Every monthly report has a "Loud Daytime Events" table, an informational ranking, no ANAC threshold attached since daytime noise isn't regulated. Went back through June, July and August after spotting the pattern in June's table. Fifteen entries across the three, five in June (EZY75QC 124.1 dB, RYR95FC 120.8 dB, TAP1040 116.3 dB, and two more at 112.3 and 112.0), four in July (topped by 49520f at 122.3 dB), six in August (topped by TAP93ZR at 124.2 dB), don't hold up anymore.
This week's new verification work went looking through the full history, not just new flights, and found the same signature on all fifteen: numbers that don't match anything else that specific aircraft has ever registered, on this route, on any other day. Three of the June ones landed within an hour of each other on the 25th. Sensor glitches, not real passes, and worth being upfront that this took three separate incidents on three separate days across three months before the pattern was obvious enough to build a real check for.
Went looking for more once the pattern was clear. Below these fifteen, both July and August settle straight back into a smooth, ordinary spread (100 to 110 dB, dozens of different aircraft and airlines, no more isolated jumps). May's report has a different, already-disclosed issue (the microphone was hardware-capped at 100 dB before June's upgrade, noted on the report itself), not the same thing.
The public archive already reflects the correction, the real numbers for these flights sit where they should now. None of the three PDFs are being reissued. Editing a report after it's been downloaded and read felt like the wrong instinct, so instead: here's the correction, in writing, attached to the record rather than hidden inside it. Everything else in those reports, the ANAC alert counts, the night figures, wasn't affected and stands as published.
Same reasoning as always here: a mistake found and written down openly is worth more than one quietly smoothed over.