• Andrews, T., , M. Doutriaux-Boucher, , O. Boucher, , and P. M. Forster, 2011: A regional and global analysis of carbon dioxide physiological forcing and its impacts on climate. Climate Dyn., 36, 783792, doi:10.1007/s00382-010-0742-1.

    • Search Google Scholar
    • Export Citation
  • Dai, A., , J. Wang, , P. W. Thorne, , D. E. Parker, , L. Haimberger, , and X. L. Wang, 2011: A new approach to homogenize daily radiosonde humidity data. J. Climate, 24, 965991.

    • Search Google Scholar
    • Export Citation
  • Dubrovsky, M., , I. Nemesova, , and J. Kalvova, 2005: Uncertainties in climate change scenarios for the Czech Republic. Climate Res., 29, 139156.

    • Search Google Scholar
    • Export Citation
  • Fasullo, J. T., 2010: Robust land–ocean contrast in energy and water cycle feedbacks. J. Climate, 23, 46774693.

  • Fischer, A. M., , A. P. Weigel, , C. M. Buser, , R. Knutti, , H. R. Künsch, , M. A. Liniger, , C. Schär, , and C. Appenzeller, 2011: Climate change projections for Switzerland based on a Bayesian multi-model approach. Int. J. Climatol., 32, 23482371, doi:10.1002/joc.3396.

    • Search Google Scholar
    • Export Citation
  • Fischer, E. M., , and C. Schär, 2010: Consistent geographical patterns of changes in high-impact European heatwaves. Nat. Geosci., 3, 398403, doi:10.1038/NGEO866.

    • Search Google Scholar
    • Export Citation
  • Jylhä, K., , H. Tuomenvirta, , K. Ruosteenoja, , H. Niemi-Hugaerts, , K. Keisu, , and J. A. Karhu, 2010: Observed and projected future shifts of climate zones in Europe and their use to visualize climate change information. Wea. Climate Soc., 2, 148167.

    • Search Google Scholar
    • Export Citation
  • Kilpeläinen, A., , S. Kellomäki, , H. Strandman, , and A. Venäläinen, 2010: Climate change impacts on forest fire potential in boreal conditions in Finland. Climatic Change, 103, 383398.

    • Search Google Scholar
    • Export Citation
  • Köppen, W. P., 1936: Das geographische System der Klimate. Handbuch der Klimatologie, W. P. Köppen and R. Geiger, Eds., Vol. I, Gebrüder Borntraeger, C1–C44.

  • Lorenz, D. J., , and E. T. Deweaver, 2007: The response of the extratropical hydrological cycle to global warming. J. Climate, 20, 34703484.

    • Search Google Scholar
    • Export Citation
  • McCarthy, M. P., , P. W. Thorne, , and H. A. Titchner, 2009: An analysis of tropospheric humidity trends from radiosondes. J. Climate, 22, 58205838.

    • Search Google Scholar
    • Export Citation
  • Orlowsky, B., , and S. I. Seneviratne, 2012: Global changes in extreme events: Regional and seasonal dimension. Climatic Change, 110, 669696, doi:10.1007/s10584-011-0122-9.

    • Search Google Scholar
    • Export Citation
  • Pennell, C., , and T. Reichler, 2011: On the effective number of climate models. J. Climate, 24, 23582367.

  • Perch-Nielsen, S. L., , B. Amelung, , and R. Knutti, 2010: Future climate resources for tourism in Europe based on the daily Tourism Climate Index. Climatic Change, 103, 363381.

    • Search Google Scholar
    • Export Citation
  • Räisänen, J., 2007: How reliable are climate models? Tellus, 59A, 229, doi:10.1111/j.1600-0870.2006.00211.x.

  • Richter, I., , and S.-P. Xie, 2008: Muted precipitation increase in global warming simulations: A surface evaporation perspective. J. Geophys. Res.,113, D24118, doi:10.1029/2008JD010561.

  • Rowell, D. P., , and R. G. Jones, 2006: Causes and uncertainty of future summer drying over Europe. Climate Dyn., 27, 281299, doi:10.1007/s00382-006-0125-9.

    • Search Google Scholar
    • Export Citation
  • 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, 193208, doi:10.1007/s10584-006-9222-3.

    • Search Google Scholar
    • Export Citation
  • Ruuhela, R., , L. Hiltunen, , A. Venäläinen, , P. Pirinen, , and T. Partonen, 2009: Climate impact on suicide rates in Finland from 1971 to 2003. Int. J. Biometeor., 53, 167175, doi:10.1007/s00484-008-0200-5.

    • Search Google Scholar
    • Export Citation
  • Seneviratne, S. I., , D. Lüthi, , M. Litschi, , and C. Schär, 2006: Land–atmosphere coupling and climate change in Europe. Nature, 443, 205209, doi:10.1038/nature05095.

    • Search Google Scholar
    • Export Citation
  • Sherwood, S. C., , W. Ingram, , Y. Tsushima, , M. Satoh, , M. Roberts, , P. L. Vidale, , and P. A. O’Gorman, 2010: Relative humidity changes in a warmer climate. J. Geophys. Res.,115, D09104, doi:10.1029/2009JD012585.

  • Simmons, A. J., , K. M. Willett, , P. D. Jones, , P. W. Thorne, , and D. P. Dee, 2010: Low-frequency variations in surface atmospheric humidity, temperature, and precipitation: Inferences from reanalyses and monthly gridded observational data sets. J. Geophys. Res.,115, D01110, doi:10.1029/2009JD012442.

  • 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.

  • Sorteberg, A., , V. Kattsov, , J. E. Walsh, , and T. Pavlova, 2007: The Arctic surface energy budget as simulated with the IPCC AR4 AOGCMs. Climate Dyn., 29, 131156, doi:10.1007/s00382-006-0222-9.

    • Search Google Scholar
    • Export Citation
  • Thum, T., and Coauthors, 2011: Soil carbon model alternatives for ECHAM5/JSBACH climate model: Evaluation and impacts on global carbon cycle estimates. J. Geophys. Res.,116, G02028, doi:10.1029/2010JG001612.

  • Trenberth, K. E., , and D. J. Shea, 2005: Relationships between precipitation and surface temperature. Geophys. Res. Lett.,32, L14703, doi:10.1029/2005GL022760.

  • Trenberth, K. E., , and J. T. Fasullo, 2009: Global warming due to increased absorbed solar radiation. Geophys. Res. Lett.,36, L07706, doi:10.1029/2009GL037527.

  • Viitanen, H., , T. Toratti, , and L. Makkonen, 2011: Development of service life model for wooden structures. Proc. 12th Int. Conf. on Durability of Building Materials and Components, Vol. 1, Porto, Portugul, Faculty of Engineering, University of Porto, 495–502.

  • Weigel, A. P., , R. Knutti, , M. A. Liniger, , and C. Appenzeller, 2010: Risks of model weighting in multimodel climate projections. J. Climate, 23, 41754191.

    • Search Google Scholar
    • Export Citation
  • Wild, M., 2008: Short-wave and long-wave surface radiation budgets in GCMs: A review based on the IPCC-AR4/CMIP3 models. Tellus, 60A, 932945.

    • Search Google Scholar
    • Export Citation
  • Wild, M., , and E. Schmucki, 2011: Assessment of global dimming and brightening in IPCC-AR4/CMIP3 models and ERA40. Climate Dyn., 37, 16711688, doi:10.1007/s00382-010-0939-3.

    • Search Google Scholar
    • Export Citation
  • Willett, K. M., , N. P. Gillett, , P. D. Jones, , and P. W. Thorne, 2007: Attribution of observed surface humidity changes to human influence. Nature, 449, 710713, doi:10.1038/nature06207.

    • Search Google Scholar
    • Export Citation
  • Willett, K. M., , P. D. Jones, , N. P. Gillett, , and P. W. Thorne, 2008: Recent changes in surface humidity: Development of the HadCRUH dataset. J. Climate, 21, 53645383.

    • Search Google Scholar
    • Export Citation
  • WMO, 1989: Calculation of monthly and annual 30-year standard normals. WCDP-10/WMO/TD-341, 12 pp. [Available online at http://www.inmet.gov.br/html/clima/OMM_WCDP_N10.pdf.]

  • Wright, J. S., , A. Sobel, , and J. Galewsky, 2010: Diagnosis of zonal mean relative humidity changes in a warmer climate. J. Climate, 23, 45564569.

    • Search Google Scholar
    • Export Citation
  • Zhang, Y., , W. B. Rossow, , A. A. Lacis, , V. Oinas, , and M. I. Mishchenko, 2004: Calculation of radiative fluxes from the surface to top of atmosphere based on ISCCP and other global data sets: Refinements of the radiative transfer model and the input data. J. Geophys. Res.,109, D19105, doi:10.1029/2003JD004457.

  • Zhou, L., , R. E. Dickinson, , P. Dirmeyer, , A. Dai, , and S.-K. Min, 2009: Spatiotemporal patterns of change in maximum and minimum temperatures in multi-model simulations. Geophys. Res. Lett.,36, L02702, doi:10.1029/2008GL036141.

  • View in gallery

    (a),(b) The deviation (%) of the baseline-period (1971–2000) multimodel mean incident solar radiation from the ISCCP FD satellite-derived estimate for the years 1983–2004. (c),(d) The difference (%) between the modeled (1971–2000) and HadCRUH observation-based (1973–2003) near-surface RH. Summer is shown in (a) and (c), winter in (b) and (d). In high latitudes, radiation is negligibly small in winter and is therefore masked out in (b). In (c) and (d), only areas with HadCRUH data available for at least 16 years out of 31 are shown (bordered by bold lines).

  • View in gallery

    Seasonal changes in incident solar radiation (%) from 1971–2000 to 2070–99 under the A1B scenario as an average of 18 GCMs: (a) summer, (b) autumn, (c) winter, and (d) spring. Areas where more than 85% of the models (at least 16 of 18 GCMs) agree on the sign of the change are hatched.

  • View in gallery

    Seasonal changes in RH (%) from 1971–2000 to 2070–99 under the A1B scenario as a mean of 7 GCMs: (a) summer, (b) autumn, (c) winter, and (d) spring. Contour interval is 2%, and the ±1% isolines are also drawn. Areas where more than 85% of the models (6 or 7 GCMs out of 7) agree on the sign of the change are hatched.

  • View in gallery

    The northern and southern European subregions used for analyzing the intervariable correlations.

  • View in gallery

    Scatter diagrams illustrating the bivariate distributions of the responses from 1971–2000 to 2070–99 under the A1B scenario for southern European summer: (a) temperature vs precipitation, (b) temperature vs solar radiation, (c) temperature vs RH, (d) precipitation vs solar radiation, (e) precipitation vs RH, and (f) solar radiation vs RH. The seven models with data available for all the four variables are marked by specific symbols [see legend in (a)]. The intervariable correlations R, based on simulations performed with 7 GCMs (in parentheses, 18 GCMs), are given on the top of the panels.

  • View in gallery

    Correlations between the temporal anomalies of (a) temperature and precipitation, (b) temperature and solar radiation, (c) temperature and RH, (d) precipitation and solar radiation, (e) precipitation and RH, and (f) solar radiation and RH calculated from the detrended annual deviations from the corresponding temporal means for the periods 1971–2000 (black bars) and 2070–99 simulated under the A1B scenario (gray bars); area averages are over southern Europe (Fig. 4) for March–May. The correlations for individual GCMs are given first, followed by the multi-GCM mean.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 279 279 86
PDF Downloads 217 217 62

Seasonal Changes in Solar Radiation and Relative Humidity in Europe in Response to Global Warming

View More View Less
  • 1 Finnish Meteorological Institute, Helsinki, Finland
© Get Permissions
Full access

Abstract

Future seasonal changes in surface incident solar radiation and relative humidity (RH) over Europe and adjacent ocean areas were assessed based on phase 3 of the Coupled Model Intercomparison Project (CMIP3) model ensemble. Under the A1B scenario, by 2070–99, summertime solar radiation is projected to increase by 5%–10% in central and southern Europe. In winter, radiation decreases in most of northern and eastern Europe by 5%–15%. RH drops in summer in the southern European inland by 8%–12%, whereas in winter a small increase of 2%–3% is projected for northeastern Europe. In spring, the change is an intermediate between those in the extreme seasons, while in autumn the patterns resemble summer. Over the northern Atlantic Ocean, RH increases in all seasons by 1%–2%. The intermodel agreement on the sign of all these shifts is good, and the patterns recur in the responses to the A2 and B1 scenarios. Substantial changes are already simulated to occur before the midcentury, for example, in summer RH decreases by more than 5% in the inner Balkan Peninsula. Projected changes in these two variables agree well and are also mainly consistent with precipitation responses both in the multimodel mean and in individual models. According to all indicators, southern European summers become more arid, while winters, in the north particularly, become moister and darker. The increasing radiation and declining RH exacerbate summertime drought in southern Europe, whereas excessive humidity in the north may, for example, inflict moisture damages in constructions.

Supplemental information related to this paper is available at the Journals Online website: http://dx.doi.org/10.1175/JCLI-D-12-00007.s1.

Corresponding author address: Kimmo Ruosteenoja, Finnish Meteorological Institute, P.O. Box 503, FI-00101 Helsinki, Finland. E-mail: kimmo.ruosteenoja@fmi.fi

Abstract

Future seasonal changes in surface incident solar radiation and relative humidity (RH) over Europe and adjacent ocean areas were assessed based on phase 3 of the Coupled Model Intercomparison Project (CMIP3) model ensemble. Under the A1B scenario, by 2070–99, summertime solar radiation is projected to increase by 5%–10% in central and southern Europe. In winter, radiation decreases in most of northern and eastern Europe by 5%–15%. RH drops in summer in the southern European inland by 8%–12%, whereas in winter a small increase of 2%–3% is projected for northeastern Europe. In spring, the change is an intermediate between those in the extreme seasons, while in autumn the patterns resemble summer. Over the northern Atlantic Ocean, RH increases in all seasons by 1%–2%. The intermodel agreement on the sign of all these shifts is good, and the patterns recur in the responses to the A2 and B1 scenarios. Substantial changes are already simulated to occur before the midcentury, for example, in summer RH decreases by more than 5% in the inner Balkan Peninsula. Projected changes in these two variables agree well and are also mainly consistent with precipitation responses both in the multimodel mean and in individual models. According to all indicators, southern European summers become more arid, while winters, in the north particularly, become moister and darker. The increasing radiation and declining RH exacerbate summertime drought in southern Europe, whereas excessive humidity in the north may, for example, inflict moisture damages in constructions.

Supplemental information related to this paper is available at the Journals Online website: http://dx.doi.org/10.1175/JCLI-D-12-00007.s1.

Corresponding author address: Kimmo Ruosteenoja, Finnish Meteorological Institute, P.O. Box 503, FI-00101 Helsinki, Finland. E-mail: kimmo.ruosteenoja@fmi.fi

1. Introduction

As a response to the increasing greenhouse gas concentrations, climate models simulate an increase in precipitation for northern Europe and a decrease for southern Europe; in central Europe, the trend is positive in winter and negative in summer (Solomon et al. 2007, their Figs. 10.9 and 11.5). Concurrently, cloudiness will increase slightly in the northern part and decrease considerably in the southern part of the continent (Solomon et al. 2007, their Fig. 10.10; Trenberth and Fasullo 2009). According to the Köppen (1936) classification, extensive areas in southern Europe will be shifted to the BS (semiarid steppe climate) or Cs (temperate Mediterranean-type climate with dry summers) climate zone by the end of the twenty-first century [Jylhä et al. (2010) provide an algorithm for determining the zone on the basis of monthly mean temperature and precipitation data].

The purpose of the present study is to find out how these reverse climatic changes in the different parts of the continent appear in the model projections for downwelling solar radiation (i.e., insolation) and relative humidity (RH). The research is based on monthly mean data from phase 3 of the Coupled Model Intercomparison Project (CMIP3) global climate models (GCMs), and the focus is on the end of the twenty-first century. In addition to Europe, adjacent ocean areas are considered.

Changes in RH and solar radiation have several practical implications. Both quantities contribute to the surface energy balance, particularly to actual and potential evaporation, and thereby influence the water balance of the soil. The scarcity of soil water increases the risk for extremely high temperatures in summer (Seneviratne et al. 2006). The availability of water is essential for the thriving of crops and natural vegetation, also impacting the ability of the soil to store carbon (e.g., Thum et al. 2011). Sunny weather with low RH favors harvesting of cereal crops, but such weather conditions likewise involve a risk for forest fire ignition (Kilpeläinen et al. 2010).

A simultaneous occurrence of high temperature and large RH affects human comfort negatively and leads to higher mortality (Fischer and Schär 2010). Strong insolation further increases the heat burden. Both the air humidity and the number of sunshine hours are taken into account in calculating the tourism climate index (Perch-Nielsen et al. 2010). Furthermore, in land areas, projected changes in solar radiation and the diurnal temperature range bear a great resemblance (Zhou et al. 2009).

In areas with humid climate, the prevailing high RH delays the drying of building materials and shortens the service life of structures exposed to weather (Viitanen et al. 2011). In northern latitudes in autumn and winter, the scarcity of visible light increases energy consumption for illumination. There is also a relationship between the darkness of winter and mental health; for example, increased suicide rates have been attributed to the low solar radiation (Ruuhela et al. 2009).

Averaged over the entire atmosphere, relative humidity is likely to remain almost unchanged (Lorenz and Deweaver 2007), whereas specific humidity would approximately follow the temperature increase according to the Clausius–Clapeyron relation. Spatially inhomogeneous changes in RH may be induced by the uneven distribution of warming, for example, the more rapid temperature increase over the continents relative to oceans. Another factor affecting the future distribution of RH is changes in the atmospheric circulation.

Several previous papers have explored changes in RH in the lower troposphere at 850–950 hPa (e.g., Rowell and Jones 2006; Fasullo 2010). For the practical applications mentioned above, however, it is essential to inspect changes in the near-surface air as these may be weaker or even opposite to those in the free atmosphere (Lorenz and Deweaver 2007). Over the subtropical oceans, in particular, RH is projected to increase near the surface, while a decrease is simulated at higher altitudes (Richter and Xie 2008). This behavior is attributed to the muted surface warming in the oceans that stabilizes the surface layer of the atmosphere, resulting in weakened vertical mixing.

Conversely, the surface air RH is anticipated to decrease over wide continental areas. Because of the slower warming of oceans (Solomon et al. 2007, their Fig. 10.8), the specific humidity of air advected from oceans to continents increases more slowly than the saturation specific humidity over land (Rowell and Jones 2006; Fasullo 2010), thereby leading to reduced RH. An earlier springtime drying of the soil in the future suppresses evapotranspiration and further reduces the RH over land.

In continents, especially in summer, changes in RH and incident solar radiation can be amplified by a positive feedback (Rowell and Jones 2006). A decrease in RH tends to restrict the formation of low and convective clouds, with the reduced cloudiness allowing more shortwave radiation to penetrate into the surface. The more intense radiation further enhances the warming and decreases the RH. This feedback process is one motivation for analyzing the future changes in solar radiation and relative humidity synchronously.

In recent history, a decreasing trend in solar radiation (dimming) was observed in Europe from the 1950s to the 1980s, followed by a partial recovery (brightening) thereafter (Wild and Schmucki 2011). Evidently, that evolution is largely explained by trends in aerosol radiative forcing. The increases observed in specific humidity are consistent with rising temperatures, while in RH no robust long-term trends have been found. This holds both for the homogenized radiosonde data (McCarthy et al. 2009; Dai et al. 2011) and the surface observations (Willett et al. 2007, 2008). In the years 2004–08, however, RH has been anomalously low in large continental areas, including southern Europe (Simmons et al. 2010). It is not clear whether this is a transient fluctuation or an indication of an inceptive long-term trend.

Future projections for solar radiation, based on Special Report on Emissions Scenarios (SRES) A1B simulations performed with 11 global models, were presented by Zhou et al. (2009). They found the annual sum of solar radiation to increase in southern Europe and decrease in northern Europe. Dubrovsky et al. (2005), using seven previous-generation (CMIP2) models forced by the 1992 Intergovernmental Panel on Climate Change (IPCC) Scenario a (IS92a), reported an increase in shortwave radiation for central Europe (Czech Republic) in summer, whereas, in other seasons, the sign of the change was uncertain.

During this century, relative humidity of the free atmosphere is projected to increase at high latitudes and in the tropical lower troposphere and to decrease in the subtropics and lower midlatitudes. This kind of zonal mean distribution has been obtained in several studies based on various sets of models and forcing scenarios (Lorenz and Deweaver 2007; Richter and Xie 2008; Wright et al. 2010). Sherwood et al. (2010) showed that, in a qualitative sense, this pattern recurs in the majority of the CMIP3 models.

Fasullo (2010) found, employing 12 CMIP3 models forced by the A1B scenario, the annual mean 950-hPa RH to decrease in central and, particularly, southern Europe, while a small increase occurred in the northern part of the continent. Fischer and Schär (2010) reported a marked decrease in summertime RH in southern Europe, up to 10%–15% by the last three decades of the twenty-first century. However, the quantity inspected in that paper was the daily minimum rather than the mean RH. Moreover, their analysis was merely based on two global models, both of which were downscaled by three regional climate models. Accordingly, their findings are not necessarily representative of a wider set of GCM simulations.

In the present study, there are four key elements that provide additional insight on the topic. First, the projections for the entire domain are given separately for all four seasons, whereas most of the previously mentioned papers only considered the annual mean or one individual season. As will be demonstrated below, the response varies substantially as a function of the season. Second, all CMIP3 models giving usable data are included in the analysis. Third, the RH data employed here represent conditions near the surface rather than in the free atmosphere. Fourth, on account of the physical reasoning presented above, as well as because of the qualitative similarity of the response patterns for both variables (discussed below), it is meaningful to explore projected changes in solar radiation and RH concurrently, both in the multimodel mean (sections 3a3c) and in individual models (section 3d). In addition, the covariability of different climate quantities (surface temperature, precipitation, solar radiation, and RH) in interannual variations is briefly discussed in section 4.

2. Climate model data

Future changes in solar radiation and near-surface relative humidity were studied here based on global atmosphere–ocean GCM simulations archived in the CMIP3 database. We chose to use data from global models rather than finer-resolution regional climate models (RCMs) to better cover the range of model uncertainty. Climate response in RCM simulations is strongly influenced by the global climate model giving the boundary forcing, and RCM simulations for Europe are only available for a limited number of driving GCMs (e.g., Fischer et al. 2011). If these few driving models represented a significantly different sensitivity to greenhouse gas forcing than the wider set of CMIP3 GCMs, the resulting RCM-based estimate would be biased severely. Moreover, as will be shown below, the simulated changes in solar radiation and relative humidity are dominated by very large-scale features, which provides a posteriori justification for the use of global models.

We chiefly examine responses to the SRES A1B scenario for the period 2070–99, relative to the baseline period 1971–2000. The 30-yr averaging period is in line with the recommendation of WMO (1989) to calculate climatological standard normals. We focus on this rather distant future period in order to ensure a maximum signal-to-noise ratio. Moreover, Wild and Schmucki (2011) suggest that, because of an imperfect representation of aerosol processes, CMIP3 models are better suited for predicting long-term than short-term changes in solar radiation; in the more distant future, greenhouse gas forcing distinctly overwhelms the aerosol forcing. Earlier future periods, as well as responses to the A2 and B1 scenarios, are discussed briefly in section 3c and in the supplemental material.

The CMIP3 archive includes data from 23 GCMs, but four of these were excluded for the criteria discussed in Jylhä et al. (2010). In addition, the Commonwealth Scientific and Industrial Research Organisation Mark, version 3.0 (CSIRO Mk3.0), model was not included because there were artifacts in the solar radiation data in northern latitudes in winter. The simulations for the A1B scenario were available for all the remaining 18 GCMs, but for B1 (A2) data were missing for one (two) model(s) (Table 1). To make the multimodel mean responses comparable for all scenarios, surrogate data were produced for these missing runs by employing the pattern-scaling technique detailed in Ruosteenoja et al. (2007).

Table 1.

Models employed in this study. “Resolution” refers to the meridional times zonal grid size in degrees and “Scenarios” to the SRES scenarios for which model runs have been performed. Scenarios for which surrogate data were created by pattern scaling are denoted by an asterisk. “SR” and “RH” express the availability of usable data for solar radiation and relative humidity, respectively. For more information on the models, see Table 8.1 of Solomon et al. (2007, and references therein).

Table 1.

For RH, we used data from the lowermost atmospheric level available in the CMIP3 datasets. For both versions of the Canadian Centre for Climate Modelling and Analysis (CCCma) Coupled Global Climate Model, version 3.1 (CGCM3.1), that level resides at σ = pressure/surface pressure = 0.995, approximately corresponding to a 40-m height. For the other models, we employed RH at 1000 hPa. Unfortunately, data covering the entire domain only existed for five of these remaining GCMs. Note that, while the 1000-hPa level resides below the surface in elevated areas, the 1000-hPa RH for these five models is actually taken from the lowest atmospheric model level in these regions (instead of extrapolating the RH below the surface, which might produce spurious results).1 Therefore, the RH projections were based on seven models in total (Table 1).

The model ensemble considered in this study is not completely independent. Six research centers have provided two parallel model versions (Table 1), and many of the remaining models exploit partially similar numerical schemes or parameterization methods (Pennell and Reichler 2011). Hence, the actual degrees of freedom in the ensemble are substantially smaller than the total number of GCMs.

The output of the GCMs was interpolated onto a common 2.5° × 2.5° grid to calculate the averages over the 18 (for radiation, temperature, and precipitation) or 7 (for humidity) models. In calculating the means, every GCM was given an equal weight. Nonequal weighting would be justifiable only if we had compelling quantitative information on the performance of individual models in simulating future climate (Weigel et al. 2010).

While a comprehensive evaluation of the performance of the models in simulating the present climate is beyond the scope of the present work, it is useful to have a general idea of the quality of the model data. For this end, downwelling solar radiation and RH in the CMIP3 baseline-period ensemble mean are compared with the corresponding quantities derived from the International Satellite Cloud Climatology Project flux data (ISCCP FD) for the period 1983–2004 (Zhang et al. 2004) and observational Hadley Centre and Climatic Research Unit global surface humidity dataset (HadCRUH) for the years 1973–2003 (Willett et al. 2008). Note that, because of data availability, the observation periods do not exactly correspond to the baseline period for model data, but the time differences are not so large that the ongoing climate change would notably impact the comparison.

In summer, the models appear to simulate a too “desertlike” climate for eastern and southern Europe, with an overestimation of solar radiation by ~10% (Fig. 1a) and an underestimation of RH by 10%–20% (Fig. 1c; all differences and changes in RH are expressed in this paper in absolute terms, that is, in percentage points). In northernmost Europe, by contrast, the models tend to produce too humid conditions: an overestimation of RH by up to 5% and an underestimation of solar radiation by up to 10%. Over oceans, RH is generally simulated well. In solar radiation, there seems to be a deficit of ~10% in the western Atlantic between 40° and 50°N.

Fig. 1.
Fig. 1.

(a),(b) The deviation (%) of the baseline-period (1971–2000) multimodel mean incident solar radiation from the ISCCP FD satellite-derived estimate for the years 1983–2004. (c),(d) The difference (%) between the modeled (1971–2000) and HadCRUH observation-based (1973–2003) near-surface RH. Summer is shown in (a) and (c), winter in (b) and (d). In high latitudes, radiation is negligibly small in winter and is therefore masked out in (b). In (c) and (d), only areas with HadCRUH data available for at least 16 years out of 31 are shown (bordered by bold lines).

Citation: Journal of Climate 26, 8; 10.1175/JCLI-D-12-00007.1

In winter, relative model errors in RH are fairly small (Fig. 1d). In central and northern Europe, a trivial reason for that is the prevailing high RH, 85%–90% in absolute terms. Modeled solar radiation mostly falls below its counterpart in the observational analysis in the continents, in northeastern Europe locally by more than 20% (Fig. 1b). In this area, the magnitude of surface insolation is certainly small, both in the model simulations and observations. Conversely, in large areas of the Atlantic Ocean, the modeled radiation exceeds the satellite analysis by 10%–20%.

In spring (not shown), the differences between the multimodel mean and satellite-based solar radiation were typically of the same order of magnitude as in summer; in autumn, they were close to those in winter or somewhat larger. For humidity, the differences were fairly small in autumn. In spring, by contrast, in wide areas in Europe, the modeled RH exceeded the observational estimate by 5%–10%.

To some extent, the model versus observation differences apparent in Fig. 1 may reflect the inaccuracy of the observational data. For example, the HadCRUH data are affected by the temporally varying and, in some areas, quite low station density, and the RH represents different heights above the surface in the models and the observations (K. Willett 2012, personal communication). Likewise, there are uncertainties in the ISCCP FD dataset because of both the radiative transfer model used in producing the flux estimates and inaccuracies in the input data (i.e., ISCCP cloud fields and other observation-based datasets). Zhang et al. (2004) reported an overall bias of 8.8 W m−2 in the downwelling solar flux relative to Global Energy Balance Archive data but only 2.0 W m−2 relative to the more accurate but geographically less extensive Baseline Surface Radiation Network measurements, with the largest regional errors in the tropics.

Wild (2008) compared incident solar radiation simulated by 14 CMIP3 models with station observation data. Globally, the GCMs overestimated surface solar radiation by 6 W m−2, or about 3%. In several models, the bias was positive in low and negative in high latitudes. Unfortunately, the radiation observations are geographically too scattered to be used for composing a map representation like Fig. 1.

3. Future projections

a. Solar radiation

Multimodel mean projections for incident solar radiation are depicted in Fig. 2. To discern the northern areas in winter, where changes are small in absolute terms, the responses are presented in percent relative to the baseline-period mean.

Fig. 2.
Fig. 2.

Seasonal changes in incident solar radiation (%) from 1971–2000 to 2070–99 under the A1B scenario as an average of 18 GCMs: (a) summer, (b) autumn, (c) winter, and (d) spring. Areas where more than 85% of the models (at least 16 of 18 GCMs) agree on the sign of the change are hatched.

Citation: Journal of Climate 26, 8; 10.1175/JCLI-D-12-00007.1

In summer, solar radiation is simulated to increase in most of Europe. This is especially true for central and southern Europe, where the increase reaches 5%–10% in extensive areas. Only in the northernmost part of the continent is there a small reduction. In winter, solar radiation will decrease everywhere apart from the Mediterranean countries; in central, northern, and eastern Europe, the reduction ranges from 5% to 15%. In autumn, the pattern is qualitatively similar to that in summer. Conversely, spring resembles winter, although the magnitude of the decline is smaller and the contour of zero change is located farther in the north.

In the areas where the response distinctly exceeds 5%, a large majority of the GCMs agree on the sign of the change (Fig. 2). In southern Europe in summer and in northern Europe in winter, even all of the 18 GCMs simulated a response of similar sign. In addition, the significance of the multimodel mean projection was evaluated using the t test. The areas where more than 85% of the GCMs agree on the direction of the trend (Fig. 2) roughly coincide with the 0.1% significance level. Nonetheless, the interpretation of the t test is not straightforward because the 18 models cannot be considered as a random independent sample, with the actual degrees of freedom being substantially smaller than the number of the models (Pennell and Reichler 2011).

Over the Atlantic, relative changes in solar radiation are generally less than ±5%. For the Arctic Ocean, by contrast, all models simulate a strong decrease of incident radiation in all seasons and in autumn up to 20%–30%. This phenomenon has been attributed to increasing cloudiness (Sorteberg et al. 2007), but the retreat of sea ice may contribute as well. At present, high-latitude sea areas are mostly ice covered, and the occurrence of snow-covered sea ice favors multiple reflection of shortwave radiation between the surface and the cloud cover. As the sea ice melts in the future, incident radiation thus tends to be reduced. A large decline in radiation is likewise projected for the Barents Sea, particularly in winter, but the agreement among the models is low (Fig. 2c). During the baseline period, many models simulate sea ice even for this area, although in reality the sea is mostly ice-free (Solomon et al. 2007, their Fig. 8.10). Thus, the strong decline in incident radiation may largely be a modeling artifact, and the discrepancy among the models reflects their different simulations for the baseline-period ice conditions.

In absolute terms, the decline of solar radiation in northern European winter is fairly small, a 15% reduction at 60°N corresponding to less than 3 W m−2 in the seasonal mean. Notwithstanding, owing to the low initial level, even such a small decline may be detrimental (see section 5). For comparison, the 10% increase projected for central Europe in summer is equivalent to more than 20 W m−2.

b. Relative humidity

Projections for RH are shown in Fig. 3. In summer, changes in RH are minor in northern Europe, while in the inland areas of southern Europe, RH drops considerably, in some places by more than 10%. In winter, RH is projected to increase in northeastern Europe by 2%–3%. Even an increase of that magnitude is remarkable as the air is already close to the saturation point at present, and the saturation deficit would thus be substantially reduced. In the intermediate seasons, especially in autumn, the pattern resembles that of summer, but the amplitude of the response is smaller.

Fig. 3.
Fig. 3.

Seasonal changes in RH (%) from 1971–2000 to 2070–99 under the A1B scenario as a mean of 7 GCMs: (a) summer, (b) autumn, (c) winter, and (d) spring. Contour interval is 2%, and the ±1% isolines are also drawn. Areas where more than 85% of the models (6 or 7 GCMs out of 7) agree on the sign of the change are hatched.

Citation: Journal of Climate 26, 8; 10.1175/JCLI-D-12-00007.1

Over the northern Atlantic, RH increases slightly in all seasons. This phenomenon can be attributed to the stabilization of the lowermost atmosphere, which suppresses ventilation near the surface (see section 1). The increase is most distinct in the areas to the southwest of Iceland between 50° and 60°N. Owing to the weakening oceanic thermohaline circulation, this area is characterized by negligible surface warming (Solomon et al. 2007, their Fig. 10.8). An opposite, arguably spurious, phenomenon occurs in high-latitude sea areas between mainland Norway and Spitsbergen, where the retreat of sea ice in the models destabilizes the boundary layer. As stated in section 3a, these ocean areas are actually ice-free. Accordingly, such a strong reduction of RH does not necessarily occur exactly in that area. We presume that a reduction in RH associated with sea ice loss would be more likely to occur farther in the north because of the more northerly location of the observed sea ice edge relative to the model simulations.

The models agree well on the main features of the distribution: an increase of RH in northeastern Europe in winter and a notable drop in the south in summer. In addition, the increasing RH over the Atlantic repeats itself in the majority of the models, although the amplitude of the change is small, generally 1%–2%. Such a high consistency indicates that the physical processes behind the increasing RH are quite robust.

In Fig. 3, the areas with more than 85% of the models agreeing on the sign of the change approximately corresponds to the 5% significance level in the t test. The significance is lower than that for solar radiation since the number of GCMs analyzed is smaller. Caveats concerning the interpretation of the t test were discussed in section 3a.

c. Other scenarios and time intervals

Projections for solar radiation and RH were calculated for two other SRES scenarios (A2 and B1) and three partially overlapping earlier 30-yr periods (2010–39, 2020–49, and 2040–69). The geographical distribution of the change proved to be very similar to that in Figs. 2 and 3, while the amplitude of the response increased monotonically as a function of the strength of forcing (supplemental Figs. S1 and S2). Hence, the key findings of (i) the northern European winters becoming darker and more humid and (ii) the southern European summers becoming sunnier and drier in the future are robust. By the period 2070–99, according to the low-emission B1 scenario, the maximum darkening over the European continent in winter will reach 11% and the maximum drying in summer will reach 7%. In the high-emission A2 scenario, the corresponding maximum reductions are 18% for solar radiation in winter and 13% for RH in summer.

Such a scalability property, that is, a similar geographical pattern of the response recurring in simulations with different forcing scenarios, has previously been found to hold for various climate quantities, for example, temperature and precipitation (Ruosteenoja et al. 2007, and references therein). Figure 10.8 of Solomon et al. (2007) vividly illustrates the scalability of the CMIP3 model temperature responses.

For practical applications, it is important to note that we do not have to wait until the end of this century in order to experience notable changes in solar radiation and humidity. As early as by the period 2020–49, a 4%–5% drop in RH is predicted for the inner Balkan Peninsula in summer (Fig. S2, top-left panel). At the same time, in places in northern Europe, winter solar radiation would decrease by more than 5% (Fig. S1, top-right panel).

Future changes in solar radiation do not arise from meteorological factors, such as changes in cloudiness, alone but are also affected by aerosol forcing and absorption of solar radiation by water vapor. In the present study, the only scenario in which aerosol forcing does not weaken by 2100 is A2, and in this scenario, because of the most intense warming, the atmospheric water vapor content is likewise larger than in the two other scenarios. Even so, the pattern of change is rather similar to that of the other scenarios (Fig. S1, middle panels). This indicates that the projected change in solar radiation is predominantly of meteorological origin, with the contributions of aerosols and atmospheric water vapor being of secondary importance.

d. Intervariable correlations among individual model projections

As can be seen in Figs. 2 and 3, in the European continent the seasonal changes in incident solar radiation and RH are almost in reverse phase. In this section, we inspect whether this kind of relationship holds for individual models. In addition, we examine the compatibility of these projections with the modeled changes in temperature and precipitation. For this purpose, we performed an intermodel correlation analysis for the projected changes of each pair of variables. Our focus is on the A1B scenario, for which simulations have been performed with all models.

The spatial patterns of the bivariate correlations across the model projections proved to be fairly noisy, and we therefore view the joint distributions of changes averaged spatially over southern and northern Europe (Fig. 4). For temperature and RH, the responses are expressed in absolute terms. In contrast, the area-averaged precipitation and solar radiation responses are given in percent, by normalizing the spatially averaged changes simulated by each model by the corresponding baseline-period means.

Fig. 4.
Fig. 4.

The northern and southern European subregions used for analyzing the intervariable correlations.

Citation: Journal of Climate 26, 8; 10.1175/JCLI-D-12-00007.1

As an example, the bivariate distributions of the responses for the southern European summer are shown in Fig. 5. All models project an increase in temperature and solar radiation and a decline in precipitation and RH (with a single exception for precipitation; see below). In general, the models projecting the strongest warming also have a tendency to simulate relatively large increases in solar radiation and reductions in precipitation and RH. This especially holds true for the seven-model ensemble, for which data are available for all variables. In this subensemble, the decreases in RH and precipitation likewise correlate, as do increases in solar radiation and decreases in RH and precipitation. The magnitudes of the correlations range from 0.82 to 0.95, thus exceeding the 5% significance limit. In the full 18-model ensemble, correlations are weaker but are of the same sign as in the small subensemble.

Fig. 5.
Fig. 5.

Scatter diagrams illustrating the bivariate distributions of the responses from 1971–2000 to 2070–99 under the A1B scenario for southern European summer: (a) temperature vs precipitation, (b) temperature vs solar radiation, (c) temperature vs RH, (d) precipitation vs solar radiation, (e) precipitation vs RH, and (f) solar radiation vs RH. The seven models with data available for all the four variables are marked by specific symbols [see legend in (a)]. The intervariable correlations R, based on simulations performed with 7 GCMs (in parentheses, 18 GCMs), are given on the top of the panels.

Citation: Journal of Climate 26, 8; 10.1175/JCLI-D-12-00007.1

In the scatterplot of temperature and precipitation changes (Fig. 5a), the strikingly low correlation in the full ensemble is largely explained by two outliers, the two versions of the MIROC3.2 model: the high-resolution version projects a 5.6% decrease and the medium-resolution version a 1.5% increase for precipitation. For solar radiation, by contrast, the simulations of these two models fit in the distribution far better.

To conclude, Fig. 5 shows that the projected changes in the different variables tend to be consistent. All models show an aridification of the southern European summer climate, but the strength of this response varies among the models. In some models, especially in the UKMO-HadCM3, increasing the greenhouse gas forcing leads to a strong reduction of air humidity, precipitation, and cloudiness. The resulting large increase in solar radiation and the reduction of water available for evaporation further intensify the warming. Some other models behave more moderately in this respect.

The correlations between the projected seasonal changes of the different variables for both subregions are presented in Table 2. The main conclusions are as follows. First, in summer, the intervariable correlations are qualitatively similar in both subregions, albeit generally weaker in the north than in the south. Second, changes in RH and solar radiation correlate negatively, apart from winter in the north. Third, in southern Europe the projections for RH and solar radiation have a statistically significant correlation with temperature changes in autumn (solar radiation only in the small ensemble) and with precipitation changes in winter and spring, with the signs of R being the same as in summer.

Table 2.

Correlation coefficients between the projected seasonal changes of the different variables in the model ensemble for the northern and southern European subregions. Correlations based on the ensemble of the 7 GCMs providing data for all variables (Table 1) are given first, and those derived from the full 18-model ensemble are given in parentheses. Correlations that are significant at the 5% level according to a two-tailed t test (R > 0.75 for 7 models and R > 0.47 for 18 models) are shown in boldface (note that the significance limits should not be interpreted literally since the model ensemble is not entirely independent; see section 2).

Table 2.

In northern Europe in winter and, to some extent, also in the intermediate seasons, models with the strongest temperature response tend to increase the precipitation and reduce the solar radiation more than the remaining GCMs. One explanation for this is the increasing moisture-holding capacity of the air in a warming climate, which enables more abundant precipitation and optically thicker clouds. In winter, the intensity of solar radiation is weak in high latitudes, and clear conditions favor low temperatures. Since the air is close to the saturation state, the RH response varies very little among the models, from 0.9% to 2.4%. Accordingly, the RH response is only weakly related to the other variables.

The projections of solar radiation and RH were further compared with the simulated changes of sea level pressure. In the multimodel mean in winter, the pressure increases in southern and decreases in northern Europe (cf. Fig. 10.9 of Solomon et al. 2007), which agrees well with the present projections for insolation (Fig. 2) and RH (Fig. 3). In summer, pressure is simulated to decrease everywhere, with the exception of northwestern Europe. This induces a more easterly time-mean geostrophic wind for southern Europe in the future, and the enhanced advection of continental air presumably contributes to the drying. Conversely, in northern Europe, a west-northwesterly inflow of maritime air would become more common.

The projected changes in RH and insolation in individual models were only weakly related to changes in sea level pressure in northern Europe and in southern Europe in summer. However, a distinct relationship was revealed for southern Europe in winter: the three GCMs simulating the largest increase in pressure (CNRM-CM3, BCCR-BCM2.0, and ECHAM5/MPI-OM) produced the most positive insolation response. Moreover, all three GCMs belonged to the five models in total, in which the RH response was negative.

A caveat concerning the correlations between the responses of RH and the other variables is the rather small sample size: only seven models could be used in the analysis, and moreover, two of them were versions of a single model (CGCM3.1) with different resolutions. The correlations between the three other variables (temperature, precipitation, and solar radiation) showed fairly large quantitative differences depending on whether 7 or 18 GCMs were used in the analysis. In particular, in the wider set of GCMs, no correlations with an absolute value larger than 0.9 exist. Accordingly, the correlations between the changes in RH and the other variables reported in Table 2 should not be considered very robust, although the dependencies are likely to be valid in a qualitative sense.

4. Coherence of interannual variations of different variables

The observation-based analysis of Trenberth and Shea (2005) revealed that the monthly mean anomalies of surface temperature and precipitation in Europe correlate negatively in summer and, apart from the southeastern part of the continent, positively in winter. The NCAR-CCSM3 model reproduced this distribution correctly in a qualitative sense, although the modeled correlations were higher than the observed ones. In this section, we explore how the interannual variations in the different climate quantities (temperature, precipitation, incident solar radiation, and RH) are related to one another in the seven-model ensemble, providing data for all these variables, and whether the dependencies are preserved in the future. We first calculated, separately for every year of the baseline period 1971–2000 and the scenario (A1B) period 2070–99, detrended anomalies of the seasonal means from the corresponding tridecadal averages. The anomalies were then used to calculate correlations between the time series of the variables.

The correlations, averaged over the southern and northern European subregions and the seven models, are listed in Table 3. There is a firm positive correlation between the anomalies of precipitation and RH, and both these variables correlate distinctly negatively with the solar radiation anomalies. Accordingly, as expected, anomalously rainy seasons are typically cloudy and humid. The correlations between the anomalies of temperature and the other variables depend on the area and season. Warm summers are mainly related to scanty precipitation, low RH, and abundant solar radiation, most notably in southern Europe. In winter, by contrast, warm weather conditions tend to coincide with high precipitation and RH and low solar radiation, particularly in northern Europe. In winter, the radiation balance at the surface is negative, with moist and cloudy conditions reducing the longwave radiative cooling. Conversely, winters with a continental circulation pattern typically have low cloudiness, precipitation, and humidity and weak wind, that is, conditions favoring radiative cooling. In the intermediate seasons, the correlations between temperature and the other variables are generally lower than in summer and winter, with the signs being similar to those in summer in the south and to winter in the north.

Table 3.

Seasonal correlation coefficients between the detrended interannual variations of surface temperature, precipitation, incident solar radiation, and RH during the baseline period 1971–2000: the seven-GCM means of the gridpoint correlations averaged over the southern and northern European subregions (see Fig. 4). The differences between the period 2070–99 under the A1B scenario and the baseline period are given in parentheses. Changes for which six GCMs at the minimum agree on the sign are italicized, and those with all the seven models agreeing are shown in boldface.

Table 3.

The spatial patterns of the multimodel mean anomaly correlations for summer and winter are given in supplemental Figs. S3 and S4. The distributions are dominated by large-scale features, and the correlations are generally most evident over the continents. Over the oceans, the surface air temperature and RH are strongly constrained by the sea surface temperature, which acts to attenuate the influence of varying synoptic conditions.

Relative to the anomaly correlations calculated for the baseline period, the projected changes in the correlations are typically quite noisy (patterns not shown) and relatively modest in magnitude, and the agreement on the sign of the change among the models is low (Table 3). The most prominent exception is the southern European spring. In that season, all of the seven GCMs project increases for the absolute values of the correlations of temperature versus precipitation, solar radiation, and RH, and the correlation between the solar radiation and RH increases in all models except one (Fig. 6). All these changes shift the correlations toward the state prevailing in the baseline climate in summer. We hypothesize that this might be related to an earlier drying of the soil in spring, along with an earlier snowmelt in elevated areas. Thus, in years with anomalously large incident radiation, the temperatures can rise and air humidity can be reduced without the evaporation from the soil being able to effectively damp the anomalies.

Fig. 6.
Fig. 6.

Correlations between the temporal anomalies of (a) temperature and precipitation, (b) temperature and solar radiation, (c) temperature and RH, (d) precipitation and solar radiation, (e) precipitation and RH, and (f) solar radiation and RH calculated from the detrended annual deviations from the corresponding temporal means for the periods 1971–2000 (black bars) and 2070–99 simulated under the A1B scenario (gray bars); area averages are over southern Europe (Fig. 4) for March–May. The correlations for individual GCMs are given first, followed by the multi-GCM mean.

Citation: Journal of Climate 26, 8; 10.1175/JCLI-D-12-00007.1

The present findings concerning the temporal covariability of the different variables should be regarded as tentative, as the number of models analyzed was small. Another interesting issue is the verification of the model simulations against the observed anomaly correlations, a theme that, however, is beyond the scope of the present work concentrating on future projections.

5. Conclusions

The geographical and seasonal distributions of the projections for the different climate variables are found to be well in line with each other. In southern Europe, all indicators exhibit a tendency toward more arid climatic conditions, with the largest changes in summer: solar radiation is simulated to increase and precipitation and RH to decrease. In northern Europe, the climate is becoming moister and darker, primarily in winter: solar radiation is reduced and RH and precipitation are increasing. In central Europe, moister conditions are projected for winter and increasing aridity for summer.

The consistency of the projections for the various climate quantities and the good agreement among the models on the sign of the change lend support to the credibility of the present findings. Even the behavior of individual models is compatible, particularly in summer: those models that project the largest increase in temperature also have a tendency to increase the solar radiation and reduce the precipitation and RH more than models with a more modest temperature response. In northern European winter, the relation is the opposite. The geographical patterns of the RH and solar radiation responses are very similar for all three SRES scenarios and the four future time spans studied.

Of course, the models are nonperfect representations of reality (as seen, for example, in the biases relative to observation-based datasets in Fig. 1). While the relationship between model biases for present climate and simulated climate changes is a complex issue, Räisänen (2007) showed, for temperature and precipitation, that the future projections vary much less than the simulated baseline-period means. Accordingly, model simulations can provide valuable information about the future changes, although there are deficiencies in the simulation of the present climate. Even so, as the models are continuously evolving, the present predictions cannot be regarded as any final truth in a quantitative sense; this caveat holds especially for RH, for which usable data were available for seven models only. Moreover, a specific problem of the current GCM simulations that may influence the future response is that many models overestimate the sea ice coverage in subpolar oceans. There, this bias may lead to a spurious reduction of RH and incident radiation.

In the future, the growth of the carbon dioxide concentration promotes photosynthesis and allows the plants to keep their stomata partially closed. The resulting increment in stomatal resistance tends to reduce transpiration, which further decreases air humidity and cloudiness and increases solar radiation (Andrews et al. 2011). In the present set of CMIP3 models, this phenomenon was not taken into account. As a result, our estimates of the drying of the southern European summers may be conservative.

Besides the temporal mean responses, we made a preliminary account of the covariability of the interannual temperature, precipitation, RH, and solar radiation fluctuations. In general, seasons with anomalously large precipitation proved to coincide with a high RH and scarce insolation. In summer, temperature anomalies correlated positively with solar radiation and negatively with RH and precipitation. In winter, the situation was reversed, particularly in northern Europe. Qualitatively, similar relationships were found in the projections of future climate. Still the findings, based on seven GCMs, need to be reassessed in the future using a larger number on models.

If the present scenarios are realized, projected changes in RH (under the A1B scenario, up to −12% in summer) and solar radiation (up to −15% in winter and +10% in summer) are so pronounced that they will have severe implications. In southern Europe, the increasing solar radiation and declining RH enhance potential evapotranspiration, further complicating the problems with water availability caused by the decreasing precipitation, lengthening dry spells (Orlowsky and Seneviratne 2012; Lehtonen et al. 2012, manuscript submitted to Int. J. Climatol.), and rising temperatures. Increasing insolation also aggravates the heat stress, although, in this respect, the lowering RH can have a mitigating effect. As an illustration of positive influences, the potential for solar energy production will improve.

In northern Europe, the higher RH predicted for winter, in conjunction with the more abundant precipitation, keeps the buildings wet for longer periods, exacerbating the risk for moisture damages. As the winters turn increasingly dark because of the reduction of solar radiation and the shortwave-reflecting snow cover, mental health problems such as the seasonal affective disorder may become more widespread.

Acknowledgments

The climate model simulations were downloaded from the CMIP3 data archive (ftp-esg.ucllnl.org). ISCCP FD data were obtained from the NASA Goddard Institute for Space Studies (isccp.giss.nasa.gov). Dr. Kate Willett at the U.K. Met Office is acknowledged for providing the observation-based HadCRUH data for the years 1973–83. This work belongs to the REFI and ACCLIM II projects that are financed by Sitra (the Finnish Innovation Fund) and by the Finnish Ministries of Transport and Communications, the Environment, and Agriculture and Forestry. Ingo Bethke, Michel Déqué, Evgeny Volodin, and Uwe Schulzweida are thanked for information regarding the treatment of near-surface RH in model output for the BCCR-BCM2.0, CNRM-CM3, INM-CM3.0, and ECHAM5/MPI-OM, respectively. Dr. Kirsti Jylhä and, in particular, the referees are acknowledged for their useful suggestions that have significantly improved the paper.

REFERENCES

  • Andrews, T., , M. Doutriaux-Boucher, , O. Boucher, , and P. M. Forster, 2011: A regional and global analysis of carbon dioxide physiological forcing and its impacts on climate. Climate Dyn., 36, 783792, doi:10.1007/s00382-010-0742-1.

    • Search Google Scholar
    • Export Citation
  • Dai, A., , J. Wang, , P. W. Thorne, , D. E. Parker, , L. Haimberger, , and X. L. Wang, 2011: A new approach to homogenize daily radiosonde humidity data. J. Climate, 24, 965991.

    • Search Google Scholar
    • Export Citation
  • Dubrovsky, M., , I. Nemesova, , and J. Kalvova, 2005: Uncertainties in climate change scenarios for the Czech Republic. Climate Res., 29, 139156.

    • Search Google Scholar
    • Export Citation
  • Fasullo, J. T., 2010: Robust land–ocean contrast in energy and water cycle feedbacks. J. Climate, 23, 46774693.

  • Fischer, A. M., , A. P. Weigel, , C. M. Buser, , R. Knutti, , H. R. Künsch, , M. A. Liniger, , C. Schär, , and C. Appenzeller, 2011: Climate change projections for Switzerland based on a Bayesian multi-model approach. Int. J. Climatol., 32, 23482371, doi:10.1002/joc.3396.

    • Search Google Scholar
    • Export Citation
  • Fischer, E. M., , and C. Schär, 2010: Consistent geographical patterns of changes in high-impact European heatwaves. Nat. Geosci., 3, 398403, doi:10.1038/NGEO866.

    • Search Google Scholar
    • Export Citation
  • Jylhä, K., , H. Tuomenvirta, , K. Ruosteenoja, , H. Niemi-Hugaerts, , K. Keisu, , and J. A. Karhu, 2010: Observed and projected future shifts of climate zones in Europe and their use to visualize climate change information. Wea. Climate Soc., 2, 148167.

    • Search Google Scholar
    • Export Citation
  • Kilpeläinen, A., , S. Kellomäki, , H. Strandman, , and A. Venäläinen, 2010: Climate change impacts on forest fire potential in boreal conditions in Finland. Climatic Change, 103, 383398.

    • Search Google Scholar
    • Export Citation
  • Köppen, W. P., 1936: Das geographische System der Klimate. Handbuch der Klimatologie, W. P. Köppen and R. Geiger, Eds., Vol. I, Gebrüder Borntraeger, C1–C44.

  • Lorenz, D. J., , and E. T. Deweaver, 2007: The response of the extratropical hydrological cycle to global warming. J. Climate, 20, 34703484.

    • Search Google Scholar
    • Export Citation
  • McCarthy, M. P., , P. W. Thorne, , and H. A. Titchner, 2009: An analysis of tropospheric humidity trends from radiosondes. J. Climate, 22, 58205838.

    • Search Google Scholar
    • Export Citation
  • Orlowsky, B., , and S. I. Seneviratne, 2012: Global changes in extreme events: Regional and seasonal dimension. Climatic Change, 110, 669696, doi:10.1007/s10584-011-0122-9.

    • Search Google Scholar
    • Export Citation
  • Pennell, C., , and T. Reichler, 2011: On the effective number of climate models. J. Climate, 24, 23582367.

  • Perch-Nielsen, S. L., , B. Amelung, , and R. Knutti, 2010: Future climate resources for tourism in Europe based on the daily Tourism Climate Index. Climatic Change, 103, 363381.

    • Search Google Scholar
    • Export Citation
  • Räisänen, J., 2007: How reliable are climate models? Tellus, 59A, 229, doi:10.1111/j.1600-0870.2006.00211.x.

  • Richter, I., , and S.-P. Xie, 2008: Muted precipitation increase in global warming simulations: A surface evaporation perspective. J. Geophys. Res.,113, D24118, doi:10.1029/2008JD010561.

  • Rowell, D. P., , and R. G. Jones, 2006: Causes and uncertainty of future summer drying over Europe. Climate Dyn., 27, 281299, doi:10.1007/s00382-006-0125-9.

    • Search Google Scholar
    • Export Citation
  • 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, 193208, doi:10.1007/s10584-006-9222-3.

    • Search Google Scholar
    • Export Citation
  • Ruuhela, R., , L. Hiltunen, , A. Venäläinen, , P. Pirinen, , and T. Partonen, 2009: Climate impact on suicide rates in Finland from 1971 to 2003. Int. J. Biometeor., 53, 167175, doi:10.1007/s00484-008-0200-5.

    • Search Google Scholar
    • Export Citation
  • Seneviratne, S. I., , D. Lüthi, , M. Litschi, , and C. Schär, 2006: Land–atmosphere coupling and climate change in Europe. Nature, 443, 205209, doi:10.1038/nature05095.

    • Search Google Scholar
    • Export Citation
  • Sherwood, S. C., , W. Ingram, , Y. Tsushima, , M. Satoh, , M. Roberts, , P. L. Vidale, , and P. A. O’Gorman, 2010: Relative humidity changes in a warmer climate. J. Geophys. Res.,115, D09104, doi:10.1029/2009JD012585.

  • Simmons, A. J., , K. M. Willett, , P. D. Jones, , P. W. Thorne, , and D. P. Dee, 2010: Low-frequency variations in surface atmospheric humidity, temperature, and precipitation: Inferences from reanalyses and monthly gridded observational data sets. J. Geophys. Res.,115, D01110, doi:10.1029/2009JD012442.

  • 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.

  • Sorteberg, A., , V. Kattsov, , J. E. Walsh, , and T. Pavlova, 2007: The Arctic surface energy budget as simulated with the IPCC AR4 AOGCMs. Climate Dyn., 29, 131156, doi:10.1007/s00382-006-0222-9.

    • Search Google Scholar
    • Export Citation
  • Thum, T., and Coauthors, 2011: Soil carbon model alternatives for ECHAM5/JSBACH climate model: Evaluation and impacts on global carbon cycle estimates. J. Geophys. Res.,116, G02028, doi:10.1029/2010JG001612.

  • Trenberth, K. E., , and D. J. Shea, 2005: Relationships between precipitation and surface temperature. Geophys. Res. Lett.,32, L14703, doi:10.1029/2005GL022760.

  • Trenberth, K. E., , and J. T. Fasullo, 2009: Global warming due to increased absorbed solar radiation. Geophys. Res. Lett.,36, L07706, doi:10.1029/2009GL037527.

  • Viitanen, H., , T. Toratti, , and L. Makkonen, 2011: Development of service life model for wooden structures. Proc. 12th Int. Conf. on Durability of Building Materials and Components, Vol. 1, Porto, Portugul, Faculty of Engineering, University of Porto, 495–502.

  • Weigel, A. P., , R. Knutti, , M. A. Liniger, , and C. Appenzeller, 2010: Risks of model weighting in multimodel climate projections. J. Climate, 23, 41754191.

    • Search Google Scholar
    • Export Citation
  • Wild, M., 2008: Short-wave and long-wave surface radiation budgets in GCMs: A review based on the IPCC-AR4/CMIP3 models. Tellus, 60A, 932945.

    • Search Google Scholar
    • Export Citation
  • Wild, M., , and E. Schmucki, 2011: Assessment of global dimming and brightening in IPCC-AR4/CMIP3 models and ERA40. Climate Dyn., 37, 16711688, doi:10.1007/s00382-010-0939-3.

    • Search Google Scholar
    • Export Citation
  • Willett, K. M., , N. P. Gillett, , P. D. Jones, , and P. W. Thorne, 2007: Attribution of observed surface humidity changes to human influence. Nature, 449, 710713, doi:10.1038/nature06207.

    • Search Google Scholar
    • Export Citation
  • Willett, K. M., , P. D. Jones, , N. P. Gillett, , and P. W. Thorne, 2008: Recent changes in surface humidity: Development of the HadCRUH dataset. J. Climate, 21, 53645383.

    • Search Google Scholar
    • Export Citation
  • WMO, 1989: Calculation of monthly and annual 30-year standard normals. WCDP-10/WMO/TD-341, 12 pp. [Available online at http://www.inmet.gov.br/html/clima/OMM_WCDP_N10.pdf.]

  • Wright, J. S., , A. Sobel, , and J. Galewsky, 2010: Diagnosis of zonal mean relative humidity changes in a warmer climate. J. Climate, 23, 45564569.

    • Search Google Scholar
    • Export Citation
  • Zhang, Y., , W. B. Rossow, , A. A. Lacis, , V. Oinas, , and M. I. Mishchenko, 2004: Calculation of radiative fluxes from the surface to top of atmosphere based on ISCCP and other global data sets: Refinements of the radiative transfer model and the input data. J. Geophys. Res.,109, D19105, doi:10.1029/2003JD004457.

  • Zhou, L., , R. E. Dickinson, , P. Dirmeyer, , A. Dai, , and S.-K. Min, 2009: Spatiotemporal patterns of change in maximum and minimum temperatures in multi-model simulations. Geophys. Res. Lett.,36, L02702, doi:10.1029/2008GL036141.

1

For four of the models (BCCR-BCM2.0, CNRM-CM3, INM-CM3.0, and ECHAM5/MPI-OM), this information was received by e-mail from the model contact persons. For UKMO-HadCM3, we confirmed that, in elevated areas, the 1000-hPa and 950-hPa RHs (in the interior of Greenland, the 850-hPa RH as well) were identical.

Supplementary Materials

Save