How Will Mets Predict Snow?

Another winter storm looks likely for the mid-Atlantic states as we go into the latter part of the week. Here's our latest projections:

You might say:

"Hey, that looks different than Carl's map from Tuesday!"or"How come my local AccuWeather.com forecast's snow accumulation keeps changing?"

There are several reasons. I hope this explanation will make you grit your teeth a little less when meteorologists adjust their snow forecasts.

1.) The Naked Truth: Models Change. Forecast models change their minds every hour of the day, forcing meteorologists to reconsider their forecasts.

Take a look at the NAM (formerly known as the ETA) forecasts for Friday morning. Below is shown the position and strength of the low-pressure system from this morning's forecast (06Z), last night's forecast (00Z), and yesterday afternoon's forecast (18Z).

Yesterday afternoon, the NAM believed that the storm would be well off the coast but would be pretty strong (pressure of 1000 mb). Last night, the model changed its mind and reported the storm would be slightly weaker (slightly higher pressure) and would be located over Connecticut. But then this morning, it adjusted again and now says the storm will be over Boston. Now imagine 25 models all jumping around every hour while meteorologists are trying to use their expertise to decide which model is right in this particular weather situation. And this is only the low pressure center -- figuring out the precipitation amounts is even harder!

2. It's The Ratio, Dummy. Take a look at the map below from the NAM for Friday morning.

It shows heavy snow over the Appalachians, rain in Tennessee and the Piedmont of North Carolina, and ice from northern Virginia through southwest North Carolina. Sounds pretty cut and dry right? Well, the problem is, we don't know how much snow will fall. Although the model can predict snowfall amounts, they are sometime inaccurate and the big question remains: What will the snow ratio be? Normally one inch of rain would be equivalent to ten inches of snow, but this varies greatly depending on temperature, elevation, and location.

3. Snow Is Evil. Here's one more wrench to throw in the process. The average person probably wouldn't know the difference between half an inch of rain and an inch. They wouldn't care either. But 5 inches of snow versus 10 (see #2) is a big deal. That's why snow amounts in a forecast might seem to vary so much. If you're on the edge of the storm, even a slight change in the low pressure's center could mean the difference between zero and five inches. Imagine you're just inside the line for 5 inches of snow. If it's a tightly packed storm, even an adjustment of 10 miles could land you in the "dusting" category. If "you" means 18 million people in New York City, some meteorologist is in trouble!

So how can meteorologists overcome these forecast shifts?

Researchers are working on several solutions, and ensemble forecasts are one. With ensemble forecasting, the same model is run many different times with slightly different initial conditions. Those conditions can trigger "the butterfly effect" with rapidly different results if you go out too far in the forecast, but in general, an average of those ensemble "members" will result in a more accurate, less-shifty forecast. Our Pro site has Ensemble forecast maps from the GFS, SREF and NAM models, as well as tools to detect model trends (see below).

Another thing that helps is looking at the model trends. Has one model been increasing the strength of the storm every time a new forecast comes out? Are all the models moving the storm northward?

These are some of the questions that meteorologists will be struggling with as we move into this latest winter tempest.

Report a Typo