Discharge and suspended sediment time series as controls on fine sediment ingress into gravel river beds

Highlights

• Time series of discharge, turbidity and sediment ingress show only weak relations.

• A redundancy approach was employed to identify the dominant facets of the regimes.

• Models incorporating multiple independent facets provided improved explanatory power.

• Temporal turbidity variations were the principal controls on ingress.

• Suspended sediment and ingress time series are needed to improve understanding.

Abstract

Fine sediment availability and channel hydraulics are two of the primary controls on the ingress of fine sediment into gravel river beds. A novel dataset consisting of fine sediment ingress measurements coupled with high-resolution turbidity and discharge time series, was analysed to investigate relations between ingress, discharge and turbidity. Discharge and turbidity demonstrated a weak association with each other, and their relations with fine sediment ingress were relatively weak. An alternative, but widely applied ‘redundancy’ approach was investigated that focused on key metrics, or facets, of the discharge and turbidity time series and their association with fine sediment ingress. Principal component analysis was used to distil the most important facets driving variation in the discharge and turbidity datasets and these were then used as independent variables in regression models with sediment ingress as the dependent variable. These models accounted for a larger amount of the statistical variation in sediment ingress over time than discharge and turbidity time series. Facets of the turbidity time series were found to be the most effective explanatory variables. The results suggest that this approach could be valuable and justify its application and testing across a range of river types in different hydrological and sedimentary settings. Application of this method could improve our generic understanding of what controls ingress at larger spatial and temporal scales and therefore complements process-based approaches, which is vital for the development of fine sediment management strategies.

Introduction

Excessive sedimentation within aquatic ecosystems is a global concern and can have detrimental consequences for all aspects of lotic ecosystem health (Heppell et al., 2009; Relyea et al., 2012; Naden et al., 2016). The deleterious effects of fine sediment on biota are well documented and are predominantly associated with sediment deposition onto, and ingress into, the river bed (Kemp et al., 2011; Jones et al., 2012a, Jones et al., 2012b, Jones et al., 2014). Effective management of fine sediment loading therefore requires understanding of the relations between deposition and ingress and their key drivers, including sediment supply and water discharge (Diplas and Parker, 1992) at scales that are relevant to catchment management.

Fine sediment deposition into a framework of gravel clasts involves a complex set of processes. Ingress rates are related to several factors including local hydraulics (Buffington and Montgomery, 1999), vertical and lateral interstitial exchange (Mathers and Wood, 2016), the relative size of the infiltrating and framework particles (Gibson et al., 2009), the concentration of suspended sediment and the settling flux (Brunke, 1999), and sediment transport capacity (Naden et al., 2016). Local hydraulic characteristics such as shear stress, flow velocity and Froude number have been associated with fine sediment accumulation, but studies often disagree regarding the gross influence of these hydraulic parameters (Petticrew et al., 2007). Beschta and Jackson (1979) found that the Froude number was positively associated with ingress, whilst Einstein (1968) and Carling (1984) found no relationship with flow parameters. It is possible that local hydraulic influences differ as a function of the dominant hydrological process. In low energy, slow-flowing waters, fine sediment ingress rates can be high because deposition rates are enhanced (Wood and Armitage, 1999), whereas in high-velocity areas sediment supply can be accentuated, enhancing the availability of fine sediment for subsequent infiltration (Frostick et al., 1984). As such, the availability of fines, as regulated by supply, transport capacity and, potentially biotic effects (e.g. Rice et al., 2016) may dominate the rate of infiltration irrespective of local hydraulics and framework size (Carling and McCahon, 1987; Sear, 1993).

Despite an enhanced understanding of the small-scale processes that control fine sediment infiltration (grains to patches; seconds to minutes) there is still no simple predictive model of fine sediment ingress than can be applied at large spatial and temporal scales. Moreover, despite a general understanding that both local hydraulics and sediment supply respond to hydrological processes that occur over longer, monthly-annual timescales, few studies have investigated the relations, over longer timescales, between variations in fine sediment ingress, suspended sediment concentrations and river discharge. This is unfortunate because there is a global need to set river management targets that maintain healthy rates of fine sediment delivery, deposition and transport (Collins et al., 2011) and gaining an understanding of the factors that influence fine sediment ingress on such time-scales is vital for developing relevant management strategies (e.g. Naden et al., 2016).

Both field and laboratory studies have identified fine sediment availability as a key determinant of ingress rates (Petts, 1984; Sear, 1993), with positive associations between suspended sediment concentration and ingress (Beschta and Jackson, 1979; Carling, 1984; Carling and McCahon, 1987). In general, fine sediment ingress rates are greatest during flood events when sediment transport rates are high and sediment is made available by scouring from pools and sub-armour deposits or is recruited to the channel via overland flow and other processes, including river bank collapse (Beschta et al., 1981; Sear, 1993; Petticrew et al., 2007). However, there is an apparent absence of studies which simultaneously investigate the relationship between flow, sediment supply and deposition to assess the potential explanatory power of different facets of these regimes (Wohl et al., 2015). Direct data on sediment transport and subsequent deposition is severely limited relative to river discharge and there is a need for more high resolution and long term suspended sediment data in order to characterise the magnitude, frequency, duration, timing and rate of change in suspended sediment levels (sensu Richter et al., 1996; Poff et al., 1997). Seeking greater understanding of the relations between the drivers and rates of fine sediment ingress over monthly-annual timescales is therefore valuable and consistent with Wohl et al.'s (2015) argument that the fine sediment regime can be managed through consideration of gross water and sediment balances.

In this regard, it is possible that ecohydrological approaches, which utilise a 'redundancy' methodology to associate key elements of hydrological time series with measurements of ecological health, may be useful (Richter et al., 1996; Olden and Poff, 2003). The purpose of such research has been to determine the ecologically relevant components or ‘facets' of discharge time series (duration, timing, frequency, magnitude, rate of change in flow events; Richter et al., 1996, Richter et al., 1997; Poff et al., 1997) that support ecologically healthy rivers, thereby facilitating the design of ‘environmental flows' (Acreman and Ferguson, 2010; Wharfe et al., 2014; Mustonen et al., 2016). Natural variability in stream processes is vital in maintaining diverse and healthy systems (Arthington et al., 2006) and these facets, rather than single simplistic metrics of a dynamic time series, are more appropriate for setting management targets (Richter et al., 1997). Given the plethora of indices that can be obtained from time series data (Poff, 1996), researchers must select which and how many indices are relevant to use for modelling purposes, particularly when many are inter-correlated (Olden and Poff, 2003).

Principal component analysis, a well-established multivariate technique, enables several variables that are inter-correlated to be analysed for the degree of similarity they characterise and subsequently transformed into a number of uncorrelated axes (variables) called ‘principal components’ which represent linear combinations of the original variables (Abdi and Williams, 2010). By identifying a reduced set of indices that represent the degree of variability in the time series, annual river management targets can be identified using a comprehensive statistical characterisation of relevant regime characteristics (Richter et al., 1997). This is an explicitly empirical method that requires careful application to avoid rejecting variables that are important, but which are not principal drivers of statistical variability (Monk et al., 2007). The method has been widely used beyond its original applications with flow discharge time series; for example to establish associations between stream temperature variability and instream communities (Jackson et al., 2007; Olden and Naiman, 2010; White et al., 2017), to group relevant instream geomorphic parameters for hydrological and ecological models (Singh et al., 2009; Faller et al., 2016) and to identify geographical properties associated with landslide susceptibility (Komac, 2006). At the core of this paper is an application of this methodology to fine sediment ingress. It is motivated by a conviction that the design and implementation of strategies that aim to manage levels of fine sediment storage in rivers would benefit from a better understanding of how facets of flow and sediment regimes relate to ingress rates.

This paper utilises novel measurements of fine sediment ingress collected over several months. These data were used with time series of discharge and turbidity, where the latter is shown to be representative of fine sediment availability, to identify key drivers of sediment ingress using the ecohydrological ‘redundancy’ approach. This analysis reveals the exploratory power of facets of the discharge and turbidity regimes as predictors of fine sediment ingress into riverbeds and seeks to establish the potential of employing simple empirical models, at temporal and spatial scales relevant to catchment management, using variables that are easily collected in the field.

A two-stage approach was employed:

• Classification of hydrological and turbidity time series into a small subset of indices that effectively characterise the dominant components (facets) of the series via a principal component analysis and redundancy reduction methodology (sensu Olden and Poff, 2003).

• Examination of the dominant facets of turbidity and discharge that influence sediment ingress using correlation matrices and the development of linear regression models using the principal component sample scores.