PROGRESS IN NATURAL SCIENCE Vol .14 , No .8 , August 2004 Key points on temperature change of the past 2000 years in China＊ GE Quansheng1 , ZHENG Jingyun1 ＊＊ , MAN Zhimin2 , FANG Xiuqi3 , 1 and ZHANG Piyuan1 (1 .Institute of Geographic Sciences and Natu ral Resources Research , Chinese Academy of Sciences, Beijing 100101 , China;2 .Institute of Chinese History Geography , Fudan University , Shanghai 200433 , China;3 .S chool of Geography , Beijing Normal University , Beijing 100875 , China) Received December 11 , 2003 ;revised February 23 , 2004 Abstract Temperatu re change of the past 2000 years in C hina is discussed based on the w inter half-year temperature series of the past 2000 years in eastern China reconstructed recently , and other related studies.The main conclusions are as follow s:(1)The Little Ice Age (LIA)in China began in the early of the 14th centu ry (1320s)and ended in the beginning of the 20th cen tury (1910s), w hich was composed of four evident cold stages and three short w arming stages.The cold period in the Wei, Jin and S outh-North dynasties(210s～ 560s)w as the only one comparable w ith LIA for the past 2000 years.(2) The Medieval W arm Period (MW P)in China began in the 930s and ended in the 1310s, w hich was composed of tw o warm stages over 100 years and a cold stage less than 100 years.(3)T he cli- mate in the Sui and Tang dynasties should be divided in to two stages:the climate in the 570s～ 770s w as as warm as that in the 20th cen- tury ;w hile the temperature in the 780s～ 920s w as low er than that in the 1950s～ 1970s.(4)In eastern China as a w hole , w inter half- year temperature variation w ith over 1 ℃ occurred betw een the cold and w arm stages on centennial scale , w hile the changing rate exceeded 1 .0 ℃ per century .(5)There exists an about 1350-year periodicity in the historical temperature change.Inferred from the periodicity , the most likely historical analogue for the w arming in the 20th century is the warm stage of the S ui and T ang dynasties(570s～ 770s), in- stead of the Medieval Warm Period .(6)Although it was critically w arm , the temperature of the 20th century in eastern China is still w ithin the threshold of the variability of the last 2000 years. Keywords: eastern China , past 2000 years, temperature change. It is important to study the temperature change from the histo rical documents is one of the most im- during the past 2000 years for understanding the is- portant issues in historical climate study in China . sues such as the g reenhouse effect and global warming The historical documents related to tem perature induced by human activities .China has advantages in change can be divided into tw o catego ries according to reconstructing historical climate change for its abun- their characteristics .The first category is “ natural dant documented historical reco rds and other natural evidence” w hich provides information about tem pera- evidence obtained from tree ring s , lake sediments , ice ture directly .It includes phenology of plants (includ- cores , and stalagmite .Since Dr .Chu laid a founda- ing some crops), the date of the first and last frost or tion o n the study of temperature change in China fo r snow , and the duration of frost or snow ;the duration the past 5000 years[ 1] , sig nificant prog ress in the and dates of freeze-up and thaw of rivers , lakes and study of temperature change of the past 2000 years seas;the distribution and its northern boundary of has been made through the g reat efforts of genera- subtropical crops and eco nomic crops (e .g .citrus, tions .These achievements not only help us to under- tea , bamboo );the spatial-tempo ral distribution of stand the characteristics of temperature change in the farming and farming systems (e .g .sow ing date , har- past 2000 years , but also to assess some key scientific vest date , the distributio n of double harvest rice). points .In this paper , temperature change of the past The second category is “impact evidence” , specifically 2000 years in China w ill be discussed based on the the effect on “ man and society ” associated w ith the w inter half-year temperature series of the past 2000 cold/w arm events (e .g .the feeling of “rather cold” years in eastern China[ 2 , 3] , as well as o ther related and “w arm w inter”).The evidence can be used to studies . extract information about changes in tem perature through comparison w ith other evidence at different 1 Improvement in reconstructing methods times[ 2] .Because of the differences of the evidence , before reconstructing the temperature changes quanti- Reconstruction on temperature change derived ＊ Supported by the Chinese Academy of Sciences (Grant Nos.KZCX3-sw-321 and KZCX2-314) ＊＊ To w hom correspondence should be addressed .E-mail:zhengjy @igsnrr .ac.cn Progress in Natural Science Vol.14 No .8 2004 731 tatively , it is necessary to strictly distinguish the cat- original records[ 4～ 9] .The quantification w as also lim- egory of each record and to answer the follow ing three ited by the inherence shortage of the “impact evi- key questions :(1)How to quantify the documented dence” .Quantificational results at a high resolution records to temperature value ? Because the sources of may be gotten from the linear regression model , but historical records and reco rd formats are different the model w as limited by strict demands to the source from one another , it is very difficult to calibrate and data , so the length of reconstruction series could not [ 10～ 12] verify these records together and m ake them compati- exceed 300 years .Recently , tem perature series ble with instrumental temperature data .(2)How to w ith a certain temporal resolution (50 years)fo r the keep the temporal continuity ?For the uneven tempo- w hole China , eastern China and w estern China w ere ral and spatial distribution of these records , it is diffi- reconstructed respectively by integration of the pub- cult to obtain continual data even in the area with lished temperature series .The leng th of these series most abundant reco rds.(3)How to assess the spatial w as only 1200 years , limited by the available series [ 13] v alidity of the reco rd ?Because China is a large coun- and evidence (Fig .1) . try w ith various kinds of environment , how to assess In order to reconstruct the continuous tem pera- the spatial validity of the reco rd from the individual ture change series w ith a fixed spatial and tem poral region , and how to com pare and combine the individ- resolution , it is necessary to use a proper model to ual series from different locations or sub-regions is the calibrate and convert reconstructed results from dif- important issue for temperature reconstruction . ferent regions, different seasons and different kinds of Four kinds of approaches have been adopted to historical reco rds, and to make comparisons between reconstruct tem perature change by Chinese scholars . them .A four-step approach is used as follow s :(1) (1)Phenology model :inferring temperature changes First , to classify the reco rds into “natural evidence” and “impact evidence” , and to wo rk out the conver- from the differences of phenology and boundaries of sion functions of the temperature fo r different types of crop distribution for different periods .It is able to re- reco rds.(2)According to the functions , to co nvert construct past tem perature quantitatively by compar- the cold/w arm event records into temperature value ing the same kind of phenology and crop distribution fo r a regio n (site).(3)To analy ze the relationship between the past and the present[ 1] .(2)Grade mod- betw een the temperature change in individual station el :giving g rades or indices based on the description of and in the domain , then to calculate the contribution- the cold/warm events in historical records, then con- ratio of the temperature changes in different seasons verting the g rades and indices into temperature depar- and from different stations to the domain tem perature ture by comparing them w ith the modern change .(4)Based on the contribution-ratio to recon- records[ 4 , 5] .(3)Ratio model :describing the temper- struct the domain mean temperature anomaly derived ature change by the occurring frequencies of the cold/ from the temperature anomaly for the individual re- w arm events, then reconstructing the temperature se- [ 2] gion . ries by the ratio of them[ 6 ～ 9] .(4)Linear reg ression model:establishing the conversion equation betw een While reconstructing the winter half-year tem- the temperature and some given w eather/climatic perature series of the past 2000 years in eastern Chi- phenomenon (such as snow ing day s in w inter), then na , w e established the equation of the variation of using this function to convert the historical records of plant phenological date with the location (latitude) the phenomena into temperature values[ 10～ 12] . and the function between the variation of tem perature and pheno-date , under different cold/warm condi- The phenology model w as used by Chu to recon- tions based on the large amounts of the pheno-date struct his quantificational tem perature series in un- data and instrumental temperature data[ 14] .These e- [ 1] fixed temporal resolution , by supposing the tem- quations and the approaches discussed above w ere perature changes are consistent in w hole China , and used as the basis fo r reconstructing the tem perature sampling data in different temporal resolutions (Fig .