Turbidity-based erosion estimation in a catchment in South Australia

doi:10.1016/S0022-1694(01)00475-9

Journal of Hydrology

Volume 253, Issues 1–4, 15 November 2001, Pages 227-238

https://doi.org/10.1016/S0022-1694(01)00475-9 Get rights and content

Abstract

An erosion estimation technique was developed in this study based on turbidity and sediment sampling data in a small catchment in South Australia. Several data sets, derived from the time sequence in which the data were collected, were used to develop a number of turbidity and suspended sediment relationships. These relationships were then used to estimate erosion from the catchment. The variability in sediment load estimation using different relationships, and how these relationships impacted on load estimation, were analyzed in detail. The study estimates erosion on a storm basis using detailed sediment sampling and turbidity data. Storm sediment loads were then accumulated to derive annual load, which distinguishes this study from volume based sediment studies. The study found that large storms dominate erosion in the catchment, and erosion rate depends more on peak storm flow than other hydrological variables. A relatively low annual erosion rate from the catchment was found, which is consistent with studies in other Australian catchments. The study found that, to establish a sound relationship between suspended sediment and turbidity for a catchment, it requires extensive data collection of large as well as small storms at short time intervals, a storm-based erosion estimation approach, and a data set that is used for interpolation rather than extrapolation. Erosion estimation based on infrequent, non-storm based or extrapolated data is exposed to potentially large errors, and the results may only be relied upon as a general guide rather than serious estimation of catchment erosion.

Introduction

Turbidity as a measure of the optical properties of water has been widely used for suspended sediment monitoring (Banasik and Walling, 1996, Jansson, 1997, Hodson et al., 1998, Jansson, 1996). The use of a turbidity meter for sediment monitoring generally requires developing a statistically significant relationship between the turbidity and suspended sediment concentration (SSC). This is not always easy as other factors such as water color, particle size, minerals and organics in water can have a significant influence on turbidity, which influences the relationship between turbidity and SSC (Gippel, 1989). Turbidity and SSC relationships are usually site and maybe time specific, and a relationship is normally unique for a particular catchment and within a particular period of time (Gippel, 1989). No theoretical analysis can determine the turbidity variation or distribution under field conditions. Using a turbidity meter can be a convenient and the most time, labour, and financially efficient method for catchment erosion monitoring and estimation. In addition, a turbidity meter can be programmed to monitor whole time series of hydrographs, which provides a cheap, continuous record of data. However, reliable use of a turbidity meter often requires a sound sediment sampling program, which has not been well addressed in previous studies (Gippel, 1989).

Large scattering of data may mean that no significant relationship can be derived from the data (Brabben, 1980). The use of a less than significant turbidity and SSC relationship may lead to misleading estimation of erosion. Some studies present good results but with low sediment concentrations. For example, Finlayson (1985) obtained a good relationship (r²=0.97) with sediment concentration up to 363 mg/l, while Ekern (1977) reported a linear turbidity and sediment relationship below 100 mg/l of sediment concentration, and a highly nonlinear one at greater concentrations in a small catchment in Hawaii. Van Bueren (1984) studied the turbidity and SSC relationship on the Yarra River, Victoria, and obtained a turbidity and SSC relationship with r²=0.71, and mean SSC of 135 mg/l. Gippel (1989) conducted a more comprehensive study of turbidity based erosion estimation in a small forest catchment near Eden, NSW. Although very detailed, the time interval for sampling was relatively large and only a few high concentration data points were observed.

These relatively good erosion estimation results are generally associated with small storm events, low range of sediment concentration, and large time intervals. It is not known whether the relationships still hold true when the sediment concentration extends beyond the data range, particularly on the upper side which can occur frequently considering the range of sediment concentration obtained. The reliability of predictions at high SSC is particularly important as studies have shown that the largest storms generally carry the majority of annual sediment loads in many Australian catchments (Edwards, 1987, Olive and Rieger, 1984, Sun, 1999). A turbidity and SSC relationship based on low or small storm monitoring may result in significant errors. Therefore, there is need to evaluate the possible errors of erosion estimation based on different turbidity and SSC relationships. Ideally, these erosion estimations should adopt a storm-based approach with the aim of sampling over an entire hydrograph and an emphasis on capturing larger storms to reflect Australian catchment erosion situations.

In this study, monitoring equipment was set up to collect sediment samples at short time intervals with the aim of capturing both the large as well as small storms. Separate turbidity and sediment relationships based on data acquired during different monitoring periods as well as relationships built on the total data were developed. These relationships were then used to estimate erosion loads in the catchment. Through this approach we explored the variability of erosion estimation based on different turbidity and sediment relationships.

Section snippets

The setting up of monitoring equipment in the Sauerbier creek catchment

The Sauerbier creek catchment is situated 20 km south of Adelaide, South Australia (Fig. 1). It has an area of 2.87 km², a high and low elevation of 370 and 180 m, and has three small creeks contributing to the main channel. A weir was constructed to measure and record flow data at the outlet of the catchment. An automatic sampler and a turbidity meter were installed approximately 100 m upslope of the weir site on the main channel. Three pluviometers were installed in or bordering the catchment.

Developing turbidity and sediment relationships

The relationship between turbidity and sediment concentration is often expressed in a linear regression equation such as $M=aT+b$ Or a non-linear equation $M=aT^{b}$ Or expressed as a polynomial function $M=a_{0} +a_{1} T+a_{2} T^{2} +a_{3} T^{3} +⋯+a_{n} T^{n}$ Where M is the concentration (mg/l), T is the turbidity usually expressed in NTU or in formazine turbidity unit (FTU), with a, b, superscript b, a₀…a_n, and n being constants. Exponential and logarithmic equations can also be used in similar ways.

A question which underlines the

Storm sediment estimation

Selected turbidity and SSC relationships developed in Section 3 were used for erosion estimation from the catchment. Most sediment studies use observed flow data to estimate suspended sediment. This study, however, used both observed and predicted flow for sediment load prediction. A storm rainfall runoff model (Sun, 1999) was used to generate the predicted flow hydrograph and then the turbidity and SSC relationships were used to estimate sediment discharges. This approach enables the

Estimating annual sediment load

The above sediment estimation was performed for the storm events with a reasonable amount of sediment sampling data covering whole hydrographs to validate erosion estimation on a storm basis. The results showed that using turbidity and SSC correlation could satisfactorily estimate sediment load on a storm basis, which is a step forward as compared to the traditional rating curve method. Since more storm events were observed by the turbidity meter and scattered sediment sampling which does not

Discussion

One of the major problems in catchment erosion estimation using turbidity observations is assessing the reliability of relationships established. This study compared alternative approaches to building such relationships, and evaluated their impacts on erosion estimation. The relationships were established based on intensive sediment sampling in the studied catchment. The study suggests that a turbidity meter can be used as an effective tool for catchment erosion estimation, but only after it is

Conclusions

The study demonstrated that a turbidity meter, well calibrated with sediment data, can be used effectively for catchment erosion estimation on a storm as well as annual basis. Successful sediment estimation depends very much on the quality of the observed sediment data and the relationship between the turbidity and sediment concentration. Intensive sediment sampling on a variable magnitude of storms covering whole hydrographs need to be monitored for reliable estimation of erosion in a

References (17)

B.L. Finlayson
Field calibration of a recording turbidity meter
Catena
(1985)
M.B. Jansson
Estimating a sediment rating curve of the Reventazon river at Palomo using logged mean loads with discharge classes
J. Hydrol.
(1996)
K. Banasik et al.
Predicting sedimentgraphs for a small agriculture catchment
Nordic Hydrol.
(1996)
Brabben, T.E., 1980. Performance and use of turbidity monitors in the Brantas River Basin, East Java, Indonesia. Rep....
Edwards, K., 1987. Runoff and soil loss studies in New South Wales. Soil Conservation Service of NSW, and Macquarie...
Ekern, P.C., 1977. Turbidity and sediment rating curves for streams on Oahu, Hawaii. Soil Erosion: Prediction and...
C.R. Fenn et al.
An evaluation of the use of suspended sediment rating curves for the prediction of suspended sedimentation in a preglacial stream
Geografis. Ann.
(1985)
Geary, P.M. 1981. Sediment and solutes in a representative basin. Australian Representative Basins Program Report...

There are more references available in the full text version of this article.

Cited by (36)

Challenges and progresses in the detailed estimation of sediment export in agricultural watersheds in Navarra (Spain) after two decades of experience
2023, Environmental Research
Soil erosion is a very serious environmental problem worldwide, with agriculture considered the main source of sediment in inland waters. In order to determine the extent and importance of soil erosion in the Spanish region of Navarra, in 1995 the Government of Navarra established the Network of Experimental Agricultural Watersheds (NEAWGN), which consists of five small watersheds representative of local conditions. In each watershed, key hydrometeorological variables, including turbidity, were recorded every 10 min, and daily samples were taken to determine suspended sediment concentration. In 2006, the frequency of suspended sediment sampling was increased during hydrologically relevant events. The main objective of this study is to explore the possibility of obtaining long and accurate time series of suspended sediment concentration in the NEAWGN. To this end, simple linear regressions between sediment concentration and turbidity are proposed. In addition, supervised learning models incorporating a larger number of predictive variables are used for the same purpose. A series of indicators are proposed to objectively characterize the intensity and timing of sampling. It was not possible to obtain a satisfactory model for estimating the concentration of suspended sediment. This would be mainly due to the large temporal variability found of the physical and mineralogical characteristics of the sediment, which would be affecting the turbidity value, independently of the sediment concentration, per se. This fact would be particularly important in small river watersheds such as those of this study, and especially if their physical conditions are spatially and temporally radically disturbed by agricultural tillage and by a constant modification of the vegetation cover, as is the case in cereal basins. Our findings suggest that better results could be obtained by including in the analysis variables such as soil texture and exported sediment texture, rainfall erosivity, and the state of vegetation cover and riparian vegetation.
Temporal variation in TiO<inf>2</inf> engineered particle concentrations in the Broad River during dry and wet weathers
2022, Science of the Total Environment
Citation Excerpt :
However, as shown here, this relationship depends on several factors such as antecedent dry periods and is likely to be site specific. It is worth noting that a strong positive relationship was observed between total suspended solid concentration and turbidity in rivers, which varied for a particular catchment and within a particular period (Daphne et al., 2011; Packman et al., 1999; Sun et al., 2001). This variation between turbidity and suspended sediment underpins our observation of the variation between turbidity and TiO2, given that TiO2 is typically associated with suspended solids.
Titanium dioxide (TiO₂) engineered particles are widely used in the urban environment as pigments in paints, and as active ingredients in photocatalytic coatings. Consequently, studies are necessary to quantify TiO₂ engineered particle concentrations and their temporal variability in surface waters to gain better understanding about their abundance and environmental fate in order to minimize their potential environmental impacts. The objective of this study was to determine the temporal variability in the concentration of TiO₂ engineered particles in the Broad River, Columbia, South Carolina, United States during dry and wet weather conditions and to examine the relationship between flow discharge, water quality indicators, and the concentration of TiO₂ engineered particles. TiO₂ engineered particle concentration in the Broad River water was determined by mass balance calculation using bulk titanium concentration and the increase in Ti/Nb ratio above the natural background ratio. The relative abundance of single metal and multi-metal Ti-bearing particles was determined by single particle-inductively coupled plasma-time of flight-mass spectrometer (SP-ICP-TOF-MS). Additionally, the elemental ratios of Ti/Nb, Ti/Al, and Ti/Fe within multi-metal Ti-bearing particles were determined at the single particle level. Discharge, bulk elemental concentrations (e.g., Ti, Al, Fe, and Nb), bulk elemental ratios (e.g., Ti/Al, Ti/Fe, and Ti/Nb), TiO₂ engineered particle concentration, and turbidity displayed the same trend of rise and fall following storm events. Linear relationships were established between turbidity and TiO₂ engineered particle concentrations in the Broad River for different flow regimes. However, no correlation was observed between TiO₂ engineered particle concentrations and flow discharge, dissolved oxygen, pH, or ionic strength. The established correlations between turbidity and TiO₂ engineered particle concentrations are important as they can be used to translate the continuously monitored turbidity to TiO₂ concentrations.
Uncertainty in the evaluation of sediment yield from badland areas: Suspended sediment transport estimated in the Araguás catchment (central Spanish Pyrenees)
2013, Catena
Citation Excerpt :
In several studies relatively good estimates of erosion from turbidity measurements at low sediment concentrations have been obtained (Finlayson, 1985; Van Bueren, 1984), but these have generally been associated with small storm events involving a short range of sediment concentrations and long time intervals. Turbidity and SSC relationships based on low sediment concentrations or infrequent storm monitoring may result in significant errors (Sun et al., 2001). In this study, uncertainty in estimates of suspended sediment yield in badland areas associated with limitations in methodologies and probes was analyzed as the main objective.
Badlands are important in terms of sediment yield, particularly in mountain areas having subhumid and humid climatic conditions. Various studies concerning erosion and hydrological processes have demonstrated that suspended sediment transport is probably the main process in sediment export from badland areas. In the Araguás catchment (central Pyrenees) there is a statistically significant positive linear relationship between maximum discharge and the maximum suspended sediment concentration (SSC). The high frequency of hyper-concentrated fluxes of SSC (> 500 g·l^− 1) recorded at two gauging stations close to badland areas in the Pyrenees (Araguás) and the Alps (Draix) indicates that these fluxes are not uncommon, and suggests that they may transport relatively large suspended particles, especially during extreme floods. In a study involving sampling during two moderate floods (August 2006 and February 2007) in the Araguás catchment it was observed that the mean suspended sediment particle size was significantly greater during the highest SSC conditions. The results showed the great heterogeneity of particle sizes that can affect suspended sediment transport, which is usually estimated from concentration determined from turbidity values obtained using infrared devices and associated with the corresponding discharge value. Infrared turbidimeters have problems in detecting particles with a diameter (D) > 0.1 mm, although discharge can be evaluated with relative high accuracy.
The combination of these factors suggests that the evaluation of sediment yield from badland areas using turbidity values involves significant uncertainty. If most suspended sediment is transported during moderate–high floods, which carry large quantities of suspended particles having D > 0.1 mm, then the sediment yield will be underestimated. The uncertainty can be calculated by determining the percentage and mean diameter of particles not detected, and the specific weight of the material. However, the uncertainty is not linear because of the exponential relationship between increasing diameter and the volume/mass, and consequently the error will increase with the growth in the suspended concentration. In this study the physical factors associated with uncertainty in the estimation of sediment yield were investigated, and quantitative estimates of the errors involved are provided.
Temporal variations of suspended sediment transport in Oneida Creek watershed, central New York
2012, Journal of Hydrology
Citation Excerpt :
Suspended sediment transport in rivers is profoundly affected by sediment supply, which varies spatially and temporally at the watershed scale and often leads to the hysteresis effect. Thus, an ideal sampling strategy is to capture variations of sediment transport by continuously monitoring turbidity (e.g., Gao et al., 2008; Olive and Rieger, 1988; Sun et al., 2001). Unfortunately, turbidity is significantly affected by sediment size distribution and other uncertainties (Navratil et al., 2010; Pavanelli and Bigi, 2005), which makes it difficult to establish a reliable turbidity-C curve for estimating continuous C in rivers transporting sediment of variable sizes.
Using event-based suspended sediment data collected from 2008 to 2010 in a medium-size agricultural watershed in Central New York, we first examined the statistical properties of suspended sediment concentration, C and the associated water discharge, Q. Next, we identified two different transport processes in two discharge ranges separated by a different threshold Q_t for each season. Different exponents of sediment rating curves (SRCs) in each season revealed that sediment was transported near capacity during lower discharges but well below capacity during higher discharges. The persistence of these two trends in all seasons suggested that suspended sediment transport was generally supply limited. Sediment loads predicted by a single seasonal SRC are similar to those predicted by the two separate SRCs for above and below Q_t, which suggests that the two transport processes are not significantly different and seasonal sediment transport may be described by a single SRC. The better fit of SRC for the combined 3-year data compared to those for individual years indicates that seasonal changes of suspended sediment transport are limited and the transport dynamics that emerge at shorter time scales (i.e., event and season) are effectively averaged out. We then calculated sediment yields for 3 years using a process-based SRC method: annual hydrograph was divided into storm and base flows using values of Q_t for all seasons and only storm flows were used to calculate sediment yields based on the developed SRCs. Comparing sediment yields of 3 years calculated using three seasonal SRCs to those using the combined 3-year SRC indicated that the discrepancies between the two were less than 5%, suggesting that sediment yield may be accurately estimated using the single combined-year SRC. Finally, we discussed the appropriate sampling strategy in the region and demonstrated geomorphological nature of sediment transport based on the calculated effective discharges and cumulative sediment loads.
Factors controlling suspended sediment yield during runoff events in small headwater catchments of the Basque Country
2007, Catena
For the first time in the Basque Country, turbidity (NTU), discharge (l/s) and precipitation (mm) have been continuously monitored in the gauging stations located at the outlet of three catchments (Aixola, Barrendiola and Añarbe) since October 2003. In this study, several data sets derived from flood events were used to develop turbidity and suspended sediment relationships for the three catchments separately, and so to estimate continuous suspended sediment concentration (SSC). Linear relationships are found in Barrendiola and Añarbe, and two curvilinear relationships for Aixola owing to changing sediment sources in the catchment. Several event (discharge, precipitation and suspended sediment concentration) and pre-event (discharge and precipitation) factors are calculated for all the events registered. With them correlation matrixes were developed for each catchment. Although some differences are found between catchments good correlation between precipitation (P), discharge (Q) and suspended sediment (SS) variables is found in general. Pre-event conditions are also well correlated with Q and SS variables in Barrendiola (higher regulation capacity) and Añarbe (larger area) but not in Aixola (“flash floods”). SSC-discharge evolutions through the events were also analysed. For Aixola four different types of hysteretic loops were observed: single lined, clockwise, counter-clockwise and eight-shaped; while for Barrendiola and Añarbe just clockwise loops were observed.
Conceptualizing turbidity for aquatic ecosystems in the context of sustainable development goals
2023, Environmental Science: Advances

View all citing articles on Scopus

View full text

Turbidity-based erosion estimation in a catchment in South Australia

Abstract

Introduction

Section snippets

The setting up of monitoring equipment in the Sauerbier creek catchment

Developing turbidity and sediment relationships

Storm sediment estimation

Estimating annual sediment load

Discussion

Conclusions

Catena

J. Hydrol.

Predicting sedimentgraphs for a small agriculture catchment

Nordic Hydrol.

An evaluation of the use of suspended sediment rating curves for the prediction of suspended sedimentation in a preglacial stream

Geografis. Ann.