Why snow forecasts disagree, and how to read the spread

Snow forecasts disagree because each weather model starts from slightly different data, divides the atmosphere and the terrain into grids of different sizes, and makes different assumptions about timing, temperature and how much snow each inch of water produces. In the mountains those small differences grow quickly. The useful question is not which number is right, but how far apart the forecasts are and how far ahead you are looking.

Where the differences come from

• Starting conditions. Every forecast begins from an estimate of the current atmosphere, and no two models start from exactly the same picture. Small differences grow over time.
• Grid size and terrain. A coarse model smooths a mountain range into a gentle bump, which shifts where the snow falls. Finer models resolve ridges and valleys better but cover shorter time ranges.
• Storm timing. A storm arriving six hours later can move snow from one day into the next, so two forecasts can agree on a storm and still disagree on any single day.
• Snow level and temperature. A degree or two decides whether precipitation falls as snow or rain at a resort's base.
• Snow-to-liquid ratio. Models convert water into snow depth in different ways. A cold storm can produce twice the snow of a warm storm with the same water; see snow-to-liquid ratio explained.
• Different questions. One source may report a calendar day, another the 24 hours from now, another overnight. Those totals are not directly comparable.

How SkiBuds compares sources

The SkiBuds app pulls forecasts from several providers, including Open-Meteo and its ensemble, the US National Weather Service, MET Norway, WeatherAPI, Visual Crossing and the high-resolution HRRR model. Before showing them side by side it drops any source whose data is out of date or returned an error. It only compares snowfall when the sources describe the same local date over the same time window, so a calendar-day total is never set against a rolling 24-hour one.

What ensemble agreement means

An ensemble runs the same model many times with slightly different starting conditions. If most runs agree, the forecast is less sensitive to small errors. If they scatter, the outcome is genuinely uncertain.

SkiBuds turns this into a simple count: how many ensemble members forecast at least your powder-alert threshold of snow in the next 24 hours, shown as a number of members out of the total. When at least 60% of members clear the threshold, the indicator is highlighted as strong agreement. A low count does not mean no snow; it means a big day is less likely than a smaller one.

How the confidence rating works

The app gives each forecast day a confidence score from 0 to 100, built from two things.

How far ahead the day is. Forecast skill fades with time, so each day starts from a lower ceiling:

Day	Starting confidence
Today	100%
Tomorrow	92%
Day 3	82%
Day 4	70%
Day 5	58%
Day 6	48%
Day 7	38%
Day 8	30%

How far apart the forecasts are. The app measures the middle half of the forecast range for snowfall, temperature and wind and lowers the score as that spread widens. Snowfall counts most (half of the penalty), then temperature (30%) and wind (20%). Having more sources adds a small bonus.

The score maps to a plain label:

Score	Label	What it suggests
85 and above	Very high	Models strongly agree.
70 to 84	High	Minor differences between models.
50 to 69	Moderate	Plan for a range of outcomes.
30 to 49	Low	Snowfall could vary a lot.
Below 30	Very low	Extended range; general planning only.

Days scoring 60 or more count toward the forecast's "reliable window", such as "first 3 of 7 days are reliable".

The rating describes agreement and lead time. It is not a calibrated probability that a particular amount of snow will fall.

How to plan around disagreement

• Commit late, plan early. Use the outlook a week out to pencil in a day, then confirm with the forecast one or two days ahead, when confidence is usually highest.
• Plan for the range, not the headline. If sources range from 4 to 14 inches, agree on a plan that works at 4 and gets better at 14.
• Watch the trend. A forecast that rises run after run is more convincing than a single big number.
• Check observations. Once snow starts, observed snowfall and the resort's report beat any forecast; see snow depth vs snowfall vs SWE.

How to read a ski resort snow forecast covers elevation and timing, and planning a powder day with friends helps a group agree on when to commit. Every mountain in the resort directory has its own forecast page on this website. The source comparison, ensemble agreement and confidence rating described here are part of the SkiBuds app, which is being rebuilt for iOS and Android and is coming soon.