UPPER CHERRY · good between 200 and 145 cfs on the creek itself (CDEC UCC)
Wrapped: Jun 11 → Jun 16 — a 5-day window
The lake-inflow number ran Jun 08 → Jun 24 — the early warning the
creek follows within ±3 days. 34% snow — every warm day counted double.
Our call
April 1 said
Jun 3 — from the snow survey alone, 9 weeks out
June 1 said
Jun 11 — 10 days out
River did
Jun 11 on the creek — dead on the call (the inflow
number crossed Jun 08, right on the early-warning schedule)
How this season played out
The run itself — live flow off Upper Cherry Creek (CDEC UCC), the number this
page is graded on. A 5-day window this year: the creek drops out while the
lake is still catching up.
The early warning: computed Cherry Lake inflow (storage + outflows, the same math
Dreamflows does) — the creek follows this number within ±3 days in clean years, and it carries the
40-year record the model trains on. Gray dash = the median year (1986–2025), literally off this
year's chart until mid-June: that's what 34% snow looks like.
Why we called it
Snowpack was 34% of normal on April 1 — the thinnest pack in the Emigrant in
years. Everything pointed early and short.
May ran warm — about 2°C over normal up here, and on a thin pack every warm
day counts double. The inflow number beat the call by 3 days; the creek landed on it.
Emigrant granite drains clean. The recession is steep and sharp, so the
crossing dates are sharp too — this is why Cherry calls land within ~3 days while the rest of
the Sierra gets ±5.
What goes into this call — and where to check it yourself
The run itself: CDEC UCC
(live since 2015) · the inflow math: Dreamflows —
Cherry above Cherry Lake · computed here daily from
Cherry Lake
storage (CHV) + the powerhouse and dam outflows. (The old Cherry-near-Hetch-Hetchy gauge
behind the classic 1,200 → 600 beta died in 1955 — every live number above the lake is computed,
ours and Dreamflows' alike.)
Every season 1986–2026 of computed inflow; 2015–2026 on the real gauge
Track record — every call since 2016
year
we said
river did
miss
2016
Jun 28
Jul 03
5 late
2017
Aug 07
Aug 06†
1 early
2018
Jun 21
Jun 11
10 early
2019
Aug 02
Aug 02†
dead on
2020
Jun 12
Jun 10
2 early
2021*
Jun 16
Jun 07
9 early
2022
Jun 13
Jun 14†
1 late
2023
Jul 31
Aug 01
1 late
2024
Jun 23
Jun 20
3 early
2025
Jun 13
Jun 16
3 late
2026
Jun 11
Jun 11
dead on
Typical miss ≈ 3 days; worst 10 (2021* aside). Graded on the creek itself now:
"river did" = the UCC gauge dropping under 200 and staying under. † years the creek gauge was
down — those grade on the computed inflow instead. Every "we said" is the June 1 call (2021*'s
is April), and the model still trains on 40 years of lake-inflow record while answering for the
creek. The monster years (2017, 2019, 2023) still graded within a day.
The one ugly one, 2018: the creek quit ten days before the lake math said it
would. We tested the obvious fixes — constant bias, snow-scaled offsets — and every one of them
grades worse than the straight call. Eight creek seasons is too few to fit a fudge factor
honestly, so we don't. Each year adds one.
*2021: dust on the snow, here too. April said Jun 16; the creek
gave a 2-day shot Jun 7–9 and that was the season (the inflow number had already crossed May 27).
The spring peaked at 1,600 cfs and the October storms out-peaked the entire snowmelt three to
one. It stays on the board and out of the math.
Wrapped for 2026 — the first 2027 call lands ~April 1
Snow survey drops, we run the model, you get the date. Four emails a year.
These are forecasts, not gauge readings — the river gets the last word.
Check the live gauge before you burn vacation days. Seen a run come in? Tell us —
every confirmation makes next year's call better.