Info

SKY VISIBILITY LOW & CLOUDINESS DARK BLUE CONVECTIVE PAINTINGS

CLOUDS

FIGURE I 1.5 Changes in the frequency of certain characteristics of European paintings in the periods 1400-1549,1550-1849,and I850-I967 (Neuberger, I970).

Early work attempted to reduce the bias due to technological changes and increasing observations in recent history by assuming that such changes affected drought and flood observations equally. An index of "raininess" or precipitation anomaly was obtained by expressing the number of droughts (D) and floods (F) as a ratio (D/F), a common approach to reducing data errors in historical climatology. Thus, Yao (1942) was able to characterize each century over the last 2200 yr as "wet" or "dry" compared to the mean ratio for different regions of China (Table 11.3).

A detailed spatial-temporal data base of floods and droughts in China over the last 500 yr was compiled by the State Meteorological Administration (1981); each regional record, year by year, was classified into one of five anomaly classes (wet to dry). Wang and Zhao (1981) subjected these matrices to principal components analysis. One hundred and eighteen stations from almost the whole of eastern China were used in the analysis of data from 1470 to 1977. The resulting eigenvectors indicated broad-scale patterns of "precipitation anomalies" and these were then compared with eigenvectors derived from instrumentally recorded summer precipitation data for the period 1951-1974 (Fig. 11.6). Summer was chosen because it is the time corresponding to most recorded droughts and floods (Yao, 1942). Similarities between the principal modes of precipitation anomaly in both the instrumental and the historical periods indicate strongly that the historical records can indeed provide a valuable proxy of the patterns of summer precipitation variation over long periods of time. One advantage of such long time series is that periodic or quasi-periodic

II DOCUMENTARY DATA Ratio of Droughts to Floods in Northern, Central, and Southern China

Northern China Central China Southern China

Century

D

F

R

Remarks

D

F

R

Remarks

D

F

R

Remarks

B.C.

2

-

11

0.0

Wet?

-

2

0.0

Wet?

-

2

0.0

Wet?

1

-

12

0.0

Wet?

-

1

0.0

Wet?

-

-

-

-

A.D.

1

5

6

0.83

Dry

-

-

-

-

-

1

0.0

Wet?

2

12

17

0.71

Dry

1

3

0.33

Wet

-

1

0.0

Wet?

3

9

27

0.33

Wet

-

27

0.0

Wet?

1

4

0.25

Wet

4

4

6

0.67

-

4

17

0.24

Wet

-

6

0.0

Wet?

5

7

20

0.35

Wet

8

39

0.21

Wet

-

-

-

-

6

13

27

0.48

Wet

4

14

0.29

Wet

-

-

-

-

7

26

57

0.46

Wet

11

26

0.42

Wet

1

3

0.33

Wet

8

15

67

0.22

Wet

5

21

0.24

Wet

2

6

0.33

Wet

9

22

56

0.39

Wet

27

60

0.45

Wet

9

12

0.75

Dry

10

74

98

0.76

Dry

20

46

0.44

Wet

8

16

0.50

Wet

11

56

118

0.47

Wet

23

50

0.46

Wet

2

15

0.13

Wet

12

37

34

1.09

Dry

103

131

0.79

Dry

36

48

0.75

Dry

13

60

83

0.72

Dry

55

112

0.49

Wet

25

33

0.76

Dry

14

110

199

0.55

Wet

65

108

0.60

Dry

32

50

0.64

-

15

40

75

0.53

Wet

42

115

0.37

Wet

33

46

0.72

Dry

16

52

49

1.06

Dry

60

104

0.58

-

66

104

0.63

-

17

77

107

0.72

Dry

96

145

0.66

Dry

55

81

0.68

Dry

18

116

131

0.89

Dry

107

164

0.65

Dry

42

72

0.58

-

19

100

89

1.12

Dry

85

114

0.75

Dry

6

16

0.37

Wet

Totals

835

1289

0.65

716

1299

0.55

318

516

0.62

D = number of droughts. F = number of floods. R = D/F.

variations in climate may be observed, which could not be resolved in the shorter instrumental records. For example, Wang et al. (1981) used the long-term precipitation anomaly data for different latitude zones in China to construct a time-space diagram that points to a recurrent pattern of precipitation anomaly beginning in northern China and migrating southward, with a period of ~80 yr. More detailed studies of records from different areas have used spectral analysis to isolate statistically significant periodicities in the data. Thus, in the precipitation anomaly series for the Shanghai region a periodicity of 36.7 yr is apparent (Wang and Zhao, 1979). This periodicity is also seen in other records from southwestern China and the east-

40

FIGURE I 1.6 Eigenvectors of drought and flood data using historical records for A.D. 1470-1977 (left) and instrumental!/ recorded precipitation data from 1951-1974 (right).The first three eigenvectors of historical data (HI to H3) account for 15, 11, and 7% of variance in the data respectively. The first three eigenvectors of instrumental data (I I to 13) account for 18, 13, and I I % of variance in the data, respectively. Eigenvectors HI and II are similar; so are H2 and 13 and H3 and 12 (inverse patterns).This indicates that historical data can be reliable indicators of climatic anomaly patterns (Wang and Zhao, 1981).

ern part of the Yangtze River Basin. Elsewhere, other periodic variations have occurred, most notably a quasi-biennial oscillation (2-2.5 yr) in the Yangtze River area and north of the Yellow River (Wang and Zhao, 1981). Both the short-term and long-term periodicities appear to be related to large, synoptic-scale pressure anomalies over eastern Asia and adjacent equatorial regions.

Apart from floods and droughts, the parameteorological phenomenon that seems to have attracted most attention in historical records is the freezing of lakes and rivers. The longest continuous series is that of Lake Suwa (near Kyoto) in Japan. Data on the time of freezing of this small (-15 km2) lake are available almost annually from 1444 to the present, although the dates are not very reliable from 1680 to 1740 (Arakawa, 1954, 1957). Gray (1974) calibrated this record with instrumental data from Tokyo for the period 1876-1953 and found the best correlation with mean December to February temperature data. Using the regression equation relating Lake Suwa freezing dates to temperatures since 1876, Gray was able to reconstruct Tokyo midwinter temperatures back to 1450. The coldest periods appear to have been -1450-1500 and -1600-1700, when winter temperatures were about 0.5 °C below the mean of the last 100 yr.

Long historical records of river and lake freezings are also available from China. Using local gazetteers and diaries, Zhang and Gong (1979) compiled records of the frequency of freezings of lakes in the mid and lower reaches of the Yangtze River, freezings of rivers and wells in the lower Yellow River Basin, the occurrence of sea ice in the Gulf of Chihli and Jiangsu Province (31-41° N), and snowfall in tropical areas of southern China. Using all this information they calculated the number of exceptionally cold winters per decade from 1500 to 1978 (Fig. 11.7). The highest frequency of cold winters occurred during the periods 1500-1550, 1601-1720, and 1830-1900, with the decade 1711-1720 being the most severely cold period in the last 480 yr. By mapping the areas most affected by severe winters it was also possible to recognize two main patterns of the anomaly — periods when cold winters predominated east of -115° E and periods when areas to the west were colder. From modern meteorological studies it appears that such large-scale anomaly patterns result from changes in the position of the upper level trough over East Asia. When the trough is in a more westerly position and fairly deep, cold air sweeps down more frequently over the area west of 115° E. Westerly flow prevails when the trough is weaker and less extensive, leading to milder conditions in the west, with cold air outbreaks more common to the east. Generally speaking, the colder periods shown in Fig. 11.7 had more frequent cold outbreaks west of 115° E, indicating that such periods were characterized by a strongly developed upper air trough over eastern Asia. Conversely, the warmer

1st cold period i

2nd cold period n

3rd cold period m

1501-10 1601-10 1701-10 1801-10 1901-10

FIGURE I 1.7 Number of cold winters per decade in central and southern China from 1501 —1510 to 1971-1980. Major cold intervals are indicated (Zhang and Gong, 1979).

periods were times of stronger westerly flow in winter and weaker upper level trough development (Zhang and Gong, 1979).

Historically, one of the most important effects of severe winter temperatures was on water transportation systems, and many records exist regarding the disruptions caused by canals and rivers freezing over for prolonged periods. In the Netherlands, for example, canals were built in the early seventeenth century to connect major cities, and records of transportation on the canals, including times of freezing over, have been kept since 1633 (de Vries, 1977). Using instrumental winter temperature data from De Bilt (Labrijn, 1945) the number of days on which the Haarlem-Leiden canal was frozen each winter was calibrated (see Fig. 11.2a), enabling De Bilt winter temperatures to be reconstructed back to 1657. Further temperature estimates, back to 1634, were possible by calibration of the canal freezing data with barge trip frequency between Haarlem and Amsterdam (1634-1682) a service that was commonly suspended due to ice cover on the canal (van den Dool et al., 1978). In this way, a complete winter temperature reconstruction for De Bilt has been obtained back to 1634 (Fig. 11.2b). Note however that this record conveys no information about how warm winters may have been in years when the record shows only that the canal was never frozen over (Fig. 11.2a).

In more northern latitudes, rivers freeze over every year, and in historical time the dates of freeze-up and break-up were both economically and psychologically important. Consequently, diaries and journals from these regions commonly contain frequent reference to the state of icing on nearby rivers and estuaries. Ironically, remote regions of northern Canada are relatively well-endowed with historical records, thanks to the efforts of Hudson Bay Company managers at various company posts around Hudson Bay and points to the west (Ball, 1992). A valuable analysis of such data from western Hudson Bay has been made by Catchpole et al. (1976). Using content analysis they analyzed journals kept by Hudson Bay Company trading post managers from the early eighteenth to the late nineteenth century. Although reference to the state of ice on nearby rivers and estuaries was often imprecise, content analysis enabled quite reliable estimates to be made of the dates of freeze-up and break-up (Fig. 11.8) and these provide a unique index of overall "winter duration" in this remote region (Moody and Catchpole, 1975). The prolonged period of both early freeze-up and late breakup in the early part of the nineteenth century is particularly noteworthy. Comparisons with modern data are difficult because the sites are no longer inhabited, but where comparisons can be made it appears that the "freeze season" (the time between freeze-up and break-up) averaged 2-3 weeks longer during the eighteenth and nineteenth centuries than in recent years (Table 11.4). Recently, the mid-eighteenth century record of first freeze-up dates has been used to calibrate white spruce tree-ring records from the area, enabling a 300-yr record of first freeze-up dates to be reconstructed (Jacoby and Ulan, 1982). Although only a limited amount of modern data was available for verification, the results were reasonably good, suggesting that some confidence can be placed in the long-term reconstruction. This is an interesting example of how one proxy data set may be used to calibrate (or verify) another.

TABLE 11.4 Comparison of Historical (H) and Modern (M) Dates of Freeze-up and Break-up (in Days after December 31)

First partial freezing (H) First complete freezing (H) First breaking (H) or or first permanent ice (M) or complete freezing (M) first deterioration of ice (M)

TABLE 11.4 Comparison of Historical (H) and Modern (M) Dates of Freeze-up and Break-up (in Days after December 31)

First partial freezing (H) First complete freezing (H) First breaking (H) or or first permanent ice (M) or complete freezing (M) first deterioration of ice (M)

Site

Earliest

Mean

Latest

Earliest

Mean

Latest

Earliest

Mean

Latest

Churchill River at Churchill (M)

273

291

318

288

319

336

141

160

169

Fort Prince of Wales

273

292

319

295

321

345

150

168

187

Hudson Bay at Churchill (M)

292

305

319

313

340

124

159

180

Moose Factory (H)

281

304

335

290

319

341

105

126

145

Moose River

304

316

331

217

330

347

103

116

Catchpoleetal. (1976)

at Moosonee(M)

Catchpoleetal. (1976)

11.2.3 Phenological and Biological Records

In this section consideration will be given to purely phenological data, that is, data on the timing of recurrent biological phenomena (such as the blossoming and leafing of plants, crop maturation, animal migrations, etc.), as well as historical observations on the former distribution of particular climate-sensitive plant species. The value of phenological records as a proxy of climate is illustrated in Fig. 11.9. From 1923 to 1953, the flowering dates of 51 different species of plants in a Bluffton, Indiana, garden were noted. For each species, the average date of flowering was computed, and individual years expressed as a departure from the 30-yr mean (Lindsey and Newman, 1956). Yearly departure values for all species were then averaged to give an overall departure index for the 51 species; in Fig. 11.9, this is plotted against the mean temperature of the period March 1 to May 16 (i.e., the start of the growth period). Clearly, the phenological data are an excellent index of spring temperatures, cool periods corresponding closely to late flowering dates and vice versa. This example illustrates well the potential paleoclimatic value of phenological observations; if they can be calibrated, they may provide an excellent proxy record of past climatic variation.

One of the longest and best known phenological records comes, like so many other long historical records, from the Far East. At Kyoto (the capital of Japan until 1869) the Governor or Emperor used to hold a party under the flowering cherry blossoms of his estate, when they were in full bloom (Arakawa, 1956b, 1957). The blooming dates can be considered as an index of spring warmth (February and March) as shown by Sekiguti (1969) using modern phenological records and instrumental data. Higher spring temperatures result in earlier blooming dates; according to Kawamura (1992) an increase in March temperature of 1 °C changes the mean date of cherry blossom flowering by 2-3 days. The Kyoto record is extremely sparse, but nevertheless is of interest as it spans such a long period of time (Table 11.5). It appears from this record that the eleventh to fourteenth centuries were relatively

3154

Fori Albany

3154

Fori Albany

Moose Factory

Moose Factory

Fort Albany

1151 1710

Fort Albany

jj

IM .*

J

r\

1

(T

i

V

30 50 70 90 1B00 10 30 50 70

30 50 70 90 1B00 10 30 50 70

Moose Factory

i> 5-

A A / \

505

V

30 50 70 90 1SOO 10 30 50 70

1710

30 50 70 90 1SOO 10 30 50 70

FIGURE I 1.8 Seven-year running means of dates of first partial freeze-up (above) and first break-up (below) of ice in estuaries at the locations indicated (all on west coast of Hudson Bay, Canada). Data obtained by content analysis of historical sources. See also Table I 1.4. Comparable dates for modern conditions shown as horizontal lines. Dates are given in days after December 31 (Catchpole et al., 1976).

-1—I—i—i—I—i—i—I—I—i—I—I—I—I—I—I—I—I—I—i—I—I—I—I—I—I—i—I—r

0 0

Post a comment