FIGURE I 1.3 August temperature anomalies in northeastern Japan (northern Honshu) from 1770 to 1840, derived from historical records of rainfall frequency.The reference period (zero line) is the mean for 1951-1970 (Mikami, 1992b).

1131-1210, when discontinuous records were made of the dates of the last snowfall in spring. Comparison with Hangchow records for the period 1905-1914 indicates that snowfalls commonly occurred 3-4 weeks later in the spring months during the twelfth century than was typical of the early twentieth century (Chu, 1926). This suggests a more prolonged, and probably more severe, winter during the Southern Song Dynasty (A.D. 1127-1279).

More modern and more complete records of snowfall (dates of first and last occurrences) were compiled for the London area by Manley (1969). These extend back to 1811 and indicate a reduction in the snow season of approximately 6 weeks between 1811-1840 and 1931-1960, though most of this change has occurred during the twentieth century, perhaps due to a marked increase in the urban heat island effect. A similar problem may have affected recent observations of first snow cover in the Tokyo area. During the period from 1632-1633 to 1869-1870 the average date of the first snow cover was January 6, whereas from 1876-1877 to 1954-1955 the mean was January 15 (Arakawa, 1956a). For both periods, the standard deviations were similar (approximately 20 days). How much of the change is due to a warmer Tokyo metropolitan area is difficult to assess, and points to the particular value of data from more rural locations. Pfister, for example, has compared the number of days with snow cover (extensive snow on the ground) at various places with similar elevations on the upper and lower plateaus of Switzerland. During the eighteenth century, snow cover was often far more persistent than in even the harshest winter of the twentieth century, 1962-1963 (Table 11.2) and this provides strong support for the record of "days with snow lying" in the Zurich-Winterthur area (Fig. 11.4). These data point to the decade 1691-1700 as having had the highest number of days with snow on the ground (> 65 annually) compared to recent

TABLE 11.2 Number of Days With Snow Cover per Year (March to May Values in Parentheses) for Selected Localities in Switzerland

Winter of

Lower plateau

Upper plateau


126 (45)


134 (51)

154 (60)


112 (35)


59 (0)

86 (12)

From Pfister (1978a).

" Harshest winter of the twentieth century. Severe winters were also noted in 1684-1685, 1715-1716, 1730-1731, and 1969-1970.

From Pfister (1978a).

" Harshest winter of the twentieth century. Severe winters were also noted in 1684-1685, 1715-1716, 1730-1731, and 1969-1970.

averages of only half that number (Pfister, 1985). It is also of significance that this decade was the coldest in central England, according to long-term instrumental records of temperature (Manley, 1974). Thus, urban growth effects, while undoubtedly exerting a progressive influence on temperature, cannot account for all the marked changes observed.

An interesting method of assessing past winter temperatures has been demonstrated by Flohn (1949), who observed that the ratios of snow days to rain days in winter months correlate well with winter temperatures during the instrumental period. Using sixteenth century observations of Tycho Brahe in Hven, Denmark (1582-1597) and of Wolfgang Haller in Zurich (1546-1576), Flohn was able to show that winters after 1564 were increasingly severe, leading to a marked increase in glacial advances in the Alps in the early seventeenth century. At Hven, winter

FIGURE I 1.4 Fluctuations in the number of days with snow lying in Zürich (note break in abscissa scale). Prior to 1800, figures have been estimated based on daily non-instrumental observations. Observations for 1721 -1738 made In Winterthur, 20 km northeast of Zürich. Value plotted as 10-yr running means with value at year n corresponding to the decade n to (n + 9) (Pfister, 1978b).

FIGURE I 1.4 Fluctuations in the number of days with snow lying in Zürich (note break in abscissa scale). Prior to 1800, figures have been estimated based on daily non-instrumental observations. Observations for 1721 -1738 made In Winterthur, 20 km northeast of Zürich. Value plotted as 10-yr running means with value at year n corresponding to the decade n to (n + 9) (Pfister, 1978b).

temperatures from 1582 to 1597 averaged 1.5 °C below those around the early twentieth century. It is also interesting that sixteenth century weather singularities (synoptic events that recur at the same time each year) continue to be observed in the twentieth century, indicating an underlying cyclicity in the general circulation that has persisted in spite of fundamental changes in the climate of the region.

In the Kanazawa area of Japan (west central Honshu) historical records of snowfall were maintained by the ruling Maeda family from 1583 until -1870. Based on detailed analysis of these and subsequent records, Yamamoto (1971) constructed an index of snowfall variation that corresponds reasonably well with instrumentally recorded winter temperatures in recent years. Although no precise calibration has been attempted, the index gives an overall impression of snowier winters, particularly in the first half of the nineteenth century, and this conclusion seems to be supported by other snowfall indices derived from Japanese historical sources.

Another observation frequently noted in historical records is the incidence of frost during the growing season, an occurrence of particular significance to agriculturalists. As with snowfall, long records of frost occurrence are available from China, in particular from the farming regions of the mid and lower sections of the Yellow River (Huang-he). Higher frequencies of frost were recorded during the periods 1551-1600, 1621-1700, 1731-1780, and 1811-1910. However, the range is small (from 1 to 6 events per decade) and the significance of such changes in the frequency of extremes for the overall growing season temperature is not clear. Other historical records indicate that the frost-free period in Inner Mongolia and northeastern China from 1440 to 1900 was -2 months shorter on average than during the twentieth century. Similarly, in southern China the frost-free season in recent decades has been 5-6 weeks longer than the long-term mean from 1440 to 1900 (Zhang and Gong, 1979).

European historical archives have provided a variety of non-instrumental records of value to the climatologist. Many of these sources were evaluated by Lamb to produce an index of winter severity and summer wetness spanning the last 1000 years (Lamb, 1961, 1963, 1977). Lamb concluded that the period A.D. 1080-1200 was characterized by dry summers throughout Europe, the like of which has not been seen since. He therefore designated this interval the "Medieval Warm Epoch" (Lamb, 1965, 1988). Recently, this has generated considerable interest as a possible analog for future, greenhouse-gas induced climates. However, subsequent studies have provided mixed support for this concept (Hughes and Diaz, 1994); the available evidence is limited (geographically) and equivocal. A number of records do indeed show evidence for warmer conditions at some time during this interval, especially in the eleventh and twelfth centuries in parts of Europe, as Lamb pointed out. However other records show no such evidence, or indicate that warmer conditions prevailed, but at different times. Indeed, not all seasons may have been warm; for example winters at this time were relatively harsh in western Europe, at least until A.D. -1170 (Alexandre, 1977). This rather incoherent picture may be due to an inadequate number of records, and a clearer picture may emerge as more and better calibrated proxy records are produced. However, it is not yet possible at this point to say whether the notion of a Medieval Warm Epoch should be considered as a worldwide episode or of no more than regional significance.

One of the most interesting and unusual studies of historical material is that of Neuberger (1970), who examined the changing climate of the "Little Ice Age" (sixteenth-nineteenth centuries) in Western Europe through artists' perceptions of their climatic environment, as depicted in contemporary paintings. Over 12,000 paintings from the period 1400-1967 were examined and wherever possible the intensity of the blue sky, visibility depicted, percentage cloudiness, and cloud type were categorized for each painting. More than half of the paintings contained some sort of meteorological information and the basic characteristics were averaged for different periods within the last 570 yr, as shown in Fig. 11.5. During the period 1400-1549, paintings have a high percentage of blue sky, good visibility, and little cloud cover. Paintings completed during the next 300 yr were generally darker, with less blue sky, showed lower visibilities, and had a much higher percentage of cloud cover and a greater frequency of low and convective type clouds. Over the last 100 yr there has been some reduction in cloudiness depicted and a drop in low and convective cloud frequency, though visibilities remain low, perhaps reflecting increasing atmospheric turbidity due to industrial and agricultural activity. It is remarkable that, in spite of many changes in style, the artists have captured in their paintings a significant record of climatic variation through time. Perhaps this perceptual record, more than any cold statistic, indicates the degree to which life during the "Little Ice Age" was affected by a deteriorating climate.

11.2.2 Historical Records of Weather-Dependent Natural Phenomena

Of all the weather-related natural catastrophes, floods and droughts appear to have had the most widespread and persistent impact on human communities, as records of these events are found in historical documents from all over the world (Pfister and Hachler, 1991; Pavese et al., 1992; Barriendos, 1997). The longest and most detailed records come from China, where regional gazetteers have been kept in many provinces and districts since the fourteenth century and many records are available for the last 2000 yr (Chu, 1973). Each of these gazetteers recorded local facts of historical or geographical interest as well as climatological events of significance to agriculture and the local economy (for example, droughts, floods, severe cold snaps, heavy snowfalls, unseasonable frosts, etc.). Not surprisingly, the gazetteers have been the focus of much interest, though there are often many difficulties in interpreting the data (see Section 11.2). In particular, technological improvements such as the building of irrigation channels or drainage ditches may drastically reduce the frequency of climatological disasters. For example, the Chinese province of Sichuan is unusual in that recorded occurrences of floods are rare, yet droughts are common. Usually, one finds over long periods of time a similar number of extremely wet and extremely dry events. The reason for this anomaly appears to be the particularly efficient flood-control measures introduced by the administrator Li Ping 2100 yr ago, which were able to reduce flood hazards but did little to alleviate the perils of droughts (Yao, 1943). In a similar way, the rise in flood and drought frequency during the Yuan Dynasty (A.D. 1234-1367) may be partly due to the destruction of irrigation and drainage systems by the Mongol invaders of the time (Chu, 1926).

Was this article helpful?

0 0

Post a comment