Figure 4 presents the estimated RVs of dry winter extremes in four regions (see rectangles in Fig. 1) with RPs of 5, 10, 20, and 50 years. The RVs are estimated for 50-year windows from 1700 to 2000 based on 1000 bootstrap samples, taking into account the uncertainty of the estimates of the gamma parameters. Over northern Ireland (Fig. 4a) the RVs gradually decreased since 1700, i.e. extremely dry winters became more severe. Additionally, some interdecadal fluctuations can be observed. The RVs over southern Spain (Fig. 4b) and over north-eastern Europe (Fig. 4c) show a weak overall negative trend since 1700 with superimposed interdecadal fluctuations. The temporal evolution of the RVs in central Europe (Fig. 4d) is characterized by a slight increase. Significance estimates of the changes 1700–2000 relative to 1951–2000 are given in Fig. 5. Over northern Ireland the RVs were significantly different from 1951–2000 only during the first part of the 18th century, while over southern Spain this was the case up to around 1880 (Fig. 5a and b). Over north-eastern Europe (Fig. 5c) the 50-year RVs were different from 1951– 2000 from 1750 to 1900, while the RVs with frequencies of 5–10 years remained the same since 1700. Over central Europe the RVs during the 18th century differed significantly from 1951–2000 (Fig. 5d).
Similar to Fig. 4, Fig. 6 displays the estimated RVs of wet winter extremes. The RVs in northern Ireland (Fig. 6a) are dominated by a positive trend, which is enhanced during the 20th century. The RVs over southern Spain (Fig. 6b) are characterized by a peak during the second half of the 18th century and by a positive trend during the 20th century. The RVs in north-eastern Europe (Fig. 6c) feature similar characteristics. In central Europe (Fig. 6d) the RVs are highest around 1750, suggesting severe wet extremes during that time. Another feature is the positive trend during the 20th century. However, the RVs do not reach the levels of 1750. Significance estimates of the changes 1700–2000 relative to 1951–2000 are given in Fig. 7. All RVs are significantly different from 1951–2000 except over southern Spain during the second part of the 18th century and over central Europe during the whole 18th century.
Figure 8 presents the spatio-temporal variation of RPs of dry extremes, which had a RP of 20 years during 1951– 2000. Only areas with reconstructive skill (RE>0) during the 1701–1720 period are considered (therefore the white areas over parts of Europe). We chose this period as benchmark because thereafter the skill generally increases (Pauling et al., 2006). The significance of changes in the displayed RPs relative to the RPs during 1951–2000 for the regions marked by the black rectangles in Fig. 8e can be obtained from Fig. 5. As Fig. 8 is based on the 20-year RVs, the horizontal line with the black dots in Fig. 5 (end of the analysed 50-year periods) can be used to determine the significance.
As seen from Fig. 8a dry winters in western Europe with a RP of 20 years during 1951–2000 occurred every 5-10 years during 1701–1750. This change is significant at the 5% level (see corresponding black dot in Fig. 5d). On the other hand, dry winters were less frequent over the Iberian Peninsula, south-eastern Europe and the British Isles. The same situation persists during 1751–1800 over Europe (Fig. 8b) except that in eastern Europe dry winters were less frequent. During 1801–1850 (Fig. 8c) almost over whole Europe dry winters were less frequent than during the 1951–2000 reference period. From 1851–1900 (Fig. 8d) the pattern is again very similar to the one during the early 18th century (Fig. 8a). The same holds for 1901–1950 (Fig. 8e), although the magnitude of the changes of the RP is less pronounced.
Figure 9 displays the same analysis as Fig. 8 for wet extremes. The significance of the RPs relative to the RPs during 1951–2000 for the regions marked by the black rectangles in Fig. 9e is given by the black dots in Fig. 7. From 1701–1750 wet winters were up to twice as probable over central Eu rope and parts of eastern Europe than during the 1951–2000 period (Fig. 9a). This change is, however, not significant at the 10% level (see corresponding black dot in Fig. 7d). From 1750–1950 wet winters were less probable compared to 1951–2000 (Fig. 9b–e) over whole Europe. This change is mostly significant (Fig. 7a–d).