The greatest variation among the four stations contributing to the mean curve
in Figure 11 occurs from mid-October to mid-November, when BET (long
dashed line) experiences locally strong but shallow boundary layer modification
due to open ocean heat fluxes in off-ice (on-shore) flow (e.g., Olsson and Har-
rington 2000). During these conditions, typified by cold advection with strong
NE winds, BET may actually warm temporarily due to maritime modification,
while more inland and higher elevation locations, most notably WKU, experience
falling temperatures. This effect at BET disappears with the onset of DC as ice is
both pushed by the prevailing northeasterly winds against the Arctic coast and
simultaneously grows in situ.
Also, with the onset of DC, the temperature relationship between IMN and
the other sites becomes decoupled. During this season IMN experiences very
dramatic warming events echoed weakly, if at all, by the mean curve. These IMN
warming events are typically concurrent with strong southerly flow not seen at
the other locations, again suggestive of compressional warming in downslope
conditions. Temperatures drop rapidly once cross-barrier pressure gradients
decrease and NE flow is again established.
Another notable feature in Figure 11 is the dramatic warm-up at all sites in
the second week of November. All of the non-coastal sites experienced warming
to above-freezing daily mean temperatures, with BET showing a weaker but still
discernible warming as well. Further investigation shows that similar, if not as
spectacular, warming events occur at least once during the latter third of EC at
most sites for each of the years in the record. The timing of this warm-up, some-
time between julian days 303 and 323 (the last week in October to the third week
in November), is consistent enough and strong enough to appear clearly in Figure
3, having survived both the multiple year and running average operations.
Figure 12 shows the unfiltered time series of daily mean temperature during
the latter half of EC at SAG for each of the five years of record. All years show
one, and sometimes two, strong warming events, with many lasting for several
days. [Though 1994 does not show a daily mean temperature that exceeds freez-
ing, there were several hours spanning two days that the temperature was within
2C of freezing. The fact that many nearby locations had experienced above-
freezing temperatures was attested to by the widespread very hard and distinct
snow stratum still quite evident at the onset of the melt season (Olsson et al.
2000)]. From the context of the cold-season snow cover, this annual event is sig-
nificant. The resulting higher-density wind slab and melt crust (Fig. 10) acts as a
"capping layer" that has structural, mechanical, and thermodynamic implications
as the snow cover accumulates and evolves through the next six months.