Early and late-season snowfalls are notoriously complicated to forecast. Temperatures swing wildly between warm and cold, and precipitation amounts are usually much larger than mid-winter storms. The type of precipitation can change quickly, but it also depends when that transition happens.
Let’s take this current storm in Minnesota and the location with the widest range of possible computer model snow outcomes: Marshall, in southwest Minnesota.
Current model forecasts range from as little as 1 inch by Thursday morning to as much as 10 inches. Right away, most seasoned meteorologists, understanding late-March weather, are going to be very skeptical of 10 inches unless temperatures are already below freezing (and staying subfreezing consistently).
Marshall was 68 degrees on Monday. The ground is warm and it is not frozen any longer. This is an area that has been snow-free for some time. Temperatures through early Thursday will be in the 30s, at or above freezing with the type of precipitation going back and forth from snow to rain.
It is VERY difficult to get significant snow to accumulate in these conditions.
While it’s also overcast, the sun angle is more than twice what it was in December and solar energy does still get through to the ground. This means that snow has a hard time accumulating – except at night – unless it’s very cold.
The essential elements for heavy late-March snowfall in southern Minnesota are:
- Night time, heavy bursts of snow
- Below freezing, preferably for hours ahead of the snow
- Consistency in both temperature and precipitation type
None of these conditions will be met in this storm for southern Minnesota (northeast Minnesota is a different story). That doesn’t mean snow won't fall in southwest Minnesota, it just means that 10 inches is highly unlikely.
Let’s look at 10 different computer model scenarios:
What’s important is to step back and look at all the models, including the extreme ones. We then look at reality, like radar and surface observations.
What’s clear is that all the models are hinting at a broken line of slightly higher snow totals across western Minnesota into southwest Minnesota (when compared to the east). This is the result of cold air on the west side of the storm system.
Let’s look at the extreme model (in this case the NAM model) which is the one in the middle-right of the graphic above. It’s creating some crazy totals. When we look at what it thinks has already happened through 8am compared with the observations we see a big red flag.
It "thinks" 1-2 inches has already fallen when it has not. Already this will lead to an exaggerated conclusion. The models cannot account for how warm the ground is, and the mixing of precipitation and surface temperatures above freezing (even though aloft it is below freezing and cold enough to create snow). These are complicating factors limiting snow already Tuesday morning and will continue to do so during the daytime hours, both Tuesday and Wednesday.
Looking at all the models together we can see that there is somewhat of a consensus, which helps to eliminate the outliers. Nine of the 10 models are predicting anywhere from 1.3 inches to 4 inches in Marshall, while the NAM is hammering Marshall with 10 inches.
We can then apply some statistics to narrow in on a most likely range. While the 10-inch model is 1 of 10, it’s not a 10% chance. Statistically, it’s so far out that it’s an outlier that probably is closer to a 1% chance of becoming reality.
It’s clear without even applying a statistical analysis that most of the models agree on 2-4 inches of snow for the Marshall area.
This example doesn't mean the NAM model is awful and never to be trusted. On the contrary, every model has good and bad times and strengths and weaknesses. It’s just that with this storm, today, the NAM is out to lunch.