**2. Materials and methods**

### **2.1 Sediment source classification**

The identification and the classification of the potential sediment sources within a catchment or river basin is of fundamental importance for a successful application of the fingerprinting approaches. The sources have been classified

based on the source type (e.g., surface or sub-surface erosion, areas under different land use) or on spatial location (e.g., sub-catchments or geological units). Despite that most existing studies have focused on discriminating sources types [10], information on spatial source, for example tributary sub-basins, might be more important at larger scales [14].

Three spatially differentiated sources have been identified in this study to apportion their contribution to the sediments deposited within the riparian zone of the Ruxi tributary channel. Firstly, the upside soils above the 175 m elevation level have the possibility to be moved downward as a result of surface erosion associated with rainfall events and agricultural disturbance, which therefore provide a potential source (Source 1) of the sediment. Secondly, the suspended sediments transported by both the mainstream of Yangtze River (Source 2) and the upstream Ruxi River (Source 3) are also likely to be deposited and stored within the riparian zone during the impounding period of the reservoir.

#### **2.2 Sample collection**

Representative samples of the deposited sediments have been collected from the riparian zone at six sections along the Ruxi River by using artificial grass mats as sediment traps. At each section, the 30 m high riparian zone has been subdivided into three intervals of elevation (145–155, 155–165, and 165–175 m). For each interval, one small piece of plastic mat with an area of 1 × 1 m has been fixed to the soil surface with steel pins prior to inundation. The time-integrated deposited sediments have been sampled between September 2015 and June 2016, encompassing a complete inundation and exposure period of the riparian zone.

At each section for deposited sediment collection, surface soils (0–2 cm) have been grabbed at locations with actively eroding signs and immediately above the 175 m elevation level. To increase the representativeness of the source samples, several subsamples have been collected within the vicinity of each sampling location and then mixed to produce a composite sample (n = 6).

The collection of fluvial suspended sediments (Source 2 and Source 3) has been undertaken monthly from September 2015 to June 2016. One time-integrating sediment trap has been deployed in the main channel of the Yangtze River, which is located at approximately 1 km upstream from the confluence with the Ruxi River (**Figure 1**). The trap consists of a floating barrel fixed at 1.5 m below the ambient water surface. Using the same method, suspended samples of the Ruxi tributary have been collected from a site near the center of the river channel cross section. This sampling site was located about 5 km upstream of the tributary. Totally, 18 samples of the suspended sediment transported by both the Yangtze mainstream (n = 9) and Ruxi River (n = 9), which representing Source 2 and Source 3, respectively, were collected.

It is noted that, however, the specific flow regulation mode of the Three Gorges Dam leads to remarkable difference in water level residence time and inundation duration between different elevations within the 30-m-height zone. Typically, water level rises rapidly from 145 to 175 m during the period spanning from mid-September to early October; then, it remains at the peak level until late January. Subsequently, water level retreats gradually to the 145 m base level in early June, which then maintain around this level until next inundation period (**Figure 2**). Therefore, deposited sediment sources from the suspended material originating from the mainstream and tributary vary temporally with the variation of water levels. In the case of the 145–155 m elevation, the inundation duration lasts for nearly 9 months. For the upper 155–165 and 165–175 m elevations, the riparian zones are submerged for about 6 and 4 months, respectively. In the case of the

**53**

*DOI: http://dx.doi.org/10.5772/intechopen.85208*

with the exception of Source 1.

**2.3 Laboratory analysis**

**Figure 2.**

before being dried.

the <63 μm fraction.

*Fingerprinting Sources of the Sediments Deposited in the Riparian Zone of the Ruxi Tributary…*

Source 1, however, the eroded material has the equal chance to be transported downward and stored within different elevations. In this context, sediment sources have been allocated according to the submerging and sampling periods

*Dynamic of water levels with the inundation period of the Three Gorges Reservoir.*

The deposited sediment collected on each mat has been rinsed using deionized water and recovered by sedimentation and centrifugation. After being retrieved from the traps, the suspended sediment samples (Source 2 and Source 3) have been also obtained by setting and centrifugation in the laboratory. In the case of the composite topsoil samples, stones and visible plant debris have been discarded

All samples have been air-dried at room temperature, gently disaggregated using a pestle and mortar and screened through a 2-mm sieve. Subsamples of the <2 mm fraction of the target deposit were measured firstly to determine the grain size distribution using a laser diffraction granulometer (Mastersizer 2000, Malvern Instruments). Prior to analysis, the samples were pretreated with 10% H2O2 and 10% HCl to remove organic matter and CaCO3, respectively, and then dispersed with ultrasound for 2 minutes. The obtained results have revealed that the deposited sediments collected from the riparian zone with different elevation levels were dominated by silt and clay particles, in which over 90% of the sediment was <63 μm in size. Consequently, all source material and deposited sediment samples were sieved to <63 μm to obtain a comparable grain-size faction between source and sediment material. Subsequent analyses were restricted to

A total of 18 potential fingerprinting properties were selected for analysis. Concentrations of TOC and TN were measured using a vario MACRO cube element analyzer (Elementar, Germany) after the removal of CaCO3 using 10% HCl. Other elements, including TP, K, Mg, Na, Ca, Fe, Al, Cr, Mn, Ti, Zn, Cd, Co, Cu, Ni, and Pb, were determined using ICP-OES and ICP-MS following a microwave-assisted digestion with HNO3 and HF. Duplicates, method blanks, and standard reference

materials (GBW07401) were used for quality assurance and control.

*DOI: http://dx.doi.org/10.5772/intechopen.85208 Fingerprinting Sources of the Sediments Deposited in the Riparian Zone of the Ruxi Tributary…*

**Figure 2.** *Dynamic of water levels with the inundation period of the Three Gorges Reservoir.*

Source 1, however, the eroded material has the equal chance to be transported downward and stored within different elevations. In this context, sediment sources have been allocated according to the submerging and sampling periods with the exception of Source 1.

#### **2.3 Laboratory analysis**

*Sedimentary Processes - Examples from Asia, Turkey and Nigeria*

important at larger scales [14].

the impounding period of the reservoir.

**2.2 Sample collection**

tively, were collected.

based on the source type (e.g., surface or sub-surface erosion, areas under different land use) or on spatial location (e.g., sub-catchments or geological units). Despite that most existing studies have focused on discriminating sources types [10], information on spatial source, for example tributary sub-basins, might be more

Three spatially differentiated sources have been identified in this study to apportion their contribution to the sediments deposited within the riparian zone of the Ruxi tributary channel. Firstly, the upside soils above the 175 m elevation level have the possibility to be moved downward as a result of surface erosion associated with rainfall events and agricultural disturbance, which therefore provide a potential source (Source 1) of the sediment. Secondly, the suspended sediments transported by both the mainstream of Yangtze River (Source 2) and the upstream Ruxi River (Source 3) are also likely to be deposited and stored within the riparian zone during

Representative samples of the deposited sediments have been collected from the riparian zone at six sections along the Ruxi River by using artificial grass mats as sediment traps. At each section, the 30 m high riparian zone has been subdivided into three intervals of elevation (145–155, 155–165, and 165–175 m). For each interval, one small piece of plastic mat with an area of 1 × 1 m has been fixed to the soil surface with steel pins prior to inundation. The time-integrated deposited sediments have been sampled between September 2015 and June 2016, encompassing a

At each section for deposited sediment collection, surface soils (0–2 cm) have been grabbed at locations with actively eroding signs and immediately above the 175 m elevation level. To increase the representativeness of the source samples, several subsamples have been collected within the vicinity of each sampling loca-

The collection of fluvial suspended sediments (Source 2 and Source 3) has been

It is noted that, however, the specific flow regulation mode of the Three Gorges Dam leads to remarkable difference in water level residence time and inundation duration between different elevations within the 30-m-height zone. Typically, water level rises rapidly from 145 to 175 m during the period spanning from mid-September to early October; then, it remains at the peak level until late January. Subsequently, water level retreats gradually to the 145 m base level in early June, which then maintain around this level until next inundation period (**Figure 2**). Therefore, deposited sediment sources from the suspended material originating from the mainstream and tributary vary temporally with the variation of water levels. In the case of the 145–155 m elevation, the inundation duration lasts for nearly 9 months. For the upper 155–165 and 165–175 m elevations, the riparian zones are submerged for about 6 and 4 months, respectively. In the case of the

undertaken monthly from September 2015 to June 2016. One time-integrating sediment trap has been deployed in the main channel of the Yangtze River, which is located at approximately 1 km upstream from the confluence with the Ruxi River (**Figure 1**). The trap consists of a floating barrel fixed at 1.5 m below the ambient water surface. Using the same method, suspended samples of the Ruxi tributary have been collected from a site near the center of the river channel cross section. This sampling site was located about 5 km upstream of the tributary. Totally, 18 samples of the suspended sediment transported by both the Yangtze mainstream (n = 9) and Ruxi River (n = 9), which representing Source 2 and Source 3, respec-

complete inundation and exposure period of the riparian zone.

tion and then mixed to produce a composite sample (n = 6).

**52**

The deposited sediment collected on each mat has been rinsed using deionized water and recovered by sedimentation and centrifugation. After being retrieved from the traps, the suspended sediment samples (Source 2 and Source 3) have been also obtained by setting and centrifugation in the laboratory. In the case of the composite topsoil samples, stones and visible plant debris have been discarded before being dried.

All samples have been air-dried at room temperature, gently disaggregated using a pestle and mortar and screened through a 2-mm sieve. Subsamples of the <2 mm fraction of the target deposit were measured firstly to determine the grain size distribution using a laser diffraction granulometer (Mastersizer 2000, Malvern Instruments). Prior to analysis, the samples were pretreated with 10% H2O2 and 10% HCl to remove organic matter and CaCO3, respectively, and then dispersed with ultrasound for 2 minutes. The obtained results have revealed that the deposited sediments collected from the riparian zone with different elevation levels were dominated by silt and clay particles, in which over 90% of the sediment was <63 μm in size. Consequently, all source material and deposited sediment samples were sieved to <63 μm to obtain a comparable grain-size faction between source and sediment material. Subsequent analyses were restricted to the <63 μm fraction.

A total of 18 potential fingerprinting properties were selected for analysis. Concentrations of TOC and TN were measured using a vario MACRO cube element analyzer (Elementar, Germany) after the removal of CaCO3 using 10% HCl. Other elements, including TP, K, Mg, Na, Ca, Fe, Al, Cr, Mn, Ti, Zn, Cd, Co, Cu, Ni, and Pb, were determined using ICP-OES and ICP-MS following a microwave-assisted digestion with HNO3 and HF. Duplicates, method blanks, and standard reference materials (GBW07401) were used for quality assurance and control.


 *The results of the Shapiro-Wilk test for normality.*

**55**

*\**

**Table 3.**

*levels in the riparian zone.*

*DOI: http://dx.doi.org/10.5772/intechopen.85208*

**Table 2.**

**Elevation level (m) Properties passed the range test** 145–155 TN, TOC, TP, Na, Ti, Mn, Cr, Zn, Pb, Cd 155–165 TN, TOC, TP, Ca, Mg, Na, Ti, Mn, Cu, Cd

**2.4 Sediment source discrimination**

*Statistically significant values at P ≤ 0.05.*

source and target samples are used for further analyses.

The measured values have been firstly tested for normality prior to proceed with the statistical selection of potential fingerprint properties [15]. Particle size and organic matter correction factors were not included in this study to avoid the risk of over-correction of the tracer values [16, 17]. **Table 1** shows that some properties exhibit non-normal distribution. Nonetheless, the majority of the fingerprint properties for source and target samples in different elevation levels have passed the Shapiro-Wilk test, confirming that these data were normal in distribution. Since the number of samples in this study is relatively small and the mean values are more sensitive than the median, the median values for individual properties in both

*The results of applying the Kruskal-Wallis H-test to the fingerprint property dataset for different elevation* 

Mn 11.000 0.004\* Cu 4.524 0.104 Cr 4.614 0.100 Pb 10.629 0.005\* Cd 10.315 0.006\*

One fundamental assumption underpinning the sediment source fingerprinting techniques is that the tracer properties, which were selected to discriminate between sources, should behave conservatively during mobilization and delivery through the catchment system [14]. Although the conservative behavior of the

*Fingerprinting Sources of the Sediments Deposited in the Riparian Zone of the Ruxi Tributary…*

165–175 TN, TOC, TP, Fe, Al, Ca, Mg, K, Na, Ti, Mn, Cu, Cr, Pb, Cd

*Fingerprinting properties passed the range test for different elevation levels in the riparian zone.*

**145–155 m 155–165 m 165–175 m**

**Property** *H***-value** *P***-value Property** *H***-value** *P***-value Property** *H***-value** *P***-value** TN 13.814 0.001\* TN 11.275 0.004\* TN 8.857 0.012\* TOC 14.658 0.001\* TOC 11.275 0.004\* TOC 8.857 0.012\* TP 14.487 0.001\* TP 12.329 0.002\* TP 9.736 0.008\* Na 9.937 0.007\* Ca 12.784 0.002\* Fe 4.295 0.117 Ti 16.275 0.000\* Mg 11.368 0.003\* Al 1.929 0.381 Mn 14.656 0.001\* Na 9.697 0.008\* Ca 11.429 0.003\* Cr 5.889 0.053 Ti 11.699 0.003\* Mg 8.000 0.018\* Zn 10.926 0.004\* Mn 12.784 0.002\* K 1.529 0.466 Pb 12.010 0.002\* Cu 3.029 0.220 Na 5.618 0.060 Cd 13.469 0.001\* Cd 12.501 0.002\* Ti 8.324 0.016\*

**Table 1.**

*DOI: http://dx.doi.org/10.5772/intechopen.85208 Fingerprinting Sources of the Sediments Deposited in the Riparian Zone of the Ruxi Tributary…*


**Table 2.**

*Sedimentary Processes - Examples from Asia, Turkey and Nigeria*

**54**

**Property** **S1**

> TN

TOC

TP Fe Al Ca Mg

K Na

Ti Mn

Co Cu

Cr Zn Pb Cd

Ni

0.114 *\*Statistically significant values at P ≤ 0.05.*

**Table 1.**

*The results of the Shapiro-Wilk test for normality.*

0.542

0.561

0.835

0.114

0.626

0.431

0.074

0.114

0.686

0.259

0.748

0.213 0.021\*

0.963

0.800

0.868

0.021\*

0.341

0.505

0.688

0.021\*

0.152

0.142

0.407

0.221

0.087

0.946

0.213

0.420

0.180

0.608

0.213

0.544

0.738

0.101

0.071

0.927

0.566

0.206

0.071

0.020\*

0.729

0.936

0.071

0.013\*

0.037\*

0.867

0.999

0.134

0.608

0.945

0.999

0.118

0.339

0.717

0.999

0.077

0.390

0.961

0.694 0.043\*

0.247

0.003\*

0.253

0.043\*

0.507

0.024\*

0.806

0.043\*

0.131

0.015\*

0.027\*

0.636

0.466

0.726

0.694

0.655

0.256

0.101

0.694

0.326

0.404

0.648

0.371

0.951

0.260

0.637

0.371

0.966

0.666

0.000\*

0.371

0.993

0.504

0.811

0.323

0.811

0.052

0.384

0.323

0.861

0.066

0.177

0.323

0.641

0.259

0.574

0.569

0.006\*

0.116

0.396

0.569

0.094

0.272

0.878

0.569

0.817

0.150

0.052

0.167

0.982

0.978

0.235

0.167

0.758

0.911

0.432

0.167

0.387

0.973

0.042\*

0.667

0.987

0.684

0.207

0.667

0.723

0.573

0.388

0.667

0.676

0.616

0.696

0.114

0.020\*

0.003\*

0.516

0.114

0.005\*

0.001\*

0.927

0.114

0.594

0.350

0.722

0.071

0.549

0.975

0.133

0.071

0.376

0.523

0.910

0.071

0.297

0.923

0.206

0.074

0.994

0.795

0.745

0.074

0.948

0.861

0.470

0.074

0.899

0.648

0.160

0.490

0.405

0.656

0.537

0.490

0.723

0.359

0.843

0.490

0.204

0.521

0.747

0.180

0.299

0.052

0.432

0.180

0.074

0.006\*

0.625

0.180

0.625

0.699

0.825

0.725

0.407

0.114

0.430

0.725

0.131

0.009\*

0.910

0.725

0.496

0.009\*

0.451

**S2**

**S3**

**Target**

**S1**

**S2**

**S3**

**Target**

**S1**

**S2**

**S3**

**Target**

**145–155 m**

**155–165 m**

**165–175 m**

*Fingerprinting properties passed the range test for different elevation levels in the riparian zone.*


*\* Statistically significant values at P ≤ 0.05.*

#### **Table 3.**

*The results of applying the Kruskal-Wallis H-test to the fingerprint property dataset for different elevation levels in the riparian zone.*

#### **2.4 Sediment source discrimination**

The measured values have been firstly tested for normality prior to proceed with the statistical selection of potential fingerprint properties [15]. Particle size and organic matter correction factors were not included in this study to avoid the risk of over-correction of the tracer values [16, 17]. **Table 1** shows that some properties exhibit non-normal distribution. Nonetheless, the majority of the fingerprint properties for source and target samples in different elevation levels have passed the Shapiro-Wilk test, confirming that these data were normal in distribution. Since the number of samples in this study is relatively small and the mean values are more sensitive than the median, the median values for individual properties in both source and target samples are used for further analyses.

One fundamental assumption underpinning the sediment source fingerprinting techniques is that the tracer properties, which were selected to discriminate between sources, should behave conservatively during mobilization and delivery through the catchment system [14]. Although the conservative behavior of the


#### **Table 4.**

*The optimum composite fingerprint for different elevation levels in the riparian zone established using stepwise discriminant function analysis.*

tracer properties is complex and difficult, if not impossible, to quantify, a modified range test was applied to ensure that the median concentration for each tracer associated with the target falls within the range of median concentrations of that tracer associated with the potential sources [18]. The potential fingerprinting properties, which passed the range test for each elevation level, are listed in **Table 2**. It has been recognized that such a range test is simply a method of selection to exclude the properties suffering a significant change during the transport within the fluvial system; thus, complete absence of tracer property transformation (i.e., conservative behavior) is not guaranteed [19].

A two-stage statistical procedure [12] was then applied to the source material properties passing the range test to identify the composite fingerprints that provide a good discrimination between sources. During the first stage, the Kruskal-Wallis *H*-test was used to select properties exhibiting significant differences (*P* ≤ 0.05) between the individual sources (**Table 3**). Properties passing the Kruskal-Wallis *H*-test are then tested, in stage two, by stepwise multivariate discriminant function analysis (DFA) to identify the optimum combination of tracers for discriminating the source groups (**Table 4**). In this stage, each fingerprint property was selected based on the minimization of Wilks' λ and a probability value for parameter entry of 0.05.

### **3. Results**

A frequentist-based multivariate mixing model [12] was applied to assess the relative contribution of the three potential sources to the deposited sediment samples collected from each designated riparian zone. In this method, the proportions *P* contributed by the *m* individual sources *s* are quantified by minimizing the sum of the squares of the residuals (*Res*) for the *n* tracer properties involved, where:

$$R\_{cs} = \sum\_{i=1}^{n} \left( \frac{C\_{di} - \left(\sum\_{s=1}^{m} C\_{sl} P\_s\right)}{C\_{di}} \right)^2 \tag{1}$$

**57**

**4. Discussion**

**4.1 Sediment source apportionment**

fingerprinting technique in this investigation.

*DOI: http://dx.doi.org/10.5772/intechopen.85208*

Hypercube sampling method.

source group [21]. The GOF function is defined as:

*GOF* = 1 − [

*Fingerprinting Sources of the Sediments Deposited in the Riparian Zone of the Ruxi Tributary…*

relative proportion from source group *s*. Note that two constraints are imposed on the mixing model: (a) the relative contributions from the individual sediment sources must lie between 0 and 100%, and (b) these contributions sum to 100%. The mixing model was optimized using the OptQuest algorithm in Oracle's Crystal Ball software. To address the uncertainties associated with representation of the sources and targets by single property values (e.g., mean or median), Student's *t*-distributions were assigned for each source (*Csi*) and target sediment property (*Cdi*) using measured median as location and the product of standard deviation and *n*<sup>−</sup>1/2 as scale parameter, where *n* is the number of samples [20]. The proportional contribution for each source was repeatedly solved for 1000 times using a Latin

The results of Latin Hypercube sampling procedure documented that the estimates of sediment contribution to the riparian deposits from the individual sources involved limited uncertainty bands of ±0.5–0.7% at the 95% level of confidence. Consequently, the proportional source contributions are reported as absolute median values in the following section when comparing and explaining the results. The robustness of the optimized solutions of the mixing model was assessed using a goodness-of-fit (GOF) function, which compares the actual fingerprint property concentration measured in target sediment with the corresponding values predicted by the model, based on the optimized percentage contribution from each

> \_\_1 *<sup>n</sup>* ∑ *i*=1

the values were comparable between the upper 155–165 and 165–175 m levels.

A good discrimination between the sources (**Table 3**) and high levels of correct classification of source type samples (**Table 4**) were provided by the two-stage statistical procedure following the normality and conservatism tests. More particularly, the source contribution estimation produced by the mixing model has generated high values of GOF and limited levels of uncertainty. The results indicate that reliable estimates of sediment source contribution were obtained by using the

The source ascription results presented in **Figure 3** clearly verify that the contribution of the sediments suspended from both the Yangtze mainstream and the Ruxi

**Figure 3** presents the median contributions from individual source types to the deposited sediment samples collected from the riparian zone with different elevation level. The GOF values range between 0.88 and 0.95, indicating that the modeling results are acceptable. The sources of the suspended material from the Yangtze mainstream (Source 2) contributed the most proportion (>40%) to the deposited sediment for all three elevation levels of the riparian zone. There were, however, clear evidence of significant contrasts between the contributions from the other two sources (i.e., sources 1 and 3) for the 145–155 m elevation level and the other two upper elevation levels. In the case of the sources of topsoil above the riparian zone (Source 1), the estimated contribution to the lowest 145–155 m deposits was only about half of those to the other two upper elevation levels. In the case of the sediment input from the Ruxi tributary (Source 3), the relative contribution decreased with increasing elevation, despite that

*<sup>n</sup>* |*Cdi* − ∑*s*=1

*<sup>m</sup> CsiPs*|

\_\_\_\_\_\_\_\_\_\_\_\_\_ *Cdi* ] (2)

and *Cdi* is the concentration of tracer property *i* in the deposited sediment sample, *Csi* is the concentration of tracer property *i* in source group *s* and *Ps* is the *DOI: http://dx.doi.org/10.5772/intechopen.85208 Fingerprinting Sources of the Sediments Deposited in the Riparian Zone of the Ruxi Tributary…*

relative proportion from source group *s*. Note that two constraints are imposed on the mixing model: (a) the relative contributions from the individual sediment sources must lie between 0 and 100%, and (b) these contributions sum to 100%.

The mixing model was optimized using the OptQuest algorithm in Oracle's Crystal Ball software. To address the uncertainties associated with representation of the sources and targets by single property values (e.g., mean or median), Student's *t*-distributions were assigned for each source (*Csi*) and target sediment property (*Cdi*) using measured median as location and the product of standard deviation and *n*<sup>−</sup>1/2 as scale parameter, where *n* is the number of samples [20]. The proportional contribution for each source was repeatedly solved for 1000 times using a Latin Hypercube sampling method.

The results of Latin Hypercube sampling procedure documented that the estimates of sediment contribution to the riparian deposits from the individual sources involved limited uncertainty bands of ±0.5–0.7% at the 95% level of confidence. Consequently, the proportional source contributions are reported as absolute median values in the following section when comparing and explaining the results.

The robustness of the optimized solutions of the mixing model was assessed using a goodness-of-fit (GOF) function, which compares the actual fingerprint property concentration measured in target sediment with the corresponding values predicted by the model, based on the optimized percentage contribution from each source group [21]. The GOF function is defined as:

$$GOF = \mathbf{1} - \left[\frac{\mathbf{1}}{\overline{n}} \sum\_{i=1}^{n} \frac{|\mathbf{C}\_{di} - \sum\_{i=1}^{n} \mathbf{C}\_{il} P\_i|}{\mathbf{C}\_{di}}\right] \tag{2}$$

**Figure 3** presents the median contributions from individual source types to the deposited sediment samples collected from the riparian zone with different elevation level. The GOF values range between 0.88 and 0.95, indicating that the modeling results are acceptable. The sources of the suspended material from the Yangtze mainstream (Source 2) contributed the most proportion (>40%) to the deposited sediment for all three elevation levels of the riparian zone. There were, however, clear evidence of significant contrasts between the contributions from the other two sources (i.e., sources 1 and 3) for the 145–155 m elevation level and the other two upper elevation levels. In the case of the sources of topsoil above the riparian zone (Source 1), the estimated contribution to the lowest 145–155 m deposits was only about half of those to the other two upper elevation levels. In the case of the sediment input from the Ruxi tributary (Source 3), the relative contribution decreased with increasing elevation, despite that the values were comparable between the upper 155–165 and 165–175 m levels.

### **4. Discussion**

*Sedimentary Processes - Examples from Asia, Turkey and Nigeria*

145–155 1 Cd 0.367 62.5

155–165 1 Mg 0.277 61.1

165–175 1 Ca 0.016 100.0

behavior) is not guaranteed [19].

tracer properties is complex and difficult, if not impossible, to quantify, a modified range test was applied to ensure that the median concentration for each tracer associated with the target falls within the range of median concentrations of that tracer associated with the potential sources [18]. The potential fingerprinting properties, which passed the range test for each elevation level, are listed in **Table 2**. It has been recognized that such a range test is simply a method of selection to exclude the properties suffering a significant change during the transport within the fluvial system; thus, complete absence of tracer property transformation (i.e., conservative

**Elevation level (m) Step Fingerprint Wilks' λ Cumulative % source type samples classified correctly**

2 Ti 0.138 91.7 3 Mn 0.088 95.8

2 Mn 0.075 94.4 3 TP 0.026 100.0 4 TOC 0.014 100.0

2 TN 0.001 100.0 3 TP 0.001 100.0

*The optimum composite fingerprint for different elevation levels in the riparian zone established using stepwise* 

A two-stage statistical procedure [12] was then applied to the source material properties passing the range test to identify the composite fingerprints that provide a good discrimination between sources. During the first stage, the Kruskal-Wallis *H*-test was used to select properties exhibiting significant differences (*P* ≤ 0.05) between the individual sources (**Table 3**). Properties passing the Kruskal-Wallis *H*-test are then tested, in stage two, by stepwise multivariate discriminant function analysis (DFA) to identify the optimum combination of tracers for discriminating the source groups (**Table 4**). In this stage, each fingerprint property was selected based on the minimization of Wilks' λ and a probability value for parameter entry of 0.05.

A frequentist-based multivariate mixing model [12] was applied to assess the relative contribution of the three potential sources to the deposited sediment samples collected from each designated riparian zone. In this method, the proportions *P* contributed by the *m* individual sources *s* are quantified by minimizing the sum of the squares of the residuals (*Res*) for the *n* tracer properties involved, where:

*Cdi* − (∑*s*=1

and *Cdi* is the concentration of tracer property *i* in the deposited sediment sample, *Csi* is the concentration of tracer property *i* in source group *s* and *Ps* is the

*<sup>m</sup> CsiPs*) \_\_\_\_\_\_\_\_\_\_\_\_\_\_ *Cdi* )

2

(1)

*Res* = ∑ *i*=1 *n* (

**56**

**3. Results**

**Table 4.**

*discriminant function analysis.*

#### **4.1 Sediment source apportionment**

A good discrimination between the sources (**Table 3**) and high levels of correct classification of source type samples (**Table 4**) were provided by the two-stage statistical procedure following the normality and conservatism tests. More particularly, the source contribution estimation produced by the mixing model has generated high values of GOF and limited levels of uncertainty. The results indicate that reliable estimates of sediment source contribution were obtained by using the fingerprinting technique in this investigation.

The source ascription results presented in **Figure 3** clearly verify that the contribution of the sediments suspended from both the Yangtze mainstream and the Ruxi

#### **Figure 3.**

*The median relative contribution of each source type to the deposited sediment collected from the riparian zone with different elevation levels.*

tributary channel dominate the sediments deposited in the riparian zone of the reservoir. The sediment contribution originated from these two sources occupied a total proportion of 69.7–84.2%, although there is some evidence of a declining trend with increased elevation. The highest proportional contribution documented for the lowest elevation level of 145–155 m was closely related to the longest water retention time at this level (**Figure 2**), which means that there will be more opportunities for sediment deposition.

In contrast, the proportion of the topsoil source contribution increased from 15.8 to 30.3% with the increasing of elevation. The higher contribution of the topsoil source to the deposits in upper elevation levels (i.e., 155–165 and 165–175 m) could probably be attributed to the proximity of sources to the target sediment sampling sites. During the periods September–October and April–June, which represent the end and the beginning of the wet season in the study area, respectively, sediment originating from the upper lands above the riparian zone caused by water erosion will be preferentially stored in adjacent fields as a result of gentle slope gradient. The relative contribution of this source type to the deposited sediment at the 145–155 m elevation level is therefore much less than those for the upper portions of the riparian zone.

The relatively high sediment contribution from the Ruxi tributary to the 145–155 m deposits might again be explained by the relatively longer submerging period, which last for nearly 9 months (**Figure 2**). Probably more importantly, the sediment mobilized from the upstream catchment and transferred to the Ruxi River during the wet periods September–October and April–June may be predominantly deposited within the 145–155 m level due to the impoundment of the reservoir.

#### **4.2 Environmental implications for sediment source contribution**

The information on the relative contribution of potential sediment sources can also be used to evaluate the relative importance of the contributions of sedimentassociated nutrients and contaminants [22]. **Figure 4** compares the mean element concentrations for the riverine suspended sediments (sources 2 and 3) and local

**59**

cyclic soaking.

**Figure 4.**

**5. Conclusions**

*DOI: http://dx.doi.org/10.5772/intechopen.85208*

*Fingerprinting Sources of the Sediments Deposited in the Riparian Zone of the Ruxi Tributary…*

topsoils (Source 1) during the whole sampling period. The presented results show that for all the nutrients (TN, TOC, TP) and some heavy metals (e.g., Mn, Zn, Pb, Cd), the average element concentrations in the suspended fluvial sediments were significantly higher than that one of the local soils. Combining this information with the relative contribution of the suspended fluvial sediments to the riparian deposits, it can be inferred that the sediment-associated nutrients and contaminants transported from both the Yangtze mainstream and the Ruxi tributary represent dominant sources of the contaminants deposited within the riparian zone. Recent studies have documented elevated levels of nutrients [23] and heavy metals [23–25] in the riparian zone of the Three Gorges Reservoir. In this context, the accumulated deposition and storage of fine-grained sediment and associated nutrients and contaminants within the riparian zone of the reservoir may exert negative environmental impacts on aquatic ecosystems and the agricultural use of riparian lands. Moreover, the nutrients and contaminants that stored within the riparian zones may have great potential to be reintroduced into the river system by future bank erosion and/or release to the water column accompanying the annually

*(Source 2) and the Ruxi River (Source 3) to that of the topsoil above the riparian zone (Source 1).*

*The ratio of mean concentrations of individual element in suspended sediment collected from the Yangtze River* 

The regular closure in dry season and the water drainage in rainy season of the Three Gorges Dam on the Yangtze River have resulted in significant sedimentation and enrichment of sediment-associated nutrients and contaminants in the reservoir riparian zone. Against this background, the fingerprinting approach was applied to assess the sources of sediment that deposited within the riparian zone of a tributary after the impoundment of the reservoir. Despite significant contrast of sediment contribution from individual sources that was documented for different elevation levels of the riparian zone, the sediments suspended from the Yangtze mainstream was demonstrated to be the primary source of the riparian deposits. From a

*DOI: http://dx.doi.org/10.5772/intechopen.85208 Fingerprinting Sources of the Sediments Deposited in the Riparian Zone of the Ruxi Tributary…*

#### **Figure 4.**

*Sedimentary Processes - Examples from Asia, Turkey and Nigeria*

tributary channel dominate the sediments deposited in the riparian zone of the reservoir. The sediment contribution originated from these two sources occupied a total proportion of 69.7–84.2%, although there is some evidence of a declining trend with increased elevation. The highest proportional contribution documented for the lowest elevation level of 145–155 m was closely related to the longest water retention time at this level (**Figure 2**), which means that there will be more oppor-

*The median relative contribution of each source type to the deposited sediment collected from the riparian zone* 

In contrast, the proportion of the topsoil source contribution increased from 15.8 to 30.3% with the increasing of elevation. The higher contribution of the topsoil source to the deposits in upper elevation levels (i.e., 155–165 and 165–175 m) could probably be attributed to the proximity of sources to the target sediment sampling sites. During the periods September–October and April–June, which represent the end and the beginning of the wet season in the study area, respectively, sediment originating from the upper lands above the riparian zone caused by water erosion will be preferentially stored in adjacent fields as a result of gentle slope gradient. The relative contribution of this source type to the deposited sediment at the 145–155 m elevation level is therefore much less than those for the upper portions of

The relatively high sediment contribution from the Ruxi tributary to the 145–155 m deposits might again be explained by the relatively longer submerging period, which last for nearly 9 months (**Figure 2**). Probably more importantly, the sediment mobilized from the upstream catchment and transferred to the Ruxi River during the wet periods September–October and April–June may be predominantly deposited within the 145–155 m level due to the impoundment of the reservoir.

The information on the relative contribution of potential sediment sources can also be used to evaluate the relative importance of the contributions of sedimentassociated nutrients and contaminants [22]. **Figure 4** compares the mean element concentrations for the riverine suspended sediments (sources 2 and 3) and local

**4.2 Environmental implications for sediment source contribution**

tunities for sediment deposition.

the riparian zone.

**Figure 3.**

*with different elevation levels.*

**58**

*The ratio of mean concentrations of individual element in suspended sediment collected from the Yangtze River (Source 2) and the Ruxi River (Source 3) to that of the topsoil above the riparian zone (Source 1).*

topsoils (Source 1) during the whole sampling period. The presented results show that for all the nutrients (TN, TOC, TP) and some heavy metals (e.g., Mn, Zn, Pb, Cd), the average element concentrations in the suspended fluvial sediments were significantly higher than that one of the local soils. Combining this information with the relative contribution of the suspended fluvial sediments to the riparian deposits, it can be inferred that the sediment-associated nutrients and contaminants transported from both the Yangtze mainstream and the Ruxi tributary represent dominant sources of the contaminants deposited within the riparian zone. Recent studies have documented elevated levels of nutrients [23] and heavy metals [23–25] in the riparian zone of the Three Gorges Reservoir. In this context, the accumulated deposition and storage of fine-grained sediment and associated nutrients and contaminants within the riparian zone of the reservoir may exert negative environmental impacts on aquatic ecosystems and the agricultural use of riparian lands. Moreover, the nutrients and contaminants that stored within the riparian zones may have great potential to be reintroduced into the river system by future bank erosion and/or release to the water column accompanying the annually cyclic soaking.

### **5. Conclusions**

The regular closure in dry season and the water drainage in rainy season of the Three Gorges Dam on the Yangtze River have resulted in significant sedimentation and enrichment of sediment-associated nutrients and contaminants in the reservoir riparian zone. Against this background, the fingerprinting approach was applied to assess the sources of sediment that deposited within the riparian zone of a tributary after the impoundment of the reservoir. Despite significant contrast of sediment contribution from individual sources that was documented for different elevation levels of the riparian zone, the sediments suspended from the Yangtze mainstream was demonstrated to be the primary source of the riparian deposits. From a

### *Sedimentary Processes - Examples from Asia, Turkey and Nigeria*

contamination perspective, however, the sediment input from the upstream tributary also represents an important source of pollution to the riparian environment.

Although it should be recognized that the result reported is limited in temporal and spatial scopes in that only one-year sampling campaign and one single catchment was involved, the findings, combining with the information on previous studies, emphasize the need to take targeted sediment and contaminant management strategies to control the potential environmental problems in the reservoir riparian zones. Further attempts were required to explore the use of sediment fingerprinting approach to assess the sources of sediment-associated nutrients and contaminants in this area.
