The GFS Eyes Superstorm, Part 2

Elliot Abrams was kind enough to speak a little bit in his blog Tuesday (PREMIUM | PRO) about Forecast Model [JessePedia] ensembles and Spaghetti Plots yesterday, so I'll follow up with that today and show you what the GFS Ensembles think about Monday's storm, which may or may not spread heavy snow into the New England coast.

As a reminder, "ensembles" means that one forecast model is run with slightly different conditions. This produces up to a dozen or two different forecasts, or "perturbations" from the same model. Our Pro site has several different ensemble tools that can give you some insight into how confident the model is of it's own forecast.

First, we show the spaghetti plot of selected isobars for Monday's storm:

So what this map is saying is, show me the center of the low pressure on Monday evening from each perturbation. You can tell by the differences here that the GFS is not confident on the placement of the low pressure system, which will make all the difference for the New England coast. Specifically, the light green ensemble "run" thinks the storm will be right off the coast of Massachusetts, but the light blue run puts it off the Canadian Maritime coast, which means either it's moving quicker than the other run, or it just formed there (you can tell which from a loop).

Another way of viewing the ensemble information is to show the "real" GFS forecast, along with what I'll call the "forecast uncertainty" -- the range of differences amongst the ensemble members.

In this case the actual GFS forecast is shown in black; a decent low pressure system is forming Monday night, but it's pretty far off the coast. The colors look at each point on the map and ask: What ranges of pressure do the different runs present? A higher number (orange and red) means that the runs are further apart, and hence forecast confidence is low. As we've already seen, this is the case all around our low pressure system, which means we'll just have to wait and see on this one. On the other hand, if I was looking to forecast the pressure in South Carolina from the GFS, I could be confident of the forecast because the ensembles pretty much agree.

Let me know if this doesn't make sense. It took me a while to comprehend the ensembles when they started making their way into modern meteorology (they didn't even have the GFS when I was in school :))