A Look at Lake Effect Snow

Cold air advection in the wake of yesterday's cyclone passage across Lake Erie is causing a very well-defined, wide band of lake effect snow south of Buffalo this afternoon:

Fig 1 -- 0.5 degree base reflectivity image from KBUF at 2113Z, Dec 1, 2010

Heavy snows are being forecast for the area as per the latest Lake Effect Snow Warning for the Buffalo area:

SNOW ACCUMULATIONS: 14 TO 20 INCHES IN THE MOST PERSISTENT
  BANDS. SNOWFALL RATES OF 1 TO 2 INCHES PER HOUR LIKELY.

So what goes into a lake effect snow event?  What conditions have to be met to see extraordinary snow bands like this one?

There are at least five main qualifications that usually must be met to get a good lake effect snow event.  This list is adapted from a presentation in Dr. Fred Carr's mesoscale meteorology class at the University of Oklahoma (though I believe the actual presentation was given by Todd Kluber).  But here are the five:

1. Lake-air temperature difference --> Generally, the lake surface water temperature should be at least 13 degrees Celsius warmer than the 850mb temperature.
2. Height of the capping inversion --> To get any real snow at all, the capping inversion needs to be at least 1.9 km above the ground.  For heavier snows, the capping inversion should be more than 2.3 km above the ground.
3. Wind direction over lake fetch length --> The winds must be blowing in a direction such that the length of open water over which they pass is at least 80 km for light flurries and at least 160 km for vigorous banded snow.
4. Vertical wind shear --> The best banding occurs with little to no directional shear of the wind underneath the capping inversion.  If there is more than 30 degrees of directional shear, the snow band location becomes difficult to pin down and with more than 60 degrees of directional shear bands are unlikely.
5. Lake-land temperature difference --> In some cases, "lake breeze" convergence zones can set up if you get offshore (or onshore) flow in association with a lake-land temperature difference (like the sea breeze near the ocean).  This convergence can enhance bands of lake effect snow.

So...let's see if these conditions are being met in our Buffalo case today.

First, the lake surface water temperature  should be at least 13 degrees Celsius warmer than the 850mb temperature.  This should be easy enough to check.  NOAA's Great Lakes CoastWatch division publishes daily estimates of lake surface water temperature.  Here's yesterday's image (today's isn't out yet, but lake temperatures don't change as fast as air temperatures...):

Fig 2 -- Analysis of Great Lake surface water temperature for Nov 30, 2010.  From NOAA CoastWatch and the Great Lakes Environmental Research Laboratory.

Now, I am not a fan of their color scale on here at all.  How are you supposed to tell the difference between those different kinds of green on that color bar? But after careful, close-up analysis I concluded that those dark and light greens in Lake Erie (the lake west of Buffalo) average out to a lake temperature of about 10 degrees Celsius.  Well above freezing.  So now we need our 850 mb temperature. We could try using this morning's 12Z sounding from Buffalo:

Fig 3 -- 12 Z sounding from Buffalo, NY on Dec 1, 2010.  From the HOOT website.

Well. If we grab our 850mb temperature from this sounding we get right about 0 degrees Celsius.  That's a lake-850mb temperature difference of only 10 degrees Celsius.  Rather large, but not the 13 degrees Celsius we wanted it to be.  Of course, this sounding was taken much earlier this morning.  Note that the inversion (and a corresponding wind shift) is also right around 850mb.  The wind direction at this level shifts from westerly to southerly with height--a backing of the winds.  If we went back to those thermal wind arguments, we might suspect that there was cold air advection going on in that narrow layer.  So perhaps our 850mb temperature has since cooled a bit. Let's see what a model sounding for this afternoon says:

Fig 4 -- RUC 21Z Analysis sounding for Buffalo on Dec 1, 2010.  From the twisterdata.com page.

This forecast sounding is actually based on the analysis for the 21Z RUC model run, so it's probably the best we are going to get for a sounding at the time the above radar snapshot was taken.  Sure enough, our 850mb temperature is forecast to have cooled to around -7 degrees Celsius.  This would make our lake-850mb temperature difference 17 degrees Celsius.  So our first condition most likely checks out.

Next, we need the height of the capping inversion to be at least 1.9 km above the ground and over 2.3 km above the ground for very heavy snow.  Fortunately in the forecast sounding in figure 4, the capping inversion starts at right around 700mb and on the left they've conveniently labeled the height of the 700mb level at 2838 meters.  That's about 2.8 km, which is greater than 2.3 km--enough for heavy snow.  Since lake effect snow is essentially convection driven by cold air over a much warmer lake, like all convection in ther atmosphere it is going to be inhibited by any inversions aloft (unless the lake is much, much, much warmer than the air).  As such, the capping inversion essentially represents the maximum height the lake effect clouds will be able to reach.  Deeper clouds tend to produce more snowfall.  This is why we need such a high capping inversion for heavy snow.

For our third condition, we need to have winds beneath the capping inversion blowing across at least 160km of open lake for heavy snow.  Well, according to the forecast sounding above, the winds beneath the capping inversion are generally out of the west southwest.  So, let's draw a line from just south of Buffalo (the middle of the snow band) to the west-southwest (actually, if you go into the numerical values for the sounding, it's more west by south, or around 250 degrees) and see how much open water that covers.

Fig 5 -- Fetch length from a point south of Buffalo west by south across Lake Erie.  The fetch distance is about 174 kilometers.  Image from Google Earth.

We see in this direction that the winds are travelling over 174 kilometers of open water--more than the 160 km we needed.  The longer the air spends over the lake, the more water vapor it can accumulate and consequently the more snow can fall once it reaches land.  The band of snow as seen on the radar above stretches well out over the lake, too, so we know we're actually reaching saturation well before we even hit land.  Also note in the image above that if the winds become even slightly more westerly, the amount of open water the air will be traveling over will increase dramatically.  Therefore wind direction (and lake geometry) can play a huge role in determining how much snow we'll get.


For the fourth condition, we want as little wind directional wind shear as possible beneath our capping inversion.  Once again, in our forecast sounding above we see almost no directional wind shear beneath the capping inversion--all the winds are generally out of the west-southwest.  Strong directional wind shear would limit the ability for convection to occur over the long fetch of the lake as winds at different levels would be advected over different lengths of warm water.  This could create spurious inversions and other inhomogeneities that would disrupted the banded structure of the snow and perhaps inhibit it all together.  To get a strong band of snow, you need uniform wind directions through the cloud layer and we have that.

The fifth condition is a bit more tricky to apply, as it only describes enhancement of precipitation due to lake breeze convergence.  It's possible to look into this more, but we've already more than satisfied our other lake effect snow conditions.  So, there may be some enhancement due to convergence of winds over the lake, but it's very difficult to quantify that.

And there you have it.  Lake effect snow in Buffalo that should be and is happening, with all of our forecasting rule-of-thumb guidelines met.  Note how a lot of this analysis was simply based on looking at a sounding (or rather, a forecast sounding) over Buffalo--and that's it.  This is why the forecasters at the National Weather Service--Buffalo forecast office developed the now widely-used forecast sounding analysis tool called BufKit.  It was originally designed to predict lake effect snow, but has now been expanded to all sorts of uses.  I encourage you to look into the software if you want to experiment with some fun sounding analysis tools.  And best of all, it's free...