- Email: [email protected]
Monitoring climate at Jungfraujoch in the high Swiss Alpine region
SC IE N CE OF T H E TOT AL E N V I RO N ME N T 3 9 1 ( 2 00 8 ) 2 6 2–2 68
a v a i l a b l e a t w w w. s c i e n c e d i r e c t . c o m
w w w. e l s e v i e r. c o m / l o c a t e / s c i t o t e n v
Monitoring climate at Jungfraujoch in the high Swiss Alpine region Christof Appenzeller a,⁎, Michael Begert a , Evelyn Zenklusena , Simon C. Scherrer b a
Federal Office of Meteorology and Climatology MeteoSwiss, Krähbühlstrasse 58, CH-8044 Zürich, Switzerland National Center for Atmospheric Research, Climate Analysis Section, Climate and Global Dynamics Division, P.O. Box 3000, Boulder CO, 80307, USA
b
AR TIC LE I N FO
ABS TR ACT
Keywords:
A homogenized temperature record measured at Jungfraujoch, the highest permanently
Jungfraujoch
manned meteorological station in Europe at 3580 m asl, is presented based on almost 70 years of
Temperature
record (1937–2005). The observed decadal variability as well as the overall trend (1.8 °C/69 years)
Homogeneity
in the homogenized data is comparable to other homogenized Swiss time series at other
Trends
altitudes. A detailed analysis of seasonal mean temperature trends revealed no significant
Thawing
height dependence for the period 1961–2005. The dominant trend features are the weaker
Alpine region
trends in autumn, significant only at low altitudes. Temperature indices such as thawing days, derived from newly homogenized minimum temperature series, exhibit strong vertical and seasonal trend dependence. Strongest relative trends occur in winter at an altitude around 1000 and 1600 m asl. For the summer season relative trends in thawing days are strongest at the highest stations, as expected. At Jungfraujoch an increase of about 50% is observed for the period 1961–2005 even when the extraordinary warm summer of 2003 is excluded. © 2007 Elsevier B.V. All rights reserved.
1.
Introduction
In recent years climate change has gained substantial attention in the media and on the political agenda. In terms of global mean surface temperature, the years in the 1990s and the beginning of the 21st century have been the warmest years, at least since the beginning of instrumental measurements (Houghton et al., 2001). In central Europe eight of the ten warmest years of the 1851–2004 temperature record have been observed from 1989 to 2003 (Scherrer et al., 2005). Climate model simulations show that most of the observed trend can only be explained when the anthropogenic greenhouse gas emissions are taken into account (Houghton et al., 2001). In the Alpine region similar temperature increases have been observed over the last ∼150 years (e.g. Begert et al., 2005; Böhm et al., 2001; Kunz et al., 2007). The amplitudes of these trends are twice or even three times as large as the global average figures for the last few decades (Philipona et al., 2004; Böhm et al., 2001). Some studies find indications for larger increases in temperature
at high altitude stations (e.g. Diaz and Bradley, 1997). Many economical, ecological and geophysical sectors in the Alpine region are sensitive in one or the other way to temperature changes (Houghton et al., 2001; Haeberli and Beniston, 1998; Studer et al., 2005; Walther et al., 2002; Scherrer et al., 2004). For such studies the vertical distribution of the observed and expected temperature changes, as will be explored in this paper, are of crucial importance. In Switzerland a high density near surface measurement network (SwissMetNet: Fig. 1, Frei, 2003) provides long term observations at altitudes ranging from 197 to 3580 m asl. The later altitude is the one of the station at Jungfraujoch operated by the Federal Office of Meteorology and Climatology MeteoSwiss. It is the highest permanently manned meteorological station in Europe. Measurements have been carried out since 1922 when the access by railroad was completed and the Jungfraujoch Commission was founded. Today, long term data series of Jungfraujoch and other stations on such altitudes are of great value to address questions of the current global change debate.
⁎ Corresponding author. Tel.: +41 44 256 93 88. E-mail address: [email protected] (C. Appenzeller). 0048-9697/$ – see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.scitotenv.2007.10.005
S CIE N CE OF T H E TOT AL E N V I RO N ME N T 3 9 1 ( 2 00 8 ) 2 6 2–2 68
263
Fig. 1 – Frequency distribution of Swiss meteorological stations with respect of their altitude (m asl). SwissMetNet weather stations (solid line), homogenized minimum and mean temperature series (dashed line) used for the current study and Swiss land surface (grey). In this paper, the results of the homogeneity assessment of the temperature record at Jungfraujoch are examined and compared with decadal variability and seasonal trend with other homogenized Swiss time series. The observed vertical and seasonal dependence of the mean annual and seasonal temperature and derived temperature indices are discussed.
2. Background: temperature record at Jungfraujoch The first years of meteorological measurements at the Jungfraujoch were dominated by finding an optimal location to place the instruments as, especially for temperature, the differences between alternative measuring sites were substantial and strongly influenced by exposition. In 1925 a meteorological pavilion (Fig. 2) was built on the firn of the Joch and led to the first satisfying measurements. However, the pavilion was destroyed several times due to strong winds and moving ground. Undisturbed measurements were only achieved when the meteorological station could be integrated in the new observatory inaugurated at the Jungfraujoch Sphinx in 1937. Henceforth temperature was measured in a screen in front of a north-faced window sheltered by stone shutters. In 1980 the station became automated. The instruments could be installed on the platform of the observatory and additional measurements such as radiation, luminosity and radioactivity were established. The automated station has been operating for 26 years now. In the near future it will become part of the SwissMetNet, MeteoSwiss new observational network which renews and unifies the ground-based networks in Switzerland. The permanent presence of qualified personnel, however, was and will always be a key factor for high quality meteorological measurements in a high Alpine environment. Even the latest generation of instruments cannot always cope with the extreme weather conditions at Jungfraujoch and need to be de-iced or cleaned. Meteorological measurements taken at such an exposed location as the Jungfraujoch highlight the basic problem of
changing measuring conditions over time in long term data series. Site relocations or changes in instrumentation can lead to non-climatic variations in the data series and distort or even hide the true climatic signal. The detection and correction of such inhomogeneities as a first step in climate change studies is an important task and well established at MeteoSwiss. The homogenization procedure applied in this study combines statistical testing and adjustment techniques with an in-depth analysis of the available metadata information. It is an interactive procedure that allows searching and adjusting multiple shifts and trend inhomogeneities. For a detailed description see Begert et al. (2005).
3.
Results
3.1.
Homogeneity
Fig. 3 shows the adjustment amounts of the monthly temperature series at Jungfraujoch for the period 1937 to 2005. The
Fig. 2 – The first meteorological pavilion at the Jungfraujoch built in 1925. Photo: Fahrni.
264
SC IE N CE OF T H E TOT AL E N V I RO N ME N T 3 9 1 ( 2 00 8 ) 2 6 2–2 68
Fig. 3 – Adjustments [°C] of the monthly temperature series at Jungfraujoch for the period 1937 to 2005.
Fig. 4 – Yearly mean temperature anomalies derived from the homogenized series at Jungfraujoch for the period 1937 to 2005 with a 20 year Gaussian low-pass filter (solid line) and the standard deviation (dashed line). All anomalies are in °C and given relative to the 1961 to 1990 period.
Fig. 5 – Yearly and seasonal mean temperature anomalies in Switzerland (black) and Jungfraujoch (grey) for the period 1937 to 2005.
S CIE N CE OF T H E TOT AL E N V I RO N ME N T 3 9 1 ( 2 00 8 ) 2 6 2–2 68
265
Fig. 6 – Trends in homogenized annual mean 2 m temperature at 45 MeteoSwiss stations as a function of geographical latitude (°N, abscissa) and station altitude (m asl, ordinate) for the period 1961–2005. Positive (negative) trends are shown as circles (triangles), trendless station series are displayed as squares. Significant trends (5% level) are filled. Annual (a) and seasonal (c–f ) trend analysis (linear least square fit) in °C per 45 years. Deviation of trends (in % per 45 years) relative to mean trend averaged over all stations (b).
measurements of the pavilion before 1937 have not been adjusted yet to the rest of the series and are not shown. Four shift inhomogeneities were discovered caused by the introduction of automated measurements (1981), change of instrument type (1982) and instrumental problems (1992 to 1994). As seen for other temperature series in Switzerland, the use of ventilated sensors in the automated network led to a smaller mean annual cycle. The reason for that is the smaller sensitivity of the measurement to radiation. In contrast to findings for other stations no systematic bias is observed for the early instrumental period at the Jungfraujoch. Special attention had to be given to the first year of automated measurement in 1981 (Fig. 3). The instrument in use was not suitable for high Alpine
conditions and led to dramatically high monthly means that had to be adjusted for.
3.2.
Decadal variability in mean temperature
The evolution of annual mean temperature at Jungfraujoch is shown in Fig. 4 for the period 1937 to 2005 expressed as yearly anomalies relative to 1961 to 1990. The main characteristics of the curve are the warm period in the late 1940s and the shift towards warmer temperatures in the second part of the 1980s (Bader and Bantle, 2004). The anomalies reveal an increase of 1.8 °C within the 69 year period. Overall, the decadal variability at the Jungfraujoch is similar to the mean temperature
266
SC IE N CE OF T H E TOT AL E N V I RO N ME N T 3 9 1 ( 2 00 8 ) 2 6 2–2 68
evolution in Switzerland. Fig. 5 shows the Jungfraujoch anomalies compared to the annual and seasonal mean Swiss temperature record described in Begert et al. (2005). It consists of 12 homogenized series representing one of the climatic regions of Switzerland each. A remarkable uniform course of both the annual and seasonal curves is observed.
3.3.
Vertical distribution of changes in mean temperature
When comparing Jungfraujoch measurements with those of other stations it becomes evident that the mean temperature evolution shows no distinct altitude dependence in Switzerland (Fig. 6a). The graph shows mean annual 2 m temperature trends for 45 homogenized stations at different altitudes for the period 1961–2005. The annual trends are statistically significant on the 5% level for all stations. There is no obvious dependence of the trends with altitude or geographical position. Relative trends expressed as deviation to the mean Swiss trend indicate slightly weaker trends at high altitudes and slightly stronger trends in the lowlands (Fig. 6 b). This is in opposition to the height dependence suggested for several mountain regions in Diaz and Bradley (1997). Note that the number of stations included in their analysis was very low in Western Europe. The height dependence found in Fig. 6 (b) is weak and in the order of 15% only. These results are in agreement with other studies which also find no strong relationship between trend magnitudes and elevation using data from many places around the world (e.g. Pepin and Seidel, 2005). Seasonal analysis (Fig. 6 c–f) shows the trends for winter (DJF), spring (MAM), summer (JJA) and autumn (SON). Again, there is no clear height dependence visible, except for autumn. More pronounced are the trend differences with season. In winter, spring and summer statistically significant increasing trends are found for all stations, except one, Samedan. This inconsistency is most likely a remaining problem in the homogenization which is doubtful for this station and could therefore lead to wrong interpretations. In autumn trends are positive as well, but statistically significant only at low altitude stations. The weaker trends in autumn are not specific for Switzerland but also found over most regions of Europe (cf. Klein Tank et al., 2005). Without including the most recent years (2003, 2004 and 2005) autumn
trends were completely absent (e.g. Scherrer, 2006). The reason for the weaker trends in autumn particularly at higher altitude is not explored in detail. It could be simply a peculiarity of decadal variability. A closer investigation of the relation between local temperature, large-scale climate variability (e.g. in terms of atmospheric blocking), the occurrence of “cold-pools” over the Swiss plateau and/or Föhn situation could be a promising approach to determine whether the missing autumn trends are due to compensating natural variability or not.
3.4. Vertical distribution of changes in temperature related indices The impact of recent warming at high and low altitudes can strongly differ. As an example Fig. 7 shows the number of days per year with minimum temperature above 0 °C (Tn N 0, referred to as “thawing days”) at the Jungfraujoch. The indices were derived from a set of newly homogenized minimum temperature series of Switzerland (Begert et al., 2003) for the period between 1961 and 2005. The most striking feature is the extraordinary warm summer of 2003, where the number of thawing days has doubled. But even without the year 2003 a logistic trend estimate in annual count (McCullagh and Nelder, 1989; see also Frei and Schär, 2001) reveals an increase of about 50%. Comparing relative trends in count of thawing days at different altitudes for the period 1961–2005 shows a strong dependence on height and season as expected (Fig. 8). Annual thawing days clearly demonstrate an overall positive trend (Fig. 8 a). Absolute trend values (Fig. 8 b) have roughly the same magnitude; however relative (Fig. 8 a) trend values are larger at higher altitudes than in the lowlands. All annual trends are significant on the 5% level. Note that stations where the index-criterion was fulfilled for every day or never are marked as squares in Fig. 8 (e.g. lowland stations in summer). The height dependence of the logistic trends in number of thawing days for the four seasons are separately shown (Fig. 8 c–f). Strong positive trends in thawing days occur in winter, with largest trend values at a height level between 1000 and 1600 m asl. The strongest trend is observed at Davos (1590 m asl) with an increase of more than 260% in winter.
Fig. 7 – Annual count of thawing days (i.e. days with minimum temperature above 0 °C) at Jungfraujoch for the period 1961 to 2005.
S CIE N CE OF T H E TOT AL E N V I RO N ME N T 3 9 1 ( 2 00 8 ) 2 6 2–2 68
267
Fig. 8 – Logistic trends in annual count of thawing days at 45 MeteoSwiss stations as a function of geographical latitude (°N, abscissa) and station altitude (m asl, ordinate) for the period 1961–2005. Positive (negative) trends are shown as circles (triangles), trendless station series are displayed as squares. Significant trends (5% level) are filled. Annual (a) and seasonal (c–f) trend analysis in % per 45 years. Absolute trend values for annual count of thawing days per 45 years (b).
In spring and autumn, trends are less extreme but still predominantly positive. Strongest relative increases occur at heights between 1500 and 2500 m asl and in inner Alpine regions. The (insignificantly) negative autumn trends at some high altitude stations agree well with the insignificant mean temperature trends in autumn. In the summer season strongest trends in thawing days are observed at the highest Alpine stations with the strongest increase (66%) at Jungfraujoch at 3580 m asl.
4.
Discussion
Studies quantifying and exploring the impact of a changing climate on the high Alpine region strongly rely on stations like
the one at Jungfraujoch. High altitude stations are and will remain an important part of the climate monitoring system in Switzerland and Europe. Since 1925 meteorological measurements have been carried out at Jungfraujoch and the homogenized record of mean temperature over the period 1937–2005 shows an increase of 1.8 °C, which is comparable to the mean temperature evolution in Switzerland. The trends in annual mean temperature for 45 Swiss stations at different altitudes for the period 1961–2005 are strongly positive and highly significant. No significant height dependence of monthly mean temperature trends was found for both, annual and seasonal values. Seasonal temperatures in winter, spring and summer are all clearly increasing. Weaker trends are found for autumn where
268
SC IE N CE OF T H E TOT AL E N V I RO N ME N T 3 9 1 ( 2 00 8 ) 2 6 2–2 68
they are statistically significant only at low altitude stations. Non-linear dependent parameters such as thawing days show that the impact of the recent warming at high and low altitudes can strongly differ. Annual trends in thawing days for 45 Swiss stations are all positive and mostly significant. Seasonal values show that the strongest trends in thawing days are observed in winter. Remarkable changes of the height levels of the maximum seasonal trends can be observed. The strongest trends in winter occur in heights between 1000 and 1600 m asl, unlike in summer when the trend maxima rise up into the highest Alpine regions to 3500 m asl. In spring and autumn thawing days trends are strongest at an intermediate level around 1500 to 2500 m asl. Regional climate models indicate that towards the end of the 21st century about every second summer could be as warm as 2003 (see e.g. Schär et al., 2004 and reference therein). Hence it can be expected that the observed increase in number of thawing days in the intermediate and higher Alpine region will continue.
Acknowledgements We would like to thank the “Stiftung Hochalpine Forschungsstationen Jungfraujoch und Gornergrat (SHF)” and the “Jungfraujoch Bahn (JB)” which enable us to carry out meteorological observations at that unique location. Special thanks also to Christoph Frei for his advice in preparing the logistic trend analysis; Thomas Schlegel for his support in data homogenization and the Swiss National Centre for Competence in Research Climate (NCCR-Climate) for partly funding this research. S. C. Scherrer is funded by the Swiss National Science Foundation.
REFERENCES Bader S, Bantle H. Das Schweizer Klima im Trend: Temperatur- und Niederschlagsentwicklung 1864–2001 (in german). Veröffentlichungen der MeteoSchweiz, Nr. 68, 48pp, 2004. Available online via http://www.meteoschweiz.ch. Begert M, Seiz G, Schlegel T, Musa M, Baudraz G, Moesch M. Homogenisierung von Klimamessreihen der Schweiz und Bestimmung der Normwerte 1961–1990 (in german). Veröffentlichungen der MeteoSchweiz, Nr. 67, 170pp, 2003. Available online via http://www.meteoschweiz.ch. Begert M, Schlegel T, Kirchhofer W. Homogeneous temperature and precipitation series of Switzerland from 1864 to 2000. Int J Climatol 2005;23:65–80. Böhm R, Auer I, Brunetti M, Maugeri M, Nanni T, Schöner W. Regional temperature variability in the European Alps: 1760–1998 from homogenized instrumental time series. Int J Climatol 2001;21:1779–801.
Diaz HF, Bradley RS. Temperature variations during the last century at high elevation sites. Clim Change 1997;36:253–79. Frei T. Designing meteorological networks for Switzerland according to user requirements. Meteorol Appl 2003;10:313–7. Frei C, Schär C. Detection probability of trends in rare events: theory and application to heavy precipitation in the Alpine region. J Clim 2001;14:1568–84. Haeberli W, Beniston M. Climate change and its impacts on glaciers and permafrost in the Alps. Ambio 1998;27:258–65. Houghton JT, Ding Y, Griggs DJ, Noguer M, van der Linden PJ, Dai X, et al. Climate change 2001: the scientific basis. contribution of working group i to the third assessment report of the intergovernmental panel on climate change. United Kingdom: Cambridge University Press; 2001. 944 pp. Klein Tank AMG, Können GP, Selten FM. Signals of anthropogenic influence on European warming as seen in the trend patterns of daily temperature variance. Int J Climatol 2005;25:1–16. Kunz H, Scherrer SC, Liniger MA, Appenzeller C. Changes in surface temperatures and total ozone in the Alpine region: An intercomparison of ERA-40 data and observed Swiss time series. Meteorologische Zeitschrift 2007;16:171–81. McCullagh P, Nelder JA. Generalized linear models. Monogr. On Statistics and Appl. Probability, 2d ed., vol. 37. Chapman and Hall; 1989. 511 pp. Pepin NC, Seidel DJ. A global comparison of surface and free-air temperatures at high elevations. J Geophys Res-Atmos 2005;110:D03104. _________________________ doi:10.1029/2004JD005047. Philipona R, Dürr B, Marty C, Ohmura A, Wild M. Radiative forcing — measured at Earth's surface — corroborates the increasing greenhouse effect. Geophys Res Lett 2004;31: L03202. _________________________ doi:10.1029/2003GL018765. Schär C, Vidale PL, Lüthi D, Frei C, Haeberli C, Liniger MA, Appenzeller C. The role of increasing temperature variability in European summer heatwaves. Nature 2004;427:332–6. Scherrer SC. Interannual climate variability in the European and Alpine region. PhD Thesis No. 16338, Eidgenössisch Technische Hochschule (ETH), Zürich, 2006, 120pp (electronic version available at http://e-collection.ethbib.ethz.ch/show? type=diss&nr=16338). Scherrer SC, Appenzeller C, Laternser M. Trends in Swiss Alpine snow days: the role of local- and large-scale climate variability. Geophys Res Lett 2004;31:L13215. doi:10.1029/2004GL020255. _________________________ Scherrer SC, Appenzeller C, Liniger MA, Schär C. European temperature distribution changes in observations and climate change scenarios. Geophys Res Lett 2005;32:L19705. doi:10.1029/2005GL024108. _________________________ Studer S, Appenzeller C, Defila C. Inter-annual variability and decadal trends in Alpine spring phenology; A multivariate analysis approach. Clim Change 2005;73:395–414. Walther G, Post E, Convey P, Menzel A, Parmesan C, Beebee TJC, et al. Ecological responses to recent climate change. Nature 2002;416:389–95.
a v a i l a b l e a t w w w. s c i e n c e d i r e c t . c o m
w w w. e l s e v i e r. c o m / l o c a t e / s c i t o t e n v
Monitoring climate at Jungfraujoch in the high Swiss Alpine region Christof Appenzeller a,⁎, Michael Begert a , Evelyn Zenklusena , Simon C. Scherrer b a
Federal Office of Meteorology and Climatology MeteoSwiss, Krähbühlstrasse 58, CH-8044 Zürich, Switzerland National Center for Atmospheric Research, Climate Analysis Section, Climate and Global Dynamics Division, P.O. Box 3000, Boulder CO, 80307, USA
b
AR TIC LE I N FO
ABS TR ACT
Keywords:
A homogenized temperature record measured at Jungfraujoch, the highest permanently
Jungfraujoch
manned meteorological station in Europe at 3580 m asl, is presented based on almost 70 years of
Temperature
record (1937–2005). The observed decadal variability as well as the overall trend (1.8 °C/69 years)
Homogeneity
in the homogenized data is comparable to other homogenized Swiss time series at other
Trends
altitudes. A detailed analysis of seasonal mean temperature trends revealed no significant
Thawing
height dependence for the period 1961–2005. The dominant trend features are the weaker
Alpine region
trends in autumn, significant only at low altitudes. Temperature indices such as thawing days, derived from newly homogenized minimum temperature series, exhibit strong vertical and seasonal trend dependence. Strongest relative trends occur in winter at an altitude around 1000 and 1600 m asl. For the summer season relative trends in thawing days are strongest at the highest stations, as expected. At Jungfraujoch an increase of about 50% is observed for the period 1961–2005 even when the extraordinary warm summer of 2003 is excluded. © 2007 Elsevier B.V. All rights reserved.
1.
Introduction
In recent years climate change has gained substantial attention in the media and on the political agenda. In terms of global mean surface temperature, the years in the 1990s and the beginning of the 21st century have been the warmest years, at least since the beginning of instrumental measurements (Houghton et al., 2001). In central Europe eight of the ten warmest years of the 1851–2004 temperature record have been observed from 1989 to 2003 (Scherrer et al., 2005). Climate model simulations show that most of the observed trend can only be explained when the anthropogenic greenhouse gas emissions are taken into account (Houghton et al., 2001). In the Alpine region similar temperature increases have been observed over the last ∼150 years (e.g. Begert et al., 2005; Böhm et al., 2001; Kunz et al., 2007). The amplitudes of these trends are twice or even three times as large as the global average figures for the last few decades (Philipona et al., 2004; Böhm et al., 2001). Some studies find indications for larger increases in temperature
at high altitude stations (e.g. Diaz and Bradley, 1997). Many economical, ecological and geophysical sectors in the Alpine region are sensitive in one or the other way to temperature changes (Houghton et al., 2001; Haeberli and Beniston, 1998; Studer et al., 2005; Walther et al., 2002; Scherrer et al., 2004). For such studies the vertical distribution of the observed and expected temperature changes, as will be explored in this paper, are of crucial importance. In Switzerland a high density near surface measurement network (SwissMetNet: Fig. 1, Frei, 2003) provides long term observations at altitudes ranging from 197 to 3580 m asl. The later altitude is the one of the station at Jungfraujoch operated by the Federal Office of Meteorology and Climatology MeteoSwiss. It is the highest permanently manned meteorological station in Europe. Measurements have been carried out since 1922 when the access by railroad was completed and the Jungfraujoch Commission was founded. Today, long term data series of Jungfraujoch and other stations on such altitudes are of great value to address questions of the current global change debate.
⁎ Corresponding author. Tel.: +41 44 256 93 88. E-mail address: [email protected] (C. Appenzeller). 0048-9697/$ – see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.scitotenv.2007.10.005
S CIE N CE OF T H E TOT AL E N V I RO N ME N T 3 9 1 ( 2 00 8 ) 2 6 2–2 68
263
Fig. 1 – Frequency distribution of Swiss meteorological stations with respect of their altitude (m asl). SwissMetNet weather stations (solid line), homogenized minimum and mean temperature series (dashed line) used for the current study and Swiss land surface (grey). In this paper, the results of the homogeneity assessment of the temperature record at Jungfraujoch are examined and compared with decadal variability and seasonal trend with other homogenized Swiss time series. The observed vertical and seasonal dependence of the mean annual and seasonal temperature and derived temperature indices are discussed.
2. Background: temperature record at Jungfraujoch The first years of meteorological measurements at the Jungfraujoch were dominated by finding an optimal location to place the instruments as, especially for temperature, the differences between alternative measuring sites were substantial and strongly influenced by exposition. In 1925 a meteorological pavilion (Fig. 2) was built on the firn of the Joch and led to the first satisfying measurements. However, the pavilion was destroyed several times due to strong winds and moving ground. Undisturbed measurements were only achieved when the meteorological station could be integrated in the new observatory inaugurated at the Jungfraujoch Sphinx in 1937. Henceforth temperature was measured in a screen in front of a north-faced window sheltered by stone shutters. In 1980 the station became automated. The instruments could be installed on the platform of the observatory and additional measurements such as radiation, luminosity and radioactivity were established. The automated station has been operating for 26 years now. In the near future it will become part of the SwissMetNet, MeteoSwiss new observational network which renews and unifies the ground-based networks in Switzerland. The permanent presence of qualified personnel, however, was and will always be a key factor for high quality meteorological measurements in a high Alpine environment. Even the latest generation of instruments cannot always cope with the extreme weather conditions at Jungfraujoch and need to be de-iced or cleaned. Meteorological measurements taken at such an exposed location as the Jungfraujoch highlight the basic problem of
changing measuring conditions over time in long term data series. Site relocations or changes in instrumentation can lead to non-climatic variations in the data series and distort or even hide the true climatic signal. The detection and correction of such inhomogeneities as a first step in climate change studies is an important task and well established at MeteoSwiss. The homogenization procedure applied in this study combines statistical testing and adjustment techniques with an in-depth analysis of the available metadata information. It is an interactive procedure that allows searching and adjusting multiple shifts and trend inhomogeneities. For a detailed description see Begert et al. (2005).
3.
Results
3.1.
Homogeneity
Fig. 3 shows the adjustment amounts of the monthly temperature series at Jungfraujoch for the period 1937 to 2005. The
Fig. 2 – The first meteorological pavilion at the Jungfraujoch built in 1925. Photo: Fahrni.
264
SC IE N CE OF T H E TOT AL E N V I RO N ME N T 3 9 1 ( 2 00 8 ) 2 6 2–2 68
Fig. 3 – Adjustments [°C] of the monthly temperature series at Jungfraujoch for the period 1937 to 2005.
Fig. 4 – Yearly mean temperature anomalies derived from the homogenized series at Jungfraujoch for the period 1937 to 2005 with a 20 year Gaussian low-pass filter (solid line) and the standard deviation (dashed line). All anomalies are in °C and given relative to the 1961 to 1990 period.
Fig. 5 – Yearly and seasonal mean temperature anomalies in Switzerland (black) and Jungfraujoch (grey) for the period 1937 to 2005.
S CIE N CE OF T H E TOT AL E N V I RO N ME N T 3 9 1 ( 2 00 8 ) 2 6 2–2 68
265
Fig. 6 – Trends in homogenized annual mean 2 m temperature at 45 MeteoSwiss stations as a function of geographical latitude (°N, abscissa) and station altitude (m asl, ordinate) for the period 1961–2005. Positive (negative) trends are shown as circles (triangles), trendless station series are displayed as squares. Significant trends (5% level) are filled. Annual (a) and seasonal (c–f ) trend analysis (linear least square fit) in °C per 45 years. Deviation of trends (in % per 45 years) relative to mean trend averaged over all stations (b).
measurements of the pavilion before 1937 have not been adjusted yet to the rest of the series and are not shown. Four shift inhomogeneities were discovered caused by the introduction of automated measurements (1981), change of instrument type (1982) and instrumental problems (1992 to 1994). As seen for other temperature series in Switzerland, the use of ventilated sensors in the automated network led to a smaller mean annual cycle. The reason for that is the smaller sensitivity of the measurement to radiation. In contrast to findings for other stations no systematic bias is observed for the early instrumental period at the Jungfraujoch. Special attention had to be given to the first year of automated measurement in 1981 (Fig. 3). The instrument in use was not suitable for high Alpine
conditions and led to dramatically high monthly means that had to be adjusted for.
3.2.
Decadal variability in mean temperature
The evolution of annual mean temperature at Jungfraujoch is shown in Fig. 4 for the period 1937 to 2005 expressed as yearly anomalies relative to 1961 to 1990. The main characteristics of the curve are the warm period in the late 1940s and the shift towards warmer temperatures in the second part of the 1980s (Bader and Bantle, 2004). The anomalies reveal an increase of 1.8 °C within the 69 year period. Overall, the decadal variability at the Jungfraujoch is similar to the mean temperature
266
SC IE N CE OF T H E TOT AL E N V I RO N ME N T 3 9 1 ( 2 00 8 ) 2 6 2–2 68
evolution in Switzerland. Fig. 5 shows the Jungfraujoch anomalies compared to the annual and seasonal mean Swiss temperature record described in Begert et al. (2005). It consists of 12 homogenized series representing one of the climatic regions of Switzerland each. A remarkable uniform course of both the annual and seasonal curves is observed.
3.3.
Vertical distribution of changes in mean temperature
When comparing Jungfraujoch measurements with those of other stations it becomes evident that the mean temperature evolution shows no distinct altitude dependence in Switzerland (Fig. 6a). The graph shows mean annual 2 m temperature trends for 45 homogenized stations at different altitudes for the period 1961–2005. The annual trends are statistically significant on the 5% level for all stations. There is no obvious dependence of the trends with altitude or geographical position. Relative trends expressed as deviation to the mean Swiss trend indicate slightly weaker trends at high altitudes and slightly stronger trends in the lowlands (Fig. 6 b). This is in opposition to the height dependence suggested for several mountain regions in Diaz and Bradley (1997). Note that the number of stations included in their analysis was very low in Western Europe. The height dependence found in Fig. 6 (b) is weak and in the order of 15% only. These results are in agreement with other studies which also find no strong relationship between trend magnitudes and elevation using data from many places around the world (e.g. Pepin and Seidel, 2005). Seasonal analysis (Fig. 6 c–f) shows the trends for winter (DJF), spring (MAM), summer (JJA) and autumn (SON). Again, there is no clear height dependence visible, except for autumn. More pronounced are the trend differences with season. In winter, spring and summer statistically significant increasing trends are found for all stations, except one, Samedan. This inconsistency is most likely a remaining problem in the homogenization which is doubtful for this station and could therefore lead to wrong interpretations. In autumn trends are positive as well, but statistically significant only at low altitude stations. The weaker trends in autumn are not specific for Switzerland but also found over most regions of Europe (cf. Klein Tank et al., 2005). Without including the most recent years (2003, 2004 and 2005) autumn
trends were completely absent (e.g. Scherrer, 2006). The reason for the weaker trends in autumn particularly at higher altitude is not explored in detail. It could be simply a peculiarity of decadal variability. A closer investigation of the relation between local temperature, large-scale climate variability (e.g. in terms of atmospheric blocking), the occurrence of “cold-pools” over the Swiss plateau and/or Föhn situation could be a promising approach to determine whether the missing autumn trends are due to compensating natural variability or not.
3.4. Vertical distribution of changes in temperature related indices The impact of recent warming at high and low altitudes can strongly differ. As an example Fig. 7 shows the number of days per year with minimum temperature above 0 °C (Tn N 0, referred to as “thawing days”) at the Jungfraujoch. The indices were derived from a set of newly homogenized minimum temperature series of Switzerland (Begert et al., 2003) for the period between 1961 and 2005. The most striking feature is the extraordinary warm summer of 2003, where the number of thawing days has doubled. But even without the year 2003 a logistic trend estimate in annual count (McCullagh and Nelder, 1989; see also Frei and Schär, 2001) reveals an increase of about 50%. Comparing relative trends in count of thawing days at different altitudes for the period 1961–2005 shows a strong dependence on height and season as expected (Fig. 8). Annual thawing days clearly demonstrate an overall positive trend (Fig. 8 a). Absolute trend values (Fig. 8 b) have roughly the same magnitude; however relative (Fig. 8 a) trend values are larger at higher altitudes than in the lowlands. All annual trends are significant on the 5% level. Note that stations where the index-criterion was fulfilled for every day or never are marked as squares in Fig. 8 (e.g. lowland stations in summer). The height dependence of the logistic trends in number of thawing days for the four seasons are separately shown (Fig. 8 c–f). Strong positive trends in thawing days occur in winter, with largest trend values at a height level between 1000 and 1600 m asl. The strongest trend is observed at Davos (1590 m asl) with an increase of more than 260% in winter.
Fig. 7 – Annual count of thawing days (i.e. days with minimum temperature above 0 °C) at Jungfraujoch for the period 1961 to 2005.
S CIE N CE OF T H E TOT AL E N V I RO N ME N T 3 9 1 ( 2 00 8 ) 2 6 2–2 68
267
Fig. 8 – Logistic trends in annual count of thawing days at 45 MeteoSwiss stations as a function of geographical latitude (°N, abscissa) and station altitude (m asl, ordinate) for the period 1961–2005. Positive (negative) trends are shown as circles (triangles), trendless station series are displayed as squares. Significant trends (5% level) are filled. Annual (a) and seasonal (c–f) trend analysis in % per 45 years. Absolute trend values for annual count of thawing days per 45 years (b).
In spring and autumn, trends are less extreme but still predominantly positive. Strongest relative increases occur at heights between 1500 and 2500 m asl and in inner Alpine regions. The (insignificantly) negative autumn trends at some high altitude stations agree well with the insignificant mean temperature trends in autumn. In the summer season strongest trends in thawing days are observed at the highest Alpine stations with the strongest increase (66%) at Jungfraujoch at 3580 m asl.
4.
Discussion
Studies quantifying and exploring the impact of a changing climate on the high Alpine region strongly rely on stations like
the one at Jungfraujoch. High altitude stations are and will remain an important part of the climate monitoring system in Switzerland and Europe. Since 1925 meteorological measurements have been carried out at Jungfraujoch and the homogenized record of mean temperature over the period 1937–2005 shows an increase of 1.8 °C, which is comparable to the mean temperature evolution in Switzerland. The trends in annual mean temperature for 45 Swiss stations at different altitudes for the period 1961–2005 are strongly positive and highly significant. No significant height dependence of monthly mean temperature trends was found for both, annual and seasonal values. Seasonal temperatures in winter, spring and summer are all clearly increasing. Weaker trends are found for autumn where
268
SC IE N CE OF T H E TOT AL E N V I RO N ME N T 3 9 1 ( 2 00 8 ) 2 6 2–2 68
they are statistically significant only at low altitude stations. Non-linear dependent parameters such as thawing days show that the impact of the recent warming at high and low altitudes can strongly differ. Annual trends in thawing days for 45 Swiss stations are all positive and mostly significant. Seasonal values show that the strongest trends in thawing days are observed in winter. Remarkable changes of the height levels of the maximum seasonal trends can be observed. The strongest trends in winter occur in heights between 1000 and 1600 m asl, unlike in summer when the trend maxima rise up into the highest Alpine regions to 3500 m asl. In spring and autumn thawing days trends are strongest at an intermediate level around 1500 to 2500 m asl. Regional climate models indicate that towards the end of the 21st century about every second summer could be as warm as 2003 (see e.g. Schär et al., 2004 and reference therein). Hence it can be expected that the observed increase in number of thawing days in the intermediate and higher Alpine region will continue.
Acknowledgements We would like to thank the “Stiftung Hochalpine Forschungsstationen Jungfraujoch und Gornergrat (SHF)” and the “Jungfraujoch Bahn (JB)” which enable us to carry out meteorological observations at that unique location. Special thanks also to Christoph Frei for his advice in preparing the logistic trend analysis; Thomas Schlegel for his support in data homogenization and the Swiss National Centre for Competence in Research Climate (NCCR-Climate) for partly funding this research. S. C. Scherrer is funded by the Swiss National Science Foundation.
REFERENCES Bader S, Bantle H. Das Schweizer Klima im Trend: Temperatur- und Niederschlagsentwicklung 1864–2001 (in german). Veröffentlichungen der MeteoSchweiz, Nr. 68, 48pp, 2004. Available online via http://www.meteoschweiz.ch. Begert M, Seiz G, Schlegel T, Musa M, Baudraz G, Moesch M. Homogenisierung von Klimamessreihen der Schweiz und Bestimmung der Normwerte 1961–1990 (in german). Veröffentlichungen der MeteoSchweiz, Nr. 67, 170pp, 2003. Available online via http://www.meteoschweiz.ch. Begert M, Schlegel T, Kirchhofer W. Homogeneous temperature and precipitation series of Switzerland from 1864 to 2000. Int J Climatol 2005;23:65–80. Böhm R, Auer I, Brunetti M, Maugeri M, Nanni T, Schöner W. Regional temperature variability in the European Alps: 1760–1998 from homogenized instrumental time series. Int J Climatol 2001;21:1779–801.
Diaz HF, Bradley RS. Temperature variations during the last century at high elevation sites. Clim Change 1997;36:253–79. Frei T. Designing meteorological networks for Switzerland according to user requirements. Meteorol Appl 2003;10:313–7. Frei C, Schär C. Detection probability of trends in rare events: theory and application to heavy precipitation in the Alpine region. J Clim 2001;14:1568–84. Haeberli W, Beniston M. Climate change and its impacts on glaciers and permafrost in the Alps. Ambio 1998;27:258–65. Houghton JT, Ding Y, Griggs DJ, Noguer M, van der Linden PJ, Dai X, et al. Climate change 2001: the scientific basis. contribution of working group i to the third assessment report of the intergovernmental panel on climate change. United Kingdom: Cambridge University Press; 2001. 944 pp. Klein Tank AMG, Können GP, Selten FM. Signals of anthropogenic influence on European warming as seen in the trend patterns of daily temperature variance. Int J Climatol 2005;25:1–16. Kunz H, Scherrer SC, Liniger MA, Appenzeller C. Changes in surface temperatures and total ozone in the Alpine region: An intercomparison of ERA-40 data and observed Swiss time series. Meteorologische Zeitschrift 2007;16:171–81. McCullagh P, Nelder JA. Generalized linear models. Monogr. On Statistics and Appl. Probability, 2d ed., vol. 37. Chapman and Hall; 1989. 511 pp. Pepin NC, Seidel DJ. A global comparison of surface and free-air temperatures at high elevations. J Geophys Res-Atmos 2005;110:D03104. _________________________ doi:10.1029/2004JD005047. Philipona R, Dürr B, Marty C, Ohmura A, Wild M. Radiative forcing — measured at Earth's surface — corroborates the increasing greenhouse effect. Geophys Res Lett 2004;31: L03202. _________________________ doi:10.1029/2003GL018765. Schär C, Vidale PL, Lüthi D, Frei C, Haeberli C, Liniger MA, Appenzeller C. The role of increasing temperature variability in European summer heatwaves. Nature 2004;427:332–6. Scherrer SC. Interannual climate variability in the European and Alpine region. PhD Thesis No. 16338, Eidgenössisch Technische Hochschule (ETH), Zürich, 2006, 120pp (electronic version available at http://e-collection.ethbib.ethz.ch/show? type=diss&nr=16338). Scherrer SC, Appenzeller C, Laternser M. Trends in Swiss Alpine snow days: the role of local- and large-scale climate variability. Geophys Res Lett 2004;31:L13215. doi:10.1029/2004GL020255. _________________________ Scherrer SC, Appenzeller C, Liniger MA, Schär C. European temperature distribution changes in observations and climate change scenarios. Geophys Res Lett 2005;32:L19705. doi:10.1029/2005GL024108. _________________________ Studer S, Appenzeller C, Defila C. Inter-annual variability and decadal trends in Alpine spring phenology; A multivariate analysis approach. Clim Change 2005;73:395–414. Walther G, Post E, Convey P, Menzel A, Parmesan C, Beebee TJC, et al. Ecological responses to recent climate change. Nature 2002;416:389–95.