Xian-Lin Luo　Eddy Y. Zeng　Rong-Yao Ji　Chao-Pin Wang

　　Introduction

　　The Pearl River Delta (PRD), located in South China and adjacent to the South China Sea (Fig.1), has become one of the fastest economically growing and urbanizing regions in China and around the world during the Last 20 years. Extensive anthrepogenic activities have brought about numerous environmental concerns within the PRD. In partic- ular, river water contamination has been a subject of many research efforts (Ho and Hui, 2001; Mai et at+, 2002). How-ever, an important issue that has been largely neglected is the Large-scale dredging and excavation of river sand, which in turn would affect the morphology of river channels within the PRD and related geomorphic processes. Consequently, it would result in substantial hydrological deformation (Luo et at., 2002).

　　   AS reported by Lagasse and Winkiey (1980), more than 4.26~ 107 metric tons of sand and gravel, were removed from the upper Mississippi River basin in 1950, and by1960 the production had almost doubled to 8.35×107 met-ric tons. These authors underscored the importance of the coarser fractions of the bed material as a bed surface armorLayer to the stability of river systems. They also concludedthat gravel dredging in the lower Mississippi River could cause major changes in bed degradation, resistance to flow, lood heights, groundwater tables, and aquatic habitats.

　　   Some researchers applied mathematical models to eval-uate river channel changes and the responses of fluvial pro-cesses to in-stream mining (Chang and Stow, 1989; Cotton and Ottozawa-Chatupron, 1990). Other researchers studied the behavior of mining pits (Gill, 1994; Lee et at., 1993; Li and Simons, 1979). Through hydraulic model tests and field investigations along the Danube River in Hungary, Kornis and Laczay (1988) suggested that a dredging pit must not be made shorter than the channel width nor wider than about half the channel width to avoid undesirable flow disturbance and bed erosion. Mas-Pla et al. (1999) examined the effects of gravel mining on the aquifer-river system in the coast of the Baix Fluvi~ area (NE Spain) and showed that in-stream mining caused a decline of the water-table head resulting in salty-water intrusion from the river into the aquifer.Han et aL (2005) scrutinized the impact of sand excavation on the hydrol.ogical regime in the East River network of the PRD, and the results showed that changes in the river regime parameters, such as lowering of water levels,alteration of surface water and groundwater recharge,

　　Figure 1　Map of the Pearl River Delta and its river network. N1-N15, W1-W12, and EI-E7 are the channels or reaches of theNorth River network, West River network, and East River network,respectively; Numbers I-8 indicate the eight river runoff outletsconnecting the Pearl River Delta to the South China Sea, include Humen, Jiaomen, Hongqimen, Hengmen, Modaomen, Jitimen,Hutiaomen, and Yamen eoin from east to west. The triangn symbols represent the locations of the hydrologic stations.

　　Figure 2　Annual budgets of water and suspended sediment within the Pearl River Delta (Pearl River Water Resources Commission,1991). The numbers in the parentheses are values for suspended sediment. The fluxes 1-8 indicate the eight major rivenne runoffoutlets shown in Fig.1.

　　and re-balance of salt and fresh water in tidal regions, couldbe induced by in-stream mining.

　　   The previous studies mentioned above have mainly fo-cused on a single river channel or limited reaches of one riv-er where the scale of sand excavation was relatively small and the resulting impact on river hydrology and morphology and human habitats was minor compared to that within thePRD. On the other hand, sand excavation has been occurring in almost all 324 rivers and streams (a total of 1,600 km stretch) around the PRD for more than 20 years. Theseuncontrolled practices have accelerated riverbed degrada-tion, leading to an atteration in the hydrodynamics of the entire river network (Jia et al., 2002; Luo et al., 2002). Sofar, very few reports are available in the literature assessing the effects of anthropogenic activities on hydrological changes of the PRD. In other regions, detailed sand excava-tion records are available (Pinter et aL, 2004). On the other hand, there has been no official record of sand excavation in

　　the PRD, which makes it difficult to assess the impact of sand excavation activities. Nevertheless, the present study,for the first time, surveyed the sand excavation activities and analyzed the correlation between the magnitude of sand excavation and the geometrical variation in riverbed. The results were used to evaluate the impact of sand exca-vation on hydrological variabl.es and other aspects, such as water levels and divided flow ratio (DFR) between various water systems. Responses of coastal hydrology, morphol-ogy, and sedimentology to sand excavation were also investigated.

　　The hydrological system in the Pearl RiverDelta

　　The Pear［ River is one of the tartest rivers in China, and the Pear［ River network is main［y comprised of the West River, East River, and North River (Fia. 1), with a total watershed area of 453,690 km2. The West River, the main stream of the Peart River network, is 2214 km in tength and merges with the East River and North River at the tower reach of the PRD, forming a depositional detta plain of 9750 km2.The deltaic region contains about 324 streams constituting a complicated river network within the middle part of Guangdong Province (Pear［River Water Resources Commis-sion, 1991). Based on the origins of freshwater, the PRD net-work is categorized into three subsystems, i.e., West Rivernetwork (W), North River network (N), and East River net-work (E) (Fig.1; numbers represent various reaches within a specific network).

　　Approximately 3.26x1011 m3 of freshwater annually dis-charge from the PRD into the South China Sea via eight ma-jor outlets (Pearl River Water Resources Commission, 1991)(Fig.2). The freshwater flow from the West River amountsto 2.3×1011 m3, or about 70% of the total. An analysis of the hydrological data acquired at the Gaoyao and Shijiao stations (Fig.1 ) from 1960 to 1997 indicates that the peren-nial variability of the Pearl River runoff flowing into the PRDwas considerably small (Luo et aL, 2002). On the other hand, the present study suggests that the hydrology within the PRD has changed substantially; therefore, the changes appear to originate from the delta itself rather than fromexternal sources.

　　Anthropogenic activities

　　Trends of population growth and urbanization

　　Within the PRD economic zone of 4.16×104 km2, the num-ber of permanent residents reached 40.77 millions in 2000(excluding those in Hong Kong and Macao), amounting to apopulation density of 980 personsl km2. Currently, the delta houses nine cities and nearly 400 towns. The City of Guangz-hou is the capital of Guangdong Province~ as welt as the cen-ter of culture, economy, and politics in the PRD. The population of Guangzhou increased from 2.77 millions in 1978 prior to the economic take-off to 9.94 millions in 2000 (Guangdong Statistical Bureau, 2001). Before 1976,there was only one skyrise higher than 100 m (Baiyun Hotel with a 113-m height), but more than 360 skyscrapers and 7000 multistory buildings (more than 18 floors) have been built since then. Floor space in Guangzhou increased from less than one million mZ/year before the economic reform (prior to 1978) to 1.58 million mz/year in 1978, 5.37 million mZ/year in 1985, 11.2 million mZ/year in 1995, and 22.3 mi-lion mZ/year in 2000 after the cities of Huadu and Panyuwere merged with Guangzhou.

　　   There are about 400 towns in the PRD, where the distance between two towns is typically less than 10 km. The town-based management system used to serve as anagricuttural administrative center for each district beforethe economic reform initiated in the early 1980s. Currently,the main economic sectors inctude a variety of manufactur-ing industries, greatty stimulating the manufacturing activ-ities and poputation growth. Heavy construction activities have substantiatty increased the demand for sand and other building materials.

　　Activities of river sand excavation

　　Background. Large-scare excavation of river sand was initi-ated in the mid 1980s, starting around the river channelsnear urban centers and of the East River network. This is be-cause the East River is dose to Hong Kong, and subsequently has long been the source of sand and gravel exported to Hong Kong. The excavating activities boomed between the early 1990s and 1995 with almost art river channets in the re-gion impacted. Such activities were somewhat stowed down by the Asian financial crisis in 1997 due to a cooling real. es-tate market. Within the PRD, sand excavation has beenmostly undertaken by smart private enterprises without reg-utatory permission. From the mid 1980s to 2000, there was no controt on the volumes of sand and graver excavated

　　from river channels. A large number of excavation boatscarried no permits, and the amount of sand and graver exca-vated was dictated by the market demand. Since 2000, the public and governmental agencies have expressed deep con-cerns  about  the  negative  environmental,  consequences resul.ting from riverbed damages because of intensive sandexcavation. Authorities began to estabtish regutations and pol.icies to control sand excavation and prohibit such activ-ities within certain important channels. As a resul.t of the tightened control and availability of sand, many excavating zones are shifted toward upstream of the del.ta. However, asmall number of illegal sand excavations currentl.y remainsactive in the PRD (Zhang and Yue, 2006).

　　    Field surveys of river sand excavation. No records ofsand excavation have been maintained by authorities, andmany excavating boats are not even registered. In 1998 and 2003, we conducted four surveys on the number of xcavating boats in the field. In August and December of1998, two surveys were conducted within the West Rivernetwork and North River network. We estimated that there were 306 (175 in the West River network and 131 in the North River network) and 148 (95 in the West River network and 53 in the North River network) excavating boats in Au- gust and December, respectively. The average amount of sand excavated by each boat was about 1.0×103 m3/day,and the total availabl.e working days were estimated to be 200 per year, 2/3 of which occurred in summer (wet weath-er season) and 1/3 in winter (dry season). The total amount of sand excavated in 1998 from the river channels of the  West  River  and  North  River  can  be  estimated  as (306×0.667+148~0.333) ×200× 1000 m3 = 5.1×107 m3.

　　    We conducted fietd surveys on the number of excavating boats in the East River network in July 2003, and atso sur-veyed the sand excavating activities in the West River net- work and North River network in October 2003. The totalnumber of excavating boats in the PRD was 187 in 2003(84 in the West River network, 34 in the North River net- work, and 69 in the East River work), a targe decrease in  the West River network and North River network comparedto that in 1998. On the other hand, bigger excavating boats with more capacities have been increasingly used. A large excavating boat may be able to excavate 5000-6000 m3 of sand per day in the West River network. As a result, the average amount of sand excavated by each boat was about2500 m3/day in the West River network, and 1500 m3/day in the North River network and East River network. The totalavailable working days were estimated to be 200 per year. Therefore, the total amount of sand excavated from thePRD was estimated at 7.3×107 m3 in 2003.

　　   Use of channel hypsonaphy to determine excavated sand volumes. In addition to the field surveys which could only be considered rough estimates, we used channel hyps-ography (with a chart scale of 1/5000) provided by the Guangdong Waterway Bureau to calcul.ate the amount of sand excavated from the channels of the PRD. We divided a river cross section into 1-km intervals and calculated the difference between the areas of each interval (cross sec-tion) before and after sand excavation on the channel hyps-ography. The area difference was mul.tiptied by the river.

　　Length to derive the expanded river volume due to sand excavation. The results (Table 1 ) indicate that the total ex-anded river capacity, with a reach of 813.8 km as verified by existing charts, was 8.7×108 m3, resulting in an average annual expanded volume of 7.3×107 m3. This number is essentially the same as the estimate based on the field sur-veys in 2003 (7.3×107 m3). In reality, the total amount of sand excavated is probably more than the estimated number8.7×108 m3, because no charts are available for the chan-nels of the West River and North River after 1999. There are also no charts available for numerous small tributarieswith a total length of more than 780 kin, but an estimate of 1.5×108 m3 of sand excavated before 1999 was obtained for these tributaries (Luo et aL, 2002). In addition, about 8-10×106 m3 of incoming sedimentary loads are transportedfrom upstream areas and deposited into the river channelsof the PRD every year (Liao and Fan, 1981). This amount was also not included in the calculation.

　　Changes in river channel geometries

　　Table 2 shows the geometrical changes of the main channelsof the West River network, North River network, and EastRiver network based on the comparison of historical channel hypsography. Before the surge in large-scale excavation, a certain level of deposition or erosion occurred at various Locations at times, as shown by the non-underlined numbers in Table 2. However, large-scale dredging and excavation have greatly changed the geometries of all river channels.The underlined numbers indicate the geometrical changes in river channels during the period of intensive excavation.The cumulative change values of the channel volumes in a 1-km stretch (columns 3 and 8 of Table 2) have been expanded by 48.0(W5)-206.4(W4)×104 m3/km, 26.5(N6)-210.5(N4)×104 m3/km, and 67.1(ES)-360.B(E2)×104 m3/km for the West River, North River, and East River, respectively. Simi-larly, the downcutting depths ranged from 0.59 to 1.73 m,0.34 to 4.43 m, and 1.77 to 6.48 m (columns 4 and 9 of Table2) for the West River, North River, and East River,respectively.

　　   The rivers and streams around the PRD are basically be-sieged by man-built banks. Consequently, the river widths

　　are constrained and have not been altered much after the1970s. As the average depth increased, the average width-to-depth ratio (WDR) decreased because the average width has not changed much due to the constraints by the man-built banks. The underlined numbers in Table 2 (columns 5and 10) show that the average WDR for all river channelshave substantially decreased due to sand excavation.

　　     The geometrical dimensions of the West River channelsdid not show any significant change from 1952 to 1991 (Wlto W7 in Table 2). This indicates that sand excavationmainly occurred after 1991. However, there was sand exca-vation in the N3 reach of the North River prior to 1990 and perhaps in N4 reach prior to 1983 because they are close to Guangzhou. At the N4 reach, the channel depth was in-creased by 0.95 m during 1983-1995 and by 3.27m during 1995-1999, suggesting that the peak of sand excavation oc-urred after 1995. Sand excavation also occurred prior to198g in the E3-5 reach of the East River, probably due to its proximity to the city of Dongguan, one of the earliest development zones in the PRD. After the sand excavation activities stopped at the E4 and E7 reaches of the East River,the channels near the river mouth became depository as re-flected in the decreases of the channel volumes by 3.g×104 m31km   (1997-2002) and 43.5×104 m31km(1997-2003) for E4 and E7, respectively (Table 2). The sand excavation activities reflected by the changes in river chan-nel geometries are consistent with those from field surveys.

　　Figure 3　Longitudingal section of the riverbed of the East River(Fig.1）and the dredging pits. The date in the figure were provided by the Sanshui Hydrologic State.

　　     As a result of the large-scale sand excavation, the longi-tudinal section of riverbed was also impacted in the PRD.For example, the longitudinal section of riverbed of the East River from the upstream to the outlet remained basically unchanged between 1964 and 1988 (Fig.3), but the longitu-dinal section in 2002 became uneven due to sand excava-tion. Numerous pits have been generated along a more than 100-km longitudinal, section of the East River from the upstream to the outlet and the riverbed slope in the channel of 80 km long from the outlet upward has been reduced to nearly zero (Fig.3).

　　Changes to the river water levels

　　 Due to the sand excavation and downcutting of the river-bed, the river water levels around the Pearl River Delta havefall.en significantl.y. For example, the flood section at the Sanshui station was I.ittle changed from 1984 to 1988, buthas been enlarged after 1988. The furthest depth of the riv-erbed has been cut down more than 7 m from 1988 to 2005(Fig.4), suggesting that large-scale excavation of river sandwas initiated after 1988.

　　Figure 4　Temporal changes of the cross section at the Sanshui station(Fig.1).The data in the figure were provided by the Sanshui Hydrologic Stae.

　　Figure 5　Reationship between flood stage and discharge rate at the Sanshui station(Fig.1)

　　 Along with riverbed degradation the water Level at the Sanshui station has been reduced substantially after excava-tion for the same discharge amount (Fig.5). The flow-stage relationships for the flood events occurred after 1989 are shifted nearly in paraU.el from upper-left to lower-right cor-ner (Fig.5) going from 1989 to 2005. It shows that the water level was lower for the same amount of discharge or larger amount of discharge was needed to maintain the same water level in 2005 compared to 1989. The flood event data measured in the summers of 1989 to 2005 at the Sanshul sta-tion indicate that the flood stage levels were reduced by 2.45 m from 1989 to 2005 at a discharge rate of 3000 mS/s, by 3.21 m from 1992 to 2005 at 7000 mS/s, by 2.11 m from1994 to 2005 at 11000 mS/s, by 1.59 m from 1994 to 2005 at15,000 mS/s. The average water level has been reduced by 0.5-0.25 m/year since the occurrence of large-scale exca-vation of river sand. On the other hand, the river water levels in the PRD slowly rose at a rate of 0.008-0.069 m/year from the 1960s to the 1970s prior to large-scale sand exca-vation due to block-out of many small tributaries (Li, 1985).

　　Changes to the divided flow ratio

　　Upper part of the Pearl River Delta

　　The water flow rates at four stations, Gaoyao, Shijiao, Ma-kou, and Sanshui (Figs.1 and 6), on the upper part of thePRD were compared to elucidate the DFR information around the delta. Geographically, waters from the West Riv-er and North River enter the

　　Figure 6　Changes of divided flow ratios at upper stream f the Pearl River Delta.

　　delta via the Gaoyao station and Shijiao station, respectively. The two flows mix and redistribute by the Sixianjiao channel. (Fig.6), and a portion of the flow enters the West River network via the Makou station and the rest goes into the North River network through the Sanshui station. As shown in Tabte 3, the DFRbetween the Makou and Sanshui stations did not changemuch from the 1960s to the 1980s, with the average relative.

　　DFR at the Makou station varying between 84.7% and 86.2%of the total flow. On the other hand, during the early days of sand excavation (1990-1992), the DFR started to chanse with the relative DFR decreasing to less than ~82% at the Makou station and increasing to ~18% at the Sanshui station. From 1993, the relative DFR continued to rise at the Sanshui station, reached a maximum value of 26.2% in 1997, and maintained higher than 22% from 1997 to 2000. This means that for the smaller and narrower waterways of the NorthRiver network, the annual average discharse increased from1240 m3/s (between the1960s and the 1980s) to 2587 m3/s(between 1996 and 1998).Outlets of the Pearl River Delta Changes to the DFR have taken place not only at the upper part of the PRD, but also at almost all the chan-nels, due to imbalanced diggin~ depths. Currently, no DFR data are available to allow a quantitative analysis on the situation. The measured river DFRs during the middle flow stage at the eight major outlets(Fig.1) of the PRD in July of 1999 indicate that the DFRs at the four eastside outlets (Humen,Jiaomen,Hongqili and Henmen) was 61.1% in 1999, but was 54.4% in 1980s, and ncrease of 7.7%(Table 4).As a result, the relative flow rates declined at the four westside outlets (Modaomen, Jidimen,Hutiaomen and Yamen).This change is clearly reflective of the change in the DFRs between the North River and West River networks, because the runoffs from the North River and West River networks mainly flow through the four eastside and four westside outlets,respectively.For example,the North River network increased by 8.8% from the early 1980s to 1999(Table 3),close to the 7.7% increase at the four eastside four outlets.

　　Discussion

　　Evolution of riverbed processes

　　Anthropogenic activities have considerably changed the evotving river processes in the PRD. Before the burst of large-scare  sand  excavations  river  processes  were  con-trotted by gentle and shallow aggradation. An annual aver-age of sediment amount at 8-10×106 m3 (Liao and Fan,1981 ) aggraded in the river channels of the Pear［ River Delta and the water lever in the Pearl River network slowly in-creased by 0.008-0.069 m/year of race (Li, 1985). How-ever, river processes were reversed by severe riverbed erosion derived from large-scale dredging and excavation.More than 7~107 m3/year of sand has been dredged from the channets around the Pearl River Detta over nearly two decades, which is seven times more than the average sedi- mentation amount before excavation. The average waterlever has been reduced by 0.15-0.25 m/year at the Shan-shui station from 1989 to 2005, opposite to the stow rise ob-served before excavation. Apparent［y, the change in river morphotogy and hydrodynamics and geomorphic processes caused by sand excavation greatly exceeded that by natural processes. As a result, the riverbed process has evolved from a stow depository one to a rapid erosive one. Therefore anthropogenic activities are deemed as the dominant factor controlling the variability of the hydrodynamic and geomor-phic processes within the PRD over the fast 20 years.

　　Effects of chanes to the river water levels

　　Flood heights, and consequently the chance of flooding damages, have been significantly decreased due to the large-scale excavation in the Pearl River Delta. For exam-pie, two huge floods occurred in June of 1994 and 2005 (des-ignated  as  “94.6”  and  “05.6”,  respectively),  were supposed to be at the frequencies of 2% and 1%, respec-tively, of the flood recurrence as they arrived in the Pear［River Delta (at the Shanshui Station). However, the flood stages of “94.6” and “05.6” were lower by 1.28m (3.85-2.57=1.28) and 2.45 m (3.85-1.4=2.45) at a dis-charge rate of 3000 m31s compared to that of the flood event in 1989 which was a relatively smart flood at the time(Fig.5). The flood stage of “05.6” was tower by 1.59m(10.09-8.5=1.59) at 15,000m31s compared to that of “94.6”. As a result, there were no flooding damages inthe PRD during the events of “94.6” and “05.6”.

　　    However, the water levels during low flows, as welt asflood stages, have been significantly decreased by thelarge-scale excavation, which negatively influences the functions of waterworks for domestic, industrial, and farm-ing purposes and navigation. Some pump intakes and bot-toms of watertocks are 1 m higher than the water levelduring low flows and consequently these waterworks cannot work property. In addition, alteration in river depthshas also resulted in changed navigating conditions. Deeper river depths generally improve navigating conditions, but in many cases the navigable pathways around large dredging pits are disrupted. Based on an investigation of the bound-ary reach between channels N1 and N2 of the North River(Fig.1) in December of 2004, the water depth was more than 4.8m in the excavated navigation channel, down-stream, but lower than 1.5 m in the navigation channel up-stream section about lOOm away. During the one-day survey, about 60 cargo ships had to run aground in the up-stream shoaling reach because the water depths were too shallow.

　　Effects of changes in river channel geometries

　　The width-to-depth ratio (WDR) of the channels in the PRDhas generally declined due to channel dredging (Table 2).For example, the WDR was 9.93 in the E3 channel of the EastRiver in 1964 and 11.05 in 1972 (Table 2). Sediment deposi-tion was predominant in the channel during this period. On the other hand, the WDRs were 8.73 in 1988, 4.36 in 1997 and 3.26 in 2002. Small. WDRs increase the grade slope and instability of the riverbank. For examples, two incidentsof fatting dikes have occurred in the W4 channel of the West River. One of the incidents, occurred in July 2005, causedmore than ten bull.dings on the riverbank to collapse into water and nearly 1000 people from 300 families had to rel.o-cate (Lun et al., 2005). Sand excavation appeared to be mainly responsible for the incidents according to our long-term field surveys, despite a different opinion by the authority (Zhang and Yue, 2005).

　　Impacts of hydrological changes on estuarine and coastal environments

　　Hydraulic al.teration, decreased sediment transport to the ocean, and reduced riverbed height due to Long-term and extensive sand excavation in the PRD may have profoundimpacts on the estuarine and coastal environments. Brack-ish-water intrusion is one of the most notable events that affect the water supply for about 15 million residents living in the coastal, region of the PRD during the dry season almostevery year (Let et at., 2004). As shown in Fig.7, present-day

　　Figure 7　Brackishwater intrusion ranges in the 1980s an the present days within the Peart River Delta.

　　brackish-water intrusion (with salinity ranging from 250 rag/L to more than 3000 mg/L) resulting from extensive sandexcavation occurs in 10-20 km more upstream areas than natural  brackish-water  intrusion  in  the  past  (satin-ity = 3000 mg/L) (Yang, 1986). The brackish-water intrusionmay fast for half a year in dry seasons. The first time brack-ish-water intrusion occurred in September was in 2004(Zhang and Yue, 2004) white normally it was supposed to oc-cur from December to March. It was suggested then that dry weather was the most important factor for brackish-waterintrusion (Xu and Luo, 2005), and apparently Long-term

　　Figure 8　Temporal change of the subsurface bar at the Modaomen outlet(Fig.1)

　　and extensive sand excavation was largely ignored as a sig-nificant contributor to brackish-water intrusion.

　　     Our data suggest that there is erosion occurring in theestuarine and coastal regions. For example, the subsurfacebarrier bar of the Modaomen outlet always moved seaward before 1994 (Fig.8), whereas it began to move backward 1994-2002 (Wang et at., 2006). We speculate herein that this change is related to extensive sand excavation in theupstream channels (WI-W7) of the Modaomen outlet, assand excavation was intensive from the early 1990s to2004. Sand excavation has reduced not only the amountsof sediment transported to the coastal region, but alsothe DRF at the Modaomen outlet (from 28.3% in the 1980sto 26.8% in 1999; Table 4).

　　Effects of changes to the DFR

　　Water shortage has been a local issue in the areas surround-ing the North River network. Sand excavation has beenimposing a positive effect on the situation. Changes to the

　　DFRs within the channels of the PRD have caused redistribu-tion of water resources in the river network. Discharge of the North River network (the channel at the Shanshui sta-tion) has increased by ~80%, from 124Clm3/s (1960s-1980s) to 2253 m3/s (1996-1998) annually (Table 3). Thismeans that additional 32 billion m3 of water is supplied to the North River network annually to support rapid economicand population growth in the region. However, these water resources have not been sufficiently utilized. The addedamount of water, on the other hand, has not caused anyflood damages, because the channel volume of the NorthRiver network has increased by 130-220% due to sand exca-vation, larger than the increased amount of water flowinginto the system.

　　Acknowledgement

　　This research was supported largety by the Science and Technology Program of the Ministry of Communications of China (Project No. 95-5-01-53). Support of E.Y.Z. by the“One Hundred Talents” Program of the Chinese Academy of Sciences is also appreciated. The authors would like to thank Q.-S. Yang, L.-W. Jia, J.-X. Peng, Y.-T. Chen, Z.-R.Luo, and G.-R. Yang for assistance in field surveys and data compilation, analysis, and interpretation. The authors ben-efited greatly from stimulating discussions with Daniel Coxwith the Department of Coastal and Ocean Engineering of Oregon State University, Oregon, USA. 

　　原載：Journal of Hydrology,2007,343:230-239.