untitled


Simulated and reconstructed temperature in China since 

1550AD 

Jian Liu1, Hans von Storch2, Eduardo Zorita2, Xing Chen3, Sumin Wang1 

1. Nanjing Institute of Geography and Limnology, Chinese Academy of Sciences, Nanjing 

210008, P.R.China  

2. Institute for Coastal Research, GKSS Research Center, 21502 Geesthacht, Germany 

3. Department of Atmospheric Sciences, Nanjing University, Nanjing 210093, P.R.China  

 

Research of climate change in Little Ice Age (LIA, 1550-1850AD) has received extensive 

concern in China. Wide-spread and severe famine and serious social turmoil have taken 

place in China during LIA. Also the warm conditions after the LIA had great impact on 

human life and national economy of China (Xu, 1998). Because of this broad interest, 

Climate (in particular temperature and precipitation) of China since LIA has been 

reconstructed by Chinese scientists with a variety of proxy data, such as historical 

documents, tree rings, ice cores, lake sediments, archaeological materials, etc. (Yang, 

2002). 

Recently two long term climatic simulation experiments have been done by a consortium 

of scientists from Institute for Coastal Research, GKSS Research Center and other 

institutions. The first run, named “Christoph Columbus” (CC) was run over 535 years 

beginning in 1450, and is described in some detail by Zorita et al. (2003). The second run, 

named “Erik den Røde” (EdR), was begun in 900 and ran over 110o years (Zorita et al., 

2005). Both simulations were done with the same global atmosphere-ocean coupled 

climate model ECHO-G with, however, different code versions adapted to different 

computer systems. Reconstructed time series of the radiative effect of the presence of 

volcanic aerosols, greenhouse gases in the atmosphere as well as solar radiation were 

used as forcing. The modeling results reveal global and regional pattern of natural and 

anthropogenic climate change, which have some similarities with the observational 

record. The purpose of the present study is to compare these simulated data with 

observational evidence for the territory of China, and to interpret the recent changes of 

Chinese temperatures in the context of the forced climate model. 

China can be divided into 10 districts, which are relatively homogeneous in terms of 

temperature, precipitation and growing season (Wang, 2001). These are northeast of 

China, north of China, east of China, middle of China, south of China, southeast of China, 

southwest of China, northwest of China, Xinjiang region and Tibet region. Based on 

proxies, such as temperature index, tree ring width and density, δ18O of ice core, 
reconstructions of temperature series have been made for all these regions via statistical 

techniques. The reconstructed temperature anomalies are relative to 1880-1979 (Wang, 

1998). 



 

 

 

 

 

2

The reconstructed decadal mean temperature anomaly series of 10 regions of China are 

compared to those generated in two multi-century simulations with the climate model 

ECHO-G, which was forced by time-dependent volcanic aerosols, solar radiation and 

atmospheric greenhouse gas concentrations. The two model simulations are rather 

similar but exhibit some differences. Both the reconstructed and simulated developments 

of the temperature exhibit a “hockey-stick” pattern, with a marked increase of 

temperatures since the beginning of the 20th century. The variations of time scales of a 

few decades are, however, mostly dissimilar in the proxy-based reconstructions and in 

the simulations. 

The reconstructions have been compared with the empirical evidence available from the 

past century. Unfortunately, this evidence is very uncertain. One analysis of decadal 

mean temperatures has been prepared by Wang et al. (2001), who used three kinds of 

data: from 1951-1996 CMA's monthly temperature series from 165 stations for whole 

China, from 1911-1950 CMA's temperature grade diagram series for 139 stations of China, 

and from 1880-1910 documentary data, ice core and tree-ring data. On the other hand, 

Jones et al. (1999) made an attempt to reconstruct Chinese temperatures from 1856 to 

1998 AD. This reconstruction is rather uncertain, in particular before 1949. Both 

reconstructions deviate markedly from each other also in recent decades. 

In all cases, the warming trend since about 1900 is shared by the reconstructed 

temperatures and the simulated changes. In all 10 regions, both simulations warm with a 

rate of about 0.4-1.1 K/(100 a), whereas in Wang et al.’s reconstructions the warming is 

considerably weaker, -0.05 – 0.4 K/(100 a), which may be due to a significant cooling 

caused by the emissions of industrial aerosols, which is not accounted for in the GCM 

simulations. An exception is region 4, in SE China, where the reconstructions reveal a 

heating of 1.3 K/(100 a), and the GCM runs 0.5 and 0.8 K/(100 a). 

In the pre-industrial time (1590-1910), the GCM simulates temperatures lower than 

Wang’s estimates. In particular in the regions 1, 2, 5 and 6, the differences are larger than 

the Root Mean Square Error (RMSE). Only in the regions 4 (SE China) and 8 (Tibet) the 

simulated temperatures vary within one plus/minus RMSE. In case of Tibet the 

coincidence between the GCM and the proxy data is excellent. Also, the development in 

the Northwest, in region 7, is similar, even if the low-frequency variations are different. In 

the GCM simulations, temperatures vary around a level of -0.5 K and less, while the 

reconstructed temperatures vary around a level of -0.3 K or decline from a value close in 

1550 to zero to a minimum in the late 19th century (regions 3 and 6 in Central and 

Southern China). 

In general, the decadal variations are dissimilar in the reconstructed data and in the 

simulated data, even though the data are already heavily smoothed with a 5 decades 

running mean filter. The correlation between the reconstructed data and the 

CC-simulation for the entire period, 1590-1980, is positive (0.47 – 0.84), but these high 



 

 

 

 

 

3

values are essentially reflecting the presence of the hockey-stick pattern. If the two 

periods, pre-industrial 1590-1910 and industrial 1920-1980 are considered separately, 

the correlations become much smaller, namely -0.32 (region 6) to 0.72 (region 8) for the 

pre-industrial times, and -0.17 (region 1) to 0.32 (region 7) in modern times. Because of 

the heavy serial correlation in the data, the determination of significance levels is not 

meaningful. Interestingly, the simulated curves are among themselves rather similar, but 

the reconstructed series are quite different from region to region. 

An attempt is made to assess whether the warming during the 20th century is within the 

range of “normal’’ variations related to solar and internal dynamical influences. It is 

found that the reconstructed data are well above the pre-industrial noise level of 

temperature fluctuations during most of the 20th century. Within the adopted framework, 

only the increased greenhouse gas concentrations can account for these significantly 

elevated temperatures. The major part of the 20th century warming can be explained only 

with the help of the anthropogenic greenhouse gas effect, whereas the solar effect can 

account only for a smaller proportion. Differently from the development in the GCM 

simulations, which is a steady upward trend throughout the 20th century, there is in the 

reconstructions a decline in the 2nd half of the 20th century. We suggest that this is 

reflecting the steady increase of industrial aerosol emissions in China. 

 

References: 

Xu J-H, 1998: Sun, climate, famine and national migration. Science in China (series D) 

41(5) 449-472 (in Chinese) 

Yang B et al., 2002: Research progress on climate change during the last 2000 years. 

Progress in Earth Science 17(1): 110-117 (in Chinese) 

Zorita E, Storch H V, González-Rouco J F, et al., 2003: Simulation of the climate of the 

last five centuries. GKSS Report 2003/12, 43 pp 

Zorita E, González-Rouco J F, Storch H V et al., 2005: Natural and anthropogenic modes 

of surface temperature variations in the last thousand years. Geophysical Research 

Letters, 32, L08707, doi:10.1029/2004GL021563 

Wang Sh-W, Gong D-Y, Zhu J-H, 2001: Twentieth century climatic warming in China in 

the context of the Holocene. The Holocene, 11(3): 313-321 

Wang Sh-W, Ye J-L, Gong D-Y, 1998: Climate of Little Ice Age in China. Quaternary 

Sciences, (1): 54-64 (in Chinese) 

Jones P D, New M, Parker D E et al., 1999: Surface air temperature and its changes over 

the past 150 years. Reviews of Geophysics, 37: 173-199