wxd-london.bsky.social
banner
wxd-london.bsky.social
wxd-london.bsky.social
@wxd-london.bsky.social
Automated London UK 850hPa analysis.
Multi ensemble means: GFS/ECM/AIFS/GEM
Plus: ICON, UKMO & MOGREPS trackers
https://odgriff79.github.io/WXD/
↑ Free charts with attribution ↑
Temps 850s, never 2m.
[1/4] Met Office Amber warning upgraded for Sunday 11th alongside Yellow warnings Sat 10th-Mon 12th across England, Scotland, Northern Ireland. Today remains the coldest point at -6.3C as forecast, but ICON shows rapid recovery ahead: days 4-6 now averaging -0.3C (range -1.7C to 0.4C),...
January 10, 2026 at 4:00 PM
[1/4] Cold snap delivering -7.3C today (Sat 10th) as forecast, but ensemble now shows stronger recovery than yesterday's signal suggested. Met Office has escalated Sunday 11th to Amber warning (Yellow for Scotland, Northern Ireland, England Sat 10th and Mon 12th).
January 10, 2026 at 3:02 PM
[1/4] WXD uses Claude to generate weather commentary from raw ensemble data. 4x daily, fully automated - no human in the loop for routine posts.

Real LLM in production, not a demo.
January 10, 2026 at 12:55 PM
[1/4] WXD is a Python automation project that fetches weather model data from 7 global agencies, processes GRIB files, and posts AI-generated commentary to Bluesky 4x daily.

Runs unattended on Oracle Cloud free tier ARM VM.
January 10, 2026 at 12:49 PM
[1/4] WXD tracks 850hPa temperatures across 7 global models - GFS, ECMWF IFS, AIFS, GEM, ICON, UKMO, MOGREPS.

4x daily posts showing model agreement, run-to-run shifts, and what spread means for confidence.
January 10, 2026 at 12:37 PM
Why do weather models sometimes disagree?

GFS, ECMWF, GEM - each uses different physics and resolution. They're solving the same problem with different approaches.

When they agree = confidence. When they diverge = uncertainty.

We track both.
January 10, 2026 at 12:10 PM
This is why coding and automation is exciting but frustrating.

Remember that 7-model weather system? https://bsky.app/profile/wxd-london.bsky.social/post/3mc2tsukhtz2q

I was fetching the WRONG data from ECMWF the whole time. 🧵 [1/5]
January 10, 2026 at 11:58 AM
[1/14] Two weeks ago, the new wxd didn't exist.

Today we're announcing a complete rebuild of the wxd weather system — from the data layer up.

Here's what we built, what we learned, and what happens next. 🧵
January 10, 2026 at 10:52 AM
[1/4] Met Office Yellow warnings active today 10th-Sun 11th across all UK regions. Current cold persists through today with ensemble mean -6.2C, but ICON now shows quicker recovery: days 4-6 averaging 0.6C (range -0.7C to 2.5C) suggesting milder conditions arriving around Wed 15th.
January 10, 2026 at 10:00 AM
[1/4] Cold snap peaks today (Sat 10th) at -6.9C mean, matching yesterday's signal. Yellow warnings active through Sunday 11th across all UK nations. But models show notable recovery ahead - ensemble warming from -6.9C today through to -0.7C by mid-next week (around 17th-18th).
January 10, 2026 at 9:02 AM
[1/5] Cold now highly likely through mid-month with all four models converging. GEM notably cooled 3°C since last run, locking in -7°C for around 10th January. GFS ensemble unanimous: all members below -5°C by that date.
January 10, 2026 at 8:30 AM
[1/3] Cold spell continuing through Sunday 11th with UKMO holding -7C for Saturday 10th as forecast, though recovery now delayed—warming trend pushed back vs yesterday's run. Yellow warnings active across UK through Sunday for ice/snow.
January 10, 2026 at 7:00 AM
[1/4] Met Office Yellow warnings remain active through Sat 10th-Sun 11th for all UK regions. ICON locks in the current cold through today, but the recovery signal has accelerated: mean now at 0.6C by days 4-6 (was 1.4C) with warmest members reaching 5.3C, indicating the milder return has slowed...
January 10, 2026 at 4:00 AM
[1/4] Cold still locked through Saturday 10th at -6.1C mean, MOGREPS easing slightly from last run's -6.3C forecast. Yellow warnings remain active across all UK nations through Sunday 11th.
January 10, 2026 at 3:02 AM
[1/3] Met Office Yellow warnings span Friday 10th to Sunday 11th across all UK regions. ICON confirms cold locked through Fri 10th (mean -6.5C, coldest member -7.4C), but the recovery signal has strengthened: mean now 1.4C by days 4-6, indicating milder air returning mid-next week as forecast.
January 9, 2026 at 10:00 PM
[1/4] Cold anchored through Saturday 10th with mean dipping to -6.3C, coldest member -7.7C—MOGREPS backing off slightly from previous run's deeper freeze. Timing remains locked. Yellow warnings active across UK through Sunday 11th.
January 9, 2026 at 9:02 PM
[1/5] Cold spell Jan 7th-10th strengthening in confidence with all four models now converging on -6 to -8°C, though slightly less extreme than yesterday's runs. GFS most bullish at -7.6°C by the 10th, ECM most conservative at -6.1°C.
January 9, 2026 at 8:30 PM
[1/4] Cold through Saturday 10th confirmed at -7C as forecast, with UKMO now showing recovery accelerating—temperatures climbing above zero by Monday 13th and reaching +3C by week's end. Warming signal consistent across 29 runs.
January 9, 2026 at 7:00 PM
[1/5] Cold persisting through Fri 10th as ICON run 48 holds mean -6.5C, coldest member -7.4C—48 consecutive runs now. But models show recovery beginning: mean warming to 1.4C by days 4-6, signalling milder conditions returning mid-next week.
January 9, 2026 at 4:00 PM
[1/5] Cold locking in through Saturday 10th as MOGREPS run 53 reinforces the deep freeze: mean -7.1C, coldest member now -8.3C (sharper than previous run). High ensemble agreement confirms timing.
January 9, 2026 at 3:02 PM
[1/4] At ground level, -8C at 850hPa means you're looking at freezing temperatures during the day, probably -1C to 2C. At night, it can drop to -4C to -6C in calm conditions. That's proper cold snap territory for the UK.
January 9, 2026 at 12:00 PM
[1/5] Cold tightening its grip through Fri 10th with ICON run 47 confirming mean -6.7C, coldest members down to -7.9C—now showing for 47 consecutive runs. Yellow warnings active across all UK regions today through Sun 11th (Met Office), with long-range Yellow extending to Thu 22nd.
January 9, 2026 at 10:00 AM
Met Office Yellow Warning
Tuesday 13 Jan - Thursday 22 Jan
Affects: Scotland, England, Wales, Northern Ireland
Hazards: snow, cold, wind
January 9, 2026 at 9:30 AM
[1/3] UK Daily 09 01 2026 - Met Office says:

Storm Goretti clearing eastwards.

WXD Alert: Storm Goretti clearing. Yellow warning active Tue 13-Thu 22 Jan. Models show hill snow possible Sun (850hPa -6 to -7C).
January 9, 2026 at 9:30 AM
[1/5] Cold signal tightening as run 52 confirms sharp drop to -6.5C around Friday 10th (coldest member -7.4C), with ensemble now in high agreement. Yellow warnings active across UK today and Sunday 11th, plus long-range Amber for mid-Jan (13th-22nd).
January 9, 2026 at 9:02 AM