All the resolution and sophisticated physics in the world will not produce good forecasts unless the forcing for the event is forecast well in time and space. The following table provides links to animations of the model-provided forcing for the 00 UTC 24 December 2001 forecast. Each link will appear in a new browser window; click on the upper right hand "X" (Windows) or box (UNIX) to close. When you are finished viewing each of the graphics, read the discussion of each model forcing mechanism below.
Model forcing mechanisms for lake effect snow |
|
|---|---|
 |
First layer wind speed and the total surface sensible and latent heat fluxes |
|
Lake surface to first model layer lapse rate (°C), lowest model layer winds (kts), 3-hour accumulated precipitation (in.) |
In the pre-storm scenario, we saw that the LST values for Lake Erie were at least 5°C above normal the day that this lake effect snow event began. At forecast initiation time, the temperature decrease from the lake surface to the lowest model layer ranged from less than 1°C at the east end to about 5-6°C over the entire western half of the lake. By six hours into the forecast, the large lapse rates have spread across the length of the lake. By nine hours, the values in the western part of Lake Erie have increased to greater than 8°C, and downwind from this area, the lake effect precipitation begins along the northeastern lake shore and downwind. By 15 hours (panel 6), 850 hPa temperatures have decreased also to less than -10°C (not shown), and a single lake effect snow band has become well-developed, as indicated by the precipitation structure.
The surface fluxes three hours into the forecast resulting from the temperature and moisture gradients at the surface, and the steady southwesterly winds which increase as they move downwind along the lake, are already in excess of 250 W-m-2, and as high as 400 W-m-2 in the west-central part of the lake. Values fluctuate between 250-500 W-m-2, comparable to the surface fluxes into the atmosphere averaged over a summer day. At 60-72 hours, the maximum surface fluxes over Lake Erie approach 600 W-m-2!
From the radar and surface observations, we can determine that this lake effect snow event started at about 8:00-9:00 UTC on 24 December. To compare the model versus actual start times, we show the 9:05 UTC radar compared to the 9:00 UTC Eta-12 forecast for 907.2-hPa cloud ice and winds. See these graphics by clicking on the "Display Graphics" button below.
09:05 UTC 24 December KBUF radar image and 00 UTC 24 December forecast
valid at 9:00 UTC 24 December
Note that there is one band visible in the radar's clear air range, crossing the northern shore of Lake Erie and the southeastern portion of the Niagara Peninsula, entering the US at Niagara Falls, and then exiting into Lake Ontario north-northeast of Buffalo. In the other graphic, the winds and cloud ice give a "snapshot" of the developing lake effect snow band. We note that the forecast matches up quite well to the radar, even down to the the location of a relative maximum in cloud ice (matched with a radar reflectivity maximum) between Long Point and Buffalo over Lake Erie. Eastward, the band crosses into the US over Niagara Falls in both the forecast and the radar reflectivities, perhaps a gridpoint or two south and east of the radar verification, but with both indicating a secondary maximum in snow band activity. The six-hour forecast shows just the beginnings of lake effect snow band development. No precipitation was forecast before the three-hour period ending 09:00 UTC by the Eta-12, so we can claim that Eta-12 predicted the event initiation and the initial snow band position quite well.
Single lake effect snow bands are very sensitive both to the vertical wind shear and the prevailing wind direction in the lowest 3-km of the atmosphere. An additional test of the Eta-12 involves seeing if it can capture the timing of actual shifts and/or redevelopment of the lake effect snow band as the forecast evolves. To see this evolution in the model forecast, click on the "Display Graphics" button below for an animation of the Eta-12 forecast.
Eta-12 0-36hr fcst. of 850-hPa omega, sfc. layer vs. obs. winds, 3-hr
accum. precip.
The first two panels show the passage of the cold front and its accompanying precipitation, with surface winds shifting from the south to the southwest. Some enhanced precipitation and vertical motion is seen in the second panel, resulting from interaction of the wind field immediately behind the cold front with the topography at the east end of Lake Erie. The six-hour forecast (panel 3) shows the beginnings of the lake effect set up, with weak upward motion over the northwestern part of Lake Erie.
By nine hours, we see the beginnings of the lake effect snow band developing on the northern end of the lake. Surface layer winds are becoming weakly convergent over the lake's northern half. Accumulated 3-hour precipitation is forecasted in the 0.01-0.03" range, on Lake Erie's northeast shore. The pattern is spreading into Niagara County NY, north of Buffalo. Over the next two 3-hour periods, vertical motion, convergence, and precipitation amounts increase, as the band shifts somewhat northward and strengthens further. This corresponds to the heavy lake effect snows which occurred on the Niagara Peninsula in Canada, and Niagara and northern Erie Counties in NY, on the morning of 24 December.
At 18 hours (panel 7), while the accumulated precipitaton amounts are relatively large (>0.07"), the vertical motions are weakening, as the winds shift to a more westerly direction and the single snow band adjusts to the new wind direction. In panels 8-13 (covering 21:00 UTC 24 December through 12:00 UTC 25 December), the lake effect band reorients itself parallel to the south shore of the lake, drifting southward through Buffalo and then back northward. This period produced heavy lake effect snow in Buffalo, as discussed previously. Up to this point, we have strong qualitative agreement in strength and location of the single lake effect band.
Observational studies show that single band lake effect snow events are characterized by specific mesoscale features related to heating from below by the lake, and latent heating from convective processes and ice microphysics. These include:
Each of these points will be illustrated with one or more graphics, to show that the Eta-12 captures some elements of these mesoscale features.
First, the warm core nature of the lake effect simulated in the Eta-12 can be seen by clicking on the "Display Graphic" button below, to reveal a snapshot of 850-hPa temperature and winds at 850-hPa.
850-hPa temp. and winds, Eta-12 24-h fcst. valid 00 UTC 12/25/01
Note the warming as the air passes over the lake and is warmed from below. At the downwind side over Buffalo, NY, the temperature is 3°C higher than on the upwind side. The warm-core nature of the lake-induced circulation can also be seen in the cross-section animations between Bradford PA on the left and Toronto, ON on the right, from 970 to 700 hPa. These can be seen by clicking on the next graphics button.
Cross-section of 00 UTC 12/24/01 Eta-12 fcst. winds, temp., grid-scale condensate, 6-84h
The lake is located at the middle of the cross-section, as can be seen in the cut-out in the upper right portion of the graphics. Even at six hours, the warming from the lake near the surface is clear. There is also a cap of relatively cold air at about 750 hPa over the lake, resulting from the beginnings of the upward motion at the top of what will become the convergent lake circulation. Both of these features will remain as long as the single lake effect band exists.
By nine hours, we see vertical development of a temperature "bulge" over the lake, as latent heating warms the area when cloud and precipitation condensate is being formed. Here, the condensate is all ice, since the cloud-top temperature is about -14°C. The Eta-12 requires temperatures of -10°C or less at the top of a below-freezing cloud to create cloud ice and snow. Note the confluence in the winds as they move across across the lake, with more southerly winds south of the lake meeting southwesterlies where the lake effect band is located, at the north shore of the lake both in the forecast and in the observations, as discussed previously. Concurrently, there is diffluence in the wind field from south to north above 850-hPa. These features are typical for observations taken during single band lake snow events (see Byrd, 1995). We can watch the single band feature meander and fluctuate in strength as we step through the animation, past the first portion of then event up through 36 hours, and then beyond. How accurate the forecast location and strength of the lake effect band is beyond 36 hours will be discussed later.
We can get a better look at the lake-generated circulation by taking the same cross-section, except looking at the vertical motions and divergence. An animation of these fields can be seen by clicking below:
Cross-section of 00 UTC 12/24/01 Eta-12 fcst omega, divergence from 970 to 700 hPa
These cross-sections tell a similar story; by six hours vertical motion centered over the lake with a maximum at 900 hPa vertically straddle a lower convergence/upper divergence couplet. By nine hours, a vertical motion maximum of 0.01 Pa/sec is at 850 hPa, with the accompanying convergence/divergence couplet strengthening to in excess of 1*10-4/sec. Maximum strength over the first 36 hours occurs at 12 UTC on 25 December (24-h forecast), with vertical motions approaching 0.02 Pa/sec and divergence approaching 1.8*10-4/sec. above the lake effect band.
Studies have shown that the surface pressure gradient is typically stronger in single band lake effect events to the right of the band itself. The graphic below shows a snapshot of the mean sea level pressure from 12 hour forecast from 0:00 UTC 24 December 2001 Eta-12 forecast, with the 3-hour precipitation to indicate the location of the snow band.
Note the weak gradient to the left and the strong gradient to the right of the snow band. These features are consistent with the faithful emulation of the lake circulation and the lake effect snow band, and result from the superposition of mesoscale low pressure, induced by the warm lake, onto the general west-southwest to east-northeast isobaric pattern.