From early evening on Sunday, December 17, 1995, through the following afternoon, a slow-moving upper-level southern stream storm from the Pacific dumped snowfall over the Oklahoma and Texas Panhandles. The heaviest snow, the majority of which fell from 0000 through 1200 UTC on the 18th, was in the Oklahoma and northwestern Texas Panhandles, where amounts ranged from 4 to 8 in. Snowfall in Amarillo ranged from nearly 3 in at the airport, east of town, to 4 in on the west side of town.
As early as Wednesday, December 13, NWSO Amarillo forecasters recognized some potential for a significant winter precipitation event for the 17th and 18th, as long-range models were forecasting a strong upper-level system to evolve over the southwestern U.S. With the long-range models showing differences in the evolution and track of this low, however, there was much uncertainty this far in advance with respect to precipitation type and amounts. By the midnight shift on Saturday the 16th, the office had an excellent handle on this storm, and mentioned in the Area Forecast Discussion the possibility for heavy snow from late on the 17th through the 18th. A Winter Storm Watch for Sunday night and Monday was issued with the Saturday afternoon forecast package. On Sunday afternoon (the 17th), the watch was upgraded to a Winter Storm Warning for heavy snow for Sunday night and Monday.
This paper provides an in-depth review of key parameters which led to the heavy snow, and demonstrates that forecasters cannot always rely on techniques that are based on traditional heavy snow climatology.
2. Synoptic Setting
The upper-level storm (Figs. 1a-c) began forming by 1200 UTC Saturday, December 16, over Nevada, as a strong 300 mb jet streak in excess of 115 kt dropped southeastward over coastal California. Over the next 48 hr, the storm continued to dig slowly southeastward, reaching southern Arizona by 1200 UTC on the 17th, and the Big Bend area of West Texas by 1200 UTC on the 18th, when the upper-level low became negatively tilted, and the 300 mb jet streak rounded the base of the trough and intensified to 130 kt.
The surface low began developing on the 17th near San Antonio and maintained intensity as it traversed northeast Texas into Arkansas on the 18th. The surface and 850 mb lows passed well southeast of the Panhandles, while the 700 and 500 mb lows moved over the Permian Basin and Low Rolling Plains into south-central Oklahoma. None of the three NCEP short-range models (NGM, Eta, and AVN) slowed the upper-level system enough as it moved from central Arizona at 0000 UTC on the 17th, through southwestern New Mexico, to near El Paso at 1200 UTC on the 18th. The overall track of this storm was forecast most accurately by the NGM. For this reason, NGM PC-GRIDDS data were chosen for analysis.
The Amarillo upper air sounding for 0000 UTC December 18 (Fig. 2) showed a nearly saturated environment from the surface through 600 mb, with a precipitable water content of 0.54 in (nearly 200 percent of normal). Twelve hours later the environment was still nearly saturated (PW of 0.40 in, 150 percent of normal). Gridded model data indicated the mean 1000-500 mb relative humidity was at or above 90 percent over the Panhandles during the time of heavy snowfall. Thus, the available moisture for this storm was excellent. The 0000 UTC sounding on the 18th showed the freezing level near 1,800 ft agl; and it was at the surface 12 hr later. Although a 0000 UTC surface analysis (not shown) indicated northerly winds, the sounding indicated an upslope component in the lower levels from just off the surface up to 750 mb.
This upper-level southern stream storm was particularly strong in view of the 500 mb heights, which lowered to an average of 550 dm as the low passed over southwestern New Mexico and across the Big Bend of Texas. The 500 mb low also tracked south of 32 deg with the assistance of an intensifying 300 mb jet streak (to 130 kt), which caused the low to become negatively tilted as it rounded the base of the trough from 0000 through 1200 UTC on the 18th. In addition, from 0000 UTC on the 17th through the afternoon on the 18th, there was substantial severe weather over southeastern Texas and much of Louisiana, including 13 tornadoes, along with numerous reports of hail and wind damage, according to unofficial reports received at the Storm Prediction Center from the 17th through the 19th. This severe weather was undoubtedly in response to the approaching and intensifying storm system and its associated strong jet streak.
As the 300 mb jet streak rounded the base of the trough, the Oklahoma and Texas Panhandles were located in the left-front quadrant, or exit region, of the jet. The NGM gridded data indicated significant divergence at 300 mb over the Panhandles from 0000 through about 0600 UTC on the 18th. An initial vorticity lobe was ejected from the low over the eastern Panhandles from 0000 through 1200 UTC on the 18th (Fig. 3), allowing for positive vorticity advection over the area. The main energy associated with the upper low tracked well southeast of the Panhandles, however, on a path from the Big Bend area across South Texas, then over northeast Texas. The 500 mb through 300 mb layer height fields indicated excellent diffluence over the Panhandles during the period of heaviest snowfall. Thus, dynamic forcing played a key role in producing significant vertical motion for the snow.
Figure 4 shows an isentropic analysis for the 296K surface at 0000 UTC on the 18th. Isentropic lift via easterly flow was occurring over all of the Panhandles at this time, but had shifted over the Oklahoma and northern Texas Panhandles by 1200 UTC on the 18th. By 0000 UTC on the 19th, isentropic lift had shut off over all of the Panhandles as winds turned northerly on the 296K surface.
Overrunning was also evident from 0000 through 1200 UTC on the 18th. Figure 5 shows 700 mb theta-e contours along with the advection of theta-e. At 0000 UTC on the 18th, theta-e advection was occurring over all of the Panhandles with east to southeasterly flow at 700 mb advecting warmer, moister air from Oklahoma and northern Texas (where surface dewpoints ranged from near 40F to the lower 50s) over an increasingly colder airmass in the lower levels. But similar to the isentropic lift, the theta-e advection shifted to the northern Panhandles by 1200 UTC on the 18th. After 1200 UTC, the theta-e advection quickly decreased in conjunction with the isentropic lift.
Occasional rain occurred over the Panhandles during the daylight hours on the 17th. A surface cold front passed from northwest to southeast across the region that evening, allowing 850 mb temperatures and the 1000-500 mb thickness to drop, and changing the rain to snow. Figure 6 shows 850 mb heights and temperatures, and the 1000-500 mb thicknesses at 0000 and 1200 UTC on the 18th. By looking at the 2 deg C isotherm at 850 mb and the 546 dm contour (both used as critical values for rain versus snow in the southern High Plains), it is evident that the southeastern Texas Panhandle missed out on any significant snowfall, since the strong overrunning, isentropic lift, and divergence at 300 mb occurred over this region when the airmass was still too warm to support snow. The 546 dm thickness line and 850 mb 2 deg isotherm apparently passed through the southeastern Texas Panhandle between 0600 and 1200 UTC, when the best overrunning and isentropic lift shifted into the northern Panhandles.
Moderate upslope flow (Fig. 2 and a cross section analysis [not shown]) occurred for a very short time centered around 0000 UTC on the 18th. Thus, orographic lift likely played only a limited role in this event.
3. Synoptic-Scale Heavy Snow Climatology Techniques and the Decision Tree for Snow/Heavy Snow in West Texas
Some heavy snow forecast techniques based on climatology were examined for applicability in this case. Goree and Younkin (1966) focused on the tracks of the surface low and 500 mb vorticity maximum, and the 500 mb height field. They show, for example, that the highest frequency of heavy snow generally occurs 2.5 deg latitude to the left of the track for a well-defined surface low. They found a greater than 20 percent frequency from 1 to 4.5 deg to the left of the surface low track. This technique would not have worked well in this event, as the surface low track was in excess of 6 deg latitude to the southeast of the heavy snow area.
Goree and Younkin (1966) found the heaviest snow is frequently 2.5 deg to the left of the track of the 500 mb vorticity maximum, with a greater than 20 percent frequency from 1 to 5 deg latitude to the left of the track. This general rule seemed to work well in the case studied here, considering the initial, weaker vorticity lobe that was ejected from the upper low over the eastern Panhandles and western Oklahoma. In addition, the track of this weaker vorticity lobe brought PVA to much of the Panhandles. If the rule were applied to the track of the strongest vorticity max, it would not have worked in this case, as the heavy snow was anywhere from 6.5 to 8 deg to the left of the track.
Goree and Younkin (1966) found the heaviest snow generally to be along and slightly to the left of the path of the 500 mb low, and slightly downstream from the point where the contour curvature changes from cyclonic to anticyclonic. Data from the NGM model indicated this general guideline was applicable in the case studied here.
Brown and Younkin (1970) emphasized the 850 mb low can be a more useful indicator of significant weather than the surface low center in cases of cold air advection in the lowest layers, where the surface low center may not be well-defined. Based on 81 cases, the highest frequency of heavy snow was found to occur 1.5 deg (90 nm) to the left of the 850 mb low track and in warm advection between the 0 and -5C isotherms. In 94 percent of the cases, the 850 mb low deepened. For the event studied here, the 850 mb low track was from 3.5 to 5.5 deg southeast of the heavy snow region. In addition, although the 850 mb low strengthened slightly during the 12-hr heavy snow period, no warm advection was noted at 850 mb. Instead, warm advection occurred primarily at 700 mb. Thus, the 850 mb low technique presented by Brown and Younkin (1970) would not have assisted in this forecast.
Alan Johnson and others at WSFO Lubbock (personal communication) developed a decision tree to assist in forecasting heavy snow in West Texas, based on a study by Cook (1977). This technique, mainly Side B of Fig. 7, would not have fared well for forecasters working this event. Only 5 of the 7 checklist factors on Side B were met, indicating that heavy snow was unlikely. One of the factors not met was significant surface to 700 mb warm air advection. The warm air advection for this event primarily occurred around the 700 mb level. The second factor not met was a 500 mb vorticity maximum equal to or greater than 18 10-5 sec-1, moving east-northeast and located upstream, maintaining or increasing in intensity. The vorticity lobe ejected from the low over the eastern Panhandles had a maximum strength near 15 10-5 sec-1. The strongest vorticity maximum, exceeding 20 10-5 sec-1, maintained its intensity through 1200 UTC on the 18th as it rounded the base of the 500 mb low, but was obviously well south of the Panhandles.
Key parameters which played a role in the heavy snow event of December 17-18, 1995, included:
Other contributing factors included strong dynamic forcing associated with the left exit region of an intensifying 300 mb jet streak; positive vorticity advection at 500 mb, along with the track of the initial, weaker vorticity lobe; and strong upper-level diffluence. The most persistent overrunning (around 700 mb) and isentropic lift occurred in the heavy snow region with 1000-500 mb thicknesses at or below 546 dm and 850 mb temperatures at or below 2C. It should also be mentioned that experience has shown overrunning to be a major contributing factor in producing most of the widespread heavy snows in the Oklahoma and Texas Panhandles. Finally, moderate upslope flow likely played a role for a few hours centered around 0000 UTC on the 18th.
An examination of climatology-based heavy snow forecast techniques shows in this case that forecasters cannot always depend on such methods. Emphasis should first be given to looking at basic key parameters such as described above.
The author wishes to thank Edward Andrade, lead forecaster at NWSO Amarillo, for his review and input into this paper.
Brown, R.F. and R.J. Younkin, 1970: Some Relationships Between 850 Millibar Lows and Heavy Snow Occurrences Over the Central and Eastern United States. Mon. Wea. Rev., 98, 399-401.
Goree, P.A. and R.J. Younkin, 1966: Synoptic Climatology of Heavy Snowfall Over the Central and Eastern United States. Mon. Wea. Rev., 94, 663-668.
Cook, Billie J., 1977: A Snow Index Using 200 mb Warm Advection. NOAA Technical Memo. NWS SR-93, National Weather Service Southern Region, Ft. Worth, 14 pp.