CORRELATION AND MULTI-TIMESCALE CHARACTERISTICS OF THE LANCANG RIVER'S MONTHLY FLOW VARIATIONS IN YUNNAN (CHINA)

CORRELATION AND MULTI-TIMESCALE CHARACTERISTICS OF THE LANCANG RIVER’S MONTHLY FLOW VARIATIONS IN YUNNAN (CHINA): He, D. et al.

Role of Water Sciences in Transboundary River Basin Management, Thailand, 2005 33

CORRELATION AND MULTI-TIMESCALE CHARACTERISTICS OF THE LANCANG RIVER’S MONTHLY FLOW VARIATIONS IN YUNNAN (CHINA)

Daming He 1), Weiyong You 2) and Darrin Magee 3)

1) Asian International Rivers Center,

Yunnan University, Kunming, 650091 China 2) School of Resources, Environment, and Earth Science,

Yunnan University, Kunming, 650091 China 3) Department of Geography

University of Washington, Seattle, WA 98195-3550 USA

E-mail of Corresponding author: [email protected]

ABSTRACT

This paper uses correlation analysis and wavelet transforms to investigate the correlation and multi-timescale characteristics of the Lancang River’s monthly flow variations in southwestern China’s Yunnan Province. The results show that flow variations in the lower reach of the Lancang not only reflect those of the upper reach, but also (and more significantly) reflect the influence of climate changes over the whole of Yunnan. The monthly variations in the Lancang’s flow in Yunnan reveal a variety of features over multiple timescales. There may exist 4- and 53-months characteristic timescales for monthly climate change over the regions of the upper and lower reaches Lancang. Flow changes at the 7-month, 30-month, and 74-month characteristic timescales downstream at Yinjinghong are due to flow variations upstream at Jiuzhou, but changes at the 14- and 21-month characteristic timescales of Yinjinghong are due to the monthly climate changes over Yunnan. For the smaller characteristic timescales, the time phases of the flow’s variations influenced by the monthly climate changes over Yunnan equal or lag behind the time phases of flow variations in the upper reaches of the Lancang. For larger characteristic timescales, the time phases of flow variations influenced by monthly climate changes over Yunnan lag behind the time phases of flow’s variations in the upper reaches.

INTRODUCTION

The Lancang-Mekong River is an important north-south river in China and mainland Southeast Asia, often called the “Danube of the East.” The river originates in the Tangula Mountains of China’s Qinghai Province, and from north to south flows through Qinghai, Tibet, and Yunnan. After leaving Yunnan through the southern prefecture of Xishuangbanna, the Lancang (now called the Mekong) then flows through Myanmar, Laos, Thailand, Cambodia, and Vietnam, emptying into the sea to the west of Ho Chi Min City. The Lancang-Mekong is the only river in Southeast Asia to cross six countries. Rational and equitable use of its water resources, along with protection of its water ecology, not only impacts sustainable development goals in China’s western regions, but also China’s relations with its downstream neighbors, prospects for regional international cooperation, and sustainable development goals for the entire Mekong region (Chen and He, 2001; He et al., 1999; Tang, 1999). Over the past 40 years of cooperative research and development in the region, rational allocation and use and coordinated management of this transboundary water resource has been of primary concern (Chen and He, 1999; He et al., 1999). In recent years, alongside growing economic cooperation in the so-called Greater Mekong Subregion, the creation of the China-ASEAN Free Trade Area, and the China-Japan-Korea ASEAN+3 Regional Cooperation, research on changes of and impacts on the hydrology and water resources of the Lancang-Mekong watershed has become a hot topic for research (Chen and He, 2000; He and Lin, 1997; You, 2001). As a result, understanding the correlation and multiple timescale characteristics of changes in the Lancang’s monthly flow rates in Yunnan becomes an important theoretical basis for rational use and protection of the



river’s water resources, as well as an important scientific basis for deciding transboundary water allocation, modeling the impact of cascade hydropower development, and eliminating international misunderstanding. Factors directly influencing Lancang flow rates include climate change within the watershed, and global climate change impacts also exhibit clear patterns when viewed over multiple timescales (Chen et al., 2002; Mann and Park, 1996; Venegas and Mysak, 2000; White et al., 1997; Zhang et al., 1998). Thus understanding the correlation and multiple timescale characteristics of changes in the Lancang’s monthly flow rates means understanding such changes from the perspective of physical processes, and will serve as a foundation for better assessment and predictions of future trends for the Lancang river. Relying on sampled data from the Lancang River in Yunnan, this paper uses correlation analysis and wavelet transforms to investigate the correlation and multi-timescale characteristics of the Lancang River’s monthly flow variations in southwestern China’s Yunnan Province.

DATA AND METHODS

Data used for this research included the following: (1) observed monthly flow data from the monitoring station at Jiuzhou (25.47°N, 99.13°E) on the upper Lancang from January 1955 to December 2001; (2) observed monthly flow data from the monitoring station at Yinjinghong (22.02°N, 100.47°E) on the lower Lancang from January 1956 to December 2001; (3) observed monthly flow data from the monitoring station at Nanla River (21.26°N, 101.33°E) on the tributary of the lower Lancang from January 1966 to December 2001. Details of the analytical methods can be found elsewhere (Huang, 2000). Following a continuous wavelet transform method (You, 1998), the wavelet transform of a time-series function )(tf can be defined as

∫+∞

∞−

∗ ′−==′ dt

atttf

atfWatf )()(1)]([),(

~21 ψ (1)

where ),(~

atf ′ is the wavelet coefficient, W is the wavelet transform operator, t is time expressed in months, t ′ corresponds to the time shift of the wavelet’s location, a is a scale elasticity variable for defining the wavelet time width, ∗ψ is the mother wavelet, the complex conjugate ofψ . Here, we use the Morlet mother wavelet (You, 1998; You et al., 1999)

2/2)( tti eet −Ω=ψ (2)

This is a plane wave of vector Ω adjusted by a unit-width Gaussian envelope. In this paper, we let π2=Ω . For a weekly oscillation tie ω , since its wavelet transform coefficient is given as

tiati eeaeW ′−Ω−= ωωω π 2/)( 2

2][ (3)

explains that for a fixed scale elasticity variable a , when the full frequency of this weekly oscillation is a/Ω=ω , its wavelet transform coefficient will have its greatest module. As a result, for a fixed time

period, if we define the scale elasticity variable a such that it satisfies

ε

ε

±

±>

max

max

max

max ]()(

a

af

a

af (4)

where f is the average wavelet transform coefficient module over the time period, and ε is any small

nonzero number. From the scale elasticity variable maxa can identify the corresponding real-space characteristic timescale as

Ω= /maxaLc π (5) We define the time wavelet transform coefficient function for characteristic timescale cL as

∫+∞

∞−

∗=

′−==′ dt

atttf

afWtf aa )()(1][)(ˆ

maxmaxmax

ψ (6)

where )(ˆ tf ′ is the time wavelet transform coefficient for characteristic timescale cL . Clearly, the wavelet transform coefficient )(ˆ tf ′ reflects the time evolution characteristics and rules of the time series data )(tf under the characteristic timescale cL .



CHANGES IN LANCANG RIVER ANNUAL FLOW Fig. 1 shows the trend lines for changes in annual flow rates at Jiuzhou on the upper Lancang, Yinjinghong on the lower Lancang, as well as the difference between the two. As can be seen, even though the changes in annual flows for Yinjinghong and those for the upstream station at Jiuzhou are similar, the trend line for annual flows at Yinjinghong is much more complex than that at Jiuzhou. This indicates that the changes in downstream annual flows at Yinjinghong not only reflect changes in upstream flows as observed at Jiuzhou, but also the effects of regional climate change. Simply put, the difference between changes in annual flow at Yinjinghong and Jiuzhou results from annual climate change in Yunnan. Fig. 1 also shows that the greatest annual flow at Jiuzhou occurred in 1962, when it was recorded as 41.06 billion m3, whereas the smallest annual flow occurred in 1973, recorded as 22.17 billion m3. The 47-year average annual flow for Jiuzhou is 30.30 billion m3. For the downstream station at Yunjinghong, the greatest annual flow of 78.74 billion m3 occurred in 1966, whereas the smallest of 41.69 billion m3. The 46-year average annual flow for Yinjinghong is 57.03 billion m3. The greatest difference between annual flow at Yinjinghong and Jiuzhou occurred in 1966, at 42.98 billion m3, while the smallest occurred in 1988 at 17.50 billion m3, and the 46-year average difference between the two stations is 26.88 billion m3. It is evident from these numbers that the changes in annual flows between brought about by annual climate change in Yunnan outweigh the changes in annual flows at the upstream station at Jiuzhou, indicating the significance of impacts of climate change in Yunnan on annual flows in the Lancang.

Fig. 1: Changes in annual flow volume on the Lancang River in Yunnan Province Units: 100 million m3

(Top: Jiuzhou; Middle: Yunjinghong; Bottom: Diff-YJ)

CORRELATION OF CHANGES IN LANCANG RIVER MONTHLY FLOW

In order to eliminate the influence of annual oscillation of monthly flows, Fig. 2 presents standardized monthly time series of monthly flow data for Jiuzhou, Yunjinghong, and the difference between the two stations. As shown in Fig. 2, there are clear correlations at several levels. In order to quantitatively explain these correlations, Table 1 shows the correlation coefficients for the three standardized monthly time series.



The correlations between these two stations and the monthly flows observed at the Manlasa station on the Nanla River, a lower Lancang tributary, are also shown. According to Table 1, samples for the 552 months between January 1956 and December 2001, the correlation coefficient for Yinjinghong and Jiuzhou is 0.673, while that for Yinjinghong and Diff-YJ (the difference obtained by subtracting Jiuzhou values from Yinjinghong values) is higher at 0.795. This clearly demonstrates that monthly flow changes at the downstream station at Yinjinghong not only reflect changes upstream at Jiuzhou, but also even more clearly reflect the influence of monthly climate change in Yunnan on monthly flows at Yunjinghong. In addition, the correlation coefficient for Diff-YJ and Jiuzhou is 0.140 (α=0.01), indicating a clear interrelation between monthly climate change within Yunnan and monthly climate change in the region surrounding the upper Lancang (within Yunnan). From Table 1 it can also be seen that for the 432 monthly time samples from January 1966 to December 2001, the correlation coefficient between Manlasa and Jiuzhou was -0.080, between Manlasa and Yinjinghong 0.274, and Manlasa and Diff-YJ 0.446. Of these, the latter two correlations hold for α=0.01. This demonstrates that localized climate change within Yunnan only influences Lancang monthly flow changes within a limited geographic range, and that such localized climate changes is also influenced by larger climate change patterns in Yunnan Province.

Fig. 2: Standardized anomaly time series of Lancang Monthly flow variations in Yunnan

（Black=Jiuzhou; Red=Yunjinghong; Green= Diff-YJ）

This correlation analysis confirms that the changes in the Lancang’s monthly flow rate brought about by climate change within Yunnan exceed those changes brought about by upstream monthly flow changes at Jiuzhou, indicating the significant influence of climate change within Yunnan on Lancang monthly flows.

MULTI-TIMESCALE CHARACTERISTICS OF LANCANG MONTHLY FLOW CHANGES Using the method in Eq. (1) to perform wavelet transform on the time series for Jiuzhou, Yunjinghong, and Diff-YJ in Fig. 2, we obtain the wavelet transform coefficients modules and real components shown in Fig. 3 and 4. In both figures, the vertical axis represents timescale in months, while the horizontal axis



represents time change in months. From the figures it is clear that monthly flow changes in all three cases exhibit clear multi-timescale characteristics. The wavelet transform coefficient modules in Fig. 3 reflects the time evolution characteristics of capacity density for monthly flow changes over various timescales on the Lancang within Yunnan. The real component of the wavelet transform coefficient in Fig. 4 reflects the time oscillation characteristics of flow magnitude for monthly flow changes over various timescales. Table 1: Correlation coefficients for monthly flows on upper Lancang and tributaries in Yunnan

Jiuzhou Yunjinghong Diff-YJ Nanla R. (Manlasa)

Jiuzhou 1.000 0.673 0.140 -0.080

Yunjinghong 0.673 1.000 0.795 0.274

Diff-YJ 0.140 0.795 1.000 0.446

Nanla R. (Manlasa) -0.080 0.274 0.446 1.000

Fig. 3: Mode of wavelet transform coefficients for Lancang monthly flow changes (Top: Jiuzhou; Middle: Yunjinghong; Bottom: Diff-YJ)

In order to further explain the multi-timescale characteristics of Lancang flow changes in Yunnan, we now use the method in Eq. (4) and in Fig. 5 provide the timescale distribution of wavelet transform coefficients time mean modules for monthly flow changes for Jiuzhou, Yunjinghong, and Diff-YJ. From Fig. 5 it can be seen that the characteristic timescales for monthly flow changes at Jiuzhou are 4 months, 7 months, 15 months, 30 months, 53 months, and 74 months. For Yunjinghong, they are 4 months, 7 months, 14 months, 21 months, 30 months, 53 months, 74 months, and 106 months. Finally, for the difference in flows between the two stations, characteristic timescales are 4 months, 14 months, 21 months, 38 months, 53 months, and 100 months. The following are evident from such characteristic timescales: (1) since Jiuzhou, Yunjinghong, and Diff-YJ all have characteristics timescales of 4 and 53 months, we can expect that monthly climate change in the upper Lancang region in Yunnan and for Yunnan overall might have characteristic time cycle changes of 8 months and 106 months; (2) flow changes occurring over the 7-,



30-, and 74-month characteristic timescales for Yinjinghong are primarily caused by the monthly flow changes upstream at Jiuzhou; and (3) flow changes occurring over the 14- and 21-month characteristic timescales for Yinjinghong are the result of monthly climate change patterns within Yunnan.

Fig. 4: Real component of wavelet transform coefficients for Lancang monthly flow changes

(Top: Jiuzhou; Middle: Yunjinghong; Bottom: Diff-YJ)

Fig. 5: Timescale spectrum of the time mean modules of wavelet transforms of monthly variations for the Lancang River flow in the region of Yunnan(Top: Jiuzhou; Middle: Yunjinghong; Bottom: Diff-YJ)



TIME PHASE CHARACTERISTICS OF LANCANG MONTHLY FLOWS In order to further analyze the time phase characteristics of Lancang monthly flows in Yunnan over characteristic timescales, we apply the method in Eq. (6) to the characteristic timescales of Jiuzhou, Yunjinghong, and Diff-YJ. In Fig. 6 and 7 we have shown the wavelet transform coefficient phases for the standard month time series for Jiuzhou, Yunjinghong, and Diff-YJ 4- and 53-month timescales. From the two figures, we can see that for the 4-month characteristic timescale, the time phase change for Diff-YJ is generally equivalent to or lagging behind that of Jiuzhou. For the 53-month characteristic timescale, the time phase change for Diff-YJ, however, lags behind that of Jiuzhou for its entirety. Thus we can assert that for smaller characteristic timescale changes, changes in Lancang monthly flows caused by monthly climate change within Yunnan generally correspond to or lag behind changes in monthly flows in the upper reach of the Lancang in Yunnan. Meanwhile, for larger timescales, changes in Lancang monthly flows caused by monthly climate change within Yunnan lag behind changes in monthly flows upstream.

Fig. 6: Wavelet transform coefficient phase for 4-month characteristic timescale (Black=Jiuzhou; Green=Diff-YJ)

CONCLUSIONS

This paper has used correlation analysis and time wavelet transform analysis to study correlation characteristics and multi-timescale characteristics of changes in monthly flows of the Lancang River in Yunnan Province. Principle finds are as follows:

(1) Monthly changes in downstream flows on the Lancang within Yunnan do not only reflect changes in upstream flows, but even more clearly reveal the influence of monthly climate changes within Yunnan.

(2) Monthly flow changes in Yunnan exhibited certain obvious traits over a variety of timescales. Characteristic timescales for the upstream station at Jiuzhou were 4 months, 7 months, 15 months, 30 months, 53 months, and 74 months. For the downstream station at Yunjinghong, characteristic timescales were 4 months, 7 months, 14 months, 21 months, 30 months, 53 months, 74 months, and 106 months. Finally, for the difference in flows between the two stations, characteristic timescales are 4 months, 14 months, 21 months, 38 months, 53 months, and 100 months.



(3) It is possible that 4- and 53-month characteristic timescale changes (i.e., 8- and 106- month characteristic time cycle changes) exist between monthly climate change in the upstream region of the Lancang in Yunnan and monthly climate change throughout the whole of Yunnan Province.

(4) Flow changes downstream at Yinjinghong occurring at the 7-, 30-, and 74-month characteristic timescales are primarily the result of flow changes upstream at Jiuzhou. Flow changes at Yinjinghong occurring at the 14- and 21-month characteristic timescales, however, are the result of monthly climate change across Yunnan Province.

(5) For smaller characteristic timescales, Lancang monthly flow changes caused by monthly climate change in Yunnan generally mirror or lag behind changes in upstream monthly flows. For larger characteristic timescales, however, such flow changes lag behind monthly flow changes upstream.

Fig. 7: Wavelet transform coefficient phase for 53-month characteristic timescale (Black=Jiuzhou; Green=Diff-YJ)

ACKNOWLEDGEMENTS

This research was supported by funding from the National Key Basic Research Development Plan 973 (Approval Number: 2003CB415105), National Key Science and Technology Plan (Approval Number: 2002BA901A22), and Yunnan Province Natural Science Foundation Key Project (Approval Number: 2001D0002Z).

REFERENCES Chen, L., & He, D. (1999) Lancang Jiang-Meigong He liuyu zhengti kaifa de qianjing yu wenti yanjiu (The

Prospect and Problem of Integrated Development in Lancang-Mekong River Basin). Dili Xuebao (Acta Geographica Sinica), 54 (Supplement), 55-64.

Chen, L., & He, D. (2000) Lancang Jiang-Meigong He shuidian tiji kaifa de shengtai yingxiang (The ecological impacts of hydropower cascade development in Lancang). Dili Xuebao (Acta Geographica Sinica), 55(5), 577-586.

Chen, L., & He, D. (2001) Lancang Jiang-Meigong He zhengti shui fenpei (Comprehensive water allocation for the Lancang-Mekong River). Jingji Dili (Economic Geography), 21(1), 28-32.



Chen, Y., Zhou, R., & Wu, H. (2002) Characteristics and impacts of rising western Pacific from Nino 1+2 cold and warm water periods on southeast monsoons Daqi kexue (Atmospheric Science), 26(3), 373-386.

He, D., & Lin, Z. (1997) Lancang Jiang-Meigong He ciquyu guoji hezuo he kechixu fazhan (International cooperation and sustainable development in the Greater Mekong Subregion). Guoji jishu jingji yanjiu xuebao (Journal of international technical economic research), 43(3), 24-30.

He, D., Liu, C., & Yang, Z. (1999) Zhongguo guoji heliu kechixu fazhan yanjiu (Study for the Sustainability of International Rivers in China). Dili Xuebao (Acta Geographica Sinica), 54 (Supplement), 1-10.

He, D., Yang, M., & Feng, Y. (1999) Xi'nan guoji heliu shuiziyuan de heli liyong yu guoji hezuo yanjiu (Study on Reasonable Utilization of Water Resources in International Rivers and International Region Cooperation in Southwest China). Dili Xuebao (Acta Geographica Sinica), 54 (Supplement), 29-37.

Huang, J. (2000) Qixiang tongji fenxi yu yubao fangfa (Statistical analysis and prediction methods in meteorology). Beijing: Meteorology Publishing.

Mann, M. E., & Park, J. (1996) Joint spatiotemporal modes of surface temperature and sea level pressure variability in the Northern Hemisphere during the last century Journal of Climate, 9(9), 2137-2162.

Tang, H. (1999) Lancang Jiang-Meigong He liuyu de shuiziyuan ji qi kaifa liyong xianzhuang fenxi (Analysis of the current situation of water resources development and use in the Lancang-Mekong watershed). Yunnan Dili Huanjing Yanjiu (Yunnan Geographic Environment Research), 11(1), 16-25.

Venegas, S. A., & Mysak, L. A. (2000) Is there a dominant timescale of natural climate variability in the Arctic? Journal of Climate, 13(9), 3412-3434.

White, W. B., Lean, J., Cayan, D. R., & Dettinger, M. D. (1997) Response of global upper ocean temperature to changing solar irradiance Journal of Geophysics Research, 102, 3255-3266.

You, L. (2001) Lancang Jiang hedao chongyou bianhua tezheng ji fazhan qinshi (Characteristics and developing trends of change in scour sedimentation in the Lancang River). Dili Yanjiu (Geographic Research), 20(2), 178-183.

You, W. (1998) Qihou bianhua de duo chidu zhenduan fenxi he yuce de duozhong jishu fangfa yanjiu (Several technical methods for multiscale diagnosis, analysis, and prediction of climate change). Beijing: Climate Publishing.

You, W., Duan, X., & Qi, M. (1999) Liannxu xiaobo bianhua zai Yunnan jin bai nian qiwen he jiangshui bianhua fenxi zhong de yingyong (Application of continuous wavelet transforms for analysis of Yunnan's climate and precipitation change over the past 100 years). Gaoyuan Qixiang (Highland Climate), 18(1), 47-54.

Zhang, X., Sheng, J., & Shabbar, A. (1998) Modes of interannual and interdecadal variability of Pacific SST Journal of Climate, 11, 2556-2569.

CORRELATION AND MULTI-TIMESCALE CHARACTERISTICS OF THE LANCANG RIVER'S MONTHLY FLOW VARIATIONS IN YUNNAN (CHINA)

Documents