Moisture Content - Green Tea
Moisture content is the single most critical quality parameter measured at tea factory intake. It determines green leaf acceptance, price paid to smallholders, and withering endpoint decisions. Our NIR-based moisture model replaces the 6-hour ISO oven drying standard with a 15-second field reading - validated against the Tea Research Institute of Sri Lanka's reference protocol in December 2025.
Findings
What We've Established
TRI Validation - December 2025
The model was independently evaluated by the Tea Research Institute of Sri Lanka against ISO 1573 oven-drying reference values across a representative range of green leaf moisture levels (60–80% wet-weight basis). Calibration performance met TRI's acceptance threshold for field deployment.
Spectral Sensitivity to Moisture
NIR absorption bands in the 1,450 nm and 1,940 nm regions - corresponding to O–H stretching overtones - provide strong and reproducible sensitivity to water content in tea leaf matrices. The model leverages these bands alongside a broader spectral fingerprint to suppress interference from leaf surface variation and ambient temperature drift.
Factory-Scale Repeatability
Repeat measurements on the same leaf samples across multiple factory environments demonstrated coefficient of variation below 0.8%, confirming that the model is stable enough for real-world lot acceptance decisions without on-site recalibration between runs.
Methodology
Technical Approach
Calibration samples were collected from multiple estates across the Uva, Dimbula, and Kandy growing regions to ensure geographic and seasonal coverage. Reference moisture values were obtained using ISO 1573 oven drying at 103 °C. Spectral preprocessing steps include SNV normalisation and second-derivative transformation to remove baseline offsets. The predictive model uses Partial Least Squares regression (PLS-R) with optimal latent variable count determined by leave-one-out cross-validation. The final model was validated on a held-out test set not used during calibration.
Status
Where We Stand
The model is production-ready and deployed. It is in active commercial use with factory and brokerage clients in Sri Lanka. Ongoing data collection from additional estates is incorporated into quarterly model updates to improve robustness across harvest conditions.
Roadmap
Next Steps
Extend coverage to CTC-process black tea moisture at rolling and drying stages
Build a temperature-compensated variant for factories without climate-controlled intake bays
Investigate transfer learning approach to reduce recalibration burden when deploying to new device units
Publish a technical summary of the TRI validation protocol and results
Explore More
Related Projects
Total Polyphenols (TPP)
Total polyphenol profiling for black tea - the leading proxy for antioxidant quality and a primary auction pricing signal.
View project Model ReadySugar in Tea
Total sugar quantification across black tea grades - a key differentiator for premium blend specification and export documentation.
View project Lite Model ReadyCaffeine Content
Rapid caffeine screening for black tea blends - lite model delivered, enabling blend certification and label compliance workflows.
View projectR&D Partnerships
Interested in This Research?
If you have relevant data, domain expertise, or a measurement problem in this area, we're open to research collaboration and data-sharing agreements.