Alta Collins precipitation, MOS round 3

Issued Mon 2026-09-28 19:00 MDT (replay, models from the ClickHouse archive) · cln_precip_r3_20261024 · HRRR 4-point mean

Issue time (local; empty for now) −6 h−1 h+1 h+6 hnow Gauge in hand1 h late2 h late3 h late6 h late12 h late18 h lateoff HRRR in hand1 h late2 h late3 h late6 h late12 h late18 h lateoff GFS in hand1 h late2 h late3 h late6 h late12 h late18 h lateoff GEFS in hand1 h late2 h late3 h late6 h late12 h late18 h lateoff AIFS single in hand1 h late2 h late3 h late6 h late12 h late18 h lateoff Cut-offs HRRR points 4-point meantorchlight's weights Show

Storms at CLN since 10-20 12:00 MDT (at least 0.05 in; current sources and cut-offs kept)

• 10-25 14:00 to 10-25 18:00 MDT, 0.08 in: as if issued 12 h · 6 h · 3 h · 1 h before
• 10-22 17:00 to 10-23 09:00 MDT, 1.19 in: as if issued 12 h · 6 h · 3 h · 1 h before
• 10-22 11:00 to 10-22 14:00 MDT, 0.29 in: as if issued 12 h · 6 h · 3 h · 1 h before
• 10-21 14:00 to 10-21 18:00 MDT, 0.17 in: as if issued 12 h · 6 h · 3 h · 1 h before
• 10-20 18:00 to 10-20 19:00 MDT, 0.16 in: as if issued 6 h · 3 h · 1 h before

Totals from the start of the issue hour, each 15-minute step included at its end; inches of water. Expected (p × the amount if wet) is the forecast total. Its areas split it into likely, from hours with p ≥ 0.5, and phantom, from hours with p < 0.5. Called is each hour as wet, max(amount if wet, 0.01) in, when p is at or above the cut-off, or dry, 0: what an hourly view shows. Its sum is not a total forecast: it counts each called hour as certain and the rest as 0, and in a storm of confident hours it runs above expected. The lower chart has each hour on its own: called at 0.5 against expected, and in replays the gauge. The gauge counts only the hours the forecast has. HRRR: the 10 newest runs in hand, oldest lightest; torchlight's weights are RTP 0.929558, albion 1.061006, castle 0.974476, superior 1.078902, summed and divided by 4. Each MOS hour is split into 15 minutes as the newest HRRR cycle in hand splits it, evenly without one. In a replay or with a source late, a datum counts only if it would have arrived by then: HRRR f(h) 52 + 2 (h − 1) minutes after the cycle, the gauge 10 minutes after the hour, GFS and GEFS 6 hours and AIFS single 7 hours after the run. Built in 0.83 s.