National Weather Service United States Department of Commerce

Bust Analysis of the 20-21 November 2000
Northwest Flow Snow Advisory

Bryan P. McAvoy
NOAA/National Weather Service
Greer, SC

Author's Note: The following report has not been subjected to the scientific peer review process.

1. Introduction

This web page records data form a forecast "northwest flow snow event" which never occurred. On this occasion, I issued a snow advisory for the spine of the western North Carolina Mountains, which encompassed the counties of Graham, Swain, Madison, Mitchell, Yancey and Avery. However, only a dusting of snow was reported in a few locations in the advisory area, mainly from Madison County and points north. Actually, I inherited the advisory, but I chose to extend it south, as the original advisory only included the four northern-most counties listed above. In the past I have had a fair amount of success with northwest flow, mountain snow forecasts. However, many of the previous events shared common features, such as an ample pool of low level moisture and strong 850 mb flow, two elements (among others) lacking in this event.

For this review, I've constructed a series of 4-panels which follow the event in time. I explain what I thought each level represented for snow potential before the event, and what I learned about the importance of these fields as predictors thereafter. Many of these images are "thumbnails". This means you must click on the image for a larger version. The model I used for this bust analysis was the 1200 UTC run of the 22-km Eta model on 20 November. This was the run I used to decide to issue the snow advisory. Most of the these images start 18 hours into the Eta run, or 0600 UTC on 21 November. This was about the time I thought the best snow would develop.

2. Synoptic Features

Working down through the atmosphere, I'll start with the 500 mb level (Fig. 1). I never thought that vorticity advection would be much of a contributor to this event. However, a sharp short wave trough, accompanied by significant positive vorticity advection, is a well known contributor to northwest flow snow, and certainly would have helped. In retrospect, seeing that most of our mountain counties were on the south side of a channeled shear axis, I might have wanted to weigh the 500 mb features a little more in my thinking, as this is quite unfavorable for northwest flow snow (as there is usually strong drying and subsidence aloft). I did not plot relative humidity as 1000-500 mb layer RH's are all but meaningless in these events. Apparently the wave was progged to be further south when the previous shift initially issued the advisory.

18Z-h5.gif (53560 bytes)24hrs-h5.jpg (82323 bytes)

36hrs-h5.jpg (66375 bytes)

Figure 1. Forecast of 500 mb geopotential height (dm; solid green contours) and vorticity (10-5 s-1; dashed blue contours) from the 22-km Eta model initialized at 1200 UTC on 20 November 2000 and valid at 0600 UTC 21 November (upper left), 1200 UTC 21 November (upper right), and 0000 UTC 22 November (lower left).

18hrs-sfc.gif (14191 bytes)24hrs-sfc.gif (14687 bytes)

30hrs-sfc.gif (14859 bytes)36hrs-sfc.gif (14045 bytes)

Figure 2. Surface plots of mean sea level pressure (mb, green contours) and dew point (deg. Celsius; black dashed contours) from the 22-km Eta model initialized at 1200 UTC on 20 November 2000 and valid at 0600 UTC 21 November (upper left), 1200 UTC 21 November (upper right), 1800 UTC 21 November (lower left), and 0000 UTC 22 November (lower right).

Surface charts showed very low dew points upstream from the mountains (Fig. 2). While temperatures aloft were also cold, the low levels are the atmosphere were quite warm and dry, leading to large dew point depressions. This is evident by the satellite data from the event (Fig. 3).

Figure 3. Visible and infrared satellite imagery from GOES-12 at 1515 UTC 20 November (upper left), 1815 UTC 21 November (upper right), 0615 UTC 21 November (lower left), and 1215 UTC 21 November (lower right).

Here is possibly the best indicator of the entire event. The decision to issue a Winter Weather Advisory had actually been made the night before. However, during the afternoon of the 20th, we decided to expand the aerial extent of the advisory by a few counties. Looking at the satellite imagery during the day, I lost track of one of the best rules of thumb that we have. Typically, if there is not a big upstream reservoir of low clouds 6 to 12 hours before an event is forecast to take place, snowfall totals will not be very high. Note the lack of low clouds over most of Tennessee and Kentucky during the day of the 20th (Monday). Monday night was not much better, with just a narrow band of clouds forming at the highest elevations.

The average relative humidity in the 1000-850 mb layer never rose above about 80% (Fig. 4). Despite strong cold advection, the 850 mb winds were only 25 to 30 knots. It appears that the strength of the wind and the projected layer average relative humidity is a much better predictor than low level thermal advection. However, our best snow in these events still needs to occur with 850 mb temps of less than -12 deg C.

18hrs-h8.gif (21375 bytes)24hrs-h8.gif (19661 bytes)

30hrs-h8.gif (19991 bytes)36hrs-h8.gif (23259 bytes)

Figure 4. As in Fig. 2, except for 850 mb geopotential height (dm; solid green contours), 1000-850 mb layer average relative humidity (percent; solid purple contours), 850 mb temperature (deg. Celsius; dashed black contours), and wind (kt; barbs).

Mon-18Z-sfc_medium.jpg (275622 bytes)

Figure 5. Surface plot at 1800 UTC on 20 November 2000. Dew point depression is indicated in blue on the station model plot.

Notice the 25 to 30 degree Fahrenheit dew point depressions behind the surface cold front at 1800 UTC on 20 November (Fig. 5). Another interesting thing is the strong winds over Upstate South Carolina. Winds gusted to 30 knots from the southwest this day. There was even some minor damage -- more than was reported with northwest winds behind the front!

The orographic omega values (or the component of lift in the low levels of the atmosphere owing to terrain effects) were quite respectable for this event, anther thing which had me going (Fig. 6). I figured that the very cold 850 mb temperatures, coupled with the reasonably strong upslope would lead to respectable snowfall at the higher elevations. For reasons previously demonstrated, this was not the case.

18hrs-orvv.gif (19504 bytes)24hrs-orvv.gif (20344 bytes)

30hrs-orvv.gif (19749 bytes)

Figure 6. As in Fig. 1, except for orograhic omega.

A time series of total vertical velocities from 1000 mb to 850 mb is shown in Figure 7. What I find interesting about this series is that the lift seen on the upshear side of the mountains is much stronger in West Virginia. This is from the combination of stronger 850 mb winds and being on the cyclonic side of the 500 mb jet. I may start using this field in lieu of orographic omega fields in the future. As I do not have any other cases posted, I do not yet know if this will work.

18hrs-vvel.gif (19015 bytes)24hrs-vvel.gif (20738 bytes)

30hrs-vvel.gif (19920 bytes)36hrs-vvel.gif (20604 bytes)

Figure 7. As in Fig. 2, except for vertical velocity in the 1000-850 mb layer.

3. Summary

In conclusion, and based on better luck last year, here are a few cook-book type ingredients that should be useful when forecasting a northwest flow snow event:

  • 850 mb wind speed of 30-40 knots.
  • considerable upstream low level clouds.
  • 1000-850 mb relative humidity greater than 85%.
  • 850 mb temperatures of -12 to -14 deg C (but only with high layer average relative humidity).
  • surface dew point depressions of 15 degrees or less during the maximum daily mixing.
  • located north of the 500 mb shear axis; even better if there is ample vorticity advection.
  • total 1000-850 mb omega values bulls-eyed along the western NC mountains.

Pat Moore and Neil Dixon reformatted the original event review to make it conform to the new standard.