seasons/regions.
• Sample heterogeneity: Although rotation and multipoint measures mitigate variability, very heterogeneous lots may still reduce prediction accuracy.
• Reference method quality: Accuracy depends on reliable reference analyses (Soxhlet, loss-on-drying, IOC maturity scoring) used during model training.
• Model maintenance: Periodic revalidation and updating is recommended to maintain performance over new harvests, cultivars or changing environmental conditions.

Future trends and potential applications

• Model generalization using larger, multi-season datasets and transfer-learning approaches to improve inter-region and inter-cultivar robustness.
• Integration of NIR with inline or at-harvest platforms (portable sensors, conveyor-mounted systems) for real-time orchard and mill analytics.
• Combining NIR with machine learning and chemometric techniques (e.g., ensemble models, domain adaptation) and with spatial/hyperspectral imaging to capture compositional and spatial heterogeneity.
• Linking spectral QC data to production planning and predictive yield models for improved supply-chain optimization and sustainability metrics.

Conclusion

The application demonstrates that NIRS on intact olives using the OMNIS NIR Analyzer Solid can reliably predict oil and moisture content and provide useful estimates of maturity index without sample preparation. The technique offers substantial workflow advantages over conventional methods—speed, absence of reagents and ease of use—making it attractive for routine QC, harvest decision support and early yield estimation in olive oil production. Ongoing model maintenance and expansion of calibration sets will further strengthen its operational value.

References

• Metrohm, Application Note AN-NIR-153: Monitoring quality of intact olives with near-infrared spectroscopy.
• International Olive Council, Standards, Methods and Guides (used as reference for maturity index determination).