Sensor calibration data, reference-method comparisons, and digital MRV documentation for Virridy's Lume water-quality sensing platform.
How a reading travels from the device through Notehub, Pumphaus, the inventory database, the baseline store and the E. coli model to a customer dashboard — plus how to reach the validation grabs, the invariants that must hold at every join, and the audit that checks them.
The synthesis across the burn-in fleet, the four-batch barrel and the vacuum degassing rig: whether temperature is correctable, whether bubbles are the everyday mechanism, and what drift actually is once windows stop straddling interventions. Carries the degassing test in full.
Paired calibration against two EPA-approved reference methods — Colilert (IDEXX, MPN) and membrane filtration (CFU) — with lab, field, and combined models.
→Calibration against the Aquagenx CBT used in international monitoring programmes, with field data from Rwanda and Kenya deployments.
→Digital Measurement, Reporting & Verification for the Amazi Meza Rwanda water treatment carbon credit programme. Includes the implementation plan, pilot report with FAR resolution evidence, and VVB findings.
→Time-of-flight sensor calibration to Nephelometric Turbidity Units using laboratory and field reference standards.
→Does distance_mm start carrying information? The channel reported a single constant value on 99.87% of readings — a sentinel emitted when ranging fails, not a measurement. Seven units on the candidate firmware.
Amplitude alone cannot separate a film on the window from particles in the path
— both lift the floor, and the streamflow guard cannot run where there is no
gauge. Now that distance_mm works, it supplies the missing axis:
turbidity moves the amplitude and leaves the distance alone, while fouling
shortens the distance as sediment builds on the cap. Tracked across
the deployed fleet against the cleaning registry.
Three sensors in one tank, each with its own ultrasonic oscillator, against water an air pump keeps continuously bubbling. Staggering the oscillators puts treated and untreated units in the same water at the same instant, so the comparison cancels temperature — and every episode is judged on whether a step survives switch-off, against a measured null.
→Tryptophan serial-dilution bench test characterizing the TLF fluorescence channel against known concentration, read live across 14 sensors.
→Developing the TLF channel into a published, standardized method and getting it recognized by EPA for recreational-water screening — via the alternative-indicator and standard-method routes, plus AOAC PTM. Background, precedents, the plan, and a working draft of the method itself.
→FDOM, chlorophyll-a, new TLF and the 13-day-old burn-in fleet in one tank: drift, temperature response and the response to a shake, read side by side, across a nine-day soak in shared DI water. Carries the full burn-in campaign. Closed 2026-08-29 09:00 and frozen — a static record ending where the spike test begins.
→Thirty-nine units and three AquaTroll 500 sondes, now dosed with challenge solutions one at a time. Which channels move, by how much, and which move that should not. Opens 2026-08-29 09:00 — updates as the test runs.
→Flowing vs. still water classification for piped systems, supporting SDWS 23 (water volume) monitoring.
→Lume sensor performance in wastewater matrices — high-turbidity, high-fluorescence environments beyond typical drinking-water ranges.
→