Twice a day, we get three different models at the same time. At noon today, the Canadian and European models looked juicy for the end of the week: snow and cold for the Birkie. The American model? Not so great, again. It is cold for Saturday (and Friday, for that matter), but suppresses the low far south, far enough that the Birkie would see little more than flurries.

The latest from the trail makes me wonder if there’s much of a chance of keeping the snow through the race on much of the trail: two inches has already been lost with four warm days to go. So I’ve upped the cancelation probability to 37% (there’s a formula for this, below), but there’s still a decent chance of having a race.

How do I calculate these guesses? First, I estimate the chance that snow will come and save us. That is calculated by taking the proportion of the models giving us enough snow by Friday evening for a race and dividing it by the model pessimism adjustment rate (generally, 2). So if two of three models give us snow, I give a 33% chance that the snow will save us. Then I take the probability that the whole course will have snow and that just the north end will. Add it all together, subtract from 100, and get a cancelation percent. My guesses right now:

• Probability of snow: 33%
• Probability of part of the trail surviving: 30%
• Probability of the whole trail surviving: 15%
• Probability of cancelation: 37%