Sampling and Analysis Plan
Boulder Creek Segment 2b

This draft is written against the example Sampling and Analysis Plan at Appendix B of EPA-820-R-14-011, which states that “you are not limited to the elements shown in this example; you may use other SAP formats and designs.” Values still to be agreed are marked TO SETTLE and collected in Section 11. Nothing marked that way is a data gap; each is a decision for the City, CDPHE or the analytical laboratory.

1 · Purpose and scope

This plan governs the collection and analysis of paired water quality data on Boulder Creek Segment 2b for the purpose of demonstrating a consistent and predictable relationship between an EPA-approved E. coli method and a continuous in-situ tryptophan-like fluorescence (TLF) indicator, as required by Step 2 of EPA-820-R-14-011.

Scope. Ambient recreational fresh water only. This plan does not cover NPDES compliance monitoring, drinking water, or wastewater. The alternative indicator is proposed for site-specific use in recreational-use assessment on Segment 2b, not for national use and not as a 40 CFR 136 compliance method.

Objective. To produce a paired dataset sufficient for the Step 3 comparison: at least 30 paired observations within the limits of quantification of both methods, distributed across the range of hydrologic, meteorological and contamination conditions occurring at the site during the recreation season.

2 · Project organization and responsibilities

Any change to this plan during execution is recorded under Section 10 and communicated to the Division before the affected sampling occurs.

3 · Study design

3.1 Monitoring locations

Six locations are instrumented, each with a continuously deployed Lume unit and co-located grab sampling. Counts below are the paired record to date and are indicative; the record is growing.

Precise coordinates, descriptions and the segment boundary determination for each location are TO SETTLE with the City, including confirmation of which sites fall inside Segment 2b (13th Street to the confluence with South Boulder Creek) and which are upstream reference locations.

3.2 Sampling frequency and duration

The plan runs for one full recreation season with sampling at each location on a routine interval, supplemented by event-based sampling. The design deliberately addresses the coverage gap identified in the existing record (Section 3.3).

• Routine sampling at a fixed interval across all six locations, spanning the full recreation season including the late-season period.
• Event-based sampling targeted at wet-weather and elevated-flow conditions, triggered on forecast or observed precipitation.
• Sensor operation is continuous and independent of the grab schedule, at the reporting interval configured in Method LUME-1.
• Routine interval, event trigger threshold and total sample target are TO SETTLE.

3.3 Condition coverage

The TSM requires samples “taken at intervals over time to cover the range of conditions at the site” and directs that the submission discuss assay type, fecal sources, source age and proximity, and hydrometeorological factors (Appendix A). Coverage in the existing record was measured by matching each grab to the mean daily discharge at USGS gauge 06730200 and placing it in the recreation-season flow distribution.

3.4 Sample size

The TSM requires at least 30 paired observations within the limits of quantification of both assays, and the AltCalc tool issues a warning below that number. The existing record clears this comfortably: applying the quantification limits in Section 4.3 and excluding non-detects leaves 326 paired observations across the bench and field record, of which 145 are the Boulder Creek in-situ set. Additional sampling under this plan serves condition coverage and seasonal completeness rather than raw count.

4 · Methods compared

4.1 Method one — the reference

IDEXX Colilert with Quanti-Tray/2000, reported as MPN/100 mL. Colilert is an ATP-approved equivalent E. coli method and therefore qualifies as method one under the TSM, which permits EPA Method 1600, 1603, 1611, or “any equivalent method to those listed above, as determined by the Alternate Test Procedure (ATP) program.”

Final confirmation of the acceptable reference rests with CDPHE. If the Division requires EPA Method 1603 specifically, paired 1603 analysis will be added rather than substituted, so that the existing Colilert record is preserved. TO SETTLE

4.2 Method two — the alternative indicator

Continuous in-situ tryptophan-like fluorescence as specified in Method LUME-1, implemented by the Virridy Lume. The indicator is a substance rather than an organism, which the TSM expressly contemplates: the alternative may be “a different fecal indicator organism or substance,” with caffeine and detergent brighteners given as examples.

The reported quantity is a fecal-contamination screening index calibrated to the reference scale. Temperature and optical-backscatter turbidity enter only as corrections to the fluorescence measurement, per Method LUME-1 Sections 4 and 10. They are not independent indicators and carry no weight of their own. Environmental predictors such as rainfall and upstream discharge, used in Virridy's operational products, are excluded from the submitted method, since combining independent measurements as a tiered set of indicators is outside what the TSM permits.

4.3 Limits of quantification

The 126 CFU/100 mL criterion sits 12.6× above the alternative method's quantification limit. The 10 CFU/100 mL figure is presently supported by drinking-water paired data; confirmation in the recreational matrix on the Boulder Creek record is outstanding. TO SETTLE

5 · Field procedures

5.1 Equipment and materials

• Sterile sample containers, single use, of a volume sufficient for the reference analysis and any duplicate. Container type and volume TO SETTLE.
• Insulated cooler with refrigerant sufficient to hold samples at 0–4 °C through transport. Coolers sanitized between uses and rejected if cracked or torn.
• Calibrated thermometer for air and water temperature, verified against a reference thermometer on a documented schedule.
• Field forms or electronic equivalent, permanent marker, and pens in blue or black ink. Pencil is not used.
• Waders or equivalent where the sampling point requires entry, used only where conditions make entry safe.

5.2 Preparation before each round

1. Label containers for each location before leaving, against the field form.
2. Record date, personnel and planned locations.
3. Record antecedent precipitation for the preceding 24 hours and current wind conditions from the designated source, before departure.
4. Confirm from the sensor telemetry that each unit is reporting and in water, so that a grab is not collected against a unit that is offline or exposed.

5.3 Sanitary survey

Completed at each location on each visit, before the sample is drawn, and recorded on the field form. The survey supports the TSM's requirement that the submission discuss fecal sources and their age and proximity.

• Counts of gulls, geese, dogs and other animals present.
• Number of people in and out of the water.
• Algal presence, floating material, litter and debris, odours, dead fish.
• Visible discharges, bank disturbance, or upstream activity relevant to loading.
• Water appearance and apparent turbidity, weather at time of sampling.

5.4 Sample collection

1. Collect at the sensor's depth and as close to the sensor's optical volume as safe access permits.
2. Open the container without touching the mouth or interior. If contaminated, discard and use a fresh container.
3. Submerge mouth-down, invert and draw the sample while moving the container away from the body, avoiding surface film and disturbed sediment.
4. Leave headspace sufficient for mixing. Seal immediately.
5. Take water temperature separately, not from the sample, to avoid contaminating it.
6. Place upright on ice immediately. Record collection time to the minute, since pairing depends on it.

5.5 Co-location with the sensor reading

A grab is paired to a sensor reading only where all of the following hold. These rules are applied in the data pipeline and are stated here so that field practice and analysis agree.

• Time. The sensor reading falls within ±20 minutes of the recorded grab time.
• Deployment validity. The grab time falls inside a valid installation window for the unit at that location. Readings before installation or after removal are never paired.
• Operating point. The reading is taken at the calibrated operating point, and where several are available within the window, the single reading nearest the target bias is used rather than an average across operating points.
• Submersion. The optical channel confirms the unit was in water at the time of the reading.
• Maximum permissible physical separation between grab point and sensor TO SETTLE.

5.6 Field records

Each sample carries: unique identifier, location, date and time to the minute, collector, container lot where applicable, air and water temperature, sanitary survey, antecedent precipitation, observed conditions, and time placed on ice. Chain-of-custody documentation and the transfer record to the analytical laboratory are TO SETTLE.

6 · Laboratory procedures

Reference analysis follows the manufacturer's procedure for Colilert with Quanti-Tray/2000 without modification: reagent addition, sealing, incubation at the specified temperature and duration, and reading of large-well and small-well positives against the MPN table.

• Holding time. Analysis begins as soon as practicable and within the reference method holding time. Method LUME-1 specifies processing within a target window and no more than 8 hours total; the example plan at TSM Appendix B states no more than 6 hours between collection and start of analysis for culture methods. The operative figure is TO SETTLE with the laboratory, and the shorter is adopted where they differ.
• Temperature control. Samples held 0–4 °C from collection to analysis, with receipt temperature recorded.
• Dilution. Where concentrations are expected to exceed the upper quantification limit, a dilution is run in parallel and the dilution factor recorded. Results are reported at the dilution-corrected value.
• Reporting. Results are reported as measured MPN/100 mL. Non-detects are reported as such and not substituted with a numeric value at the point of reporting; see Section 8.3.

7 · Quality assurance and quality control

7.1 Sensor QC

Governed by Method LUME-1 Section 9 and recorded on the per-unit conformance certificate. Acceptance values shown as provisional in the method remain provisional here and are listed in Section 11.

7.2 Reference method QC

• Field duplicates at a defined proportion of samples, to characterize combined sampling and analytical variability.
• Field and reagent blanks at a defined frequency, to confirm no carryover or contamination.
• Holding time and receipt temperature recorded for every sample; excursions flagged and the affected result qualified.
• Duplicate and blank frequencies are TO SETTLE with the laboratory.

Reference-method imprecision is not treated as zero. Replicate work on these two culture methods shows membrane filtration and Colilert disagreeing with each other more than the alternative indicator disagrees with either, and that disagreement bounds what any comparison can achieve.

7.3 Data quality objectives

• Completeness. A target proportion of planned samples successfully collected, analyzed within holding time and paired to a valid sensor reading. Target TO SETTLE.
• Representativeness. Samples distributed across flow bands and across the full recreation season, assessed as in Section 3.3.
• Comparability. A single reference method, a single frozen calibration model, and consistent pairing rules throughout.
• Sufficiency. At least 30 paired observations within both quantification limits.

8 · Data management

8.1 Time base

All timestamps are normalized to UTC before pairing. Sensor telemetry and installation windows are recorded in UTC. Field-collected times are converted from the recorded local time using the recorded time zone rather than an assumed offset, because a stored local clock value combined with an assumed zone has previously produced multi-hour pairing errors.

8.2 Pairing

Pairing applies the rules in Section 5.5 deterministically: install-window validity first, then the ±20 minute time window, then selection of the single reading nearest the calibrated operating point. A grab with no valid in-window reading is left unpaired and reported as such. It is not paired to a nearby unit or to a reading outside the window.

8.3 Non-detects and values below quantification

This is the single most consequential data-handling rule in the plan, because it determines which pairs enter the statistical comparison.

• A reference non-detect is reported and retained as a non-detect. It is not substituted with a small positive value in the submitted dataset.
• A pair is marked BLOQ where either measurement falls below its method's lower quantification limit in Section 4.3.
• BLOQ pairs and zero values are excluded from the Step 3 computation. The AltCalc tool will not accept zero or negative values, and pairs entered as blank or as BLOQ are excluded from the analysis by the tool itself.
• Any substitution used internally for log-scale model fitting is documented as such and does not travel to the submitted worksheet.
• Counts of excluded pairs are reported alongside the result, so the surviving n is transparent.

8.4 Exclusions

Exclusions are documented, justified individually, and reported with the analysis. Permissible grounds are limited to: a documented instrument fault, a reading outside a valid installation window, a sample outside holding time or temperature control, or a failed QC check bracketing the sample. A result is never excluded because it worsens a performance metric. Fouled intervals are corrected under the method's fouling procedure and retained where correction is valid, rather than discarded.

9 · Data analysis

Analysis follows Step 3 of EPA-820-R-14-011, computed in the EPA Alternative Methods Calculator Tool. A workbook preloaded with the current indicative Boulder Creek pairs is provided for review: AltCalc workbook, Segment 2b (Excel) and the same pairs as CSV for pasting into a clean copy of the tool. The workbook is indicative and will be rebuilt on the final dataset and the frozen calibration model before submission.

1. Assemble paired measured values, reference in the Method 1 column and alternative indicator in the Method 2 column, as actual measured values rather than log-transformed.
2. Mark BLOQ pairs per Section 8.3; the tool excludes them.
3. Compute the index of agreement and R-squared on log₁₀-transformed values.
4. Record the tool's outputs together with the surviving pair count, and complete the tool's data documentation tab naming this plan, the QA project plan, and the sanitary survey record.

Decision rule. If IA ≥ 0.7, the alternative indicator may be used and the site-specific criteria take the same numeric values as the EPA method, with no further statistical analysis required. If IA < 0.7 but R-squared > 0.6, new criteria values are derived by least-squares regression of the log-transformed data. If both fall below their thresholds, this pathway cannot be used.

Supporting analyses reported alongside the acceptance statistics, for interpretation rather than for acceptance: classification performance at the action limit, ranking performance, and absolute error in log units. Where an evaluation sample contains few sites, short windows or repeated measures, R-squared is reported but not read as a standalone verdict, because its reference point is the pooled mean of that particular sample.

10 · Records, reporting and revision

• Retention. Field forms, sanitary surveys, chain of custody, laboratory records, calibration and conformance certificates, raw sensor telemetry and the analysis workbook are retained for the life of the standard and any subsequent review.
• Reporting. The submission package comprises this plan, Method LUME-1, the paired dataset, the completed calculator workbook, and the derivation of the site-specific criteria, per the WQS submission checklist at Section 4 of the TSM.
• Triennial reevaluation. The TSM directs that site-specific alternative criteria be reevaluated every three years, aligned to the state's triennial review of water quality standards, to confirm that the relationship between the indicator and the reference method has remained valid. The reevaluation re-runs the Step 3 comparison on the accumulated record; it is an analysis rather than a new field campaign.
• Revision. Changes to this plan are versioned, dated, and communicated to the Division before the affected sampling occurs.

11 · Items to settle before execution

Each is a decision rather than a data gap, and each has an owner. The first three were added on 29 August 2026 in response to the City’s second round of questions and rank above the rest, because each of them changes what the subsequent work should look like. The reasoning is set out in the second-round answers.

12 · References

1. U.S. EPA. Site-Specific Alternative Recreational Criteria Technical Support Materials for Alternative Indicators and Methods. EPA-820-R-14-011, Office of Water, December 2014. PDF
2. U.S. EPA. Alternative Methods Calculator Tool User Guide. EPA 821-B-21-001, November 2021. Tool: EPA 821-B-21-002. Populated for this study: Excel workbook.
3. U.S. EPA. Recreational Water Quality Criteria. 2012.
4. U.S. EPA. Method Validation of U.S. Environmental Protection Agency Microbiological Methods of Analysis. 2009.
5. Virridy. Method LUME-1: E. coli in ambient fresh water by continuous in-situ tryptophan-like fluorescence. PDF
6. Virridy. Per-Unit Conformance and Calibration Certificate. PDF
7. Sorensen, J.P.R. et al. Real-time detection of faecally contaminated drinking water with tryptophan-like fluorescence. 2018.
8. Colorado Water Quality Control Commission. Regulation No. 31, The Basic Standards and Methodologies for Surface Water, 5 CCR 1002-31 (effective 31 December 2024) — Table I footnote 7, two-month geometric mean; Regulation No. 38, 5 CCR 1002-38 — COSPBO02B; and Regulation No. 93.
9. Colorado Water Quality Control Division. Colorado’s Section 303(d) Listing Methodology, 2026 Listing Cycle. March 2024.
10. Colorado Water Quality Control Commission. Policy 25-1: Advancing External Proposals for Revised Water Quality Classifications and Standards (in Regulations Nos. 31-38) before the Water Quality Control Commission. Adopted 12 May 2025.
11. GEI Consultants for the City of Boulder. E. coli TMDL Implementation Plan, June 2019, for the Boulder Creek Segment 2b TMDL approved by EPA 27 September 2011.
12. 40 CFR 131.20(a); 40 CFR 136.1.

Prepared by Virridy with the Mortenson Center in Global Engineering and Resilience, University of Colorado Boulder. Draft for review by the City of Boulder Utilities Department prior to submission to CDPHE. Figures describing the existing paired record are indicative and were current at the time of writing; the record is growing.