National Weather Service United States Department of Commerce

The Palouse

On the afternoon of January 4th, 2007, a small-scale snowstorm moved over the Palouse. There were 2 interesting features about this storm that make it noteworthy. The first was the scale of the storm, which was largely isolated to the Washington and Idaho Palouse region. The second aspect was the significant impact it made to travel on the Palouse highways.

A loop of the Spokane radar shows the relatively small size of this snow storm.


Snow began in Pullman (KPUW) around 1230 pm PST (2030 UTC) and continued for about 4 hours. Spotter reports from the region show that this area of snow was rather intense, with a few sites picking up over 4" in a short time.

Snow amounts

Examination of the typical 500mb (~18,000 ft) height and vorticity from the GFS initialized at 1200 UTC that morning shows a fairly vigorous short wave in Montana and North Dakota which had moved through Washington the previous day. In its wake the atmosphere was rather unstable (see more on this below). A broader trough was located to the west of the area with several embedded vorticity centers. By 0000 UTC (4pm PST), there is an obvious but small vorticity center that has moved into eastern Washington and intensified. Still, tracking this feature is somewhat difficult. Examination of the 300mb (~30,000 ft) height and vorticity (not shown) was even less definitive with this feature. This is somewhat unusual, since these waves are typically more intense at higher levels in the atmosphere.

500 mb Height

Looking at the height and vorticity at 700mb (~10,000 ft) was rather surprising. The vorticity max moving through the area was very obvious and intensifying as it moved east. Thus, the main dynamic forcing for this storm was in the lower to mid-troposphere, which is lower than for most storms. This, coupled with the instability, had important ramifications (see below).

700 mb Height

Both the NAM/WRF and the GFS models had indicated the potential for precipitation over southeast Washington, but were focused more on the evening (after 00Z) hours. But the GFS had been increasing the speed of this feature over the last few runs. Below is the dprog/dt for the NAM/WRF model for the precipitation valid for the evening of the 4th (00Z-06Z 5 Jan). This loop shows how the last 3 runs of the WRF model (initialized at Jan 3 12Z, Jan 4 00Z, and Jan 4 12Z) were all forecasting precipitation, but there wasn't a great deal of consistency.

NAM change over time

The next loop is similar to the previous one. However, this is the GFS model and is valid for the afternoon (18Z-00Z) of the 4th of January. The first frame of the loop shows that the earlier runs of the GFS had the precipitation out in the Columbia Basin, but subsequent runs had been increasing the speed of this feature, moving into the Palouse. Note that even the precipitation has a somewhat comma-shape to it. Total precipitation amounts from the GFS were around 0.25" of liquid, which agreed well with the ASOS observations at Pullman.

GFS change over time

The previously mentioned, the atmosphere was rather unstable in the wake of the previous day's storm. The image below is the forecast sounding for Pullman from the GFS valid at 00Z (4 pm PST). A few things are evident. The sounding was very unstable. Lapse rates between 900mb and 500mb were 6-7 C/km, which is essentially moist adiabatic. The omega (thin white line) was very strong (-20 mb/s) in the low levels, with a max aoudn 750 mb. Note that this is nearly coincident with the dendritic layer (yellow portion of the sounding). This pointed to the possibility of convective snow with heavy snow rates. Another interesting fact is that the convective schemes from both the WRF and GFS turned on at the very start of the event, but then all of the model-generated precipitation thereafter was from the stable scheme.

Sounding data

The Washington State Patrol report literally hundreds of accidents from this event. Highway 195 (Spokane to Pullman and Lewiston) and Highway 95 (Coeur d'Alene to Lewiston) were both closed for a short time. State Highway 26 west of Colfax was closed for 3 hours due to accidents. While the accumulation of 4-8" of snow in a short period was a major, the roads were also reported to be icy, not snow packed.

Air temperatures during this event were 32F at the Pullman airport. Other mesonet observations on the Palouse showed air temperatures 32-34F during the event. Since this event took place during the afternoon, it was somewhat surprising to see such icy road conditions given these temperatures.

Recent Washington DOT road sensors now allow us to observe the temperatures of the actual road surface. Two such sensors exist on the Palouse. Spangle is located about 10 miles south of Spokane at the north end of the Palouse, and the Uniontown observation is located between Pullman and Lewiston at the south end of the Palouse. Alpowa Summit sensor was also in the area, just south of the Snake River.

Road temps

Road Temperatures

Road temperatures

As you can see from the graphs above, pavement temperatures at all sites were around 28F in the morning, but warmed to around 40F at the start of the snow event. Skies were actually mostly sunny in the morning allowing for good radiational warming of the road surfaces. Thus, as the snow began to fall, it initially melted on the roads. But note the sharp drop in pavement temperatures during the early afternoon. This drop is likely due to three causes:

  • Decreasing solar angle. While sunset at this time is shortly after 4pm PST, the road sensors on days before and after this event showed a peak temperature around 2pm.
  • Increasing clouds. The sunny skies of the morning gave way to cloudy skies by noon as the storm moved in from the west. This thick cloud deck diminished the incoming solar radiation reaching the roads.
  • Melting. As the cold snow contacted the warm roads, the heat flux required to melt the snow would have taken heat out of the pavement, thus cooling it.

The result was pavement temperatures cooling to the freezing mark as the event progressed. Thus the initially melted snow later froze to the roads, creating the icy conditions, even during the daytime hours with air temperatures around 32F. It should be noted that none of the 3 road sensors were located in the area of the heaviest snowfall, and in fact were largely on the fringe of the precipitation (see map above). Thus it is likely that in the area with the intense snow, pavement cooling was more rapid than what is seen in these sensors.

Thus, one of the most important aspects of this event was the timing. If the short wave arrived about 12 hours earlier or later (i.e., during the late night or early morning hours), the road conditions would have been markedly different. Road temperatures would have been well below freezing during these hours, so the snow would have just accumulated on the roads rather than melting and then freezing.

By Ron Miller