P(A) Cold = 15%
P(B) Snow = 40%
P(AuB) = 55% Snow and cold are not mutually exclusive. It can be too cold and snow simultaneously. Remove this overlap
P(AnB) = During the 40% chance of snow, 15% of the time it will be under 20 degrees. Remove 15% of 40% or 6%. (.15 * .4 = .06)
P(AuB) = P(A) + P(B) - (AnB) = 49% chance of cancellation
Although I don't like this problem, it's unrealistic.
If it's already going to snow, if that 40% is going to happen then the 15% chance of cold has already drastically improved its chances.
That 15% cold probability is taking into account whatever the entire forecast is that day lets say the 15% chance is based on reports that it could be anywhere from 15 degrees to 40 degrees. If the 40% snow is going to happen then its automatically under 32 degrees and the chance of cold has shifted from 15% to 48% chance.
You would need to limit the problem by saying its automatically cold enough to snow and that rain isn't an option.
Same thing the other way around if its already the 15% cold hasn't the snow changes improved?
Was the likelihood of the 40% weather snow forecast contingent on it being cold enough (100% chance of snow or rain, X% chance of too cold) or climate weather (x% chance of snow or rain, 100% chance of too cold)