Meets EPA’s criterion for an alternative recreational-water monitoring method (IA ≥ 0.70, EPA-820-R-14-011).
Ready now for operational screening and early warning; local calibration and state-level approval required for use in compliance reporting.
Substantiation: the Lume clears EPA’s Alternative Methods Calculator index-of-agreement threshold (IA ≥ 0.70) against EPA reference methods, the acceptance bar in EPA-820-R-14-011. The qualifier must always accompany the headline; do not claim “EPA-approved,” “EPA-certified,” or “EPA-compliant.”
Overview
The Lume is a continuous, in-situ tryptophan-like fluorescence (TLF) sensor that indicates fecal-contamination risk in water. This project develops it into a formal, standardized method and secures EPA recognition for recreational-water screening and early warning.
Why now: a broad search finds no standardized TLF method anywhere (not Standard Methods, ASTM, or ISO), despite a decade of peer-reviewed TLF research. That is white space and a first-mover opportunity.
Background
The monitoring gap
Recreational-water safety is governed by bacteria criteria (E. coli in freshwater, enterococci in marine water). Compliance monitoring relies on grab samples cultured in a lab (Colilert, membrane filtration), which take 18–24 hours and sample only a moment in time. Contamination events are episodic and flow-driven, so grab sampling routinely misses them. A continuous, in-situ indicator that flags high-risk conditions in near-real-time fills that gap — not to replace the compliance measurement, but to target and prioritize it.
What TLF is
Tryptophan-like fluorescence is the optical signal (excitation ~273 nm, emission ~350 nm) of tryptophan-like fluorophores — intrinsic components of microbial cells and their metabolic byproducts. It correlates with fecal-indicator bacteria and is measured optically, continuously, with no reagents. It is not specific to E. coli, does not distinguish viable from non-viable cells, and is predictive (a model estimate), which is exactly why it is an indicator rather than an enumeration.
What the Lume adds
The Lume 1.2 pairs a UV-LED / silicon-photomultiplier TLF channel with an optical turbidity (time-of-flight) channel and temperature, and applies per-sensor calibration plus a model that corrects the two dominant confounders (temperature quenching and turbidity inner-filter / scattering) to convert the raw signal into an E. coli estimate and an exceedance-risk classification.
Validation evidence
The core evidence base, from the Boulder Creek environmental dataset (paired Lume, Colilert, and membrane-filtration samples), now extended across a multi-site field program. This is exactly the kind of environmental-sample correlation the alternative-indicator framework asks for, and it maps directly onto the metrics that framework uses.
The reference-vs-reference ceiling
The two EPA-approved culture methods disagree with each other more than the Lume disagrees with either. On the dedicated replicate study (n = 153), membrane filtration vs Colilert agree at only R-squared = 0.52, with MF reading ~2.2× higher, and by ISO 17994:2014 they are not equivalent (mean relative difference +79%). No indicator can be expected to beat the reference methods’ own mutual disagreement.
Lume vs the reference methods
| Comparison | n | IA | R-squared | Bal. acc (≥126 / median) |
|---|---|---|---|---|
| MF vs Colilert (two EPA methods) | 153 | 0.79 | 0.52 | 0.86 / 0.80 |
| Lume vs Colilert | 209 | 0.96 | 0.861 | 0.90 / 0.81 |
| Lume vs MF | 206 | 0.91 | 0.702 | 0.98* / 0.75 |
The Lume clears the EPA IA ≥ 0.70 bar against both reference methods (0.96 vs Colilert, 0.91 vs MF) and tracks whichever culture method it is trained on (R-squared 0.86 vs Colilert, 0.70 vs MF) — at or above the agreement the two EPA methods reach with each other. Replicate precision favors the sensor (RPD ~14% vs ≥26% for culture duplicates). The ≥126 classifier is reported at a screening operating point (sensitivity-first).
Beyond Boulder Creek
The method is now paired to 661 reference samples across 12 sites, including an independent recreational deployment on the Seine & Marne (Paris) that classified the 900 CFU/100 mL bathing threshold at 96.8% accuracy / 94% balanced accuracy on a forward-in-time split — corroborating the Boulder result in a different matrix and criterion. Data collection is ongoing across a widening set of water bodies, with marine / estuarine (saltwater) waters next; those add enterococci (Enterolert) as the reference and will extend this validation to coastal criteria.
Full interactive analysis (regression, Bland–Altman, both classifier thresholds, per-point CI overlap, downloadable data): the Validation Data page.
Regulatory landscape
EPA — Final Water Quality Standards Bacteria Rule (BEACH rule) ›
The target is recreational water bacteria criteria: EPA’s Water Quality Standards for Coastal and Great Lakes Recreation Waters (2004 BEACH rule, 40 CFR 131) and the 2012 Recreational Water Quality Criteria (RWQC). These are Water Quality Standards, not 40 CFR 136 analytical-method approvals — and that distinction is our way in.
Pathways to recognition
- Pathway 1 · Site-specific alternative recreational criteria. A state / tribe adopts the Lume as an alternative indicator into site-specific WQS, validated by R-squared and Index of Agreement on environmental samples (EPA-820-R-14-011). Health linkage is inherited from EPA’s NEEAR epidemiology, so no new epi study is required.
- Pathway 2 · Method → peer-reviewed paper → EPA guidance reference. Standardize and publish the method, then get EPA to reference it in guidance (precedent: Bolton & Linden 2003 → UV Disinfection Guidance Manual).
- Pathway 3 · Standard Methods. Publish the method in Standard Methods (APHA / AWWA / WEF, Part 9000), which EPA references in 40 CFR 136 / 141; pursue ASTM D19 and ISO TC 147 in parallel.
- Pathway 4 · AOAC Performance Tested Methods (PTM). Independent third-party certification of the Lume as a proprietary device — the fastest recognized-method credential and an on-ramp to AOAC’s CFR-level Official Methods. See the dedicated section below ›
These four pathways are not alternatives to choose between; they stack into a three-layer scaling model, below.
How this scales — not one filing per beach
| Layer | Use | EPA involvement | Adoption unit | Per-site study? |
|---|---|---|---|---|
| 1. Operational screening | Early warning, prioritize sampling, precautionary advisories; culture stays the official number | None required | The operator / beach | No |
| 2. Regulatory replacement | TLF becomes the reported regulatory number for a waterbody | Alternative-indicator adoption (EPA-820-R-14-011); EPA Region approves the state standard | The state (covers many beaches) | Yes: ≥30 paired samples, a deployment byproduct |
| 3. Recognized method | Method accepted generally, adopted by reference | Consensus body (AOAC PTM → Standard Methods / ASTM); EPA references it | National | Collapses to a lightweight local verification |
Layer 1 · Most use needs no adoption at all
The large majority of deployments do not replace the culture number. They run TLF as a continuous screening and early-warning layer: when to sample, when to post a precautionary advisory, where to send crews, filling the days between grabs. This needs zero EPA paperwork, and a beach manager can do it today. It is also the data engine: every one of these deployments collects paired grabs that become the evidence for Layer 2.
Layer 2 · Regulatory replacement is per-state, not per-beach
When TLF is to become the official regulatory number, the site-specific framework applies, but the actor is the state, not the beach. A state agency adopts the alternative indicator/method into its water quality standards; the beach files nothing. Two things make this tractable: a state can adopt an alternative-method provision under which individual sites qualify by meeting the IA ≥ 0.70 bar (one state rulemaking, per-site data), and the ≥30 paired samples per site are the Colilert grabs already collected at each install. The scaling unit is one state or program covering many beaches, which is also who buys: states, utilities, and watershed authorities that manage many sites, not individual beaches.
Layer 3 · The real unlock: get the method recognized, then adopt by reference
This is how Colilert went from one product to universal: it became Standard Methods 9223, after which every state and lab could use it by reference under OMB Circular A-119, with no per-site federal blessing. The Lume equivalent is the AOAC PTM → Standard Methods / ASTM track (see the AOAC PTM section). Once TLF screening is a recognized consensus method, the heavy per-site study shrinks to a standardized local verification the operator self-performs, because the measurement is recognized generally and only the local proxy relationship needs a quick confirmation.
What the method can accurately claim
Precision in the claim protects the program. The defensible claim escalates by stage:
| Stage | Claim |
|---|---|
| Never | “EPA-approved,” “EPA-certified,” or “EPA-compliant” method. No such status exists for a monitoring method, and the 40 CFR 136 Alternate Test Procedure is closed to a correlative indicator (it requires measuring the same analyte as the reference). |
| Now | Evaluated under EPA’s Recreational Water Quality Criteria Alternative Indicators and Methods framework (EPA-820-R-14-011); meets EPA’s Alternative Methods Calculator index-of-agreement criterion (IA ≥ 0.70) as an alternative recreational indicator method against an EPA-approved E. coli method. This is a factual statement about meeting EPA’s published bar on EPA’s own tool, not a claim of approval. |
| After a site adoption | Adopted by [State] and approved by EPA Region [X] as an alternative monitoring method for [waterbody] recreational water quality standards. Site- and state-specific regulatory acceptance, which accumulates as adoptions grow. |
Net: you never certify tens of thousands of beaches. You certify the method (Layer 3), you win states (Layer 2), and you sell screening to everyone else with no gate (Layer 1). The staged sequence is in The plan.
AOAC Performance Tested Methods (PTM)
What it is
The AOAC Research Institute (AOAC INTERNATIONAL) runs the Performance Tested Methods program: an independent, third-party validation that certifies proprietary commercial test methods and devices against a reference method. A certified method carries an AOAC PTM certificate number and a published validation report, and is harmonized with ISO 16140-2, the international standard for validating alternative microbiological methods. PTM certifications are a widely recognized mark of independent validation across food and water microbiology and are referenced in federal method frameworks.
Why it fits the Lume where the consensus routes don’t
The Lume is a proprietary instrument with per-unit calibration — exactly the case the consensus bodies shy away from and the case PTM is designed for. Instead of a multi-year, multi-laboratory collaborative study, PTM turns on a single independent-laboratory validation against the reference method, on the order of ~6–12 months. It converts the Lume’s proprietary, device-specific nature from an obstacle into the normal case.
Proprietary device, fast
- Single independent-lab validation
- ISO 16140-2 aligned
- ~6–12 months
- Built for single-vendor methods
- On-ramp to AOAC OMA (CFR-level)
Consensus method, slow
- Multi-laboratory collaborative study
- Avoids single-vendor methods
- ~1.5–4 years
- Method-, not device-, oriented
- Directly EPA-referenced once published
How it works
Pre-submission call with AOAC RI
Define the certification claim (fecal-contamination screening / E. coli exceedance against a reference method) and the study design with the AOAC Research Institute.
Method + SLV study
Submit the documented method (Method LUME-1) and a single-laboratory validation characterizing accuracy, precision, and threshold classification against Colilert / membrane filtration.
Independent-laboratory study
An independent laboratory reproduces the validation against the reference method under an AOAC study director / expert reviewer.
PTM certificate + annual renewal
On review, the method receives an AOAC PTM certificate number and published report, maintained by annual renewal and ongoing oversight.
Why now — the strategic fit
- Fastest credential. The quickest route to being an independently validated, recognized method — ahead of the multi-year consensus routes, which it complements rather than replaces.
- On-ramp to CFR-level. PTM is the recognized on-ramp to AOAC Official Methods of Analysis (OMA), which are referenced at the CFR level.
- Procurement lever. OMB Circular A-119 and the NTTAA direct federal agencies (including DoD and NASA) to prefer voluntary consensus-body methods over vendor-unique ones. A PTM certification places LUME-1 inside a recognized body, removing the “not a recognized method” objection for operational and agency adoption.
- Evidence reuse. The validation package PTM requires is largely what we have already built — the reference-uncertainty framework, the Index of Agreement, and the threshold-classification results in the manuscript and validation data. Mostly re-wrappering, not new science.
Working documents
Two documents are prepared for the AOAC engagement (generated from the Overleaf source):
📄 AOAC PTM Pre-Submission Scoping Brief (PDF) — the device, the certification claim, existing validation evidence, and the questions for the AOAC RI consulting call.
📄 Draft AOAC PTM Validation Study Plan (PDF) — a starting study outline (developer + independent-lab), mapping ISO 16140-2 / PTM elements onto the Lume and crediting existing data.
Key documents
The governing EPA guidance and the framework documents. EPA documents (public-domain US government works) are embedded; the copyrighted Standard Methods SOP is linked to its source.
EPA · Site-Specific Alternative Recreational Criteria TSM (2014)
EPA-820-R-14-011. The framework for Pathway 1: how a state adopts an alternative indicator/method into site-specific recreational WQS (three steps, IA ≥ 0.7 or R-squared > 0.6 decision rule, criteria derivation, WQS checklist).
📄 View / download — EPA-820-R-14-011 (PDF, 85 pp.)
EPA · Alternative Methods Calculator Tool (AltCalc)
EPA 821-F-21-005 (Nov 2021). The Excel tool that computes the Index of Agreement and Pearson R-squared for two paired methods, per the 2014 TSM (Appendix E). This is precisely the tool we run our Lume-vs-Method-1603 paired data through for a site-specific WQS submission: IA ≥ 0.7 keeps the EPA criteria unchanged, else R-squared > 0.6 derives new site-specific criteria. Needs ≥30 paired points within the limits of quantification.
EPA source: epa.gov · technical support: Shamima Akhter (akhter.shamima@epa.gov).
Other documents (linked)
- Standard Methods SOP (SM SOP-2026-2.1). The method development / balloting procedure for Pathway 3. Copyrighted APHA / AWWA / WEF; not re-hosted. standardmethods.org
- 2004 BEACH rule. Water Quality Standards for Coastal and Great Lakes Recreation Waters (40 CFR 131). EPA
- EPA Draft Method C (rapid E. coli qPCR). Aw et al. 2019, Water Research 156:465–474. doi · EPA record
- Bolton & Linden (2003). UV fluence standardization, adopted into EPA’s UV Disinfection Guidance Manual. doi
Prior art & references
- Sorensen et al. TLF as a measure of microbial contamination risk in groundwater; real-time detection thresholds (BGS). PubMed
- Bridgeman, Baker et al. TLF fundamentals and organic-matter / microbial fluorescence in water.
- Nowicki et al. Evaluating TLF to quantify E. coli in a combined-sewer-overflow watershed. PubMed
- Thomas et al. / Virridy. Continuous in-situ TLF sensor + machine-learning model; drinking-water field validation. Water Research · TLF research
- Sensor design & validation (environmental waters). Water Research
Standards venues (no TLF standard yet)
Precedents
EPA Draft Method C · rapid E. coli qPCR
EPA developed a rapid, non-culture, non-viability E. coli method (qPCR) for recreational freshwater and validated it via a multi-lab QC-criteria study. Precedent that EPA pursues fast, non-culture quantification. Distinction: qPCR still targets E. coli-specific genes (a molecular measure); TLF is neither specific nor a direct measure, so we sit further along the indicator spectrum.
Method → paper → EPA use · Bolton & Linden (2003)
A peer-reviewed standardization paper that became the de facto regulatory standard as EPA guidance, without a promulgated method rule: (1) standardized the UV-fluence bench method, (2) published it, (3) EPA adopted it by reference into the UV Disinfection Guidance Manual (2006) for LT2ESWTR. The clearest model for Pathway 2.
SCCWRP · alternative indicators in practice
The 2014 alternative-indicator TSM is the framework SCCWRP and others already use to relate alternative indicators (qPCR, Bacteroides, Enterolert) to EPA methods by R-squared / Index of Agreement and adopt site-specific criteria. Worked example in the TSM: Doheny Beach, Enterolert vs EPA Method 1600, IA = 0.94, R-squared = 0.79.
The plan
Develop the Lume TLF sensor into a published, standardized method and get it recognized by EPA for recreational-water screening — via the alternative-indicator and standard-method routes.
Roadmap
Position, validate, and document
Lock the positioning (indicator / MST; target = recreational WQS / BEACH rule). Extend the Boulder Creek validation to the metrics the frameworks use: R-squared + Index of Agreement + classifier performance (balanced accuracy, sensitivity, specificity) + precision & bias, on ≥30 paired environmental samples across conditions. Write the Lume up as a documented method (Draft below).
Peer-reviewed method paper
Publish in a top water journal (Water Research or ACS ES&T Water). Establishes the science and creates the citable method that every downstream route builds on.
Into a standards body
AOAC PTM first — the fastest recognized-method credential for a proprietary device (single independent-lab validation, ~6–12 mo; see above). In parallel, the consensus routes: Standard Methods (Part 9000) — the EPA-referenced US venue (SM Manager → JEB → charge / JTG → multi-lab study → balloting) — plus ASTM D19 and ISO TC 147 (international).
EPA listing / state adoption
Pathway 1: partner with a state / tribe (Colorado / CDPHE, Boulder Creek), show IA ≥ 0.7 or R-squared > 0.6 vs EPA Method 1603, adopt into a site-specific WQS. Pathway 2: EPA guidance reference. Eventually EPA cites the SM / ASTM / ISO method for 40 CFR 136 / 141 use.
Customers adopt the recognized method
With a published standard and a regulatory pathway, states, utilities, and monitoring programs adopt the Lume as the recognized screening method.
Deliverables to build
- Documented method (Standard Methods / EPA format) — the Draft below.
- Environmental validation package: R-squared + Index of Agreement + classifier metrics + precision / bias, ≥30 paired samples across wet / dry / storm / bloom conditions.
- Multi-laboratory validation study (required for Standard Methods).
- Peer-reviewed method paper.
- AOAC PTM validation study & certification (fastest recognized-method credential).
- SM / ASTM / ISO method proposals.
- A state / tribe site-specific WQS pilot submission.
Key contacts
- EPA: Lemuel Walker, CWA methods coordinator (walker.lemuel@epa.gov, 202-566-1077).
- Standard Methods: SM Manager (1-303-347-6241) → Joint Editorial Board.
- State partner: Colorado / CDPHE, Boulder Creek as the validation matrix.
The method & companion documents
Three authored deliverables, all maintained in one LaTeX source (Overleaf) and auto-compiled here: the method itself, the peer-review manuscript that presents and validates it, and the per-unit acceptance/calibration certificate that shows an individual sensor meets it. The web copies are generated from the source and stamped with the source revision, so they cannot drift from the documents of record.
1 · Method LUME-1 — the method
A working draft of the Lume TLF method in U.S. EPA method-report format (scope, summary, interferences, apparatus, QC, calibration, procedure, calculations, performance), modeled on EPA Methods 1611/1696. It positions TLF as a continuous, real-time screening method for fecal contamination calibrated to the 126 CFU/100 mL recreational criterion — not an organism-specific E. coli enumerator.
📄 Read the method (web) › · Method LUME-1 (PDF)
2 · Journal manuscript — method description & validation
The peer-review paper (draft, target ACS ES&T Water) presenting the method and validating it against both EPA reference methods. By the EPA Alternative Methods Calculator (Willmott index of agreement on log₁₀ values), the screen reached IA = 0.96 vs Colilert (n = 209) and IA = 0.91 vs membrane filtration (n = 206) — both above the 0.70 threshold that permits using the same criterion, and higher than the two EPA methods agree with each other (IA = 0.79). Against Colilert: R² ≈ 0.86, mean bias 0.00 log₁₀, 95% limits of agreement [−0.42, +0.42], exceedance balanced accuracy 86–90%. It makes reference-method uncertainty explicit (the two EPA methods disagree at R² ≈ 0.52; 8.1% of paired grabs are unresolvable at the 126 CFU/100 mL boundary by the reference’s own 95% interval) and names inter-unit reproducibility (σ ≈ 0.16 log₁₀) as the gating limitation before multi-lab validation.
📄 Manuscript — ES&T Water draft (PDF)
3 · Per-unit conformance & calibration certificate
The complementary, ongoing artifact: where the method’s one-time validation establishes that the Lume as a class agrees with EPA methods, this document is the protocol and record by which each serialized unit is shown to meet that performance before deployment and at intervals thereafter. It operationalizes Method LUME-1’s QC (§9) and calibration (§10) per unit: configuration control, the per-unit calibration steps (SiPM dark characterization, L-tryptophan standard curve, temperature coefficient ρ, turbidity coefficient, clean-water baseline, QCS), an acceptance-criteria table (e.g. tryptophan linearity R² ≥ 0.98, LOD ≤ 0.1 ppb, agreement within ±0.42 log₁₀ of Colilert, inter-unit reproducibility limit), continuing-conformance triggers, and a fillable certificate produced per unit.