WeatherBlend

Multi-model forecast blending for Bonehill Rocks, Dartmoor

Dry-window forecast

Per-station, per-window blender. The label asks "is there a contiguous N-hour dry block somewhere within 09:00–18:00 local time?" — the realistic outdoor-walking window at Bonehill year-round (DST handled per target day). Phase 3b (53-feature LightGBM) ships as champion; Phase 3g (parameter-free MC over Phase 3a hourly P(wet)) ships alongside as challenger and guarantees cross-window monotonicity by construction. Cards show today and the next 4 forecast days.

Bellever Dartmoor

3-hour dry window

Phase 3b — lean (53 features)

Day-aggregate per-model precip totals, wet-hour counts, run-length stats, EA persistence, climatology, calendar encodings. Production champion.

Target date (UTC) +24h +48h +72h Model agreement MC longest dry run (3g only) Conformal (90% set) Best start (UTC, calibrated %)
99%91%81% 71% confident dry day 08:00Z (16%)
99%100%70% 71% confident dry day 14:00Z (15%)
100%33% 71% confident dry day 10:00Z (17%)
100% 57% confident dry day 08:00Z (10%)
Phase 3g — Monte Carlo over Phase 3a hourly P(wet) marginals

Parameter-free. For each daytime hour, sample 10,000 Bernoullis using Phase 3a's hourly P(wet); count the fraction of samples whose longest dry run reaches the target window length. No LightGBM, no learned weights — the prediction is purely 3a's per-hour view + the structural rule that longer windows are rarer. Cross-window monotonicity P(N=3) ≥ P(N=4) ≥ P(N=6) holds by construction (single MC pass, three indicators read off the same Bernoulli sequence).

Target date (UTC) +24h +48h +72h Model agreement MC longest dry run (3g only) Conformal (90% set) Best start (UTC, calibrated %)
100% med 9h (80%: 4-9h) confident dry day 08:00Z (16%)
100%99% med 9h (80%: 4-9h) confident dry day 14:00Z (15%)
98%64% med 6h (80%: 4-9h) confident dry day 10:00Z (17%)
99% med 7h (80%: 4-9h) confident dry day 08:00Z (10%)

4-hour dry window

Phase 3b — lean (53 features)

Day-aggregate per-model precip totals, wet-hour counts, run-length stats, EA persistence, climatology, calendar encodings. Production champion.

Target date (UTC) +24h +48h +72h Model agreement MC longest dry run (3g only) Conformal (90% set) Best start (UTC, calibrated %)
100%59%75% 71% confident dry day 08:00Z (18%)
100%96%60% 71% confident dry day 13:00Z (18%)
63%31% 71% confident dry day 09:00Z (19%)
100% 57% confident dry day 08:00Z (12%)
Phase 3g — Monte Carlo over Phase 3a hourly P(wet) marginals

Parameter-free. For each daytime hour, sample 10,000 Bernoullis using Phase 3a's hourly P(wet); count the fraction of samples whose longest dry run reaches the target window length. No LightGBM, no learned weights — the prediction is purely 3a's per-hour view + the structural rule that longer windows are rarer. Cross-window monotonicity P(N=3) ≥ P(N=4) ≥ P(N=6) holds by construction (single MC pass, three indicators read off the same Bernoulli sequence).

Target date (UTC) +24h +48h +72h Model agreement MC longest dry run (3g only) Conformal (90% set) Best start (UTC, calibrated %)
97% med 9h (80%: 4-9h) confident dry day 08:00Z (18%)
98%92% med 9h (80%: 4-9h) confident dry day 13:00Z (18%)
91%34% med 6h (80%: 4-9h) confident dry day 09:00Z (19%)
92% med 7h (80%: 4-9h) confident dry day 08:00Z (12%)

6-hour dry window

Phase 3b — lean (53 features)

Day-aggregate per-model precip totals, wet-hour counts, run-length stats, EA persistence, climatology, calendar encodings. Production champion.

Target date (UTC) +24h +48h +72h Model agreement MC longest dry run (3g only) Conformal (90% set) Best start (UTC, calibrated %)
88%38%36% 57% confident dry day 08:00Z (10%)
77%55%79% 43% confident dry day 11:00Z (20%)
29%29% 29% confident wet day 08:00Z (23%)
89% 43% confident dry day 08:00Z (11%)
Phase 3g — Monte Carlo over Phase 3a hourly P(wet) marginals

Parameter-free. For each daytime hour, sample 10,000 Bernoullis using Phase 3a's hourly P(wet); count the fraction of samples whose longest dry run reaches the target window length. No LightGBM, no learned weights — the prediction is purely 3a's per-hour view + the structural rule that longer windows are rarer. Cross-window monotonicity P(N=3) ≥ P(N=4) ≥ P(N=6) holds by construction (single MC pass, three indicators read off the same Bernoulli sequence).

Target date (UTC) +24h +48h +72h Model agreement MC longest dry run (3g only) Conformal (90% set) Best start (UTC, calibrated %)
78% med 9h (80%: 4-9h) confident dry day 08:00Z (10%)
78%62% med 9h (80%: 4-9h) confident dry day 11:00Z (20%)
65%10% med 6h (80%: 4-9h) confident dry day 08:00Z (23%)
67% med 7h (80%: 4-9h) confident dry day 08:00Z (11%)

A dry "hour" requires all four 15-min EA gauge readings to be ≤ 0.1 mm. Search is bounded to 09:00–18:00 local time (Europe/London, DST-aware) — overnight dry stretches don't count, and a dry block that bridges 18:00 into the evening isn't credited. Cross-midnight dry stretches are not credited (UTC-day boundary).