Interpretation of the observed changes in extreme frequencies together with atmospheric circulation may yield insights into important climate mechanisms. Figures 4–7 focus on the temporal evolution of the changes of the RPs. The fluctuations of the RVs of dry extremes as displayed by Fig. 4 can be linked to changing prevalence of circulation types. In southern Spain (Fig. 4b) the dry RVs were significantly less severe between 1750 and 1850 than during the most recent 50 years. This may be due to the anomalously dry condi tions in the 1990s that were linked to the well-known positive states of the NAO. This feature appeared to be unique for at least the past 300 years in that region (Fig. 4b). An interesting feature are the very severe RVs of dry winter extremes in central Europe at the beginning of the 18th century.
This can be well interpreted as the end of the Maunder Minimum (1645–1715), a period known for its dry and cold winters in central Europe (e.g. Wanner et al., 1995; Luterbacher et al., 2001, 2002, 2004). It is well known that during that period the influence of the Russian high increased during some winters which led to persisting advection of cold and dry continental air (e.g. Wanner et al., 1995; Luterbacher et al., 2001, 2002, 2004; Xoplaki et al., 2001; Shindell et al., 2001; Jacobeit et al., 2003). Moreover, this coincides with few extremely dry winters in southern Spain (Fig. 4b). Negative NAO states may provide a physical explanation as they are connected with dry and cold winters in central Europe and wet winters in southern Europe (e.g. Hurrell, 1995;Wanner et al., 1995; Luterbacher et al., 2001, 2002, 2004, 2006; Xoplaki et al., 2004). On the other hand, also wet winters became more severe at the end of the Maunder Minimum especially over central Europe (Fig. 6d), leading to the observed increase in the year-to-year variability of winter precipitation. The physical mechanism leading to this change is not completely understood. Shindell et al. (2001) conclude from modelling studies that low solar irradiance forces the NAO toward the low index state. Wanner et al. (2000) argue that sea surface temperatures (SSTs) may have been high in the north Atlantic promoting high pressure over the Atlantic which helped block the westerlies and facilitate the advection of cold and dry continental air to central Europe. Luterbacher et al. (2001) and Shindell et al. (2001) further suggest that increasing solar irradiance at the end of the Maunder Minimum might lead to a strengthening of the NAO through complex interactions between the troposphere and stratosphere, and thus to a general continental warming and wetter conditions in northern Europe.
Figures 8 and 9 provide a spatially detailed view of the changes of RPs of 20-year return values the back to 1700. Figure 8 shows that most of western Europe has experienced more frequently dry extremes back to 1700 compared to 1951–2000 except during the first part of the 19th century. However, only the 1701–1750 period was significantly different from 1951–2000 (see black dots in Fig. 5d). Other regions such as Iceland, western Norway, parts of eastern Europe and Turkey experienced less frequently dry winters. These results suggest that the circulation changes over the last centuries affected all regions but in a different way. While in central and eastern Europe circulation changes caused more frequent dry extremes during the 18th century, over most other European regions dry extremes happened less frequently during the same period. During the 1800–1850 period the dry extremes were less frequent than during the 18th century and the second part of the 19th century in central Europe. This is in line with Jacobeit et al. (2001, 2003) who found north-westerly flow over Europe to be dominant during 1830–1850. This circulation type advects moist and cold air to central Europe which may also have contributed to the marked and well documented advances of Alpine glaciers such as the Lower Grindelwald Glacier, Switzerland (Zumb¨uhl, 1980; Zumb¨uhl et al., 1983; Zumb¨uhl and Holzhauser, 1988; Steiner et al., 2005). However, less dry extremes do not necessarily mean wetter conditions on average (Katz and Brown, 1992; Schaeffer et al., 2005) which may be more important for the mass balance of glaciers than changes in the extremes. This argument is supported by Steiner et al. (2007)1 who found from a sensitivity study using neural networks that summer temperature was the dominant factor leading to the advance of the Lower Grindelwald Glacier during the first part of the 19th century. An important issue is the link of observed (reconstructed) changes in extremes to the underlying causes. However, the link to forcing factors such as solar variability, volcanic activity and greenhouse forcing is hard to establish (Easterling et al., 2000; Wanner et al., 2001; Yoshimori et al., 2005). Solar activity may play a certain role as the above discussion on circulation during the end of the Maunder Minimum (reduced solar irradiance) has shown. Still, the possible physical mechanisms leading to this is not well understood. Volcanic eruptions may also influence the presented results. However, the influence of volcanic events typically persists for only 1–3 years (Robock, 2002; Shindell et al., 2004). Hence, it is not surprising that there is no obvious connection between volcanic events and the observed changes in the extremes, which are analysed using 50-year periods. Additionally, all possible influences by the forcing factors are superimposed by internal variability. The clearest interpretation of the influence of the forcing factors is via circulation changes as proposed above. As the example of the end of the Maunder Minimum has shown, the increase of cold and dry winters due to circulation changes have affected very much the probability of extremely dry winters. These circulation changes are linked in a complex way to both solar irradiance and volcanoes (Shindell et al., 2001) on different time and space scales, but also to other factors such as the distribution of sea surface temperatures (Wanner et al., 2000; Luterbacher et al., 2001). The observed increase in wet extremes during the recent decades appears to be unique for northern Ireland, southern Spain and eastern Europe for the last 300 years (Fig. 6a–c). This may be a result of the intensification of the water cycle which can be a result of anthropogenic greenhouse forcing. On the other hand the presented analyses have shown that there have been substantial fluctuations in extremes before the onset of the human influence on climate. However, establishing the chain of causation is not trivial because many interactive factors influence the occurrence of extremes. One way to address this task is experiments using physical climate models, an objective for further investigations.