1 Introduction

Against the background of durative global economic and societal development, continuous population growth and increasing urbanization, global climate warming has become a major focus of scientific research (Ahmed 2011; Habib 2011; Nordhaus 2011; Scheffran and Battaglini 2011; Liu et al. 2012). According to the IPCC Fourth Assessment Report (IPCC 2007), global air temperature has risen by 0.74 °C in the past 100 years. As a result, glaciers, the largest store of freshwater, are increasingly diminishing (Scambos 2011). Rivers that are mainly supplied by glacial meltwater are directly impacted, especially in the arid and semi-arid areas of northwestern China, where glacial meltwater is the most important natural source for maintaining the existence and development of oases in the river basins (Xu et al. 2004b; Hao et al. 2008; Li et al. 2010; Xu et al. 2010; Yu et al. 2011; Bao and Fang 2012; Ling et al. 2012). Regional acceleration of water circulation is also a result of global warming, bringing about frequent environmental disasters such as drought, rainstorms, floods, heat waves and chilling damage (Scheffran and Battaglini 2011). Consequently, hydrologic circulation processes and their response to climate change have received much attention and have become a focus of research.

Xinjiang lies in the Eurasian hinterland and belongs to the arid and semi-arid area of northwestern China. It is a long distance from the sea, includes a large part of the Gobi desert and has a fragile environment. The water vapour of Xinjiang comes mainly from the west, northwest and north. Xinjiang is characterized by high air temperature, dry climate and uneven temporal-spatial distributions of precipitation and surface water resources that are determined by unique geomorphic conditions. As a result of intensive disturbances due to global change and human activities, the climate of Xinjiang has been drastically transformed from a warm-dry to a warm-wet one, and this has caused a large increase in runoff (Xu and Wei 2004; Zhang et al. 2010). However, the farmland area and the population size increased by 63.5 % and 80.1 % from 1949 to 2011, resulting in accelerated consumption of water resources in irrigated areas, diversion of water around irrigated areas and a series of ecological and environmental problems such as severe natural vegetation degradation, desertification, biodiversity decline, secondary salinization of soil and lake withering (Bruelheide et al. 2003; Feng et al. 2005). Therefore, the coordinated growth of the regional social economy and environment has been seriously impacted. For this reason, analysis of the temporal-spatial variations of climate in Xinjiang and their impact on the regional surface runoff against the background of global warming is important for sustainable utilization of water resources and ecological preservation in the region. Previous studies (Xu and Wei 2004; Sun et al. 2010; Wu et al. 2010) have focused mostly on climate change in the whole of Xinjiang or in certain regions; the variation characteristics and temporal-spatial differences in air temperature and precipitation at each station of the study area were of less concern. Accordingly, on the basis of air temperature and precipitation data from 56 meteorological stations in Xinjiang and runoff data from 8 hydrologic stations distributed in the Tarim River Basin (southern Xinjiang) and the Manas River Basin (northern Xinjiang), the present study examined the trend and period of runoff and climate factors in the study area by using nonparametric tests and wavelet analysis and investigated the temporal-spatial variation characteristics of climate factors through geostatistics. The purpose was to understand the long-term variation rules of climate factors and runoff processes and provide a scientific basis for realizing a reasonable allocation of water resources in Xinjiang.

2 Study Area

Xinjiang (73°40′E–96°18′E, 34°25′N–48°10′N) accounts for one-sixth of the total land area of China. It has a total area of 166.04 × 104 km2 and a population of 2.181 × 107. It is also the largest provincial administrative region in China (Fig. 1). Distinguishing it from other arid regions, Xinjiang is characterized by three mountain ranges (Altai Mountains, Tianshan Mountains and Kunlun Mountains) and two basins (Junggar Basin and Tarim Basin) among the mountains. Xinjiang is divided into two regions, the north and the south, separated by the Tianshan Mountains as the boundary. High mountains intercept the water vapour headed inland, and the water vapour rises, condenses and generates precipitation in the mountain area. As a result, numerous rivers are supplied and provide water resources to the plains area. Taklamakan Desert, lying in the central section of the Tarim Basin, is dominated by migratory dunes of 100–150 m. It is the largest desert in China and the second largest desert around the world, with a total area of 32.4 × 104 km2. Gurbantunggut Desert, located in the middle of Junggar Basin, is the second largest desert in China and is dominated by fixed and semi-fixed dunes (the dunes make up 97 % of the area of the whole desert); the coverage of vegetation is 40–50 % in the fixed dunes and 15–25 % in the semi-fixed dunes.

Fig. 1
figure 1

Map of Xinjiang showing the locations of selected meteorological and hydrologic stations

Full size image

The average annual precipitation in Xinjiang is 147 mm, 23 % of the average annual precipitation of China as a whole. There are significant differences in precipitation among regions: there is more precipitation in the north than in the south, more in the west than in the east, more in the mountains than on the plains and more on the windward slopes than on the leeward slopes (Table 1). With a smaller area than southern Xinjiang, northern Xinjiang has an average precipitation of 255 mm, giving a total water resource of 1150 × 108 m3 (47 % of the total precipitation in Xinjiang). The precipitation in southern Xinjiang is 106 mm, providing a water resource of 1280 × 108 m3 (53 % of the total precipitation in Xinjiang). The average precipitation is 240 mm in the west and is 60 mm in the east. In the mountain area, the precipitation is 2048 × 108 m3, 84 % of the total precipitation in Xinjiang, and in the plains area it is 381 × 108 m3, 16 % of the total.

Table 1 Geographical coordinates, annual mean air temperature, and precipitation in the meteorological stations across Xinjiang province
Full size table

The air temperature of northern Xinjiang is higher than that of southern Xinjiang in winter, while in summer it is higher in the south than in the north. The lowest temperature is in January, and the mean temperature in Junggar Basin is below −20 °C. Fuyun County, located on the northern margin of Junggar Basin, is the coldest region in China; its absolute minimum temperature has once reached −50.15 °C. The highest temperature in Xinjiang is in July, and Turpan has the highest temperature in China; in the past 50 years, its mean air temperature has been over 33 °C, and the absolute highest temperature has once reached 49.6 °C.

The mountain area is the runoff formation area of Xinjiang, and the plains area is the runoff loss area. The water resources of northern Xinjiang are distributed mainly in Ili, Altay and Tacheng, with water quantities of 170.9 × 108 m3, 129 × 108 m3 and 61.6 × 108 m3, respectively (Chen et al. 2009). The total quantity of surface water in northern Xinjiang is 435.7 × 108 m3. Manas River is one of the largest rivers in northern Xinjiang, with a total length of 400 km and a total area of 3.35 × 104 km2; its average runoff over many years is 12.464 × 108 m3. The water resources have a relatively uniform distribution in southern Xinjiang. The water resources of Aksu, Hotan, Kashgar, Bayangol Mongol Autonomous Prefecture and Kezileshukeer are 98.6 × 108 m3, 86.2 × 108 m3, 84.4 × 108 m3, 83.2 × 108 m3 and 78.3 × 108 m3, respectively (Chen et al. 2009). Tarim River Basin is the largest river in southern Xinjiang, with a total length of 1321 km, a total area of 102 × 104 km2 and a total water quantity of 429 × 108 m3. The average natural runoff over many years is 3.983 × 1010 m3, and the non-repeated underground water quantity is 30.7 × 108 m3.

3 Data Sources

This paper collected the yearly data of air temperature and precipitation in 56 meteorological stations in Xinjiang during 1961–2007 (provided by Xinjiang Meteorological Administration). The hydrologic data were measured during 1957–2008 from the 7 hydrologic stations (Xiehela, Shiliguilank, Wuluwat, Yzmeilek, Dashankou, Tongguzlek and Kaqun) in the headstreams of the Tarim River Basin in the southern Xinjiang and Kenswat Station in the headstream of Manas River Basin in the northern Xinjiang (Table 2). The above hydrologic stations were taken as the natural runoff without any artificial drainage or irrigation, and the runoff data were offered by Tarim River Basin Administration and Manas River Basin Administration.

Table 2 Geographical coordinates and annual runoff of 8 hydrologic stations across Xinjiang province
Full size table

4 Study Methods

4.1 Mann-Kendall Monotonic Trend Test

Mann-Kendall monotonic trend test is generally considered in hypothesis testing of a climate and runoff time series, and it can detect trend without requiring the statistical distribution (e.g. normality and linearity) before conducting the test (Xu et al. 2011a). Therefore, this method is superior to the t-test in testing capacity and recommended for wide use by the World Meteorological Organization (Mitchell et al. 1966; Huang et al. 2012).

In the Mann-Kendall monotonic trend test, the null hypothesis assumes that there is no significant increasing or decreasing trend in the time series, while the time series has a significant variation trend based on the alternative hypothesis.

Time series (X 1 , X 2 , X 3 ,…, X n ) are compared in turn, and the results are recorded as sgn(θ):

$$ \operatorname{sgn}\left( \theta \right)=\left\{ \begin{array}{*{20}c}\;\;1,\quad \theta >0 \hfill \\ \;\;0,\quad \theta =0 \hfill \\ -1,\quad \theta <0 \hfill \\\end{array} \right.\ . $$
(1)

The Mann-Kendall statistic is calculated as:

$$ S=\sum\limits_{i=1}^{n-1 } {\sum\limits_{k=i+1}^n {\operatorname{sgn}\left( {x_k -{x_i}} \right)} }, $$
(2)

where, x k and x i are random variables and n is the length of the selected data sequence. The test statistic Z c is

$$ {Z_c}=\left\{ \begin{array}{*{20}c} \frac{s-1 }{{\sqrt{{\operatorname{var}(s)}}}},\quad s>0 \hfill \\ \quad \quad 0,\quad \quad s=0 \hfill \\ \frac{s+1 }{{\sqrt{{\operatorname{var}(s)}}}},\quad s<0 \hfill \\\end{array} \right.. $$
(3)

When Z c is −1.96 ≤ Z c  ≤ 1.96, the null hypothesis is accepted, which indicates that there is no obvious trend in the samples. The trend is significant at the 95 % confidence level if |Z| > 1.96 and at the 99 % confidence level if |Z| > 2.58. A positive Z c indicates that the sequence has an increasing trend, while a negative Z c reflects a declining trend (Kendall 1975).

In Mann-Kendall test, the Kendall inclination is usually used to detect the monotonic trend and indicate the variable quantity in the unit time section; the corresponding value is calculated as

$$ \begin{array}{*{20}c} {\beta =Median\left( {\frac{{x_i -{x_j}}}{i-j }} \right),} & {\forall j<i} \\ \end{array}\quad \quad $$
(4)

where 1 < j < i < n, positive β denotes an increasing trend, and negative β indicates a decreasing trend (Kendall 1975).

4.2 Morlet Continuous Wavelet Transforms

A wavelet analysis is appropriate for determining the periodicity of a time series and analysis of the phase change and periodic intensity of the time series on different time scales (Farge 1992; Deng et al. 2011). In wavelet analysis, the main periods have poor reliability if they are not subjected to a chi-square test at the 0.05 significance level (Torrence and Compo 1998). Therefore, a significance test for the periodicity of the time series was conducted in this study by employing a chi-square test and detecting significant time sections from red-noise standard spectra (Torrence and Compo 1998).

The Morlet wavelet (ψ t ) represents a wave modulated by a Gaussian function with the expression

$$ \psi (t)={\pi^{{-{1 \left/ {4} \right.}}}}{e^{ict }}{e^{{{{{-{t^2}}} \left/ {2} \right.}}}}, $$
(5)

where i is the unit of the imaginary number, t is the nondimensional time parameter, and c is a constant number (Farge 1992; Torrence and Compo 1998).

The continuous wavelet transformation of a discrete signal f (t) with a Morlet wavelet (ψ t ) is

$$ {W_f}\left( {a,b} \right)=\frac{1}{a}\int\limits_R {f(t){\psi^{*}}\left( {\frac{t-b }{a}} \right)} dt, $$
(6)

where W f (a, b) is the transformation coefficient, a and b are a scale parameter and translation parameter, respectively, t is the time scale and ψ* is the complex conjugate.

The wavelet variance is used to determine the main periodicity on different scales of the time domain, and is expressed as

$$ Var(a)=\int_{{-\infty}}^{{+\infty }} {{{{\left| {W_f \left( {a,b} \right)} \right|}}^2}db}, $$
(7)

where, Var(a) is the wavelet variance and a the time scale. The peak values of wavelet variance correspond to the main periods of different time scales (Farge 1992; Torrence and Compo 1998).

5 Results and Analysis

5.1 Variation Trend and Temporal-Spatial Distribution of Air Temperature

5.1.1 Temporal-Spatial Distribution of Variation Trend of Air Temperature

According to previous work (Xu and Wei 2004), the air temperature of western China rose from the late 19th century to the early 20th century, reached a maximum in the 1940s, declined to a minimum in the early 1970s and has been in a steady uptrend since then. Based on the air temperature data from 56 meteorological stations for the period 1961–2007, this study examined the temporal-spatial distribution of air temperature variation trend in Xinjiang by using the Mann-Kendall monotonic trend test and geostatistics (Figs 2 and 3).

Fig. 2
figure 2

Spatial patterns of the Mann-Kendall monotonic trend test results of air temperature in Xinjiang

Full size image
Fig. 3
figure 3

Spatial patterns of air temperature slopes in Xinjiang

Full size image

According to the Mann-Kendall monotonic trend test (Fig. 2), the air temperature of Beitaishan (BTS) in northern Xinjiang increased significantly between 1961 and 2007, as indicated by a test statistic that reached a significance level of 0.05 (Z 0 =2.53 > Z 0.05 =1.96). The rest of the stations all reached the very strong significance level of 0.01 (Z 0 > Z 0.01 = 2.58), which meant that the air temperature was in a very significant uptrend. In the Tianshan Mountains, Wuqia (WQ) and Wrumqi (WRMQ) both had increasing trends that were significant at the test level of 0.05 (1.96 < Z 0 < 2.58), while the increasing trend at Bayinbluk (BYBLK) is not statistically significant (Z 0 = 1.44 < Z 0.05). The air temperatures at the other stations all rose noticeably at the significance level of 0.01. In southern Xinjiang, the temperature at Kuche (KC) had a downtrend, but it was not statistically significant (Z 0 = −1.09). Alar (ALR) and Yutian (YT) showed increasing trends that were not statistically significant, but the other stations showed increases that were highly significant (at the test level of 0.01).

Figure 3 shows the temporal-spatial distribution characteristics of the change rate (β) of air temperature in Xinjiang obtained from the Mann-Kendall monotonic trend test. The change rates decline from the Taklimakan hinterland northward to the Tianshan Mountains through KC and rise to a certain degree in other regions. The air-temperature increase rate in northern Xinjiang was between 0.014 °C/year and 0.082 °C/year and mostly in the range 0.035–0.047 °C/year. The lowest increase rates, making up a center of low values, occurred in Beitashan (BTS) and Qitai (QT), where the values were 0.020 °C/year and 0.024 °C/year, respectively. The highest rates, making up a center of high values, were in Fuyun (FY) and Tuoli (TL), where the values were 0.060 °C/year and 0.061 °C/year, respectively. On the whole, the air temperature increase rate in northern Xinjiang rose gradually from the southeast to the northwest. The minimum value of air-temperature increase rate in the Tianshan Mountains was in BYBLK (0.016 °C/year), followed by Wenquan (WQ), Krmay (KRMY), Dabancheng (DBC) and Kumishi (KMS). However, the increase rate became higher from the rim of five stations to the center of them. The maximum of increase rate was in Balikun (BLK), with a value of 0.062 °C/year, and the increase rate decreased toward Dabancheng (DBC). In the Turpan Basin of eastern of Xinjiang, the air-temperature increase rate declined gradually from Turupan (TRP) in the west to Hami (HM) in the east. The minimum change rate of air temperature in southern Xinjiang appeared in Alar (ALR) (0.006 °C/year), followed by Yutian (YT) with a rate of 0.008 °C/year. The maximum was 0.034 °C/year at Hotan (HT) and Langan (LG). The air-temperature annual change rate in southern Xinjiang increased gradually from Tashikurgan (TSKRG), the Taklamakan hinterland and Ruoqiang (RQ) to the surrounding regions.

5.1.2 Temporal-Spatial Distribution of Periodic Variation of Air Temperature

Wavelet analysis is widely used to study the periodic characteristics of climate change, and it will detect periodic variation on different time scales (Xu et al. 2010, 2011b). The wavelet variance was put into a chi-square test at a confidence level of 95 % (Torrence and Compo 1998), and the spatial distribution of significant periodic variations of air temperature in Xinjiang during 1961–2007 was obtained (Fig. 4).

Fig. 4
figure 4

Spatial patterns of air temperature periods in Xinjiang

Full size image

According to Fig. 4, the air-temperature period in Xinjiang was mainly 4 years. In northern Xinjiang, the significant period was 4 years in the northwest and increased to 6–7 years in the southeast. Fuyun, (FY) in the northeast, also had a significant 4-year period. In the western area of the Tianshan Mountains, with the exception of Wenquan (WQ), with a period of 7 years, the stations all presented 4-year periods. The air temperature in the middle of the Tianshan Mountains had a 4-year period, whereas in the east it had a period of 6–7 years. In southern Xinjiang, the periods at Turgat (TRGT) and Tashikurgan (TSKRG) in the western mountain area were 6 years and 7 years, while the period in the eastern plains area decreased to 4 years. In addition, Alar and Tieganlik (TGLK) in the upper and lower reaches of the Tarim River Basin had significant periods of 9 years and 7 years.

5.2 Variation Trend and Periodic Temporal-Spatial Distribution of Precipitation

5.2.1 Temporal-Spatial Distribution of Variation Trend of Precipitation

Xinjiang is located in the Eurasian hinterland and far from the sea, so the quantity of vapour transport is restricted by not only the vapour resource but also the impact of terrain on air circulation. The southwest monsoon from the Indian Ocean is obstructed by the Himalaya Mountains and the Tibet Plateau, which make it difficult for water to reach Xinjiang. The southeast monsoon and tropical storms from the Pacific Ocean make up the meridional circulation and mainly affect the east of China because of the obstruction of the mountains and plateau. Xinjiang lies in the mid-latitudes of the Northern Hemisphere and prevailing westerlies, and the major source of water vapour in Xinjiang is airflow from the Atlantic Ocean, followed by dry-cold airflow from the Arctic Ocean (the quantity of water vapour is 25–33 % of that of the west wind airflow). However, because it is affected by the airflow circulation described above, the temporal-spatial distribution of the variation trend of precipitation in Xinjiang needs to be discussed (Fig. 5).

Fig. 5
figure 5

Spatial patterns of the Mann-Kendall monotonic trend test results of precipitation in Xinjiang

Full size image

Figure 5 shows that the precipitation at all stations in the Altai Mountains of northern Xinjiang had a significant increasing trend at the test level of 0.05, and the precipitation reached a very high significance level at the test level of 0.01 at Fuyun (FY). Furthermore, the test statistic of precipitation at Tuoli (TL) accepted the null hypothesis (Z0 = −0.08), so the precipitation at that station decreased slightly. The precipitation at Alashankou (ALSK) and Jinghe (JH) rose significantly at the level of 0.05, while the increasing trends at other stations were not notable. In the Tianshan Mountains, BYBLK presented a slight declining trend, while at the other stations precipitation rose. Specifically, Wenquan (WQ), Yining (YN), Kenswat (KSWT), Wrumqi (WRMQ) and Dabancheng (DBC) showed the most significant increases. In the Turpan Basin, Shisanjianfang (SSJF) had a statistically insignificant decreasing trend, while the others presented increases. In the Tarim Basin of southern Xinjiang, the precipitation had a more significant increasing trend on the northern margin than on the southern margin. Only Ruoqiang (RQ), on the southern margin, had an extremely significant increasing trend at the test level of 0.01.

Figure 6 shows the temporal-spatial distribution characteristics of the annual change rate (β) of the precipitation in the Mann-Kendall monotonic trend test. In northern Xinjiang, the annual change rate of precipitation in the northern mountain area was larger than that in the southern plains area, and a negative change rate occurred in the southeast region and around Tuoli (TL). In the Tianshan Mountains, a relatively large increase rate of precipitation existed at Wenquan (WQ) and Yining (YN) in the northwest, Wrumqi (WRMQ) and Kenswat (KSWT) in the middle and Wuqia (WQ) in the east. The increase rate at Wrumqi (WRMQ) reached an especially high maximum of 3.209 mm/year. The precipitation change rate in northern Xinjiang increased from the low-value center (Zhaosu (ZS) and BYBLK) to the high-value area in the surrounding vicinity. The precipitation change rate in southern Xinjiang was negative in the Taklamakan hinterland, from which the rate increased gradually to the surrounding regions. The maximum increase rate (2.073 mm/year) occurred at Ahqi (AHQ), followed by Baicheng (BCH) with a change rate of 1.548 mm/year. On the whole, the regional precipitation had a larger change rate in the north than in the south; the rate in the north was between 0.910 and 2.642 mm/year, and the rate in the south was between 0 and 0.910 mm/year.

Fig. 6
figure 6

Spatial patterns of precipitation slopes in Xinjiang

Full size image

5.2.2 Temporal-Spatial Distribution of Periodic Change of Precipitation

Figure 7 shows the temporal-spatial distribution characteristics of the periodic variation of precipitation in Xinjiang for the period 1961–2007. At the significance level of 0.05, there was a 4-year period at four of the meteorological stations, a 5-year period at three stations and a 6-year period at six stations. There was an 8-year period at Jinghe (JH) and Wusu (WS) and a 9-year period in the west and at Kenswat (KSWT). The significant periods were 3–4 years in the middle of the Tianshan Mountains, 7–9 years in the west and mainly 3–7 years in the east. The shortest period (3 years) appeared at Hongliuhe (HLH), Kumishi (KMS) and Dabancheng (DBC), and the longest period (9 years) was at Wenquan (WQ). In Turpan Basin, Hami (HM) in the east had a 4-year period, Shisanjianfang (SSJF) in the middle was dominated by a 6-year period, and at Turupan (TRP) in the west the length increased to 8 years. In southern Xinjiang, KC had the shortest period (3 years). In the west, the period increased to 8 years, and Aksu (AKS) had the longest period (10 years). In the western mountain area the period was 4–5 years. The precipitation in the eastern Tarim River Basin had periods of 3–4 years, but in Luntai (LT) and Kurle (KRL), the periods were 6 years and 10 years. The precipitation at Langan (LG), in the southern mountain area, had a 9-year period.

Fig. 7
figure 7

Spatial patterns of precipitation periods in Xinjiang

Full size image

According to the above analysis and a previous study (Xu et al. 2010), climate change in Xinjiang is a complicated nonlinear process influenced by a combination of factors such as global climate change, regional geographic location, topography, human activities etc. Therefore, the mechanisms of change in regional climate variables (e.g. air temperature and precipitation) are still not obvious and should be investigated more deeply in future work.

5.3 Variation Characteristics of Runoff Change in Xinjiang

5.3.1 Variation Trend of Surface Runoff in Xinjiang

According to existing studies (Chen et al. 2009; Zhang et al. 2010; Ling et al. 2011; Xu et al. 2011a), the runoff change in Xinjiang is intimately related to air temperature and precipitation. Tarim River Basin and Manas River Basin, on the north and south slopes of the Tianshan Mountains, are typical representatives of pure-dissipation continental river basins in arid and semi-arid areas. The two basins have similar characteristics of runoff generation: the main stream has no runoff yield, and the runoff is mainly supplied by precipitation and glacier meltwater in the mountain area. Therefore, the annual runoff in the two basins is especially sensitive to the impact of climate change (Ling et al. 2011; Xu et al. 2011a). In the past 50 years, because of overexploitation of water resources in the upper reaches of the two river basins, biodiversity has declined, desertification has been aggravated, and the environment has deteriorated in the lower reaches (Xu et al. 2007; Ling et al. 2010). Accordingly, discussion of the trend and periodicity of runoff in the Tianshan mountain area against the background of global change is meaningful in developing reasonable water-utilization plans and reducing environmental problems resulting from climatic and hydrological fluctuations. Therefore, this study made use of the annual runoff data from eight hydrologic stations in two headstream mountain areas during the period 1957–2008 and analysed the variation trend of runoff at these stations (Table 3).

Table 3 Mann-Kendall monotonic trend test results of annual runoff in Xinjiang
Full size table

According to Table 3, the annual runoff at Dashankou (DSK), Shiliguilank (SLGLK), Xiehela (XHL) and Yzmeilek (YZMLK) in the Tarim River Basin had significant increasing trends at the test level of 0.05, while Kaqun (KQ) and Wuluwat (WLWT) had slight increasing trends. The test statistic for Tongguzlek (TGZLK) is −0.21, indicating that the local annual runoff was in a slight downtrend. Consequently, the increasing trend of runoff was larger in the northern Tarim River Basin than in the southern, the main reason being that the west and north are the sources of water vapour in Xinjiang, so there is a larger precipitation supply in the north than in the south. The spatial difference in runoff between the north and south is remarkable. In the Manas River Basin, the test statistic for Kenswat (KSWT) was 3.95 (Z0 > Z0.01 =2.58). Precipitation and air temperature both increased significantly, so the local runoff showed a highly significant increasing trend at the test level of 0.01.

5.3.2 Periodic Variation of Surface Runoff

Under the combined impacts of periodic variations of precipitation and air temperature in Xinjiang, the periodic variation of runoff can be analyzed by using wavelet analysis. The significance of wavelet variance is tested through the chi-square test in wavelet analysis (Fig. 8). However, previous studies in Xinjiang (Xu et al. 2010, 2011b) have not employed a significance test for runoff periodicity. Wavelet analysis is characterized by timeliness that limits certain sections. Therefore, the significant sections of runoff are tested by the specific value of the wavelet power spectrum to the overall red noise spectrum at the confidence level of 95 % (Torrence and Compo 1998), and the periodicity is significant when the specific value is more than 1 (Fig. 8).

Fig. 8
figure 8

Wavelet analysis of annual runoff in Xinjiang from 1957 to 2008

Full size image

As shown in Fig. 8, there existed a significant 5-year period at Kenswat (KSWT) in the Manas River Basin and at Dashankou (DSK) in the Tarim River Basin during the period 1957–2008. Figure 9 indicates that in a 5-year period, Kenswat (KSWT) had a remarkable high-frequency oscillation in 1965–1967, 1973–1982, 1991–1993 and 2004–2007; Dashankou (DSK) had one in 1957–1971 and 1998–2004. Xiehela (XHL) and Tongguzlek (TGZLK) had a significant 6-year period, and the period of runoff was significant at Xiehela (XHL) during 1974–1978 and 1989–2000 and at Tongguzlek (TGZLK) during 1966–1984, 1990–1997 and 2002–2007. Shiliguilank (SLGLK) had a significant 7-year period and showed marked periodic variation in 1967–1975 and 1993–1999. A significant 8-year period appeared at Yzmeilek (YZMLK) and Wuluwat (WLWT), and the periodic change at Yzmeilek (YZMLK) was remarkable in 1963–1968, 1982–1983 and 2002–2007 and at Wuluwat (WLWT) in 1960–1968 and 1973–1984. Kaqun (KQ) had the longest significant period of 9 years and had significant periodic oscillation during 1958–1967. Overall, there were significant periods of 5–9 years in those rivers in Xinjiang, and they were coordinated well with the air temperature (significant periods of 4–9 years) and precipitation (significant periods of 3–10 year). The periodic variations of the surface runoff in the two basins in Xinjiang resulted from the combined actions of additive and counteracting effects in different periods and sections of precipitation and air temperature.

Fig. 9
figure 9

Significant time sections of significant periods for annual runoff in Xinjiang

Full size image

6 Conclusions

From analysis of the variation trend and the periodic temporal-spatial changes in air temperature, precipitation and surface runoff in Xinjiang over the past 50 years, the following conclusions can be drawn:

  1. (1)

    The air temperatures at all stations in northern Xinjiang had significant increasing trends. In the Tianshan Mountains, BYBLK had only a slight increasing trend in air temperature, while the other stations all had significant increases. The maximum increase rate occurred in Balikun (BLK), and the increase rate slowed in the direction of Dabancheng (DBC). In southern Xinjiang, the air temperature in KC had only a slight downtrend; the uptrends in air temperature at Alar (ALR) and Yutian (YT) were not statistically significant, while those at the other stations were significant. The annual change rate of air temperature increased gradually from Tashikurgan (TSKRG), the Taklamakan hinterland and Ruoqiang (RQ) to the surrounding regions.

  2. (2)

    In the air temperature record of northern Xinjiang, the significant period was 4 years in the northwest and became 6–7 years in the southeast. A 4-year period also appeared at Fuyun (FY). In the Tianshan mountain area, Wenquan (WQ) in the western high mountain area had a 7-year period, while the others all had 4-year periods. The period was 4 years in the middle of the Tianshan Mountains and 6–7 years in the eastern part. In southern Xinjiang, Turgat (TRGT) and Tashikurgan (TSKRG) in the western mountain area had 6-year and 7-year periods, while the eastern plain area had 4-year periods. In addition, Alar (ALR) in the upper reaches of the Tarim River had a significant 9-year period, and Tieganlik (TGLK) in the lower reaches had a 7-year period.

  3. (3)

    In northern Xinjiang, precipitation in the Altai Mountains showed significant increasing trends. Tuoli (TL) had a statistically insignificant decreasing trend, Alashankou (ALSK) and Jinghe (JH) both had marked increasing trends, and the rest of the stations had increasing trends that were not statistically significant. The precipitation at BYBLK in the Tianshan mountain area had a decreasing trend that was not statistically significant, while the other stations in the area showed increasing trends. As for southern Xinjiang, the northern rim of the Tarim River Basin had a more significant increasing trend than did the southern rim.

    As for the change rate of precipitation in northern Xinjiang, in the northern mountain area it was larger than it was in the southern plain area, and a negative change rate occurred in the southeast and in the area around Tuoli (TL). In the Tianshan mountain area, the increase rate of precipitation increased gradually to the high-value area from the low-value area (ZS and BYBLK). The northern region has a larger change rate than the southern in southern Xinjiang.

  4. (4)

    In northern Xinjiang, there existed significant 4-year periods at four weather stations, 5-year periods at three stations, 6-year periods at six stations, 8-year periods mainly at Jinghe (JH) and Wusu (WS) and 9-year periods in the western region and at Kenswat (KSWT). The middle of the Tianshan mountain area had significant periods of 3–4 years, the west had longer periods of 7–9 years, and the east was dominated by periods of 3–7 years. In southern Xinjiang, KC had the shortest period (3 years), the significant periods in the west were mainly 8 years, Aksu (AKS) had the longest period (10 years), and periods of 4–5 years dominated in the western mountain area. The eastern Tarim Basin had significant periods mainly of 3–4 years, but Luntai (LT) and Kurle (KRL) had 6-year and 10-year periods. Additionally, Langan(LG) in the southern mountain area had a 9-year period.

  5. (5)

    In the Tarim River Basin, the runoffs at Dashankou (DSK), Shiliguilank (SLGLK), Xiehela (XHL) and Yzmeilek (YZMLK) all showed apparent increasing trends; Kaqun (KQ) and Wuluwat (WLWT) had increases that were not statistically significant; and Tongguzlek (TGZLK) showed a slight downtrend. In the Manas River Basin, the increasing trend at Kenswat (KSWT) was highly significant. Tarim River Basin and Manas River Basin both had significant periods of 5–9 years.

  6. (6)

    Because of the significant periodic nature of surface runoff in Xinjiang, water resource utilization should be managed in accordance with the observed fluctuation periods. In addition, it is very important to require the local administrative department to institute a better system for early warning of drought, and corresponding measures should be taken to achieve reasonable allocation of water resources and coordinated development of the environment and social economy.