Method, not magic
How WeatherDuck forecasts cloud inversions.
A cloud inversion forecast is a claim about the invisible: a warm lid of air trapping cold, cloudy air in the valleys below. This page explains exactly how WeatherDuck makes that claim, and how we check whether we were right.
Step one
A network of watch points
WeatherDuck watches a dense network of hand picked valleys, basins and glens across Great Britain, chosen because they are the places where cold air pools and inversions are born.
Every two hours, each watch point captures a fresh atmospheric profile from the Met Office’s high resolution UK model, supplied through Open-Meteo: temperature and humidity at many heights through the lower atmosphere, wind, cloud at low, mid and high levels, and surface pressure. Capturing around the clock means the forecast is built from the newest model run, not this morning’s leftovers.
Step two
Four questions, one score
For every night and dawn hour, each watch point’s profile is scored on the four things a sea of cloud actually needs.
- Is there a lid? A layer of warmer air sitting above colder valley air, and how strong that temperature flip is
- Is it calm enough to keep it? Wind mixes the layers and destroys an inversion, so still air scores well and gusty air scores badly
- Is there cloud to trap? Moist air low down can become the sea of cloud, while bone dry valleys give a clear sky inversion with nothing to photograph
- Is the weather settled? Steady, high pressure holds the whole arrangement in place; a changing pattern breaks it up
Only night and dawn hours are scored, because that is when inversions form and hold. The result becomes the bands you see on the map and in alerts: Possible, Likely and Very likely.
Step three
From watch points to your place
Search any UK location and WeatherDuck blends the nearest captured profiles, weighted by distance, into a local outlook. Regional signals then look for agreement: one foggy basin is a curiosity, while a dozen neighbouring valleys all scoring high at dawn is a forecast worth setting an alarm for.
Step four
Checked against the real sky
A forecast that never faces the morning after is just marketing. WeatherDuck saves its headline forecasts before the event, then compares them with what actually happened: photographs, webcams, satellite imagery and reports from people who were on the hill.
- Pre-event forecast reports are stored before dawn, so a hit can be proven and a miss cannot be quietly forgotten
- A growing archive of replayed historical events tests every formula change against mornings we already know the answer to
- Public write-ups in the forecast verification hub cover hits and misses alike, including events forecast more than a day ahead
The honest bit
Where it can go wrong
No inversion forecast is certain, including ours. These are the failure modes we know about and work on.
- Very shallow fog layers can slip under the model’s eye, so some marginal mornings score lower than they deserve
- Confidence grows as the night approaches; a promising signal two or three days out can fade as newer model runs arrive
- Tight, steep sided glens can behave differently from the nearest watch points that represent them
- The cloud top height is an estimate; a viewpoint close to the estimated top can be in the murk rather than above it
The practical advice never changes: treat the band as a probability rather than a promise, recheck the forecast the evening before, and always read a dedicated mountain weather forecast before committing to a route. More on that in our safety information.
Put it to work
See tonight’s signal
The cloud inversion forecast and predictor shows the next 24 hours for any UK location, free and without an account. If you want WeatherDuck to watch a mountain for you, a free account adds one email alert for the place you choose.