Validated sensor performance, calibration data, and the peer-reviewed science behind continuous microbial water quality monitoring.
Tryptophan-like fluorescence (TLF) is an optical water quality parameter centered on excitation/emission wavelengths of approximately 275/340 nm. It reflects concentrations of compounds with fluorescence characteristics similar to the amino acid tryptophan, which is associated with microbial activity and fecal contamination.
A positive relationship between TLF intensity and fecal indicator bacteria (FIB), particularly E. coli, has been demonstrated across groundwater, surface water, estuarine, and urban watershed contexts on multiple continents. This makes TLF a promising real-time proxy for microbial contamination risk—complementing or replacing traditional culture-based methods that require 24–48 hours for results.
The Virridy Lume addresses many of the limitations identified in the TLF literature by coupling optical TLF sensing with machine learning models that account for environmental confounders. Rather than relying on static TLF thresholds, the Lume's algorithm adapts to site-specific conditions.
The Lume returns a TLF reading in 60 seconds, compared to 24–48 hours for traditional culture-based methods.
The Lume's minimum detection limit for tryptophan dissolved in deionized water.
The Lume's binary classification at the 1 CFU/100 mL drinking water regulatory threshold yields 91% overall accuracy (Cohen's kappa = 0.82).
Site-specific calibrated categorical classifications of microbial contamination risk.
Out-of-the-box accuracy on a continuous scale across 0–1,000 CFU/100 mL.
Mean absolute percentage error in log-transformed concentration space vs. culture-based methods.
Machine learning quantification: Unlike earlier TLF sensors that report relative fluorescent units, the Lume quantifies actual E. coli concentrations using gradient-boosted decision tree models that capture nonlinear patterns in noisy sensor data. The ML model was validated across freshwater and coastal environments, including saline settings where enterococci are the regulatory indicator.
Multi-parameter correction: Built-in turbidity and temperature sensors allow the algorithm to correct for environmental interference that confounds raw TLF readings.
Drinking water performance: Binary classification at regulatory thresholds (1 and 10 CFU/100 mL) achieves 91–92% overall accuracy with Cohen's kappa of 0.82–0.84, indicating strong agreement with laboratory classifications.
Ocean transferability: In initial ocean deployments, the Lume achieved over 76% categorical accuracy using just six training samples, demonstrating transferability across microbial indicators and water matrices.
Adaptive contamination detection: The Lume uses machine learning to learn normal conditions and trigger contamination alerts based on learned patterns rather than fixed thresholds. (US Patent 11,506,606 B2 — Bedell, Fankhauser, Sharpe, Wilson & Thomas)
Automated system-state classification: Time-series data from water infrastructure sensors are analyzed to classify system states and support operational decisions without manual inspection or rule-based logic. (US Patent 11,507,861 B2 — Wilson, Coyle, Thomas & Croshere)
The Lume has been validated for drinking water monitoring across chlorinated and unchlorinated supplies. Binary classification at regulatory thresholds of 1 and 10 CFU/100 mL yields 91–92% overall accuracy with Cohen's kappa of 0.82–0.84, indicating strong agreement between sensor and laboratory classifications with minimal bias toward either class. The sensor also detects chlorine residual presence with 85% accuracy.
The Lume algorithm has been developed and extensively validated against Colilert® E. coli in freshwater systems. The platform has also been successfully deployed in saline coastal environments, where regulatory compliance is based on enterococci. In these initial ocean deployments, Lume achieved over 76% categorical accuracy using just six training samples.
How does the Lume compare to both EPA-approved laboratory methods? We analyzed paired samples measured by Colilert (IDEXX), membrane filtration (MF), and the Lume sensor across regression, Bland-Altman agreement, and categorical classification frameworks. The fourth panel additionally refits the Lume regression directly against MF, isolating the contribution of reference-method choice from sensor performance.
The Colilert-trained Lume achieves stronger agreement with Colilert (R² = 0.881, n = 209) than the two EPA-approved methods achieve with each other (R² = 0.572, n = 153).
When refit against MF as the training target, the Lume achieves equivalent quantitative performance (R² = 0.872, n = 303) — showing the sensor is method-agnostic.
Against Colilert, the Lume achieves “almost perfect” agreement (κ = 0.88), versus only “fair” agreement between the two lab methods (κ = 0.40).
The headline result: Sensor-to-reference agreement is bounded by reference-method reproducibility, not by Lume hardware. When the Colilert-trained Lume is tested against MF (a reference it was never trained on), R² drops to 0.514 — almost exactly the disagreement between the two lab methods themselves. When the Lume is instead calibrated directly against MF, performance jumps back to R² ≈ 0.87. Whichever culture method you hand the Lume as truth, it fits that method as well as or better than the two EPA-approved methods fit each other.
Microbial water quality is most frequently assessed using E. coli as a risk indicator. However, relying exclusively on culture-based E. coli measurement is limiting: it is slow, expensive, requires trained personnel, and captures only a single point in time. Risk assessments can be significantly improved by integrating TLF as a complementary, continuously monitored parameter.
TLF sensing is purely optical. No consumables, reagents, or lab infrastructure needed for each measurement.
In Kenya groundwater studies, TLF sampling showed 14% average relative percent difference between duplicates, compared to ≥26% for culture-based methods.
Sensors can operate autonomously for months, capturing short-duration contamination events that weekly grab samples miss.
Research in Malawi showed TLF indicates broader contamination risk than microbial culturing, making it a useful high-level screening tool.
For the cost of a single lab-processed grab sample, a TLF sensor can deliver thousands of in-situ microbial estimates over weeks to months of continuous deployment.
TLF has been validated across groundwater, freshwater rivers, chlorinated piped systems, and saline coastal environments on multiple continents, with studies spanning Kenya, Malawi, the US, and France.
TLF cannot be used as a direct proxy for E. coli on an individual sample basis. The TLF signal can be influenced by dissolved organic carbon (DOC), humic-like fluorescence (HLF), turbidity, and temperature. Standardization of TLF thresholds associated with different risk levels remains an active area of research. Performance is best in groundwater, which typically has low DOC, consistent temperature, and negligible turbidity.
Baker et al. demonstrate in-situ tryptophan-like fluorescence as a real-time indicator of faecal contamination in drinking water supplies. (Water Research)
Sorensen et al. evaluate TLF as a measure of microbial contamination risk in Kenyan groundwater across 37 water points. (Sci. Total Environ.)
Khamis et al. establish real-time detection of faecally contaminated drinking water with TLF, defining threshold values. (Sci. Total Environ.)
Bedell, Sharpe, Purvis, Brown & Thomas demonstrate low-cost TLF sensor concepts for fecal exposure detection. (Sustainability) • Virridy
Nowicki et al. conduct a nine-month monitoring program in Malawi, finding TLF is a more precautionary risk indicator. (Sci. Total Environ.)
Bedell, Harmon, Fankhauser, Shivers & Thomas field-validate a continuous in-situ fluorescence sensor coupled with ML. 83% accuracy. (Water Research) • Virridy
Multiple groups evaluate TLF for combined sewer overflow watersheds and estuarine systems. (Sci. Total Environ.; ACS ES&T Water)
Knopp, Klaus, Wilson et al. advance continuous in-situ quantification with Lume V1.2 sensor design and multi-site validation. (EarthArXiv) • Virridy
Demaree, Fankhauser, Cole, Ross & Thomas develop sensor-informed predictive models for TOC and nutrients on the Upper Yampa River. (ES&T Water) • Virridy