Saturday, February 5, 2011

Temperature forecasting over the snow cover

How high will the temperature get across OK over the snow pack?
The sounding above at OUN was taken at the morning low according to the Norman mesonet site. Now it snowed another inch or two yesterday  to make the compacted snow depth anywhere from 3 to 5 inches. The snow we had was an icy consistency. I walked on it and only in the drifts was it actually loose snow.

The forecast high for today was 36 (as of this writing was upped to 38) from the NWS, and I thought that was pretty good. But it is always difficult to predict the high temperature when the ground is snow covered. The high for the day in these situations depends how far from the bare ground you are relative to the low level air mass that is moving in. The 1845 UTC visible satellite image shows where the snow is. The whiter the ground the deeper the snow:

I can't mark where Norman is but trust that we are not very far from the edge of the snowpack and points to the west of us didn't get that much snow in the first storm and barely received any in yesterdays.

However, a forecasting technique I learned at UAlbany was the 850 hPa method. This method assumes that air from above the surface will serve as the upper limit as the boundary layer grows upward.  Of course the atmosphere aloft is not usually stagnant so you need to account for that. This mornings OUN sounding had an 850 T of 1.6 C with even warmer temperatures of 6.4C at 911 hPa.  This suggests a temperature of 14.3 C from 850, or a temperature of 13.6 C from 911 hPa ... again assuming these temperatures change very little during the day. So the mid to upper 50's!

So the forecast problem is the battle between surface effects: how the incoming radiation can heat the ground and help that mixing process along versus how the mixing occurring elsewhere (in this case to the northwest and west) brings in warmer air over the snowpack. The latter is usually referred to as an internal boundary layer.

Now, the temperature is obviously controlled by the snow depth and snow cover and there should be an obvious temperature contrastor gradient across the plains:


Note, the yellow shading which tells us the potential temperature gradient pretty much aligned along the edge of the snow pack from this morning.
I will add the 18UTC and 21UTC maps when available below:
The potential temperature gradient shifted somewhat east as the snow below has melted and temperatures have warmed.

So far at 4pm local time the temperatures in western OK have risen to 62+F and Norman is currently around 45F! This is pretty much going to be the high for the day.
 I will add the mesonets max T plot when it is available:


Now where did this warm air layer aloft come from? Downslope off the mountains as the large scale trough passed through our region. Temperatures at 850 hPa warmed from -15C to 2C in 24 hours at Dodge City, and from -18C to 2C at Amarillo.

A comparison of soundings from Lamont indicated a slow warmup primarily in the boundary layer with just a minor change in the height of warm layer. However by 00 UTC it was clear that the entire column was warming significantly downstream from the mountains. LMN and OUN  were clearly situated in a warm advection regime at 00 UTC. Given the sounding structures, you could have used the 850 method with forecast soundings, not upstream soundings given how much the entire temperature profile changed.

Of course, it is very interesting that the sounding at AMA can not be used to show the large surface warming over western OK. AMA only reached 12 C which is around 50F, while western OK reached 62F. The implication is that the warm pocket over Oklahoma this morning was bigger (or may have even developed aloft) than could be revealed by the sounding network.  Either way these mesoscale details (snow cover or lack thereof, warm pockets of air aloft from downsloping) made it difficult for forecasters to utilize observations for short term forecasts of temperature. This is a much more complex scenario for what seems like a straightforward case of temperature forecasting over snowpack.


As a side note, the temperature warmed above freezing and so all the snow in the rain gauge is melting. Now a ton of it was blown around but what was secured to the gauge from the last two snowfalls should tell us something about the sleet that fell, then the snow, and then yesterdays heavier snow. The liquid total from the melt is: as of 9:40pm is 0.32".