Sweden Sweden Sweden Sweden

Climate change Sweden

Air temperature changes until now

Over the period 1961-1990 the average temperature in January was 0°C in the southernmost part of the country, while the coldest valleys in northern Sweden had an average temperature of -17°C. In July the maximum average 24-hour temperature was around +17°C in South-Eastern Sweden and just over 10°C in the north (1).

Over the period 1991-2007 the average temperature was around one degree higher than in the period 1961-1990. The increase was greatest during the winter at just over two degrees in the central and northern parts of the country. The increase was least during the autumn, with an almost unchanged temperature in south-western Sweden. As a consequence of the rise in temperature the densely populated areas, including Stockholm, have seen a shift from a cold-temperate to a warm-temperate climate, which reduces the frequency of winters bringing heavy snowfalls. The winter of 2007/08 was the warmest of all winters since 1858/59 in southeastern Sweden.

They estimated that the recent increase in temperature and in precipitation with respect to 1961-1990  had both about a 6-7% chance to occur solely as a result of natural variability (2).

Part of the major climate changes observed in the Baltic Sea region during the late 20th century can be related to changes in atmospheric circulation (3). It has been suggested that the increased frequency of anti-cyclonic circulation and westerly wind types have resulted in a slightly warmer climate with reduced seasonal amplitude and reduced ice cover (3). Recent research using tree-ring based chronologies indicates that the variability of recent decades may lie within the natural range (4).

In Scandinavia, a heatwave event associated with a 100-year return period in 1981 is estimated to happen once in 20–40 years in 2022 (30).

Urban heat island

The urban heat island effect has been studied for Fennoscandia, the northern half of Norway, Sweden and Finland, and including the adjacent part of Russia. This study includes all 57 cities located above 64° N in this region. Data covering the period 2001-2017 show that the mean urban heat island intensity is found in the range 0-5°C. The intensity is larger for the largest cities of Murmansk and Oulu (3-5°C) (28).

Precipitation changes until now

Passing low-pressure systems result in fairly abundant precipitation which falls throughout the year, although most heavily in the summer and autumn. Annual precipitation is around 1000 mm. As most low-pressure systems move in across the country from the west or southwest, the western parts of the country receive the heaviest precipitation. In the mountains close to the Norwegian border there are local annual precipitation levels of 1500-2000 mm. The lowest annual precipitation is along the eastern coasts, with just under 400 mm per year (1).

Precipitation shows an increase of about 2% in global land precipitation since the beginning of the 20th century. This increase, however, is neither temporally nor spatially uniform. For the Northern Hemisphere the increasing trend is significant. The TAR concludes that the present increase in precipitation over the middle and high northern latitudes will continue at a rate between 0.5 and 1% per decade (5). According to other studies precipitation may rise by up to 2% per decade (6).

Annual precipitation over Northern Europe has increased by between 10 and 40% in the last century; the strongest increases are found in Scandinavia and Western Russia. It was found that the trend towards increasing precipitation in Northern Europe would continue at a rate of 1 to 2% per decade (5).

A recent decrease in the duration of snow cover and its water equivalent has been observed in the southern parts of all Fennoscandian countries, while the opposite trend prevails in the north. In the Scandic mountains, the increase in precipitation has overshadowed increases in temperature in the past two decades, and the snow cover has become thicker. In Finland, increasing temperatures have intensified the wintertime snow melt in the western and southern parts of the country towards the end of the period 1946–2001, in contrast to eastern and northern Finland, where the maximum snow storage has increased. A similar distribution is evident in Sweden, where there is more snow in the north and the snow cover has become thinner in the southern part of the country (7).

No statistical evidence was found that White Christmas is arriving more infrequently in Sweden; this conclusion is based on data from six locations in southern parts of Sweden during 1904–2012 (22). 

Ice cover changes rivers and lakes until now

Ice cover of lakes in southern Sweden is more sensitive to climate change than those in the north, where mean winter temperatures are below zero most of the winter. A study of 196 Swedish lakes along a latitudinal temperature gradient revealed that a 1°C air temperature increase caused an up to 35 days earlier ice break-up in Sweden's warmest southern regions with annual mean air temperatures around 7°C. It caused only about 5 days earlier break-up in Sweden's coldest northern regions where annual mean air temperatures are around − 2°C (19).

Wind climate changes until now

Over the past decades annual mean wind speed in Sweden has reduced by -0.14 m/s/decade (24). Similar results were found for Finland (25).

Air temperature changes in the 21st century

The UN’s climate panel, the IPCC, has reached the conclusion that global warming up until now has been around 0.7°C over the last 100 years. Warming has occurred almost twice as fast during the last 50 years compared with the 100-year period as a whole, and it is extremely likely that this has mostly been caused by human activity. It is very likely that, compared with 1990, the average global temperature will increase by a further 1.8–4.0°C by the end of this century (8).

Most climate change scenarios indicate that the highest and most rapid temperature increases will occur in the arctic, sub arctic, and alpine regions (9).

Temperature will rise more in Sweden and Scandinavia than the global mean. A rise in mean temperature in Sweden has been estimated of about 0.4°C per 10-year period (6).

Average temperatures increase gradually and climate zones shift north. Looking at average temperatures, by the 2020s warming will be around 2°Cin comparison with the period 1960−1990, mostly during winter, slightly less in spring and autumn and least in summer. By the 2050s, warming is put at 2−3°C, with the same seasonal distribution. By the 2080s, warming is up to around 3−5°C, mostly in the northeasterly parts of the country. In terms of temperature, Mälardalen’s climate will be similar to that of northern France today (8).

The summers will be much warmer. Depending on the chosen scenario, the average temperature in July will rise by 0.5−1.5°C by the 2020s in comparison with the period 1960−1990, by around 1.5–2.5°C by the 2050s and by 2−4°C by the 2080s (according to the RCA3-EA2 scenario). Generally, the increases are greater along the coasts, particularly around and above the Baltic. The increases are of a similar magnitude in June and August. There will be an increased number of days with extremely high temperatures. The number of tropical nights, that is to say 24-hour periods when the temperature never falls below 20 degrees, will increase substantially in southern and central parts of the country and along the Norrland coast. Towards the end of the century there may be as many such nights as there are at present in Southern Europe. There will be fewer really cold days (8).

The Ministry of the Environment of Sweden (1) reports a rise in temperature in summer-time of 2.9°C in northern Sweden and 2.8°C in southern Sweden from 1961-1990 to 2100, based on a number of scenarios.

Winters will be up to 7°C warmer by the end of the century in comparison with the period 1960−1990. By 2020, the average temperature in January will increase by between 1.5°C and 2.5°C across large parts of the country. By the 2050s, the increase is around 2.5−4°C and by the 2080s we are looking at an increase of 5–6°C in Götaland and 6–7°C in large parts of Norrland (according to the RCA3-EA2 scenario). One of the main causes of this considerable warming is a reduction in the duration and thickness of the snow cover (8).

The Ministry of the Environment of Sweden (1) reports a rise in temperature in winter-time of 5.7°C in northern Sweden and 4.4°C in southern Sweden from 1961-1990 to 2100, based on a number of scenarios.

Urban Heat Island

Urban development increases the urban heat island effect. Scenario results, focused on the summer of 2014 as an example, indicate that summer temperatures in Stockholm may increase over the new built-up areas by, on average, 0.29 °C in 2030 and 0.46 °C in 2050, up to a local maximum of 1.35 °C in the latter, as a consequence of urbanization (27).

Changes in summer and winter length in the 21st century

For northern Europe, season lengths have been quantified for the period 2040-2069, under a moderate scenario of climate change (the RCP4.5 scenario) and based on a large number of global climate models. Changes have been compared with the seasons in 1971-2000 for reference. This scenario corresponds to 2°C global warming in 2040-2069 relative to the preindustrial climate. In northern Europe, warming exceeds this global mean substantially, however (26).

According to these model projections, the summer (daily mean temperature > 10°C) will last about a month longer by mid-century in most of Northern Europe. The summer is projected to start 2 weeks earlier and last two weeks longer. In the very coldest areas, the mountains, northern Lapland and the coasts of the Arctic Ocean, summer lengthening may be even more than 30 days. Concurrently, the projections show that winter (daily mean temperature < 0°C) will become shorter by 30-60 days. Winters are projected to start 15-30 days later and end 15-30 days earlier. Changes are largest near the coasts of the Arctic Ocean and the Baltic Sea, and relatively modest in the northern inland areas. Standard deviations of the model projections are about 10 days for the lengths of the spring, summer and autumn, and 10-25 days for winter length. In Denmark and southernmost Sweden, the average winter is already quite short now and there is little room for further shortening. By mid-century, the probability of missing winters will increase considerably, particularly in southern Sweden and the Baltic countries (26).

Precipitation changes in the 21st century

Precipitation patterns will also change. Precipitation will increase in most of the country during the autumn, winter and spring. In summer-time the climate will be warmer and drier, particularly in southern Sweden. Precipitation increases most during the winter. The mean value of a number of scenarios gives an increase in precipitation from 1961-90 to 2100 of 25% in Northern Sweden and 21% in Southern Sweden (1).

As the winters become shorter, snowfall in autumn and spring months is reduced. In the middle of the winter, snowfall may increase in the coldest regions. Even in the areas where snowfall is projected to increase in the middle of the winter, the total annual snow- fall is generally projected to decrease, although the change is small in the coldest regions of northern Europe (23,29)

By the end of this century, at an altitude near the current treeline (around 800 m), the number of days with significant snow depth – at least 30 cm of snow – is projected to diminish rapidly in winter and spring, by around 25% under a moderate (RCP4.5) and by around 50% under a high-end (RCP8.5) scenario of climate change. At that time, below 500 m, days with significant snow depth occur very rarely under the moderate and become principally absent under the high-end scenario of climate change (29).

Wind climate changes in the 21st century

Mean and extreme geostrophic wind speeds in Northern Europe have been projected for the future periods 2046–2065 and 2081–2100, and compared with the baseline 1971–2000 (based on nine global climate models and the SRES A1B, A2 and B1 scenarios) (20). The geostrophic wind speed is a theoretical, calculated wind speed indicative of true surface wind speed. The results show:

  • Mean wind geostrophic speeds: During the windiest time of the year, the monthly mean wind speeds will start to increase in the Baltic Sea already in 2046–2065. In Finland, increases are largest (5–7%) in November and January by 2081–2100. In November–February 2081–2100, a positive shift of 5–10% is projected to materialize in the Baltic Sea.
  • Extreme geostrophic wind speeds: The extreme wind speeds (10-year return level estimates) will increase on average by 2–4% in the southern and eastern parts of Northern Europe, whereas a decrease of 2–6% dominates over the Norwegian Sea. These results agree with results on the future projections of 20-year return level estimates of gust winds that showed that the increase in winds is dominant in a zone stretching from northern parts of France over the Baltic Sea towards northeast (21).

Sea water temperature changes in the 21st century

As air temperatures rise, the surface temperature of the sea will also rise. Changes on an annual basis vary from scenario to scenario. In RCA3-EA2, the temperature increases by over 4°C across the whole of the Baltic Sea. In terms of how the temperature changes across the seasons, taking an average of the climate scenarios, the increase will be small in the winter in northern parts of the Gulf of Bothnia and along the coast throughout the Gulf, as ice cover is still expected by the end of the century. In the summer, however, the increases will be greatest in the Gulf of Bothnia, at more than 4°C in many places. Elsewhere, the temperature increases vary between just over 2°C and 3.5°C (8).

Uncertainties in climate projections

There is relatively widespread agreement between a large number of models that the temperature in Sweden and Scandinavia will rise by more than the global average. The precipitation patterns will also change, with increased precipitation in Scandinavia, but there is a greater degree of uncertainty here. There is also greater uncertainty about how trends relating to winds and storms will change in our part of the world.

Uncertainties in model results - GCM’s

Different GCMs often give diverging warming estimates on regional scales such as the north European and the Arctic region (10). The range of GCM estimates, however, may contain useful information about the GCMs in general since the scatter in the climatic trend estimates may be taken as a crude measure of GCM uncertainties (11). It is important to keep in mind that whereas a large scatter implies large uncertainties, a small scatter does not mean that the uncertainty is small because it is possible that all the GCMs are wrong (12).

Little confidence should be attached to a single-integration scenario, and an ensemble approach is required in order to extract the climate-change signal from the internal decadal variability (12).

It has been argued that the use of multi-model ensembles may be problematic because the GCMs do not necessarily span the full range of known climate system behaviour, and it is important to keep this in mind when interpreting the conditional probability estimates (13).

The spatial resolution of the most sophisticated global GCMs is typically around 3 × 3 deg2 and important features for local climates, such as mountain ranges and valleys, are overly smoothed (14). The Scandinavian climate is strongly influenced by the local geography, which comprises mountain ranges, fjords, valleys, forests, with different characteristic climates. The present state-of-the-art GCMs are incapable of giving a realistic description of the local climate in regions with a complex physiography (18), partly because they have too coarse a spatial resolution to adequately represent the important geographical features. Another issue is that GCMs are believed to have a lower limit on their skillful spatial scale (15).

Uncertainties in model results - Downscaling

Local climate scenarios can nevertheless be derived using (12)

  • empirical downscaling techniques where geographical information (implicitly) as well as the relationship between large-scale climatic anomalies and the local climate (explicitly) are taken into account.
  • higher-resolution models of a limited region and take the results from the global climate models as boundary conditions (dynamical downscaling) to improve the description of the regional climate.

The empirical downscaling may introduce additional uncertainties in the scenarios due to shortcomings in the analysis (16) as linear statistical downscaling models merely provide an approximate description of the relationship between the different spatial scales. It is also possible that these statistical relationships are non-stationary in time (17). It is therefore important to take into consideration not only the uncertainties associated with the GCMs but also those accompanying the downscaling. Furthermore, the trends are often not linear, and therefore a linear fit may not be the most appropriate description for the past climatic evolution (12).


The references below are cited in full in a separate map 'References'. Please click here if you are looking for the full references for Sweden.

  1. Ministry of the Environment of Sweden (2009)
  2. Räisänen and Alexandersson (2003)
  3. Omstedt et al. (2004)
  4. Esper et al. (2002), in: Omstedt et al. (2004)
  5. Eisenreich (2005)
  6. Räisänen et al. (2004), in: Eisenreich (2005)
  7. HELCOM (2007)
  8. Swedish Commission on Climate and Vulnerability(2007)
  9. McCarthy et al. (2001); Arctic Council (2004), in: Eisenreich (2005)
  10. Räisänen (2001a,b); Benestad et al. (2002b), both in: Benestad (2004)
  11. Räisänen and Palmer (2001), in: Benestad (2004)
  12. Benestad (2004)
  13. Allen et al. (2000), in: Benestad (2004)
  14. Benestad (2002b), in: Benestad (2004)
  15. Grotch and MacCracken (1991); Von Storch et al. (1993a), both in: Benestad (2004)
  16. Benestad (2001), in: Benestad (2004)
  17. Wilby (1997), in: Benestad (2004)
  18. Houghton et al. (2001), in: Benestad (2004)
  19. Weyhenmeyer (2007), in: EEA, JRC and WHO (2008)
  20. Gregow et al. (2011)
  21. Nikulin et al. (2011), in: Gregow et al. (2011)
  22. Rydén (2015)
  23. Räisänen (2016)
  24. Minola et al. (2016), in Laapas and Venäläinen (2017)
  25. Laapas and Venäläinen (2017)
  26. Ruosteenoja et al. (2020)
  27. Amorim et al. (2020)
  28. Miles and Esau (2020)
  29. Lind et al. (2023)
  30. Berghald et al. (2024)