Temperature variability and trends
Mean summer TMAX ranged from approximately 20 °C in the Interior (McGrath, Fairbanks, Northway) to 16–18 °C in the south and about 13 °C at westernmost stations to less than 7 °C at Barrow (Table 2). On the other hand, mean TMIN in the winter ranged from less than − 4 °C at southernmost stations to approximately − 12 °C at Anchorage and King Salmon and about − 18 °C at Bethel and Nome to less than − 25 °C in the Interior and Barrow. Both in the summer and winter, the mean diurnal temperature range (mean TMAX less mean TMIN) is the highest in the Interior and at King Salmon (10–12 °C).
Table 2 Mean TMAX, mean TMIN, linear regression coefficients (ΔTMAX, ΔTMIN; °C/10 years), and highest and lowest percentile-based thresholds (TMAX95 and TMIN5) for the summer (JJA) and winter (DJF) for Alaska (1951–2015) Large differences in the spatial distribution of air temperatures are also visible in the spatial distribution of extreme values. Table 2 lists the highest and lowest percentile-based thresholds (TMAX95 and TMIN5) calculated for every weather station for the summer and winter. The lowest TMAX95 values are most often detected in late summer, with the exception of early summer for Barrow and Cold Bay, while the highest values are most often noted for early July, with late July and early August in the case of Barrow and Cold Bay. The highest TMIN5 values in the north and along the western coast are noted in the early winter—and late winter in other regions of Alaska. The lowest TMIN5 values at most of the weather stations are noted in the first half of January, but in the north and along the western coast in late January and early February. Percentile-based thresholds vary the most in the summer at the Barrow station (12.1 °C), while in the winter, at Barrow and in the Interior (7.2 to 7.8 °C).
Air temperature in Alaska is characterized by significant variability from year to year, although higher variability is noted in the winter than in the summer (Fig. 2). Higher variability is also noted in the Interior relative to the other regions (Fig. 2b). The best linear fit over the period 1951–2015 indicates a statistically significant increase in TMAX, which occurs at the highest rate in the summer at Barrow (0.5 °C/10 years) and in the winter at King Salmon (1.0 °C/10 years) and in the Interior (Table 2, Fig. 2a, b). Summer TMIN tends to decline at most of the studied stations, although the trend is usually statistically insignificant. On the other hand, winter TMIN increases at almost all of the stations (max. 0.7 °C/10 years at King Salmon), and the increase is statistically significant in most cases (Table 2).
In general, the observed warming occurs faster in the winter than in the summer, and also TMAX increases faster than TMIN, which results in an increase in the mean diurnal temperature range in both seasons. This increase occurs fastest in western Alaska and in the Interior (Table 2, Fig. 2b, c).
Long-term patterns in TMAX and TMIN variability in Alaska were produced by calculating trends for multi-year sub-periods of variable length (Fig. 3). The results of this calculation made it possible to observe common characteristics for the state as a whole as well as differences for individual regions. In general, TMAX tended to decline in the summer until the early 1960s, while afterwards it tended to increase, with a few cooler periods (Fig. 3a, c). The TMIN trend for the summer consisted of a decline in most sub-periods, while the only readily observable increases were noted between the years 1970 and 2000 (Fig. 2). The TMAX and TMIN trends for the winter were largely the same, but the magnitude of TMAX was usually higher (Fig. 2). A strong statewide cooling period was observed in the following two sub-periods: 1956/57 to 1975/76 and 1976/77 to 2000/01 (Fig. 3b, d). This illustrates the effect of an unusually warm winter in 1976/77 on the results of trend analysis (Figs. 2 and 3b, d). A rapid decline in air temperature also occurred in the first decade of the twenty-first century, especially in the winter, which may be observed for all weather stations except Barrow, where air temperature values were characterized by uninterrupted increases since the mid-1980s (Figs. 2 and 3b, d).
Summertime temperature extremes
Temporal and spatial variability
An average of four WDs occurred at each weather station per season, while their maximum number in one summer ranged from 15 at Nome to 30 at Cold Bay (Table 3). The number of WSs at most stations ranged from 30 to 34, with the exception being Juneau at 42; their maximum number per one season ranged from 3 to 6 at Cold Bay (Table 3).
Table 3 Characteristics of summer warm extremes in Alaska (1951–2015); ∆WDs—change in the number of WDs per 10 years, ∆TS95—change in the TS95 per 10 years The share of WDs in the formation of WSs was greatest at weather stations located farthest to the east (54 and 55%), somewhat smaller in the Interior and southern and western Alaska, and the smallest at Barrow (40%) (Table 3). Data for all weather stations indicate that WSs occurred during all summer months, with the largest number occurring in July (124) followed by August (110) and June (104).
The long-term pattern of WDs and WSs variability differed significantly for each weather station (Fig. 4). A positive WDs trend was noted only in the north and at some stations in the west—statistically significant at Nome and Barrow, reaching 0.7 WD per 10 years (Table 3). No trend was detected for the remaining stations, while summer seasons with a large number of WDs and WSs occurred in different years depending on region. At the beginning of the study period, many warm extremes were noted in southern and central Alaska (Fig. 4b, d), while in the 1970s and early 1980s at stations located farthest east and west (Fig. 4d). The 1990s were characterized by a relatively large number of WDs at Barrow and Fairbanks (Fig. 4a, b). Between 2000 and 2015, summers with a relatively large number of warm extremes were noted in all regions of the state. In addition, each weather station recorded a season with the largest number of warm extremes during the same period of time. An exceptionally large number, and in some instances a record number, of WDs and WSs as well as a long WS total duration was recorded at all stations except Cold Bay in the summer of 2004 (Fig. 4). A large number of WDs and WSs also occurred in the Interior and at Bethel and King Salmon in the summer of 2013 (Fig. 4b, c). Finally, the largest number of warm extremes were noted at Cold Bay in the summer of 2014 (Table 3).
WS duration
The most frequently occurring WSs lasted 3 days (41% WSs at Northway to 67% at Barrow) and 4 days (15% at McGrath to 37% at Nome). Every weather station recorded at least one 5-day WS (Table 3). WSs lasting 1 week or more occurred mostly in the Interior and in western Alaska. In addition, long duration WSs most often occurred in June and August rather than in July. The longest WSs lasted 10 days and occurred at McGrath and Bethel in the summers of 1957 and 2004, respectively (Table 3).
Both longer and shorter WSs occurred throughout the entire study period (Fig. 4). WSs lasting 5 days or more occurred in various years throughout the study period starting with the very beginning of the study period (e.g., Cold Bay, McGrath, Fairbanks, Juneau), through its middle (e.g., Bethel, King Salmon, Fairbanks), and end period (e.g., King Salmon, Anchorage, Northway). More than half the weather stations, scattered throughout Alaska, recorded the longest WS in the summer of 2004, while Anchorage and King Salmon did so in the second half of June 2015. The remaining weather stations recorded their longest WSs in different years (Table 3).
Severity
Mean summer TS95 (cumulative temperature excess above the 95th percentile) ranged from 6 °C at Anchorage and Northway to 9 °C at Barrow, Nome, and King Salmon. Over the long term, TS95 were strongly associated with the long-term pattern of the occurrence of warm extremes; hence, the number of WDs and WSs usually directly translated into the extremeness of a given summer season (Figs. 4 and 5). A comparison of Figs. 4 and 5 makes it possible to identify seasons characterized by a relatively large number of warm extremes of relatively low severity (e.g., Juneau 1951) as well as those with a relatively small number of WDs and WSs but high severity (e.g., Barrow 1977).
Over the long term, TS95 was characterized by an increase whose largest rate was noted at Barrow, Nome, Bethel, and Anchorage (statistically significant change) and smallest rate in the Interior (Table 3, Fig. 5a–c). The largest TS95 was noted at each weather station during summers with some of the largest, although not necessarily the largest, numbers of warm extremes. The relationship between the highest TS95 and the occurrence of the longest WSs is readily observable—they both occurred at six weather station during the same season (Table 3). The summers characterized by the greatest severity of warm extremes were noted at Nome in 1977 and at Juneau in 2004 (about 60 °C; Table 3, Fig. 5d).
The highest daily TMAX values exceeded 30 °C and occurred most often in the Interior, but were also noted in the west (except Cold Bay) and at Juneau (Fig. 5b–d). At the Anchorage weather station, TMAX exceeded 29 °C only in 1969, while at Barrow and Cold Bay, it exceeded 25 °C only in 1960 and 1993, respectively (Fig. 5a).
Of the highest seasonal TMAX values noted during warm extremes, anywhere from 30% (Nome) to 44% (Juneau, King Salmon) occurred in the course of a WS (Fig. 5).
Wintertime temperature extremes
Temporal and spatial variability
An average of four CDs were noted per weather station, but their maximum number per winter ranged from 13 at King Salmon to 25 at Fairbanks (Table 4, Fig. 6b). The number of CSs ranged from 27 at Barrow to 43 at Northway, and their maximum number per season ranged from 2 at King Salmon to 5 at Fairbanks (Table 4).
Table 4 Characteristics of winter cold extremes in Alaska (1951/1952–2015/2016); ∆CDs—change in the number of CDs per 10 years, ∆TS5—change in the TS5 per 10 years The share of CDs in CSs formation exceeded 70% in the Interior and equaled about 60% in western Alaska, with the exception of Cold Bay, and stood at about 60% at Anchorage (Table 4). CDs least often formed CSs at Barrow and at stations located farthest to the south. CSs most often occurred in December, with a total of 132 at all weather stations, and in January, with 130, and the least often in February at 92.
A statistically significant decrease in the number of CDs occurred at all stations except for Nome, King Salmon, and Cold Bay, where the number of CDs did not change over the long term (Table 4). The decline in the number of CDs was greatest at Anchorage, with a loss of one CD per 10 years, while at other weather stations, the decline rate ranged from 0.4 to 0.6 per 10 years.
In the early part of the study period, the number of cold extremes in the Interior and at Barrow and Juneau was large (Fig. 6a–d). In the 1960s, a very large number of cold extremes were noted in the Interior and at Anchorage, especially the winters of 1961/62 and 1964/65 (Table 4, Fig. 6b). The 1970s were characterized by a large number of CDs and CSs throughout the state, with the coldest winters in 1970/71 (Interior), 1974/75 and 1975/76 (western and northern Alaska), and 1978/79 (eastern Alaska). The winters of 1988/89 and 1989/90 were very cold in the western part of the state (Table 4, Fig. 6c), while winters in the late 1990s may be described as rather cold. Since the year 2000, many cold extremes were recorded at King Salmon and Cold Bay, while the only very cold winter season to affect a larger area was that of 2011/12 in western Alaska (Fig. 6c).
CS duration
The most frequently occurring CSs lasted 3 days (from 30% CS at Barrow to 56% at Nome), 4 days (from 7% at Cold Bay to 37% at Barrow), and 5 days (from 9% at Northway to 25% at Cold Bay). Each weather station recorded at least one CS lasting 8 days or more, while 10-day CSs and longer CSs were noted at each station except those located farthest south. The longest CSs lasted 17 days and occurred at Bethel and McGrath in the winter of 1988/89 (Table 4).
Both long and short CSs occurred throughout the multi-year study period (Fig. 6). CSs lasting 6 days or more occurred both in the beginning of the research period (e.g., Bethel, Fairbanks, Anchorage) and in the middle (e.g., Barrow, Nome, Juneau) as well as towards the end (e.g., Bethel, Nome). The longest CSs at each weather station were noted in the second half of January in 1989 (western Alaska—longest CSs during the study period), in February of 1979 (Interior, Juneau), and in the following seasons: 1961/1962, 1973/1974, 1974/1975, and 2005/2006 (remaining stations; Table 4).
Severity
Mean winter TS5 (cumulative temperature excess below 5th percentile) at the studied stations ranged from 10 °C at Barrow, Cold Bay, Fairbanks, and Nome to 14 °C at King Salmon. As in the case of warm extremes, the long-term pattern of TS5 reflects the long-term pattern of cold extremes, and in the vast majority of cases, a large number of CSs and CDs per season translate into significant extremeness of the given winter season (Figs. 6 and 7). In some cases, a fairly small number of cold extremes were associated with high severity (e.g., Barrow 1959/1960, Fairbanks 1968/1969) and vice versa; a relatively large number of CDs and CSs was characterized in some cases by low severity (e.g., Bethel 2011/2012).
TS5 for the study period followed a declining tendency at most of the weather stations. The decline rate was highest and statistically significant in the Interior and at Juneau, Anchorage, and Barrow (Table 4, Fig. 7a–d). A statistically insignificant increase in the TS5 was noted at King Salmon and Cold Bay (up to 0.9 °C/10 years); hence, cold extremes were characterized by increasing severity (Table 4). The largest TS5 at each weather station occurred in winters with a large, although often not the largest, number of cold extremes, while at seven stations, this occurred in seasons with the longest CSs (Table 4). The highest TS5 occurred at Anchorage in the winter of 1961/62 (almost 123 °C) as well as at McGrath in the winter of 1988/89 and Northway in the winter of 1978/79 (slightly more than 100 °C; Table 4).
The lowest daily TMIN values fell below − 50 °C in the Interior (down to − 58.9 °C at McGrath; Fig. 7b). At Barrow, the lowest daily TMIN fell slightly below − 49 °C only twice (Fig. 7a). In western Alaska, TMIN rarely fell below − 40 °C (Fig. 7c), while at Cold Bay, TMIN rarely fell below − 20 °C. The lowest TMIN values at Anchorage and Juneau were about − 30 °C, although these occurred at Juneau much less frequently (Fig. 7d).
Of the lowest seasonal TMIN values recorded during cold extremes, between 20% (Cold Bay) and 68% (Fairbanks) occurred during a CS (Fig. 7).
Circulation background
Figure 8 shows the frequency of WDs for different advection types (ATs) in relation to the total number of days with a given advection type during summer (1979–2015). We combined these two characteristics to show indirectly the conditional probability of occurrence of warm extremes for different ATs. Figure 9 is the same but for ATs and CDs in winter.
The occurrence of warm extremes at Barrow independently of pressure systems was accompanied by an influx of air masses from the south and southwest (greatest probability) as well as west and occasionally southeast (Fig. 8a). In the western part of the state, except for Cold Bay, warm extremes occurred most frequently along with advection from the east and to a lesser degree north, although the probability of their actual occurrence was greater for anticyclonic ATs (Fig. 8c). The distribution of warm extremes in the Interior varied strongly for different ATs, occurring at McGrath with different ATs, with the greatest probability for anticyclonic types with advection from the east. The largest number of warm extremes at Fairbanks occurred along with types Ec and Bc, and while their probability was greatest during advection from the east, it was also the least variable among the different ATs present (Fig. 8b). Warm extremes at Northway occurred mainly with advection from the west and with advection-free types, but the greatest probability was noted with types Sa, Ea, and Ba. At Anchorage, Cold Bay, and Juneau, warm extremes occurred most often along with types characterized by an influx of air masses from the S, SE, N, and NE (Anchorage); Ea, SWa, and Wa (Cold Bay); and NE, E, SWa, and Wa (Juneau), and with advection-free types, although the probability of their occurrence was greatest for types NEa and Ea (Fig. 8d). WSs occurred with both cyclonic and anticyclonic types; the higher probability of their occurrence for anticyclonic types is evident only at the easternmost stations, i.e., Northway and Juneau (Fig. 8d).
Cold extremes at Barrow occurred in the presence of advection from different directions, although more often with anticyclonic than cyclonic types (most favorable type: Bc, Fig. 9a). The occurrence of cold extremes in western Alaska (except Cold Bay) was facilitated by advection from the north and northwest, and especially by anticyclonic ATs (Fig. 9c). In the Interior, cold extremes were accompanied by both cyclonic and anticyclonic types with advection from the northeast (especially at McGrath), east (Fairbanks, Northway), and the SEa type (Northway), with the greatest probability of their occurrence being associated with types NEa and Ea (Fig. 9b). Cold extremes at Cold Bay occurred most often along with cyclonic types with advection from the north; however, type Ea was most favorable. The occurrence of cold extremes at Anchorage was favored with advection from the north and northeast, while the probability of their occurrence was similar to that for ATs with an influx of air masses from the north. Cold extremes at Juneau were facilitated by advection from the east and northeast (Fig. 9d), and this is especially true of anticyclonic ATs. CSs occurred with both cyclonic and anticyclonic types; however, the probability of their occurrence is generally higher for anticyclonic types. This is especially evident in the Interior (Fig. 9b).