Abstract

A Web site questionnaire survey in Finland suggested that maps illustrating projected shifts of Köppen climatic zones are an effective visualization tool for disseminating climate change information. The climate classification is based on seasonal cycles of monthly-mean temperature and precipitation, and it divides Europe and its adjacent land areas into tundra, boreal, temperate, and dry climate types. Projections of future changes in the climatic zones were composed using multimodel mean projections based on simulations performed with 19 global climate models. The projections imply that, depending on the greenhouse gas scenarios, about half or possibly even two-thirds of the study domain will be affected by shifts toward a warmer or drier climate type during this century. The projected changes within the next few decades are chiefly located near regions where shifts in the borders of the zones have already occurred during the period 1950–2006. The questionnaire survey indicated that the information regarding the shifting climatic zones as disseminated by the maps was generally interpreted correctly, with the average percentage of correct answers being 86%. Additional examples of the use of the climatic zones to communicate climate change information to the public are included.

1. Introduction

During the twentieth century, from 1901 to 2000, the average surface air temperature in European land areas increased by 0.8 ± 0.3°C (Luterbacher et al. 2004), alongside a global mean warming of 0.74° ± 0.18°C from 1906 to 2006 (Solomon et al. 2007). At the same time, the most extreme climatic zones of the earth in the widely used Köppen classification system showed statistically significant shifts: the global areas covered by tropical climate expanded, whereas the tundra regions were reduced in size (Fraedrich et al. 2001; Wang and Overland 2004). Shifts from colder to warmer climate types have occurred in Europe as well (Fraedrich et al. 2001; Beck et al. 2006; Gerstengarbe and Werner 2008). In the future, the anthropogenic global warming, with its associated changes in precipitation, is projected to move the boundaries of the climatic zones farther still (Lohmann et al. 1993; Kalvová et al. 2003; de Castro et al. 2007; Gao and Giorgi 2008; Lemke and Stein 2008).

The Köppen climate classification combines two climate parameters of high practical importance: namely, temperature and precipitation (Köppen 1936). Many effects of climate change are related to these two variables either directly or indirectly, through evaporation, soil moisture, and runoff. Changes in their average values, variability, and extremes have multiple impacts on the environment, society, and human life. Many of the impacts, although not all, will be adverse regionally or even globally. With the aid of the classification, one can assess how the projected changes in monthly-mean temperature and precipitation might alter the spatial distribution of climate zones and thereby potentially influence terrestrial ecosystems and climate-sensitive sectors of society.

A major challenge for raising awareness of climate change and motivating active mitigation and adaptation measures is the dissemination of scientific information in terms that are understandable to the layman: that is, anyone without scientific training, including many decision makers. Planning of cost-effective measures to mitigate climate change and to adapt to its impacts requires fluent communication between scientists and decision makers. For the climate policy measures to be implemented successfully, it is also essential to have the support of public opinion. Social and behavioral factors are likely to have an increasingly important role in activating people to favor climate policy actions and to consider climate protection in their daily decisions (e.g., Staats et al. 1996; Retallack et al. 2007; Moser 2008). In parallel, information from the physical sciences about climate change is essential and should be given in a way that is both compact and easily absorbable.

The potential of visualization in communicating complex issues such as climate change has been examined, for example, by Nicholson-Cole (2005). According to her, taking a visual approach has many advantages, but there are also critical issues that one needs to consider. Visualizations can put across strong messages in a way that makes them easy to remember, condense complicated information, and communicate content that is not necessarily familiar to the viewer. Illustrations also provide a foundation for people’s personal reflections and observations, making people more aware of the issues at hand and helping them to remember them. However, people do not respond to new visual material in a purely rational way but also depending on their prior expectations and attitudes. To provide a meaningful and motivating message with the aid of images, one has to be careful not to awaken disturbing or even misleading conceptions.

An introduction to world climatic zones is included in the syllabus for geography in basic and secondary education in many countries. Besides, everyone has a sense of the climate type of the region where they live or have visited. This suggests that maps illustrating the projected future shifts of the climatic zones might help to make the regional impacts of global climate change more easily imaginable. However, if the maps are interpreted incorrectly, the attempt to strengthen people’s knowledge about global change by this means is likely to fail.

The objectives of this study are twofold: first, to update the projections of climatic zones in Europe, and second, to consider their use mainly from the point of view of visualizing climate change information. We employ a recently developed high-resolution observational dataset for Europe (Haylock et al. 2008) and a large suite of simulations by global climate models (GCMs) (Meehl et al. 2007) that were utilized to prepare the Fourth Assessment Report of the Intergovernmental Panel on Climate Change (IPCC; Solomon et al. 2007). As to the second objective, we have conducted a questionnaire survey about the visualization of climate change with the aid of maps showing the shifting climatic zones in Europe. The aim of the survey was not to test whether the maps can serve as an efficient tool in motivating and activating people to favor climate protection. Rather, the questions and statements dealt with the comprehensibility of the maps.

In the next section, we describe the Köppen climate classification system and document the methods and materials used to construct the future projections. Our findings for recent past and anticipated future changes in the climatic zones are then presented and briefly compared with previous studies. Regarding interpretation of the maps, we introduce the questionnaire and show the main results. We also give some examples of how a previous set of animations of migrating Köppen zones—first introduced in Heureka, the Finnish Science Centre, in 2003—has subsequently been employed in climate change dissemination.

2. Methods and material

a. Köppen climate classification

The Köppen climate classification, with its revisions, is the most widely used classification of climates in modern atlases and geography textbooks (e.g., Lewis and Geelan 1998). Global maps of the classification, which were recently updated by Kottek et al. (2006) and Peel et al. (2007), were originally developed during the period 1900–36 by Wladimir Köppen (1846–1940), a German scientist. In the Köppen system, there are five principal categories (A–E), which were originally selected on the basis of vegetation boundaries (Sanderson 1999). The classification depends on annual variations in monthly-mean temperature and precipitation.

Four of the principal classes—tropical climates (A), temperate humid climates (C), boreal climates (D), and polar climates (E)—are defined by the mean surface air temperature of the warmest and coldest months of the year. For A, the coldest month is warmer than 18°C; for E, the warmest month is colder than 10°C. The remaining principal classes, C and D, both have their highest monthly-mean temperature above 10°C, but they are differentiated based on the mean temperature of the coldest calendar month (between −3° and 18°C or below −3°C, respectively).

The criteria for a dry climate (B) are more complicated. The limit between dry (B) and humid (A, C, and D) climates is defined by empirical relations that aim to approximate the effect of evaporation. The relations compare the average annual precipitation to the average annual temperature, considering the seasonality of precipitation (e.g., Conrad and Pollak 1950; Critchfield 1966). If there is a rain season in summer, the annual precipitation limit between the B category and the humid climates is higher than when the precipitation season occurs in winter or when precipitation is distributed evenly around the year.

The second and third levels of the classification generally describe the annual cycles of precipitation and temperature. Combinations of the first and second levels are referred to here as the main climatic classes or types. They are further divided into subtypes by the third level of the classification. Table 1 indicates the climatic types that prevail in our study domain covering most of Europe, Asia Minor, and northernmost Africa (35.125°–74.375°N, 27.125°W–64.875°E, excluding a corner in the southeast). The exact procedure to determine the types is shown in the appendix.

Table 1.

The Köppen climate classification (Critchfield 1966). Only those classes found in the domain discussed in the present study are shown. The first level of the classification system, indicating the principal categories, is shown by capital letters on the left. The second and third levels of the classification are denoted by the indented letters.

The Köppen climate classification (Critchfield 1966). Only those classes found in the domain discussed in the present study are shown. The first level of the classification system, indicating the principal categories, is shown by capital letters on the left. The second and third levels of the classification are denoted by the indented letters.
The Köppen climate classification (Critchfield 1966). Only those classes found in the domain discussed in the present study are shown. The first level of the classification system, indicating the principal categories, is shown by capital letters on the left. The second and third levels of the classification are denoted by the indented letters.

The aim of the Köppen system is to represent areas having similar climatic conditions. Like any classification based on fixed thresholds, it simplifies the actual spatial variability of climate. The merits of the Köppen classification, however, outweigh its deficiencies; this is proved by its wide acceptance (Stern et al. 2000). Although far less sophisticated than dynamic vegetation models, it can be regarded as a nonlinear filter on the temperature and precipitation data that produces a first-order picture of biome distributions (Gnanadesikan and Stouffer 2006). In the context of climate models, the system or its modifications [e.g., the Köppen–Trewartha (K-T) classification] have been utilized, for example, by Manabe and Holloway (1975), Hansen et al. (1984), Lohmann et al. (1993), Kalvová et al. (2003), AchutaRao et al. (2004), Gnanadesikan and Stouffer (2006), de Castro et al. (2007), Gao and Giorgi (2008), and Lemke and Stein (2008) either to explore how closely they can reproduce the current climate or to construct scenarios for future climatic conditions, or both. Furthermore, Guetter and Kutzbach (1990) used a modified Köppen climate classification to interpret GCM simulations of glacial and interglacial climates, whereas, in a GCM-based sensitivity study about the influence of vegetation on climate, Kleidon et al. (2000) illustrated the results of their model estimates with the aid of the Köppen system.

Although we focus here on the Köppen system, it should be mentioned that other classifications for climate, vegetation, and/or terrestrial ecosystems have also been developed. In the context of climate change, the Holdridge life zones classification, for example, has recently been applied, for example, by Yates et al. (2000) and Yue et al. (2006), and a modified Thornthwaite climate classification scheme has been used by Grundstein (2009). Regarding the second topic of the current work (i.e., dissemination of climate change information), these classifications (or more sophisticated vegetation models) could be utilized as well.

b. Emission scenarios

Projections of future climatic zones are based, from a socioeconomic and technological point of view, on alternative storylines of the future world, as defined in the IPCC Special Report on Emissions Scenarios (SRES; Nakićenović and Swart 2000). Each storyline is quantified by a number of scenarios of greenhouse gas (GHG) and aerosol emissions in the future. In this paper, we focus on three SRES illustrative scenarios: A1B, A2, and B1. These cover a reasonably large range of the full set of SRES scenarios, and most climate model experiments nowadays apply at least one of these scenarios.

In the B1 scenario, emissions of carbon dioxide (CO2) and other GHGs increase only slowly up to the 2040s and then start to reduce in time, resulting in a not larger than 1.5-fold concentration of CO2 in the atmosphere and a threefold total radiative forcing in 2100 compared to those in 2000 (Fig. 1). Hence, B1 may be used as a surrogate for mitigation scenarios. In the A2 scenario, GHG emissions increase continuously, producing as large as 2.3-fold CO2 concentrations and a sixfold radiative forcing in 2100 relative to 2000. The A1B scenario first exceeds A2 but soon falls to between B1 and A2, with 1.9-fold concentrations and 4.5-fold forcing by the end of this century. The best estimates (with likely ranges in parenthesis) of global mean warming by 2090–99 under the B1, A1B, and A2 scenarios are 1.8°C (1.1°–2.9°C), 2.8°C (1.7°–4.4°C), and 3.4°C (2.0°–5.4°C), respectively, relative to 1980–99 (Solomon et al. 2007).

Fig. 1.

The three illustrative SRES scenarios for CO2 applied in this study: A1B (dashed line), A2 (thin solid line), and B1 (dotted line). The scenarios represent the outcome of different assumptions about the future course of economic development, demography, and technological change. Shown are (a) the anthropogenic emissions from fossil fuel and industrial processes, deforestation and land use and (b) the resulting CO2 concentrations as projected by a fast carbon cycle model [Bern Carbon Cycle (Bern-CC)]. For more information on the SRES scenarios, see Nakićenović and Swart (2000) and Houghton et al. (2001). Values prior to the year 2001 (thick solid line) are taken from observations provided by the University of East Anglia [for emissions in (a); available online at http://lgmacweb.env.uea.ac.uk/lequere/co2/carbon_budget.htm; see Le Quéré et al. (2009) for more details] and by NOAA/Earth System Research Laboratory [ESRL; for atmospheric concentrations in (b); available online at ftp://ftp.cmdl.noaa.gov/ccg/co2/trends/co2_annmean_gl.txt; for more details, see Conway et al. 1994].

Fig. 1.

The three illustrative SRES scenarios for CO2 applied in this study: A1B (dashed line), A2 (thin solid line), and B1 (dotted line). The scenarios represent the outcome of different assumptions about the future course of economic development, demography, and technological change. Shown are (a) the anthropogenic emissions from fossil fuel and industrial processes, deforestation and land use and (b) the resulting CO2 concentrations as projected by a fast carbon cycle model [Bern Carbon Cycle (Bern-CC)]. For more information on the SRES scenarios, see Nakićenović and Swart (2000) and Houghton et al. (2001). Values prior to the year 2001 (thick solid line) are taken from observations provided by the University of East Anglia [for emissions in (a); available online at http://lgmacweb.env.uea.ac.uk/lequere/co2/carbon_budget.htm; see Le Quéré et al. (2009) for more details] and by NOAA/Earth System Research Laboratory [ESRL; for atmospheric concentrations in (b); available online at ftp://ftp.cmdl.noaa.gov/ccg/co2/trends/co2_annmean_gl.txt; for more details, see Conway et al. 1994].

c. Observational and simulated climate datasets

For specifying the Köppen climatic zones, the spatial distributions of monthly-mean air temperature and precipitation first have to be defined. To smooth out year-to-year variations, we computed floating averages over periods of three decades for each calendar month and then applied the methodology presented in the appendix to perform the classification.

Recent shifts in climatic zones were inferred from a European land-only high-resolution gridded dataset (version 1.1) of daily surface temperature and precipitation for the period 1950–2006 (Haylock et al. 2008). The dataset has a spatial resolution of about 25 km (0.25° by 0.25° in latitude and longitude). The number of observing stations employed to create the dataset varied in time between about 1000 and 2000 for precipitation and between about 500 and 1000 for temperature. Fewer stations were available in the 1950s and in the 2000s than during the middle decades of the period (for details, see Haylock et al. 2008). According to the data, the wettest calendar month varied spatially quite strongly (e.g., in France and the British Isles), as did the driest month as well (Fig. 2). Changes in the climate regions were examined both with the aid of a time series based on 30-yr floating averages as well as by dividing the observational period into two parts, 1950–78 and 1979–2006. The period 1971–2000 (referred to here as the 1980s) was selected to represent a baseline climatology on which to compose future projections; this climatology was modified on the grounds of climate model projections (the so-called delta-change method).

Fig. 2.

Calendar month of the (a) smallest and (b) largest average monthly precipitation sum during 1971–2000 on the basis of observations (data source: Haylock et al. 2008).

Fig. 2.

Calendar month of the (a) smallest and (b) largest average monthly precipitation sum during 1971–2000 on the basis of observations (data source: Haylock et al. 2008).

Estimates of future changes in temperature and precipitation, separately for each emission scenario, were calculated as multimodel means based on the output data from 19 global climate models (GCMs). The models were the following (for more information, see Solomon et al. 2007, 597–599):

  • Bjerknes Centre for Climate Research Bergen Climate Model version 2.0 (BCCR-BCM2.0);

  • Canadian Centre for Climate Modelling and Analysis Coupled Global Climate Model version 3.1 at T47 resolution (CGCM3.1-T47); at T63 resolution (CGCM3.1-T63);

  • Centre National de Recherches Meteorologiques Coupled Global Climate Model version 3 (CNRM-CM3);

  • Commonwealth Scientific and Industrial Research Organisation Mark version 3.0 (CSIRO-Mk3.0);

  • Geophysical Fluid Dynamics Laboratory Climate Model version 2.0 (GFDL-CM2.0); version 2.1 GFDL-CM2.1;

  • Goddard Institute for Space Studies Model E-R (GISS-ER);

  • Institute of Numerical Mathematics Coupled Model, version 3.0 (INM-CM3.0);

  • Institut Pierre Simon Laplace Coupled Model, version 4 (IPSL-CM4);

  • Model for Interdisciplinary Research on Climate 3.2, high-resolution version [MIROC3.2(hires)]; medium-resolution version [MIROC3.2(medres)];

  • Meteorological Institute of the University of Bonn, Institute of KMA, and Model and Data Group coupled model (ECHO-G);

  • Max Planck Institute for Meteorology coupled model (ECHAM5/MPI-OM);

  • Meteorological Research Institute Coupled General Circulation Model, version 2.3.2 (MRI CGCM2.3.2);

  • National Center for Atmospheric Research (NCAR) Community Climate System Model, version 3 (CCSM3);

  • NCAR Parallel Climate Model (PCM);

  • third climate configuration of the Met Office (UKMO) Unified Model (HadCM3); and

  • Hadley Centre Global Environmental Model version 1 (HadGEM1).

The model data were taken from the World Climate Research Programme (WCRP) Coupled Model Intercomparison Project phase 3 (CMIP3) multimodel dataset (Meehl et al. 2007). Of the 23 models in the CMIP3 archive, 4 were discarded because they did not fulfill the following criteria: the land–sea distribution is realistically described over the European and North Atlantic region; the model performance in representing the current climate is satisfactory; and useable model data are available for at least two of the three SRES scenarios. In the absence of simulations by a GCM for one of the scenarios, a pattern-scaling method (Ruosteenoja et al. 2007) was applied to approximate the missing simulation.

To apply the delta-change method, the multimodel mean changes were interpolated onto the same 0.25° × 0.25° grid as used for the observed climatology. The projected increases of monthly-mean temperatures were added to the observed baseline period climatological monthly means. For precipitation, the observed baseline period climatological means were multiplied by the ratios between the simulated precipitation amounts for the future period and those for the baseline period. The purpose of this procedure is to minimize the influence of biases that commonly exist in model simulations. Precipitation changes are treated in relative rather than absolute terms to avoid negative values of future precipitation in regions that have a significant projected reduction in precipitation and a negative bias in the modeled baseline period precipitation compared to observations (for further discussion, see, e.g., Graham et al. 2007).

The animations of future climatic zones were constructed based on projected 30-yr running averages of temperature and precipitation. The results are illustrated by maps of Köppen zones for three discrete 30-yr periods: 2010–39 (referred to as the 2020s), 2040–69 (referred to as the 2050s), and 2070–99 (referred to as the 2080s). The changes in the spatial coverage of the climate regions (in km2) were derived from the continuous time series.

d. Survey design and methods

The questionnaire about the visualization of climate change aimed to get empirical data about how easy or difficult it is to correctly interpret the maps of Köppen climatic zones in Europe. The participants in the survey were shown the maps for the 1980s, 2050s, and 2080s. A total of 13 checkbox questions were provided; a respondent could answer these as true or false or could skip to the next. An opinion question about the climatic zones as a tool to visualize climate change was asked twice; the first time was after four basic questions, to which the respondent replied before the color shading in the maps was explained, and the second time was after more detailed questions, for which the respondents could see brief definitions of the colored zones. The participants were asked to give answers to the questions and the statements based on the maps only, utilizing their prior knowledge of the current climate but independently of what they may have previously learned about climate change. An attempt was made to reduce the influence of personal attitudes to climate change on the responses, as the respondents were told that the purpose of the questionnaire was to evaluate the maps from the point of view of comprehensibility and that in the future the climate may in fact evolve in a somewhat different way than depicted in the charts. There were also a few open questions about opinions concerning the maps and the questionnaire itself. In the planning phase of the survey, the Swedish School of Social Science of the University of Helsinki (UH), the Communication Research Centre within the Department of Communication at the UH, the Media Laboratory of the University of Art and Design Helsinki, and the communications unit of the Finnish Meteorological Institute (FMI) were consulted.

The questionnaire was carried out online on the Internet, and the target group consisted of visitors to the home page of the FMI (available online at http://www.fmi.fi). A short announcement of the questionnaire was made on the home page, and then the aims of the online questionnaire were described on a separate Web page, where a link to the questionnaire itself was given. Responses to the questionnaire were gathered for 19 days, with the total number of responses received being 404. On average, 70 000 persons visited the FMI home page daily during the time span of the online survey. Hence, the sample represents only a tiny fraction of all visitors and an even tinier fraction of Finland’s total inhabitants (5.3 million). Although no information about the ongoing survey was given, for example, through any kinds of scientific networks in Finland, some attempts were made to increase the proportion of women, elder, and/or less educated people among those replying.

The age of the respondents ranged from less than 14 yr to more than 65 yr, but the main age classes were 25–44 (46%) and 45–64 (38%) yr. The percentage of women was 35%. About half (51%) of the viewers had tertiary education, whereas 18% of them had primary or lower vocational education. In the analysis of the responses, the potential impacts of education were briefly considered. Another grouping was made based on the responses to two arguments that were presented near the end of the questionnaire. The first stated that the respondent had answered the questions purely on the basis to the maps, and the second argued that he/she could have answered the questions even without the maps; the scale ranged from 1 (fully disagree) to 5 (fully agree).

3. Climatic zones in Europe

a. Recent climatic zones

Apart from the tropical humid zone (A), all the principal climatic zones exist in Europe, including the Mediterranean, with the areal coverage of the boreal (D) and temperate (C) climates prominently exceeding those of the dry (B) and polar (E) climates (Fig. 3a). The total number of main climate classes is seven, the dominant ones being Df, Cf, and Cs (Table 2). The temperate humid subtype Cfb is most abundant in western and central Europe, with mild winters, at least four months warmer than 10°C, and precipitation in all seasons. A climate subtype with approximately similar summers but colder winters, with the average temperature of the coldest month below −3°C, prevails in eastern Europe (Dfb). Cold winters and short summers, with at most three months warmer than 10°C, constitute the dominant climate type (Dfc) north of 60°N. Mediterranean Europe is mainly characterized by four types of temperate climates, with wetter (Cf) or drier (Cs) and warm (b) or hot (a) summers. In southeastern parts of the Iberian Peninsula, as well as locally on the coasts of the Aegean Sea and the Black Sea, the criteria for a dry semiarid climate (BS) are met. Areas assigned to a tundra climate type (ET), with the mean temperature of the warmest month below 10°C but above 0°C, can be found in the Scandinavian mountains and the Alps.

Fig. 3.

Köppen climate zones in Europe deduced (a) from observations for the period 1971–2000 and from future projections for the periods (b) 2010–39, (c) 2040–69, and (d) 2070–99, assuming the A1B scenario. The projections are based on the delta-change method with the mean estimates of future changes given by 19 models. For the acronyms of the classes, see the legend below each map. More descriptive explanations, analogous to those presented in Finnish in the Web site questionnaire, are also shown for the most common classes. White areas inside the study domain indicate sea areas and areas with missing data.

Fig. 3.

Köppen climate zones in Europe deduced (a) from observations for the period 1971–2000 and from future projections for the periods (b) 2010–39, (c) 2040–69, and (d) 2070–99, assuming the A1B scenario. The projections are based on the delta-change method with the mean estimates of future changes given by 19 models. For the acronyms of the classes, see the legend below each map. More descriptive explanations, analogous to those presented in Finnish in the Web site questionnaire, are also shown for the most common classes. White areas inside the study domain indicate sea areas and areas with missing data.

Table 2.

Spatial coverage of Köppen classes and their trends in Europe, Asia Minor, and northernmost Africa. Column 1: main classes; column 2: average area based on observations during 1971–2000 (103 km2); column 3: observed difference between the periods 1979–2006 and 1950–78 [103 km2 (10 yr)−1]; column 4: as in column 3, but as a percentage; columns 5–7: projected differences between the periods 2010–39 (2020s) and 1971–2000 (1980s) [103 km2 (10 yr)−1] for the SRES B1, A1B, and A2 scenarios; columns 8–10: as in columns 5–7, but between the periods 2040–69 (2050s) and 2010–39; columns 11–13: as in columns 5–7, but between the periods 2070–99 (2080s) and 2040–69.

Spatial coverage of Köppen classes and their trends in Europe, Asia Minor, and northernmost Africa. Column 1: main classes; column 2: average area based on observations during 1971–2000 (103 km2); column 3: observed difference between the periods 1979–2006 and 1950–78 [103 km2 (10 yr)−1]; column 4: as in column 3, but as a percentage; columns 5–7: projected differences between the periods 2010–39 (2020s) and 1971–2000 (1980s) [103 km2 (10 yr)−1] for the SRES B1, A1B, and A2 scenarios; columns 8–10: as in columns 5–7, but between the periods 2040–69 (2050s) and 2010–39; columns 11–13: as in columns 5–7, but between the periods 2070–99 (2080s) and 2040–69.
Spatial coverage of Köppen classes and their trends in Europe, Asia Minor, and northernmost Africa. Column 1: main classes; column 2: average area based on observations during 1971–2000 (103 km2); column 3: observed difference between the periods 1979–2006 and 1950–78 [103 km2 (10 yr)−1]; column 4: as in column 3, but as a percentage; columns 5–7: projected differences between the periods 2010–39 (2020s) and 1971–2000 (1980s) [103 km2 (10 yr)−1] for the SRES B1, A1B, and A2 scenarios; columns 8–10: as in columns 5–7, but between the periods 2040–69 (2050s) and 2010–39; columns 11–13: as in columns 5–7, but between the periods 2070–99 (2080s) and 2040–69.

Although these basic features of the European climate have remained unaltered, the observational dataset reveals noticeable shifts in climatic zone borders between the periods 1950–78 and 1979–2006 (Fig. 4a); 12.1% of the land areas (or 1.17 × 106 km2, corresponding to an area twice as large as the French mainland) was affected by shifts toward warmer or drier and 2.2% (0.22 × 106 km2) toward cooler or wetter climate types. Most markedly, the boundary between the temperate Cf and boreal Df zones has moved to the east in Poland and to the north in southern Sweden. In a sector extending from Romania via Moldova and the Ukraine to southwestern Russia, Cf has replaced Df there as well. The total area occupied by Df has decreased by 138 × 103 km2 decade−1 (comparable to an area of almost half of Poland), whereas areas assigned to Cf have increased by 175 × 103 km2 decade−1 (see Table 2 and the left-hand side of Fig. 5). On the other hand, the temperate climate zone with dry summers (Cs) has been reduced by 65 × 103 km2 decade−1. This reduction mainly took place in Spain, where Cs was replaced by a wetter climate type Cf at places in the north and by a drier type BS in the middle of the country. Altogether, these shifts from one main climate class to another (the first and second levels of the Köppen classification) took place in 10.3% of the study domain between the former and latter halves of the observational period 1950–2006.

Fig. 4.

Regions exposed to shifts in climatic zones (a) from the period 1950–78 to the period 1979–2006 based on observations and (b) from the period 1971–2000 to the period 2070–99 under the A1B scenario. The colors labeled 1–4 indicate shifts toward warmer and/or drier climatic zones, as shown: 4: toward drier, a shift in principal category to B (from D or C); 3: toward warmer, a shift in principal category to C (from D or E) or to D (from E); 2: toward drier, a shift in the second level to s (from f) or to BW (from BS); 1: toward warmer, a shift in the third level to a (from b or c), to b (from c), or to h (from k); 0: unaltered; and −1: toward cooler (but not drier) or wetter (but not warmer). Note that the changes labeled 2–4 may be accompanied with any changes in the lower levels of the classification.

Fig. 4.

Regions exposed to shifts in climatic zones (a) from the period 1950–78 to the period 1979–2006 based on observations and (b) from the period 1971–2000 to the period 2070–99 under the A1B scenario. The colors labeled 1–4 indicate shifts toward warmer and/or drier climatic zones, as shown: 4: toward drier, a shift in principal category to B (from D or C); 3: toward warmer, a shift in principal category to C (from D or E) or to D (from E); 2: toward drier, a shift in the second level to s (from f) or to BW (from BS); 1: toward warmer, a shift in the third level to a (from b or c), to b (from c), or to h (from k); 0: unaltered; and −1: toward cooler (but not drier) or wetter (but not warmer). Note that the changes labeled 2–4 may be accompanied with any changes in the lower levels of the classification.

Fig. 5.

Evolution in time of the spatial coverage of the main climatic zones in Europe, Asia Minor, and northernmost Africa (a) for the Df, Cf, Cs, and BS types and (b) for the ET, Ds, Cs, BS, and BW types. Year-to-year variations are smoothed by considering 30-yr central moving averages of monthly-mean temperature and precipitation. Because of this 30-yr window, the values based on observations in 1950–2006 are truncated into the period 1964–91. The projections from 1985 onward are based on a delta method in which the baseline climate in 1971–2000 has been modified using the simulated temporal multimodel mean change. The shading indicates uncertainty ranges across the A1B, A2, and B1 scenarios. Note that the Cs and BS types in (a),(b) are shown in both and that the vertical scales of the axes are different.

Fig. 5.

Evolution in time of the spatial coverage of the main climatic zones in Europe, Asia Minor, and northernmost Africa (a) for the Df, Cf, Cs, and BS types and (b) for the ET, Ds, Cs, BS, and BW types. Year-to-year variations are smoothed by considering 30-yr central moving averages of monthly-mean temperature and precipitation. Because of this 30-yr window, the values based on observations in 1950–2006 are truncated into the period 1964–91. The projections from 1985 onward are based on a delta method in which the baseline climate in 1971–2000 has been modified using the simulated temporal multimodel mean change. The shading indicates uncertainty ranges across the A1B, A2, and B1 scenarios. Note that the Cs and BS types in (a),(b) are shown in both and that the vertical scales of the axes are different.

Among the principal categories (the first level of the classification), the dry climate zone B and the polar climate zone E were those with the largest percentage changes. The B category expanded by more than half (Table 2), mainly because of the enlargement of dry semiarid regions (BS) but also because there were increases in areas allocated to the dry arid type (BW), a very rare type in our study region. The E zone shrank by about 30% as the Df climate took over tundra (ET) regions in the Scandinavian and Alpine mountains. The C category increased by 9%, whereas the D category decreased by 8%.

Regarding shifts between the climate subtypes alone (the third level), the Dfb type penetrated slightly northward in Russia and southern Finland at the expense of the Dfc type with short and cool summers. Changes in the third level alone also occurred in the Mediterranean area (Fig. 4a). Integrated over the study domain, the fraction of areas indicating changes solely in third level subdivisions was 4%.

Two likely contributors to the detected shifts of the climate zones in Europe are human-induced warming and fluctuations of the North Atlantic Oscillation (NAO), the latter possibly also having an anthropogenic component (Solomon et al. 2007, 709–710). The NAO index, illustrating the strength of the westerly flow across the Atlantic Ocean into Europe, was considerably higher during the period 1979–2006 than during the period 1950–78: the average winter value of the index increased from −0.30 to 0.38 and the average annual value increased from −0.11 to 0.11 (based on data provided by the NOAA/National Weather Service). As shown by Gerstengarbe and Werner (2008), the NAO index and the area changes of almost all climate zones in Europe in 1901–2003 were strongly correlated. The correlation was especially clear for the winter value of the NAO index; however, for the dry climate type, correlation with the annual NAO index was even higher.

b. Projected future changes

On the basis of the multimodel mean climate projections, notable shifts toward warmer and/or drier climatic zones will occur in Europe during the ongoing century as a result of increasing GHG concentrations. This is illustrated by maps of Köppen zones for the 2020s, 2050s, and 2080s (Figs. 3b–d); by an anomaly map for the 2080s relative to the baseline period 1971–2000 (Fig. 4b); and by time series of the spatial coverage of the climate regions (Fig. 5). The main geographical features of the projected changes are as follows:

  • The tundra climate (ET) will contract in the Scandinavian mountains and will disappear in the Alps, at least on the large scale;

  • The dry semiarid zone (BS) will expand in the Iberian Peninsula, in Italy, on the western and northern coasts of the Aegean Sea and the Black Sea, and in the vicinity of the Caspian Sea;

  • The temperate rainy zone (Cf) will penetrate northeastward in a wide sector between the surroundings of the Black Sea and the Baltic Sea, at the expense of the boreal rainy climate (Df);

  • The temperate rainy climate (Cf) will be replaced by the temperate dry-summer climate (Cs) in western France;

  • Although the temperate rainy climate will remain elsewhere in France and the Balkan Peninsula, summers will be hotter there (a shift from Cfb to Cfa);

  • In areas of Russia and Fennoscandia where the boreal rainy climate (Df) will still prevail, the zones of warm and hot summers (Dfb and Dfa) will extend northward.

The projected changes in the climatic zones, in terms of relocation and coverage, are in accord with the observed trends during the past 50 yr. During the next few decades, regions assigned to either the tundra (ET) or boreal (Df and Ds) climates are projected to shrink at a rate of about 180 × 103 km2 decade−1 (analogous to an area of half of Germany), whereas the temperate (Cf and Cs) and dry climate (BW and BS) zones will expand at rates of about 140 × 103 and 40 × 103 km2 decade−1, respectively (Table 2). The most extensive main classes in our study area remain Df and Cf, but their mutual rank is likely to be switched in midcentury (Fig. 5a). In the 2050s, the C category is projected to have a 21%–28% larger extent and the D category is projected to have a 21%–30% smaller extent than in the 1980s, depending on the emission scenarios. The projected alterations to the E and B categories by the 2050s are even larger: a decrease of 47%–54% and an increase of 88%–150%, respectively.

Future GHG emissions strongly influence the climatic shifts during the latter half of the century. If the A2 scenario materializes, the retreat rate of the Df zone will be doubled by the end of this century (Table 2). Concurrently, the dry BS climate zone will become as extensive as or even more extensive than the Cs zone (Fig. 5b). The movement of the other climate zones’ boundaries would likewise be faster than in the case of the A1B scenario. Relocation of the zones under the B1 scenario is also pronounced but takes place more slowly. The projected pattern of the climatic zones in the 2080s under the B1 scenario is very close to that in the 2050s under the A1B scenario. This indicates that reductions in GHG emissions would serve to delay shifts in the climate zones.

The projected total extent of areas subject to shifts from one main climate class to another amounts to 3.2 × 106 km2 (or 33% of the study domain, corresponding to more than 6 times the area of Spain) by the 2080s for the A1B scenario. For the B1 and A2 scenarios, the corresponding figures are 2.1 × 106 km2 (22%) and 3.8 × 106 km2 (39%), respectively. Besides these major alterations, changes in the climatic subtypes alone (the third level) will likewise be common; corresponding to the B1, A1B, and A2 scenarios, respectively, they will take place over areas of 2.3 × 106, 2.8 × 106, and 3.1 × 106 km2 (24%, 29%, and 32%), mainly because of increases in summer temperatures. Altogether, shifts toward a warmer (but not drier) first or third level of the classification are projected to occur in 40%, 52%, and 58% of the study domain, shifts toward a drier first or second level take place in 6%, 11%, and 13% of the domain, depending on the scenario (see Fig. 4).

Based on these projections, about half or even two-thirds of Europe will be affected by some reallocations of the climate type. It is worth keeping in mind, however, that the Köppen classification is based on fixed thresholds. Hence, the areas in Fig. 4 not experiencing movements of the borderlines between the climate types should not be interpreted as regions not exposed to climate changes at all. Other issues to be mentioned here are natural climate variability and intermodel differences; we smoothed them by considering 30-yr averages of temperature and precipitation, as we also did for the multimodel mean changes in them.

Somewhat unexpectedly, Fig. 3d for the A1B scenario suggests a small area of BSk climate emerging in Poland by the 2080s. The area is a slightly larger in the A2 scenario but missing in the B1 scenario. A somewhat wider region of BSk climate appears to emerge in Hungary, possibly related to the Puszta, a steppe landscape in that country. Based on the observational data, these two areas show local minima in present-day annual-mean precipitation sums. However, projections performed with high-resolution regional climate models (RCMs) might be needed to assess the reliability of these details.

c. Comparisons with previous studies

1) Observational studies

There are a number of recent studies about present-day climatic zones in Europe or globally (e.g., Fraedrich et al. 2001; Kalvová et al. 2003; Wang and Overland 2004; Kottek et al. 2006; Beck et al. 2006; de Castro et al. 2007; Peel et al. 2007; Gerstengarbe and Werner 2008). To the knowledge of the authors, however, the most up-to-date European-scale observational dataset on a 25-km grid interval by Haylock et al. (2008), used here, has not been previously adopted for the purpose of climate classification. This dataset improves on previous gridded products in its spatial resolution and extent, time period, and number of contributing stations.

A comparison of our results with previous studies indicated that the general current pattern of climatic zones in Europe is rather congruently captured, but one could also find deviations. These deviations were mostly located near regions experiencing climatic shifts in the late twentieth century. They can mainly be explained by differences in the observational datasets and time spans considered. Expectedly, the improved spatial resolution and increased number of contributing stations in the observational data added details in our results. The outcomes in dry areas that are getting even drier were particularly affected. Slight differences between the studies also appeared in the Alpine and Scandinavian mountains and in a scattered way along the border sector between the C and D climates in central Europe. Comparisons of the results were to some extent complicated by the fact that, instead of the Köppen system, some studies applied the K-T classification with slightly different classes. Besides, the criterion for the boundary between the Köppen C and D climates (or between the K-T Do and Dc classes; i.e., the mean temperature of the coldest month) appeared to be somewhat variable: in a few studies it was raised from −3° to 0°C.

Our results for recent past changes in the climate classes were largely consistent with previous studies by Fraedrich et al. (2001), Beck et al. (2006), and Gerstengarbe and Werner (2008). According to Fraedrich et al. (2001), climate shifts occurred in 8.4% (1.179 × 106 km2) of European land areas during 1981–95, compared to 1901–95. Based on Beck et al. (2006), up to 10% of the European land area experienced shifts of the borderline between the C and D climates during the period 1950–2000. Gerstengarbe and Werner (2008) stated that, between the periods 1901–15 and 1989–2003, changes from colder to warmer climate types took place in areas with a total size of 1.061 × 106 km2 (14.6%). Moreover, dry areas had a tendency to become even drier. Taking into account differences in the study domains and observational periods, our results, with 12.1% (1.17 × 106 km2) of the study domain having experienced shifts toward warmer or drier between the periods 1950–78 and 1979–2006, are broadly in line with the previous studies.

2) Modeling studies

Applying the Köppen climate classification to the outputs of four GCMs, Kalvová et al. (2003) examined shifts in the climate zones in greenhouse gas warming simulations between the periods 1961–90 and 2036–65. In continental areas, increases were found for the tropical A and dry B climates and decreases were found for the boreal D and polar E climates, but the changes in the coverage did not exceed 5% for the land areas of the whole globe. Although the models disagreed on the sign of the changes for the temperate C climates, a poleward shift of the boundaries of C in both hemispheres was the dominant feature revealed in all simulations.

A remarkable alteration in the distribution of climate types in Europe from the period 1961–90 to the period 2071–2100 was reported by de Castro et al. (2007) and Gao and Giorgi (2008). In the former study, the K-T classification was applied to outputs from experiments performed with nine RCMs nested with a common GCM under the A2 emission scenario. The percentage of land grid points (50-km grid interval) with a shift from the current climate zone toward a warmer or a drier one in 2071–2100 ranged from 51% to 59% across the RCMs. In Gao and Giorgi (2008), the K-T classification was used as a measure of aridity in the Mediterranean region under the A2 and B2 scenarios, as simulated with a high-resolution RCM (20-km grid interval). Although differences in the classification systems complicate comparisons of our results for the A2 scenario with those obtained in the two previous studies, the main features appear quite similar. For example, all three studies suggest a northward expansion of the steppe (BS) climate in Spain and even a shift to a desert (BW) climate in a small area in the southeastern corner of the country. A major disagreement can be found in southeastern Europe, where areas occupied by the BS climate during the last third of this century are much wider in the present study.

In addition to the effects of emission scenarios, uncertainties in the speed of movement of the climatic borderlines and in regional-scale details are caused by deficiencies in model formulation and by natural climate variability. The current work improves on the previous study by Kalvová et al. (2003) in that the projections have been constructed on the basis of as many as 19 GCMs. Additionally, three different SRES scenarios have been considered. The construction of climate change scenarios on the basis of a large suite of the CMIP3 GCMs has the advantage of reducing natural climate variability and including a wide range of plausible future climate responses. Our assumption here is that a model ensemble offers more reliable projections of climate change than an arbitrarily chosen individual global or regional model. Nonetheless, it must be emphasized that the spatial resolution of the GCM simulations is typically 200–300 km and hence much coarser than that of the observational dataset. Besides, averaging was made over a large number of GCMs with different land–sea masks and topography fields. The spatial details in our results for future projections in coastal and mountainous areas should therefore be regarded with caution. A more realistic consideration of local and regional-scale features in areas of complex topography and coastlines requires the downscaling of GCM simulations with high-resolution RCMs, as demonstrated by de Castro et al. (2007) and Gao and Giorgi (2008).

4. Potential applications of the projections of the climatic zones

One example of application of the Köppen system, surveyed by us with the aid of the online questionnaire, is to consider shifts of the climatic zones from the point of view of communicating climate change information to the public. Because the system is closely linked with vegetative diversity, migrations of the climatic zones are of relevance for ecosystem services. Other impacts on society can also be identified.

a. Comprehensibility of the climatic zones as a dissemination tool

The maps of climatic zones for the 1980s, 2050s, and 2080s (Figs. 3a,c,d) were shown to the online survey respondents to examine whether these maps could be interpreted correctly and to get feedback about the maps as a visualization tool for climate change. The 13 checkbox queries measuring the comprehensibility of the maps are shown in Table 3. The first four statements aimed to test if the viewers could interpret the maps simply by means of the color patterns and in question 3 by utilizing their prior knowledge of the present-day climate in their own country (Finland). On average, 91% of the respondents were able to answer the questions (column 2 in Table 3). At that stage, the majority (89.5%) of the respondents found the maps clear and easy or rather clear and easy to understand. Only 3.4% considered them difficult or rather difficult (Fig. 6). On a scale from 1 to 5, the mean score given to the visualization method was 4.3 (standard deviation of 0.8), regardless of gender.

Table 3.

The questionnaire about interpreting the maps of Köppen climatic zones in Europe. The respondents were shown the observationally based map for the period 1971–2000 (Fig. 3a) and the projections for the 2050s and 2080s under the A1B scenario (Figs. 3c,d). The 13 checkbox questions and statements, originally in Finnish but translated into English, are given in column 1. The alternatives, with the correct one underlined, are indicated in parentheses. The geographical locations referred to in the questions were marked in the maps, and the definitions of the zones (see bottom of Fig. 3) were shown to the respondents after questions 1–4. The percentages of correct answers by all the respondents are shown in column 2. Columns 3 and 4 give the results of those who responded differently to an additional statement that they had based their answers purely on the maps (Table 4). The sample for column 3 consists of those who agreed with this statement to some extent or fully (4 or 5 on a scale from 1 to 5), whereas the sample for column 4 consists of those who did not (scores 1–3).

The questionnaire about interpreting the maps of Köppen climatic zones in Europe. The respondents were shown the observationally based map for the period 1971–2000 (Fig. 3a) and the projections for the 2050s and 2080s under the A1B scenario (Figs. 3c,d). The 13 checkbox questions and statements, originally in Finnish but translated into English, are given in column 1. The alternatives, with the correct one underlined, are indicated in parentheses. The geographical locations referred to in the questions were marked in the maps, and the definitions of the zones (see bottom of Fig. 3) were shown to the respondents after questions 1–4. The percentages of correct answers by all the respondents are shown in column 2. Columns 3 and 4 give the results of those who responded differently to an additional statement that they had based their answers purely on the maps (Table 4). The sample for column 3 consists of those who agreed with this statement to some extent or fully (4 or 5 on a scale from 1 to 5), whereas the sample for column 4 consists of those who did not (scores 1–3).
The questionnaire about interpreting the maps of Köppen climatic zones in Europe. The respondents were shown the observationally based map for the period 1971–2000 (Fig. 3a) and the projections for the 2050s and 2080s under the A1B scenario (Figs. 3c,d). The 13 checkbox questions and statements, originally in Finnish but translated into English, are given in column 1. The alternatives, with the correct one underlined, are indicated in parentheses. The geographical locations referred to in the questions were marked in the maps, and the definitions of the zones (see bottom of Fig. 3) were shown to the respondents after questions 1–4. The percentages of correct answers by all the respondents are shown in column 2. Columns 3 and 4 give the results of those who responded differently to an additional statement that they had based their answers purely on the maps (Table 4). The sample for column 3 consists of those who agreed with this statement to some extent or fully (4 or 5 on a scale from 1 to 5), whereas the sample for column 4 consists of those who did not (scores 1–3).
Fig. 6.

Distribution (in %) of opinions among the questionnaire respondents about the maps of climatic zones as a tool to visualize climate change. The score on a scale from 1 (unclear and difficult to understand) to 5 (clear and easy to understand) was given twice, for the first time after replying to the first four basic questions (1–4) and for the second time after answering questions 5–13 (see Table 3 for the questions).

Fig. 6.

Distribution (in %) of opinions among the questionnaire respondents about the maps of climatic zones as a tool to visualize climate change. The score on a scale from 1 (unclear and difficult to understand) to 5 (clear and easy to understand) was given twice, for the first time after replying to the first four basic questions (1–4) and for the second time after answering questions 5–13 (see Table 3 for the questions).

In the next phase, explanations for the climatic zones, translated into English in the legend of Fig. 3, were made available and more detailed questions were presented (questions 5–13). The percentage of correct answers to all 13 checkbox questions was 86% on average, ranging from 65% to 93% among the questions. Question 10 about changes in summer precipitation in France and question 6 on finding the reason for Belarus belonging to a different climatic zone in the future had the lowest proportion of correct responses. This suggests that more comprehensive explanations for the three levels of categorization in the Köppen system might be needed.

After answering the further questions, the respondents were asked again about the usability of these charts as tools to visualize projections of climate change. Now, 80.7% regarded the maps as clear and easy or rather clear and easy to understand, and 8.8% regarded the maps as difficult or rather difficult to interpret (Fig. 6). The mean grade given was 3.9 (standard deviation of 0.9). Presumably, the somewhat worse results reflect the difficulty of the questions preceding this evaluation.

Most of the viewers recalled having seen graphical representations of climatic zones previously, typically at two of the following alternatives: at school (60.1%); in newspapers and other mass media (73.3%); in textbooks, nonfiction books or professional literature (56.4%); or elsewhere (e.g., on the Internet; 20.3%). Only 4.2% of the viewers could not recall having seen them before. A somewhat larger proportion of the respondents agreed fully or partly with the statement about having a clear picture in their mind about what the climate is like in most of the zones (72.7%) than about what kind of vegetation is peculiar to the various climatic zones (61.1%). Expectedly, the scores 1–5 given to these two statements were rather closely linked, with a correlation coefficient of 0.74 (p < 0.001).

To take account of the respondents’ previous knowledge about the subject and to assess the process of answering, they were asked to give an opinion about two statements. The former (S1) argued that their answers were based purely on the maps. Although 89.2% agreed fully or partly, only 10.8% disagreed fully or partly or neither agreed nor disagreed (the bottom row in Table 4). Answers to the 13 true/false questions by these two groups were considered separately (columns 3–4 in Table 3). The average percentage of correct answers was considerably lower (65%) among those who disagreed with the statement or did not have an opinion, in comparison with the value of 88% for those who had based their answers on the maps, the difference being statistically significant at a level higher than 99.95%.

Table 4.

Two-dimensional distribution (in %) of replies to arguments S1 and S2 measuring the respondents’ previous knowledge about climate change and assessing the process of answering. S1 states that the respondent answered the questions purely on the basis of the maps (in columns); S2 argues that he/she would have been able to answer the questions even without the maps (in rows). The scale is as follows: 1 (fully disagree), 2 (partly disagree), 3 (neither agree nor disagree), 4 (partly agree), and 5 (fully agree).

Two-dimensional distribution (in %) of replies to arguments S1 and S2 measuring the respondents’ previous knowledge about climate change and assessing the process of answering. S1 states that the respondent answered the questions purely on the basis of the maps (in columns); S2 argues that he/she would have been able to answer the questions even without the maps (in rows). The scale is as follows: 1 (fully disagree), 2 (partly disagree), 3 (neither agree nor disagree), 4 (partly agree), and 5 (fully agree).
Two-dimensional distribution (in %) of replies to arguments S1 and S2 measuring the respondents’ previous knowledge about climate change and assessing the process of answering. S1 states that the respondent answered the questions purely on the basis of the maps (in columns); S2 argues that he/she would have been able to answer the questions even without the maps (in rows). The scale is as follows: 1 (fully disagree), 2 (partly disagree), 3 (neither agree nor disagree), 4 (partly agree), and 5 (fully agree).

The latter statement (S2) argued that the respondents would have been able to answer the questions even without the maps. The correlation between responses to this and to statement S1 was negligible (−0.05). Remarkably, 30.6% agreed fully or partly with S2 (the last column in Table 4). However, as regards the 13 true/false questions, there was no statistically significant difference between their results and those of them who needed the maps to answer and therefore disagreed with S2. The inference can be made that the level of prior knowledge about climate change was not crucial, whereas the use of the charts notably helped the viewers to reply to the questions correctly.

The answers to the 13 true/false questions were weakly affected by educational background. The mean percentage of correct answerers was 81% among respondents having primary or lower vocational education and 88% among those with tertiary education. The grades given to the visualization method were virtually equal, 4.2 versus 4.3 after the first four basic questions and 3.9 versus 4.0 after all the questions. Although the proportions of correct answers differed significantly (statistically at a level higher than 99.95%) between the two subsamples, even the lower percentage (81%) can be considered quite high.

A larger proportion of respondents in the lower educational group, compared to those having higher educational background, fully disagreed that they would have been able to answer the questions even without the maps (23% versus 16%). However, the mean grades given to the statement S2 were virtually equal (2.7 versus 2.8). Considering the actual process of answering, those more educated based their results more thoroughly on the maps, with 91% of them fully or partially agreeing with statement S1 compared to 82% for those less educated. The mean grades given to S1 were both high (4.5 versus 4.2) and differed statistically only at a significance level of 95%.

Regarding the relations of age and gender with statements S1 and S2, age only affected responses to S1 and gender only affected responses S2. Although 91% of the respondents between the ages of 25 and 64 yr fully or partly agreed with the statement of having answered the questions purely on the basis of the maps, the percentage for younger or older people was only 78%. The mean scores given to S1 by the middle aged and by others (4.4 and 4.1, respectively) deviated statistically significantly (at a level higher than 97.25%). The statement about being able to answer the questions even without the maps was fully or partly agreed with by 39% of men but only 17% of women. The difference between the mean scores given to S2 by women and men (2.4 and 2.9, respectively) was statistically significant (at a level higher than 99.95%). However, the percentage of correct answers to the 13 checkbox questions (86%) was the same for men and women.

It can be concluded that for this questionnaire survey, demographic factors and the respondents’ previous knowledge about the subject had a clearly smaller influence than the process of answering: that is, whether the respondent had utilized the maps when replying to the questions. The use of the charts notably helped the viewers to find the correct answers.

Inferring from the feedback that we received, our audience agreed with the potential of the maps of projected climatic zones to communicate climate change. However, there are a few aspects to notice. First, the respondents do not represent the general public in Finland but a tiny fraction of those who visited the FMI home page. Second, judging from the comments given to the open questions, there were a couple of climate change skeptics among the respondents, but the number of persons highly concerned about climate change was harder to assess. Because of their goals and expectations, disbelievers and eager mitigators could have very different ideas on how well these maps communicate climate change; even their interpretations could vary. Third, we did not examine to what extent the audience would have preferred separate maps of seasonal or annual temperature and precipitation changes or maps of future temperature and precipitation patterns rather than the climatic zones. Fourth, there may be a risk of misinterpretation, because someone might mistakenly think that regions with no shifts in climatic zones (like Germany in Fig. 3) do not experience climate changes at all. By providing guidance for the viewers of the illustrations, this risk can be avoided.

b. Examples of climate change dissemination activities

Based on the questionnaire, it can be argued that maps illustrating the projected future shifts of the climatic zones are fairly easy to understand and help to make the regional impacts of global climate change more easily imaginable. In Finland, the potential of projected climatic zones as a dissemination tool for climate change information was first utilized in a 1-yr exhibition entitled “Open Questions” in Heureka, the Finnish Science Centre in Vantaa. The exhibition in 2003–04 was aimed at the general public, particularly young people and their teachers. In the exhibition, a personal computer interactively displayed animations of maps of projected future climatic zones in Europe during the twenty-first century. The animated maps year by year were based on alternative storylines of the future world (Nakićenović and Swart 2000) and on climate change experiments using two global climate models, HadCM3 (Gordon et al. 2000; Pope et al. 2000) and ECHAM4/Ocean Isopycnal Model 3 (OPYC3; Roeckner et al. 1999), together with a global observational dataset (Legates and Willmott 1990a,b). To see an animation, a visitor to the exhibition needed to choose one of the storylines and either of the two GCMs applied. By selecting various combinations, he could get an impression of the uncertainty attached to the evolution of the future climate. The visitor was also provided background information about constructing climate change scenarios.

Since the exhibition, the Heureka animations of future climatic zones have been made freely available via the Internet in a Finnish climate change virtual Web school. In a Web school component entitled “What is projected?” pupils are asked to identify regions exposed to notable alterations in climatic classes under the alternative GHG scenarios and also to ponder the underlying storylines of the future world. Furthermore, one task given is to compare climatic changes in one’s home district under larger emissions versus smaller future emissions. In staff training days for school teachers, feedback about the Web school was collected. Teachers particularly appreciated the clarity of the animations (T. Alakoski 2008, personal communication).

Besides students and teachers, particularly important audiences for global change information and motivators for climate protective actions are journalists. They have an important role in distributing relevant information on the scientific issues to the general public. Accordingly, climate change educational courses for journalists have been annually organized in Finland since 2006. In these courses, a selection of the Heureka maps depicting shifts in the climatic zones has proved to be very illustrative. The maps have also been displayed in more than 50 lectures and seminar presentations around Finland, many of them contributing to the Finnish Climate Change Communications Programme in 2002–07 and the European Commission’s “You Control Climate Change” campaign in 2006–07.

The updated maps of climatic zones presented in this paper are based on a state-of-the-art ensemble of global climate models and a new high-resolution gridded observational dataset. For some climate change outreach purposes, it may be thought that the number of climate types shown in Fig. 3 is too large. An alternative is to neglect the third level of the classification (i.e., the right-hand side of the legend of Fig. 3). This may be particularly relevant if the projected evolutions of the climatic zones in time are shown as animated maps instead of frozen ones. Although the focus of this paper is on Europe, the present approach is also readily applicable to other world regions.

c. Climatic zones, ecosystem services, and society

Shifts in climatic zones in Europe during the coming decades are likely to have multiple impacts on the environment and ecosystem services and thereby on society and human life: on water resources, agriculture, forestry, transportation, energy production and demand, insurance, health, tourism, and recreation. Actually, many changes in natural and managed ecosystems have already been attributed to recent temperature and precipitation trends in Europe. These include, among others, the lengthening of the growing season; the upward shift of the tree line; the increasing productivity and carbon sink of forests; decreases in glacier volume; changes in high mountain vegetation types; the increased crop stress during hotter, drier summers; the earlier onset and extension of the allergenic pollen season; and the movement of tick vectors northward (Alcamo et al. 2007).

It is noteworthy that the Köppen classification system is based on the strong dependence of natural vegetation on climate. Natural plants integrate many effects of climate, particularly expressing the adequacy of moisture under a given set of temperature conditions (Critchfield 1966). However, because some adaptation of plant species to changes in climate may occur, because the length of day will remain unaltered, and because the expansion of plants to new regions is also affected by nonclimatic factors, we do not know how appropriate the threshold values used in the current procedure will be in the future from the point of view of natural vegetation patterns. As emphasized by de Castro et al. (2007), four aspects must be kept in mind if climate classifications are used to assess the possible broad-scale impacts on vegetation of a projected climatic change. First, there is the uncertainty in the simulations of the future climate. Second, the relationships between climate and vegetation may not be the same in the future scenarios as in the current conditions. Third, the feedback of vegetation distribution changes to climate via altering surface characteristics is ignored. Fourth, because climate–vegetation schemes only consider a limited number of divisions, they hardly represent the entire current vegetative diversity.

In addition to natural vegetation, other approaches of climate categorization may be suggested. Optimum and limiting values of climatic factors give a basis for classification in terms of, for example, cultivated plant species; insect pests; or human health, clothing, and comfort. As stated by Critchfield (1966), the crucial point is the purpose for which a systematic arrangement of climates is intended. An interesting application is shown by Lemke and Stein (2008), who considered only the temperature-dependent first and third levels of the Köppen classification in their estimates of changes in the energy demand of buildings. They arrived at 12 temperature-dependent climate types globally, the projected shifts of which by 2050 were utilized to assess changes in heating and cooling degree days.

Climate classification utilizes a powerful tool used in both science and education: analogy. Another approach using analogs to describe climate change impacts is to look for regions where the present-day climate resembles the anticipated future climate of the study area. This method of spatial analogs has been applied, for example, by Hallegatte et al. (2007). To assess the economic impacts of climate change in urban areas, they relocated some European cities to places where the current climate is the same as what the projected climate will be for that city in the future. Unlike maps of climatic zones, this kind of spatial analog projections may, however, distort the geographical configuration between locations (see, e.g., Fig. 1 in Hallegatte et al. 2007).

5. Conclusions

The Köppen climate classification is based on seasonal cycles of monthly-mean temperature and precipitation, dividing Europe and neighboring land areas into tundra, boreal, temperate, and dry climate types. During the latter half of the twentieth century, 12.1% of the land areas (or 1.17 × 106 km2, analogous to an area twice as large as the French mainland) was affected by shifts toward warmer or drier climate types. These shifts were more than 5 times more widespread than changes toward cooler or wetter climate types. Among the seven main climate types prevailing in the study domain, those assigned to dry or tundra climates indicated the largest percentage changes, with the former extending and the latter retreating. In absolute terms, the widest extension in area coverage took place for the moist midlatitude climate with mild winters (the Cf class).

These recent changes in the climate types were chiefly located near regions where the most notable reallocations of the climate types are projected to occur during this century as a result of the enhancing greenhouse gas forcing. On the basis of multimodel mean climate projections, major changes in the climatic zones will take place, particularly in northeastern Europe, in the Iberian Peninsula, around the Black Sea, and in the Alps. Altogether, around 45%–70% of the study domain will be affected by shifts toward a warmer or drier climate type, depending on the emission scenarios. Within the next few decades, regions assigned to either tundra or boreal climates are projected to shrink at a rate of about 180 × 103 km2 decade−1 (roughly corresponding to an area half that of Germany), whereas the temperate and dry climate zones are projected to expand at rates of about 140 × 103 and 40 × 103 km2 decade−1, respectively. The projected pattern of climatic zones in 2070–99 under the SRES B1 scenario is very close to that in 2040–69 under the A1B scenario, whereas the A2 scenario produces the fastest and most significant alterations. These changes are likely to have multiple impacts on the environment and on ecosystem services and thereby on society and human life.

In this study, we considered the projections of climatic zones in Europe for the purpose of visualizing climate change information. Based on the feedback gathered by a Web-based questionnaire, maps showing projected temporal evolutions of the climatic zones appeared to be an easily comprehensible means for dissemination of climate change information. The percentage of correct answers to the 13 checkbox questions in the survey was 86% on average, ranging from 65% to 93% among the questions; 80.7% of the respondents regarded the maps as an effective tool to visualize projections of climate change. The educational background of the respondents and their level of prior knowledge about climate change were clearly less influential than the process of answering: that is, whether the respondent had utilized the maps when replying to the questions. The use of the charts notably helped the viewers to find the correct answers.

According to Nicholson-Cole (2005), visualizations of climate change should have five characteristics in order to be attention grabbing, easy to remember, and likely to motivate viewers to make behavioral changes. Our maps are already aimed at fulfilling two of them: they make an effort to be easy to relate to and to be personally applicable (because our maps are based on climatic zones that are, in general, familiar to people) and they are scientifically as accurate as possible [they were based on simulations performed with 19 state-of-the-art climate models employed by Solomon et al. (2007)]. The remaining three: being instructive with a clear message so that people would undertake mitigation, having an affective dimension (e.g., keeping the attention of the viewers), and being aimed and tailored at the target audience are all issues with which we could continue working.

Fig. A1. The flow diagram for defining the five principal Köppen categories A–E. The second level of the classification is also shown for A, B, and E. For the variables, see the legend. The main classes BW and BS have subtypes h and k defined as follows: for h, T ≥ 18°C; for k, T < 18°C (based on Critchfield 1966).

Fig. A1. The flow diagram for defining the five principal Köppen categories A–E. The second level of the classification is also shown for A, B, and E. For the variables, see the legend. The main classes BW and BS have subtypes h and k defined as follows: for h, T ≥ 18°C; for k, T < 18°C (based on Critchfield 1966).

Fig. A2. The flow diagram for defining the main classes and subtypes for the principal Köppen categories C and D. For the variables, see the legend (based on Critchfield 1966).

Fig. A2. The flow diagram for defining the main classes and subtypes for the principal Köppen categories C and D. For the variables, see the legend (based on Critchfield 1966).

Acknowledgments

This work was partially financed by the Finnish Climate Change Adaptation Research Programme ISTO, funded by the Finnish ministries of Transport and Communications, the Environment, and Agriculture and Forestry and by the Finnish Road Administration; by the ALARM and ENSEMBLES projects, funded by the European Commission’s 6th Framework Programme through Contracts GOCE-CT-2003-506675 and GOCE-CT-2003-505539, respectively; and by the Climate and Energy Systems—Risks, Potential and Adaptation (CES) project, funded by Nordic Energy Research. Observational climate data used by us were made available by the ENSEMBLES project and the European Climate Assessment and Dataset (ECAandD) project. For climate model data, we acknowledge the modeling groups, the Program for Climate Model Diagnosis and Intercomparison (PCMDI) and the WCRP Working Group on Coupled Modelling (WGCM) for their roles in making available the WCRP CMIP3 multimodel dataset. Dr. Jouni Räisänen is thanked for assistance in the climate model data processing. Kaisa Vähähyyppä, Tomi Alakoski, emeritus Prof. Juhani Rinne, and Esa Liekovuori are thanked for cooperation and provision of information related to the Heureka animations of climatic zones (section 4b). The valuable advice given by Prof. Tom Moring [the Swedish School of Social Science of the University of Helsinki (UH)], Prof. Hannu Nieminen (Department of Communication, UH), Prof. Lily Diaz and Markku Reunanen (the Media Lab of the University of Art and Design Helsinki), and the communications unit of the Finnish Meteorological Institute is gratefully acknowledged.

REFERENCES

REFERENCES
AchutaRao
,
K.
,
C.
Covey
,
C.
Doutriaux
,
M.
Fiorino
,
P.
Gleckler
,
T.
Phillips
,
K.
Sperber
, and
K.
Taylor
,
2004
:
An appraisal of coupled climate model simulations.
Lawrence Livermore National Laboratory Tech. Rep. UCRL-TR-202550, 197 pp. [Available online at https://e-reports-ext.llnl.gov/pdf/305293.pdf]
.
Alcamo
,
J.
, and
Coauthors
,
2007
:
Europe.
Climate Change 2007: Impacts, Adaptation and Vulnerability, M. L. Parry et al., Eds., Cambridge University Press, 541–580
.
Beck
,
C.
,
J.
Grieser
,
M.
Kottek
,
F.
Rubel
, and
B.
Rudolf
,
2006
:
Characterizing global climate change by means of Köppen climate classification.
Klimastatusbericht
,
2005
,
139
149
.
Conrad
,
V.
, and
L. W.
Pollak
,
1950
:
Methods in Climatology.
Harvard University Press, 459 pp
.
Conway
,
T. J.
,
P. P.
Tans
,
L. S.
Waterman
,
K. W.
Thoning
,
D. R.
Kitzis
,
K. A.
Masarie
, and
N.
Zhang
,
1994
:
Evidence of interannual variability of the carbon cycle from the NOAA/CMDL global air sampling network.
J. Geophys. Res.
,
99
,
22831
22855
.
Critchfield
,
H. J.
,
1966
:
General Climatology.
2nd ed. Prentice Hall, 420 pp
.
de Castro
,
M.
,
C.
Gallardo
,
K.
Jylhä
, and
H.
Tuomenvirta
,
2007
:
The use of a climate-type classification for assessing climate change effects in Europe from an ensemble of regional climate models.
Climatic Change
,
81
, (
Suppl.
).
329
341
.
doi:10.1007/s10584-006-9224-1
.
Fraedrich
,
K.
,
F-W.
Gerstengarbe
, and
P. C.
Werner
,
2001
:
Climate shifts during the last century.
Climatic Change
,
50
,
405
417
.
doi:10.1023/A:1010699428863
.
Gao
,
X.
, and
F.
Giorgi
,
2008
:
Increased aridity in the Mediterranean region under greenhouse gas forcing estimated from high resolution simulations with a regional climate model.
Global Planet. Change
,
62
,
195
209
.
doi:10.1016/j.gloplacha.2008.02.002
.
Gerstengarbe
,
F-W.
, and
P. C.
Werner
,
2008
:
A short update on Koeppen climate shifts in Europe between 1901 and 2003.
Climatic Change
,
92
,
99
107
.
doi:10.1007/s10584-008-9430-0
.
Gnanadesikan
,
A.
, and
R. J.
Stouffer
,
2006
:
Diagnosing atmosphere-ocean general circulation model errors relevant to the terrestrial biosphere using the Köppen climate classification.
Geophys. Res. Lett.
,
33
,
L22701
.
doi:10.1029/2006GL028098
.
Gordon
,
C.
,
C.
Cooper
,
C. A.
Senior
,
H.
Banks
,
J. M.
Gregory
,
T. C.
Johns
,
J. F. B.
Mitchell
, and
R. A.
Wood
,
2000
:
The simulation of SST, sea ice extents and ocean heat transports in a version of the Hadley Centre coupled model without flux adjustments.
Climate Dyn.
,
16
,
147
168
.
doi:10.1007/s003820050010
.
Graham
,
L. P.
,
S.
Hagemann
,
S.
Jaun
, and
M.
Beniston
,
2007
:
On interpreting hydrological change from regional climate models.
Climatic Change
,
81
,
97
122
.
doi:10.1007/s10584-006-9217-0
.
Grundstein
,
A.
,
2009
:
Evaluation of climate change over the continental United States using a moisture index.
Climatic Change
,
93
,
103
115
.
doi:10.1007/s10584-008-9480-3
.
Guetter
,
P. J.
, and
J. E.
Kutzbach
,
1990
:
A modified Köppen classification applied to model simulations of glacial and interglacial climates.
Climatic Change
,
16
,
193
215
.
doi:10.1007/BF00134657
.
Hallegatte
,
S.
,
J-C.
Hourcade
, and
P.
Ambrosi
,
2007
:
Using climate analogues for assessing climate change economic impacts in urban areas.
Climatic Change
,
82
,
47
60
.
doi:10.1007/s10584-006-9161-z
.
Hansen
,
J.
,
A.
Lacis
,
D.
Rind
,
G.
Russel
,
P.
Stone
,
I.
Fung
,
R.
Ruedy
, and
J.
Lerner
,
1984
:
Climate sensitivity: Analysis of feedback mechanisms.
Climate Processes and Climate Sensitivity, Geophys. Monogr., Vol. 29, Amer. Geophys. Union, 130–163
.
Haylock
,
M. R.
,
N.
Hofstra
,
A. M. G.
Klein Tank
,
E. J.
Klok
,
P. D.
Jones
, and
M.
New
,
2008
:
A European daily high-resolution gridded data set of surface temperature and precipitation for 1950–2006.
J. Geophys. Res.
,
113
,
D20119
.
doi:10.1029/2008JD010201
.
Houghton
,
J. T.
,
Y.
Ding
,
D. J.
Griggs
,
M.
Noguer
,
P. J.
van der Linden
,
X.
Da
,
K.
Maskell
, and
C. A.
Johnson
,
Eds.
2001
:
Climate Change 2001: The Scientific Basis.
Cambridge University Press, 881 pp
.
Kalvová
,
J.
,
T.
Halenka
,
K.
Bezpalcová
, and
I.
Nemešová
,
2003
:
Köppen climate types in observed and simulated climates.
Stud. Geophys. Geod.
,
47
,
185
202
.
doi:10.1023/A:1022263908716
.
Kleidon
,
A.
,
K.
Fraedrich
, and
M.
Heimann
,
2000
:
A green planet versus a desert world: Estimating the maximum effect of vegetation on the land surface climate.
Climatic Change
,
44
,
471
493
.
doi:10.1023/A:1005559518889
.
Köppen
,
W. P.
,
1936
:
Das geographische System der Klimate.
Handbuch der Klimatologie, W. P. Köppen and R. Geiger, Eds., Vol. 1, Part C, Gebrüder Borntraeger, C1–C44
.
Kottek
,
M.
,
J.
Grieser
,
C.
Beck
,
B.
Rudolf
, and
F.
Rubel
,
2006
:
World map of the Köppen-Geiger climate classification updated.
Meteor. Z.
,
15
,
259
263
.
doi:10.1127/0941-2948/2006/0130
.
Legates
,
D. R.
, and
C. J.
Willmott
,
1990a
:
Mean seasonal and spatial variability global surface air temperature.
Theor. Appl. Climatol.
,
41
,
11
21
.
doi:10.1007/BF00866198
.
Legates
,
D. R.
, and
C. J.
Willmott
,
1990b
:
Mean seasonal and spatial variability in gauge-corrected, global precipitation.
Int. J. Climatol.
,
10
,
111
127
.
doi:10.1002/joc.3370100202
.
Lemke
,
C. R.
, and
B.
Stein
,
2008
:
Population–climate–energy: Scenarios to 2050.
Klima 2008, Hamburg, Germany, Hamburg University of Applied Sciences, 8 pp
.
Le Quéré
,
C.
, and
Coauthors
,
2009
:
Trends in the sources and sinks of carbon dioxide.
Nat. Geosci.
,
2
,
831
836
.
doi:10.1038/ngeo689
.
Lewis
,
H. A. G.
, and
P. J. M.
Geelan
,
1998
:
The Times Atlas of the World.
9th ed. Times Books, 123 pp. + app. 218 pp
.
Lohmann
,
U.
,
R.
Sausen
,
L.
Bengtsson
,
U.
Cubasch
,
J.
Perlwitz
, and
E.
Roeckner
,
1993
:
The Köppen climate classification as a diagnostic tool for general circulation models.
Climate Res.
,
3
,
177
193
.
Luterbacher
,
J.
,
D.
Dietrich
,
E.
Xoplaki
,
M.
Grosjean
, and
H.
Wanner
,
2004
:
European seasonal and annual temperature variability, trends, and extremes since 1500.
Science
,
303
,
1499
1503
.
doi:10.1126/science.1093877
.
Manabe
,
S.
, and
J. L.
Holloway
,
1975
:
The seasonal variation of the hydrologic cycle as simulated by a global model of the atmosphere.
J. Geophys. Res.
,
80
,
1617
1649
.
Meehl
,
G. A.
,
C.
Covey
,
T.
Delworth
,
M.
Latif
,
B.
McAvaney
,
J. F. B.
Mitchell
,
R. J.
Stouffer
, and
K. E.
Taylor
,
2007
:
The WCRP CMIP3 multimodel dataset: A new era in climate change research.
Bull. Amer. Meteor. Soc.
,
88
,
1383
1394
.
Moser
,
S. C.
,
2008
:
Toward a deeper engagement of the U.S. public on climate change: An open letter to the 44th president of the United States of America.
Int. J. Sustainability Commun.
,
3
,
119
132
.
Nakićenović
,
N.
, and
R.
Swart
,
2000
:
Special Report on Emissions Scenarios.
Cambridge University Press, 599 pp
.
Nicholson-Cole
,
S. A.
,
2005
:
Representing climate change futures: A critique on the use of images for visual communication.
Comput. Environ. Urban Syst.
,
29
,
255
273
.
doi:10.1016/j.compenvurbsys.2004.05.002
.
Peel
,
M. C.
,
B. L.
Finlayson
, and
T. A.
McMahon
,
2007
:
Updated world map of the Köppen-Geiger climate classification.
Hydrol. Earth Syst. Sci.
,
11
,
1633
1644
.
Pope
,
V. D.
,
M. L.
Gallani
,
P. R.
Rowntree
, and
R. A.
Stratton
,
2000
:
The impact of new physical parametrizations in the Hadley Centre climate model: HadAM3.
Climate Dyn.
,
16
,
123
146
.
doi:10.1007/s003820050009
.
Retallack
,
S.
,
T.
Lawrence
, and
M.
Lockwood
,
2007
:
Positive energy: Harnessing people power to prevent climate change: A summary.
Institute for Public Policy Research, 23 pp
.
Roeckner
,
E.
,
L.
Bengtsson
,
J.
Feichter
,
J.
Lelieveld
, and
H.
Rodhe
,
1999
:
Transient climate change simulations with a coupled atmosphere–ocean GCM including the tropospheric sulfur cycle.
J. Climate
,
12
,
3004
3032
.
Ruosteenoja
,
K.
,
H.
Tuomenvirta
, and
K.
Jylhä
,
2007
:
GCM-based regional temperature and precipitation change estimates for Europe under four SRES scenarios applying a super-ensemble pattern-scaling method.
Climatic Change
,
81
, (
Suppl.
).
193
208
.
doi:10.1007/s10584-006-9222-3
.
Sanderson
,
M.
,
1999
:
The classification of climates from Pythagoras to Koeppen.
Bull. Amer. Meteor. Soc.
,
80
,
669
673
.
Solomon
,
S.
,
D.
Qin
,
M.
Manning
,
M.
Marquis
,
K.
Averyt
,
M. M. B.
Tignor
,
H. L.
Miller
Jr.
, and
Z.
Chen
,
Eds.
2007
:
Climate Change 2007: The Physical Science Basis.
Cambridge University Press, 996 pp
.
Staats
,
H. J.
,
A. P.
Wit
, and
C. J. H.
Midden
,
1996
:
Communicating the greenhouse effect to the public: Evaluation of a mass media campaign from a social dilemma perspective.
J. Environ. Manage.
,
46
,
189
203
.
Stern
,
H.
,
G.
de Hoedt
, and
J.
Ernst
,
2000
:
Objective classification of Australian climates.
Aust. Meteor. Mag.
,
49
,
87
93
.
Wang
,
M.
, and
J. E.
Overland
,
2004
:
Detecting Arctic climate change using Köppen climate classification.
Climatic Change
,
67
,
43
62
.
doi:10.1007/s10584-004-4786-2
.
Yates
,
D. N.
,
T. G. F.
Kittel
, and
R. F.
Cannon
,
2000
:
Comparing the correlative Holdridge model to mechanistic biogeographical models for assessing vegetation distribution response to climatic change.
Climatic Change
,
44
,
59
87
.
doi:10.1023/A:1005495908758
.
Yue
,
T. X.
,
Z. M.
Fan
,
J. Y.
Liu
, and
B. X.
Wei
,
2006
:
Scenarios of major terrestrial ecosystems in China.
Ecol. Modell.
,
199
,
363
376
.

APPENDIX

The Köppen Classification System

The procedure used to carry out the Köppen classification is shown by the flow diagrams in Figs. A1 and A2.

Footnotes

Corresponding author address: Kirsti Jylhä, Finnish Meteorological Institute, P.O. Box 503, FI-00101 Helsinki, Finland. Email: kirsti.jylha@fmi.fi