Journal of Hydrology

Transcription

1 Journal of Hydrology 388 (2) Contents lists available at ScienceDirect Journal of Hydrology journal homepage: Variability of effective discharge for suspended sediment transport in a large semi-arid river basin Yuanxu Ma a, *, He Qing Huang a, Jiongxin Xu a, Gary J. Brierley b, Zhijun Yao a, ** a Institute of Geographic Sciences and Natural Resources Research, China Academy of Sciences, A Datun Road, Chaoyang District, Beijing, China b School of Environment, University of Auckland, Private Bag 929 Auckland, New Zealand article info summary Article history: Received 3 February 2 Received in revised form 4 May 2 Accepted 8 May 2 This manuscript was handled by K. Georgakakos. Keywords: Effective discharge Class intervals Flow duration Flow regime Suspended sediment concentration Wuding basin The variability of effective discharge is analysed for three geomorphological zones (gullied hilly loess, valley hill loess and eolian sand) in the Wuding basin, China, based on mean daily flow discharge and mean daily suspended sediment discharge from 959 to 969, a period when human disturbance in this catchment was less intensive. A modified approach to the determination of discharge class intervals is developed, framed in terms of equal arithmetic intervals of the standard deviation S for all the discharges, such as S,.75S,.5S, and.25s. The average flow duration of effective discharge in the river basin ranges primarily from.26% to 3.6% in the two loess regions (corresponding to large flood events), and from 8.75% to 9.5% in the eolian sand region (corresponding to low or moderate flows). The average flow duration of effective discharge is significantly influenced by the size of class intervals and by characteristics of the flow and sediment regime. Using the most appropriate class interval of.25s, the average flow duration of effective discharge is about.26% in the two loess regions (other than.4% at Hengshan), but in the eolian sand region it reaches 24.5% at Yulin and 52.66% at Hanjiamao, respectively. Histograms of suspended sediment transport indicate that there is a bimodal dominant discharge for suspended sediment transport, with one peak in the range of low flows and the other in the range of large floods. Drainage density and specific sediment yields are lower in the eolian sand region, where effective discharge events occur more frequently and suspended sediment concentration is much lower than that carried by events of the same discharge in the loess region. In contrast, drainage density is higher in the two loess regions, where infrequent hyperconcentrated flows generate high specific sediment yields. Effective discharge differs significantly from bankfull discharge across the whole Wuding basin. Ó 2 Elsevier B.V. All rights reserved.. Introduction As alluvial channels respond to variations in water and sediment discharges, channel shape is the product of a range of flow conditions (Osterkamp, 24). Despite this assertion, determination of a characteristic discharge has frequently been preferred in river channel rehabilitation design and environmental management (Shields et al., 23; Vogel et al., 23; Doyle et al., 25; Schmidt and Morche, 26). Many scientists have argued that in the long-term, a characteristic discharge within a complete flow duration curve produces channel dimensions that reflect steady flow conditions (e.g. Schaffernak, 922; Inglis, 949; Federal Interagency Stream Restoration Working Group (FISRWG), 998). This characteristic discharge, which is considered to play an important role in forming and maintaining a stream s morphology or sediment transport, is referred to as the channel-forming discharge or * Corresponding author. ** Corresponding author. addresses: mayuanxu@pku.org.cn (Y. Ma), yaozj@igsnrr.ac.cn (Z. Yao). dominant discharge (Simon et al., 24; Crowder and Knapp, 25; Lenzi et al., 26). In recent decades, there has been debate over whether bankfull discharge (Q b ), effective discharge(q e )ora specific recurrence interval flow (Q ri ) is the best measure of channel-forming or dominant discharge (Wolman and Miller, 96; Benson and Thomas, 966; Pickup and Warner, 976; Andrews, 98; Ashmore and Day, 988; Nash, 994; Emmett and Wolman, 2; Simon et al., 24; Crowder and Knapp, 25; Lenzi et al., 26; Gomez et al., 27). Among many factors, this uncertainty reflects the lack of specificity in the definition of terms, along with indiscriminate applications to streams of differing sizes, flow characteristics and sediment discharges along the spectrum from ephemeral to fully perennial systems. Bankfull discharge, defined as the flow stage at which river channel flow just fills the channel to the top of the banks (Williams, 978), is considered to have morphological significance because it represents the boundary between the channel and the floodplain (Lenzi et al., 26). Dunne and Leopold (978) described bankfull discharge as the most effective flow in forming and/or maintaining average channel dimensions. In a sense, bankfull /$ - see front matter Ó 2 Elsevier B.V. All rights reserved. doi:.6/j.jhydrol.2.5.4

2 358 Y. Ma et al. / Journal of Hydrology 388 (2) discharge marks the transition in hydraulic and related physical changes along most alluvial stream channels when flow increases from in channel to overbank conditions (Osterkamp, 24). However, there is great uncertainty in how to estimate bankfull discharge. Williams (978) outlined 6 ways to depict bankfull elevation, none of which can be used alone effectively and extensively. Xia et al. (2) compared four different methods of determining bankfull discharge in the lower Yellow and found that a method using a stage discharge relation from one-dimensional hydrodynamic model is of higher prediction accuracy than the other three methods. Many researchers have related flow events of a specific recurrence interval to bankfull discharge by using empirical relationships derived from stable reaches. Based on the annual peak discharge series, Q.5 has been found to be a good proxy of bankfull discharge (Dury et al., 963; Dury, 977; Leopold et al., 964; Dunne and Leopold, 978; Hickin, 968; Harman et al., 999; Odem et al., 999; Castro and Jackson, 2). Williams (978) showed that recurrence intervals for bankfull discharge vary widely ranging between and 32 years, although an average of.5 years is exhibited in the annual maximum flow series. Undoubtedly, this range results in part from observer s misidentification of notionally bankfull surfaces that are higher and lower Fig. a. Location of Wuding basin: (A) eolian sand region; (B) gullied hilly loess region; and (C) valley hill loess region. The numbers denote hydrometric stations:. Chuankou; 2. Dingjiagou; 3. Lijiahe; 4. Dianshi 5. Hengshan; 6. Qingyangcha; 7. Jingbian; 8. Zhaoshiyao; 9. Yulin; and. Hanjiamao.

3 Y. Ma et al. / Journal of Hydrology 388 (2) Fig. b. Geological map of Wuding basin. than the floodplain itself (Osterkamp, 24). Segura and Pitlick (in press) found that the flow duration of bankfull discharge in 32 snowmelt-dominated streams in Colorado and Idaho has scaling dependence on drainage area deriving from a broken power law function. Effective discharge was originally defined as a discharge or a range of discharges that are able to transport the largest portion of sediment load in the long-term (Wolman and Miller, 96; Andrews, 98; Lenzi et al., 26). In general terms, effective discharge duration occurs between.4% and 3% of the time (.5 days year )(Wolman and Miller, 96; Pickup and Warner, 976; Andrews, 98; Webb and Walling, 982; Nolan et al., 987). Nevertheless, Sichingabula (999) identified that the average duration of effective discharge for suspended sediment transport in the temperate Fraser basin varies in the range of.3 6.%. Ashmore and Day (988) also showed that the percentage of effective discharge for suspended sediment transport in Saskatchewan basin ranges from <.% to over 5%. Wolman and Miller (96) concluded that effective discharge for suspended sediment transport in perennial streams of the moist eastern United States approaches bankfull discharge and has a recurrence interval of 2 years. By extension, effective discharge at bankfull stage has been considered to be the key control upon channel formation and/or maintenance (e.g. Dury, 973; Andrews, 98; Carling, 988; Andrews and Nankervis, 995; Rosgen, 996; Simon et al., 24; Torizzo and Pitlick, 24). However, other studies have noted significant differences between effective and bankfull discharges (e.g. Benson and Thomas, 966; Pickup and Warner, 976; Ashmore and Day, 988; Whiting et al., 999; Chen et al., 26; Gomez et al., 27). Indeed, several researchers indicate that flow events of a single recurrence interval cannot be considered to be representative of effective or bankfull discharge for all streams because discharge values are influenced by basin morphology, drainage area, hydrologic regime and sediment transport mode (bedload and suspended sediment load) (Ashmore and Day, 988; Nash, 994; Whiting et al., 999; Phillips, 22; Lenzi et al., 26). Methods used to derive the values of flow discharge and suspended sediment load may impact greatly upon the ways in which data are interpreted. Although it is simple to compute effective discharge according to its original definition, the procedure contains empirical elements, wherein estimates of sediment load are derived for various discharge class intervals, the number of which is based purely on empirical judgment (see Biedenharn et al., 999; Copeland et al., 2; Crowder and Knapp, 25; Lenzi et al., 26). Recognizing serious shortcomings that have been identified in many of the methodological procedures used to assess effective discharge, this study develops a modified approach to analyse and compare the effective discharge for suspended sediment transport in three different geomorphological regions of the Wuding basin, a tributary of the Yellow in China (Fig. a). Physiographic setting and related attributes such as flow regime, sediment transport, specific sediment yield (SSY) and drainage density vary markedly in these geomorphological regions (Table ). The aims of this paper can be generalized as: (i) to present a modified approach to calculating effective discharge; (ii) to apply this approach to loess and eolian sand regions of China (areas that are under-represented in other studies); and (iii) to appraise the variability in the magnitude and frequency of effective discharge in these regions. 2. Study area Wuding in China drains a total area of 3,26 km 2 in the middle drainage basin of the Yellow. The length of the main stem is c. 49 km. The average bed slope is approximately.97%. This temperate/semi-arid region had an average annual precipitation of 45.9 mm for the period of study ( ) (Zhang et al., 23). Heavy storms occur during summer.

4 36 Y. Ma et al. / Journal of Hydrology 388 (2) Table Main characteristics of rivers in Wuding basin. Station name Chuankou Dingjiagou Zhaoshiyao name Wuding Wuding Wuding Xiaoli Heimutou Geomorphological region A (km 2 ) Q a ( 6 m 3 ) Q s ( 6 t) Q max (m 3 s ) Q min (m 3 s ) C max (kg m 3 ) C min (kg m 3 ) SSY (t/ (km 2 a)) Whole river basin 24, Whole river basin 23, Whole river basin 5, Lijiahe Gullied hilly loess , (2) region Dianshi Gullied hilly loess ,2 region Hengshan Lu Gullied hilly loess region Qingyangcha Dali Valley hill loess , (28) region Jingbian Lu Valley hill loess ,483 region Yulin Yuxi Eolian sand region < (5) Hanjiamao Hailiutu Eolian sand region Drainage density (km km 2 ) (and relative relief (m)) Notes: A-drainage area, Q a -mean annual runoff ( ), Q s -mean annual suspended load ( ), Q max -maximum mean daily discharge ( ), Q min - minimum mean daily discharge ( ), C max -maximum mean daily suspended sediment concentration ( ), C min -minimum mean daily suspended sediment concentration ( ); Drainage density from Chen (989); relative relief from Google Earth. Fig. 2. Aerial view of geomorphological regions in Wuding basin: (a) gullied hilly loess region; (b) valley hill loess region; and (c) eolian sand region. Wuding basin can be divided into three geomorphological regions: () gullied hilly loess; (2) valley hill loess; and (3) eolian sand (Figs. and 2). Three types of landforms make up the gullied hilly loess region: Liang (elongate loess hills in Chinese), Gullies and Mao (isolated hills in Chinese) (Fig. 2a) (Zhang et al., 23). The land surface is deeply dissected with a drainage density of about 5 7 km km 2. The loess thickness ranges from 5 to m. Gullies typically range in depth from 5 to 25 m. Jurassic and Neogene sandstones and silty sandstones are exposed in the headwaters of several tributaries (Fig. b). The valley hill loess region lies in the headwater basin of Wuding. The larger valleys and extensive dissection of Liangs into Maos indicate that this surface has been subjected to more intensive erosion than in the gullied hilly loess region (Fig. 2b). Gullies have been cut into bedrock 5 3 m below the loess surface. Cretaceous sandy gravel-stone and sandstone are exposed in the headwater basin. The north-west eolian sand region makes up by far the largest portion (54.3%) of the basin. Vegetation cover is sparse, limited to sporadic bushes. The region is covered by eolian sand with small areas of alluvium and loess (Fig. 2c). Bed incision, headward retreat and channel expansion are the primary fluvial processes in these highly sediment-charged suspended load rivers. In this study, effective discharge estimates are derived for hydrometric stations in the Wuding basin that have long- Table 2 Summary of hydrological data observed in Wuding basin. Station name Stream name Study period Years Data points Chuankou Wuding Dingjiagou Wuding Zhaoshiyao Wuding Lijiahe Xiaoli Dianshi Heimutou Hengshan Lu Qingyangcha Dali Jingbian Lu Yulin Yuxi Hanjiamao Hailiutu

5 Y. Ma et al. / Journal of Hydrology 388 (2) Table 3 Procedures used for hydrometric measurement and data processing. Attribute Water level Discharge Suspended sediment concentration Extrapolation of mean daily discharge and suspended sediment concentration Procedures During the dry season, when the water level is stable, it was measured once a day. During the winter freezing season, water level was measured once every 2 5 days. Water levels on unmeasured days were calculated from straight line functions that linked the preceding and subsequent days. During the flood season when flow level varied rapidly, water level was measured at least four times per day. Values were averaged as mean daily water level using the time-weighted method Discharge was calculated using the velocity area method. At least five verticals were spaced from one bank to the other. The depth in each vertical and the width between adjacent verticals was measured. Mean velocity in corresponding verticals was measured using one-five point method (Boiten, 23). The discharge of each subsection was calculated as the product of the average of the velocity, depth and width between the two adjacent verticals. The total discharge was the sum of the discharge of each subsection. For flood events, more than five measurements were made. The mean daily discharge is the time-weighted value of all measured discharge during a day Suspended sediment concentration was sampled whenever the velocity in each vertical was measured. The method for calculating suspended sediment transport rate in the whole cross-section is similar to procedures used in the discharge calculation. Average suspended sediment concentration is the suspended sediment transport rate divided by the total discharge in the whole cross-section. At least 7 measurements were taken during flood events. Mean daily suspended sediment concentration is the time-weighted value of all measured data points during a day. On days when the river was frozen, suspended sediment concentration was recorded as zero Mean daily discharge was usually obtained using a discharge stage relationship. Stage discharge relationships can be grouped into two categories: unique rating curve and non-unique rating curve. For stable channels, the stage discharge relationship is typically a unique rating curve. When the measurements are plotted on logarithmic paper, the stage discharge relationship was plotted as a straight line for the lowest water stages and is smoothly curved at high water stages. The rating curve is fitted using a power equation. Mean daily discharge is computed from the equation using available water level data. However, in channels subjected to intensive erosion or sedimentation, the discharge stage relationship is more irregular, generally expressed as a loop or a more complex shape. For these non-unique relationships, discharge stage relationships were differentiated into several groups, and appropriate curves of best fit were developed for each group. Mean daily discharge was extrapolated from these different rating curves a 3. b f(q) (%).5 f(q) (%) c. 7 6 d f(q) (%) 4 3 f(q) (%) Fig. 3. Examples of the frequency of mean daily discharge in each geomorphological region: (a) Chuankou, mouth of Wuding (class interval = m 3 s ); (b) Lijiahe, gullied hilly loess region (class interval =. m 3 s ); (c) Hanjiamao, eolian sand region (class interval =. m 3 s ); and (d) Qingyangcha, valley hill loess region (class interval =. m 3 s ).

6 362 Y. Ma et al. / Journal of Hydrology 388 (2) term mean daily discharge and sediment concentration records ( ; Table 2). The location and key attributes of these stations are shown in Fig. a and Table, respectively. Dingjiagou and Chuankou are in the lower reach of Wuding with the latter station being close to the confluence with the Yellow. These stations summarize flow discharge trends across the basin as a whole. However, as the majority of the lower basin falls within the gullied hilly loess region, these two stations are considered to be representative of this geomorphological region (Table ) (Wang, 27). Three other stations also fall within this region: Lijiahe (Xiaoli ), Dianshi (Heimutou ) and Hengshan (Lu ). The Xiaoli and Heimutou rivers lie entirely within the gullied hilly loess region, whereas the upstream part of Lu lies within the eolian sand and valley hill loess region. Two stations are located close to the downstream margin of the valley hill loess region, namely Jingbian on Lu and Qingyancha on Dali (Fig. a). Although Zhaoshiyao lies within a gullied hilly loess region, the water and suspended sediment load originate primarily from the eolian sand region (Wang, 27). Two further sites are located in the eolian sand region: Yulin in the upper reach of Yuxi and Hanjiamao near the mouth of Hailiutu. Flow and sediment discharge data vary markedly in the three different geomorphological regions (Table ). Specific sediment yield and drainage density are much lower in the eolian sand region. In addition, rivers in this region have much small variation in mean daily suspended sediment concentration relative to the two loess regions. Since 97, application of large scale soil conservation measures across the basin has greatly altered runoff and sediment yield regimes (Xu, 24). This study is concerned only with the variation of effective discharge for suspended sediment transport in the basin prior to the implementation of these measures, from 959 to 969, assessing the variability in effective discharges over a large basin during a period of less intensive human disturbance. 3. Data and methods 3.. Data Regular daily hydrometric measurements of water stage, flow discharge, and suspended sediment concentration were made by the Yellow Conservancy Commission during , following national standards issued by the Hydrological Bureau, Ministry of Water Resources (the former Ministry of Water Conservancy and Electric Power of China, 962). Although data with a higher temporal resolution may be more desirable as they more closely reflect natural flow regime and process (Lenzi et al., 26), previous studies have indicated that the dataset used here has a sufficient size for calculating effective discharge (Biedenharn et al., 999). Procedures used to undertake the hydrometric measurement and data processing are summarized in Table 3. SSC (kg m -3 ) SSC (kg m -3 ). (a) Chuankou. (c) Hanjiamao SSC (kg m -3 ) SSC (kg m -3 )... (b) Lijiahe. (d) Qingyangcha Fig. 4. Relationship between mean daily suspended sediment concentration (SSC) and mean daily discharge (Q) in rivers in different geomorphological regions: (a) mouth of Wuding, (b) gullied hilly loess region, (c) eolian sand region, and (d) valley hill loess region.

7 Y. Ma et al. / Journal of Hydrology 388 (2) Table 4 The size of class intervals (CI) and numbers of discharge classes (N) used in the computation of effective discharge. Station name Range of discharge S.75S.5S.25S CI (N) CI (N) CI (N) CI (N) Chuankou (3) 49.95(4) 33.3(6) 6.65(9) Dingjiagou (28) 25.9(37) 7.6(55) 8.53(9) Zhaoshiyao (8).72(23) 7.5(35) 3.57(69) Lijiahe (27) 3.32(36) 2.22(54).(7) Dianshi (36) 2.5(47).43(7).72(4) Hengshan (25) 3.66(33) 2.44(49).22(97) Qingyangcha (4) 2.7(54).8(8).9(62) Jingbian (24).7(32).4(48).57(85) Yulin (3) 6.99(42) 4.66(62) 2.33(24) Hanjiamao (47) 2.27(63).5(94).76(87) N refers to the number of class intervals in which discharge events occur (i.e. any of the equally sized high flow class intervals in which no events occurred in the period of record are not included in the determination of N) Method The Wuding and its larger tributaries transport very large amounts of fine-grained bedload. The treatment undertaken here is incomplete because bedload is not considered. Rather, emphasis is placed upon the suspended sediment load, which accounts for a significant majority of the total sediment load. Although many studies assume that the frequency of the logarithm of mean daily discharge is log-normally distributed (Stedinger et al., 992; Nash, 994; Doyle et al., 25), the flow frequencies (f(q)) at stations in the Wuding basin do not show a log-normal distribution (Fig. 3). Also, the suspended sediment rating curves at these stations are not characterized by a power function, as shown in the graphs of mean daily suspended sediment concentration (SSC) and mean daily flow discharge (Q)(Fig. 4). These mean that the traditional approaches used to derive effective discharge, such as the procedures described by Nash (994) and Doyle et al. (25) for example, cannot be used directly at these stations. Nevertheless, the total suspended sediment load of all sample points falling within a class interval can be determined by direct computation, thereby avoiding systematic under- or over-estimation of sediment load caused by using power law predictions. As noted in previous studies, the choice of the size of class interval (CI) or the number of flow discharge classes (N) has generally been empirical (Pickup and Warner, 976; Andrews, 98; Lenzi et al., 26). Typically, Yevjevich (972) stated that the class interval of flow discharge should not be larger than S/4, where S is the standard deviation of flow discharge for the sample concerned, and that the number of classes should be between and 25, depending on the sample size. Using S/4 as the class interval, the number of classes at all ten stations in the Wuding basin exceeds 25 (Table 4), indicating that Yevjevich s two standards are not met at these stations. Furthermore, Biedenharn et al. (999) and Crowder and Knapp (25) argued that each class interval should contain at Table 5 Variation of flow discharge and suspended sediment concentration in Wuding basin. Station name R ed C vd R es C vs Chuankou Dingjiagou , Zhaoshiyao , Lijiahe Dianshi Hengshan , Qingyangcha Jingbian , Yulin Hanjiamao Means infinitely large. least one flow event. Following their suggestion, the number of discharge classes in most cases in the Wuding basin would be <, because the differences among the high flow discharges at these stations are generally very large. For example, the difference between the largest and second largest discharges (respectively 97 m 3 /s and 24 m 3 /s) at Chuankou station is nearly 7 m 3 / s, which makes the number of discharge classes no larger than 4. This clearly fails to satisfy the requirement of Crowder and Knapp (25) s method. To overcome the disadvantages of the methods proposed by Yevjevich (972) and Crowder and Knapp (25), this study assumes that the variation of each flow discharge class varies within the range between and S, where S is the standard deviation of all flow discharge observations. Within the range, equal arithmetic intervals of S,.75S,.5S, and.25s are used to divide flow discharge records into classes. This enables an analysis of the effect of interval width on magnitude-frequency analysis so that a suitable interval can be determined among the four. For simplicity, the representative discharge of each class is considered to be the midpoint of the corresponding interval. Then, the total suspended sediment load (SSL) transported by flow discharge in each class interval is calculated by summing the suspended sediment load of all sample points that fall within the corresponding class interval. This integrating approach can be considered equivalent to the approach described by Crowder and Knapp (25). Finally, the suspended sediment loads in all corresponding classes are calculated and plotted in a histogram against the representative discharges. The flow discharge corresponding to the peak of suspended load in the histogram is then determined to be the effective discharge. 4. Results 4.. Flow variability in the Wuding basin Flow and sediment regimes may vary significantly in different physiographical settings, even if the annual precipitation is similar. The ratio of maximum mean daily discharge to minimum mean daily discharge (R ed ) and the coefficient of variation (C vd ) (the ratio of the standard deviation of all discharges to their mean value; Yevjevich, 972) for the three geomorphological regions in the Wuding basin are shown in Table 5. Values of C vd indicate that flow variability is notably smaller in the eolian sand region than in the other regions. This reflects the relative lower magnitude of extreme flow events in this region Effective discharge for suspended sediment transport The variability of effective discharge in the Wuding basin for is summarized in Table 6. Regardless of which class

8 364 Y. Ma et al. / Journal of Hydrology 388 (2) Table 6 Effective discharges for suspended sediment transport in Wuding basin. Station name S.75S.5S.25S Q e (m 3 s ) f(q) (%) Q e (m 3 s ) f(q) (%) Q e (m 3 s ) f(q) (%) Q e (m 3 s ) f(q) (%) Chuankou Dingjiagou Zhaoshiyao Lijiahe Dianshi Hengshan Qingyangcha Jingbian Yulin Hanjiamao s.75s.5 s.25s s.75s.5 s.25s SSL ( 8 kg) SSL ( 8 kg) s.75s.5 s.25s (a) Heimutou at Dianshi. s.75s.5 s.25s (c) Yuxi at Yulin SSL ( 8 kg) SSL ( 8 kg) (b) Dali at Qingyangcha (d) Wuding at Chuankou Fig. 5. Effective discharge curves for suspended sediment transport derived for different flow class intervals. interval is chosen, significant differences in effective discharge are exhibited in the three geomorphological regions of the basin. In the gullied hilly loess region and the valley hill loess region, the average flow duration of effective discharge ranges from.26% to 3.6%, which corresponds to large flood events. Despite the low frequency, the extremely high suspended sediment concentration of flood events plays a critical role in the determination of effective discharge. In the eolian sand region, the average flow duration of effective discharge varies from 8.75% to 9.5%. The higher frequency of low and moderate flow and lower suspended sediment concentration of flood events account for effective discharge occurring in the range of low and moderate flows. On the main stream at Chuankou and Dingjiagou stations, influenced by the combination of the two loess regions with the eolian sand region, the effective discharge is not significantly representative. This indicates that effective discharges in different regions are dominated by different flow and sediment regimes. Fig. 5 shows the response of effective discharge curve to the size of class intervals at the selected stations. Whichever class interval is used, not all class intervals contain at least one flow event. Discontinuities inevitably appear in the discharge frequency distribution. A comparison of the average flow duration of effective discharge from different sizes of class intervals indicates that the effective discharge for suspended sediment transport in the eolian

9 Y. Ma et al. / Journal of Hydrology 388 (2) SSL ( 8 kg) SSL ( 8 kg) (a) Yuxi at Yulin (c) Lu at Dingjiagou SSL ( 8 kg) SSL ( 8 kg) (b) Xiaoli at Lijiahe (d) Wuding at Chuankou Fig. 6. Suspended sediment transport histograms using the.25s class interval. sand region appears more sensitive to the size of class intervals. However, when the size of class intervals is reduced, effective discharge in rivers of the same region converges on similar values of flow duration (Table 6). Therefore, the smaller the size of class interval is, the more accurate the results are. Among the four class intervals of S,.75S,.5S, and.25s,.25s appears to be the most appropriate class interval, because effective discharges are approximately the same for all streams in the same geomorphological region when this class interval is used. At this class interval, effective discharge has an approximate flow duration of.26% for most of streams in the gullied loess region and about 24.5% or 52.66% as shown at Yulin and Hanjiamao stations in the eolian sand region. The effective discharge in the gullied loess region is in the range of higher discharges while in the eolian sand region it is in the range of low flow events. The effective discharges for transporting suspended sediments at Chuankou station on Wuding, at Dianshi station on Heimutou, at Lijiahe station on Dali, at Qingyangcha station on Dali and at Jingbian station on Lu all correspond to the largest floods Histograms of suspended sediment transport Based on the S/4 class interval, histograms of suspended sediment transport in the Wuding basin generally have multipeak forms (Fig. 6). These can be differentiated into four types. In the first type, effective discharge is within the range of low or moderate flows with a notable peak (Fig. 6a). These events have high flow frequency and small magnitude, with an average flow duration ranging from 24.5% to 52.66%. The other peaks reflect large suspended sediment loads during extreme flow events. Yulin, Hanjiamao, and Zhaoshiyao stations, which are primarily gauged in the eolian sand region, belong to this type. Stations of the second type have an erratic form, with an effective discharge dominated by events at the upper end of the discharge range (Fig. 6b). Although several peaks are common in low flow events, the peak of suspended sediment transport for effective discharge is also well-defined. In some cases, the largest measured flood is the effective discharge. This type contains stations located in the gullied hilly loess region and valley hill loess region. Effective discharges at the third type of stations are within the normal duration range of flow events but the large flood events during the period of suspended sediment transport are almost comparable to the effective discharge (Fig. 6c). These findings are similar to the results of Ashmore and Day (988). Dingjiagou station is a typical example of this type. The fourth type of stations has a form similar to the third type but effective discharges at this type of stations are dominated by flow events at the upper end of the discharge range (Fig. 6d). Stations in this type include Jingbian and Chuankou. The histograms of suspended sediment transport indicate that the suspended sediment transport in the Wuding basin is characterized by a bimodal dominant discharge, with one peak in the range of low flows and another during large floods.

10 366 Y. Ma et al. / Journal of Hydrology 388 (2) a 34 a bed stage (m) bed stage (m) Distance (m) Distance (m) b 34 b 24 bed stage (m) bankfull discharge.%flow %flow effective discharge Distance (m) bed stage (m) bankfull discharge effective discharge.%flow %flow Distance (m) Fig. 7. Channel cross-sectional profiles of Yuxi at Yulin: (a) channel crosssectional profile at different time interval and (b) channel cross-sectional profile and the corresponding bankfull stage, effective discharge stage and %,.% flow stage observed on May 6, 967. The toe of the bank is composed of some stones and the channel bed covered by a m thick layer of sand. The channel bed gradient is about.3%. Four measurements of cross-section made in 967 are shown in (a). Mean daily flow discharges during May 6 September 6 range from.72 to 288 m 3 s. Mean daily flow discharges during September 2 6 range from 45. to 288 m 3 s. These are larger than the effective discharge. The changes of crosssection during different time intervals reflect intensive scour and fill. Channel scour and fill make it difficult to identify the bankfull stage and the corresponding discharge in the eolian sand region. In order to compare effective discharge and bankfull discharge, the channel cross-sectional profile on May 6, 967 is used Adjustments in channel morphology Bankfull stage is determined through identification of bank inflection proposed by Navratil et al. (26). This allows comparison of bankfull discharge and effective discharge in relation to flow stage. Channel scour and fill make it difficult to identify bankfull stage and the corresponding discharge in the eolian sand region because cross-sections there adjust in a complex manner (Fig. 7). Regardless of challenges faced in the determination of bankfull stage, effective discharges appear at stages notably less than bankfull. Whether the effective discharge is assessed at the scour phase ( ) or at the fill phase ( ) at Yulin station, it occurs much less frequently than % of the time. Channel geometry is more stable at Lijiahe station (gullied hilly loess region, Fig. 8), but effective discharge also occurs significantly below bankfull stage, <% of the time. In this instance, the bankfull stage of about 9.6 m indicated by the channel morphology measured on is much higher than the stage of effective discharge, which is about 4.5 m. Fig. 8. Channel cross-section profiles of Xiaoli at Lijiahe (bed stage is elevation relative to base cross-section): (a) channel cross-sectional profile at different time interval and (b) channel cross-sectional profile and the corresponding bankfull stage, effective discharge-stage and %,.% flow stage observed on April 2, Discussion 5.. Variation of flow duration of effective discharge Results from this study show that the average flow duration of effective discharges in the eolian sand region differs significantly from the counterpart in the loess region in the Wuding basin. The effective discharges for suspended sediment transport in streams running through the loess region vary among extreme flow events. In some cases they represent the largest recorded flood discharges, with a percentage duration ranging from.26% to.27% (.9.98 days year ). In the eolian sand region, the percentage duration of effective discharges varies from 24.5% to 52.66% (26 92 days year ). These findings are similar to the results reported by Ashmore and Day (988) on streams in the Saskatchewan basin in Canada, in which both extreme and more frequent flow events are defined as effective discharges. However, results from this study differ considerably with some studies performed in other areas, where average effective discharge duration occurs between.4% and 3% of the time (.5 6 days year )(Wolman and Miller, 96; Pickup and Warner, 976; Andrews, 98; Webb and Walling, 982; Nolan et al., 987; Simon et al., 24; Segura and Pitlick, in press). This likely reflects the relatively dry climate, limited ground cover and high sediment availability in the Wuding basin conditions that differ notably from humid-temperate and/or humid-tropical set-

11 Y. Ma et al. / Journal of Hydrology 388 (2) Lijiahe Yulin SSC=67.77Q.68 R 2 =.49 Table 7 Sediment concentration for discharges of different flow durations in different geomorphological regions. Geomorphological regions Sediment concentration for discharges of different flow duration (kg m 3 ) % 5% % Gullied hilly loess region Valley hill loess region Eolian sand region SSC (kg m -3 ) SSC=.69Q.963 R 2 = Fig. 9. Relationship between suspended sediment concentration and discharge at Lijiahe and Yulin ( ). tings in which most previous analyses of effective discharges have been undertaken Variability of the relationship between flow variability and sediment concentration Findings from this study support the assertion of Wolman and Miller (96) that the more variable the flow regime is, the larger the percentage of total sediment load that is likely to be carried by infrequent flows. Runoff generation in the Wuding basin appears to exert a major impact on flow regime. In the eolian sand region, the high porosity of loose eolian sand at the land surface promotes high infiltration. Runoff is generated when rainfall exceeds a threshold. This reduces the occurrence of large floods (Wang, 27). As a result, the flow regime is dominated by low or moderate flow events, with limited variability in both annual and inter-annual flows. In the gullied hilly loess region and the valley hill loess region, soil crusts and low porosity reduce infiltration and accelerate the generation of surface runoff (Cheng et al., 27). As a consequence, floods generally have high magnitudes. These conditions, coupled with high sediment availability, the semi-arid climate and sparse vegetation cover, generate hyperconcentrated flows during extreme events (Xu, 998). In these instances, high magnitude flow events are the effective discharges. In contrast, such hyperconcentrated flows do not occur in the eolian sand area, despite similar physical conditions. The wind-blown sand that covers this area is relatively uniform in texture, with insufficient finegrained material in the water sediment mixed liquid phase. Flow capacity for transporting eolian sand is much lower than that produced by hyperconcentrated flows. Relationships between mean daily sediment concentration and mean daily discharge vary significantly between the gullied loess region and the eolian sand region (Fig. 9). In the gullied loess region, suspended sediment concentrations can exceed 3 kg/m 3 when flow discharges exceed a threshold. The maximum suspended sediment concentration at Lijiahe station on Xiaoli is around kg/m 3. In contrast, suspended sediment concentrations in streams are much lower in the eolian sand region. At Yulin station on Yuxi, the maximum suspended sediment concentration is <3 kg/m 3. To quantify the variation of suspended sediment concentration in the study area, parameters similar to those used to define flow regime have been derived, namely the ratio of maximum mean daily suspended sediment concentration to minimum mean daily suspended sediment concentration (R es ) during the study period and the coefficient of variation (C vs ) which is the ratio of the standard deviation of all mean daily suspended sediment concentrations to their average (Yevjevich, 972). The value of C vs in the gullied loess region is larger than in the eolian sand region (Table 5). During high flow events, high suspended sediment concentrations can result in high suspended sediment loads Sensitivity of effective discharges to the size of class interval Effective discharge is sensitive to the number of flow classes used (Crowder and Knapp, 25; Lenzi et al., 26). The flow duration of effective discharges in streams in the eolian sand region has a wider magnitude of variation than in the loess region (Table 6). In other streams in the gullied hilly loess region and valley hill loess region, effective discharges generally increase when the number of flow classes increases, but the flow duration changes less, with a value of about.26%. When different class intervals are adopted, the flow frequency of class intervals in the range of all mean daily discharges varies considerably. This indicates that effective discharges for suspended sediment transport in the eolian sand region appear more sensitive to the size of class intervals. This regional scale variability indicates that evaluation of effective discharges depends on many factors such as flow regime, suspended sediment concentration, computational approaches, and drainage basin morphology Effective discharge, dominant discharge, and channel morphology Several researchers provide evidence for the existence of a bimodal dominant discharge (Pickup and Warner, 976; Carling, 988; Phillips, 22; Lenzi et al., 26). On the one hand, frequent low flows below bankfull stage are able to transport quite a large proportion of sediment load and prevent significant channel sedimentation. On the other hand, large floods that approximate or exceed bankfull discharge are necessary to transport coarser bed sediments and maintain channel capacity. Histogram plots shown in Fig. 6 reveal a multi-modal dominant discharge for suspended sediment transport in this region, with a large portion of the total suspended sediment load being transported by both one or two peaks in the range of extreme flow events and several peaks at the lower end of flow events. These discharges can be regarded as dominant discharges for suspended sediment transport. Effective discharge events vary markedly in different geomorphological regions in the Wuding basin. Flow discharge and suspended sediment concentration are higher but more variable in the loess region than in the eolian sand region. Suspended sediment load for a given discharge is higher in the loess region than in the eolian sand region (Table 7). Events that occur % of the time approximate the maximum mean daily suspended sediment concentration in the loess region (roughly 7 kg m 3 relative to a maximum concentration of kg m 3 ), but in Yuxi (Yulin station in the eolian sand region) these events possess a relatively small proportion of the maximum mean daily suspended sediment concentration (roughly 45 kg m 3 relative to a maximum concentration of 265 kg m 3 ).

12 368 Y. Ma et al. / Journal of Hydrology 388 (2) Table 8 Landscape attributes of different geomorphological regions. Geomorphological regions Tributary valley cross-sections (derived parallel to the trunk stream) * Relative relief ** (m) SSY *** (t km 2 a ) Gullied hilly loess region 2 8, 5 7 Drainage density (km km 2 ) Valley hill loess region 28 6, 4 6 Eolian sand region 5 9 < * Measured from google Earth. ** Calculated from column 2. *** Zhang et al. (23). Effective discharge events occur more frequently in the eolian sand region than in the loess region. In neither region, however, does effective discharge equate to bankfull discharge or the.5 year flood recurrence interval, as suggested by Wolman and Miller (96) and subsequently confirmed by Andrews (98) and Simon et al. (24). In this area, effective discharge events are not the primary determinant of channel morphology, as they occur much less frequently than bankfull discharge events. From this, it is inferred that channel shape is controlled by a range of flow events, as all flows transport and sort sediments, thereby modifying channel morphology (Osterkamp, 24). Nevertheless, the non-accordance between effective discharge and bankfull stage determined in this study is consistent with results reported by Benson and Thomas (966), Pickup and Warner (976) and Gomez et al. (27). This highlights inherent dangers in the prescriptive use of a particular recurrence interval to determine effective discharges in the designation of channel maintenance flows and/or river rehabilitation planning. Although the sediment concentration per unit area is much higher in the loess region, the effective discharge events in the region are more than ten times frequent in the eolian sand region. Seemingly, pulsed hyperconcentrated flow events in the loess region rapidly transport high volumes of sediments. This accelerates the dissection of the land surface, resulting in much higher drainage density and relative relief than in the eolian sand region (Table 8). The contrasting relationships exhibited in the loess and eolian sand regions reflect the nature of available materials. Although abundant sediments are available in each region, the porous sandy materials that make up the eolian sand region have markedly different textural properties to the loess region. This, in turn, affects the relief and the degree of landscape dissection. Collectively, these factors influence the dominance of different sediment transport mechanisms, with the occurrence of hyperconcentrated flows being restricted to the loess region. In both regions, however, channel-forming flows occur much less frequently than the corresponding effective discharges in highly sediment-charged, transport-limited, semi-arid environments. Different sets of relationships are likely to be observed in a supply-limited condition. 6. Conclusions Effective discharge is an important concept in fluvial geomorphology. Many studies have demonstrated that effective discharge has a regular recurrence interval of 2 years. In this study, a method for selecting a suitable class interval, modified from those proposed by Yevjevich (972) and Crowder and Knapp (25), is used to calculate effective discharges for streams transporting suspended sediment in the Wuding basin in China. In two loess regions of the study area, the flow duration of effective discharge ranges primarily from.26% to 3.6%. In the eolian sand region, the flow duration of effective discharge varies from 8.75% to 9.5%. The effective discharge for suspended sediment transport in the eolian sand region is sensitive to the size of class intervals. For the most reliable class interval of.25s, the effective discharge occurs approximately.26% of the time in the loess regions and 24.5% at Yulin and 52.66% at Hanjiamao respectively in the eolian sand region. Effective discharges for suspended sediment differ significantly in rivers located in different geomorphological regions. Histograms of suspended sediment transport indicate that there is a bimodal dominant discharge for suspended sediment transport, with one peak in the range of low flows and the other in the range of large floods. Hydrologic regime and suspended sediment concentration are important determinants of effective discharge. Flow discharge and suspended sediment concentration are more variable in the loess region than in the eolian sand region. In neither instance, however, does effective discharge equate to bankfull discharge or the.5 year flood recurrence interval. Acknowledgements This research was supported by the National Natural Science Foundation of China (No. 4788) and the Chinese Academy of Sciences. Field measurements were made by the Yellow Con-

13 Y. Ma et al. / Journal of Hydrology 388 (2) servancy Commission of China. John Pitlick, Stuart Lane, Waite Osterkamp and an anonymous reviewer are gratefully acknowledged for their constructive suggestions. References Andrews, E.D., 98. Effective and bankfull discharges of streams in the Yampa Basin, Colorado and Wyoming. Journal of Hydrology 46, Andrews, E.D., Nankervis, J.M., 995. Effective discharge and the design of channel maintenance flows for gravel-bed rivers. In: Costa, J.E., Miller, A.J., Potter, K.W., Wilcock, P.R. (Eds.), Natural and Anthropogenic Influences in Fluvial Geomorphology. American Geophysical Union, pp Ashmore, P.E., Day, T.J., 988. Effective discharge for suspended sediment transport in streams of the Saskatchewan Basin. Water Resource Research 24 (6), Benson, M.A., Thomas, D.M., 966. A definition of dominant discharge. Bulletin of the International Association of Scientific Hydrology, Biedenharn, D.S., Thorne, C.R., Soar, P.J., Hey, R.D., Watson, C.C., 999. A practical guide to effective discharge calculation. In: Watson, C.C., Biedenharn, D.S., Thorne, C.R. (Eds.), Demonstration Erosion Control: Design Manual. Engineer Research and Development Center, US Army Corps of Engineers, Vicksburg, MS, pp Boiten, W., 23. Hydrometry, second ed. Swets and Zeitlinger BV, Lisse, The Netherlands. Carling, P., 988. The concept of dominant discharge applied to two gravel-bed streams in relation to channel stability thresholds. Earth Surface Processes and Landforms 3, Castro, J.M., Jackson, P.L., 2. Bankfull discharge recurrence intervals and regional hydraulic geometry relationships: patterns in the Pacific Northwest, SA. Journal of the American Water Resources Association 37 (5), Chen, W., 989. The statistical features trend surface and cluster analysis of the geomorphy in the loess region of the drainage area of the Wuding. Journal of Arid Resources and Environment 3 (4), 2 32 (in Chinese). Chen, J., Hu, C., Dong, Z., Liu, D., 26. Change of bankfull and bed-forming discharges in the Lower Yellow. Journal of Sediment Research 3 (5), 6 (In Chinese). Cheng, Q., Cai, Q., Zheng, M., 27. Effects of soil crust on critical rainfall intensity of runoff production. Scientia Geographica Sinica 27 (5), (in Chinese). Copeland, R.R., McComas, D.N., Thorne, C.R., Soar, P.J., Jonas, M.M., Fripp, J.B., 2. Hydraulic design of stream restoration projects. US Army Corps of Engineers, ERDC/CHL TR--28. Crowder, D.W., Knapp, H.V., 25. Effective discharge recurrence intervals of Illinois streams. Geomorphology 64, Doyle, M.W., Stanley, E.H., Strayer, D.L., Jacobson, R.B., Schmidt, J.C., 25. Effective discharge analysis of ecological processes in streams. Water Resources Research 4, W4. doi:.29/25wr4222. Dunne, T., Leopold, L.B., 978. Water in Environmental Planning. Freeman, San Francisco. p. 88. Dury, G.H., 973. Magnitude-frequency analysis and channel morphology. In: Morisawa, M. (Ed.), Fluvial Geomorphology. Suny, Binghamton, pp Dury, G.H., 977. Underfit streams: retrospect, perspect and prospect. In: Gregory, K.J. (Ed.), Channel Changes. Wiley, Chichester, pp Dury, G.H., Hails, J.R., Robbie, M.B., 963. Bankfull discharge and the magnitudefrequency series. Australian Journal of Science 26, Emmett, W.W., Wolman, M.G., 2. Effective discharge and gravel-bed rivers. Earth Surface Processes and Landforms 26, Federal Interagency Stream Restoration Working Group (FISRWG), 998. Stream corridor restoration: principles, processes, and practices. GPO Item No. 2-A; SuDocs No. A 57.6/2:EN 3/PT.653, ISBN , < stream_restoration/>. Gomez, B., Coleman, S.E., Sy, V.W.K., Peacock, D.H., Kent, M., 27. Channel change, bankfull and effective discharges on a vertically accreting, meandering, gravelbed river. Earth Surface Processes and Landforms 32, 5. doi:.2/ esp.424. Harman, W.H., Jennings, G.D., Patterson, J.M., Clinton, D.R., Slate, L.O., Jessup, A.G., Everhart, J.R., Smith, R.E., 999. Bankfull hydraulic geometry relationships for North Carolina streams. In: Olsen, D.S., Potyondy, J.P. (Eds.), Proceedings of the Wildland Hydrology Symposium. AWRA, Bozeman, MT. Hickin, E.J., 968. Channel morphology, bankfull stage and bankfull discharge of streams near Sydney. Australian Journal of Science 3, Inglis, C.C., 949. The behavior and control of rivers and canals. Research Publication, vol. 3. Central Waterpower Irrigation and Navigation Research Station, Poona, India. pp Lenzi, M.A., Mao, L., Comiti, F., 26. Effective discharge for sediment transport in a mountain river: computational approaches and geomorphological effectiveness. Journal of Hydrology 326, Leopold, L.B., Wolman, M.G., Miller, J.P., 964. Fluvial Processes in Geomorphology. W.H. Freeman, San Francisco. p Hydrological Bureau, Ministry of Water Conservancy and Electric Power (PRC), 962. National Standards for Hydrological Survey, vols. 7. China Industry Press, Beijing. (in Chinese). Nash, D.B., 994. Effective sediment-transporting discharge from magnitudefrequency analysis. Journal of Geology 2, Navratil, O., Albert, M., Hérouin, B.E., Gresillon, J.M., 26. Determination of bankfull discharge magnitude and frequency: comparison of methods on 6 gravel-bed river reaches. Earth Surface Processes and Landforms 3 (), Nolan, K.M., Lisle, T.E., Kelsey, H.M., 987. Bankfull discharge and sediment transport in northwestern California. Erosion and Sedimentation in the Pacific Rim, IAHS Publ 65, Odem, W.O., Moody, T., Knight, K., Wirtanen, M., 999. Stream Channel Morphology in New Mexico: Regional Relationships. Department of Civil and Environmental Engineering. Northern Arizona University, Flagstaff, AZ. Osterkamp, W.R., 24. Bankfull discharge. In: Goudie, A.S. (Ed.), Encyclopedia of Geomorphology. Routledge, London, pp Phillips, J.D., 22. Geomorphological impacts of flash flooding in a forested headwater basin. Journal of Hydrology 269, Pickup, G., Warner, R.F., 976. Effects of hydrologic regime on magnitude and frequency of dominant discharge. Journal of Hydrology 29, Rosgen, D.L., 996. Applied Morphology. Wildland Hydrology, Pagosa Springs, Colorado, USA. p. 39. Schaffernak, F., 922. Neue Grundlagen fur die Berechung der Qeschisbefuhrung in Flusslaufen. Franz Deuticke, Leipzig, Germany. Schmidt, K.H., Morche, D., 26. Sediment Output and Effective Discharge in two small High Mountain Catchments in the Bavarian Alps, Germany. Geomorphology 8 ( 2), Segura, C., Pitlick, J. in press. Scaling Frequency of Channel-Forming Flows in Snowmelt-Dominated Streams. Water Resources Research. Shields Jr., F.D., Copeland, R.C., Klingeman, P.C., Doyle, M.V., Simon, A., 23. Design for stream restoration. Journal of Hydraulic Engineering 29 (8), Sichingabula, H.M., 999. Magnitude-frequency characteristics of effective discharge for suspended sediment transport, Fraser, British Columbia, Canada. Hydrological Processes 3, Simon, A., Dickerson, W., Heins, A., 24. Suspended-sediment transport rates at the.5-year recurrence interval for ecoregions of the United States: transport conditions at the bankfull and effective discharge? Geomorphology 58, Stedinger, J.R., Vogel, R.M., Foufoula-Georgiou, E., 992. Frequency analysis of extreme events. In: Maidment, D. (Ed.), Handbook of Hydrology. McGraw-Hill, New York, pp Torizzo, M., Pitlick, J., 24. Magnitude-frequency of bed load transport in mountain streams in Colorado. Journal of Hydrology 29, Vogel, R.M., Stedinger, J.R., Hooper, R.P., 23. Discharge indices for water quality loads. Water Resources Research 39 (), 273. doi:.29/22wr872. Wang, S., 27. Comparison of sediment and runoff yield processes between different geomorphological regions in the Wuding basin. Geographical Research 26 (3), (In Chinese). Webb, B.W., Walling, D.E., 982. The magnitude and frequency characteristics of fluvial transport in a Devon drainage basin and some geomorphological implications. Catena 9, Whiting, P.J., Stamm, J.F., Moog, D.B., Orndorff, R.L., 999. Sediment-transporting flows in headwaters streams. Geological Society of America Bulletin (3), Williams, G.P., 978. Bankfull discharge of rivers. Water Resources Research 23 (8), Wolman, M.G., Miller, J.P., 96. Magnitude and frequency of forces in geomorphological processes. Journal of Geology 68, Xia, J., Wu, B., Wang, G., Wang, Y., 2. Estimation of bankfull discharge in the lower Yellow using different approaches. Geomorphology 7, Xu, J., 998. A Study of physio-geographical factors for the formation of hyperconcentrated flows in the Loess Plateau of China. Geomorphology 24, Xu, J., 24. Response of erosion and sediment producing processes to soil and water conservation measures in the Wuding basin. Acta Geographica Sinica 59 (6), (in Chinese). Yevjevich, V., 972. Probability and Statistics in Hydrology. Water Resources Publications, Fort Collins, Colorado. p. 32. Zhang, J., Ji, W., Feng, X., 23. Water and sediment changes in the Wuding : present state, formative cause and tendency in the future. In: Wang, G., Fan, Z. (Eds.), A Study of Water and Sediment Changes in the Yellow, vol. 2. Publishing House of Yellow Water Conservancy, Zhengzhou, pp (In Chinese).