Recent Changes of Sediment Yield inthe Upper Yangtze, ChinaXIXI LUDAVID L. HIGGITT*Department of GeographyUniversity of Durham, Science LaboratoriesSouth RoadDurham DH1 3LE, UK

ABSTRACT / Reservoir sedimentation is one of the manyenvironmental problems associated with the Three GorgesProject in China. The rate and characteristics of sedimenta-tion that directly affect the operating life of the reservoir areclosely related to soil erosion and sediment transport dy-namics in the upstream catchment and to the ability to man-age the throughput of sediment-laden waters. The recentchanges in sediment yield were examined using gaugingdata from 187 stations of varying sizes from less than 100km2 to larger than 1,000,000 km2 in the Upper Yangtze basinbetween 1956 and 1987. Whereas many previous studies

have concentrated on the trends in the main channel of theYangtze, the distributed pattern of changes across the wholecatchment is complex. Results from time series analysis indi-cate ten stations, mainly located in the Dadu and Wu tributar-ies (with a total incremental catchment area of 78,963 km2)have shown increasing trajectories of sediment yield, and sixstations, located in the upper Jialing and Tuo tributaries (witha total incremental area of 27,816 km2) have experienceddecreasing trajectories. By dividing the time series into threecomponents, it is possible to map significant decadalchanges in sediment yields that can be related to phases ofdeforestation and the construction of water conservancyprojects. Most of the observed decreases in sediment yieldare associated with large reservoir schemes on tributary riv-ers. The lack of evidence for increasing sediment input to theThree Gorges area masks a considerable variation in sedi-ment conveyance and storage within the Upper Yangtzecatchment.

The dynamics of soil erosion and fluvial sedimenttransport in the Upper Yangtze River is a matter ofenvironmental and engineering concern because of thepotential impact of sedimentation on the Three GorgesProject (TGP). Temporal variability of sediment trans-port is mainly due to climate and human activity such asdeforestation, soil conservation, and urbanization (Wall-ing and Webb 1996). Long-term change in sedimentyields has been investigated mainly through the recon-struction of the variable sediment input to the sea(Milliman and others 1987, Degens and others 1991)and lakes or reservoirs (Oldfield and others 1980,Dearing and others 1987), while direct measurement ofsediment concentrations at gauging stations facilitatesanalysis of recent changes.

There have been a number of attempts to relateglobal or regional sediment yields to controlling factors.Langbein and Schumm (1958) produced a theoreticalmodel of sediment yield in relation to effective precipi-tation, reaching a maximum in semiarid environments(effective precipitation 300 mm) and declining as vegeta-tion cover protects the land surface. As the availability

of gauging station data has increased, the complexrelationship between sediment yield and precipitation(or runoff) has been noted (e.g., Wilson 1972, Wallingand Webb 1983), not least because of the impact ofhuman activity in disturbing natural vegetation coverand promoting soil erosion. The role of topography hasalso been highlighted (Milliman and Syvitski 1992,Summerfield and Hulton 1994), and the importance ofsmall mountainous catchments has been identified as amajor contributor to global continental sediment ex-port. However, prediction of sediment yields is compli-cated by the interaction between controlling variables,the influence of human interference in the hydrologi-cal system, and by scale effects of catchment size. Inmost areas, the proportion of sediment mobilized oncatchment slopes that is exported to the catchmentoutlet decreases as catchment size increases (Walling1983). This sediment delivery ratio makes direct com-parison of specific sediment yields (load per unit timeper unit area) difficult. Furthermore, human activitymay serve to increase sediment mobilization on slopesthrough soil erosion, but decrease sediment deliveryratios through the impoundment of sediments behindcheck dams and within reservoirs. The legacy of pastsediment transport dynamics is also important. In theupper Mississippi sediment yields have remained highdespite successful soil conservation, as material origi-

KEY WORDS: Sediment yield; Reservoir sedimentation; Three GorgesProject; Time series analysis; China

*Author to whom correspondence should be addressed.

Environmental Management Vol. 22, No. 5, pp. 697–709 r 1998 Springer-Verlag New York Inc.

nally eroded in the nineteenth century has been remo-bilized from valley floors (Trimble 1981). A similar pat-tern has been identified in the Piedmont (Phillips 1993).

Identifying the controls on recent changes in soilerosion, sediment delivery, and sediment yield is anissue of concern to policy makers engaged in the man-agement of rivers with high sediment loads. Althoughseveral examples of accelerated erosion and increasedsediment loads are associated with anthropogenic activ-ity, such as deforestation, vegetation clearance, or land-use change, some human activity may not lead toobvious trends, particularly in large catchments (Wall-ing 1997). For example, Alford (1992) reported that theChao Phraya basin (14,028 km2), draining the high-lands of northern Thailand, showed no evidence of asignificant increase in sediment yield during the periodfrom the late 1950s to the mid-1980s, despite substantialdeforestation and extensive expansion of agriculturalactivity within the basin. A similar situation has alsobeen found in the Upper Yangtze river and severalauthors have noted that there is no systematic trend insediment loads in the Upper Yangtze (Gu and others1987, Gu and Douglas 1989, CIPM 1988, Zhuo andXiang 1994, Dai and Tan 1996). Most of the analysis hasbeen based on large stations, particularly that at Yichang,and there has been relatively little attention paid to theentire gauging network within the catchment.

Rural China has experienced profound changes inland use in recent decades, including widespread defor-estation and extension of agricultural land. For ex-ample, forest cover in Sichuan decreased from 19% to12% between the 1950s and the 1980s, while in neighbor-ing Guizhou it declined from 23% to 13% between the1960s and the 1980s (Yu and others 1991). Conse-quently the reported extent of soil erosion in bothprovinces has dramatically increased. Inventories oferosion status undertaken in the 1950s and 1980sindicate a rise from 16% to 67% of land area affected inSichuan and from 11% to 31% in Guizhou (Chen andGao 1988). Such a large increase in soil loss on slopesmight be expected to lead to increasing sediment loadsin the Upper Yangtze catchment. On the other hand,the initiation of water conservancy projects, includingthe construction of ponds, small check dams, ditchesand headwater reservoirs, may trap a considerableproportion of eroded sediment and prevent its transferto the river either temporarily or permanently (Luk andWhitney 1993). Together with larger HEP schemes, it isestimated that the by the mid-1980s the total reservoircapacity in the Upper Yangtze exceeded 16 billion m3

(Gu and others 1987) with some 0.3 billion m3 capacitylost to sedimentation each year (Chen and Gao 1988).The interaction between factors controlling the input of

sediment into the fluvial network and those controllingthe efficiency of transmission, interposed with year-to-year hydrological variability, cause problems for theinterpretation of trends of sediment yield based on datafrom the Yichang gauging station alone. However, it islikely that trends in the relative importance of parts ofthe catchment will emerge from analysis based on allgauging station data. This paper investigates temporalchanges of sediment yield and the spatial locations ofthese changes using sediment yield data from theUpper Yangtze gauging stations collected between 1956and 1987. Reasons for the observed changes are dis-cussed.

Study Area and Data Sources

The Upper Yangtze traditionally refers to the catch-ment upstream of Yichang, Hubei Province. The catch-ment generally experiences a subtropical monsoonclimate, apart from the relatively arid, mountainouswest where the Yangtze rises on the Qinghai–TibetPlateau. The plateau, resulting from collision of theIndian and Asian plates, contains four fifths of theworld’s land surface higher than 4000 m and stronglyinfluences climate by affecting atmospheric circulation(Ruddiman and others 1989). Precipitation varies from,250 mm in the north of Qinghai–Tibet to .1000 mmin the east of the basin. Population densities aresimilarly varied ranging from ,10 people/km2 in themountainous west to .500 people/km2 in the SichuanBasin, which is one of the most populous areas of China.The Upper Yangtze catchment has a number of impor-tant tributaries: Upper Jinsha, Yalong, Dadu, Min, Tuo,Fu, Jialing, Qu, and Wu (Figure 1). The first four ofthese mainly drain the mountainous west while the lastfive flow through areas of higher population density.

The sediment yield data are from Chinese hydrologi-cal yearbooks. There are five volumes for the UpperYangtze relating to the main tributaries: Jinsha–Yalong,Min–Dadu (including Tou), Jialing, Wu, and MainChannel. The yearbooks date back to the 1930s for a fewstations but the majority date back to the 1950s. Theoriginal records for each station are coordinates (lati-tude and longitude), catchment area, mean monthlyloads, mean annual loads, and maximum daily dis-charge and its date of occurrence. The yearbook datawere transcribed from paper to a series of linkedspreadsheets. A few problems with data quality in theoriginal books were identified by recalculating meanannual load from monthly load data (Higgitt and Lu1996). It should be recognized that the sediment loadsare based on discrete, rather than continuous measure-ments, and they do not account for bedload transport.

X. X. Lu and D. L. Higgitt698

Figure 1. The Upper Yangtze catchment.

Figure 2. Distribution of gauging stations in the Upper Yangtze.

Sediment Yield in Upper Yangtze 699

The spatial distribution of stations is also uneven with aconcentration in the populated and humid east relativeto the mountainous, semiarid west. The period ofoperation of gauging stations is also variable. Of the 250stations that have operated within the Upper Yangtzesince 1956, 187 stations (74.8%) have measurementseries 5 years or longer, and 56 stations (22.4%) 25 yearsor longer.

Methods

The study employed two simple methods for evaluat-ing trends. First, in order to examine long term changes

of sediment yield, the linear regression model,

SY 5 a1 1 b1T

was used to examine the relationship between sedimentyield (SY) and the year of measurement (T). If thecorrelation coefficient was larger than the critical valueat a 95% significance level, it is assumed that SY hadexperienced a significant change (increase or decrease)over the measurement period. The method, applied tothe 187 stations with 5 years or more of measurementseries, provides information about the overall directionof trend. As year-to-year variations in sediment yield arelikely to be affected by annual runoff, the stations that

Figure 3. Plot of time series correla-tion coefficients against catchmentarea.

Table 1. Stations with significant time series trends at significant level a 5 0.05

Station TributaryStationname

Catchmentarea (km2)

Lengthof record(years)

SYcorrelationcoefficient,

r b1

Runoffcorrelationcoefficient,

r b2 tau P

14 Jinsha-Yalong Huidongqiao 590 21 0.52 60.1 20.13 210.6 0.61 0.0001210 Wu Niuchishui 2,210 20 0.60 58.9 0.12 1.9 0.54 0.001

23 Jinsha-Yalong Ninnan 3,074 25 0.49 45.3 20.12 21.8 0.43 0.003249 Main Channel Xingshan 1,900 14 0.60 44.4 0.01 0.2 0.54 0.0086114 Dadu-Min Yanruen 3,302 23 0.45 30.1 0.01 0.2 0.35 0.020116 Dadu-Min Shaping 75,016 21 0.76 18.0 0.25 2.0 0.53 0.0008180 Jialing Guodukou 31,626 30 0.49 17.9 0.33 8.3 0.24 0.087216 Wu Huoshiba 1,527 19 0.52 13.3 0.05 1.1 0.36 0.036207 Wu Wulong 83,035 27 0.51 9.0 0.27 4.0 0.28 0.052

40 Jinsha-Yalong Xiaohekuo 649 7 0.77 7.2 0.70 70.585 Dadu-Min Zagunao 2,404 26 20.42 28.3 20.11 21.6 20.25 0.08082 Dadu-Min Heishui 1,720 23 20.64 211.5 20.35 24.4 20.52 0.0006

154 Jialing Sanleiba 29,247 29 20.45 212.2 20.25 22.3 20.40 0.0004128 Tou Denyenyan 14,484 31 20.36 214.4 20.21 23.0 20.24 0.057

36 Jinsha-Yalong Dianwei 630 29 20.59 216.2 20.17 20.8 20.25 0.066153 Jialing Bikou 26,086 27 20.53 219.3 20.04 20.3 20.54 0.0001

X. X. Lu and D. L. Higgitt700

recorded significant trends in sediment yield wereinvestigated for underlying trends in runoff (Q). Plotsof cumulative runoff against cumulative sediment yieldhave been used elsewhere (e.g., Walling and Webb1996) to illustrate graphically the changing relationshipbetween sediment yield and runoff. This can be supple-mented by examining the temporal trend in the resi-duals from the SY–Q regression (Helsel and Hirsch1992).

Having examined evidence for trends throughoutthe measurement period, the second method aimed toexamine temporal variations within the time period.Analysis was confined to the 56 stations that have at least25 years of data plus the addition of six stations from theWu tributary, which is not represented by long-termstations. The t test for the difference of means betweentwo samples has been applied in studies of climaticchanges (Sneyers 1992, Loureiro and Coutinho 1995)for evaluating the significance of change between de-fined periods. The sediment yield data were dividedinto three periods: 1956–1967 (P1), 1968–1977 (P2),and 1978–1987 (P3), and a series of t tests undertaken.

The division gives approximately 10 years for eachperiod, but more interestingly, coincides with the threehistorical incidents; the Great Leap Forward, the Cul-tural Revolution, and the Land Responsibility Reform.Each of these incidents is considered to have hadprofound influences on land use and hence on erosionand sediment transport potential. During the GreatLeap Forward from 1958 to 1961 nationwide deforesta-tion occurred for small-scale iron smelting (Biot and Lu1993). During the decade of the Cultural Revolution(1966–1975), many water conservancy projects in fields(terraces and ditches) and in waterways (dams orreservoirs) were constructed. Inventories of agriculturalland indicate an increasing extent of soil erosion duringthis period, but the water conservancy schemes will haveaffected sediment delivery ratios. Land ResponsibilityReform, from the end of 1970s, resulted in a furtherphase of widespread deforestation, as the harvestingand selling of timber became lucrative. More detaileddiscussion about the environmental effects of modelChinese socioeconomic policies is available elsewhere(Smil 1987, 1993, Edmonds 1994).

Figure 4. Specific sediment yieldtime series for six selected sta-tions with large catchment areas.

Sediment Yield in Upper Yangtze 701

Results and Discussion

Long-Term Trends in Sediment Yields

The correlation coefficients of the linear regressionequations indicate that the majority of stations did notexperience a significant change in sediment yield acrossthe time period. Among the 187 stations, however, therewere 16 stations displaying significant change at the95% confidence level. Of these, ten stations, with a totalincremental area of 78,963 km2 (7.9% of catchmentarea), experienced increasing sediment yields and six,with a total incremental area of 27,816 km2 (2.8%),decreasing yields (Table 1). The stations are ranked indescending order of the slope coefficient of the regres-sion (b1). It can be noted that the rate of change inupwardly trending sediment yields is considerably higherthan the rate of change exhibited in downwardlytrending sediment yields. The locations of the 16stations exhibiting significant change were marked inFigure 2. The sharpest increases are recorded fromdiverse geographical locations but feature catchmentssmaller than 4000 km2, which may suggest the impact ofrelatively localized land-use changes. In order to estab-lish whether the likelihood of identifying a trend isscale-dependent, correlation coefficients for the SY–T

relationship have been plotted against catchment area(Figure 3). There is no apparent relationship betweenstrength of trend and catchment size, although themajority of correlation coefficients are positive.

Of the stations exhibiting significant change, it canbe noted that six, Bikou (No. 153), Sanleiba (No. 154),Denyenyan (No. 128), Shaping (No. 116), Guodukou(No. 180), and Wulong (No. 207), were located onprincipal tributaries (Figure 2). Time series plots ofthese stations are shown in Figure 5. Bikou (No. 153)and Sanleiba (No. 154) on the Jialing and Denyenyan(No. 128) on the Tuo experienced a decline in sedi-ment yields. For the Jialing tributaries the experiencecan be explained by the construction of the Bikou Damin 1975 with a 300-MW capacity. The impact of largereservoirs can be seen readily by comparing the timeseries of gauging stations upstream and downstream ofthe catchment’s four largest dams: Bikou, Gongzui,Wujiadu, and Gezhouba (location in Figure 3; plot inFigure 5). Numerous smaller reservoir and other waterconservancy projects have been developed within thecatchment over the last 30 years with clear implicationsfor the conveyance of fluvial sediment load. The distri-bution of water conservancy projects, in terms ofreservoir capacity is also unevenly distributed (Table 2).

Figure 5. Impact of large reservoirs on sediment yield time series. Solid lines indicate stations above the reservoir and dashed linestations downstream of reservoir.

X. X. Lu and D. L. Higgitt702

Figure 6. (A) Sediment yield–runoff rela-tionships, and (B) time series plot of regres-sion residuals for stations experiencing in-creasing sediment yields. Key: Ln(SY): logsediment yield (t/km2/yr), Ln(Q): log run-off (mm), tau: Mann-Kendall rank correla-tion, p: significance level.

The Tuo, Fu, and Qu have high ratios of reservoircapacity relative to catchment area, while the density ofreservoir capacity is more limited in the Wu andJinsha–Yalong tributaries. The significant decrease insediment yield at Denyenyan station (No. 128) on theTuo might similarly be attributed to high aggregatereservoir capacity introduced to its catchment areaduring the period.

In contrast, the gauging stations at Shaping (No.116), Guodukou (No. 180) and Wulong (No. 207)experienced increasing sediment yields during theperiod. These are the principal gauging stations for theDadu, Qu, and Wu, respectively (Figure 2). The Daducatchment spans the transition between the mountain-ous western Sichuan and the intensive agriculture of theSichuan Basin, an area that has experienced bothdeforestation and the extension of arable land. The Qu,draining eastern Sichuan has witnessed a dramatic risein its importance as a sediment source during theperiod. By the 1980s its specific sediment yields wereroughly double the catchment average (Higgitt and Lu1996). The third tributary experiencing increasingsediment yield is the Wu, which mainly flows throughGuizhou Province. This area, which is relatively poor,has experienced rapid deforestation and a markedincrease in soil erosion during the period.

As noted elsewhere (Gu and Douglas 1989, Zhuo andXiang 1994, Dai and Tan 1996), the sediment yieldmeasured at Yichang station has not exhibited anysignificant trend, and the long-term mean sedimentload has formed the basis for most investigations of theprospective Three Gorges Dam sedimentation prob-lem. The implication is that changes in upstreamsediment transport dynamics are damped or lagged bythe time they reach the main channel. In addition, the

construction of the Gezhouba Dam on the main chan-nel of the Yangtze upstream of Yichang in 1981 has hadsome impact on the sediment yields recorded at Yichang.Comparison between Yichang and Chuntan station(located upstream end of the Gezhouba Reservoir,Figure 1) show that from 1970 to 1980 specific sedimentyield at Yichang was higher than that at Chuntan but thedifference was reduced after 1981 when the reservoirbegan to store water (Figure 5).

The discussion so far has considered evidence for theimpact of land-use changes and the construction ofreservoirs. Year-to-year variability related to hydrologi-cal conditions is important since this may mask moresubtle changes in sediment transfer. As in many studieselsewhere, sediment yield in the Upper Yangtze ishighly dependent on annual streamflow (Zhu andothers 1993). In order to consider the effect of waterdischarge on sediment loads, a similar regression model,Q 5 a2 1 b2T, is applied to the runoff for those stationsexperiencing significant trends of sediment yield deter-mined above. The results were also summarized inTable 1. None of the correlation coefficients are signifi-cant at the 95% level for runoff. Two stations, Huidong-qiao (No. 14) and Niushishui (No. 210), have negativecorrelation coefficients for runoff but positive correla-tion coefficients for sediment yield, indicating strongersediment yield increase trends. However, most stationshave runoff and sediment yield trending in the samedirection. The comparative increase in runoff andsediment yield can also be examined by plotting cumu-lative runoff against cumulative sediment yield as adouble mass plot (Walling, 1997). Helsel and Hirsch(1992) suggest a more robust method to remove theeffect of runoff variation by examining trends in theresidual of the SY–Q relationship. Runoff data wereobtained for all of the stations in Table 1 and logarith-mic least-squares linear regression equations betweensediment yields and runoff were developed (Figure 6).All the regressions are significant, except for Zagunao(No. 85). The residuals from the equations (the differ-ences between the observed and predicted values ofsediment yield in logarithmic units) can be used tomeasure the variations in the sediment yield caused byother factors rather than water discharge. The regres-sion residuals were plotted against time (Figure 7).Significant trends were determined with the Mann-Kendall test (Helsel and Hirsch 1992). It is found thatthe level of significance in the trend of residuals fallsslightly below the 95% confidence level for five of thestations: Guodukou (No. 180) and Wulong (No. 207)with increasing yields and Zagunao (No. 85), Deyenyan(No. 128), and Dianwei (No. 36) with decreasing yields.

Table 2. Water conservancy capacities within UpperYangtze catchmenta

Maintributary

Totalreservoircapacity

(million m3)

Percentageof UpperYangtze

reservoircapacity

Ratio ofreservoir

capacity tocatchment area(103 m3/km2)

Jinsha-Yalong 1,802 10.8 3.6Dadu-Min 1,737 10.4 13.1Tuo 2,549 15.2 91.1Fu 1,900 11.4 60.0Jialing 2,951 17.6 29.8Qu 1,548 9.3 52.6Wu 578 3.5 8.0Main channel 3,654 21.8 32.6Total 16,719 100.0 16.6

aReservoir capacity data derived from Gu and others (1987).

X. X. Lu and D. L. Higgitt704

The station at Xiaohekuo (No. 40) has been omittedfrom residual analysis because of the limited length ofthe record.

Upper Yangtze sediment yields viewed from thecatchment outlet at Yichang alone do not appear tohave changed markedly in recent times. Further investi-gation from stations within the catchment, however,demonstrates some coincidence between evidence ofincreasing soil erosion and fluvial sediment transportbut also that the Upper Yangtze has a huge capability tostore sediment and buffer the effects of increasingsediment supply, particularly due to recent water conser-vancy projects.

Changes Among the Three Periods

Having examined evidence for trends across thewhole period, fluctuations in mean sediment yieldsamong the three time periods discussed previously[1956–1967 (P1), 1968–1977 (P2), 1978–1987 (P3)]were investigated. Again, the majority of stations, includ-ing Yichang show no significant changes between eachof the periods, which suggests that year-to-year hydro-logic variability was more significant than the impact ofland-use changes (Zhou and Xiang 1994). However, anumber of stations do exhibit significant changes be-tween each of the periods (listed in Table 3). These

Figure 7. (A) Sediment yield–runoff rela-tionships, and (B) time series plot of regres-sion residuals for stations experiencing de-creasing sediment yields. Key as in Figure 6.

Sediment Yield in Upper Yangtze 705

feature both increases and decreases in different partsof the basin but, as reported elsewhere (Gu and Douglas1989, Higgitt and Lu 1996), the main change insediment yields was decreasing from P1 to P2 andincreasing from P2 to P3 for most tributaries. Thepattern for stations within the Wu tributary was gener-ally reversed.

The areas experiencing significant difference ofmean sediment yield are mapped in Figure 8. Themapping units are based on incremental catchmentareas that deduct nested catchment boundaries unlessthey also exhibit significant change. From P1 to P2,mean sediment yield significantly increased at twostations, Wulong (No. 207) and Gongtan (No. 208) with

a total incremental catchment area of 33,450 km2

(3.3%) and decreased at 11 stations representing 132,684km2 (13.2%) mainly in the Jialing, Tuo, and the lowerDadu–Min tributaries. These decreases reflect a combi-nation of massive deforestation in first period and theestablishment of water conservancy projects in secondperiod. The decrease in yield in the Dadu–Min may bedue to trapping of sediment in the Gongzhui reservoir(see also Figure 6). Runoff data for those stations werechecked and indicated that the stations did not experi-ence a significant change in runoff coincidental to thechanges in sediment yield, except for Dianwei (No. 36)from P2 to P3 (Table 3).

From P2 to P3, mean sediment yield increased for

Table 3. Stations exhibiting significant change between successive time periodsa

Station TributaryStationname

Catchmentarea (km2)

Time periods t test

P11956–1967

P21968–1977

P31978–1987

P2 vP1

P3 vP2

36 Jinsha–Yalong Dianwei 120 SY (t/km2/yr) 490 134 186 **Runoff (mm) 134 112 113

98 Dadu–Min Gaochang 135378 SY (t/km2/yr) 457 262 352 **Runoff (mm) 676 603 631

105 Dadu–Min Dajin 40484 SY (t/km2/yr) 100 77 140 *Runoff (mm) 415 384 415

116 Dadu–Min Shaping 75016 SY (t/km2/yr) 302 305 546 **Runoff (mm) 581 555 595

128 Tuo Denyenyan 14484 SY (t/km2/yr) 848 431 550 **Runoff (mm) 720 600 638

129 Tuo Lijiawan 23283 SY (t/km2/yr) 760 352 480 **Runoff (mm) 603 477 531

140 Jialing Liuanyang 19206 SY (t/km2/yr) 2186 990 1824 **Runoff (mm) 244 136 219

153 Jialing Bikou 26086 SY (t/km2/yr) 1012 531 407 *Runoff (mm) 360 300 343

156 Jialing Tinzhikou 61089 SY (t/km2/yr) 1426 622 1031 **Runoff (mm) 402 274 321

164 Jialing Wusheng 79714 SY (t/km2/yr) 1187 620 951 **Runoff (mm) 398 287 336

165 Jialing Beipei 156142 SY (t/km2/yr) 1130 742 937 *Runoff (mm) 474 384 421

180 Jialing Guodukou 31626 SY (t/km2/yr) 423 586 882 *Runoff (mm) 572 575 716

195 Jialing Shehong 23574 SY (t/km2/yr) 931 507 658 *Runoff (mm) 669 518 525

197 Jialing Xiaoheba 29420 SY (t/km2/yr) 839 419 690 **Runoff (mm) 585 459 479

201 Wu Wujiangdu 26496 SY (t/km2/yr) 462 696 208 **Runoff (mm) 540 608 570

206 Wu Shinan 50791 SY (t/km2/yr) 364 441 169 **Runoff (mm) 497 571 527

207 Wu Wulong 83035 SY (t/km2/yr) 287 490 415 **Runoff (mm) 574 638 600

208 Wu Gongtan 58346 SY (t/km2/yr) 284 449 296 * *Runoff (mm) 559 638 555

243 Main Channel Wanxian 974881 SY (t/km2/yr) 450 445 559 *Runoff (mm) 434 416 427

a**significant at a 5 0.01; *significant at a 5 0.05.

X. X. Lu and D. L. Higgitt706

four stations (116,363 km2, 11.6%) and decreased forthree stations (39,146 km2, 3.9%). Sediment yieldsincreased in the Dadu and upper Qu, neither of whichrecorded changes between the earlier periods. It is mostlikely that this reflects deforestation for timber supplyand expansion of new arable lands triggered by LandResponsibility Reform. The affected areas are located atthe transition between predominantly agricultural andforestry land uses.

Conclusion

The study has demonstrated that sediment loadsexported from large catchments mask a considerableamount of variability in the trajectories of soil erosion

and sediment transport within different parts of thecatchment. Oscillation around a nontrending meanvalue is usually attributed to hydrological variability, butfurther investigation can demonstrate underlyingchanges in the sediment yield–runoff relationship.Examination of sediment yield data from stations withinthe Upper Yangtze collected between 1956 and 1987reveal some patterns of temporal change. Sedimentyield had mainly increased for the Dadu, Qu, and Wutributaries and decreased in the Tuo and upper Jialing.Most of the observed decreases in sediment yield can berelated to the construction of large reservoir schemesinterrupting the downstream conveyance of sediment.

Comparison of sediment yield and runoff data sug-gest that detectable changes in sediment delivery occur-

Figure 8. Incremental catch-ment areas undergoing signifi-cant change between(A) 1956–1967 and 1968–1977;(B) 1968–1977 and 1978–1987.

A

B

Sediment Yield in Upper Yangtze 707

ring in the Upper Yangtze reflect an increasing supplyinduced by human activities over the last 30 years.Comparison of three time periods framed by majorshifts in the nature of resource exploitation (the GreatLeap Forward, the Cultural Revolution, and the LandResponsibility Reform) show changing patterns of sedi-ment yield that are most likely related to the changinggeographical expression of deforestation and of waterconservancy project construction. Linking soil erosionon slopes to sediment transport in the fluvial system ishampered by the limited amount of quantitative dataon soil erosion rates. Many geomorphological systemsare controlled by threshold conditions such that adjust-ment to controlling variables is likely to be nonlinear.Although the patterns of changing sediment yieldwithin the Upper Yangtze can be matched to knownhistories of human disturbance, the possibility thatsediment yields are responding to subtle changes inclimate cannot be dismissed. However, suggestions thatsediment yield patterns in China essentially reflectclimatic controls (Xu 1994) clearly do not hold withinthe upper Yangtze basin. Sediment yield variation mod-eling of the Upper Yangtze within a GIS frameworkindicates the importance of topographic variables, par-ticularly slope and altitude in explaining sediment yieldvariability. The latter may be partly acting as a surrogatefor human activity and precipitation, which are bothinversely related to altitude in the Upper Yangtze.

Detailed examination of spatial and temporal yieldvariation may have implications for policy makers con-cerned with the management of potential sedimenta-tion. Discussion of the TGP reservoir sedimentationissue has largely focused around river regulation proce-dures and the design of sluices to maximize the dis-charge of sediment-laden waters. The lack of evidencefor an increasing sediment yield at Yichang has justifiedthe use of the long-term mean load and limited atten-tion to long-term changes in sediment supply to thereservoir. The evidence for increasing sediment supplyin the catchment buffered by increasing storage inintermediate reservoirs raises further concerns. As thisreservoir storage begins to fill, the trap efficiency ofindividual reservoirs will decline and the conveyance ofsediment downstream will increase. The long-termimpact of the exhaustion of artificial reservoir capacityand the remobilization of sediments from storage couldbe significant. Analysis of time series for individualstations suggests that 7.9% of the Upper Yangtze catch-ment had an increase in sediment yields over themeasurement period while 2.8% experienced decreas-ing yields. The variable response of large catchments tophases of human disturbance and river engineeringprovides many challenges to basin management, to

which the detailed analysis of intracatchment sedimentyields may contribute.

Acknowledgments

The work presented in this paper was undertakenwhile X. X. Lu was in receipt of a Durham Universityscholarship and an Overseas Research StudentshipAward. The authors would like to acknowledge thehelpful comments of three reviewers—Dr. M. A. Madej,Dr. F. D. Shields, Jr., and Dr. R. E. Turner. TheCartographic Unit, University of Durham, preparedsome of the maps.

Literature Cited

Alford, D. 1992. Streamflow and sediment transport frommountain watersheds of the Chao Phraya basin, northernThailand: a reconnaissance study. Mountain Research andDevelopment 12:257–268.

Biot, Y., and X. X. Lu. 1993. Assessing the severity of theproblem and the urgency for action. Pages 165–191 in E.Baum, P. Wolff, and M. A. Zobisch (eds.), Acceptance of soiland water conservation strategies and technologies. TheGerman Institute for Tropical and Subtropical Agriculture(DITSL), Witzenhausen.

Chen, G. J., and F. H. Gao. 1988. The ecology and environ-ment of the Yangtze and the Three Gorges, and the ThreeGorges project. Pages 1–15 in The leading group of theresearch project of Chinese Academy of Sciences (ed.), Theeffect of the Yangtze Three Gorges project on ecology andenvironment and countermeasures. Science Press, Beijing(in Chinese).

CIPM (Canada International Project Managers Ltd.). 1988.Yangtze joint venture: Three Gorges water control projectfeasibility study, Vols. 1–11.

Dai, D. Z., and Y. Tan. 1996. Soil erosion and sediment yield inthe Upper Yangtze River basin. Pages 191–203 in D. E.Walling and B. W. Webb. (eds.), Erosion and sediment yield:Global and regional perspectives (Proceedings of the ExeterSymposium, July 1996). IAHS Publ. No. 236. InternationalAssociation of Hydrological Sciences, Wallingford.

Dearing, J. A., H. Hakansson, B. Liedberg-Johnsson, A. Pers-son, S. Skansjo, D. Widholm, and F. El Daoushy. 1987. Lakesediments used to quantify the erosional response to landuse change in southern Sweden. Oikos 50:60–78.

Degens, E. T., S. Kempe and J. E. Richey. 1991. Summary:Biogeochemistry of major world rivers. Pages 323–347 in E.T. Degens, S. Kempe, and J. E. Richey (eds.), Biogeochemis-try of major world rivers. Wiley, Chichester, UK.

Edmonds, R. L. 1994. Patterns of China’s lost harmony: Asurvey of the country’s environmental degradation andprotection. Routledge, London.

Gu, H. Y., and I. Douglas. 1989. Spatial and temporal dynamicsof land degradation and fluvial erosion in the middle andupper Yangtze River basin, China. Land Degradation &Rehabilitation 1:217–235.

X. X. Lu and D. L. Higgitt708

Gu, H. Y., N. S. Ai, and H. L. Ma. 1987. Sediment sources andtrend of sedimentation in the Three Gorges reservoir area.Pages 522–541 in Leading group of the Three Gorgesproject ecology and environment research project, ChineseAcademy of Sciences (ed.), Collected papers on ecologicaland environmental impact of the Three Gorges project andcountermeasures. Science Press, Beijing (in Chinese).

Helsel, D. R., and R. M. Hirsch. 1992. Statistical methods inwater resources: Studies in environmental science 49.Elsevier, Amsterdam.

Higgitt, D. L., and X. X. Lu. 1996. Patterns of sediment yield inthe Upper Yangtze Basin, China. Pages 205–214 in D. E.Walling and B. W. Webb. (eds.), Erosion and sediment yield:Global and regional perspectives (Proceedings of the ExeterSymposium, July 1996). IAHS Publ. No. 236. InternationalAssociation of Hydrological Sciences, Wallingford.

Langbein, W. B., and S. A. Schumm. 1958. Yield of sediment inrelation to mean annual precipitation. Transactions of theAmerican Geophysical Union 39:1076–1084.

Loureiro, N. S., and M. A. Coutinho. 1995. Rainfall changesand rainfall erosivity increase in the Algarve (Portugal).Catena 24:55–67.

Luk, S. H., and J. Whitney. 1993. Unresolved issues: Perspec-tives from China. Pages 89–99 in M. Barber and G. Ryder(eds.), Damming the Three Gorges. Earthscan, London.

Milliman, J. D., and J. P. M. Syvitski. 1992. Geomorphic/tectonic control of sediment discharge to the ocean: Theimportance of small mountainous rivers. Journal of Geology100:525–544.

Milliman, J. D., Y. S. Qin, M. E. Ren, and Y. Saito. 1987. Man’sinfluence on the erosion and transport of sediment by Asianrivers: The Yellow River (Huanghe) example. Journal ofGeology 95:751–762.

Oldfield, F., P. G. Appleby, and R. Thompson. 1980. Palaeoeco-logical studies of lakes in the Highlands of Papua NewGuinea. I. The chronology of sedimentation. Journal ofEcology 68:457–477.

Phillips, J. D. 1993. Pre- and post-colonial sediment sourcesand storage in the Lower Neuse Basin, North Carolina.Physical Geography 14:272–284.

Ruddiman, W. F., W. L. Prell, and M. E. Raymo. 1989. LateCenozoic uplift in southern Asia and the American West:Rationale for general circulation modelling experiments.Journal of Geophysical Research 94:18,379–18,391.

Smil, V. 1987. Land degradation in China: an ancient problemgetting worse. Pages 214–222 in P. Blaikie and H. Brookfield(eds.), Land degradation and society. Methuen, London.

Smil, V. 1993. China’s environmental crisis: An inquiry intothe limits of national development. M. E. Sharpe, New York.

Sneyers, R. 1992. Use and misuse of statistical methods for thedetection of climatic change. Report on the informalplanning meeting on statistical procedures for climatechange detection. World Climate Data Programme Series20, Annex 3, WMO, Geneva.

Summerfield, M. A., and N. J. Hulton. 1994. Natural controlsof fluvial denudation rates in major world drainage basins.Journal of Geophysical Research 99:13871–13883.

Trimble, S. 1981. Changes in sediment storage in the CoonCreek basin, Driftless Area, Wisconsin, 1853–1975. Science214:181–183.

Walling, D. E. 1983. The sediment delivery problem. Journal ofHydrology 65:209–237.

Walling, D. E. 1997. The response of sediment yields toenvironmental change. Pages 77–89 in D. E. Walling andJ.-B. Probst (eds.), Human impact on erosion and sedimen-tation (Proceedings of the Rabat Symposium S6, April1997), IAHS Publ. No. 245. International Association ofHydrological Sciences, Wallingford.

Walling, D. E., and B. W. Webb. 1983. Patterns of sedimentyield. Pages 61–100 in K. J. Gregory (ed.), Background topalaeohydrology. Wiley, Chichester, UK.

Walling, D. E., and B. W. Webb. 1996. Erosion and sedimentyield, a global overview. Pages 3–19 in D. E. Walling andB. W. Webb. (eds.), Erosion and sediment yield: Global andregional perspectives (Proceedings of the Exeter Sympo-sium, July 1996). IAHS Publ. No. 236. International Associa-tion of Hydrological Sciences, Wallingford.

Wilson, L. 1973. Variation in mean annual sediment yield as afunction of mean annual precipitation. American Journal ofScience 273:335–349.

Xu, J. X. 1994. Zonal distribution of river basin erosion andsediment yield in China. China Science Bulletin 39:1356–1361.

Yu, J. R., L. R. Shi, M. H. Feng, and R. H. Li. 1991. The surfaceerosion and fluvial silt in the upper reaches of ChangjiangRiver. Bulletin of Soil and Water Conservation 11(1):9–17 (inChinese).

Zhou, G. Y., and Z. Xiang. 1994. Sediment transport in theYangtze basin. In L. J. Oliver, R. J. Loughran and J. A. Kesby(eds.), Variability in stream erosion and sediment transport(Proceedings of the Canberra Symposium, December 1994).IAHS Publ. No. 224. International Association of Hydrologi-cal Sciences, Wallingford.

Zhu, A. G., C. H. Lin, and H. S. Zhu. 1993. Research on thelaws of sediment yield in the middle and upper reaches ofthe Wujiang River. Journal of Soil and Water Conservation7(3):53–57 (in Chinese).

Sediment Yield in Upper Yangtze 709