WeatherBlend

Multi-model forecast blending for Bonehill Rocks, Dartmoor

Dry-window forecast

P(at least one N-hour dry block in 09:00–18:00 local). 3b champion plus Monte-Carlo challengers.

Princetown

2-hour dry window

Phase 3p — Gaussian copula MC over Phase 3o hourly P(wet)

Gaussian copula MC over the rich+orographic Phase 3o's hourly P(wet) marginals. Single empirical Σ per station, fit on train-split observed daytime wet/dry binary sequences (daytime-shape autocorrelation is lead-independent for truth, so per-lead Σs pool into one). Captures within-day wet/dry autocorrelation that an iid sampler misses. Bake-off Brier 0.1064 aggregate on the 9 (station, lead) Bonehill cells.

Target date (UTC) +24h +48h +72h Confidence Best start (UTC, P dry block)
100%100%100% confident dry (P=100% · run 7–9h) 13:00Z (98%)
100%100%100% confident dry (P=100% · run 9–9h) 13:00Z (99%)
—100%100% confident dry (P=100% · run 7–9h) 13:00Z (98%)
——100% confident dry (P=100% · run 9–9h) 11:00Z (98%)
Monte-Carlo P(an 2-hour dry block runs from each start hour) — one line per forecast horizon, soonest target day shown. The peak is the best time to set off; the height is how likely it stays dry. Each hour is its own probability, so the lines need not sum to the daily "any dry window" figure.

3-hour dry window

Phase 3b — lean (53 features)

53-feature LightGBM. Champion.

Target date (UTC) +24h +48h +72h Model agreement Confidence
99%99%96% 67% confident dry (P=99% · τ=9%)
100%98%96% 100% confident dry (P=100% · τ=9%)
—96%95% 100% confident dry (P=96% · τ=24%)
——94% 100% confident dry (P=94% · τ=40%)
Phase 3p — Gaussian copula MC over Phase 3o hourly P(wet)

Gaussian copula MC over the rich+orographic Phase 3o's hourly P(wet) marginals. Single empirical Σ per station, fit on train-split observed daytime wet/dry binary sequences (daytime-shape autocorrelation is lead-independent for truth, so per-lead Σs pool into one). Captures within-day wet/dry autocorrelation that an iid sampler misses. Bake-off Brier 0.1064 aggregate on the 9 (station, lead) Bonehill cells.

Target date (UTC) +24h +48h +72h Confidence Best start (UTC, P dry block)
100%100%100% confident dry (P=100% · run 7–9h) 13:00Z (96%)
100%100%100% confident dry (P=100% · run 9–9h) 13:00Z (98%)
—99%99% confident dry (P=99% · run 7–9h) 13:00Z (97%)
——100% confident dry (P=100% · run 9–9h) 14:00Z (97%)
Monte-Carlo P(an 3-hour dry block runs from each start hour) — one line per forecast horizon, soonest target day shown. The peak is the best time to set off; the height is how likely it stays dry. Each hour is its own probability, so the lines need not sum to the daily "any dry window" figure.

4-hour dry window

Phase 3b — lean (53 features)

53-feature LightGBM. Champion.

Target date (UTC) +24h +48h +72h Model agreement Confidence
94%99%99% 67% confident dry (P=94% · τ=23%)
99%99%99% 100% confident dry (P=99% · τ=23%)
—98%91% 100% confident dry (P=98% · τ=25%)
——98% 100% confident dry (P=98% · τ=29%)
Phase 3p — Gaussian copula MC over Phase 3o hourly P(wet)

Gaussian copula MC over the rich+orographic Phase 3o's hourly P(wet) marginals. Single empirical Σ per station, fit on train-split observed daytime wet/dry binary sequences (daytime-shape autocorrelation is lead-independent for truth, so per-lead Σs pool into one). Captures within-day wet/dry autocorrelation that an iid sampler misses. Bake-off Brier 0.1064 aggregate on the 9 (station, lead) Bonehill cells.

Target date (UTC) +24h +48h +72h Confidence Best start (UTC, P dry block)
99%98%99% confident dry (P=99% · run 7–9h) 13:00Z (95%)
100%100%100% confident dry (P=100% · run 9–9h) 12:00Z (98%)
—99%97% confident dry (P=99% · run 7–9h) 13:00Z (96%)
——99% confident dry (P=99% · run 9–9h) 13:00Z (97%)
Monte-Carlo P(an 4-hour dry block runs from each start hour) — one line per forecast horizon, soonest target day shown. The peak is the best time to set off; the height is how likely it stays dry. Each hour is its own probability, so the lines need not sum to the daily "any dry window" figure.

5-hour dry window

Phase 3p — Gaussian copula MC over Phase 3o hourly P(wet)

Gaussian copula MC over the rich+orographic Phase 3o's hourly P(wet) marginals. Single empirical Σ per station, fit on train-split observed daytime wet/dry binary sequences (daytime-shape autocorrelation is lead-independent for truth, so per-lead Σs pool into one). Captures within-day wet/dry autocorrelation that an iid sampler misses. Bake-off Brier 0.1064 aggregate on the 9 (station, lead) Bonehill cells.

Target date (UTC) +24h +48h +72h Confidence Best start (UTC, P dry block)
96%96%98% confident dry (P=96% · run 7–9h) 12:00Z (94%)
98%99%99% confident dry (P=98% · run 9–9h) 12:00Z (97%)
—97%85% confident dry (P=97% · run 7–9h) 12:00Z (95%)
——98% confident dry (P=98% · run 9–9h) 12:00Z (96%)
Monte-Carlo P(an 5-hour dry block runs from each start hour) — one line per forecast horizon, soonest target day shown. The peak is the best time to set off; the height is how likely it stays dry. Each hour is its own probability, so the lines need not sum to the daily "any dry window" figure.

6-hour dry window

Phase 3b — lean (53 features)

53-feature LightGBM. Champion.

Target date (UTC) +24h +48h +72h Model agreement Confidence
76%91%94% 33% confident dry (P=76% · τ=39%)
98%96%95% 100% confident dry (P=98% · τ=39%)
—84%83% 83% confident dry (P=84% · τ=40%)
——91% 83% confident dry (P=91% · τ=41%)
Phase 3p — Gaussian copula MC over Phase 3o hourly P(wet)

Gaussian copula MC over the rich+orographic Phase 3o's hourly P(wet) marginals. Single empirical Σ per station, fit on train-split observed daytime wet/dry binary sequences (daytime-shape autocorrelation is lead-independent for truth, so per-lead Σs pool into one). Captures within-day wet/dry autocorrelation that an iid sampler misses. Bake-off Brier 0.1064 aggregate on the 9 (station, lead) Bonehill cells.

Target date (UTC) +24h +48h +72h Confidence Best start (UTC, P dry block)
95%92%96% confident dry (P=95% · run 7–9h) 11:00Z (92%)
97%98%98% confident dry (P=97% · run 9–9h) 11:00Z (96%)
—95%78% confident dry (P=95% · run 7–9h) 11:00Z (94%)
——97% confident dry (P=97% · run 9–9h) 11:00Z (95%)
Monte-Carlo P(an 6-hour dry block runs from each start hour) — one line per forecast horizon, soonest target day shown. The peak is the best time to set off; the height is how likely it stays dry. Each hour is its own probability, so the lines need not sum to the daily "any dry window" figure.

Dry hour = all four 15-min readings ≤ 0.1 mm. Search bounded to 09:00–18:00 local.