Mountain forecasts are issued in hours rather than days, and plans are built around narrow windows. The terrain itself is generating much of the weather being forecast.
Air forced upward produces cloud locally
When wind meets a mountain range it has nowhere to go but up, and rising air cools and releases its moisture as cloud and precipitation.
This happens regardless of what the wider weather pattern is doing, which is why a valley can be clear while the ridge above it is in cloud all day.
The effect is strongest on the windward side, and the leeward side may stay dry, producing entirely different conditions on two sides of the same summit.
Daytime heating drives a reliable daily cycle
Slopes warm faster than the free air at the same altitude, and the warm air rises along them through the morning.
By early afternoon that convection frequently builds into towering cloud and thunderstorms, which is why so many mountain days are planned to finish before the middle of the afternoon.
The cycle is predictable enough that it is treated as a planning assumption rather than a forecast, particularly in summer.
Wind behaves differently at altitude
Ridges and cols accelerate wind by squeezing the airflow, so a col can be far windier than either the summit above or the valley below.
Wind determines what is possible more often than precipitation does, because it affects balance on exposed ground and drives heat loss sharply.
Forecast wind speeds are usually quoted for a stated altitude, so reading which altitude is being described matters as much as reading the number.
Models resolve terrain poorly
Weather models divide the atmosphere into cells, and a cell may be wider than an entire valley, so the model sees a smoothed version of the mountains.
Specialist mountain forecasts add local interpretation to model output, which is why they differ from a general forecast for the nearest town.
The practical consequence is that forecasts become substantially more reliable within about a day, and much less so beyond two or three.
How windows are actually used
Parties plan for the earliest feasible start, because the morning is the part of the day the models predict best and the terrain disturbs least.
Flexible itineraries with alternative objectives allow a party to use a window wherever it appears rather than waiting for one on a fixed route.
Refuge wardens and local guides read the same forecasts alongside what they can see, and their reading of a marginal day is worth more than the raw output.