Coherence of Climatic Reconstruction from Historical Documents in China by Different Studies

Home , Dongzhi (solar term), Lunar calendar

INTERNATIONAL JOURNAL OF CLIMATOLOGY Int. J. Climatol. (2007) Published online in Wiley InterScience (www.interscience.wiley.com) DOI: 10.1002/joc.1552

Coherence of climatic reconstruction from historical documents in China by different studies

Quansheng Ge,a* Jingyun Zheng,a Yanyu Tian,a Wenxiang Wu,a Xiuqi Fanga,b and Wei-Chyung Wangc a Institute of Geographic Sciences and Natural Resources Research, Chinese Academy of Sciences, Beijing, China 100101 b School of Geography, Beijing Normal University, Beijing, China 100875 c Atmospheric Sciences Research Center, State University of New York, Albany, New York, USA 12203

ABSTRACT: Much effort has been spent in the last few decades to reconstruct the climate over China using a variety of historical documents. However, differences in the results of reconstructions exist even when people are using similar documents. In order to address this issue, 14 published temperature series by different studies were analyzed for coherence and mutual consistency. The analyses on their temporal fluctuations indicate that for the individual time series (standardized) on the 10-years time scales, 57 of the 91 correlation coefficients reach the significance level of 99%. The spatial patterns among the different time series also show high coherency. In addition, consistency also exhibit when comparing the reconstructions with other available natural climate change indicators. Above information was subsequently used to synthesize the temperature series for the last 500 and 1000 years. Copyright  2007 Royal Meteorological Society

KEY WORDS coherence; climatic reconstruction; historical documents; China Received 30 May 2006; Revised 11 March 2007; Accepted 25 March 2007

1. Introduction Ge et al. (2003), however, shows that the warmth dur- China has abundant continuous historical documents with ing the Medieval Warm Period is more evident than a large number of records describing the weather and that in the 20th century. Therefore, it is natural that climate information (Wang, 1979; Zhang and Gong, some doubt be raised on the objectivity and quality of 1980; Wang and Zhang, 1988; Zhang, 1991; Bradley reconstruction, even the reliability of the historical doc- et al., 1993; Zhang, 2004; Ge et al., 2005b). As proxy uments as data source. What is the main factor that data with high temporal and spatial resolution, these results in the discord between different reconstructed records have been used to reconstruct climate time series series? What kind of temporal resolution would be suit- since the 1970’s (Chu, 1973; Zheng and Feng, 1986; able while using these reconstructed series with a lack Zhang and Crowley, 1989; Gong and Hameed, 1991; of ‘complete’ historical information? Is there signifi- Wang, 1991; Wang and Zhang, 1992; Zhang, 1994; Zhang cant difference between the climate series reconstructed et al., 1997; Riches et al., 2000; Ge et al., 2003), and from the natural proxy data and the historical docu- have provided valuable datasets for studying past climatic ments? change (Bradley and Jones, 1995; Pfister and Wanner, 2002). This paper attempts to answer the above questions However, compared with instrumental records and nat- through comparative analysis on climatic series recon- ural evidence, the meaning of historical documents is structed with historical documents as well as natural evi- not always so definite (Pfister and Wanner, 2002). Some dence, and furthermore, to illustrate the quality of climate discrepancies have been found between different recon- reconstruction using Chinese historical documents. The structed climatic series derived from Chinese historical study has taken following steps. Firstly, the relationship documents. For example, the warming intensity in the between the different climatic series reconstructed by dif- Medieval Warm Period in China is so moderate when ferent studies using Chinese historical documents is anal- compared with the warmth in the 20th century accord- ysed. Secondly, the coherence between document-based ing to some proxy temperature reconstructed by Chu series and natural-evidence-based series is examined. (1973); Wang and Gong (2000a). The reconstructed tem- Based on the discussion on the validation of histori- perature by Zhang (1993); Man and Zhang (1993) and cal documents, the synthesized temperature series are achieved. * Correspondence to: Quansheng Ge, Institute of Geographic Sciences and Natural Resources Research, Chinese Academy of Sciences, 11A To provide a general idea on the work basis of Datun Road, Beijing 100101 China. E-mail: [email protected] the reconstructed series discussed in this article, we

Copyright  2007 Royal Meteorological Society Q. GE ET AL. have annexed an introduction on the climatic informa- 110–120 °E) during 1500’s-1990’s, and the 30-year tion recorded in Chinese historical documents. More- mean winter-half-year temperature series for the past over, because some proper nouns may cause confusion 2000 years. Most of these above series focus on last to readers who are unfamiliar with ancient Chinese lit- 500 years because the climatic information in historical erature, and some words may need a clear deﬁnition, documents is more abundant in this period (see Appendix a glossary is annexed as well to help the understand- I for detail). In our study, 14 document-based temperature ing. Both appendices are attached at the end of this series over 500 years are selected and analysed (Table I). paper. These series all have 10-year-resolution, and cover large area of Eastern China (Figure 1). Besides the half-millennial series above, some scien- 2. Research Materials tists have reconstructed the millennial-scale temperature series of China, based on either historical records, or In past decades, Chinese researchers have reconstructed natural proxy data (Table II). Wang and Gong (2000a) many temperature series from Chinese historical docu- collected documentary records on abnormal meteorolog- ments. For example, based on the records of frost, heavy ical and hydrological phenomena, from which the cold snow, river and lake freezing, excessive rain, cold disas- index can be deﬁned, and reconstructed the tempera- ter, etc, Wang and Wang (1990) reconstructed the winter ture series for every 50-year in Eastern China during temperature of the East China (25–35 °N, 115–120 °E) A.D. 800–2000 (WG00). With similar method used in during 1470’s–1970’s. In this study, the description of 500-year temperature series reconstruction (Table I), Ge cold/warm in historical documents was ranked and trans- et al. (2003) reconstructed the winter-half-year tempera- formed into corresponding anomaly temperature series, ture series for every 30-year in Eastern China during A.D. by calibration with modern data. Similar method was 1–2000 (Ge03). Tan et al. (2003) analysed the stalag- used by Wang et al. (1998) to reconstruct the decadal mite near Beijing, and reconstructed the annual summer- temperature in the other area of Eastern China for the past temperature series during A.D.1–1985 (Tan03). Liu et al. 500 years. Zhang (1980); Zheng and Zheng (1993) calcu- (2004) reconstructed the tree ring width chronology in the lated the frequency of cold/warm year, and reconstructed Qilian Mountain in Qinghai Province (near Tibet), and the decadal winter temperature in different regions dur- estimated the temperature variation of the region during ing 1470’s–1970’s. This method is also used by Shen A.D. 1000–2000 (Liu04). Yang et al. (2003) analysed the and Chen (1993) to reconstruct relatively longer (1100’s- ice core samples from Qinghai-Tibet plateau, and recon- 1970’s) temperature series in the drainage area of the structed the relative temperature series for every 50 years Taihu Lake. Based on a collection of the phenologi- during A.D. 1–2000 (Yang03). Yang et al. (2002) syn- cal records on plant, hydrometeorology and agriculture, thesized several reconstructed series based on tree ring, Ge et al. (2003) estimated the difference between ancient ice core, lake sediments, and historical documents, and and modern phenology, and reconstructed the decadal developed relative decadal temperature series for China winter-half-year (October–April, similar herein after) during A.D. 1–2000 (Yang02). These six series are also temperature series for Eastern China (25–40°N, compared in this study.

Figure 1. Locations and area of 14 temperature series from Chinese historical documents by different studies.

Copyright  2007 Royal Meteorological Society Int. J. Climatol. (2007) DOI: 10.1002/joc COHERENCE OF CHINESE DOCUMENTARY CLIMATIC RECONSTRUCTION series of the Abbreviation ZSDW ZMYW WNCY WECY ., 1998 WFTY ., 1998 WSCY ., 1998 WCCY ., 1998 ., 1998 year and the Authors et al et al et al et al et al publishing Shen and Chen, 1993 STHW Zhang, 1980 ZXGW Zhang, 1980 ZGGW Wang Zheng and Zheng, 1993 Zhang, 1980 Zhang, 1980 ZLYW Wang method Reconstruction Rank the description of temperature variations, and transform the rank into corresponding anomaly temperature series by calibration with modern data Calculated the frequency of cold/warm years in every 10then years, reconstruct the series with the regression equation of the records Locality Taihu Lake Hengyang, Fuling Zhangzhou, Nanning, Haikou Beijing, Tianjin, Hebei, Shanxi, northwest of Shandong, north of Henan Anhui, Jiangsu, Zhejiang, Jiangxi, Shanghai Shandong Hankou, Xiangyang, Changsha Dongtai, Anqing, Hefei records Historical Frost, heavy snow, (rivers and lakes) freezing, over rain, cold disaster, etc. As above Frost, heavy snow, (rivers and lakes) freezing Frost, heavy snow, (rivers and lakes) freezing, over rain, cold disaster, etc. Temporal resolution 10 years 10 years 10 years 10 years Table I. Brief information of 14 reconstructed series at 10-year resolution. series Length of 1100’s–1970’s 10 years As above Drainage area of the 1470’s–1960’s 10 years As above Ji’an, Ganzhou, 1470’s–1960’s 10 years As above Fuzhou, Yong’an Zhang, 1980 ZMOW 1470’s–1960’s 10 years As above Changzhou, 1470’s–1960’s 10 years As above Shanghai, Nanjing, 1380’s–1990’s 1380’s–1990’s 1470’s–1980’s 1470’s–1960’s 1500’s–1990’s 10 years As above Fujian, Taiwan Wang 1500’s–1990’s 10 years As above Guangdong Wang 1470’s–1990’s 10 years As above Hunan, Hubei Wang indexes Winter temperature Winter temperature temperature Winter temperature Winter temperature Annual temperature Annual temperature Winter temperature Winter temperature Annual temperature temperature temperature Drainage area of the Taihu Lake Hunan and Jiangxi Provinces Fujian Province Winter Guangdong and Guangxi Provinces Lower reaches of Yangtze Region Reconstructed North China East China Shandong Province Middle reaches of Yangtze Fujian and Taiwan Provinces South China Annual Central China Annual

Copyright  2007 Royal Meteorological Society Int. J. Climatol. (2007) DOI: 10.1002/joc Q. GE ET AL. series of the Abbreviation WECW year ., 2003 GECW and the Authors publishing et al Ge Wang and Wang, 1990 method Reconstruction between ancient phenology and modern one, then reconstruct the temperature series basedonthe relationship between modern phenology and temperature temperature, and transform the rank into corresponding anomaly temperature series by calibration with modern data E Compare the difference E Rank the description of ° ° of the records Locality N, 110–120 N, 115–120 ° ° ). 25–40 25–35 Continued Table I. ( records Historical hydrometeorology and agricultural phenology (rivers and lakes) freezing, cold disaster, etc. Temporal resolution series Length of 1500’s–1990’s 10 years Plant, 1470’s–1970’s 10 years Frost, heavy snow, indexes year temperature temperature Region Reconstructed Eastern China Winter-half- East China Winter

Copyright  2007 Royal Meteorological Society Int. J. Climatol. (2007) DOI: 10.1002/joc COHERENCE OF CHINESE DOCUMENTARY CLIMATIC RECONSTRUCTION

Table II. Six millennial-scale temperature series from historical records and natural evidence in China.

Region Recon- Recon- Temporal Evidence and Reconstruction Authors Abbreviation (Site) structed struction resolution records method & published indexes period (year) year (AD)

Eastern China Temperature 1–2000 30 Records in historical Compare the difference Ge et al., 2003 Ge03 of winter- documents about plants, between the present and half-year hydrology, meteorology ancient times, and and agricultural reconstruct temperature phonological series based on the phenomenon, etc. relationship of modern climate and phenology Eastern China Annual 800–2000 50 Records of abnormal Deﬁne cold index Wang and Gong, WG00 temperature meteorological according to warm/cold 2000a hydrological phenomena description, then build such as frost, heavy the regressive equation snow, (river and lake) of cold index and annual freeze, cold hazards, etc. temperature China Relative 1–2000 10 Width of tree ring, δ18O Calculate the mean Yang et al., 2002 Yang02 temperature of ice core, δ18Oof standardized value of peat, TOC & C/N of original series, with a lake sediments, weight based on area temperature reconstructed from historical documents Beijing Temperature 1–1985 1 Thickness of annual Reconstruct by Tan et al., 2003 Tan03 between July laminae in stalagmite regressive relationship and August between the thickness of annual stalagmite level and summer temperature Qilian Mt., Temperature 1000–2000 1 Width of tree ring Estimate temperature Liu et al., 2004 Liu04 Qinghai from Dec. to variation indicated by Province Apr. and from tree ring width. No June to Sep. reconstruction Qinghai-Tibet Relative 1–2000 50 δ18O of ice core, density Calculate the mean Yang et al., 2003 Yang03 plateau temperature of palynomorph, content standardized value of of CH4, width of tree original series, with a ring, stable carbon weight based on area isotope, TOC in lacustrine sediment, water temperature, palynological assemblage, magnetic parameters, pigment, δ18O of peat, ﬂuctuation of glacier in high mountains

For the comparison of these series, each series is 3. Comparison between document-based series standardized firstly by following equation: The relationship among the 14 document-based series is checked with correlation analysis and cluster analysis. − = xi x According to Table III, most (57 of 91) of the corre- xi (1) σ lation coefficients reach the significance level of 99%. Furthermore, the coefficient values are usually attributed where xi denotes the standardized value of the series, to the adjacency or overlapping of the regions. The closer Xi is the original value in the series, xandσ is the regions are, or the more overlapping the two regions the mean value and standard deviation of the common have, the higher the correlation coefficient is. According reference period of the series. For the 14 document- to the location or area, the 14 document-based series can based temperature series, the common reference period be grouped into six regions and 10 pairs. For 10 pairs of is 1500’s–1940’s. For the six millennial-scale series, the reconstructed temperature series, nearly all (9 of 10) the common reference period is A.D. 1000–1950. correlation coefficients (the bold text in Table III) reach

Table III. Correlation coefﬁcients between the reconstructed series with a 10-year resolution.

Series WNCY ZSDW WECY STHW ZLYW ZXGW WCCY ZMYW WFTY ZMOW WSCY ZGGY WECW Abbr.

ZSDW 0.519∗∗∗ WECY 0.536∗∗∗ 0.406∗∗ ∗ ∗ ∗∗∗ STHW 0.287√ 0.356 0.661 ∗ ∗∗∗ ∗∗∗ ZLYW 0.221 0.315 0.518 0.722 √ √ ZXGW 0.146 0.288∗ 0.373∗∗ 0.240 0.260 WCMY 0.606∗∗∗ 0.411∗∗ 0.746∗∗∗ 0.343∗ 0.490∗∗∗ 0.330∗ ZMYW 0.356∗∗ 0.448∗∗ 0.466∗∗∗ 0.325∗ 0.486∗∗∗ 0.510∗∗∗ 0.653∗∗∗ WFTY 0.461∗∗∗ 0.415∗∗ 0.514∗∗∗ 0.295∗ 0.038 0.099 0.386∗∗ 0.157 ∗∗ ∗∗ ∗∗∗ ∗∗ ∗ ∗∗ ∗∗∗ ∗∗ ∗∗ ZMOW 0.407 0.422√ 0.531 0.371 0.288 0.380 0.458 0.418 0.428 ∗∗ ∗∗∗ ∗ ∗∗∗ ∗∗∗ ∗∗ ∗ ∗ WSCY 0.443 0.238√ 0.471 √ 0.171√ 0.295√ 0.508 0.495 0.423 0.291√ 0.343 ∗∗ ∗ ∗∗ ∗∗∗ ZGGW 0.180√ 0.267 0.255 0.215 0.261 0.367 0.322 0.407 0.208√ 0.199 0.729 WECW 0.255 0.426∗∗ 0.740∗∗∗ 0.666∗∗∗ 0.638∗∗∗ 0.496∗∗∗ 0.518∗∗∗ 0.480∗∗∗ 0.269 0.460∗∗∗ 0.382∗∗ 0.387∗∗ GECW 0.459∗∗∗ 0.541∗∗∗ 0.546∗∗∗ 0.409∗∗ 0.418∗∗ 0.367∗∗ 0.516∗∗∗ 0.627∗∗∗ 0.452∗∗ 0.295∗ 0.334∗ 0.285∗ 0.463∗∗∗

∗∗∗ ∗∗ ∗ √Notes: Marks on the right top of correlation coefficients represent levels of significance, and is for 99.9%, is for 99%, is for 95%, and is for 90%. The boldfaced one means that two series come from same region. the significance level of 99%, and most (7 of 10) correla- This result indicates that the reconstructed series from tion coefficients reached the significance level of 99.9%. the Chinese historical documents could reflect tempera- Moreover, the clustering spectrum system of the series ture variations of 30-year scale better. Therefore, 30-year in Figure 2 indicates that the series for the same or over- temporal resolution might be reasonable for temperature lapping regions will cluster earliest, and series for closer change studies by using Chinese historical documentary regions will cluster relatively earlier. Therefore, it is rea- data. sonable to conclude that the coherence of temperature series in different regions actually depends on regional climate similarity, and regional climatic difference should 4. Comparison between millennium series derived be responsible for the discrepancy between reconstructed from historical documents and natural evidence climate series. To further explore the discrepancy of the paired series, To further check the validity of the document-based the wavelets for each paired series are compared as well. series, the authors also compared the document-based The results indicate that the similarity for each paired series and the natural-evidence-based series. Figure 4 series exists mainly on the correspondence of lower (left) shows the six temperature series over 1000 years frequency signals, especially on the scale over 30 years, in China derived from the historical documents and i.e., the fluctuations on multi-decadal to centennial scale natural evidence respectively. To check the coherence show a greater similarity, while more discrepancies will between the two types of series, they are divided into be detected on the scale of shorter than 30 years. Figure 3 ‘documentary group’ and ‘natural group’. Taking a 30- shows an example of the comparison of the paired series year resolution, most correlation coefficients between in North China Plain. the ‘documentary’ series and ‘natural’ series reach the significance level of 90% (Table IV). ‘Ge03’ and ‘WG 00’ derived from historical documents show higher correlation with ‘Yang 02’ from synthesized natural and documentary evidence, and show lower correlation with ‘Tan03’ and ‘Liu04’ derived from single natural evidence.

Table IV. Correlation coefﬁcients between the reconstructed temperature series from historical records and natural evidence.

Series Resolution Yang02 Tan03 Liu04 Yang03 Abbr. (year) √ ∗∗∗ Ge03 30 0.471 0.216 0.203 – √ ∗∗∗ 50 0.507 0.218√ 0.146√ 0.293 WG00 50 0.700∗∗∗ 0.330 0.432 0.441∗

Figure 2. The clustering spectrum system of temperature series in Notes: marks on the right top of the coefﬁcients represent√ credibility different regions. level, and ∗∗∗ is for 99.9‰, ∗∗ is for 99‰, ∗ is for 95‰, is for 90‰.

Copyright  2007 Royal Meteorological Society Int. J. Climatol. (2007) DOI: 10.1002/joc COHERENCE OF CHINESE DOCUMENTARY CLIMATIC RECONSTRUCTION

Figure 3. Comparison of the wavelet transformation results of the paired series of WNCY (a) and ZSDW (b) in North China Plain.

Table V. Correlation coefficients between the two groups wavelet transformation for each series are shown in (documentary and natural). Figure 4 (right). Despite of the amplitude differences between different series in individual time intervals, the Series Coefficients Coefficients 50–100 years cold/warm stages agree with one other. resolution for the first for the second On century scale, the cold stages revealed from these eigenvector eigenvector series agree well with the cold events in historical docu- 30 a 0.738 0.422 ments, as well as the natural evidence for glaciers and 50 a 0.763 0.497 periglacier activities (Fang et al., 2004). On the scale 100 a 0.913 0.588 of 100–300 years, the cold-warm fluctuations in these series are almost simultaneous except for ‘Tan03’. On the scale of 300–500 years, the cold-warm fluctuations The canonical correlation analysis is also used to in these series are coherent with each other except for examine the coherence between ‘documentary group’ ‘Yang03’ and ‘Liu04’. On the period of 500–800 years, (‘Ge 03’ and ‘WG00’) and ‘natural group’ (‘Yang02’, there are four warm and three cold episodes during the ‘Tan03’, ‘Liu 04’ and ‘Yang03’), under trials of three past 2000 years, except that the fluctuation in ‘Yang different temporal resolutions: 30, 50, and 100-year. 03’ is relatively advanced than the others. According to High correlation is detected between the two groups, above comparison, it’s concluded that the past 2000-year and the correlation coefficient grows while taking lower temperature variations in China derived from historical temporal resolution (Table V). With 30-year resolution, documents and natural evidence agree with each other the canonical correlation coefficient between the first approximately. maximal eigenvectors of two groups’ series is 0.738, reaching a significance level of 99.9%. With the 50- year resolution, the coefficient is 0.763, reaching a 5. Synthesization of different temperature series significance level of 97.5%. With the 100-year resolution, Since the document-based series, as well as the series the coefficient is 0.913, reaching a significance level of based on historical documents and natural evidence are 85.0%. coherent to each other in general, it will be meaningful to Besides correlation analysis, Morlet wavelet transfor- synthesize current results so that a common conception mation is used as well to reveal their coherence vari- on the historical temperature variation can be achieved. ation on different time-scales. The real parts of Morlet For the document-based series, the 14 series and their

Figure 4. Six reconstructed temperature series from historical documents and natural evidence (left) and the wavelet of these series (right). (a) Temperatures of winter-half-year in the East of China (Ge et al., 2003); (b) Annual temperatures in the East of China (Wang and Gong, 2000a); (c) Average temperatures in China (Yang et al., 2002); (d) Temperatures from July to August in Beijing, China (Tan et al., 2003); (e) Width indexes of tree ring in Qilian Mt., Qinghai, China (Liu et al., 2004); (f) Average temperature in Qinghai-Tibet plateau, China (Yang et al., 2003). weighted average are plotted to describe the general it is difficult to estimate the errors/uncertainty for this temperature trend for Eastern China in last 500 years synthesized series. However the 14 standardized values (Figure 5). All the series are taken 30-year mean values, in different series for the same time are regarded as 14 and the series in the same regions are taken arithmetic “samples”, so that the standard deviation value for the means to form different regional series (north China, period can be calculated to indicate the internal coherence east China, central China, south China, southeast China, of the synthesized series. In the synthesized series, it is Taiwan Island, and southwest China). Taking reference easy to identify some periods with special significance to the study by Wang and Gong (2000b), the series of such as the Little Ice Age. these regions are given different weights (0.247 for north To describe the millennial-scale temperature trend in China, 0.126 for east China, 0.212 for central China, China, the six standardized millennium series mentioned 0.174 for southeast China, 0.032 for Taiwan Island, and in Section 4 as well as their area-weighted average value 0.209 for southwest China) and taken the area-weighted are plotted in Figure 6, with a unified temporal resolution average, so that they form a synthesized series of Eastern of 50 years. Taking reference to the weights value used China. As most of the studies cited in this article did by Wang and Gong (2000b), the weights are assigned not provide the error estimation or uncertainty analysis, to each series according to their coverage area (taking

Copyright  2007 Royal Meteorological Society Int. J. Climatol. (2007) DOI: 10.1002/joc COHERENCE OF CHINESE DOCUMENTARY CLIMATIC RECONSTRUCTION

Figure 5. Temperature trend for Eastern China in last 500 years by 14 document-based series. This ﬁgure is available in colour online at www.interscience.wiley.com/ijoc

Figure 6. Millennial-scale temperature trend for China by document-based and natural-evidence-based series. This figure is available in colour online at www.interscience.wiley.com/ijoc whole China as 1.00): Ge03, 0.232, for north China, the Medieval Warm Period and Little Ice Age in China east China, central China and southwest China; WG00, can be identified. But the amplitude of temperature 0.471, for whole China except Xinjiang, Northwest China fluctuation is smaller than the one in Yang02. and Tibet; Tan04, 0.084, for north China; Yang03, 0.182, for Tibet; Liu04, 0.198, for northwest China. However, Yang02, the series reconstructed by Yang et al. (2002), 6. Conclusion and discussions have an extraordinary large coverage (whole China) that It has been pointed out that Chinese historical docu- might ‘hijack’ the synthesized series. Therefore Yang02 is ments are valuable data source for reconstruction of plotted separately from the synthesized series constructed past climatic changes. This study reaffirms the valid- by the other five series, which covers most area (over ity of the climatic series reconstructed with these his- 85%) of China except Xinjiang. In the synthesized series, torical documents, according to the comparison among

Copyright  2007 Royal Meteorological Society Int. J. Climatol. (2007) DOI: 10.1002/joc Q. GE ET AL. document-based climate series, and the comparison will be smaller, compared with the series based on objec- between ‘documentary’ and ‘natural’ records. The con- tive records such as phonology, in which the relationship clusion can be further summarized as followed: between temperature and original records can be directly expressed with the regression equations. (1) There are significant correlations between the 14 temperature series derived from Chinese historical documents by different studies in Eastern China for last Acknowledgements 500 years. The coefficients are close related to the This research was jointly funded by NSFC (Grant No. adjacency or overlapping of the regions. The coher- 40625002, 40571007, 40331013), Chinese Academy of ence of climatic series in different region actually Sciences (KZCX2-YW-315), and Ministry of Science and depends on regional climate similarity, and regional Technology, P. R. China (2005DFA20010). climatic difference should be responsible for the discrepancy between reconstructed climate series. Com- parison on wavelet suggests that 30-year tempo- Appendix AI. Introduction of climatic information ral resolution might be reasonable for temperature recorded in the Chinese historical documents. change study by using Chinese documentary data. AI.1. Types of historical documents as data source (2) It is found that there are approximately simultaneous temperature fluctuations between most of Climate information can be mined from four types of the millennial-scale series derived from historical Chinese historical documents: Chinese classical docu- ∗ documents and the ones from natural evidence in ments, local gazettes, archives (Asterisked words have China, especially under a lower temporal resolu- more detailed explanation in appendix II Glossary, simi- ∗ tion. Cold/warm stages of 50–100 years can be well larly hereinafter.) of the Qing Dynasty and the Republic ∗ revealed by both ‘documentary’ and ‘natural’ series. of China , and private diaries. The climate information derived from documentary (1) Chinese classical documents, called Jing, Shi, Zi, ∗ ∗ records and natural evidence is coherent approxi- Ji in Chinese. According to Si-Ku-Quan-Shu (The mately. Complete Collection in the Four Branches of Literature) published in 1787 A.D., there are 44 categories∗ in Jing, Despite of the coherence listed above, it is also Shi, Zi, Ji ∗ (Editor Committee of Indexes of Catalogue for noticed that the temporal resolution will significantly the Imperial Complete Collection in the Four Branches affect the discrepancy between different series. Though of Literature, 2002). Among them, 28 categories (1531 some document-based series can reach a 10-year temporal books, 32251 volumes∗ in total) have contained cli- resolution, their ability in revealing the decadal temper- matic information such as changes in temperature, cloud ature features may be unsatisfactory. As the number of cover, precipitation situation, etc. (Table VI). From these literature records is limited, the data abundance is sig- books, 22567 items∗ recorded during 30 B.C.–1470 A.D. nificantly affected by the interval length. When a higher (Figure 7), which provide us climatic information with temporal resolution, such as 10-year, is used, the data definite times and localities, have been extracted (Zhang, amount will decrease. Thus, the reconstruction results by 1996). In addition, some other books such as Ming-Shi- different researchers could differ due to the inconsistency Lu∗ (the Veritable Records of the Ming Dynasty)and between small-sample-size evidence. When a lower tem- Qing-Shi-Lu∗ (the Veritable Records of the Qing Dynasty) poral resolution is used, higher data abundance will be compiled during the Ming∗ and Qing∗ Dynastic peri- achieved with the larger sample size, and the ‘real’ tem- ods respectively, reported important political and social perature features are more likely to be revealed, even affairs, natural disasters, and abnormal climatic events. with different evidence by different researchers. There- Qing-Shi-Gao∗ (the Manuscript of History of the Qing fore, reconstruction results by different authors would Dynasty) compiled during the period of the Republic have better consistency. of China∗ contained much information about climate The discrepancy might also result from different changes as well. methodology, especially different indicators for recon- (2) Local gazettes. Local gazettes are those compre- struction. For example, the maximal amplitude of tem- hensive books that recorded both the nature and soci- perature change in WG00 is significantly smaller than ety conditions in a given district (County, Prefecture or that in the other millennial-scale series. That may be Province). The climatic information involved is mainly explained by several reasons. First, the magnitude of about drought/flood, frost/snow, severe cold, phenology, winter-temperature variation is larger than annual tem- agricultural conditions, plagues and locusts, harvest, and perature. Second, 50-year variation magnitude would be changes in the river systems, etc. The times and local- smaller than that of 30-year variation. Third, the extreme ities of climatic events were clearly recorded, and their events and subjective records used to reconstruct WG00 impacts were described in detail (Figure 8). The local need to be quantified into the cold indices before temper- gazettes first appeared approximately around the time ature reconstruction. Because of the variance loss during of the Zhou∗ and Qin∗ Dynasties (from 11th century this data conversion process, the percentage of variance B.C.–to 206 B.C.), formed the stylistic rules in the Song∗ explanation as well as variation in the reconstructed series Dynasty (from 960 A.D.–to 1279 A.D.), and flourished in

Copyright  2007 Royal Meteorological Society Int. J. Climatol. (2007) DOI: 10.1002/joc COHERENCE OF CHINESE DOCUMENTARY CLIMATIC RECONSTRUCTION

Table VI. Records of climatic information in Chinese ancient documents of Jing, Shi, Zi, Ji.

Document Categories of books Categories of books Number of volumes Group containing records of consulted consulted climatic information

Jing (the Classics) Shi-Jing (The Book of Songs) 1– The Rites of the Zhou Dynasty 1– The Book of Rites 1– The Rites of Dadai 1– Biography Edited by Zuo in the 1– Spring-Autumn Period Biography Edited by Gongyang in the 1– Spring-Autumn Period Biography Edited by Guliang in the 1– Spring-Autumn Period Zi (the Schools) The various schools of thought and 14 157 their exponents during the period from pre-Qin times to the early years of the Han Dynasty The Confucian 5 47 The books on the art of war 1 40 On agriculture 53 107 On Medicine 3 43 Arithmetic 2 131 Miscellaneous 140 870 Allusion 59 4344 Story 173 941 Shi (the History) Ofﬁcial history 23 2 650 Non-ofﬁcial history 48 645 Chronicle 35 1 413 Events Memorial 3 410 Miscellaneous history 155 233 Record 45 392 Biography 19 34 Governmental document 43 2 408 Titbits 13 16 Geography 153 1 230 Ji (the Collections) Other collections 527 13 213 General collections 10 2 927 In total 1 531 32 251

Figure 7. The number of climatic information records in Chinese ancient documents for every decade from 30 B.C. to 1470 A.D. the Ming∗ and Qing∗ dynasties (from 1368 A. D.–to 1911 (110 000 volumes∗ in total) regarding the period from the A.D.), when they were edited and revised almost every Northern Song Dynasty∗ (from 960 A.D. to 1127 A.D.) 30–60 years. According to the statistics in The United to 1949 A.D. have been preserved. Among them, 973 Catalogue of China’s Local Gazettes, 8264 local gazettes gazettes belong to the Ming∗ Dynasty (from 1368 A.D. to

Figure 8. Example of climatic information recorded in local gazettes (Quoted from Gazettes of Yangzhou Prefecture published in 1874 A.D.) The two pages list the disasters and abnormal events in the region from 1842 A.D. to 1874 A.D. (from right to left), dated with Chinese lunar calendar∗. The numbers in brackets are added by the author of this article. The seasons here (in Chinese lunar calendar) are slightly different from modern meteorological conception, see ‘Chinese lunar calendar’ in annexed Glossary for details. (1) Twenty-eighth year (of Daoguang, 1848 A.D.), 6th (lunar) month, strong wind and heavy rain, the Yangtze River overflowed; 7th month, strong wind and thunder storm, field and houses submerged. (2) Eighth month (of the same year), strong wind and heavy rain, the Yangtze River, the Huaihe River, the (Gaoyou) Lake, and the sea abnormally rose at the same time, citation from Gazettes of Yizheng County. (3) Twenty-ninth year (of Daoguang, 1849 A.D.), autumn, flood, the Yangtze River and the (Gaoyou) Lake overflowed at the same time, below are newly collected (which suggests that above records are quoted from earlier local gazettes). (4) First year of Xianfeng (1851 A.D.), Jiaoxiechang (the name of a salt field) of Dongtai County, the tide overflowed and broke the Sea Wall of Fan. (5) Sixth year (of Xianfeng, 1856 A.D.), 5th–8th month, severe drought, the Grand Canal dried up. (6) Twelth year (of Tongzhi) (1873 A.D.), autumn, flood. Daoguang, Xianfeng and Tongzhi here are titles of the three emperor’s reign, similarly hereinafter..

1644 A.D.), 5685 ones belong to the Qing∗ Dynasty (from (The Everyday Records of Emperor Life), Shang-Yu- 1644 A.D. to 1911 A.D.), when almost every county has Dang∗ (The Archives of Imperial Decrees), Qing-Yu-Lu∗ ∗ its own local records, and 1255 local gazettes belong to (Records of Sunny or Rainy Days), Yu-Xue-Fen-Cun the Republic of China∗ (from 1911 A.D.–to 1949 A.D.) (The Depths of Rainfall infiltration and Snowfall)are (Beijing Astronomical Observatory CAS, 1985). We esti- about emperors’ decrees, lives, and some important mate that there may be more than 200 000 items∗ that government affairs. These archives contain abundant could provide climatic information with accurate times climatic, environmental, and agricultural information. For and localities in China’s local gazettes during the past example, some documents have recorded the influences 1000 years. of the Tambora Volcano eruption on China in 1815 (3) Archives of the Qing∗ Dynasty and the Republic (Zhang et al., 1992). ∗ ∗ Among the archives, some are special reports on of China . There are about 10 million files of Qing weather or climate, such as Qing-Yu-Lu∗ and Yu-Xue- Dynasty archives in the Chinese First Historical Archive Fen-Cun∗. Qing-Yu-Lu∗ provides daily weather records in the Beijing Palace Museum. Among them, 0.6 million on sky conditions, wind direction, precipitation type files are Zou-Zhe∗ (Memos) with written comments from ∗ (clear, light rain/snow, rain/snow, etc.) and the duration emperors, over 0.7 million Lu-Fu-Zou-Zhe (Duplicates of precipitation events. For Beijing, the records are from of Memos) are the approved Memos, 1.4 million files are 1724 A.D. to 1903 A.D., lasting 174 years with only 6 Tiben and Zouben (Routine Memos and Private Memos), missing years. These records are consistent with instru- 0.4 million files belong to Zong-Ren-Fu∗ (Royal Family ∗ mental meteorological records beginning in 1841 A.D. Affair Office), 2.2 million files belong to Nei-Wu-Fu in Beijing (Institute of Meteorological Sciences, Central (Palace Internal Affair Office), 1.5 million files belong Meteorological Bureau, 1975). Besides Beijing, Nan- to the six governmental ministries (administrations on jing, Suzhou and Hangzhou also observed and reported official personnel affairs, population and finance, rites and the daily weather conditions in Qing-Yu-Lu∗ with the cultural affairs, defense, criminal law, construction and same criterion, with time spans of 1723–1798 A.D., engineering), 2 million compiled files, such as Qi-Ju-Lu∗ 1736–1806 A.D. and 1723–1773 A.D. respectively.

Copyright  2007 Royal Meteorological Society Int. J. Climatol. (2007) DOI: 10.1002/joc COHERENCE OF CHINESE DOCUMENTARY CLIMATIC RECONSTRUCTION

Yu-Xue-Fen-Cun∗ (Figure 9) recorded the depth of warm in winter, no freezing.’ quoted from Wu-Xing-Zhi ∗ ∗ infiltration into the soil for each rainfall event and the (the Chronology of the Five Elements)inSong-Shi (the depth of each snowfall in 273 prefectures of 18 provinces History of the Song Dynasty). ‘In the 23rd year of Wanli (1595 A.D., Wanli is the title from 1693 A.D. to 1911 A.D., with the same criterion all ∗ over the country. The measured value was recorded in of an emperor’s reign in the Ming Dynasty), Wujiang ∗ ∗ (in current Jiangsu Province), on 1st day of the fourth terms of the Chinese unit of Fen (≈3.2mm)andCun ∗ ≈ month (Chinese lunar calendar , similarly hereinafter): ( 3.2 cm) and was submitted directly to the emperors by sunny, slight cloudy; 2nd, sunny; ...; 8th, cloudy and the local officials (Ge et al., 2005a). rain; ...; 18th, sunny and slight cold; ...;7thdayof (4) Private diaries. Up to now, about 200 private the fifth month, light rain; ...; 27th of the sixth month, diaries containing daily weather records or weather- sunny and hot; ...; 2nd of the seventh month, sunny in related natural phenomenon have been found. The Diary the morning and heavy lightning storm in the afternoon; of Gengzi-Xinchou (1180 A.D.–1181 A.D.) by Lu¨ Zuqian ...’, quoted from The Diary of Kuaixuetang written by (1137–1181) is one of the earliest among them. These Feng Mengzhen in the Ming∗ Dynasty. diaries had detailed descriptions on the time, locality and the climate events, which could be used to reveal (2) Weather-dependent natural disasters. Such informa- historical climate changes (Gong et al., 1984; Hameed tion is about drought and flood, severe heat or cold, and Gong, 1994) (Table VII). snow or hail disasters. These records were common in the official histories and local gazettes. For example: AI.2. Types of climate information derived from historical records ‘In 3rd year of Jianlong (962 A.D., Jianlong is the ∗ There are four kinds of weather and climate related title of an emperor’s reign in the Song Dynasty), ° ° information identified from historical records. the Capital (Kaifeng, 114 E, 35 N) had droughts in spring and summer. Hebei region encountered severe (1) Direct weather/climate information. This type of drought. The seedlings of crops died away in Bazhou Prefecture. Besides, there are no rains from spring to information is about changes of temperature, cloud summer in many prefectures such as Henan, Hezhong, cover, type of precipitation, etc., with detailed time, Mengzhou, Zezhou, Puzhou, Yunzhou, Qizhou, Jizhou, locality and severity of the events. The typical ∗ ∗ Huazhou, Yanzhou, Xizhou, Suzhou, etc.’ quoted from records are in Yu-Xue-Fen-Cun , Qing-Yu-Lu ,and Wu-Xing-Zhi ∗ (the Chronology of the Five Elements)in private diaries. Some records embedded in local Song-Shi (the History of the Song Dynasty). gazettes and official histories provide such informa- ‘In 1st (lunar) month of the 13th year of Shunzhi (1656 tion as well. For example: A.D., Shunzhi is the title of an emperor’s reign in the Qing∗ Dynasty), severe frost struck Qionghai of Hainan ‘In the 2nd year of Dazhong Xiangfu (1009 A. D., (110 °E, 19°N), resulting in a famine. People were hungry Dazhong Xiangfu is the title of an emperor’s reign in and frozen to death. Many animals, livestock, fish and the Song∗ Dynasty), the Capital (Kaifeng, 114 °E, 35°N), birds died. Betel palms and coconut palms withered,

Figure 9. The example of memorial to the emperors in Yu-Xue-Fen-Cun This Yu–Fen–Cun (infiltration depth of rainfall into the ground) example is dated the 20th day, the fifth month, in the eighth year of the Qianlong Reign, with the Chinese lunar calendar (11 July 1743). The two pages are the first and last page (right to left) of the original 12-page memo prepared by Gao Bin, Governor of Zhili Province (a province around Beijing). The thin handwritten words are Gregorian calendar converted from Chinese lunar calendar by researchers. Texts bracketed on the right are rainfall records listed by Fu∗ (prefecture) in the Zhili Province (from right to left): On 2nd day of the fifth month (June 23rd), Yongping Fu: Linyu county, 3 cuns; on 3rd day of the fifth month (June 24th), Jizhou Fu: Xinhe county, 3 cuns’. Red text on the left is a comment by the emperor: I see, the capital Beijing was not in severe drought, but I still expect the rainfall. The figure was published on Bulletin of the American Meteorological Society (Ge et al., 2005a). This figure is available in colour online at www.interscience.wiley.com/ijoc

Table VII. The catalogue of some historical Chinese diaries containing weather and phenology information since 1500 A. D.

Source Author Time of records Locality of records

The Diary of Huaifeng Lu Shen 1512, 1529, and 1535 Beijing, and North of Jiangsu Province The Diary of Kuaixuetang Feng Mengzhen 1587–1605 Hangzhou The Diary of Weishuixuan Li Rihua 1609–1616 Hangzhou The Diary of Yuan Xiaoxiu Yuan Zhongdao 1614 Gong’an, in Hubei Province The Diary of Jiaxing Ye Shaoyuan 1628–1634 South of Jiangsu Province The Diary of Mr. Qi Zhongmin Qi Jiabiao 1631–1645 Hangzhou The Diary of Beiyou Tan Qian 1653–1656 Beijing The Diary of Xunhua Unknown 1660 Suzhou The Reserved Diary of SunZhengjun Sun Qifeng 1660–1666 Xinxiang and Anyang, in Henan Province The Diary of Xunletang Dou Keqin 1675–1689 Beijing The Diary of Nanqian Unknown 1676 Hangzhou The Diary of Zhuting Qian Daxin 1778 Hangzhou The Diary of Mr. Wu Tuchuang Wu Qian 1780–1808 Hangzhou The Compose of Mr. Wu Tuchuang Wu Qian 1790–1800 Hangzhou The Diary of Youzhengweizhai Wu Xilin 1793–1797 Beijng The Diary of Huanjing Wu Xilin 1794–1796 Beijing The Diary of Lin Zexu Lin Zexu 1816–1842 Beijing, Suzhou, ect. The Diary of Zhizhizhai Wong Xincun 1825–1863 Beijing, ect. The Diary of Zeng Wenzheng Zeng Guofan 1841–1872 Beijing The Diary of Yuemantang Li Ciming 1854–1889 Beijing, Hangzhou The Diary of Outang Zhou Xingyu 1856 Beijing The Diary of Zhixiang Guan Tingfen 1860–1864 Vicinity of Hangzhou The Diary of Mr. Wong Wengong Wong Tonghe 1858–1904 Beijing, ect. The Diary of Xiangqilou Wang Heyun 1869–1916 Changsha, Xiangtan and Hengyang, in Hunan Province The Diary of Fan Zanchen Fan Dixiang 1866–1933 Wuhan, Beijing The Diary of Yuandulu Ye Changchi 1889 Beijing The Diary of Dongling Unknown 1890 Beijing The Diary of Gao Jijian Gao Dan 1901–1902 Beijing The Diary of Tianyinlu Unknown 1909 Beijing The Diary of Ren Guozhai Unknown 1910–1911 Beijing The Diary of Binglu Unknown 1927–1928 Beijing The Phenology Zhu Kezhen 1921–1931 Nanjing and Beijing The Diary of Zijiangjingshe Wu Chengti 1932–1944 Beijng The Diary of Zhu Kezhen Zhu Kezhen 1936–1949 Nanjing, Hangzhou, etc.

trees and grasses died away’, quoted from the Gazette Prefecture, Chizhou Prefecture and Taizhou Prefecture, of Lehui County published during Kangxi’s reign (1662 farmers plant early, mid, and late paddy, or only plant A.D.–1722 A.D.) of the Qing∗ Dynasty (The seasons and early and late paddy. To the north of Yangtze River, e.g. months above are deﬁned with Chinese lunar calendar, Fengzhou, Yingzhou, farmers plant early, mid, and late see “Chinese lunar calendar” in annexed Glossary for paddy. Sorghum, legume and coarse cereals are planted details). as well based on the soil conditions.’ quoted from the Memo to the emperor by Tuo Yong, the Anhui governor, (3) Effect and impact of climate events. This type of on 2 September 1762 A.D. information includes climate events’ political, social ‘(In Jiangsu) now the paddy is just sowed in summer, and and economical impacts, especially for agriculture, the wheat is sowed late autumn’, ‘Farmers in Suzhou district only plant one crop of wheat and one crop of such as the distribution and boundary of Mandarin paddy in a year. After the wheat has been harvested, Orange, tea, bamboo, pomegranate, winter wheat, the land is ploughed, the rice seedlings are transplanted double-cropping paddy, farming methods, and the in summer, grow in autumn, and are harvested in impacts on some climate-sensitive animals or plants. winter.’ quoted from the Comments by the governor of For example: Suzhou Prefecture in 1829 A.D. (The seasons here are deﬁned with Chinese lunar calendar∗, see ‘Chinese lunar ‘The time of rice seedlings in Anqing is the earliest calendar’ in annexed Glossary for details). in Anhui Province, where double-cropping paddy can be planted. In the other regions to the south of the (4) Phenological and biological records. The records pro- Yangtze River, such as Huizhou Prefecture, Ningzhou viding such information mention the time of recurrent

Copyright  2007 Royal Meteorological Society Int. J. Climatol. (2007) DOI: 10.1002/joc COHERENCE OF CHINESE DOCUMENTARY CLIMATIC RECONSTRUCTION

biological phenomena, e.g. the blossoming and leaf- be inferred quantitatively based on historical plant zones ing of plants, crop maturation, animal migration, as (Zhu and Chu, 1973; Ge et al., 2003). well as the time of recurrent meteorological or hydro- The subjective records are mainly the records on logical phenomena, e.g. the first frost, the last frost, human feelings related to climate changes, especially to and lake freezing. For example: those abnormal climatic events, as well as the descriptions on various kinds of climate disasters, climate events ‘(Hangzhou, 120 °E, 30°N, 1603 A.D.) 11th day of and their impacts. These descriptions like ‘severe cold’, second month (23rd March), the landscape is as beautiful ‘bitter cold’, ‘extremely hot’, ‘surging floods’, etc., are as a painting, with some peach blooming a little by often embodied in historical documents, and often mixed chance... 12th (24th March), sunny, northwest wind, with records of abnormal hydrological and meteorologi- three tenths of the peaches on the new bank around cal phenomena. It is difficult to merely use these records the Solitary Hill Island are blooming, while on the to reconstruct climate series directly, due to the lack of old bank only a few peaches bloom. 13th (25 March), a comparable measurement. However, for some climatic sunny, windy, another four tenths peaches on the new bank blooming... 16th (28 March), peaches around the factors, the mean value and the probability of abnormal new bank and the Broken Bridge blooming brilliantly’, events are close related, e.g. the temperature versus the quoted from Kuaixuetang Diary by Feng Mengzhen frequency of warm/cold events, and the precipitation ver- (1548–1605). sus the frequency of drought/flood events. Therefore, it ‘The winter in the 11th year of Shunzhi (1654 A.D.) was is possible to use some methods to reconstruct the rela- severely cold, and the thickness of ice in Taihu Lake tive series, so that the climate trends in a given region (120 °E, 31°N) was two Chis∗ (≈64 cm). Such weather during a given period can be presented. For instance, the lasted 20 days, over half of Mandarin Orange and pomelo grading series of drought and flood can be reconstructed died.’ quoted from Disasters, volume 40, the Gazette of with numerous drought/flood records in Chinese histor- Wujiang County published in Qianlong’s reign (1736 ical documents (Academy of Meteorological Science of A.D.–1795 A.D.). Chinese Central Meteorological Administration, 1981). “Zhangzhou Prefecture (118 °E, 25°N) is known for Relative annual temperature series can be reconstructed its fertile soil and warm climate. Both winter wheat and spring wheat are sowed before Dongzhi∗ (Winter as well with the records of cold disasters, such as frost Solstice). Early harvest is prior to Qingming∗ (Fresh and abnormal snowfalls (Zhang and Gong, 1979; Zhang, Green) (Dongzhi and Qingming are Chinese solar terms, 1980; Zheng, 1982). These records can also provide cir- see Glossary in appendix for detail), late harvest is after cumstantial evidence for the changes of other climatic Qingming, and the general harvest time is within the factors. middle ten days of the 3rd month (see ‘Chinese lunar calendar’ in Glossary for details)... Thericeseedlings are transplanted soon after the wheat harvest...’quoted Appendix AII. Glossary. from the Memo to the emperor by Wu Jinsheng, Garrison Commander of Zhangzhou Prefecture, on 5 April, 1736. Underlined words in context are the entries in this glossary. In summary, the types of information described above Archives: can be grouped into ‘objective’ records, which can be compared directly among the different sources, and Official documents or reports, especially for the govern- the ‘subjective’ records, which are difficult to compare ment. quantitatively. Category: The objective records include Qing-Yu-Lu, Yu-Xue- Fen-Cun and the records of natural phenology. They can The books distinguished according to their content, be used directly to reconstruct the precipitation or tem- such as books on the art of war, books on medicine, perature variations (Chu, 1973; Wang et al., 1992; Zhou chronicles, etc. et al., 1994; Wang and Zhang, 1995; Zhang and Liu, 2002; Ge et al., 2003, 2005a; Zheng et al., 2004, 2005). Chi: The distribution and boundaries of Mandarin Orange, tea, bamboo, pomegranate, winter wheat, double-cropping A traditional Chinese length unit, approximately 32 cm. paddy, as well as farming methods, are limited by the length of the growing season and the minimum tempera- Chinese lunar calendar: ture. Therefore, the distribution is usually coherent with certain climate zones. Once the temperature varies, the The lunar calendar traditionally used in China, which has 29 or 30 days in one month, and 12 months in distribution of these plants will change accordingly, and one year (354 days). Seven leap years (with one more the adaptation is usually in accordance with the average month) are set in 19 years to adapt the solar year. This climate conditions in several decades (Man and Zhang, calendar set 24 solar terms corresponding to fixed dates 1990). Therefore, with the analog relationship between in solar calendar, which determine the month, and the modern climate and modern plant zones, historical mean “Winter Solstice” is always in the 11th month. The date temperature in several decades for a certain period could in Chinese lunar calendar is usually (but varies) about

a month earlier than Gregorian calendar, e.g. 15 April Zi (the Schools) means the various schools of thoughts in Gregorian calendar is usually in 3rd or 2nd month and their annotation since the Spring-Autumn Period of the lunar calendar. The seasons in this calendar are and Warring states Period (770 B.C. to 221 B.C.), such also different from modern meteorological conception. as Xun-Zi (the thoughts of Xun Kuang, an ideologist The “spring” refers to the time from 1st month to 3rd of Confucianism, Zi is a term of respect), Lao-Zi (the month, the ‘summer’ is from 4th month to 6th month, thoughts of Li Er, who is called Lao-zi and brought out the ‘autumn’ is from 7th month to 9th month, and the the basis of Taoism philosophy), Zhuang-Zi (the thoughts ‘winter’ is from 10th month to 12th month. of Zhuang Zhou, another important ideologist of Taoism), Lun-Heng (On Measurement, by Wang Chong, a materi- Cun: alist), and so on. Ji (the Collections) are ancient Chinese books of literary art, such as poetry, prose, Ci (Lyric, a A traditional Chinese length unit, approximately 3.2 cm. Chinese poetic genre), Fu (a Chinese descriptive prose interspersed with verse). They are divided into Zong- Dongzhi (Winter Solstice): ji (General Collections), Xuan-Ji (Selected Collections) and Bie-Ji (Other Collections), such as Yue-Fu-Shi-Ji (the One of the 24 solar terms in Chinese lunar calendar, Poetry Collections of Music Ofﬁce), Gu-Shi-Ji (the Col- usually on 22 December or 23rd. lections of Ancient Poetry), Quan-Tang-Shi (the Poetry of the Tang Dynasty), and Gu-Wen-Guan-Zhi (a selected Fen: collection of ancient Chinese prose), etc.

A traditional Chinese length unit, 1/10 of Cun, approxi- Lu-Fu-Zou-Zhe (Duplicates of Memos): mately 0.32 cm. The duplicates of Zou-Zhe commented by the emperor, Fu: copied by the Military Department before they were sent to the ofﬁcials to be carried out. Ancient Chinese administration unit, prefecture. A Fu usually consists of several counties. Ming dynasty:

Han Dynasty: The second last dynasty in China from 1368 A. D. to 1644 A.D., replaced by the Qing Dynasty. A dynasty in China from 206 B.C. to 220 A.D., which brought China international reputation as a super power Ming-Shi-Lu (the Veritable Records of the Ming for the ﬁrst time. Dynasty):

Item: Compiled daily records of the emperor’s orders and activities in the Ming Dynasty. The records were sorted A record. Here it refers to a sentence or a paragraph in the by time. context that is able to provide some climatic information independently. Nei-Wu-Fu (internal affair ofﬁce): Jing, Shi, Zi, Ji (the Classics, the Histories, the Schools, A department managing internal affairs in the Palace the Collections). during the Qing Dynasty (1644–1912 A.D.).

The joint name of Chinese classical ancient literature. Northern Song Dynasty: Jing (the Classics) are known as the doctrines about politics, religion, cardinal guides, constant virtues, ethics, See Song Dynasty. and codes of ethics in ancient societies of China, dominated by Confucian books, such as Si-Shu-Wu-Jing Qi-Ju-Lu (The Records of Emperor Life in Everyday): (the Four Books and the Five Classics), Liu-Shu (the Six Classics), Shi-San-Jing (the Thirteen Books), and Bai- The records about the emperor’s everyday life. In this Hu-Tong (Comprehensive Discussions in the White Tiger article, they are mainly in the Qing Dynasty. Hall), etc. Shi (the Histories) consist of Er-Shi-Si-Shi (24 official histories for the dynasties in China), such as Shi-Ji (the Qin Dynasty: History), Han-Shu (the History of the Han Dynasty), Zi-Zhi-Tong-Jian (A General History Helpful to Gover- Also Ch’in Dynasty, the first centralized dynasty in nors), as well as other kinds of historic books. The China, from 221 B.C. to 206 B.C. chronicle of Er-Shi-Si-Shi begins from the legendary “Emperor Huang” (2550 B.C.), and lasts to the end of the Qing Dynasty: Ming Dynasty (1644 A.D.). The 24 official histories have 3213 volumes with about 40 million Chinese characters, The last dynasty in China founded by Manchurian, from which were written in the same style of chronological 1644 A.D. to 1912 A.D., replaced by the Republic of biography. China.

Copyright  2007 Royal Meteorological Society Int. J. Climatol. (2007) DOI: 10.1002/joc COHERENCE OF CHINESE DOCUMENTARY CLIMATIC RECONSTRUCTION

Qing-Shi-Gao (the Manuscript of History of the Qing territory of the empire, which is called the Southern Song Dynasty): Dynasty (From 1127 A.D. to 1279 A.D.). This empire was finally conquered by Mongols in 1279 A.D. A history on the Qing Dynasty written during early time of the Republic of China. This book is only a Song-Shi (the History of the Song Dynasty): manuscript and not listed into the classical Er-Shi-Si-Shi (24 histories,seeShi in Jing, Shi, Zi, Ji). the History of the Song Dynasty (960–1279 A.D.). One of the Er-Shi-Si-Shi (see Shi in Jing, Shi, Zi, Ji). Qing-Shi-Lu (the Veritable Records of the Qing Dynasty): Volume : Compiled daily records of the emperor’s orders and activities in the Qing Dynasty. The records were sorted One of the books in literature works. by time. Wu-Xing-Zhi (the Chronology of the Five Elements): Qing-Yu-Lu (Records of Sunny or Rainy Days): A routine part of Chinese ancient histories, which con- Daily weather record in the Qing Dynasty, which pro- cerns much about disasters and abnormal events, such as vides detailed information on sky conditions, wind direc- solar eclipse. According to ancient theory, these events tion, precipitation type (rain or snow, light, heavy, torren- are determined by the balance and circulation of five tial, etc.) and the duration of precipitation events. Many basic elements (Metal, Wood, Water, Fire, and Soil). of these records are already lost. But in four places, the records are left and found. In Beijing, the records are Yu-Xue-Fen-Cun (the Depths of Rainfall infiltration and from 1724 A.D. to 1903 A.D. In Nanjing (capital of Snowfall): Jiangsu Province), they are from 1723 A.D. to 1798 A.D. In Suzhou (in Jiangsu Province), they are from 1736 A.D. Precipitation records in the Qing Dynasty, which covered to 1806 A.D. In Hangzhou (capital of Zhejiang Province), 273 prefectures in 18 provinces from 1693 A.D. to they are from 1723 A.D. to 1773 A.D. 1911 A.D. They can be divided into Yu-Fen-Cun (Depth of Rainfall Infiltration, Yu means rainfall) and Xue- Qingming (fresh green): Fen-Cun (Depth of Snowfall, Xue means snow). The measurements were recorded in terms of the Chinese unit One of the 24 solar terms in Chinese lunar calendar, of Fen (≈3.2mm)andCun (≈3.2 cm), and reported by usually on Apr. 4th, 5th or 6th. local officials to the emperors.

Republic of China: Zhou Dynasty:

The first republic founded in China. This republic was Also Chow Dynasty or Chou Dynasty, a feudalistic set after the Qing Dynasty was overthrown in 1912 A.D., dynasty traditionally dated 11th century B.C. – to 221 and was replaced by current The People’s Republic of B.C., replaced by the Qin Dynasty. China in 1949 A.D. Zong-Ren-Fu (Royal Family Office): Shang-Yu-Dang (the Archives of Imperial Decrees): A department serving the royal family during the The archives about the emperor’s decrees in the Qing Qing Dynasty (1644–1912 A.D.). Dynasty. Zou-Zhe (Memos): Si-Ku-Quan-Shu (The Complete Collection in the Four Branches of Literature): The memorial to the throne, usually are the confidential reports written by senior officials to the emperor. A huge collection of all kinds of literatures compiled at the time of Qianlong Emperor (1736–1795 A.D.) in the References Qing Dynasty. This collection embodied or listed most literatures available at the time. Si-Ku, which means the Academy of Meteorological Science of Chinese Central Meteorological Administration. 1981. Yearly Charts of Dryness/Wetness in China for four Branches, represents Jing, Shi, Zi, and Ji. the Last 500-year Period (in Chinese). SinoMaps Press: Beijing. Beijing Astronomical Observatory CAS. 1985. Unified Catalogue of Song dynasty: Local Gazettes in China (Zhongguo di fang zhi lian he mu lu) (in Chinese). China Publishing House: Beijing. Bradley RS, Jones PD. 1995. Climate since A. D. 1500, introduction. A dynasty from 960 A.D. to 1279 A.D, which is In Climate Since A. D. 1500, Bradley RS, Jones PD (eds). Routledge: divided into two stages. The earlier dynasty called the London, New York; 1–16. Northern Song Dynasty (From 960 A.D.to 1127 A.D.) Bradley RS, Shieu D, Wang WC. 1993. High-resolution paleoclimate had set the capital in northern China, and was destroyed records from monsoon Asia. Eos, Transactions-American Geophysi- cal Union (EOS) 74: 601, 603–604. by Nuzhen¨ (Nuchen, or Jurchens) people, the ancestors Chu KC. 1973. A preliminary study on the climatic fluctuations during of Manchurians, in 1127 A.D. A prince of the Dynasty the last 5000 years in China. Scientia Sinica (Series A) 16(2): reorganized an imperial government in the southern 226–256.

Editor Committee of Indexes of Catalogue for the Imperial Complete based on the Clear and Rain Records. In Climate Since A. D. Collection in the Four Branches of Literature. 2002. Indexes of 1500, Bradley RS, Jones PD (eds). Routledge: London, New York; Catalogue for the Imperial Complete Collection in the Four Branches 194–209. of Literature (in Chinese). Shanghai Chinese Classics Publishing Wang SW, Gong DY. 2000a. Climate in China during the four special House: Shanghai. periods in Holocene. Progress in Natural Progress 10(5): 379–386. Fang XQ, Ge QS, Zheng JY. 2004. Cold events during Holocene and Wang SW, Gong DY. 2000b. Enhancement of the warming trend millennial climate rhythm (In Chinese). Progress in Natural Progress in China. Geophysical Research Letters 27(16): 2581–2584, DOI: 14(4): 456–461. 10.1029/1999GL010825, 2000. Ge QS, Zheng JY, Fang XQ, Man ZM, Zhang XQ, Zhang PY, Wang SW, Ye JL, Gong DY. 1998. Climate in China during the little Wang WC. 2003. Winter half-year temperature reconstruction for the ice age (in Chinese). Quaternary Sciences 1: 54–64. middle and lower reaches of the Yellow River and Yangtze River, Wang WC, Portman D, Gong GF, Zhang PY, Karl T. 1992. Beijing China, during the past 2000 years. Holocene 13(6): 933–940, DOI: summer temperatures since 1724. In Climate Since A. D. 1500, 10.1191/0959683603h1680rr. Bradley RS, Jones PD (eds). Routledge: London, New York; Ge QS, Zheng JY, Hao ZX, Zhang PY, Wang WC. 2005a. Reconstruc- 210–223. tion of historical climate in China, High-resolution precipitation data Yang B, Braeuning A, Shi YF. 2003. Late Holocene temperature from Qing Dynasty archives. Bulletin of the American Meteorologi- ﬂuctuations on the Tibetan Plateau. Quaternary Science Reviews cal Society 86(5): 671–679, DOI: 10.1175/BAMS-86-5-671. 22(21): 2335–2344, DOI: 10.1016/S0277-3791(03)00132-X. Ge QS, Zheng JY, Zhou AL, Zhu SY, Zhang PY, Zhang JR, Han YF. Yang B, Braeuning A, Johnson KR, Shi YF. 2002. General charac- 2005b. Agriculture and Environment-selected Memorials to Emperor teristics of temperature variation in China during the last two in the Qing Dynasty (in Chinese). The Commercial Press: Beijing. millennia. Geophysical Research Letters 29(9): 381–384, DOI: Gong GF, Hameed S. 1991. The variation of moisture conditions 10.1029/2001GL014485, 2002. in china during the last 2000 years. International Journal of Zhang DE. 1980. Winter temperature changes during the last 500 Years Climatology 11(3): 271–283. in South China. Chinese Science Bulletin 25(6): 497–500. Gong GF, Zhang PY, Zhang JR. 1984. The variation of phenodate in Zheng SZ. 1982. Climate in the little ice age and its effects in Beijing district. Chinese Science Bulletin 29(12): 1650–1652. Guangdong, China (in Chinese). Chinese Science Bulletin 27(5): Hameed S, Gong GF. 1994. Variation of spring climate in lower- 302–304. middle Yangtse River Valley and its relation with solar-cycle length. Zhang DE. 1991. Historical records of climate change in China. Geophysical Research Letters 21(24): 2693–2696. Quaternary Science Reviews 10(6): 551–554. Institute of Meteorological Sciences, Central Meteorological Bureau. Zhang DE. 1993. A Study on the medieval warm period in China (in 1975. Rainfall during the past 250 years in Beijing (in Chinese). Chinese). Quaternary Sciences 13(1): 7–15. Restricted paper, housed in the library of Institute of Geographic Zhang DE. 1994. Evidence for the existence of the Medieval Sciences and Natural Resources Research, Chinese Academy of Warm Period in China. Climatic Change 26(2–3): 289–297, DOI: Sciences, Beijing, China. 10.1007/BF01092419. Liu XH, Qin DH, Shao XM. 2004. Temperature variations recovered Zhang PY. 1996. Climate Change in China During Historical Times (in from tree-rings in the middle Qilian Mountain over the last Chinese). Shandong Science & Technology Press: Jinan. millennium (In Chinese). Science in China Series D-Earth Sciences Zhang DE. 2004. Total Collection of Meteorological Records of Three 34(1): 89–95. Thousand Years in China (in Chinese). Phoenix Publishing House Man ZM, Zhang XG. 1990. Shift of natural zones during the warm and Jiangsu Educational Publishing House: Nanjing. periods in the thirteenth century in eastern China (in Chinese). Fudan Zhang PY, Gong GF. 1979. Characteristics of climate variation in Journal (Social Sciences Edition) 5: 34–37. China since the sixteenth century (in Chinese). Acta Geographica Man ZM, Zhang XG. 1993. Research on the documental evidence and Sinica 34(3): 238–247. the characteristics of MWP in the east part of China (In Chinese). In Zhang PY, Gong GF. 1980. Ancient precipitation records in China. Research on the Past Life-supporting Environment Change of China, World Meteorological Organization Bulletin 29: 7–11. Lansheng Zhang (ed.). Ocean Press: Beijing; 95–103. Zheng SZ, Feng LW. 1986. Historical evidence of climatic instability Pﬁster C, Wanner H. 2002. Documentary data. Pages News 10(3): 2. above normal in cool periods in China. Scientia Sinica (Series B) Riches MR, Wang WC, Chen PQ, Tao SY, Zhou SG, Ding YH. 29(4): 107–114. 2000. Recent progress in the joint agreements on global and Zhang JZ, Crowley TJ. 1989. Historical climate records in China and regional climate change studies between the United States Reconstruction of Past Climates. Journal of Climate 2(8): 833–849, and the People’s Republic of China. Bulletin of the American DOI: 10.1175/1520-0442(1989)002<0833:HCRICA>2.0.CO;2. Meteorological Society 81(3): 491–499, DOI: 10.1175/1520- Zheng JY, Zheng SZ. 1993. An analysis on cold/warm and dry/wet 0477(2000)081<0491:RPITJA>2.3.CO;2. in Shandong Province during historical times (in Chinese). Acta Shen XY, Chen JQ. 1993. Grain production and climatic variation in Geographica Sinica 48(1): 348–357. Taihu Lake Basin. Chinese Geographical Science 3(2): 173–178. Zhang DE, Liu YW. 2002. A new approach to the reconstruction Tan M, Liu TS, Hou JZ, Qin XG, Zhang HC, Li TY. 2003. Cyclic of temporal rainfall sequences from 1724–1904 Qing-Dynasty rapid warming on centennial-scale revealed by a 2650-year weather records for Beijing (in Chinese). Quaternary Sciences 22(3): stalagmite record of warm season temperature. Geophysical Research 199–208. Letters 30: 1617, DOI: 10.1029/2003GL017352. Zhang PY, Wang WC, Hameed S. 1992. Anomalous weather in China Wang PK. 1979. Meteorological records from ancient chronicles of 1815–1817. In The Year Without a Summer? World Climate in 1816, China. Bulletin of the American Meteorological Society 60(4): Harington CR (ed.). National Museum of Natural Sciences: Ottawa; 313–318, DOI: 10.1175/1520-0477(1979)060<0313:MRFACO> 436–447. 2.0.CO;2. Zhang PY, Ge QS, Wang Z. 1997. Past climate in China, derived from Wang SW. 1991. Reconstruction of temperature series of North China ancient writings. In Space Science in China, Hu WR (ed.). Gordon from 1380s to 1980s. Science in China (Series B) 34(6): 751–759. and Breach Science Publisher: Amsterdam; 269–279. Wang PK, Zhang DE. 1988. An introduction to some historical Zheng JY, Hao ZX, Ge QS. 2004. Method for seasonal precipitation governmental weather records of China. Bulletin of the American derived from snow and rainfall archives in Qing Dynasty, a case Meteorological Society 69(7): 753–758, DOI: 10.1175/1520- study in Shijiazhuang (in Chinese). Progress in Natural Science 0477(1988)069<0753:AITSHG>2.0.CO;2. 14(8): 738–744. Wang RS, Wang SW. 1990. Reconstruction of winter temperature in Zheng JY, Man ZM, Fang XQ, Ge QS. 2005. Temperature variation eastern China during the past 500 years using historical documents in the eastern China during Wei, Jin and South-North Dynasties (in Chinese). Acta Meteorologica Sinica 48(2): 180–189. (220-580A.D.) (in Chinese). Quaternary Sciences 25(2): 129–140. Wang PK, Zhang DE. 1992. Recent studies of the reconstruction of Zhou QB, Zhang PY, Wang Z. 1994. Reconstruction of annual winter East Asian monsoon climate in the past using historical documents of mean temperature series in Hefei area during 1736–1991 A.D (in China. Journal of the Meteorological Society of Japan 70: 423–446. Chinese). Acta Geographica Sinica 49(4): 332–337. Wang PK, Zhang DE. 1995. Reconstruction of the 18th century summer precipitation of Nanjing, Suzhou, and Hangzhou, China,