Climate warming in the Swedish sub-Arctic since 2000 has reached a level at which statistical analysis shows for the first time that current warming has exceeded that in the late 1930's and early 1940's, and has significantly crossed the 0°C mean annual temperature threshold which causes many cryospheric and ecological impacts. The accelerating temperature increase trend has driven similar trends in the century-long increase in snow thickness, loss of lake ice, increases in active layer thickness, lake water TOC (total organic carbon) concentrations and the assemblages of diatoms, and changes in tree-line location and plant community structure. Some of these impacts were not evident in the first warm period of the 20th Century. Changes in climate are associated with reduced temperature variability, particularly loss of cold winters and cool summers, and an increase in extreme precipitation events that cause mountain slope instability and infrastructure failure. The long term records of multiple, local environmental factors compiled here for the first time provide detailed information for adaptation strategy development while dramatic changes in an environment particularly vulnerable to climate change highlight the need to adopt global mitigation strategies.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 Climate changes [Trenberth et al., 2007] and their impacts [Hassan et al., 2005; Parry et al., 2007] are evident in several parts of the world. Recent Northern Hemisphere temperatures are probably the highest for the last 2000 years [e.g., Moberg et al., 2005; Jansen et al., 2007]. However, few integrated, multidisciplinary studies exist for one location, particularly in remote and vulnerable regions where climate and direct human impacts can be clearly separated. Also, rigorous statistically-orientated analyses are few because inter-annual variability is great while records are often too short-term for trends to become significant [Weatherhead et al., 1998]. To understand recent decadal changes in climate in sub-Arctic Sweden and their environmental impacts, and to compare these with environmental changes in the earlier 20th Century warm period, we applied statistical techniques (see auxiliary material) to unique, long-term data on climate, the cryosphere and ecology. The data have been gathered at the Abisko Scientific Research Station (68°20′ N, 19°02′ E) since 1913. Smoothed mean annual temperatures (MATs) have historically been below 0°C [Alexandersson et al., 1991], and this is an important threshold temperature above which permafrost is particularly vulnerable [Smith and Riseborough, 1983] and rates of most biological activity and biogeochemical cycling markedly increase [Grogan et al., 2004; Mastepanov et al., 2008]. We expect warming that crosses this threshold to have disproportionate impacts compared with a similar degree of warming, but below the threshold, at higher latitudes. However, our findings are relevant to large areas of the high latitudes where MATs will exceed 0°C in the future.
 Climatic, abiotic and biotic parameters have been monitored at Abisko, in the Torneträsk catchment, sub-arctic Sweden (68°20′N, 19°02′E, 385 m a.s.l.). The area is dominated by Lake Torneträsk which covers an area of 330 km2 and mountains that reach 1750 ma.s.l. The vegetation extends from relict, localised pine forest, through extensive mountain birch forests to alpine regions with permanent snow. The study area covers a strong climatic gradient, ranging from a maritime climate in the West to a more continental climate in the East. In general, regional precipitation decreases and seasonal temperature differences increase towards the East. The lowest precipitation is however, found around Abisko (∼300 mm/yr for most of the 20th Century) due to a rain shadow and the highest precipitation is found near the Norwegian border (∼900 mm/yr) [Alexandersson et al., 1991].
 Mean daily air temperature and precipitation data were derived from three measurements made each day at 0700, 1300, 1900 hrs local time between 1913 and 2008 and were quality controlled using the methods by the Swedish Meteorological Hydrological Institute (SMHI) [Alexandersson et al., 1991]. The lake ice freeze up (1915–2007) and break up (1916–2007) dates are based on ocular daily observations. A well-defined part of the lake should be ice-covered or ice-free before it is defined as frozen or ice-free. The active layer monitoring has been conducted annually between 1978 and 2005 during the third week of September, when annual maximum thaw depth occurs [Akerman and Johansson, 2008]. Snow depth was measured at 07.00 hrs local time every 5 days after 1956 and at varying intervals from daily to 5 times a month prior to 1956 [Kohler et al., 2006]. Snow depth at two sets of 10 stakes together with a single stake have been measured during 1913 to 2006. Data has been analysed using fitted uncorrelated polynomials, i.e., Legendre polynomials and also 2nd degree B-spline polynomials (see auxiliary material).
3. Results: Climate Trends
 Climate warming in sub-Arctic Sweden since 2000 has exceeded the warming in the late 1930's and early 1940's and has accelerated. Statistical smoothing of MAT data showed a maximum of -0.3 °C in 1940 before the recent warming started in about 1975 (Figure 1a). Smoothed MATs have risen between 1913 and 2006 by 2.5 °C and by 1.5 °C between the minimum of the cooling period in ca. 1974 and present (2006). The current interpolated temperature (any single year after 2005) is statistically significantly (p = 0.01) higher (0.9°C) than that in 1940. Also, these interpolated temperatures significantly (p = 0.06) exceeded the 0°C threshold for the first time in the 20th and 21st Centuries. Inter-annual variability has decreased: the local maxima and minima have converged over the Century (Figure 1a), mainly because there have been fewer extremely warm years and fewer extremely cold years within the past decades.
 The seasonal evolution of mean annual temperature is complex. Autumn (SON) temperatures show a similar pattern to mean annual temperatures (Figure 1c) whereas winter (DJF: December year 1, January and February year 2) temperatures (Figure 1e) show a peak in the 20th Century that is approximately 7 years earlier than that in MATs (Figure 1a). In contrast, spring (MAM) temperatures (Figure 1d) show a gradual, consistent increase over the period 1913 to 2007 (Table 1 of Text S1 of the auxiliary material). Summer (JJA) temperatures (Figure 1b) showed no significant difference between the late 1930s and early 1940s and recent years (p = 0.20). Spring is the most, whereas winter is the least, important source of mean annual temperature rise during the period of warming from the early 20th Century maxima to 2006 (Table 1). This pattern contrasts with that at the beginning of the 20th Century when winter warming was substantial and in other studies from other places which identify winter as one of the most important sources of mean annual warming since the mid 19th Century [Jones et al., 1999; Trenberth et al., 2007].
Table 1. Annual and Seasonal Temperature Changes ± One Standard Error at Abisko Over the 20th Century With Comparisons of the Two Warming Periods, i.e. Degree of Warming and Maximum Temperatures Experienceda
Period of Comparison
1913-Early 20th Century Maxima
Mid 20th Century Minima to 2006
Early 20th Century Maxima to 2006
1913 to 2006
The comparisons are based on temperature differences between pairs of years interpolated from smoothed trends (see text and auxiliary material). Values in bold font have confidence limits >95%. The early 20th Century smoothed maxima for annual, autumn and summer temperatures occurred in 1940 ± 3 years, whereas those for winter and spring occurred in 1933 ± 2 years and 1945 ± 2 years respectively. The mid 20th Century smoothed minima occurred in 1974 ± 4 years. The exact choice of year is not crucial for the indicated intervals.
 Some temperature changes have been abrupt while others have accelerated. After 1988, cold winters which were common previously, ceased abruptly and relatively cold summers became rare. During 1997–2008, with the exception of 1999 and 2000, each mean summer temperature has been above the average for the 20th Century. Both spring and autumn had significant, accelerating (non-linear) trends (Figures 1c and 1d and auxiliary material). April exhibits a particularly steep (3 degrees), significantly accelerating rise between 1986 and 2007 (data not shown).
 Summer precipitation (Figure 1f) was relatively constant for the first half of the 20th Century and then increased (117 mm in 1959 to 143 mm in 2006; P = 0.09) after approximately 1980, in concert with total annual precipitation (auxiliary material). This increase occurred one decade before an abrupt increase in MAT's after 1989. Recently, dry winters, and fairly dry springs coincide with wetter summers and autumns (auxiliary material). Winter precipitation increased over the 20th Century with two periods of heavy snow fall (centred on the 1930's warm period and mid 1990's (auxiliary material)). The periods of heavy snow fall were associated with greater snow depth [Kohler et al., 2006] (Figure 1g) which has doubled over the Century. However, there has been an extremely dry period with relatively shallow snow since the late 1990's.
 In contrast to temperature, variability in extreme precipitation has increased since 1913 (see Figure 1 in Text S1 of the auxiliary material). Also, the magnitude of outliers (“extremes of extremes”) has increased. Since an extreme event of 39.7 mm/day in 1915, precipitation extremes have risen from 33.4 mm/day in 1930 to values that increase successively over the century to exceed earlier records. The most recent extreme of 61.9 mm/day in 2004 is the record event of the instrumental period.
4. Impacts on Abiotic and Biotic Environment
 The climatic changes have affected the landscape and particularly the cryosphere. Ice freeze up date for Lake Torneträsk has become significantly later from December 6th to January 5th (P = 0.005) by 30 days since 1913 (Figure 1h). Concurrently, ice break up day has become significantly earlier from the middle of June to the end of May by 17 days (P = 0.005) (Figure 1i). Break up day became earlier after about 1955 and this trend has accelerated dramatically in the past 8 years. Surprisingly, there was no recognisable response of break up day to the warm period in the 1930's and 1940's. Also, freeze up date in December–January showed only a weakly significant (P = 0.05) association with the early 20th Century warm period followed by the cool period and recent warming. Currently (2002–2006), ice freeze up date is weakly significantly later than in the earlier warm period (1942) by 12 days (P = 0.07). The duration of ice cover shows a decrease early in the 20th Century followed by a long period of relative stability and then a sudden, accelerating decrease starting around 1978–1982 that continues up to present, with maximum curvature in the early 1990's, that led to a Century-long decrease of 40 days (Figure 1j). Increases in active layer thickness in nine lowland mires with discontinuous permafrost show similar trends between 1978 and 2006 [Akerman and Johansson, 2008] and three of them show an abrupt onset of accelerating thaw depth increase in 1996 (Figure 1k). This pattern was explained by increased summer temperatures and thawing degree days from May to September. At two mires, permafrost was absent from over 80% of the sample points after 2004.
 Changes in climate and the cryosphere have had many environmental consequences. Thawing permafrost has affected hydrology. Areas of ponds have increased while the area of dry palsa tops has decreased [Malmer et al., 2005], leading to increased methane emissions [Christensen et al., 2004; Johansson et al., 2006]. The most recent precipitation extreme of 61.9 mm (20th to 21st of July 2004) caused extensive mountain debris flows and destroyed a road-bridge (road E10 at Pessisjåkka). If the extreme precipitation pattern continues, extensive environmental and infrastructure damage could occur within the next decades.
 Climate change impacts on ecology have been driven by crossing thresholds, extreme events and general warming. Thawing permafrost, related to crossing the 0°C threshold, has affected plant community distribution: wetland graminoid vegetation and tall shrubs have increased in extent [Malmer et al., 2005]. Eggs of the forest-defoliating moth caterpillars (Epirrita autumnata and Operopthera brumata) are killed below a temperature threshold of −35°C [Nilssen and Tenow, 1990]. Since cold winters have been absent since 1988, probably increased survival led to significant outbreaks in 2004 [Babst et al., 2010]. Extreme weather events in winter and their impacts are increasing. Temperatures rise above 0°C for several days, snow melts and ice layers form. Experimental simulation and a natural event over >1400 km2 showed damage to vegetation [Bokhorst et al., 2009]. General temperature increases are associated with densification of the forest [Van Bogaert et al., 2010], increase in pine radial growth [Grudd et al., 2002], upward displacement of tree-line [Van Bogaert et al., 2009, 2010] and increase in shrub abundance [Olofsson et al., 2009] (Tables 2a–2c). The abruptly increased MAT after 1989 has been correlated with an increased concentration of lake water total organic carbon (TOC) and associated changes in diatom assemblages [Rosén et al., 2009]. Increased precipitation and runoff after the late 1970s were the main drivers of increased transport of TOC to lakes close to permafrost (Tables 2a–2c) and precipitation variability was further found to control soil moisture in peat above permafrost [Kokfelt et al., 2009].
Table 2a. Early 20th Century Warm Period Climate Characteristics and Their Impacts Leading Up to and During The Period Circa 1930 –1945a
Approximate Year of Maxima
Data interpolated from B-spline and Legendre polynomial smoothing of annual and seasonal data. Statements without references are supported in the text of the paper.
Changes in diatom assemblages occurred post 1989 in a lake close to thawing permafrost [Rosén et al., 2009]
 Although 20th Century warming and cooling periods from the Abisko record are similar to European and global mean annual temperature trends [Hurrell, 1995; Jones et al., 1999], some important seasonal details (e.g., winter) differ. Also, our findings demonstrate that the recent warming period is very different in character from that in the late 1930s and early 1940's and that we could be now entering a new climate era. Already, some of the impacts caused by changes in climate, and extremes such as effects of increased rainfall on transport infrastructure and recent lack of snow on ski slopes, have required adaptations by local communities. Our long term records of local multiple environmental factors provide detailed information required to attribute particular environmental changes to various climatic drivers and to further develop specific adaptation strategies. Also, dramatic changes in an environment particularly vulnerable to climate change, for example impacts resulting from crossing the 0°C mean annual temperature threshold, highlight the importance of developing global mitigation strategies.
 We are grateful to former directors and past and current staff at the Abisko Scientific Research Station for implementing and maintaining the long term environmental monitoring programme. In particular we thank Björn Holmgren, Nils Åke Andersson and Mats Sonesson. We thank Jonas Åkerman who implemented the active layer thickness observations for making these available. This study forms part of the International Polar Year Project “Back to the Future” supported the Swedish Research Council (VR) (grant 327-2007-833) who also provided funds for infrastructure maintenance and development (grant 821-2007-3903). Authors' contributions statement: Terry Callaghan and Margareta Johansson led the drafting of the manuscript and extracted the data sets. Terry Callaghan, Christer Jonasson and Margareta Johansson were responsible for maintaining the data capture and data collation. Fredrik Bergholm statistically analyzed the long term data sets. Torben Christensen, Christer Jonasson, Fredrik Bergholm and Ulla Kokfelt contributed to formulating the structure of the study and drafting the text.