Réf. Auer & al. 2007 - A

Référence bibliographique complète

AUER, I., BÖHM, R., JURKOVIC, A., LIPA, W., ORLIK, A., POTZMANN, R., SCHÖNER, W., UNGERSBÖCK, M., MATULLA, C., BRIFFA, K., JONES, P.D., EFTHYMIADIS, D., BRUNETTI, M., NANNI, T., MAUGERI, M., MERCALLI, L., MESTRE, O., MOISSELIN, J.-M., BEGERT, M., MÜLLER-WESTERMEIER, G., KVETON, V., BOCHNICEK, O., STASTNY, P., LAPIN, M., SZALAI, S., SZENTIMREY, T., CEGNAR, T., DOLINAR, M., GAJIC-CAPKA, M., ZANINOVIC, K., MAJSTOROVIC, Z., NIEPLOVA, E. 2007. HISTALP – Historical instrumental climatological surface time series of the greater Alpine region 1760-2003. International Journal of Climatology, Vol. 27, 17–46.

Abstract: This paper describes the HISTALP database, consisting of monthly homogenised records of temperature, pressure, precipitation, sunshine and cloudiness for the ‘Greater Alpine Region’ (GAR, 4–19 °E, 43–49 °N, 0–3500m asl). The longest temperature and air pressure series extend back to 1760, precipitation to 1800, cloudiness to the 1840s and sunshine to the 1880s. A systematic QC procedure has been applied to the series and a high number of inhomogeneities (more than 2500) and outliers (more than 5000) have been detected and removed. The 557 HISTALP series are kept in different data modes: original and homogenised, gap-filled and outlier corrected station mode series, grid-1 series (anomaly fields at 1° × 1°, lat × long) and Coarse Resolution Subregional (CRS) mean series according to an EOF-based regionalisation. The leading climate variability features within the GAR are discussed through selected examples and a concluding linear trend analysis for 100, 50 and 25-year subperiods for the four horizontal and two altitudinal CRSs. Among the key findings of the trend analysis is the parallel centennial decrease/increase of both temperature and air pressure in the 19th/20th century. The 20th century increase (+1.2 °C/+1.1 hPa for annual GAR-means) evolved stepwise with a first peak near 1950 and the second increase (1.3 °C/0.6hPa per 25 years) starting in the 1970s. Centennial and decadal scale temperature trends were identical for all subregions. Air pressure, sunshine and cloudiness show significant differences between low versus high elevations. A long-term increase of the high-elevation series relative to the low-elevation series is given for sunshine and air pressure. Of special interest is the exceptional high correlation near 0.9 between the series on mean temperature and air pressure difference (high-minus low-elevation). This, further developed via some atmospheric statics and thermodynamics, allows the creation of ‘barometric temperature series’ without use of the measures of temperature. They support the measured temperature trends in the region. Precipitation shows the most significant regional and seasonal differences with, e.g., remarkable opposite 20th century evolution for NW (9% increase) versus SE (9% decrease). Other long- and short-term features are discussed and indicate the promising potential of the new database for further analyses and applications.

Mots-clés
Multiple climate database; Homogeneity; Instrumental period; Greater Alpine Region; Gridded data sets; Climate variability

Organismes / Contact

• ZAMG-Central Institute for Meteorology and Geodynamics, Vienna, Austria
• CCRM-Climate Research Branch, Downsview, Toronto, Canada
• CRU-Climatic Research Unit, University of East Anglia, Norwich, UK
• Istituto ISAC-CNR, Bologna, Italy
• Istituto di Fisica Generale Applicata, Universit`a di Milano, Milan, Italy
• SMI, Societ´a Meteorologica Italiana, Torino, Italy
• M´et´eo France, Toulouse, France
• MeteoSwiss, Federal Office of Meteorology and Climatology, Zurich, Switzerland
• DWD-Deutscher Wetterdienst, Offenbach, Germany
• CHMI-Czech Hydrometeorological Insitute, Prague, Czech Republic
• SHMU-Slovak Hydrometeorological Institue, Bratislava, Slovakia
• Comenius University, Bratislava, Slovakia
• OMSZ-Hungarian Meteorological Service
• ARSO-Environmental Agency of the Republic of Slovenia, Ljubljana, Slovenia
• DHMZ-Meteorological and Hydrographical Service of Croatia, Zagreb, Croatia
• METEOBIH, Federal Meteorological Institute, Sarajevo, Bosnia and Herzegovina
• Bratislava, Slovakia


(1) - Paramètre(s) atmosphérique(s) modifié(s)
(2) - Elément(s) du milieu impacté(s)
(3) - Type(s) d'aléa impacté(s)
(3) - Sous-type(s) d'aléa
       

Pays / Zone
Massif / Secteur
Site(s) d'étude
Exposition
Altitude
Période(s) d'observation
Whole territory of Switzerland, Liechtenstein, Austria, Slovenia and Croatia, and parts of France, Germany, Italy, Czech Republic, Slovakia, Hungary, Bosnia and Herzegovina. European Alps and its wider surroundings (4 to 19 °E, 43 to 49 °N) called ‘Greater Alpine Region’ or ‘GAR’        

(1) - Modifications des paramètres atmosphériques
Reconstitutions

 

Observations

OUTLINE OF THE LEADING CLIMATE VARIABILITY FEATURES IN THE GAR:

Temperature

The smoothed low-elevation subregional annual temperature series clearly show that the GAR-subregions only show small differences in their low-frequency temperature variability. This is also the case for the high- versus low-elevation subgroups. The remote observatories in the Alpine summit, at altitudes of 1500 to 3500 m, all show the same longterm temporal evolution as those in the long-term urban or rural sites in the valleys and plains in and around the Alps. The mountains neither show weaker warming (which would point at the low-elevation sites to be possibly not representative for the global background) nor show any warming stronger than the low-elevation regions (as sometimes claimed, for example by Beniston et al. (1997) and others). It is the GAR as a whole that has warmed twice as much since the late 19th century compared to the global or Northern Hemispheric average. This has also been confirmed by a study of the independently homogenised Swiss series (Begert et al., 2005) and Italian data (Brunetti et al., 2006). The difference of the series in the GAR minus CRU-N-HEM further differentiates this relative trend. In the first years of the common measuring period (1850s and 1860s) and in the 80 years from 1900 to 1980 there is no trend between the GAR and the hemispheric average, and only a stronger variability of the GAR is visible (which is not surprising as the GAR is a smaller region). The long-term effect is produced more or less exclusively by a cool GAR in the 1870s, 1880s and 1890s in contrast to a warm GAR in the 1980s and 1990s. A discussion of such outstanding climate events at the decadal scale is fully described in Matulla et al. (2005) on the basis of the HISTALP database and crosschecked against long-term runs of climate models.

The long-term temperature evolution in the GAR shows pronounced seasonal differences. The summer half-year series shows a typical subdivision with more than 100 years of cooling prior to the 1910s followed by warming in two steps (until 1950 and in the recent decades). In the winter half-year, the initial cooling was much weaker and ended near 1890, with the 20th century warming being more regular and less steplike.

Temperature and pressure

There is a striking similarity between the long-term trends of air temperature and air pressure. The 100- and 50-year trends of the annual data have the same sign throughout the 19th and the 20th century. The spring, summer and autumn trends also show this similarity of pressure and temperature trends for most of the 50- and 100-year subperiods. Interestingly, winter temperatures seem to have been influenced differently by air pressure in the course of the past two centuries. In the 19th century, the temperature trends are opposite to the trends of air pressure. This is in accordance with our traditional way of understanding Central European winters to be cold at high pressure, caused by an extension of the Asian winter anticyclone. In the 20th century, temperature and pressure trends increasingly tended to show the same sign rather than the opposite. This was most accentuated in the second part of the century, when a 4.5-hPa increase of winter air pressure went along with a 1.6-K warming. The uniform subregional pressure trends observed in the region point to the possibility that the increase in pressure do not necessarily have to be in connection with the continental East but may as well emerge from the South. Such a pattern, with reduced cyclonic activity during the Mediterranean winters, would enhance advection from the Atlantic and cause the observed warming of GAR-winters. It would also explain the observed centennial precipitation increase/decrease in the NW/SE of the Alps.

The fact that each variable has been homogenised separately makes the described pressure–temperature similarities all the more remarkable. Moreover, the homogenised air pressure data also give more confidence to the early instrumental period in general and may also be used as a first simple explanation of some of the regional GAR versus global climate features. There is also some divergence of the air pressure and the temperature curves, for example, the recent decoupling of the summer curves since around 1990. Taking into account the problems of detecting and adjusting recent inhomogeneities (adjusting the necessary length of an unaffected subperiod is not possible for this period) and also the current conversion of the networks (automation), it may well be an artefact. Anyway this interesting feature should be kept under close observation during future updations.

High-elevation minus low-elevation difference in air pressure and air temperature series show one of the strongest relationships between any pair of climate element combinations in the GAR. The respective correlation coefficients are 0.89/0.87 for AMJJAS/ ONDJFM. This is due to the thermodynamic expansion/ compression of the warming/cooling air masses below the high-elevation observatories, producing an increase/decrease of high-altitude air pressure relative to the one measured at low elevations. The effect is not new, but its manifestation in long-term series is small in relation to the inhomogeneities present in the original data. A feasibility study (Böhm et al., 1998) tried to use this approach to calculate the ‘non thermometric virtual air temperature series’ for air columns between some single high–low station pairs in the Eastern Alps. The model worked well for an annual ‘Eastern Alpine Standard Air Column’ temperature anomaly series. It confirmed the reality of the systematic bias in the ‘as measured’ temperature series and also provided a striking argument against those who doubted the warming, not believing it to be real, and considering it only as an artefact of urbanisation. In the Alps (the only place where long-term series of high- and low-elevation air pressure exist at altitude differences of 2 to 2.5 km), the warming of the last 100 to 120 years calculated from the air pressure series corresponds exactly to that measured by homogenised long-term series from the urban and rural sites of the region. This is a clear indication that a possible ‘systematic urban bias’ cannot be used as an argument to doubt climate warming. It is obviously possible to eliminate urbanization problems through careful homogenising – an experience not only gained by the authors (comp. Böhm, 1998) but also by others (Peterson, 2003; Parker, 2004). A more sophisticated relevant study on ‘barometric temperature series’ (making use of the quantitatively enlarged and qualitatively improved database) is currently in progress.

Sunshine, clouds and precipitation

Sunshine duration and cloudiness are two climate elements expected to be closely inversely correlated. Any similarities, however, are not as easy to determine as it may seem. Sunshine duration is measured as a total over a variable length of a day, whereas the sky coverage by clouds is estimated by an observer at fixed points in time each day. Also, the fact that the two elements have never been merged during the process of homogenisation makes results like the selected cases shown for the month of February in NW and for the month of May in SE rather satisfying. The inverse correlations of −0.87/−0.79 represent typical results and not specially selected examples. It is verified for all the respective monthly correlations in the five coarse resolution subregions of the GAR, for sunshine-cloudiness as well as for the positively correlated pair of precipitation and cloudiness.

The annual precipitation and cloudiness series are shown for the CRS series NW and SE. The increase in precipitation in the NW of the GAR over the last 140 years has its counterpart in a similar increase in sky coverage, while the long-term drying in the SE is accompanied by a decrease in cloudiness trend. The monthly correlation analysis also shows slightly lower but still strongly significant correlations than those of the sunshine–cloudiness pair. For both cases, correlations are more distinct from March to October and slightly, but clearly, lower during November to February. In the cold season, the situation becomes more complicated owing to the frequent occurrence of a decoupling of low-elevation local (stable) air masses (accompanied by inversions and low stratiform clouds or fog) from the higher parts of the atmosphere that are more exposed to advection from remote (maritime) regions. The existence of mountains with enclosed valleys and plains further stimulates and strengthens the decoupling in the cold season.

The comparison between sunshine duration, precipitation and high- and low-elevation cloudiness series also serves to illustrate and argue in favour of such a mechanism that is driven by vertical effects. At first sight, the parallel long-term increase of winter precipitation and sunshine duration in subregion NW seems to be inconsistent and appears to be an artefact of low data quality. A probable explanation is offered by the two cloudiness series observed at different altitudes. At low elevations in the GAR, high averages of cloud-cover during winter result from long-lasting episodes of stratiform low-level clouds. These produce no, or very little, precipitation and completely shade the lower parts of the GAR from sunshine. However, they do not reach the altitudes of the high-elevation observatories. If these stratus episodes are broken by some frontal passages (in the north-western part of the GAR typically advecting from the Atlantic) it brings rain or snow and increases the monthly total of sunshine hours. At high elevations (which have clear sky during the stratus-periods), above-normal frontal (precipitating) activity produces increased cloudiness. At low elevations the resulting effect of more high-level and less low-level clouds may produce a zero-result because there is only an exchange of one form of cloud against another form.

Relative humidity and vapour pressure The last example selected from the bulk of information present in the HISTALP CRSM series deals with the two climate elements representing air humidity – relative humidity and vapour pressure. Although they are linked by a non-linear relation (the Magnus Equation), their (linearly calculated) monthly means have been treated in HISTALP as independent climate elements. The different long-term evolutions of the two humidity measures confirms this. Of special interest is the comparison of the low- and the high-elevation humidity (shown for CRS mean north) in combination with the respective temperature curves. It shows that vapour pressure has more or less simultaneously increased with temperature, both at high and low altitudes. This corresponds to the general simple line of argument that ‘in a warmer world, there is more moisture in the atmosphere and precipitation is increased’. There is no doubt about this at the global scale. At the regional scale, on the other hand, the extent to which the humidity surplus can be transported from the (mainly oceanic) source regions into the continents is of fundamental importance. Yet the apparently nearly identical annual vapour pressure curves for high and low elevations of the GAR have to be set into the context of the significantly lower water vapour content (in an absolute sense) of the (colder, less dense) high-elevation air masses. Thus, the identical 20th century increase of approximately 0.6 hPa tells us about a more effective moisture transport towards the Alpine peaks compared to the less intensive one (in relative terms) into the valleys, basins and plains. This assumption is clearly supported by the long-term curves of relative humidity. Here we see that, in the upper levels, the moisture transport is able to balance the drying potential of the warming trend happening equally at high and low altitudes. Over the whole measuring period, relative humidity has remained relatively stable in the mountains. Only a few oscillations at the decadal scale are visible, the most prominent being a dry period in the 1890s and early 1900s, which has its logical counterparts in a high-elevation decrease of vapour pressure and an increase of temperature. In the valleys, basins and plains, in contrast, a long-term decrease of relative humidity has occurred and it has steepened in recent decades. An outstanding period of interest is the one on the decadal scale in terms of the concurrent warming near 1950 at high and low altitudes, which was accompanied by a relative humidity minimum at low elevations and a moisture peak (for vapour pressure and relative humidity) at high elevations. This period seems to have been warm and dry in the lowlands but typically one of increased transport of convective moisture towards the elevated observatories in the high mountains. A similar effect is also visible for the recent warming of the 1980s and 1990s. The increase in low-elevation vapour pressure of approximately 0.5 hPa is not enough to balance the increase in temperature of more than 1 °C. Therefore, relative humidity has decreased considerably in the north low-elevation subregion. In contrast, the smaller increase of only 0.3 hPa in the high-elevation subgroup has to be set in the context of the (much lower) saturation vapour pressure in the colder air at high elevations. Therefore, this 0.3-hPa increase in the last 20 years was sufficient to keep relative humidity rather stable in the Alpine mountains.

Multiple-element trend overview

The selection of examples of interesting features illustrated and discussed in this section so far was intended to raise interest in the broad field of applications that was supported by the new HISTALP database. The selection is far from systematic and does not cover the whole spectrum of four-dimensional climate variability in the region. The tables given in the paper outline a more systematic, but still compressed, GAR-overview of trends present in the annual/seasonal means/totals of the different climate elements. These tables show linear trends for fixed subperiods of 100, 50 and 25 years. To allow for comparisons with other regions and with global mean values the most common classification of the periods into centuries and half-centuries were used. The most recent half-century was again subdivided into two 25-year segments. The chosen subperiods also make sense in some respects in terms of major changes or reversals in climatic trend (e.g. the Atlantic Multi-decadal Oscillation). The two centennial subsections, for example, show a decrease in the 19th century in temperature (significant) and air pressure (not significant) and significantly increasing trends for both elements in the 20th century. The subdivision of the 19th century into two 50-year segments, on the other hand, is in accordance with a change in the mid 19th century, from decreasing precipitation in all subregions of the GAR to stable or increasing tendency in the second part. The stepwise increase in temperature in the 20th century is well described by using the 50-year segment from 1900 to 1950 (with significant warming in all subregions), and then the two subsequent 25-year section, with stable to slight (insignificant) cooling until the mid 1970s and the trend towards a strong warming of the recent 25 years.

The structure of the two tables (for annual and seasonal trends) also allows for quick geographical attribution. The left four trend values in each framed box are those of the low-elevation subregions NW (top-left), NE (top-right), SW (bottom-left) and SE (bottom-right). The two values in the right columns of each box are the high-elevation (top) and low-elevation trends. The tables provide quantitative information on many of the features already discussed in the examples given in the figures cited in this section. Key values (concerning the annual trends) are the −0.5 to −0.8 °C temperature decrease (significant in all subregions) in the 19th century and the subsequent 1.0 to 1.4 °C increase (also significant in all subregions) in the 20th century. Other cases of significant trends in all subregions are given for air pressure for the 20th century (increase of 0.9 to 1.3 hPa) and for the 50-years from 1950 to 2000, for temperature for both 50-year subsections of the 20th century and for the recent 25-year (all increasing), for vapour pressure (0.4 to 0.9 hPa increase in the recent 25 years) and for relative humidity with the drying of −1.8 to −10.4% from 1950 to 2000 (however, it has to be remembered that the two humidity elements are not available for all subregions).

Another outstanding event is the sharp trend-reversal for autumn-precipitation in the 1970s, from a long-term decreasing tendency (significant in the 19th century) to a sudden increase of 23 to 35% in the last 25 years (significant in the NE and for the low-elevation mean). A recent reverse is visible for winter. GAR-winters became drier in the last 25 years, significantly (44% in relation to the 20th century mean) in the SW and for the low-elevation mean (−27%). The recent trends in winter precipitation have been accompanied by respective trends in sunshine (significant increase in all subregions of 17 to 29%), cloudiness (non significant decrease of 3.1 to 7.4%) and relative humidity (significant decrease for all available low-elevation subregions, strongest in the SE with −4.8%).

Most of the other trends are variable from subregion to subregion. The element with the highest spatial variability of trends is precipitation. There is a 10% drying in the SW with an 8% wetting in the NE during the recent 25 years (annual totals). There is also a 10% drying in summer from 1950 to 1975 in the NW compared to the parallel 20% wetting in the SE, or the 16% decrease in long-term spring precipitation versus the contemporaneous increase in 27% spring precipitation during the 19th century.

For the other elements, some striking differences in trends exist between high and low elevations. For example, the centennial evolution of total sunshine in summer had a significant decrease (−9%) at low elevations, but a significant increase (+8%) at high elevations. Another ‘vertical decoupling’ is given for both 50-year segments and the 100-year trends of the 20th century for relative humidity in all four seasons. The decoupling is most pronounced in the more continental eastern subregions.

The indicated examples cover only a part of the information present in the two tables. Moreover, the idea of linear trend analysis for fixed subperiods only provides a restricted view on the full array of four-dimensional climate variability in the GAR. A more sophisticated respective analysis has already been undertaken for GAR-precipitation (Brunetti et al., 2005) and will be extended to all climate elements of the HISTALP database soon.

Modélisations

 

Hypothèses

 


Informations complémentaires (données utilisées, méthode, scénarios, etc.)

This paper describes the HISTALP database, consisting of monthly homogenised records of temperature, pressure, precipitation, sunshine and cloudiness for the ‘Greater Alpine Region’ (GAR, 4–19 °E, 43–49 °N, 0–3500m asl). The longest temperature and air pressure series extend back to 1760, precipitation to 1800, cloudiness to the 1840s and sunshine to the 1880s. A systematic QC procedure has been applied to the series and a high number of inhomogeneities (more than 2500) and outliers (more than 5000) have been detected and removed. The 557 HISTALP series are kept in different data modes: original and homogenised, gap-filled and outlier corrected station mode series, grid-1 series (anomaly fields at 1° × 1°, lat × long) and Coarse Resolution Subregional (CRS) mean series according to an EOF-based regionalisation. The leading climate variability features within the GAR are discussed through selected examples and a concluding linear trend analysis for 100, 50 and 25-year subperiods for the four horizontal and two altitudinal CRSs.

HISTALP DATA MODES

The paper provides a first comparative outline of the leading climate variability features in the GAR on the basis of the concept of ‘Coarse Resolution Subregional Means’ (CRSMs). The results of the regionalisation are discussed in the paper.

The climate information in HISTALP is stored in station mode and grid mode. The former provides two kinds of data: ‘stmod-ori’ (monthly/seasonal/annual means or totals of the original data) and the subsequent ‘stmodhom’ data. Stmod-hom series have passed the homogeneity tests, outlier correction and gap filling procedures.

Three climate elements, air pressure means (P01), air temperature means (T01) and precipitation totals (R01) have been also transferred into monthly anomaly fields (with respect to 1901–2000 averages) at a grid resolution of 1° latitude and longitude. These ‘grid-1’ series will reduce the remaining undetected inhomogeneities/outliers through the process of averaging, overcome any remaining inhomogeneities of the network in terms of spatial distribution of the sites and allow for easier mathematical treatment in different kinds of analyses. The method applied was a modified Gaussian weighted inverse distance (IDW) interpolation. The filter width of the weighting function was set with respect to spatial de-correlation of the respective climate element (least for precipitation, largest for air pressure). A few steep climate gradients in the Alpine main chain for all elements, coastal versus inland for temperature and a few others, were initially defined as barriers to information transport. Also, the search radius was set according to the specific spatial decorrelation. This was a measure to avoid transfer of information to a grid point over unrealistically long distances, especially in the earlier times when network density was less. For temperature, two such grid-1 data sets were produced. The version ‘high-elevation’ is only present for grid points in the direct Alpine realm and was calculated from a selection of high-elevation or summit series. The version ‘low-elevation’ comprises all grid points in the GAR from 4 to 19 °E and 43 to 49 °N.

For precipitation, the series from the wind exposed summit sites have already been excluded from the stmod data set (Auer et al., 2005) owing to the well known uncertainties of precipitation measurements at high-elevation Alpine sites [see references in the paper], consequently, no high-elevation grid-1 version was produced.

For air pressure, only sites up to a maximum altitude of 650 m asl were used for grid point-interpolation. These low-elevation fields can be expected to carry the main information usable for questions of circulation. Higher-elevation sites were excluded owing to their systematic bias originating from the temperature and humidity of the atmospheric layers beneath. However, studies targeting such effects directly remain in the low- and high-elevation pairs of the stmod series (e.g. Böhm et al., 1998).

At present, a second kind of a gridded data set has also been produced for precipitation. R01-grid-2 consists of absolute monthly precipitation totals at a high spatial resolution of 1°/6° lat/long. Efthymiadis et al., 2006 describe the procedure in detail. The high-resolution data set is principally based on two main sources. The HISTALP stmod-hom data set provides the long-term information back to 1800. It has been merged with the shorter (1971–1990) but higher resolved ETHZ-precipitation climatology (Schwarb, 2000; Schwarb et al., 2001). The basic assumption is the relative stability of the (highly resolved) ETHZ-patterns of spatial variability allowing the statistically-sophisticated extension back in time via the (spatially less resolved) HISTALP stmod data.

REGIONALISATION – THE CONCEPT OF COARSE RESOLUTION SUBREGIONS

This paper intends not only to describe the acquisition, quality enhancement and general structure of the HISTALP database but also to give a first survey of the general climate variability features in the region. It is clear that there have been significantly different short- and long-term climate evolutions in the GAR (e.g. for precipitation, as described by Brunetti et al., 2005). Therefore, it was necessary to reduce the number of localto- subregional differences, from element to element and from subperiod to subperiod, to some principal features.

A complete spatial compression to an average series over the entire GAR was not feasible because they would not be representative for any subregion, particularly for spatially variable elements like precipitation. To provide maximum information of a manageable size, the results of single-element regionalisations were optimised to a limited number of subregions that were identical for all climate elements. Regionalisation of each single element is based on principal component analysis (PCA) applied between all station records for the longest possible common period (approximately 1930 to 2000, with slight changes of a few years based on the availability of each element). The resulting empirical orthogonal functions (EOFs) were entered into an orthonormal transformation, which is subject to the so-called ‘varimax-criterion’ rotated empirical orthogonal functions, REOFs. Experience of the author's group from earlier attempts determined the way (Matulla et al., 2003) PCA was used (rotated EOFs) to investigate the spatial and seasonal variabilitiy of Austria’s precipitation climate throughout the 20th century. REOFs were also used to regionalise the HISTALP temperature (Matulla et al., 2005) and the precipitation data set (Brunetti et al., 2005).

Here, PCA were applied to all available climate variables on an annual basis (normalised air pressure, air temperature, precipitation, cloudiness and sunshine duration data). Regionalisation based on seasonal combinations (not shown) supports the annual results, but it also shows some interesting smallscale differences. The map presents the single-element regionalisations (thin lines) that are based on the four leading EOFs for each element, respectively. The loadings on all elements, except precipitation, also suggested a vertical stratification (e.g. a strong decoupling of lower and higher atmospheric levels is evident for winter temperature). Therefore, an additional subgroup were introduced for high-elevation summit sites (stations within this Alpine subgroup are highly dispersed and are therefore not mapped as an area on the subregions map). For precipitation, the isolation of a specific high-elevation subgroup was not attempted owing to the exclusion of summit sites from the data set for this climate element.

In order to allow comparisons between different climate elements, a regionalisation into uniform subregions (coarse resolution subregions – (CRSs)) that are the same for all climate elements was necessary. These CRSs are semi-subjectively optimised solutions in order to gain the best possible common regions for all climate elements. In general, the adjustments of single-element regionalisations are considered to be small and physically reasonable. Most of the climate elements show a favourable subdivision into four horizontal subregions of approximately similar size. They correspond to the northwest– northeast–southwest–southeast scheme.

In detail, there were no compromises necessary along a section of the Central Alpine chains, from the La Grave–Les Ecrins group in the West to the Hohe Tauern group in the East. This is the sharpest climate border existing in the GAR identical for all five elements. It is part of the continental scale transition zone, from the temperate westerly to the Mediterranean subtropical climate. Further to the west, this climate border splits into three branches. Two (for air pressure and temperature) bend to the south and continue to follow the Alpine bow a little further, whereas the other elements leave the Alpine crest and cross the Rhone valley in a zonal direction towards the Massif Centrale.

The strongest blurring of this divide is seen in the zonal climate in the east. For cloudiness and sunshine, it bends to the south and generally follows the Dalmatian coast. For air pressure, it moves further inland and separates at the Dinaric Alps and Bosnian Mountains from the Slavonian–Hungarian plains. For temperature, it continues more or less in the zonal direction as it did in the Alpine parts. For precipitation, surprisingly, southern influences reach as far to the north as southern Slovakia. Here in the east, the strongest compromises had to be accepted to define a common CRS-border, which was finally chosen similar to the temperature line along the 46th parallel.

A second continental scale climate border could be anticipated between (western) oceanic influences of the Atlantic and (eastern) continental features of the Eurasian continent. Surprisingly, the climate transitions emerged to be rather uniform and steep for the different climate elements, although no additional topographic forcing through mountain chains exists here. They roughly follow the 12th degree eastern meridian in the parts of the GAR north of the Alps. In the south, a separate diversification along 12 °E appeared only for temperature and precipitation. This splitting into a southwestern ‘Tyrrhenian’ and a southeastern ‘Adriatic’ subregion is not visible for air pressure, sunshine and cloudiness. These three elements split into three horizontal subgroups only – most likely a stable solution for air pressure, and a preliminary estimate for sunshine and cloudiness owing to network deficiencies that still exist in parts of the GAR. Anyway, the subdivision into four horizontal subgroups was maintained also for these elements and a compromise between temperature and precipitation was followed in the south. This should cause no problems for pressure, cloudiness and sunshine – simply producing very similar respective SW- and SE-series – and the existing SW–SE differences for precipitation and temperature are not destroyed.

Four low-elevation coarse resolution subregions were finally chosen. The fifth (high-elevation) subregion is indicated as a chain of triangles following the main crest-line of the Alps. The different combinations of subregions used in the paper are also shown (schematic graphs). Together, ten subregions are defined as various combinations of the five main CRSs (NE: Northeast; NW: Northwest; SE: Southeast; SW: Southwest; H: High; L: Low; N: North; S: South; W: West; E: East). All low-elevation subregions of air pressure and temperature go back to the 18th century, and for precipitation to the year 1800. Sunshine series start in the 1880s, with some cloudiness subregional series starting earlier, in the 1840s and 1850s. The four low-elevation subregions for which humidity series are available point to the potential of this climate element that is not yet fully processed and used – neither in the GAR nor in most other regions of the world. Relative humidity series go back as far as the 1860s, those of vapour pressure go back even further in some cases. The reason vapour pressure is a longer series is that it is more easily homogenised compared to the more scattered relative humidity fields. The typical starting time of high-elevation series is in the 1860s to 1880s, when the classical Alpine summit observatories were founded (e.g. Wege, 2004; Eckert, 2004; Böhm, 2004; Auer et al., 2002).

For each of the ten CSRs, monthly, seasonal, halfyear and annual CRSM series have been calculated for air pressure, air temperature, cloudiness and sunshine duration. Nine CRS sets of series could be produced for precipitation (all except group 5) and four (NW, NE, H and N) for relative humidity and vapour pressure. The following section discusses a sample of typical and interesting cases from the total of more than 1000 single graphs that resulted from the reduction of the data sets to the few leading principal components.

OUTLINE OF THE LEADING CLIMATE VARIABILITY FEATURES IN THE GAR

The HISTALP database achieved its present state in spring 2005. There has been no comprehensive study as yet making full use of its currently available potential. One study that deals with one single climate element has been undertaken. Brunetti et al. (2005) analysed the precipitation variability in the GAR on the basis of the HISTALP database released recently. Böhm et al. (2001) made use of an earlier version of the temperature data set with a lower station density and without any outlier correction. Detailed analyses of the other elements of the database have not yet been undertaken. The current potential of HISTALP is promising, as it makes possible an advanced physical understanding on long-term climate variability arising from comparative analysis of different climate elements over one to two centuries.

Users of these early data should be aware that the general quality level of the series during the last decades of the 18th century and the first two of the 19th does not match that of the recent parts of the instrumental period. However, considering the amount of work invested throughout the past decade and the lessons learned, the authors are convinced that their early homologous series is valuable when used with care. The network density in relation to spatial de-correlation has allowed quite important conclusions to be drawn for temperature, air pressure and precipitation for decades in which only indirectly derived climate information was available so far. As the series originate from different sources, institutes and administrations, they are also optimistic of the detection potential of systematic biases.


(2) - Effets du changement climatique sur le milieu naturel
Reconstitutions

 

Observations

 

Modélisations

 

Hypothèses

 


Sensibilité du milieu à des paramètres climatiques
Informations complémentaires (données utilisées, méthode, scénarios, etc.)

 

 


(3) - Effets du changement climatique sur l'aléa
Reconstitutions

 

Observations

 

Modélisations

 

Hypothèses

 


Paramètre de l'aléa
Sensibilité des paramètres de l'aléa à des paramètres climatiques
Informations complémentaires (données utilisées, méthode, scénarios, etc.)

 

 

 


(4) - Remarques générales

 


(5) - Syntèses et préconisations

CONCLUSION AND OUTLOOK

The main objective of this paper was to introduce a new database, which is the result of approximately 10 years of data collection, digitising and quality improvement. The target area is the ‘Greater Alpine Region’, which is of special climatological interest owing to its location at the intersection of four principal climate regimes, and additionally modified by vertical effects. The efforts were targeted at the elimination of shortcomings due to inadequate spatial coverage and resolution, exploitation of the data potential in the early instrumental period, followed by a multiple variable approach that was carried out under strict conditions with respect to data quality. To date, the authors succeeded in fulfilling these requirements for five leading climate elements at monthly resolution, but the database is in permanent development. The task for the future will be to continue the systematic high-resolution climate monitoring in the region with adherence to quality requirements. An extension of the database to other climate elements and an increase of the time resolution from monthly to daily are underway. In addition to these monitoring activities, the authors hope that the HISTALP database will be intensively used for climate variability analysis and for all kinds of climate impact research in the European Alps – a region of high vulnerability to climate change. Grid-1 data and CRSMs are available to the public via the ZAMG homepage (http://www.zamg.ac.at).

Références citées :

Auer et al., 2002

Auer et al., 2005

Begert et al., 2005

Beniston et al. (1997)

Böhm, 2004

Böhm et al., 1998

Böhm et al. (2001)

Brunetti et al., 2005

Brunetti et al., 2006

Eckert, 2004;

Efthymiadis et al., 2006

Matulla et al., 2003

Matulla et al. (2005)

Parker, 2004

Peterson, 2003;

Schwarb, 2000

Schwarb et al., 2001

Wege, 2004;