Winter Climate Trends

Not like the days of yore

Winter Solstice seems like a like a good time to post about winter temperature trends in the Arctic and Alaska. December though February is conventionally the “winter” season in the northern hemisphere, though that has less meaning in and near the Arctic than at mid-latitudes, what with far below freezing temperatures and snow cover that may persist for five to eight months a year. However, the average temperature for virtually all land (and most of the Arctic oceans) areas for these three months are the lowest of the year, so I often refer to December through February as “mid-winter”.

---

Arctic Temperature

It’s no surprise that over the past 50 years average temperatures have increased nearly everywhere in and near the Arctic. For the Arctic as a whole, the 50-year increase is about 3.1°C (5.6°F). This is not as large as the change in autumn (3.8°C, 6.9°F). The greatest mid-winter warming has been in the Barents Sea area, thanks to the loss of sea ice. The Pacific side of the Arctic (East Siberian Sea to Beaufort Sea) shows significantly less warming than in autumn because sea ice (in the current climate) has reformed in most areas by early December, though of course the ice is far thinner than it was in the late 20th century. The major exception to the warming is a sizable area in eastern Siberia and the north coast of the Sea of Okhotsk, with no significant change or even slight (not statistically significant) cooling.

Fig 1. Change in December-February average temperature, 1973-74 to 2022-23. Data from ERA5 reanalysis.

Here in Alaska the decreased frequency of really low winter temperatures in recent years has been very noticeable: it just doesn’t get as cold as it used to. Or at least that’s my impression. So here’s a look at how the lowest temperatures have changed over the decades. Figure 2 plots the change in lowest hourly temperature during the mid-winter season during the past 50 years.¹

Fig. 2 Change in the December-February lowest hourly temperature 1973-74 to 2022-23. Data from ERA5 hourly reanalysis.

Like the season average, the largest changes are over ocean areas where sea ice loss is most pronounced in the winter. On land the trend is upward in most areas, with the most significant increases in the European Arctic and northwest North America. However, a significant fraction of the Asian Arctic has seen little change in lowest mid-winter temperatures.

---

Arctic Precipitation

Arctic-wide precipitation has increased about 8 percent over the 50 years but there is much larger sub-regional variability than with temperatures, as seen in Fig 3. Very large precipitation increases have occurred in the Atlantic Arctic and over parts of western Alaska and northeastern-most Russia, while substantial parts of Siberia have seen modest decreases in mid-winter precipitation since the mid-1970s.

Fig. 3 Percent change in December-February total precipitation, 1973-74 to 2022-23. Stippled areas highlight areas where the 50-year change is larger than typical year-to-year variability. Data from ERA5 reanalysis.

Winter precipitation in the most of the Arctic is quite low by mid-latitudes standards. For Arctic land areas, the 1991-2020 December through February average total precipitation in only 83mm (3.3 inches). Interpreting changes or differences in quantities that are bounded by zero and are at least sometimes at or close to zero can be tricky, because small changes or differences in absolute values can show up as large percentage values but may not be meaningful. As a start to help with that confound, on Fig. 3 I’ve highlighted areas where the change over the past 50 years is larger than the typical year-to-year variability in total precipitation.²

---

Alaska Mid-Winter Trends

This is the same data as in shown above for the Arctic except zoomed into Alaska and vicinity, and of course temperatures change is shown in °F. The large increase in temperatures on the North Slope are not unexpected, but notice the maximum change is on land and not, as in autumn, over the ocean. Another interesting feature is the area of extreme warming in Southeast Alaska, in the Petersburg and Wrangell areas extending eastward into the coastal mountains. In looking at this in detail, the magnitude of this increase appears to be partly tied to extremely cold mid-winters in 1977-78, 1978-1979 and again in 1981-82. There is very little station data in this area for these years but what there is supports this analysis.

Fig. 4 Change in December-February average temperature, 1973-74 to 2022-23. Data from ERA5 reanalysis.

For Alaska as a whole, mid-winter precipitation has increased about 9 percent in the past 50 years, but with significant increases across much of the Bering Sea region and parts of the North Slope and Interior as well as much of the southern Yukon Territory. The lack of much change in Southcentral and the Gulf of Alaska coastal regions is striking, especially given the temperatures increases.

Fig. 5 Percent change in December-February total precipitation, 1973-74 to 2022-23. Stippled areas highlight areas where the 50-year change is larger than typical year-to-year variability. Data from ERA5 reanalysis.

Figure 6 shows the change in mid-winter snowfall over the past half century. Across the northern two-thirds of Alaska this is almost identical to the precipitation change since almost all winter precipitation falls as snow. The obvious difference is over the Gulf of Alaska and adjacent land areas as well as the eastern and central Aleutians, where there has been a decrease in snow. In Southeast Alaska there’s distinct northeast to southwest gradient to the change (not north to south), with the most significant decreases on the outer coast south of Baranof Island. This reflects the inner channels proximity to the colder continental air that is typically lurking just east of coastal mountains.

Fig. 6 Percent change in December-February total snow (water equivalent), 1973-74 to 2022-23. Data from ERA5 reanalysis

---

Technical details:

Technical details: Temperature and precipitation analysis spatial maps and time series from ERA5 reanalysis data from ECMWF/Copernicus.

ERA5 monthly temperatures, precipitation and snow data available here.

Seasonal minimum hourly temperatures were extracted from ERA5 hourly data with the “Daily statistics calculated from ERA5 data” available here.

Code by B. Brettschneider/NWS Alaska Region allowing rapid ERA5 regional analysis is invaluable for my work.

1. All operational weather models struggle with very low temperatures over snow covered areas due to the frequent surface based temperature inversion; that is, temperatures increasing with height above the ground. This same facet of Arctic winter climate makes station based analysis difficult because of the extreme sensitivity of microscale factors on the lowest temperatures. So while ERA5 is the best tool we currently have for assessing regional scale changes in very low temperatures, I take this as “suggestive” rather than “definitive”, i.e. the patterns are in likely to be in the ballpark but the magnitudes may not be right.

2. Technical detail: stippling shown where the 50-year change is larger than the mean absolute value of the regression residuals.