Some days at Lisbon Finals aren't about a new sensor or a new page. They're about making sure the ones already running are telling the truth, clearly. Today was one of those days.
A chart on Archive's Expert Mode had been quietly lying by omission. Four measurements (noise, peak level, wind, exposure time) were squeezed into the same cramped strip, each one flattened to fit a shared scale that belonged to none of them. It now gets four cards of its own, each reading at the scale its numbers actually deserve.
Elsewhere, a "sensor coverage" number on Air Quality was scoring itself against days before the sensor even existed: 93.4% of a test it was never eligible to take. Recalculated against its real start date, it turned out to be 99.4% reliable all along. A separate correction restored Landing and Takeoff figures on Compare that had quietly gone missing after a pipeline consolidation earlier in the day, the kind of regression that only shows up once you go looking.
None of this changes what the station measures. It changes whether the numbers are being read correctly. For a project with nothing to offer but its own numbers, that's not a small thing.
One small addition for anyone who's never had a reason to know what a decibel costs: a quick reference on the Data page ("how loud is that, really") walking through everyday levels, from the threshold of hearing to an aircraft passing 100 metres overhead.