During a period of less than 24 hours on 2002-3-17, over two feet of snow fell at Anchorage, AK. This storm was rare in its intensity and duration. Heavy snow fell for 18 consecutive hours from 170700Z to 180000Z. In addition, the large and small scale pattern was one that might cause forecasters to expect subsidence and drying over Anchorage. The topography of South Central Alaska along with weather reporting sites is shown in Figure 18 while Figure 19 displays topographic features mentioned in this discussion.
Prior to the onset of the heavy snow, a moisture plume extended from Hawaii across central Alaska. Figure 1 shows the infra-red satellite image from 0030 UTC 17 Mar 2002. The ETA model appeared to effectively capture this warm moist band, as can be seen on the 700 mb model initialization valid 0000 UTC 17 Mar 2002 (Figure 2). The flow was strong southerly with a narrow upper ridge over central Alaska and a strong meridional flow as can be seen by the AVN (used here only as ETA model data was not readily available further south) 300 mb model initialization at 170000Z (Figure 3). Quasi-geostrophic forcing demonstrates that significant upward vertical motion would be expected at mid and higher levels over central Cook Inlet. Figure 4 (a at170600, b at 171200, c at 171800, d at 180000) shows strong q-vector convergence sustained for at least 12 hours. The heavy snow began near 170600Z and ended just near 180000Z. This is in contrast to the same fields at 700 mb, at the height of the storm, where the strongest forcing appears to be much further south (Figure 5, 700 mb as in fig 4b).
Figure 6 shows a series of cross sections parallel to the flow or along the theta-e ridge (a at 170000Z along N-S line, b same as a at 171200Z, 6c same as 6b but rotated SSE-NNW). Given the saturated nature of this flow, equivalent potential temperature is chosen rather than potential temperature. In a southerly flow, the airmass flows over the Kenai Mountains and typically subsides. Figure 6 shows that, with the model resolution (much coarser than the topography shown in the cross-section), the isentropes do not subside significantly to the north of the southern portion of the Kenai Mountains. In a typical case, it would be expected that much more subsidence would occur than would be “seen” by the model. However, in this case, there is compelling evidence to suggest that cold air not only covered Anchorage but most or all of Cook Inlet.
The upper air soundings from 171200Z from Anchorage and Kodiak show that cold, modified arctic air, was present, up to 850mb at Kodiak and up to 700 mb at Anchorage (Fig 7a shows both Anchorage and Kodiak, b for Kodiak and c for Anchorage. In addition, surface observations demonstrate that the coast, south of the Kenai Mountains, was also under the influence of modified arctic air (Figure 8). Dewpoints are at or below freezing with very cold dry air evident at Valdez (temperature 29, dewpoint 9). At the surface, the combination of a weak, but deepening low to the south and a ridge to the northeast resulted in a continuous supply of cold air to Cook Inlet throughout the day on 17 March (see Figure 9a-170000Z, b-170600Z, c-171200Z, d-171800Z).
A map showing snow accumulations for the period from 170000Z to 180000Z is shown in Figure 17. The highest recorded snow accumulation was 28 inches in the City of Anchorage. It is noted that the snowfall to the north and northwest of Anchorage cannot be precisely determined due to the lack of data in the area. The anomaly in this case was the lack of significant snowfall at Girdwood and at Whittier and the lower amounts at Kenai, Soldotna and Seward. The Kenai radar demonstrates the sharp edge to the heavy precipitation area over Anchorage and Cook Inlet. Figure 10 displays 0.5 degree radar reflectivity along with 1.5 degree radar velocities (170000Z reflectivity-a and velocity-b, 170600z reflectivity-c and velocity-d, 171200Z reflectivity-e and velocity-f, 171800Z reflectivity-g and velocity-h, 180000Z reflectivity-i and velocity-j). The stationary nature of the heavy precipiation area is evident from the radar reflectivities at 6 hour intervals.
As can be seen earlier, the ETA model appeared to initialize fairly well at 170000Z. This would have instilled some confidence in the Q-G forcing which was shown in Figure 4. In terms of precipitation, the ETA model predicted a significant event with local precipitation maxima both south and north of Anchorage Figure 11a shows the ETA, initialized at 17000Z, model 12 hour precipitation for the period ending at 171200Z. Figure 11b shows the 12 hour precipitation for the period from 171200Z to 180000Z. Of note are the two stationary maxima just south of the Kenai Peninsula and just north of the northern end of Cook Inlet. The ETA model also predicted a local precipitation minimum, south of Anchorage, very close to where a minimum was observed.
After considering the available data, several questions remain.
· Why was there heavy snow at Anchorage when typically a southerly flow aloft results in subsidence?
· With advisory-level easterly winds in the Chugach Mountains, why was there not a mountain wave and substantial drying in the subsident flow?
· Why was there a large area of lighter precipitation to the south of Anchorage, a region with a propensity for heavy snow?
The first question can be answered using the observational data in the region. The Anchorage upper air sounding at 170000Z shows cold air up to approximately 700 mb. It can be argued that this cold air extends as far as Kodiak where the sounding at the same time shows cold air up to nearly 800 mb. This air is certainly modified from its arctic origin but is, nonetheless, relatively cold. At the same time, as discussed earlier (see Figure 8), surface observations around Prince William Sound show evidence of arctic air. Even offshore at Middleton Island (PAMD in Figure 8), a dewpoint of 27 degrees is an indication of air of arctic origin. The mean height of the Kenai Mountains, between Seward and Anchorage, if approximately 4000 feet with the peaks near 5000 feet or 850 mb. The sounding data show that the cold air is deeper than the barrier height and indeed has leaked to the south. This density gradient likely nullified the effect of the Kenai Mountains to an air parcel in a southerly flow. In fact, the southerly flow was quite likely still ascending over Anchorage. As an example of the isentropic uplift, follow the 292K theta-e contour in the cross-section in figure 6b.
During the snow event, this cold air became deeper due to the intensifying surface low to the south and the arctic ridge to the northeast. Figures 10d and 10f show the 1.5 degree doppler velocities at 170600z and 171200Z. It can be seen that the cold easterly or northeasterly winds deepened during this interval .
The second question, regarding the strong easterlies in the Chugach Mountains, has a similar answer to that in the preceding paragraph. Although very strong easterlies were recorded to the east of Anchorage (surface obs for Glen Alps), these winds did not descend into Anchorage where winds generally remained light (PANC obs). Although the easterlies were subsident, they did not bring a typical mountain-wave response and low level warming. The source for these easterly winds was likely in the region of northern Prince William Sound where the air was more arctic than maritime and thus did not undergo as much diabatic warming as might generally occur. Thus, the resulting wave would have been much lower in amplitude than a “typical” chinook case. Figure 21 proposes a model of the easterly flow in this case. It appears that the easterly winds at higher level did indeed have a profound impact on the snowfall pattern. Less than half the snowfall was reported at Hillside sites than that in Anchorage. Given that the easterlies likely were not subsident over Anchorage and the fact that the drying was overcome and the easterlies saturated by the heavy precipitation, these easterlies likely had no impact whatsoever in reducing the precipitation at lower levels. Model wind profiles as well as the Anchorage sounding exhibit veering winds with height in low levels; however, this warming was more than overcome by diabatic effects from the snowfall. Even in saturation at below freezing temperatures, the intensity of the snow likely cooled the column through conductive processes.
The third question is not as easy to answer although a viable theory is proposed. Satellite imagery from the snow event shows what appears to be lee cirrus over the Kenai Mountains during the snow event. The POES image from 171423Z appears as Figure 22. However, the earlier argument about cold air resident on both sides of the Kenai Mountains would essentially modify this barrier. A possible shape, as seen by a parcel moving with the southerly flow, is sketched in a conceptual diagram Figure 23. Figure 14 shows the model soundings from the 170000Z integration for a point just south of Seward (a-170000Z, b-170600Z, c-171200Z, d-171800Z, e-180000Z). Figure 16 is similar to figure 14 except for a point much further south, likely more representative of the upstream sounding. Above 400 mb, there is a nearly saturated (with respect to ice), neutral layer. As the entire column underwent isentropic ascent, it is possible that this layer became unstable and, adding latent heat of deposition, induced a wave response at high levels producing what appears to be cirroform cloud in Figure 22. If this wave were of relatively short wavelength, its return to equilibrium could have induced a region of subsidence just to the lee of the Kenai Mountains and a secondary area of ascent over northern Cook Inlet.
This heavy snow event was indeed rare and it is unlikely that its magnitude could have been forecast with the current state of the science and observation networks. Three primary processes came together to produce the area of heavy snow.
· Cold air filling Cook Inlet and preconditioning the coast south of the Kenai Mountains. This effectively eliminated the Kenai Mountain Barrier from the perspective of an air parcel in a southerly flow.
· Strong easterly winds off the Chugach Mountains were not as subsident as is generally the case. This process was predictable by diagnosing the source region of the easterly flow.
· Although the Kenai Mountain barrier was effectively replaced by a cold wedge, it appears that a gravity wave was formed through the gradual ascent, over the cold wedge, of a moist, neutral layer at high levels. This gravity wave process retarded precipitation rates between Prince William Sound and northern Cook inlet while enhancing precipitation rates over Anchorage and the northern portion of the inlet.
The numerical models for this case, within their limitations, performed remarkably well given the long fetch over the datasparse Pacific.
Although a case such as this is unlikely to occur again in the near future, understanding of the various processes will help the operational forecaster in predicting future snow events in Alaska and similar regions, especially in the mountains of western North America.
