Theoretical and Applied Climatology

, Volume 113, Issue 1, pp 259–269

Reconstruction of integrated temperature series of the past 2,000 years on the Tibetan plateau with 10-year intervals

Authors

    • Key Laboratory of Qinghai-Tibetan Plateau Environment and Resource, MOE, School of Life and Geographic ScienceQinghai Normal University
  • E Chongyi
    • Key Laboratory of Qinghai-Tibetan Plateau Environment and Resource, MOE, School of Life and Geographic ScienceQinghai Normal University
  • Liu Xiangjun
    • Qinghai Institute of Salt LakesChinese Academy of Sciences
  • Zeng Fangming
    • Qinghai Institute of Salt LakesChinese Academy of Sciences
Original Paper

DOI: 10.1007/s00704-012-0783-y

Cite this article as:
Guangliang, H., Chongyi, E., Xiangjun, L. et al. Theor Appl Climatol (2013) 113: 259. doi:10.1007/s00704-012-0783-y

Abstract

Using 1,981 pieces of temperature records extracted from a selection of tree rings, ice cores, sediments, and other materials with high-resolution historical temperature proxy data, a temperature series of the past 2,000 years on the Tibetan Plateau (TP) with 10-year intervals was reconstructed by the method of single sample correction—multi-sample average integration equations. This series shows that the warm periods mainly appeared before 235 A.D., 775–1275 A.D. and 1845–2000 A.D., while the cold periods occurred 245–765 A.D., 1045–1145 A.D., and 1285–1835 A.D. The Little Ice Age left clear evidence on the TP and its coldest period was between 1635 and 1675 A.D. The Medieval Warm Period on the TP was not as warm as that in the late twentieth century. During the nineteenth century, overall temperature tends to be warmer with a clear rising trend, and in the late twentieth century new highs broke the record of the past 2,000 years. Power spectrum analysis shows that temperature on the TP changes consistently and evidently in a 150-year cycle. This integrated series also shows clear correlations with sunspot activity and solar radiation, as high sunspot activities generally led to warmer periods, and vice versa. Solar activities and intense radiation of recent years are naturally conducive to the global warming since the nineteenth century. The combination of greenhouse gases and natural fluctuations in climate has been the main culprit behind the global warming in the twentieth century.

1 Introduction

Global warming, arguably the greatest world-changing phenomenon in the twentieth century, has a profound and significant impact on modern human society across the planet. Climate changes in the past 2,000 years have been research focus of the PAGES. Data from the past 20 centuries are also important climate background information for the IPCC in their assessment on global warming (PAGES 2009; Jamsen et al. 2007) because it requires ten times the volume of background information on past climates to evaluate and predict the climate changes for the decades or even centuries in the future.

The Tibetan Plateau (TP) is a hotspot in the field of climate changes research because of its unique topographic and natural features and is regarded as exerting major control on the climate changes in the East Asia. With its huge size and towering terrain, the TP significantly affects the global atmospheric circulation and highly sensitive to climate changes, and some researches indicate that changes on the scale of the twentieth century global warming was unprecedented on the TP over the past 2,000 years (Holmes et al. 2009), which is likely a natural response to human activities.

Observation data shows that the TP is one of the most sensitive and eminent regions responding to global warming. Analyzing annual variation trend of annual mean temperature using observation records from ground stations between 1961 and 2006, it is found that annual mean temperature in the 1961–2006 period on the TP increased substantially at the rate of 0.32 °C/10 a (Liu and Chen 2000; Chen et al. 2009), and that of the Qinghai Plateau climbed at the rate of 0.36 °C/10 a (Chen et al. 2009). Both are higher than the country’s average annual mean temperature, which was up only by 0.28 °C/10 a in the same period (National Assessment Report on Climate Change by the Committee 2007). It agrees well with the pattern that global changes on high-elevation plateaus are greater than those on lower-elevation plains and that global changes in high latitudes areas are greater than those at low latitudes.

Several scholars have performed integrated researches on the historical temperature changes on the TP. Yang et al. (2003) reconstructed temperature series with 50-year intervals of the TP in the past 2,000 years using ice cores, tree rings, sediments, and other evidences. The results point out that the Cold Period between the third and fifth centuries, the Medieval Warm Period (1150–1400 A.D.) and the Little Ice Age (1400–1900 A.D.) did exist on the TP, and that temperature changes on the plateau feature unique regional characteristics. Wang et al. (2007) recreated temperature series with 10-year intervals of the TP in the last 1,000 years through two sets of ice core data and found that eminent Medieval Warm Periods emerged in eastern China from 1040s to 1130s and from 1780s to 1260s, but none occurred in the Tibet region. Wang et al. also discovered that the Little Ice Age of the Tibet and Qinghai areas took place from the late sixteenth century to the early seventeenth century, while that of eastern China arose mainly from the 1600s to the 1690s and between the 1780s and the 1870s.

A series of new achievements in the studies on temperature variations on the TP were gained in recent years (Liu et al. 2007; Zhu et al. 2007; Liu et al. 2009; Wang et al. 2006). The Qilian Juniper trees in the northeastern domain of the TP provided vital records for the past climate reconstruction. In the past, dendroclimatology most often associated with precipitation records, but recently some scholars reconstructed temperature series by measuring ring widths from trees grown in central Qilian Mountain. The results indicate that there was a Little Ice Age from 1580 to 1890 A.D., in which the coldest period was between 1680 and 1720 A.D., while temperature displayed an upward trend in the late twentieth century (Liu et al. 2007). Tree rings from Ulan County in Qinghai Province helped the reconstruction of the average temperature changes from September in the previous year to April of the following year in the past 1,000 years. Results show that there was no conspicuous Medieval Warm Period, the seventeenth century was the coldest century in the past 1,000 years and the coldest 10 year-period was from 1642 to 1651, while temperature in the twentieth century was the highest and the warmest decade was reported to be the 1990s (Zhu et al. 2008). Annual mean temperature series of the past 2,485 years calculated from tree ring width records in Ulan and Dulan of Qinghai Province show that the temperatures in the four periods of 401–413 A.D., 604–609 A.D., 864–882 A.D., and 965–994 A.D. are closed to or higher than those of the 1970 to 2000 A.D. interval. Global warming of the late twentieth century isn’t the first time in the past 20 centuries, but temperature rise in the central and eastern regions of the TP since 1950 was higher than the rest of China (Liu et al. 2009).

Furthermore, figures obtained from ice core were also applied to the reconstruction of temperature variations in historical periods of the TP. Take the Malan ice core from the northern plateau as example, it revealed that the Little Ice Age began at the middle of the thirteenth century and ended in the late nineteenth century, the seventeenth century was also the warmest hundred years in the past thousand, and that global warming in the twentieth century did not exceed the natural fluctuation range of climate changes in the past 1,000 years (Wang et al. 2006).

Although many studies have been made, some results still cannot be quantified and lack grounds for comparison with each other, discrepancies exist in the recognitions of some important climate events, and researches are scattered throughout various sub-regions of the TP, lacking effective integration and therefore not accurately emulating the overall temperature changes on the TP in the past 2,000 years. In view of this, this paper used a variety of evidences to reconstruct the temperature series on the TP for the past 2,000 years, so as to offer a quantified material and a general academic consensus and mainstream viewpoints regarding the temperature changes on the TP over the past 2,000 years.

2 Data source

Twenty-four series of materials with high-resolution historical environmental records of the TP were collected (Table 1). First of all, a variety of evidences were used, including tree rings (Liu et al. 2007; Zhu et al. 2008; Liu et al. 2009), ice cores (Thompson et al. 2003; Yao et al. 1996, 2002, 2006), organic carbon contents and other sediments. These evidences have clear implications for acquiring temperature data and some series have been converted to temperature value through statistical methods. Temperature indexes reconstructed from different series of evidences do differ, for instance, ice cores mostly indicate the temperature during rainy seasons, and some tree-ring records display winter temperatures while others are able to illustrate temperatures in the summer. In any case, these series of materials are manifestations of the past temperature changes from certain aspects. The current domestic and international practices of the integration for data from various sources do not consider much about what they indicated (Mann et al. 1999). Most attempts to reconstruct hemispheric or regional temperature changes have used some variants on the “composite plus scale” methodology, in which proxy data (such as tree rings, ice cores, or corals) considered to be sensitive to past surface temperature variations are standardized and centered, potentially weighted, and then composite to form a regional or hemispheric temperature series (Mann et al. 1999; Mann and Jones 2003; Mann et al. 2008).
Table 1

Data sources for temperature reconstruction

Site

Latitude

Longitude

Evidence

Reconstruction index

Reconstruction method

Length of temperature series (A.D.)

Number of points

Central Qilian Mountain (Liu et al. 2007)

38.4

99.9

Tree ring

Temperature of December in Last Year and April in Current Year

Index of annual ring width

1000–2000

98

Ulan (Zhu et al. 2008)

37.0

78.6

Tree ring

Temperature of September in Last Year and April in Next Year

Index of annual ring width

1000–2000

118

Dulan–Ulan (Liu)

37

98.5

Tree ring

Annual Mean Temperature

Conversion function

484BC–2000

106

Dunde (Thompson et al. 2003)

38.1

96.4

Ice core

Precipitation Season Temperature

δ18O

0–2000

78

Guliya (Yao et al. 1996)

35. 3

81.5

Ice core

Precipitation Season Temperature

δ18O

300–1990

124

Dasuopu (Yao et al. 2002)

28.3

85.8

Ice core

Precipitation Season Temperature

δ18O

0–2000

125

Malan (Wang et al. 2005)

35.8

90.8

Ice core

Precipitation Season Temperature

δ18O

1129–2000

85

Puruogangriin, Qiangtang Plateau (Yao et al. 2006)

33.8

89.5

Ice core

Precipitation Season Temperature

δ18O

1901–2000

56

Delingha, Ulan (Zhu et al. 2007)

37.5

97.5

Tree ring

Temperature of June–August

Index of annual ring width

900-2000

139

Lhasa (Wu and Lin 1981)

29

91.1

Tree ring

Annual Mean Temperature

Index of annual ring width

0–2000

22

Tibetan Plateau (Wang et al. 2007)

26–40

78–103

Ice core, apparatus measuring

Annual Mean Temperature

Integration

1000–2000

100

Qinghai Plateau (Wang et al. 2005)

32–40

90–103

Tree ring

Annual Mean Temperature

Conversion function

1735–2000

49

Qamdo in Tibet (Helle et al. 2002)

31

96.9

Tree ring δ 13C

Summer Temperature

Tree ring δ13C

1000–2000

95

Lhünzhub–Nyingchi in Tibet (Feng et al. 2001)

30

91

Tree ring

Temperature Index

Conversion function

1400–2000

32

Qumalai–Zhiduo in Qinghai (Jin et al. 2002)

33.8

96.1

Tree ring

Monthly Mean Highest Temperature of April–June

Conversion Function

1550–2000

27

Dulan (Kanget al. 2000)

36.35

98.4

Tree ring

Temperature of September–December

Conversion function

159–1993

130

Western Sichuan Plateau (Qin et al. 2008)

30

101

Tree ring

Monthly Mean Highest Temperature in June

Conversion function

1617–1994

156

Cuoe Area (Wu et al. 2003)

31.5

91.5

Oxygen isotopes of calcites 

Temperature of May–September

Conversion function

1750–2000

61

Xingcuo Lake in Western Sichuan (Wu et al. 2001)

33.9

102.3

Oxygen isotopes of shells of freshwater snails

Temperature of April–August

Conversion function

1820–2000

26

Qinghai Lake (Zhang et al. 2002)

36.6

110.5

Total content of the organic carbon

Temperature

Conversion function

1100–2000

58

Qilian Mountain(Liu et al. 1984)

37.9

101

Tree ring

Temperature

Index of tree ring

1100–2000

66

Anemaqen Mountain (Gou et al. 2006)

34.8

99. 8

Tree ring

Monthly Mean Highest Temperature of April–September

Conversion function

1172–2001

110

Western Sichuan Plateau (Shao and Fan 1999)

30.3

100.9

Tree ring

Winter (December–February) Mean Lowest Temperature

Conversion function

1650–2000

60

Jinchuan in Sichuan (Wu et al. 2005)

31.5

102

Tree ring

Summer (June–September) Temperature

Maximum density index

1915–2000

60

Secondly, another merit of these data is the high-resolution, which are 10-year (Liu et al. 2009), or even annual series (Zhu et al. 2008). They provide guarantees for the precision of the reconstruction of the 10-year interval temperature series on the TP of the past 2,000 years. In addition, these evidences were collected from locations in different regions of the TP, making them highly demonstrative of the overall warming, which is over 50 % of variance, according to studies on modern temperature of the TP. Result shows that temperature changes on the plateau are roughly consistent. However, as the north is bounded by the Tanggula Mountain, temperature changes in northern TP are different from those in southern TP (Wei et al. 2003). The variety in sampling type and locations of these 24 series of materials can also reflect the general trend in temperature changes of the whole TP (Fig. 1), recording the southern and northern parts of the plateau proportionately. In essence, these temperature records are representative of the region. Finally, 1,981 pieces of temperature and time records extracted from these 24 series of high-resolution evidences using numerical methods will be utilized as data source for the synthetic reconstruction process.
https://static-content.springer.com/image/art%3A10.1007%2Fs00704-012-0783-y/MediaObjects/704_2012_783_Fig1_HTML.gif
Fig. 1

Types and distributions of temperature related records on TP used in this study

3 Reconstruction methods

This study takes the following steps to reconstruct the 10-year interval temperature series on the TP of the past 2,000 years.

(1) Standardization of data; (2) Division of time interval; (3) Single sample correction—the multi-sample average integration.

3.1 Standardization of data

The original temperature records extracted from the 24 series of materials are standardized with the sequence as a unit to eliminate the dimensional differences between different evidences so that all items can be compared with each other. After the standardization procedure, the original data becomes converted into standardized numerical values. Here min–max method is applied. The formula is shown as following:
$$ {X_{ij }}={{{\left( {{x_{ij }}-{x_{{i-\min }}}} \right)}} \left/ {{\left( {{x_{{i-\max }}}-{x_{{i-\min }}}} \right)}} \right.} $$
(1)

In this formula, Xij is the new numerical value standardized from record j in the sequence i. xi-min ,xi-max represent respectively the minimum value and the maximum value of sequence i.

3.2 Division of time interval

Integrate the 1,981 pieces of standardized records with 1 A.D. as the starting point and 10 years as interval. 1–10 A.D. is marked by 5 A.D., 11–20 A.D. by 15 A.D.…1990–2000 A.D. by 1995 A.D. In this way, the standardized data sets of ancient temperatures on the TP are established.

3.3 Single sample correction—the multi-sample average integration

The basic principle is to establish the functional relationship between the sample area where the series are distributed and the annual mean temperature changes on the TP. Then, this functional relationship will be applied to transform the standardized numerical values of ancient temperatures in a historical period to the standardized numerical values of ancient temperatures on the TP. The final step is to combine the values and reconstruct a temperature series with 10-year intervals of the TP during the past 2,000 years.
  1. (1)

    The functional relationship between annual mean temperature changes in a single sample and annual mean temperature changes in the whole nation and its sub-regions

     
This paper adopts grid point data sets of annual surface temperature from 1951 to 2007 in China (grid point size of 1° × 1°),1 then obtains the series of annual mean temperature from 1951 to 2007 on the TP from each grid point temperature value. After that, allowing the location of the series to act as the sample regions, the author uses the sample regions and strikes the 30-year moving regression relationship between the temperature series of the grid points of the sample regions from 1951 to 2007 and the temperature series on the TP from 1951 to 2007. The formula is below:
$$ Y=a{X_i}+b $$
(2)

Y stands for the annual mean temperature on the TP; Xi is the temperature in sample i; a and b represent the regression coefficient and the constant respectively. Regression coefficient can be considered to be the contribution rate of temperature. This approach was applied by Ge et al. (2003) to the reconstruction of temperature series of eastern China in the past 2,000 years. Based on the geographical elements distribution, this approach is under the principle of spatial correlation and geographic environment harmony. It means that the attribute in a certain region must coincide with the geographic location and be compatible to the surrounding attributes. The temperature changes on the TP should conform to those of the rest of the country. However, under the influences of atmospheric circulation and terrain, temperature variations have regional characteristics and the temperature contribution rate of each sample to the whole region is different. Through the analysis of the statistic relationship between the temperature in plateau samples and that on the general TP, and the conversion of the samples temperature as the temperature of the whole region, high representativeness of the region and the series’ continuity and high-resolution can all be ensured.

According to the regression relationship, there is a pretty strong relativity between most annual mean temperature in sample areas and on the TP. Additionally, the 30-year moving average series generally has a higher relativity than annual mean temperature series (Table 2). The correlation coefficient between the 30-year moving average series at most sites and that on the TP is over 0.90. Meanwhile, the explained variances (R2) of regression equation built by the 30-year moving series mostly exceed 0.90, and its residual standard deviation is smaller in order of magnitude than the year-by-year series. Therefore, due to the stability and accuracy of the 30-year moving average series, the statistical relationship is applied to the reconstruction of the temperature series with 10-year intervals on the TP for the past 2,000 years.
Table 2

Regression relationship between annual mean temperature at sample locations and the synthesized annual mean temperature of Tibetan plateau from 1951 to 2007

Location

Regression coefficient

R2

Central Qilian Mountain

0.923

0.963

Ulan

0.462

0.906

Dulan–Ulan

0.77

0.976

Dunde

0.666

0.99

Guliya

0.855

0.971

Dasuopu

0.693

 

Malan

0.887

0.957

Puruogangri in Qiangtang Plateau

0.766

0.926

Delingha, Ulan

0.732

0.973

Lhasa

0.691

0.878

Tibet Plateau

1.424

0.929

Tibet Plateau

0.733

0.881

Qamdo in Tibet

1.074

0.762

Linzhou–Linzhi in Tibet

1.074

0.965

Qumalai–Zhiduo in Qinghai

1.901

0.921

Dulan

1.151

0.796

Western Sichuan Plateau

0.78

0.968

Cuoe Area

0.954

0.984

Xingcuo Lake in Western Sichuan

0.677

0.979

Qinghai Lake

0.892

0.909

Qilian Mountain

1.901

0.921

Anemaqen Mountain

1.383

0.949

  1. (2)

    Single Sample Correction—the Multi-sample Average Integration

     
Single sample correction is to correct the multiplied product between standardized data in each sample and the contribution rate as the standardized values of the temperature on the TP, namely, convert the ancient temperature records from the local scale into the plateau scale according to the relationship between Formula 2 and Table 2. In this way, the reconstruction results from different sample areas can be compared with each other, and the high-resolution continuous series can be achieved as well. Multi-sample average integration is to take the average of the corrected temperature data within the same time interval. Firstly, we average the data of the same type and then average the data of different types to get the reconstruction result of plateau temperature in each time interval. This will enable the reconstructed numerical values of ancient temperatures in each century time interval to be applied to the calculation and allow errors to be reduced through increasing samples.
  1. (3)

    Deletion of the singular values

     
Since there are probably some relatively deviated and outlying singular values in ancient temperature records, it is necessary to delete these singular records with a comprehensive judgment method based on scattered diagram, i.e., compose a scattered diagram of the ancient temperature data rectified as the national average mean temperature anomaly in each time interval and delete the temperature values that are two times standard deviation larger than the series. In total, 112 temperature singular values are deleted from 1,981 original records and 1,869 items are retained.
  1. (4)

    Calculation of the temperature anomaly

     

Wang et al. (2007) standardized the observation data series from 1880s to 1970s with the application of Puruogangri ice core and Dasuopu shallow ice core δ18O average series (the correlation coefficient observed from the average temperature anomalies during the 10 years from 1880s to 1990s is 0.66). The standardized ice core series (before 1880s) and observed temperature data series (after 1880s) are utilized to get the 10-year interval temperature series of the TP in the past 1,000 years. This study will multiply the integrated standardized series by standard deviation of the series Wang et al. (2007) reconstructed, and then convert it to a temperature series to strike the temperature anomaly series relative to the modern timeframe (1951–1980).

4 Analysis of reconstruction results

4.1 Temperature changes over the past 2000 years on the TP

In terms of warm and cold fluctuation, if the temperature in a period is higher than the average level in the past 2,000 years, it is regarded as a relatively warm period, and vice versa. During the past 2,000 years, there were several century-scale of warm and cold fluctuation periods on the TP. The warm periods occurred from 10s to 220s A.D., from 810s to 1290s A.D. and from 1860s to 1990s A.D. The temperature in the first two warm periods was 0.1 °C higher than the average value in the past 2,000 years, and the temperature in the third period was 0.3 °C above the 2,000 years average value (Fig. 2). The cold periods happened between 230s and 800s A.D., and between 1410s and 1870s A.D., during those two periods the air temperature was 0.1 °C lower than the average value in the past 2,000 years.
https://static-content.springer.com/image/art%3A10.1007%2Fs00704-012-0783-y/MediaObjects/704_2012_783_Fig2_HTML.gif
Fig. 2

Reconstructed integrated temperature series with 10-year interval on the Tibetan Plateau in the past 2,000 years (thick blue line is the filter series, three points weighted coefficients successively are: 0.25, 0.5, 0.25; red line is the original integrated temperature series; anomaly is relative to the average value of the past 2,000 years)

In Fig. 2, a long relatively warm period from 810s to 1290s A.D., is roughly equivalent to the Medieval Warm Period. The warmest time in the Medieval Warm Period was from 840s to 980s A.D., (0.3 °C higher than average temperature). When viewed on 10-year intervals, there were several periods (such as 850s A.D.) when temperatures are fairly similar to the temperature of the warm periods in the twentieth century, but its overall air temperature is still cooler than that of the warm periods in the twentieth century (GW). Therefore, the Medieval Warm Period on the TP is one that is relatively less warm, and data from the various areas indicate that the temperature in that warm period is close to or slightly lower than that at present. The temperature records reconstructed from the historical references by Ge et al. (2003) shows that the Medieval Warm Period in eastern China occurred from 930 to 1300 A.D. Wang et al. (2002) integrated some records such as tree rings, glaciers, lakes and deserts, and found that in western China the Medieval Warm Period was not evident. The air temperature from the seventh to the ninth century was warm and wet, and the stretch between the tenth to twelfth centuries saw large temperature fluctuation. The causes that led to such phenomenon need to be examined further together with the factors of regional differences in climate. In the integrated plateau series, GW represents the global warming in the twentieth century. Temperature on the TP began to rise rapidly since 1820s. From 1800s to 1900s A.D., the temperature increased by more than 1 °C, and in the late twentieth century, the steadily rising trend reached its peak. Especially from the 1980s to the 90s, it was the warmest period during the past 2,000 years.

WJCP in Fig. 2 is a distinctively cold period from 230s to 800s A.D., during which the coldest period was between 260s and 360s A.D. This is popularly referred to as the Wei Jin Cold Period in Chinese history when the average temperature is 0.2 °C lower than average, with 360s A.D. being the coldest decade. The temperature records of the northeastern boundary of the TP in the past 2,000 years recreated from tree rings demonstrate that the period from 348 to 366 A.D. was one of the coldest periods (Liu et al. 2009). Historical records show that between 260s and 360s A.D. was a time of major chaos, when many northern peasants, grazing-based groups and inhabitants of the Yellow River basin were forced to move south, invading the land of southern farmers and leading to the notorious Yongjia Chaos of 311 A.D. In Fig. 2, LIA is the well-known Little Ice Age (1410s–1870s A.D.), reached its prime from 1630s to 1680s A.D., when the plateau’s temperature was 1 °C lower than that of present stage (1950s–1970s A.D.) and 0.4 °C lower compared to the average of the past 2,000 years. The 1680s A.D. being the coldest 10 years in the past 2,000 years with temperature 0.6 °C below average. During the Little Ice Age, the severe and long-lasting Chongzhen Drought (1637–1643 A.D.) broke out, which crippled 23 provinces throughout the whole nation, ultimately resulting in a peasant uprising that marked the beginning of the end of a dynasty.

What should be noted is that Fig. 2 also lists the number of records used in the reconstruction of temperature series. It can be seen that more records are available and used when approaching the later period, meaning that the series increase in reliability and decrease in errors.

4.2 Comparison with other temperature series

Comparing the integrated series of this study with other series of the TP (Yang et al. 2003), eastern China in winter (Ge et al. 2003), the temperature series in China (Yang et al. 2002), and the northern hemisphere temperature series over the past 2,000 years (Mann and Jones 2003), it is found that overall, the temperature series in this study exhibits a trend of changes similar to other series. It proves that the temperature changes on the TP during the past 2,000 years have strong correlations and are closely related to, or keeping in line with temperature shifts in eastern China, the whole China, or even the entire northern hemisphere. For instance, Fig. 3 shows that all temperature series have the heyday of Little Ice Age. Both integrated series (MWP5) and Tibetan Plateau series of Yang et al. (2003) (MWP4) have regarded that the Medieval Warm Period on the TP was not an obviously warm period as its temperature was only higher than annual mean temperature in the past 2,000 years but its warmth intensity was lower than that in the warm period of the twentieth century. Meanwhile, Mann’s northern hemisphere series (MWP1) has not shown a clear Medieval Warm Period. These series unilaterally illustrate that global temperatures have universally risen again after the Little Ice Age in the nineteenth century and stepped into the global warming phase in the twentieth century with a linear upward trend, and the peaks of the data from all the series, except Ge’s, reaching or exceeding the warmest state of the past 2,000 years (Fig. 3 GW).
https://static-content.springer.com/image/art%3A10.1007%2Fs00704-012-0783-y/MediaObjects/704_2012_783_Fig3_HTML.gif
Fig. 3

Comparisons between our integrated series and other temperature series. Key: (a) the northern hemisphere temperature (Mann and Jones 2003); (b) temperature in China (Yang et al. 2002); (c) winter temperature in eastern China (Ge et al. 2003); (d) temperature series on the Tibetan Plateau (Yang et al. 2003); (e) integrated series of this study

However, there are also some differences among these temperature series that must be addressed. In Yang’s Tibetan Plateau series, the apex of the Little Ice Age formed around the beginning of the seventeenth century and the level of coldness was weaker than those presented in other series. Both Ge’s and Yang’s series in China show that the Medieval Warm Period was pronounced, but the two deviates in the intensity of the Medieval Warm Period, which in Ge’s series is stronger than that of the global warming in the twentieth century (MWP3) while that in Yang’s series are equivalent to the late twentieth century (MWP2), and yet the series of this study shows that the temperature in Medieval Warm Period on the TP is lower than that in the late twentieth century, with only several 10-year intervals that approach the intensity of the late twentieth century (MPW5). Studies referencing glaciers in western China, tree rings and some historical materials from eastern China have shown that the Medieval Warm Period in western China was not as obvious as that in eastern China (Zheng and Wang 2005). From the perspective of frequency of occurrence, the warmest period of Medieval Warm Period in both series of this paper and Yang’s series in China are considered to be from 800 to 1000 A.D., while Ge’s series and Yang’s Tibetan Plateau series deem that it was generally in the stage between 1200 and 1400 A.D. Furthermore, for the time segment before 200 A.D., in the series of this paper and Yang’s series in China it is viewed as a weak warm period compared with the global warming in the twentieth century (HWP5, HWP2), but conversely, Ge’s series and Yang’s Tibetan Plateau series express that it was equivalent to, or even slightly higher than, the degree of warmness in the late twentieth century (HWP3, HWP4).

What should be pointed out is that even though both the series of this paper and Yang’s series are related to the temperature series of the TP in the past 2,000 years, but the time interval used for measurement in this paper is an improvement. Since the time interval in Yang’s series is 50 years and in this study it is 10 years, it means that the integrated temperature series in this study can record more accurate climate changes. Further, this study makes use of more types of environmental evidences and refers to some new achievements published in recent decades, which are integral components in the reconstruction of this high-precision integrated temperature series. In this way, these integrated results can better showcase the latest researches and shed light on the more refined processes of climate changes. Finally, another difference between this study and Yang’s series is the integration methods. The approach of integration between the process of the modern temperature changes on the plateau and the ancient temperature materials was utilized and different regional responses to the temperature changes on the plateau were taken into consideration in this study. Thus, this integrated series of temperature changes was reconstructed to more accurately represent the details of the entire plateau.

4.3 Cycle analysis

Power spectrum analysis was administered to the five-point filtering of the temperature series with 10-year interval on the TP in the past 2,000 years. The result shows that a cycle of around 150 a passed the reliability test of 0.05, indicating that applying such a cycle of around 150 a to document temperature changes on the TP is more clear and consistent (Fig. 4). The cycle of around 150 a has been recorded widely in China, and it has been proven that the distribution of organic carbon and sporopollen series exhibited clear periodic changes of 180 a and 131 a in the study on sediments at Minqin basin and Shiyang River valley (Chen et al. 2001). Also, the change cycle of 136 a existed in the records of stalagmite in China (Tan et al. 1998).
https://static-content.springer.com/image/art%3A10.1007%2Fs00704-012-0783-y/MediaObjects/704_2012_783_Fig4_HTML.gif
Fig. 4

Power spectrum of integrated filtering series (step size 30, the red noise confidence level 0.05)

4.4 Discussion for reasons of temperature changes

There was a relatively correlated relationship between solar radiation changes and sunspot activities changes over the past 2,000 years. Comparing the sunspot activity series and temperature integration series on the TP, it can be found that the temperature changes during the past 2,000 years were undeniably driven by the sun. Stronger solar radiation at phases of vibrant solar activity correspond with the warm periods, and vice versa. The most evident period was the Little Ice Age, which coincided with the Maunder Minimum, when solar radiation was lowest and sunspots rarest in the past 2,000 years. Activities of sunspots and solar radiation were intense during the Medieval Warm Period (from 810 to 1290 A.D). Steady temperature rises beginning in the 1820s and still continuing to present day (Fig. 5 GW-a/b/d). We compare the temperature series with the global emission of carbon dioxides. It is discovered that the recent warming on the TP began in the early nineteenth century, while the major CO2 emissions resulting from human activities took place mainly since the twentieth century, so it can be inferred that the recent warming precedes the artificial carbon dioxides emission spikes and that there are other natural causes for the temperature rises on the TP during the nineteenth century. However, the large amount of carbon dioxides emission also reveal a high degree of relevance with the global warming trend in the twentieth century, and the level of temperature increase in the twentieth century has surpassed the range of natural climate changes of the past 2,000 years, meaning that the tremendous volume of CO2 produced must driving global warming. Therefore, it can be said that recent intense solar activities were the natural forces behind global warming in the nineteenth century, the effect of which is then compounded by the growing emission of greenhouse gases and consequently producing the global warming of the twentieth century.
https://static-content.springer.com/image/art%3A10.1007%2Fs00704-012-0783-y/MediaObjects/704_2012_783_Fig5_HTML.gif
Fig. 5

Comparisons between integrated series, sunspot activities and CO2 Emissions. Key: (a) solar radiation (Steinhilber and Beer 2009); (b) sunspot activities (Solanki et al. 2004); (c) global CO2 emissions (Carbon Dioxide Information Analysis Center 2009); (d) series of this paper

5 Conclusion

  1. (1)

    Twenty-four series of high-resolution environmental records associated with the temperature changes on the TP have been collected; 1,981 pieces of temperature records have been extracted from these evidences to reconstruct the integrated temperature series with 10-year intervals on the TP of the past 2,000 years, through methods such as standardization of data, division of time interval, and single sample correction—multi-sample average integration equation.

     
  2. (2)

    Integrated temperature change series shows that during the past 2,000 years, there were century-scale groups of warm and cold fluctuation events on the TP. Among them the warm periods mainly occurred before 235 A.D., from 775 to 1285 A.D. and from 1845 to 2000 A.D., while the cold periods took place from 245 to 765 A.D., from 1045 to 1145 A.D. and from 1275 to 1835 A.D. The Little Ice Age, during which the coldest period was between 1635 and 1675 A.D., was prominent on the TP. The Medieval Warm Period on the TP was not conspicuous. The obvious warming started in the early nineteenth century and broke through historical records in the late twentieth century. This integrated series is basically the same as the series on the TP, eastern China, China and northern hemisphere temperature constructed by other scholars previously.

     
  3. (3)

    Power spectrum analysis of this integrated series illustrates that a cycle of around 150 a displays evident and consistent temperature changes on the TP, and it is related to the approximately 115 a and 190 a cycles of sunspot activities. Through the comparison between our integrated series and the sunspot activities, it can be found that the active periods of solar activity in general correspond to the warm periods, while the quiet periods of solar activity are linked to the cold periods. The recent more intense solar activities were the natural reason of the global warming in the nineteenth century, and significant greenhouse gases emissions later added to the natural climate fluctuations, resulting in the global warming of the twentieth century.

     
Footnotes
1

Zhang Qiang, Grid points data set of annual surface temperature. 2007.

 

Acknowledgments

This work was supported by the China NSF (grant 41161018 and 41201014).

Copyright information

© Springer-Verlag Wien 2012