The "Arctic Outbreak"

Nearly every weather blog (and forecast discussion) that I've been reading as of late has been talking about this greatly hyped "arctic outbreak" that is supposed to occur in the latter half of next week.  I thought I'd add a few of my own perspectives on this discussion.  Let's start by taking a look at current 500 mb heights across the entire polar region.

Fig 1 -- 500 mb heights (shaded) with MSLP contours for the entire Northern Hemisphere at 12Z, Nov 11 2010. From the HOOT site.

 Note the north pole is in the middle of this image. We're going to focus on the shading here which represents 500 mb heights.  We can use 500 mb heights as a rough proxy for temperature (though it would be better if we were using thickness...more on that later).  How can we do this?  This somewhat gets back to the reasoning behind my very first blog post.  To put it simply, imagine a column of air stretching between the ground and the 500 mb pressure level.  When we warm air, it expands--simple physics (or chemistry, if you like).  So if we warm that column of air, it expands.  Since the ground is fixed and cannot move, the only way for the column to expand is in the vertical direction, which is going to lift the 500 mb height higher as the column expands.  Therefore warmer air tends to be found under higher heights and colder air tends to be found under lower heights (cooling air makes it contract, pulling down a pressure surface at the top of the column.)  Thus in the image above, higher heights (conveniently represented by the warmer colors) can be thought of as areas of generally warmer air and lower heights (the cooler colors) can be thought of as areas of generally cooler air.

 As noted by Patrick Marsh in a blog post a few days ago, often when we see large intrusions of warm air into the polar regions, the response after a few days is typically for a large airmass of cold air to move southward somewhere. Last week there was a large plume of tropical air evident over the central Atlantic stretching well north to near Iceland.  This week, as seen in figure 1 above, there's a decently large tongue of warm air that is intruding across the eastern US and Canada as far north as northern Baffin Island.  With these surges of warm air north, it makes sense that we would be expecting a surge of colder air south.

But what colder air?  The really deep, cold arctic air is going to be found where we have the dark blues and purples in the above image.  It's clear that our coldest areas are directly over the north pole and in a trough near the British Isles north of Europe.  There's cool air, but nothing frigidly cold on the North American side of the arctic.  For the US to see an arctic outbreak, we would want to see a really, really cold air mass building on our side of the globe.  Much colder than anything we see there now.  Why so cold to begin with?  We have to remember that as an arctic airmass is dislodged and moves southward, it is constantly being modified.  It's moving over land that has been warmed by the sun, the air itself is being heated by the sun every day and many other things contribute to warming the air mass as it moves south.  For it to still be really cold when it gets here, it had to be REALLY REALLY cold to begin with.  And we're not really seeing that yet...

Of course, we're still a week away.  So, we'll keep watching the arctic to see if such a cold air mass does indeed build on our side of the arctic.  So what's everyone getting so hyped about?  Models.  Long-range models.  Which are not the most reliable things to look at.  However, they are pretty much the only thing to look at.  Below are the 850mb temperatures for a 168-hour forecast from this morning's European (ECMWF) and GFS models.  This would be a forecast for 12 Z next Thursday, or sometime early that morning.

Fig 2 -- ECMWF (above) and GFS (below) 850 mb height and temperature 168-hour forecasts for 12Z, Nov. 18, 2010. From the HOOT site. 

Note that in these images, the color shading represents the temperatures explicitly, not the heights (which are contoured).  Even at this rather low level, you can still see how the wavy pattern separating the warm and cold air just about follows the wavy pattern of the height contours.  It gives us better faith in our 500 mb analysis above.  Anyhow, at 850mb in both models for next Thursday you see temperatures in the -5 to -10 degree Celsius range across much of the upper midwest.  This represents temperatures that are several degrees below freezing in the very low levels of the atmosphere.  Also note the strong northwesterly to northerly winds in the cold air on the western side of the trough.  With even colder air to the north, this shows strong cold air advection, indicating that by Friday it will probably get even colder (it does in the GFS model, but I didn't have an ECMWF image from them to show for comparison).  If you remember the relationship between cooling air and changing heights as mentioned above, you might also suspect that with all that cold air advecting into the base of the trough, the heights there would fall (cooling air contracts) and the trough would be getting deeper.  We'd have to check multiple levels to confirm this, but it's a pretty good bet this is a strengthening trough.  (I note this is only a model forecast...).

So we do indeed see an impinging mass of colder air.  What does this mean for surface temperatures (what we actually feel)? Here is a GFS model forecast for those at the same time:

Fig 3--GFS model 168-hour forecast surface temperatures for 12Z, Nov. 18, 2010. From the HOOT site.

 The color shading for temperatures here is in Fahrenheit.  Often times, the 12Z temperatures are used as a good guess for what the nightly low temperature is going to be, so this represents about the coldest temperatures next Thursday morning.   We can see that a broad area of the upper midwest is forecast to drop into the teens that morning.  That's a big contrast from the current lows in the mid 40s that a lot of the upper midwest is enjoying.  And if that cold air advection is continuing (note the winds at the surface are also bringing down colder air), it will probably get colder...

One more thing to look at quickly.  This is a contour plot of the GFS forecast 1000-500mb thickness (which is technically what we should be using as a proxy for temperature instead of just heights) for that same 168-hour forecast.

Fig 4--GFS 168-hour forecast of 1000-500mb thickness for 12Z, Nov. 18, 2010. From the HOOT site.

The thickness between two pressure levels is simply the height above ground of one level minus the height above ground of the other level, or how thick (in meters) the layer between those two pressure levels happens to be.  It's pretty straightforward.  Because both levels are pressure levels which can move up and down (unlike the ground, which is fixed), it turns out that the thickness is an even better proxy for mean temperature in the layer between the two pressure levels than raw height would be.  Since 1000 mb is nearly at the surface (or often below the surface...which makes for an interesting story that I'll explain in another blog), the 1000-500mb thickness is a good indication of the mean temperatures in the lowest half of the atmosphere.

So what is this map above telling us?  Thickness is often used to help decide which type of precipitation is going to fall.  Is it going to be rain?  Snow?  Something frozen in between?  The colder the lowest half of the atmosphere is, the more likely there is going to be snow falling instead of rain--that makes sense.  Since colder air implies a lower thickness (cold air contracts), we'd want to look for lower thickness values to find areas where snow is likely.  It turns out, from lots and lots of analytical experience, that a good dividing line between rain and snow is usually around the 5400 m 1000-500mb thickness line (conveniently (or rather intentionally) contoured as the thick blue line on the map above).   Areas to the north of that line (the blue dashed contours) have a lower thickness and areas to the south (the red dashed contours) have a deeper thickness.

Therefore, areas with thicknesses less than 5400 m (in this case anywhere north of the 5400 m contour) could expect to have an atmosphere cold enough to support snow.  This means, in figure 4, that anywhere north of the 5400 m line could see some snow if any precipitation falls with this system.  This could mean the first real snowfall for much of the upper midwest.

So, there you go.  In summary, IF frigid arctic air can build on our side of the globe and be cold enough to survive modification on its trip down here, then MAYBE our really long range models will verify and we'll see some colder air move in to the central US.  Our models don't have the best track record in forecasts at this long of a range, so they shouldn't be interpreted literally.  But in terms of general trends, it does give you something to think about...