Changes in daily climate extremes in the eastern and central Tibetan Plateau during 1961–2005

Authors


Abstract

[1] Changes in indices of climate extremes are analyzed on the basis of daily maximum and minimum surface air temperature and precipitation at 71 meteorological stations with elevation above 2000 m above sea level in the eastern and central Tibetan Plateau (TP) during 1961–2005. Twelve indices of extreme temperature and nine indices of extreme precipitation are examined. Temperature extremes show patterns consistent with warming during the studied period, with a large proportion of stations showing statistically significant trends for all temperature indices. Stations in the northwestern, southwestern, and southeastern TP have larger trend magnitudes. The regional occurrence of extreme cold days and nights has decreased by −0.85 and −2.38 d/decade, respectively. Over the same period, the occurrence of extreme warm days and nights has increased by 1.26 and 2.54 d/decade, respectively. The number of frost days and ice days shows statistically significant decreasing at the rate of −4.32 and −2.46 d/decade, respectively. The length of growing season has statistically increased by 4.25 d/decade. The diurnal temperature range exhibits a statistically decreasing trend at a rate of −0.20°C per decade. The extreme temperature indices also show statistically significant increasing trends, with larger values for the index describing variations in the lowest minimum temperature. In general, warming trends in minimum temperature indices are of greater magnitude than those for maximum temperature. Most precipitation indices exhibit increasing trends in the southern and northern TP and show decreasing trends in the central TP. On average, regional annual total precipitation, heavy precipitation days, maximum 1-day precipitation, average wet days precipitation, and total precipitation on extreme wet days show nonsignificant increases. Decreasing trends are found for maximum 5-day precipitation, consecutive wet days, and consecutive dry days, but only the last is statistically significant.

1. Introduction

[2] The Fourth Assessment Report of the Intergovernmental Panel on Climate Change (IPCC) shows an increase in global mean temperature of approximately 0.74°C during the latest century [Intergovernmental Panel on Climate Change, 2007]. In the context of global warming, variations and trends in extreme climate events have recently received much attention because extreme climate events are more sensitive to climate change than their mean values [Katz and Brown, 1992]. At the same time, global climate change is expected to have a considerable impact on the global hydrological cycle. It has been seen that the economy, human health and the natural environment are becoming vulnerable to the extreme climate events [Easterling et al., 2000; Kunkel et al., 1999].

[3] Precipitation and temperature extremes have been studied in many regions around the world, such as in the Asia-Pacific region [Griffiths et al., 2005; Manton et al., 2001], Caribbean region [Peterson et al., 2002], southern and west Africa [New et al., 2006], South America [Haylock et al., 2006; Vincent et al., 2005], Middle East [X. Zhang et al., 2005], Central America and northern South American [Aguilar et al., 2005], and central and south Asia [Klein Tank et al., 2006]. Global changes in daily climate extremes have been analyzed [Alexander et al., 2006; Frich et al., 2002]. These studies concluded that widespread significant changes in temperature extremes are associated with warming, while the changes in precipitation extremes present much less spatially coherence compared with temperature change. For China, precipitation has increased by 2% and the frequency of precipitation events has decreased by 10% from 1960 to 2000 [B. Liu et al., 2005], mean minimum temperature has increased significantly and mean maximum temperatures display no statistically significant trend between 1950 and 1995 [Zhai et al., 1999].

[4] The Tibetan Plateau (TP), with an average elevation of over 4000 m above sea level (asl), is the highest and largest highland in the world and exerts a great influence on regional and global climate through its thermal forcing mechanisms [Duan and Wu, 2005; Yanai et al., 1992; Yeh and Gao, 1979]. Previous studies [Duan and Wu, 2006; Kang et al., 2007; Lin and Zhao, 1996; X. Liu and Chen, 2000; Niu et al., 2004] showed a significant warming in the TP during the last half century, in phase with the global trends derived from the increasing anthropogenic greenhouse gases emissions. The TP region is expected to be one of the most seriously impacted areas in the world by global warming effects [Chen et al., 2003; Duan et al., 2006]. However, there have been few studies in temperature and precipitation extremes in the TP, primarily owing to the lack of easily available data collection for the region. For precipitation a clear regional signal has not been identified mainly due to the complex terrain and sparse meteorological stations [Du and Ma, 2004; Lin and Zhao, 1996; X Liu and Yin, 2001]. Du [2001] analyzed monthly maximum and minimum temperatures for 16 stations in the TP from 1961–2000 and found that the magnitude of trend in minimum temperature was greater than that in maximum temperature. X. Liu et al. [2006] also confirmed the asymmetric pattern of greater warming trends in nighttime temperatures as compared to the daytime temperatures.

[5] The objective of this study is to investigate the climate change in temperature and precipitation extremes during the period 1961–2005 in the eastern and central TP, through the analysis of indices generated by the Commission for Climatology (CCl)/Climate Variability and Predictability (CLIVAR)/Joint WMO-IOC Technical Commission for Oceanography and Marine Meteorology (JCOMM) Expert Team (ET) on Climate Change Detection and Indices (ETCCDI) (http://cccma.seos.uvic.ca/ETCCDI/), a widely used approach (see section 2 for more). We have analyze the relationship between trends in temperature extremes and elevation using the same method in this region [You et al., 2008]. Analyzing these indices will hopefully lead to a better understanding of variability and changes in the frequency, intensity and duration of extreme climate events in the TP. Spatial and temporal variability of the changes in temperature and precipitation extremes are discussed in this work.

2. Data and Methods

[6] Data including daily precipitation, maximum temperature and minimum temperature is provided by the National Climate Center, China Meteorological Administration. The TP in China ranges from 26°00′12″ to 39°46′50″N and from 73°18′52″ to 104°46′59″E, and is distributed in 6 provinces, namely, the Tibet Autonomous Region, the Qinghai Province, Yunnan Province, Sichuan Province, Gansu Province and Xinjiang Uigur Autonomous Region [Y. Zhang et al., 2002]. There are 156 stations in the original data. A total of 124 stations maintain daily data since 1961, of these stations, 38 stations are excluded owing to the elevation below 2000 m asl, then 12 stations are also excluded owing to problems in data (10 stations stopped operation during the 1980s—1990s; 2 stations showed abnormity due to the discontinuity of data record). The distribution of the stations is uneven and very sparse in the western TP, which may influence the regional trends. Therefore 3 stations in the western TP are also excluded. The remaining 71 stations with elevation above 2000 m asl were selected, which observation started no later than 1961 and they were located in the eastern and central TP and in very close places which are not administratively in the region, but are relevant to this study (Figure 1).

Figure 1.

The distribution of 71 stations used in this study in the eastern and central Tibetan Plateau (TP) and adjacent territories.

[7] Stations are identified by their World Meteorological Organization (WMO) number and the stations name, along with longitude, latitude, elevation and missing period during 1961–2005 (Table 1). Most meteorological stations were established during the 1950s, and the selected 71 stations are located with station altitudes varying between 2109.5 m (56684-Huize) and 4700 m (55279-Bange). Fourteen stations are located above 4000 m and nine stations are situated between 2000 m and 2500 m (Figure 2). In order to obtain comparable time series we select data only covering the period of 1961 to 2005, excluding the spares data available for earlier periods. The stations with shorter records are not selected in this study but they are still available for assessing data quality and homogeneity at nearby stations.

Figure 2.

Number of selected stations with observation start year (top) and number of selected stations above the categorized elevation (bottom) in the eastern and central TP.

Table 1. List of the Selected Stations Above 2000 m Above Sea Level in the Eastern and Central Tibetan Plateau, Including the World Meteorological Organization (WMO) Number, Station Name, Latitude, Longitude, Elevation, and Data Missing Period During 1961–2005
WMO NumberStation NameNorth LatitudeEast LongitudeElevation (m)Data Missing Period
52787Wuqiaoling37°12′102°52′3045.1 
56080Hezuo35°102°54′2910 
56093Minxian34°26′104°01′2315 
51886Mangya38°15′90°51′2944.8 
52602Lenghu38°45′93°2′2770 
52633Tuole38°48′98°25′3367 
52645Yeniugou38°25′99°35′3320 
52657Qilian38°11′100°15′2787.4 
52707Xiaozaohuo36°48′93°41′2767Apr–Dec 1974
52713Dachaidan37°51′95°22′3173.2 
52737Delingha37°22′97°22′2981.5 
52754Gangcha37°2′100°08′3301.5 
52765Menyuan37°23′101°37′2850 
52818Germu36°25′94°54′2807.6 
52825Nuomuhong36°26′96°25′2790.4 
52836Dulan36°18′98°06′3191.1 
52856Gonghe36°16′100°37′2835 
52866Xinning36°43′101°45′2295.2 
52868Guide36°02′101°26′2237.1 
52908Wudaoliang35°13′93°05′4612.2 
52943Xinghai35°35′99°59′3323.2 
56004Tuotuohe34°13′92°26′4533.1 
56018Zaduo32°54′95°18′4066.4 
56021Qumalai34°08′95°47′4175Aug–Dec 1962
56029Yushu33°01′97°01′3681.2 
56033Maduo34°55′98°13′4272.3 
56034Qingshuihe33°48′97°08′4415.4 
56046Dari33°45′99°39′3967.5 
56067Jiuzhi33°26′101°29′3628.5Apr–May 1962
56125Nangqian32°12′96°29′3643.7 
56151Banma32°56′100°45′3530Apr 1962–1965
55279Bange31°23′90°01′4700Apr 1965
55299Naqu31°29′92°04′4507 
55472Shenzha30°57′88°38′4672 
55578Rikeze29°15′88°53′3836 
55591Lhasa29°4′91°08′3648.7Jun–Oct 1968
55598Zedang29°15′91°46′3551.7 
55664Dingri28°38′87°05′4300Nov 1968–1969, Aug 1969–Sep 1970
55680Jiangzi28°55′89°36′4040 
55696Longzi28°25′92°28′3860 
55773Pali27°44′89°05′4300 
56106Suoxian31°53′93°47′4022.8 
56116Dingqing31°25′95°36′3873.1Jun–Aug 1969
56137Changdu31°09′97°1′3306 
56227Bomi29°52′95°46′2736 
56312Linzi29°4′94°2′2991.8 
56038Shiqu32°59′98°06′4200 
56079Ruoergai33°35′102°58′3439.6 
56144Dege31°48′98°35′3184 
56146Ganzi31°37′100°3393.5 
56152Seda32°17′100°2′3893.9 
56167Daofu30°59′101°07′2957.2 
56172Maerkang31°54′102°14′2664.4 
56173Hongyuan32°48′102°33′3491.6 
56178Xiaojing31°102°21′2369.2 
56182Songpan32°39′103°34′2850.7 
56247Batang30°99°06′2589.2May–Dec 1968
56251Xinlong30°56′100°19′3000 
56257Litang30°100°16′3948.9Sep 1967, Jan–Jul 1968, May–Aug 1969
56357Daocheng29°03′100°18′3727.7May 1968
56374Kangding30°03′101°58′2615.7 
56385Emeishan29°31′103°2′3047.4 
56459Muli27°56′101°16′2426.5 
56462Jiulong29°101°3′2987.3 
56479Zhaojue28°102°51′2132.4 
56565Yanyuan27°26′101°31′2545 
56444Deqin28°29′98°55′3319 
56543Zhongdian27°5′99°42′3276.1 
56548Weixi27°1′99°17′2325.6 
56651Lijiang26°52′100°13′2392.4 
56684Huize26°25′103°17′2109.5 

[8] Data quality control is a necessary step before the calculation of indices because erroneous outliers can seriously impact the indices calculation and their trends. Data quality control and calculation of the indices are performed using the computer program RClimDex, which is developed and maintained by Xuebin Zhang and Feng Yang at the Climate Research Branch of Meteorological Service of Canada. Software and documentation are available online for downloading (http://cccma.seos.uvic.ca/ETCCDI/). The software identifies on a first run erroneous temperature and precipitation data, such as precipitation values below 0 mm or days with Tmax< Tmin. Additional execution, identify potential outliers, which have to be manually checked, validated, corrected or removed. For temperature, they are defined as values outlying a user-defined threshold determined by mean plus/minus a number of standard deviations. In our case, we choose 3 standard deviations as the thresholds for a finer quality control of the data. Both for precipitation and temperature, data plots are available for visual inspections to reveal more outliers as well as a variety of problems that cause changes in the seasonal cycle or variance of the data. Also, histograms of the data are created which reveal problems that show up when looking at the data set as a whole [Aguilar et al., 2005; New et al., 2006]. Figure 3 is an example of the plots used to quality control precipitation data. It explains the data density in two different ways: a histogram (bars) and a Kernel-filtered (line) which is a nonparametric approach to density fitting [Aguilar et al., 2005]. Both show that precipitation data in the station is fine.

Figure 3.

Example of precipitation successful quality control procedures using RClimdex. Histogram (vertical bars) and Kernel-filtered density (line).

[9] Homogeneity assessment and adjustment can be quite complex and it often requires close neighbor stations, detailed station history and a great amount of time [Vincent et al., 2005]. Data homogeneity is assessed using the RHtest software (available from the ETCCDI Web site), which uses a two-phase regression model to check for multiple step-change points that could exits in a time series [Wang, 2003; Wang and Zhou, 2005]. The two-phase regression model was applied to annual mean daily maximum, minimum temperature as well as daily temperature range, to identify potential inhomogeneities in the data [X. Zhang et al., 2005]. Once a possible step change is identified in the annual series, it is also checked against the station history. There are 13 stations with a potential step in annual maximum temperature and 8 stations with a potential step in annual minimum temperature. Historical explanations for the cause of the step, such as the relocation, are found for only two stations. Therefore, we removed them from our final data set. Figure 4 shows an example where a step change has been detected in the Tibet Autonomous Region. The station shows a large inhomogeneity in 1983, corroborated by the station history, which shows a piece of metadata saying that it relocated that year.

Figure 4.

Homogeneity assessment results for annual mean daily minimum (top) and maximum (bottom) temperature for Jiali station (30°40′, 93°17′, 4488.8 m above sea level). The largest, statistically significant discontinuity around 1983 is verified by the original station data, which indicate that the station relocated in 1983.

[10] After data quality control and homogeneity assessment, RClimDex is used to calculate climate indices from the daily data. Expert Team for Climate Change Detection and Indices (ETCCDI) has been coordinating a suite of 11 precipitation and 16 temperature indices. For percentile indices a bootstrap procedure has been implemented to ensure that the percentile-based temperature indices do not have artificial jumps at the boundaries of the in-base and out-of-base period [X. Zhang et al., 2004]. Some of the indices, such as the number of tropical nights, the number of warm or cold duration and so on, are not relevant to the studied region and have not been used, leading to a final selection of 12 temperature indices and 9 precipitation indices (Table 2). They have been calculated over the quality controlled data of the stations that passed the homogeneity assessment. Table 2 provides their descriptions.

Table 2. Definitions of 12 Temperature Indices and 9 Precipitation Indices Used in This Studya
IndexDescriptive NameDefinitionUnits
  • a

    All the indices are calculated by RClimDEX. Abbreviations are as follows: TX, daily maximum temperature; TN, daily minimum temperature; TG, daily mean temperature; RR, daily precipitation. A wet day is defined when RR ≥ 1 mm, and a dry day is defined when RR < 1 mm. Indices are included for completeness but are not analyzed further in this article.

Temperature
TXxwarmest dayannual highest TX°C
TNxwarmest nightannual highest TN°C
TXncoldest dayannual lowest TX°C
TNncoldest nightannual lowest TN°C
TN10cold night frequencypercentage of days when TN < 10th percentile of 1961–1990%
TX10cold day frequencypercentage of days when TX < 10th percentile of 1961–1990%
TN90warm night frequencypercentage of days when TN > 90th percentile of 1961–1990%
TX90warm day frequencypercentage of days when TX > 90th percentile of 1961–1990%
DTRdiurnal temperature rangeannual mean difference between TX and TN°C
IDice daysannual count when TX < 0°Cd
FDfrost daysannual count when TN < 0°Cd
GSLgrowing season lengthannual count between first span of at least 6 days with TG > 5°C after winter and first span after summer of 6 days with TG < 5°Cd
 
Precipitation
PRCPTOTwet day precipitationannual total precipitation from wet daysmm
SDIIsimple daily intensity indexaverage precipitation on wet daysmm/d
CDDconsecutive dry daysmaximum number of consecutive dry daysd
CWDconsecutive wet daysmaximum number of consecutive wet daysd
R10mmnumber of heavy precipitation daysannual count of days when RR ≥ 10 mmd
R95very wet day precipitationannual total precipitation when RR > 95th percentile of 1961–1990 daily precipitationmm
R99extremely wet day precipitationannual total precipitation when RR > 99th percentile of 1961–1990 daily precipitationmm
RX1daymaximum 1-day precipitationannual maximum 1-day precipitationmm
RX5daymaximum 5-day precipitationannual maximum consecutive 5-day precipitationmm

[11] The selected indices are calculated on a monthly and/or annual basis. Some indices are based on threshold defined as percentiles. The percentiles are calculated from the reference period 1961–1990, which is a climate normal period widely used. Monthly indices are obtained if no more than 3 days are missing in a month and annual values are calculated if no more than 15 days are missing in a year. Threshold indices are computed if at least 70% of the data are present in the reference period.

[12] In this study, linear trends for indices are calculated using a nonparametric approach, a Kendall's τ-based Sen's robust slope estimator [Sen, 1968], which is adapted and applied in a study of annual temperature and precipitation change over Canada [ X Zhang et al., 2000], and in extreme wave heights over Northern Hemisphere oceans [Wang and Swail, 2001]. Annual missing value is excluded from the analysis when calculating the linear trend. The 95% confidence intervals are calculated from tabulated values [Kendall, 1955]. The significance of the trends is determined using an iterative procedure [Wang and Swail, 2001; X. Zhang et al., 2000] to compute the trends and to test the trends significance taking account of a lag-1 autocorrelation effect. For the eastern and central TP as a whole, the regional series are calculated by averaging anomalies relative to 1961–2005. In order to avoid average series being dominated by those stations with high precipitation, regional series for precipitation indices are calculated again: standardizing the simple anomaly through dividing by the station standard deviation during the studied period. All the regional series are converted into trends per decade when describing linear regression trends. A trend is considered to be statistically significant if it is significant at the 5% level.

3. Results

[13] The analysis of temperature and precipitation reveal a variety of changes in extreme values during 1961–2005 in the eastern and central TP. Spatial patterns of trends in temperature extremes have a much higher degree of coherence while precipitation in the region has more variability. The results for indices in climate extremes are described along this section.

3.1. Temperature

3.1.1. Cold Extremes (TX10, TN10, TXn, TNn, FD, ID)

[14] Figure 5 shows the spatial distribution pattern of the temporal trends in cold extremes for the 71 meteorological stations and Figure 6 demonstrates the regional annual anomalies series for indices of cold extremes in the eastern and central TP. The regional trends in indices of cold extremes are in Table 3. Table 4 shows the number of stations with significant negative, nonsignificant, and significant positive trends for cold extremes indices during 1961–2005. For cold nights (TN10) and cold days (TX10), about 77% and 44% of stations have decreasing trends that are statistically significant. Stations in the northern and southwestern TP, especially around the Qaidam Basin, have larger trend magnitudes, while there are still a few stations that have increasing trends for cold days (TX10) and occur mainly in the southeastern TP. The cold days (TX10) have fluctuant variations before the mid-1980s and decrease annually after that, but the cold nights (TN10) have continually decreasing trends during the period of 1961–2005. The regional trends (in percentage of days) for these two indices are −2.38 and −0.85 d/decade, respectively.

Figure 5.

Spatial patterns of trends per decade for indices of cold extremes. Positive trends are shown as solid dots, negative trends as open dots. The size of the dot is proportional to the magnitude of the trends.

Figure 6.

Regional annual anomalies series relative to 1961–2005 for indices of cold extremes. The smoother line is the 9-year smoothing average.

Table 3. Trends Per Decade for Regional Indices of Temperature and Precipitation Extremesa
IndexUnits1961–2005
  • a

    Parentheses are 95% confidence intervals. Values for trends significant at the 5% level (t test) are set in bold. The bottom rows give the trends for the ratios R95/RR and R99/RR.

Temperature
TN10d/decade−2.38 (−2.85 to −1.91)
TX10d/decade−0.85 (−1.35 to −0.37)
TN90d/decade2.54 (1.84–3.11)
TX90d/decade1.26 (0.62–1.87)
DTR°C/decade−0.20 (−0.26 to –0.14)
TNn°C/decade0.69 (0.51–0.87)
TNx°C/decade0.25 (0.17–0.34)
TXn°C/decade0.30 (0.06–0.53)
TXx°C/decade0.28 (0.12–0.42)
FDd/decade−4.32 (−5.53 to −3.28)
GSLd/decade4.25 (2.87–5.69)
IDd/decade−2.46 (−3.45 to −1.23)
 
Precipitation
PRCPTOTmm/decade6.66 (−0.08–12.54)
SDIImm/decade0.03 (−0.01–0.07)
RX1daymm/decade0.27 (−0.03–0.63)
RX5daymm/decade−0.08 (−0.92–0.70)
R10mmd/decade0.23 (−0.02–0.50)
R95mm/decade1.28 (−1.55–4.15)
R99mm/decade1.09 (−0.30–2.33)
CDDd/decade−4.64 (−7.21 to −2.33)
CWDd/decade−0.07 (−0.22–0.08)
R95/RR%/decade0.15 (−0.26–0.56)
R99/RR%/decade0.17 (0–0.41)
Table 4. Number of Stations With Significant Negative, Nonsignificant, and Significant Positive Trends for the Annual Temperature and Precipitation Indices During 1961–2005a
IndexNegativeNonsignificantPositive
  • a

    Significant at the 0.05 level. Numbers of stations with negative and positive are also known in parentheses.

Temperature
TN1055 (70)160 (1)
TX1031 (63)391 (8)
TN900 (1)1160 (70)
TX901 (6)3535 (65)
DTR31 (59)391 (12)
TNn0 (5)3140 (66)
TNx0 (4)3536 (67)
TXn0 (14)5615 (56)
TXx1 (11)4228 (59)
FD52 (70)190 (1)
GSL0 (4)4328 (67)
ID21 (62)500 (3)
 
Precipitation
PRCPTOT1 (19)637 (52)
SDII0 (26)683 (44)
RX1day1 (29)673 (41)
RX5day2 (34)681 (36)
R10mm1 (20)646 (49)
R950 (24)674 (46)
R990 (23)692 (48)
CDD13 (62)571 (8)
CWD2 (39)681 (31)

[15] Similarly the temperatures of coldest days and coldest nights in each year (TXn and TNn) show increasing trends at approximately 80–90% of stations. But only 21% and 56% of stations for these two indices have statistically significant trend due to the higher variance of this index. The stations presenting larger trend magnitudes are also situated in the northern and southeastern TP for these two indices, while there are a few stations with decreasing trends in the middle and southeastern TP only for TXn. It can be seen that TXn has a slight decreasing trend from 1961 to 1980 then turns to increasing trend after 1980, while TNn has a clear decreasing trend during 1961–2005, the regional trends in TXn and TNn are 0.30 and 0.69°C/decade, respectively, which is compatible with decreasing trend in TX10 and TN10.

[16] The number of ice days (ID) has decreased at a rate of −2.46 d/decade. The overall trend is significant at the 0.001 level and intensifies after 1990. Around 30% of stations have a statistically significant decreasing trend mainly occurring in Qinghai Province. In the southern TP, the number of ice days is very little because of the low latitude, resulting to the feeble trend magnitudes compared with the high latitude. Frost days (FD) has also generally decreased over the analysis period, at a regional rate of −4.32 d/decade, significant at the 0.001 level. About 39% of stations show a statistically significant decreasing trend and stations with larger trend magnitudes are distributed in the southern and northwestern TP.

[17] Table 5 shows the proportion of stations where trends in indices are of a particular relative magnitude. About 85% of stations show larger trend magnitudes in TN10 than TX10. For TXn and TNn, 75% of stations have greater trend magnitudes in TNn. About 70% of stations show larger trend magnitudes in FD than ID.

Table 5. Number and Proportion of Individual Stations Where the Trend in One Index Is of Greater Magnitude Than the Trend in a Seconda
IndexComparisonNumberProportion
  • a

    Abbreviations are as follows: abs, indicates that the absolute magnitudes of trends are compared; rel, indicates that the signs of trends are retained during comparison.

TX90 > TX10abs560.79
TN90 > TN10abs410.58
TXx > TXnrel340.48
TNx > TNnrel110.15
TXx > TNxrel430.61
TXn > TNnrel180.25
ID > FDabs210.30
TX90 > TN90abs160.23
TX90 > TN10abs200.28
TX10 > TN10abs110.15
TN90 > TX10abs640.90

3.1.2. Diurnal Temperature Range (DTR)

[18] Many previous studies in the TP [Du, 2001; Duan et al., 2006; X. Liu et al., 2006] show that the maximum and minimum temperatures both have increasing trends, but minimum temperature increases more rapidly than maximum temperature in the recent decades. Larger trends in minimum temperature than maximum temperature should bring declining trends in DTR. The regional trend is −0.20°C/decade with significant at the 0.001 level (Table 3), which drastically declines from 1961 to 1980. In contrast, there are comparable increases in minimum and maximum temperature since the 1980s, altering recent DTR trend (Figure 7). Compared with the global, the tendency of decrease in the study conforms to it but the rate of decline is much higher [Vose et al., 2005].

Figure 7.

Same as Figure 6, but for trends in diurnal temperature range.

[19] Approximately 83% (44% statistically significant) of stations show a decrease in DTR (Table 4), the largest DTR diminished areas, such as the northwestern and southeastern TP, are in accordance with the areas of the strongest warming (Figure 8).

Figure 8.

Same as Figure 5, but for trends in diurnal temperature range.

3.1.3. Warm Extremes (TX90, TN90, TXx, TNx, GSL)

[20] For warm extremes indices during 1961–2005 in the eastern and central TP, spatial distribution of temporal trends and the regional anomalies series are shown in Figures 9 and 10. The regional trends and number of stations with significant negative, nonsignificant, and significant positive trends are also listed in Tables 3 and 4.

Figure 9.

Same as Figure 5, but for warm extreme indices.

Figure 10.

Same as Figure 6, but for warm extreme indices.

[21] For the percentage of days exceeding the 90th percentiles (TX90 and TN90), about 49% and 85% of stations show statistically significant increasing trends, respectively. Areas in the northern and southwestern TP have larger trend magnitudes, while a few stations have decreasing trends for warm days (TX90) and mainly occur in the west of Sichuan Province. The regional trends for these two indices are 1.26 and 2.54 d/decade, respectively. About 39% and 51% of stations have statistically significant increasing trends for extreme temperatures (TXx and TNx), which areas of larger trend magnitudes are accordance with TX90 and TN90. The regional trends for these two indices also show statistically increasing trends with the rate equivalent to 0.28 and 0.25°C/decade, respectively. For TX90 and TN90, about 77% of stations have greater magnitude in TN90, and approximately 61% of stations have greater magnitude in TXx than in TNx. (Table 5).

[22] The regional trend for growing season length (GSL) is 4.25 d/decade, although the GSL is not monotonic, as the increasing trend observed during the 1960s and the 1980s onward, was reversed during the 1970s. About 39% of stations show statistically significant increasing trends, with larger values at the stations in the northern, southwestern and southeastern TP. This spatial pattern is similar to the observed in other warm indices.

3.1.4. Comparison of Warm and Cold Extremes

[23] In order to learn more about the relative changes in the daily temperature distribution, it is necessary to compare trends in warm and cold indices. Comparison between warm and cold extremes is shown in Table 5.

[24] For TX90 and TX10, about 79% of stations have larger trend magnitudes in TX90 than in TX10, and the regional trend in TX90 is more than 1.5 times that of TX10. For minimum temperature, the regional trend in TN90 (2.54 d/decade) is of greater magnitude than that of TN10 (−2.38 d/decade), but the difference is not as marked as for maximum temperature. When looking at individual stations a greater proportion (58%) of stations have higher trend magnitudes in TN90 than in TN10.

[25] For TXx and TXn, regional trend in TXn is higher than in TXx (0.30 and 0.28°C/decade, respectively), and roughly half of the stations show larger trends in TXn. The magnitude of the regional trend in TNn is more than 2.8 time that of TNx. At individual stations, about 85% of stations have greater trend magnitudes in TNn. Therefore, we can conclude that changes in some warm extremes (TN90 and TX90) seem to be larger than changes in some cold extremes (TN10 and TX10), while some warm extremes (TNx and TXx) seem to have smaller trend magnitudes than that in some cold extremes (TNn and TXn).

3.2. Precipitation

[26] In contrast to the temperature extremes, the significance of changes in precipitation extremes during 1961–2005 is low as trends are difficult to detect against the larger interannual and decadal-scale variability of precipitation in the eastern and central TP. The spatial distribution of temporal trends and the regional annual standardized anomalies series of precipitation indices are shown in Figures 11 and 12. The regional trends for precipitation indices are also listed in Table 3.

Figure 11.

Same as Figure 5, but for trends in precipitation indices.

Figure 12.

Regional annual standardized anomalies series relative to 1961–2005 for precipitation indices. The smoother line is the 9-year smoothing average.

[27] For the eastern and central TP as a whole, annual total precipitation (PRCPTOT) shows positive correlations with precipitation indices except consecutive dry days (CDD) (Table 6). When looking at regional trends, only consecutive dry days (CDD) have statistically significant trend (−4.64 d/decade). Two other indices show nonsignificant decreasing trends: maximum 5-day precipitation (RX5day) and consecutive wet days (CWD) (Table 3). For these three indices, 87%, 49% and 55% of stations have decreasing trends and most of them are located in the central TP.

Table 6. The Correlation Coefficients of Precipitation Indices (n = 45, when r = ±0.29, P = 0.05)
 CDDCWDPRCPTOTR10mmR95R99RX1dayRX5daySDII
CDD1.00        
CWD0.031.00       
PRCPTOT−0.260.191.00      
R10mm−0.230.130.941.00     
R95−0.110.010.780.741.00    
R99−0.150.040.650.620.851.00   
RX1day−0.04−0.120.500.390.750.781.00  
RX5day−0.110.170.530.480.680.700.701.00 
SDII−0.10−0.050.650.760.780.660.520.561.00

[28] PRCPTOT shows larger trend magnitudes and the regional increasing trend is 6.66 mm/decade with significant at the 0.1 level, with a decreasing trend in the 1960s and increases slightly since the 1970s. About 73% of stations have increasing trends mostly occurring in the southern TP and the north of Qinghai Province while 27% of stations have decreasing trends located in the central TP.

[29] About 69% of stations for heavy precipitation days (R10mm) have increasing trends which the distributions are similar to PRCPTOT. About 87% of stations for CDD have decreasing trends and stations in the northeastern and southwestern TP have larger trend magnitudes.

[30] In additions, maximum 1-day precipitation (RX1day), average wet days precipitation (SDII), total precipitation on extreme wet days (R95 and R99) show nonsignificant increasing regional trends during the period. The proportion of stations with positive trends for these indices is 58%, 62%, 65% and 68%, respectively. These indices almost have the similar distributions that stations located in northern, southeastern and southwestern TP show a pattern of increasing trends.

[31] Figure 13 presents the regional series of the ratio between the precipitation amount on very (extremely) wet days and total precipitation. The contribution of very wet days (above the 95th percentile) to total amounts varies between 15%—24% and increases slightly over time. The trend in this ratio is 0.15%/decade (not significant at the 5% level). The contribution of extremely wet days (above the 99th percentile) to the total amounts varies between 4%—9%, and the trend (0.17%/decade) is not significant at the 5% level.

Figure 13.

Regional series (a) for the ratio between the index of precipitation falling on very wet days (R95) and total precipitation and (b) for the ratio between the index of precipitation falling on extremely wet days (R99) and total precipitation. The smoother line is as in Figure 6.

4. Discussion and Conclusions

[32] With the help of a set of widely spread descriptive indices, a better understanding of observed change in temperature and precipitation extremes is gained for the eastern and central TP during 1961–2005. For most stations, statistically significant increases in the percentage of warm nights/days and decrease in the percentage of cold nights/days are observed during the period 1961–2005, the trend magnitudes in cold/warm nights are larger than those in cold/warm days. Therefore, the (daytime) trends in maximum temperature extremes are smaller than the (nighttime) trends in minimum temperature extremes, which can be in line with the observed decrease in the DTR. The warming climate cause the number of the ice days and frost days to decrease significantly and the number of growing season length to increase significantly. These results generally agree with what has been observed in the world during the second half of the 20th century [Alexander et al., 2006; Frich et al., 2002]. For temperature extremes, the annual highest/lowest minimum temperature and maximum temperature also has statistically significant increasing trend, the magnitude in lowest of minimum temperature shows greater change. These temperature indices show spatially uniform patterns, even though the climate varies across the region. Stations in the northwestern, southwestern and southeastern TP have larger trend magnitudes, in accordance with the average warming in the regions.

[33] The TP has experienced statistically significant warming since the mid-1950s, and the linear rate of the annual mean temperature during the period 1955–1996 is about 0.16°C/decade [X Liu and Chen, 2000]. Most probably, there are two main reasons accounting for the surface warming in the TP. One is that the increasing anthropogenic greenhouse gases emissions contribute to the recent warming in the TP [Duan et al., 2006], and a model study also testifies that enhanced climatic warming in the TP due to doubling carbon dioxide [Chen et al., 2003]. The other is the change of cloud amount. The low-level cloud amount in the TP exhibits a significant increasing trend during the nighttimes, leading to the strong nocturnal surface warming, and both the total and low-level cloud amounts during daytime display decreasing trends, resulting in surface warming [Duan and Wu, 2006]. In the context of unprecedented global warming, temperature extremes show regional trends that agree with the average warming in the region, and changes in temperature extremes can be used for climate change. Analyzing the characteristics of the regional time series in the TP, it can be found that in the mid-1980s, the TP experiences a climatic jump [Niu et al., 2004], which also reflects in some temperature extremes. It is found that grain production in Qinghai Province exhibits strong correlations with the temperature, and the tree growth in Sichuan Province is closely related to the change in temperature [ X. Liu et al., 2006]. However, the influences of temperature extremes on the ecosystems are not discussed. Further works should be done to assess the aspects.

[34] Compared with change in temperature, there is no agreement yet for precipitation change in the TP which mainly contribute to the complex terrain and sparse meteorological stations [Du and Ma, 2004; Li and Kang, 2006; Wu et al., 2007]. Some experts divide the TP into nine subregions in terms of precipitation variation regimes and find that some subregions became drier but others wetter [Lin and Zhao, 1996]. Precipitation in the TP mostly happens in the summer monsoon season, and the summer precipitation in the TP is closely associated with the North Atlantic Oscillation (NAO). During the summer of low NAO index values, summer precipitation is usually above normal in the southern TP but below normal in the northern TP, and vice versa [X. Liu and Yin, 2001]. During the summer monsoon season, precipitation in the southern TP is influenced by the monsoon strength, while in the northern of TP, air masses from the Atlantic Ocean bring moisture to the TP, the dividing line which separates the regions influenced by different air masses is located around 34–35°N in the central TP [Yeh and Gao, 1979], which is in accordance with the results derived from the isotope of precipitation [Tian et al., 2007; Yeh and Gao, 1979].

[35] Changes in precipitation extremes could be detected for the variation of precipitation, although a small fraction of station trends are statistically significant for indices. There is a consistent pattern of trends in precipitation indices and the majority of precipitation indices are correlated with the annual total precipitation. Most precipitation indices exhibit increasing trends in the southern and northern TP and show decreasing trends in the central TP, which is located around 34–35°N. These suggest that the change of precipitation indices is connected with the summer monsoon and westerly, the latter is associated with the NAO. Despite the spatial and temporal variations of precipitation indices in the eastern and central TP have been examined, much work remains to be done in the future.

Acknowledgments

[36] This study is supported by the “Talent Project” of the Chinese Academy of Sciences, the National Natural Science Foundation of China (40771187, 40401054), the National Basic Research Program of China (2005CB422004), and the Sixth Framework Program Priority (036952). The authors thank the National Climate Center, China Meteorological Administration, for providing the historical climate data for this study. We are very grateful to the two anonymous reviewers for their constructive comments and suggestions.

Ancillary