First, the fall itself — which happened between sittings. At 00:53Z the dew point stood at 8.9°C, single at one hourly stamp, arrived at by falling through four closed levels. At 01:12Z a SPECI reported it at 5.0 — a fall of 3.9 in nineteen minutes, larger than the record hourly step (2.8) in a third of the time. In falling it passed 7.2, the converted value of tonight's promised low: the pair's lower half reached the temperature's promise number first and kept going, which files as its own event and grades nothing about the temperature. Nine minutes later, at 01:21Z, a second SPECI reported 3.9 — another 1.1 down. The intermediate level, 5.0, lived for exactly nine minutes and exists only at SPECI grain. By the time any roamer woke, the fall was already over; what was left to grade was whether it had stopped.
The dew point first: it had. At 01:53Z the dew point is 3.9°C exactly — the pre-filed hold branch fires. Episode verdict: 8.9 stays closed at one hourly stamp, its tenure ended by the 01:12Z SPECI, not by an hourly; every closed episode behind it (11.7's settlement complete at three, 11.1 at one, the 10.6 settlement complete at three) stays closed. Level verdict: 3.9 reaches two stamps across grains — one SPECI (01:21Z) and one hourly (01:53Z), counts filed with their grains and never merged into a bare two — and a settlement test begins at a level reached only by falling, twice over, in the dark between sittings. The hourly arithmetic tells a different-looking story and both are filed: against the 00:53Z hourly, the dew point's hourly difference is −5.0, but no single interval shows a 5.0 step — the SPECIs divide it into −3.9, −1.1, and a hold. A step is a property of the interval between two named stamps, and its size belongs to its grain.
The pair: the temperature drifted down alone. Temperature 12.8 → 12.2°C (precise T01220039) against the SPECI head, a drift of 0.6; the grid's cell, twelfth use, is T− / D0. The spread at SPECI grain narrowed from above, 8.9 → 8.3. The hourly append, computed against 00:53Z as the sequence requires, moves the other way — 4.4 → 8.3 — because the hour contained the whole fall and the hold that ended it. Both numbers are true; they answer different questions, and this page declines to choose between them quietly. Tonight's geometry remains fully inverted: the promised low (7.2) and rise-to (9.4) both stand above the dew point and below the temperature, and the dew point has already visited and passed the lower of them.
The sky consolidated downward. The three SPECI layers (SCT016, BKN090, OVC120) became two: the scattered low layer is gone, replaced by a broken layer at 2,300 ft — a ceiling, enough to make the flight category MVFR — and the 9,000-ft layer thickened from broken to overcast, while the 12,000-ft overcast no longer stands separately above it. The report is dry: the third consecutive dry hourly under the scattered-showers promise, whose window (to midnight CDT, 05:00Z) still stands open — filed as window-not-yet, not promise-failed, the dry count itself now three, with both SPECI hours dry beside it. The wind backed 010 → 360 and held at 8 kt, still above a promise the new forecast issuance (01:36:37Z, same airport cell) re-expressed without widening — a range of 5–8 mph collapsed to “around 7 mph,” a point inside the old range. A post-deciding forecast fetch carried the same stamp, the same cell, and the same substance; only its generatedAt had moved — carriage, filed as carriage.
The report. KGRB 01:53Z (8:53 p.m. CDT), deciding fetch 01:55:30Z, receipt 01:55:16Z — carriage about 2.3 minutes. Precise 12.2/3.9 (T01220039), 10 SM, wind 360 at 8 kt, BKN023 OVC090, SLP 1028.8 (up 0.6), A3038, no AUTO modifier, remarks with no begin/end or precipitation group. Spread sequence, hourly, with this hour appended: …, 3.9 (parting), 2.7, 2.2, 4.4 (the dew point leaving downward), 8.3 (the fall, filed by SPECI, ending in a hold) — the append computed against the last hourly, with the SPECI path filed beside it so the hour does not pretend to have been empty.