- Email: [email protected]
Use of continuous turbidity sensor in the prediction of fine sediment transport in the turbidity maximum of the Trent Estuary, UK
Estuarine, Coastal and Shelf Science 58 (2003) 645–652
Use of continuous turbidity sensor in the prediction of fine sediment transport in the turbidity maximum of the Trent Estuary, UK S.B. Mitchella,*, D.M. Lawlerb, J.R. Westc, J.S. Couperthwaiteb a
School of the Environment, University of Brighton, Cockcroft Building, Lewes Road, Brighton BN2 4GJ, UK b School of Geography, Earth and Environmental Sciences, University of Birmingham, Edgbaston, Birmingham B15 2TT, UK c Department of Civil Engineering, University of Birmingham, Edgbaston, Birmingham B15 2TT, UK Received 22 December 2002; accepted 19 May 2003
Abstract Results from continuous monitoring of turbidity and water level at Burringham, on the tidal River Trent, UK, are presented for the period 18 May 1997 to 9 February 1998. These measurements, together with detailed readings of velocity and suspended sediment concentration over an individual tidal cycle near the opposite bank at Derrythorpe, help to describe the mechanisms and behaviour of the turbidity maximum (TM). It is demonstrated that there is a distinct pattern of fine sediment movement that reflects a predictable system response to changing hydraulic features. It is shown that the TM in this system is highly mobile, and its location depends on antecedent fresh water flow, and tidal range. Approximate representative flood and ebb tide suspended sediment concentrations of up to 13 g/l over this nine-month period have been derived from the data and plotted against fresh water flow and tidal range, in order to show the relationship between these parameters. Three semi-empirical polynomial regression models have been tested for goodness of fit against available data. It was found that a partitioning approach, whereby data are grouped into different categories depending on antecedent fresh water flow, yielded the lowest standard error for the period analysed. Analysis of detailed observations of suspended sediment concentration and velocity measured over an individual tidal cycle also help to elucidate the mechanism of tidal pumping within this system. These results also help to give an estimate of the relative magnitude of suspended sediment fluxes during typical low fresh water flow conditions. It is estimated that for low fresh water flow conditions, a typical spring tide can mobilise at least an order of magnitude more sediment than a neap tide. Ó 2003 Elsevier Ltd. All rights reserved. Keywords: turbidity maximum; macrotidal; Humber Estuary; fine sediment transport; UK
1. Introduction Understanding the transport processes of fine, cohesive sediment within the turbidity maximum (TM) in macrotidal estuaries is important for the effective management of these complex systems. Sediment con-
* Corresponding author. E-mail addresses: [email protected] (S.B. Mitchell), [email protected] (D.M. Lawler), [email protected] (J.R. West). 0272-7714/03/$ - see front matter Ó 2003 Elsevier Ltd. All rights reserved. doi:10.1016/S0272-7714(03)00176-8
centrations found within this region may vary greatly between different estuaries and depend on the shape and bathymetry of the estuary, tidal range and river flow, as well as the type and availability of sediment. Recent advances in the reliability of continuous monitoring technology in aquatic environments means that it is generally easier to obtain data over longer time periods and at greater resolutions than was previously possible. Results from these continuous monitoring regimes, when calibrated against ground-truth observations of suspended sediment concentrations, can yield information both on the response of the system and on the
646
S.B. Mitchell et al. / Estuarine, Coastal and Shelf Science 58 (2003) 645–652
mechanisms by which the TM is maintained. Data yielded from field studies of this nature are essential for the calibration of mathematical models. In this paper, results will be presented from continuous monitoring of water level and suspended solids concentration (SSC) for a fixed monitor deployed from a culvert platform at Burringham, R. Trent, UK. The comparatively long-term nature of this deployment is important in determining the nature of seasonal variations in SSC, as well as quantifying the natural variability in SSC caused by cyclic variations in tidal range and fresh water flow patterns. From this and from other surveys carried out in the tidal Trent and (Yorkshire) Ouse rivers, a picture has been built up of the response of the system to variations in fresh water flow and tidal range. In particular, data describing the migration of the TM region upstream under low fresh water flow conditions, and downstream under high fresh water flow conditions, will be presented and discussed.
2. Turbidity maxima in macrotidal estuaries There are many descriptions of turbidity maxima in terms of their magnitude and migration in response to fresh water flow and tidal range in medium to high tidal range estuaries. These include the Scheldt, Belgium (Fettweis et al., 1998), Weser, Germany (Grabemann and Krause, 2001), Seine, France (Guezennec et al., 1999), Tamar, UK (Uncles et al., 1994) and Tay, UK (Dobereiner and McManus, 1983). Grab samples taken from the bed of the Tamar reveal the dependence of the position of the TM on the location of an area or ÔpoolÕ of mobile bed sediment which forms the source of the TM (Uncles et al., 1996). The continuing processes of erosion and deposition over each tidal cycle prevent this pool from settling for long enough to become part of a consolidated bed. The high ebb velocities caused by high fresh water flow conditions after a prolonged heavy rainfall event lead to a ÔflushingÕ effect whereby the residual (tidal-average) transport of sediment is downstream thus effecting a seaward migration of the TM (Nichols, 1993). Conversely, low dry-season fresh water flows lead to a relocation of the TM zone landwards. In the absence of significant changes in fresh water flow an estuary can exhibit substantially different patterns of sediment transport purely in response to changes in tidal range (Castaing and Allen, 1981; Vale and Sundby, 1987). Preliminary findings on the TM response in the Trent–Ouse system indicate a strong dependence of SSC values on tidal range (Arundale et al., 1997) and fresh water flow (Mitchell et al., 1998; Uncles et al., 1998). The concept of an easily erodible mobile Ômud reachÕ as proposed by Wellershaus (1981) for the Weser Estuary is important in explaining the magnitude and location of the TM. It may therefore be concluded that the SSC on
any given tide is related to the tidal range, the fresh water flow, the proximity of the mobile mud reach, and the availability of sediment within the mud reach. Since the availability of the mud reach is itself proportional to the antecedent river flow conditions, an approach taking into account mean flows prior to the day itself is justified.
3. Study site and methods Monitoring was undertaken on the tidal Trent, which is a canalised estuary approximately 80 km in length, stretching from Cromwell Weir in the south to the Humber confluence in the north (Fig. 1). The two contributing rivers Trent and Ouse have a combined catchment area of 24,240 km2, and are connected at the downstream end at Trent Falls (Fig. 1). Typical values of fresh water flow under high and low flow conditions, as well as tidal ranges, are given in Table 1. The main site of Burringham (0.75 W, 53.5 N) is located 15 km from the confluence of the Trent and Ouse rivers at Trent Falls (Fig. 1). At this site, a continuous fixed-point monitor was installed to measure turbidity and water level. The period of the deployment was 18 May 1997 to 9 February 1998. Water level was recorded at the site by a Dynamic-Logic Pressure Transducer mounted on a vertical support structure connected to an Environment Agency culvert platform. This was levelled to UK Ordnance Datum. Similarly, a Partech IR15C turbidity meter was mounted on the same support structure. This operated within a range of 0–13 g/l and was calibrated using pumped samples of suspended sediment concentration from the same location taken at the same time. These pumped samples were stored in darkness at 4 C before passing a known aliquot through dried, pre-weighed GF/C filter papers in the laboratory. The filter papers were then dried at 105 C for 24 h then re-weighed. Both the turbidity sensor and pressure transducer were positioned in the main channel flow. Campbell Scientific CR10X data loggers recorded turbidity and water level data at 1-min intervals, and stored as 15-min averages. Both sensors were exposed approximately 0.5 m above the flow at low water, dependent on hydraulic conditions of the day. Measurements were also made in the main channel, near to the opposite bank from the sampling platform at Burringham, near the village of Derrythorpe, during a complete individual tidal cycle on 30–31 July 1996. For this deployment, the sampling facilities aboard the UK Environment Agency vessel ÔSea VigilÕ were used. This vessel was tied to mooring dolphins for the whole of the tidal cycle. A manual winch was used to collect pumped samples from the main channel, and to measure velocity in situ by using an array of Braystoke impeller connected to a counter device. For both velocity and
647
S.B. Mitchell et al. / Estuarine, Coastal and Shelf Science 58 (2003) 645–652
Fig. 1. The Humber Estuary system.
pumped water samples, sampling was carried out at approximately half-hourly intervals at approximately 0.5 m depth intervals. The depth of sample was measured using a Valeport CTD probe attached to the deployment array, which contained the pumps and impellers. A fin was also attached to ensure that the sensors faced the flow direction, and a weight was attached to ensure verticality of the array. Despite this, under the highest velocities observed during the flood tide, the array was prevented from reaching down to the near-bed region due to excessive drag. Due to the combination of high tidal range and channel bathymetry the tidal cycle is characterised by short flood tides with high mean velocity, and longer, slower ebb tides. Slack water at the end of the flood tide is characterised by a gradual slowing of the flow, followed by a period of relatively stationary flow, followed by a gradual acceleration into the ebb tide. By contrast,
low slack water is characterised by a rapid change in flow direction, often accompanied by a steep fronted tidal bore during spring tides. For both the Trent and Ouse tidal rivers, this pattern of hydraulic activity leads to a more homogeneous distribution of sediment suspended in the water column during the flood tide than during the ebb tide (Uncles et al., 1998). The water may thus be considered to be well mixed under the turbulent flow of the flood tide, and weakly stratified during ebb
Table 1 Typical fresh water flows and tidal ranges, R. Trent Typical winter fresh water flow (m3/s)
Typical summer fresh water flow (m3/s)
Mean tidal range (springs) (m)
Mean tidal range (neaps) (m)
400 (N. Muskham)
30 (N. Muskham)
6.4 (Immingham)
3.2 (Immingham)
648
S.B. Mitchell et al. / Estuarine, Coastal and Shelf Science 58 (2003) 645–652
tides, with a considerable degree of settling during high slack water (HSW; Mitchell et al., 1998). 4. Results Calibration of pressure and turbidity sensors was carried out and the resulting linear relationships were incorporated into the analysis. Further details may be found in Couperthwaite et al. (1998). In general good correlation was found in both cases, except for SSC for the period around HSW, which in general occurred around 1 h after high water. This was thought to be due to the effect of preferential settling below the level of the sensor by larger particles, leaving a higher proportion of finer material at the level of the sensor itself (Mitchell and West, 2002). Results from the continuous monitor at Burringham, showing maximum tidal water level, fresh water flow, flood tide and ebb tide SSC from 18 May 1997 to 9 February 1998, are shown in Fig. 2. Representative flood tide SSC values have been calculated by taking the mean of the four values from the hour before high water at Burringham. Representative ebb tide SSC values have been obtained by calculating the mean value for the hour before the sensor emerges from the water. In this way, values for SSC have been obtained which are thought to be approximately representative of wellmixed conditions, thus the effects of sediment-induced density stratification on the observed values may to a first approximation be ignored. Values of SSC greater than 13 g/l have in most cases caused the sensor to return an error value, and where this has occurred, the point has been omitted. Of particular interest in Fig. 2 is the strong dependence of flood tide SSC on tidal range under low fresh water flow conditions, as described by other authors (Fettweis et al., 1998; Guezennec et al., 1999). However, a significant hysteresis effect, whereby SSC values during periods of declining tidal range differ from those observed during rising tidal range (Guezennec et al., 1999), is not observed. Under high fresh water flow conditions, SSC is suppressed due to the migration of the TM region downstream from the study site. It is important to note that this feature may be observed to depend on the antecedent flow for the period, rather than the flow on the day itself, owing to the lag between changes in hydraulic and sediment transport regimes. In order to help interpret the results from the fixedpoint turbidity sensor attached to the platform at Burringham, observations of depth variations in velocity and SSC were made for Derrythorpe, close to the opposite bank, as described previously. Results of depth profiles of SSC and velocity for a single tidal cycle, measured for the overnight tidal cycle on 30–31 July 1996, are shown in Fig. 3(a–h). A positive velocity represents a velocity in the downstream (ebb) direction. Each plot represents vertical profiles measured at a time
relative to HSW. Thus Fig. 3(a) represents a time during the flood tide, 1 h and 15 min before HSW, when velocities are strongly negative, and the water column is well mixed. A high degree of vertical stratification, coupled with a relatively slow moving near-bed layer, may be noted for the 2 h after HSW (Fig. 3b–d), after which the SSC stratification is broken down and the water becomes well mixed over the late ebb (Fig. 3e–h). This helps to explain the mechanism of landward dispersion of sediment by tidal pumping under high tidal range, low fresh water flow conditions, as described by Dyer (1997) and outlined for this system in Mitchell et al. (1998). 5. Discussion The marked difference in flood and ebb SSC depending on antecedent fresh water flow conditions may be explained in terms of the availability of sediment on the bed. Thus for the similar (spring tide) hydraulic conditions, the high antecedent flow of 6 July (identified as ÔAÕ on Fig. 2) has flushed the mobile pool of sediment downstream of the study site, resulting in low observed values of SSC. By 23 July (ÔBÕ on Fig. 2), the spring tide flood currents are more successful in suspending material as there is now more of it available, following the lower antecedent fresh water flow, coupled with tidal pumping of sediment under the influence of spring tides. By 23 August, peak flood tide SSC values are less than at the end of the previous month, pointing to the possibility of tidal pumping of the mobile sediment pool to a location upstream of the observation point. In general the lower ebb tide currents cause less sediment to be suspended than the higher flood currents. However, results from detailed vertical profiling (Fig. 3) demonstrate that care must be taken when interpreting the results from fixed monitors. For example, it is shown (Fig. 3b–d) that for the early part of the ebb tide, suspended sediment does not reach the upper part of the water column due to stratification causing damping of vertical turbulent transport processes (Geyer, 1993). Since the water level falls below the level of the sensor for the latter part of the ebb tide, it is possible that the sensor generally fails to detect ebb tide transport to the same degree of accuracy as for the flood tide. The results of Fig. 2 enable the formulation of simple empirical relationships yielding approximate representative flood and ebb tide SSC based on observed peak tidal water level and the observed fresh water flow for the previous day. A one-day time lag was selected as this appears to offer the best correlation with SSC. Values of representative flood tide and ebb tide SSC (as outlined in the previous section), were correlated against observed peak tidal water level. For the purposes of the analysis, ÔhighÕ and ÔlowÕ fresh water flows are those above and below mean fresh water flow for the period (83.8 m3/s). As in Couperthwaite et al. (1998), a second
S.B. Mitchell et al. / Estuarine, Coastal and Shelf Science 58 (2003) 645–652
649
Fig. 2. (a) Daily mean fresh water flow (N. Muskham) and observed peak tidal water level and (b) calculated representative flood and ebb tide suspended sediment concentration.
order relationship has been assumed for each line, in order to take into account the dependence of sediment suspension on bed shear stress, which is proportional to the square of flow velocity. This simple partitioning model, based on using different formulae for SSC, depending on whether fresh water flow is above or below a threshold value, is referred to as MODEL 1 in subsequent discussion.
Two further approaches have been used. The first (MODEL 2) is a simple regression analysis to obtain the Ôbest fitÕ curve linking these three quantities of the form: SSC ¼ AðTHÞ2 þ BðTHÞ þ C lnðQt1 Þ þ D
ð1Þ
where A, B, C and D are constants obtained by regression analysis, TH (m) is the peak observed tidal height and Qt1 (m3/s) the previous day’s observed fresh
650
S.B. Mitchell et al. / Estuarine, Coastal and Shelf Science 58 (2003) 645–652
Fig. 3. Velocity and SSC, vertical profiles, Derrythorpe, R. Trent, 30–31 July, 1996. Top horizontal axes and squares: velocity (m/s). Bottom horizontal axes and diamonds: SSC (g/l). Ebb velocities positive. Vertical axes: depth (m).
651
S.B. Mitchell et al. / Estuarine, Coastal and Shelf Science 58 (2003) 645–652
water flow. The second (MODEL 3) makes use of a series of polynomial regression equations for a range of different antecedent fresh water flow conditions, together with estimation of coefficients for the polynomial equation based on linear interpolation between those obtained for the primary fresh water flow values. Thus, all the flood tide SSC values were sub-divided into six different categories (having approximately 70 points in each), and second order polynomial regression lines of the form: 2
SSC ¼ AðTHÞ þ BðTHÞ þ C
ð2Þ
were obtained for each of these six categories. Values of A, B and C were obtained by linear interpolation between the values of these quantities obtained for the six different categories for which regression analysis was carried out. The same procedure was adopted for ebb tide SSC values. This method may be termed a Ôdynamic partitioningÕ approach. Standard errors for each of the predictors (square root of the mean of the squares of the error between observed and predicted) are given in Table 2. It can be seen from Table 2 that MODEL 3 offers the best approach generally for estimating the SSC for known values of maximum tidal height and antecedent fresh water flow. This is clearly a reflection of the complexity of the physical relationship between the hydraulics and the fine sediment transport processes in the estuary. Further work is planned to test the range of applicability of this modelling approach in the light of other data collected from the continuous monitor at Burringham. Carrying out estimates for flood and ebb tide suspended sediment flux is highly problematic when these are based on a fixed-point sensor. However, it is clear from the results presented that there must be a lower suspended sediment flux during ebb tides, based on the observation that the mobile pool of available sediment moves landwards under high tidal range, low fresh water flow conditions. It is equally clear that under high fresh water flow, sediment moves rapidly seawards. Comparing tides that occur during low fresh water flow periods, however, it is clear that a typical spring flood
tide transports an order of magnitude more sediment than a typical neap tide. The same relationship is much less clear for the ebb tides, as the suspended sediment concentrations that are observed appear to depend on the location of the sediment pool relative to the observation point.
6. Conclusions Results from observations of turbidity and water level at Burringham, R. Trent have been made for the period 18 May 1997 to 9 February 1998. These results, together with measurements of vertical profiles of velocity and SSC in a nearby location have enabled some conclusions to be drawn relating to the response of the TM in the tidal Trent. 1. The use of fixed point, continuous monitors over time periods of several months or more helps to build up a long-term picture of how sediment transport in estuaries is affected by tidal range and fresh water flow in macrotidal estuaries. 2. The temporal variation in vertical profiles of velocity and SSC means that interpretation of results from continuous monitors must be carried out with caution, owing to the high degree of suspended sediment stratification at HSW and over the early ebb tide. 3. A semi-empirical polynomial regression model may be used to predict SSC for flood and ebb phases estimated from maximum tidal water level and antecedent fresh water flow. Of the various approaches attempted in this analysis, it was found that an algorithm based on a series of best-fit polynomials for different ranges of fresh water flow, gave the least values of standard error. 4. Estimates of mass transport rate suggest that most of the sediment transport takes place for spring tide conditions. Under low fresh water flow conditions, a typical flood tide will lead to a suspended sediment concentration an order of magnitude more than that of a typical neap tide.
Table 2 Details of three approaches used to predict flood and ebb tide SSC Model number
Tide state
Model approach
Algorithm details
Standard error (g/l)
MODEL 1 MODEL 2 MODEL 3
Flood Flood Flood
Simple partition Simple regression Dynamic partition
0.106 0.082 0.074
MODEL 1 MODEL 2 MODEL 3
Ebb Ebb Ebb
Simple partition Simple regression Dynamic partition
See text for details SSC = 2.23(TH)2 + 35.2(TH) 5.79 ln(Qt1) 109.0 SSC = A(TH)2 + B(TH) + C where A, B, C obtained by linear interpolation between values obtained by regression for six different values of Qt1 See text for details SSC = 1.28(TH)2 14.0(TH) 2.76 ln(Qt1) + 49.0 SSC = A(TH)2 + B(TH) + C where A, B, C obtained by linear interpolation between values obtained by regression for six different values of Qt1
0.071 0.068 0.061
652
S.B. Mitchell et al. / Estuarine, Coastal and Shelf Science 58 (2003) 645–652
Acknowledgements Data collection for this research was funded by NERC under the LOIS initiative [code: RACS 231 (C)/ (R)]. The authors are very grateful to the UK Environment Agency for the provision of data and access to sampling sites. The assistance of Ray Hodson and Richard Johnson in managing fieldwork programmes is also gratefully acknowledged.
References Arundale, A.M.W., Darbyshire, E.J., Hunt, S.J., Schmitz, K.G., West, J.R., 1997. Turbidity maxima formation in four estuaries. In: Burt, N., Parker, R., Watts, J. (Eds.), Cohesive Sediments. Wiley, Chichester, UK, pp. 135–146. Castaing, P., Allen, G.P., 1981. Mechanisms controlling seaward escape of suspended sediment from the Gironde: a macrotidal estuary in France. Mar. Geol. 40, 101–118. Couperthwaite, J.S., Mitchell, S.B., West, J.R., Lawler, D.M., 1998. Cohesive sediment dynamics on an inter-tidal zone on the tidal Trent. Mar. Pollut. Bull. 37, 144–154. Dobereiner, C., McManus, J., 1983. Turbidity maximum migration and harbour siltation in the Tay estuary. Can. J. Fish. Aquat. Sci. 40 (Suppl. 1), 117–129. Dyer, K.R., 1997. Estuaries: A Physical Introduction, second ed. Wiley, Chichester, UK, 195 pp. Fettweis, M., Sas, M., Monbaliu, J., 1998. Seasonal, neap–spring and tidal variation of cohesive sediment concentration in the Scheldt estuary, Belgium. Estuar. Coast. Shelf Sci. 47, 21–36.
Geyer, W.R., 1993. The importance of suppression of turbulence by stratification on the estuarine turbidity maximum. Estuaries 16(1), 113–125. Grabemann, I., Krause, G., 2001. On different time scales of suspended matter dynamics in the Weser estuary. Estuaries 24, 688–698. Guezennec, L., Lafite, R., Dupont, J.-P., Meyer, R., Boust, D., 1999. Hydrodynamics of suspended particulate matter in the tidal freshwater zone of a macrotidal estuary (the Seine estuary, France). Estuaries 22, 717–727. Mitchell, S.B., West, J.R., 2002. Particle size distribution in an estuarine turbidity maximum region. In: Winterwerp, J.C., Kranenburg, C. (Eds.), Fine Sediment Dynamics in the Marine Environment. Elsevier, Amsterdam, pp. 251–263. Mitchell, S.B., West, J.R., Arundale, A.M.W., Guymer, I., Couperthwaite, J.S., 1998. Dynamics of the turbidity maxima in the upper Humber estuary system. Mar. Pollut. Bull. 37, 190–205. Nichols, M.M., 1993. Response of coastal plain estuaries to episodic events in the Chesapeake Bay region. In: Mehta, A.J. (Ed.), Nearshore and Estuarine Cohesive Sediment Transport. American Geophysical Union, Washington, DC, pp. 1–20. Uncles, R.J., Barton, M.L., Stephens, J.A., 1994. Seasonal variability of fine-sediment concentrations in the turbidity maximum region of the Tamar estuary. Estuar. Coast. Shelf Sci. 38, 19–39. Uncles, R.J., Barton, M.L., Stephens, J.A., 1996. Seasonal variability of mobile mud deposits in the Tamar estuary. In: Pattiaratchi, C. (Ed.), Mixing in Estuaries and Coastal Seas. American Geophysical Union, Washington, DC, pp. 374–387. Uncles, R.J., Easton, A.E., Griffiths, M.L., Harris, C., Howland, R.J.M., King, R.S., Morris, A.W., Plummer, D.H., 1998. Seasonality of the turbidity maximum in the Humber–Ouse estuary. Mar. Pollut. Bull. 37, 206–215. Vale, C., Sundby, B., 1987. Suspended sediment fluctuations in the Tagus estuary on semi-diurnal and fortnightly time scales. Estuar. Coast. Shelf Sci. 25, 495–508. Wellershaus, S., 1981. Turbidity maximum and mud shoaling in the Weser estuary. Arch. Hydrobiol. 92, 161–198.
Use of continuous turbidity sensor in the prediction of fine sediment transport in the turbidity maximum of the Trent Estuary, UK S.B. Mitchella,*, D.M. Lawlerb, J.R. Westc, J.S. Couperthwaiteb a
School of the Environment, University of Brighton, Cockcroft Building, Lewes Road, Brighton BN2 4GJ, UK b School of Geography, Earth and Environmental Sciences, University of Birmingham, Edgbaston, Birmingham B15 2TT, UK c Department of Civil Engineering, University of Birmingham, Edgbaston, Birmingham B15 2TT, UK Received 22 December 2002; accepted 19 May 2003
Abstract Results from continuous monitoring of turbidity and water level at Burringham, on the tidal River Trent, UK, are presented for the period 18 May 1997 to 9 February 1998. These measurements, together with detailed readings of velocity and suspended sediment concentration over an individual tidal cycle near the opposite bank at Derrythorpe, help to describe the mechanisms and behaviour of the turbidity maximum (TM). It is demonstrated that there is a distinct pattern of fine sediment movement that reflects a predictable system response to changing hydraulic features. It is shown that the TM in this system is highly mobile, and its location depends on antecedent fresh water flow, and tidal range. Approximate representative flood and ebb tide suspended sediment concentrations of up to 13 g/l over this nine-month period have been derived from the data and plotted against fresh water flow and tidal range, in order to show the relationship between these parameters. Three semi-empirical polynomial regression models have been tested for goodness of fit against available data. It was found that a partitioning approach, whereby data are grouped into different categories depending on antecedent fresh water flow, yielded the lowest standard error for the period analysed. Analysis of detailed observations of suspended sediment concentration and velocity measured over an individual tidal cycle also help to elucidate the mechanism of tidal pumping within this system. These results also help to give an estimate of the relative magnitude of suspended sediment fluxes during typical low fresh water flow conditions. It is estimated that for low fresh water flow conditions, a typical spring tide can mobilise at least an order of magnitude more sediment than a neap tide. Ó 2003 Elsevier Ltd. All rights reserved. Keywords: turbidity maximum; macrotidal; Humber Estuary; fine sediment transport; UK
1. Introduction Understanding the transport processes of fine, cohesive sediment within the turbidity maximum (TM) in macrotidal estuaries is important for the effective management of these complex systems. Sediment con-
* Corresponding author. E-mail addresses: [email protected] (S.B. Mitchell), [email protected] (D.M. Lawler), [email protected] (J.R. West). 0272-7714/03/$ - see front matter Ó 2003 Elsevier Ltd. All rights reserved. doi:10.1016/S0272-7714(03)00176-8
centrations found within this region may vary greatly between different estuaries and depend on the shape and bathymetry of the estuary, tidal range and river flow, as well as the type and availability of sediment. Recent advances in the reliability of continuous monitoring technology in aquatic environments means that it is generally easier to obtain data over longer time periods and at greater resolutions than was previously possible. Results from these continuous monitoring regimes, when calibrated against ground-truth observations of suspended sediment concentrations, can yield information both on the response of the system and on the
646
S.B. Mitchell et al. / Estuarine, Coastal and Shelf Science 58 (2003) 645–652
mechanisms by which the TM is maintained. Data yielded from field studies of this nature are essential for the calibration of mathematical models. In this paper, results will be presented from continuous monitoring of water level and suspended solids concentration (SSC) for a fixed monitor deployed from a culvert platform at Burringham, R. Trent, UK. The comparatively long-term nature of this deployment is important in determining the nature of seasonal variations in SSC, as well as quantifying the natural variability in SSC caused by cyclic variations in tidal range and fresh water flow patterns. From this and from other surveys carried out in the tidal Trent and (Yorkshire) Ouse rivers, a picture has been built up of the response of the system to variations in fresh water flow and tidal range. In particular, data describing the migration of the TM region upstream under low fresh water flow conditions, and downstream under high fresh water flow conditions, will be presented and discussed.
2. Turbidity maxima in macrotidal estuaries There are many descriptions of turbidity maxima in terms of their magnitude and migration in response to fresh water flow and tidal range in medium to high tidal range estuaries. These include the Scheldt, Belgium (Fettweis et al., 1998), Weser, Germany (Grabemann and Krause, 2001), Seine, France (Guezennec et al., 1999), Tamar, UK (Uncles et al., 1994) and Tay, UK (Dobereiner and McManus, 1983). Grab samples taken from the bed of the Tamar reveal the dependence of the position of the TM on the location of an area or ÔpoolÕ of mobile bed sediment which forms the source of the TM (Uncles et al., 1996). The continuing processes of erosion and deposition over each tidal cycle prevent this pool from settling for long enough to become part of a consolidated bed. The high ebb velocities caused by high fresh water flow conditions after a prolonged heavy rainfall event lead to a ÔflushingÕ effect whereby the residual (tidal-average) transport of sediment is downstream thus effecting a seaward migration of the TM (Nichols, 1993). Conversely, low dry-season fresh water flows lead to a relocation of the TM zone landwards. In the absence of significant changes in fresh water flow an estuary can exhibit substantially different patterns of sediment transport purely in response to changes in tidal range (Castaing and Allen, 1981; Vale and Sundby, 1987). Preliminary findings on the TM response in the Trent–Ouse system indicate a strong dependence of SSC values on tidal range (Arundale et al., 1997) and fresh water flow (Mitchell et al., 1998; Uncles et al., 1998). The concept of an easily erodible mobile Ômud reachÕ as proposed by Wellershaus (1981) for the Weser Estuary is important in explaining the magnitude and location of the TM. It may therefore be concluded that the SSC on
any given tide is related to the tidal range, the fresh water flow, the proximity of the mobile mud reach, and the availability of sediment within the mud reach. Since the availability of the mud reach is itself proportional to the antecedent river flow conditions, an approach taking into account mean flows prior to the day itself is justified.
3. Study site and methods Monitoring was undertaken on the tidal Trent, which is a canalised estuary approximately 80 km in length, stretching from Cromwell Weir in the south to the Humber confluence in the north (Fig. 1). The two contributing rivers Trent and Ouse have a combined catchment area of 24,240 km2, and are connected at the downstream end at Trent Falls (Fig. 1). Typical values of fresh water flow under high and low flow conditions, as well as tidal ranges, are given in Table 1. The main site of Burringham (0.75 W, 53.5 N) is located 15 km from the confluence of the Trent and Ouse rivers at Trent Falls (Fig. 1). At this site, a continuous fixed-point monitor was installed to measure turbidity and water level. The period of the deployment was 18 May 1997 to 9 February 1998. Water level was recorded at the site by a Dynamic-Logic Pressure Transducer mounted on a vertical support structure connected to an Environment Agency culvert platform. This was levelled to UK Ordnance Datum. Similarly, a Partech IR15C turbidity meter was mounted on the same support structure. This operated within a range of 0–13 g/l and was calibrated using pumped samples of suspended sediment concentration from the same location taken at the same time. These pumped samples were stored in darkness at 4 C before passing a known aliquot through dried, pre-weighed GF/C filter papers in the laboratory. The filter papers were then dried at 105 C for 24 h then re-weighed. Both the turbidity sensor and pressure transducer were positioned in the main channel flow. Campbell Scientific CR10X data loggers recorded turbidity and water level data at 1-min intervals, and stored as 15-min averages. Both sensors were exposed approximately 0.5 m above the flow at low water, dependent on hydraulic conditions of the day. Measurements were also made in the main channel, near to the opposite bank from the sampling platform at Burringham, near the village of Derrythorpe, during a complete individual tidal cycle on 30–31 July 1996. For this deployment, the sampling facilities aboard the UK Environment Agency vessel ÔSea VigilÕ were used. This vessel was tied to mooring dolphins for the whole of the tidal cycle. A manual winch was used to collect pumped samples from the main channel, and to measure velocity in situ by using an array of Braystoke impeller connected to a counter device. For both velocity and
647
S.B. Mitchell et al. / Estuarine, Coastal and Shelf Science 58 (2003) 645–652
Fig. 1. The Humber Estuary system.
pumped water samples, sampling was carried out at approximately half-hourly intervals at approximately 0.5 m depth intervals. The depth of sample was measured using a Valeport CTD probe attached to the deployment array, which contained the pumps and impellers. A fin was also attached to ensure that the sensors faced the flow direction, and a weight was attached to ensure verticality of the array. Despite this, under the highest velocities observed during the flood tide, the array was prevented from reaching down to the near-bed region due to excessive drag. Due to the combination of high tidal range and channel bathymetry the tidal cycle is characterised by short flood tides with high mean velocity, and longer, slower ebb tides. Slack water at the end of the flood tide is characterised by a gradual slowing of the flow, followed by a period of relatively stationary flow, followed by a gradual acceleration into the ebb tide. By contrast,
low slack water is characterised by a rapid change in flow direction, often accompanied by a steep fronted tidal bore during spring tides. For both the Trent and Ouse tidal rivers, this pattern of hydraulic activity leads to a more homogeneous distribution of sediment suspended in the water column during the flood tide than during the ebb tide (Uncles et al., 1998). The water may thus be considered to be well mixed under the turbulent flow of the flood tide, and weakly stratified during ebb
Table 1 Typical fresh water flows and tidal ranges, R. Trent Typical winter fresh water flow (m3/s)
Typical summer fresh water flow (m3/s)
Mean tidal range (springs) (m)
Mean tidal range (neaps) (m)
400 (N. Muskham)
30 (N. Muskham)
6.4 (Immingham)
3.2 (Immingham)
648
S.B. Mitchell et al. / Estuarine, Coastal and Shelf Science 58 (2003) 645–652
tides, with a considerable degree of settling during high slack water (HSW; Mitchell et al., 1998). 4. Results Calibration of pressure and turbidity sensors was carried out and the resulting linear relationships were incorporated into the analysis. Further details may be found in Couperthwaite et al. (1998). In general good correlation was found in both cases, except for SSC for the period around HSW, which in general occurred around 1 h after high water. This was thought to be due to the effect of preferential settling below the level of the sensor by larger particles, leaving a higher proportion of finer material at the level of the sensor itself (Mitchell and West, 2002). Results from the continuous monitor at Burringham, showing maximum tidal water level, fresh water flow, flood tide and ebb tide SSC from 18 May 1997 to 9 February 1998, are shown in Fig. 2. Representative flood tide SSC values have been calculated by taking the mean of the four values from the hour before high water at Burringham. Representative ebb tide SSC values have been obtained by calculating the mean value for the hour before the sensor emerges from the water. In this way, values for SSC have been obtained which are thought to be approximately representative of wellmixed conditions, thus the effects of sediment-induced density stratification on the observed values may to a first approximation be ignored. Values of SSC greater than 13 g/l have in most cases caused the sensor to return an error value, and where this has occurred, the point has been omitted. Of particular interest in Fig. 2 is the strong dependence of flood tide SSC on tidal range under low fresh water flow conditions, as described by other authors (Fettweis et al., 1998; Guezennec et al., 1999). However, a significant hysteresis effect, whereby SSC values during periods of declining tidal range differ from those observed during rising tidal range (Guezennec et al., 1999), is not observed. Under high fresh water flow conditions, SSC is suppressed due to the migration of the TM region downstream from the study site. It is important to note that this feature may be observed to depend on the antecedent flow for the period, rather than the flow on the day itself, owing to the lag between changes in hydraulic and sediment transport regimes. In order to help interpret the results from the fixedpoint turbidity sensor attached to the platform at Burringham, observations of depth variations in velocity and SSC were made for Derrythorpe, close to the opposite bank, as described previously. Results of depth profiles of SSC and velocity for a single tidal cycle, measured for the overnight tidal cycle on 30–31 July 1996, are shown in Fig. 3(a–h). A positive velocity represents a velocity in the downstream (ebb) direction. Each plot represents vertical profiles measured at a time
relative to HSW. Thus Fig. 3(a) represents a time during the flood tide, 1 h and 15 min before HSW, when velocities are strongly negative, and the water column is well mixed. A high degree of vertical stratification, coupled with a relatively slow moving near-bed layer, may be noted for the 2 h after HSW (Fig. 3b–d), after which the SSC stratification is broken down and the water becomes well mixed over the late ebb (Fig. 3e–h). This helps to explain the mechanism of landward dispersion of sediment by tidal pumping under high tidal range, low fresh water flow conditions, as described by Dyer (1997) and outlined for this system in Mitchell et al. (1998). 5. Discussion The marked difference in flood and ebb SSC depending on antecedent fresh water flow conditions may be explained in terms of the availability of sediment on the bed. Thus for the similar (spring tide) hydraulic conditions, the high antecedent flow of 6 July (identified as ÔAÕ on Fig. 2) has flushed the mobile pool of sediment downstream of the study site, resulting in low observed values of SSC. By 23 July (ÔBÕ on Fig. 2), the spring tide flood currents are more successful in suspending material as there is now more of it available, following the lower antecedent fresh water flow, coupled with tidal pumping of sediment under the influence of spring tides. By 23 August, peak flood tide SSC values are less than at the end of the previous month, pointing to the possibility of tidal pumping of the mobile sediment pool to a location upstream of the observation point. In general the lower ebb tide currents cause less sediment to be suspended than the higher flood currents. However, results from detailed vertical profiling (Fig. 3) demonstrate that care must be taken when interpreting the results from fixed monitors. For example, it is shown (Fig. 3b–d) that for the early part of the ebb tide, suspended sediment does not reach the upper part of the water column due to stratification causing damping of vertical turbulent transport processes (Geyer, 1993). Since the water level falls below the level of the sensor for the latter part of the ebb tide, it is possible that the sensor generally fails to detect ebb tide transport to the same degree of accuracy as for the flood tide. The results of Fig. 2 enable the formulation of simple empirical relationships yielding approximate representative flood and ebb tide SSC based on observed peak tidal water level and the observed fresh water flow for the previous day. A one-day time lag was selected as this appears to offer the best correlation with SSC. Values of representative flood tide and ebb tide SSC (as outlined in the previous section), were correlated against observed peak tidal water level. For the purposes of the analysis, ÔhighÕ and ÔlowÕ fresh water flows are those above and below mean fresh water flow for the period (83.8 m3/s). As in Couperthwaite et al. (1998), a second
S.B. Mitchell et al. / Estuarine, Coastal and Shelf Science 58 (2003) 645–652
649
Fig. 2. (a) Daily mean fresh water flow (N. Muskham) and observed peak tidal water level and (b) calculated representative flood and ebb tide suspended sediment concentration.
order relationship has been assumed for each line, in order to take into account the dependence of sediment suspension on bed shear stress, which is proportional to the square of flow velocity. This simple partitioning model, based on using different formulae for SSC, depending on whether fresh water flow is above or below a threshold value, is referred to as MODEL 1 in subsequent discussion.
Two further approaches have been used. The first (MODEL 2) is a simple regression analysis to obtain the Ôbest fitÕ curve linking these three quantities of the form: SSC ¼ AðTHÞ2 þ BðTHÞ þ C lnðQt1 Þ þ D
ð1Þ
where A, B, C and D are constants obtained by regression analysis, TH (m) is the peak observed tidal height and Qt1 (m3/s) the previous day’s observed fresh
650
S.B. Mitchell et al. / Estuarine, Coastal and Shelf Science 58 (2003) 645–652
Fig. 3. Velocity and SSC, vertical profiles, Derrythorpe, R. Trent, 30–31 July, 1996. Top horizontal axes and squares: velocity (m/s). Bottom horizontal axes and diamonds: SSC (g/l). Ebb velocities positive. Vertical axes: depth (m).
651
S.B. Mitchell et al. / Estuarine, Coastal and Shelf Science 58 (2003) 645–652
water flow. The second (MODEL 3) makes use of a series of polynomial regression equations for a range of different antecedent fresh water flow conditions, together with estimation of coefficients for the polynomial equation based on linear interpolation between those obtained for the primary fresh water flow values. Thus, all the flood tide SSC values were sub-divided into six different categories (having approximately 70 points in each), and second order polynomial regression lines of the form: 2
SSC ¼ AðTHÞ þ BðTHÞ þ C
ð2Þ
were obtained for each of these six categories. Values of A, B and C were obtained by linear interpolation between the values of these quantities obtained for the six different categories for which regression analysis was carried out. The same procedure was adopted for ebb tide SSC values. This method may be termed a Ôdynamic partitioningÕ approach. Standard errors for each of the predictors (square root of the mean of the squares of the error between observed and predicted) are given in Table 2. It can be seen from Table 2 that MODEL 3 offers the best approach generally for estimating the SSC for known values of maximum tidal height and antecedent fresh water flow. This is clearly a reflection of the complexity of the physical relationship between the hydraulics and the fine sediment transport processes in the estuary. Further work is planned to test the range of applicability of this modelling approach in the light of other data collected from the continuous monitor at Burringham. Carrying out estimates for flood and ebb tide suspended sediment flux is highly problematic when these are based on a fixed-point sensor. However, it is clear from the results presented that there must be a lower suspended sediment flux during ebb tides, based on the observation that the mobile pool of available sediment moves landwards under high tidal range, low fresh water flow conditions. It is equally clear that under high fresh water flow, sediment moves rapidly seawards. Comparing tides that occur during low fresh water flow periods, however, it is clear that a typical spring flood
tide transports an order of magnitude more sediment than a typical neap tide. The same relationship is much less clear for the ebb tides, as the suspended sediment concentrations that are observed appear to depend on the location of the sediment pool relative to the observation point.
6. Conclusions Results from observations of turbidity and water level at Burringham, R. Trent have been made for the period 18 May 1997 to 9 February 1998. These results, together with measurements of vertical profiles of velocity and SSC in a nearby location have enabled some conclusions to be drawn relating to the response of the TM in the tidal Trent. 1. The use of fixed point, continuous monitors over time periods of several months or more helps to build up a long-term picture of how sediment transport in estuaries is affected by tidal range and fresh water flow in macrotidal estuaries. 2. The temporal variation in vertical profiles of velocity and SSC means that interpretation of results from continuous monitors must be carried out with caution, owing to the high degree of suspended sediment stratification at HSW and over the early ebb tide. 3. A semi-empirical polynomial regression model may be used to predict SSC for flood and ebb phases estimated from maximum tidal water level and antecedent fresh water flow. Of the various approaches attempted in this analysis, it was found that an algorithm based on a series of best-fit polynomials for different ranges of fresh water flow, gave the least values of standard error. 4. Estimates of mass transport rate suggest that most of the sediment transport takes place for spring tide conditions. Under low fresh water flow conditions, a typical flood tide will lead to a suspended sediment concentration an order of magnitude more than that of a typical neap tide.
Table 2 Details of three approaches used to predict flood and ebb tide SSC Model number
Tide state
Model approach
Algorithm details
Standard error (g/l)
MODEL 1 MODEL 2 MODEL 3
Flood Flood Flood
Simple partition Simple regression Dynamic partition
0.106 0.082 0.074
MODEL 1 MODEL 2 MODEL 3
Ebb Ebb Ebb
Simple partition Simple regression Dynamic partition
See text for details SSC = 2.23(TH)2 + 35.2(TH) 5.79 ln(Qt1) 109.0 SSC = A(TH)2 + B(TH) + C where A, B, C obtained by linear interpolation between values obtained by regression for six different values of Qt1 See text for details SSC = 1.28(TH)2 14.0(TH) 2.76 ln(Qt1) + 49.0 SSC = A(TH)2 + B(TH) + C where A, B, C obtained by linear interpolation between values obtained by regression for six different values of Qt1
0.071 0.068 0.061
652
S.B. Mitchell et al. / Estuarine, Coastal and Shelf Science 58 (2003) 645–652
Acknowledgements Data collection for this research was funded by NERC under the LOIS initiative [code: RACS 231 (C)/ (R)]. The authors are very grateful to the UK Environment Agency for the provision of data and access to sampling sites. The assistance of Ray Hodson and Richard Johnson in managing fieldwork programmes is also gratefully acknowledged.
References Arundale, A.M.W., Darbyshire, E.J., Hunt, S.J., Schmitz, K.G., West, J.R., 1997. Turbidity maxima formation in four estuaries. In: Burt, N., Parker, R., Watts, J. (Eds.), Cohesive Sediments. Wiley, Chichester, UK, pp. 135–146. Castaing, P., Allen, G.P., 1981. Mechanisms controlling seaward escape of suspended sediment from the Gironde: a macrotidal estuary in France. Mar. Geol. 40, 101–118. Couperthwaite, J.S., Mitchell, S.B., West, J.R., Lawler, D.M., 1998. Cohesive sediment dynamics on an inter-tidal zone on the tidal Trent. Mar. Pollut. Bull. 37, 144–154. Dobereiner, C., McManus, J., 1983. Turbidity maximum migration and harbour siltation in the Tay estuary. Can. J. Fish. Aquat. Sci. 40 (Suppl. 1), 117–129. Dyer, K.R., 1997. Estuaries: A Physical Introduction, second ed. Wiley, Chichester, UK, 195 pp. Fettweis, M., Sas, M., Monbaliu, J., 1998. Seasonal, neap–spring and tidal variation of cohesive sediment concentration in the Scheldt estuary, Belgium. Estuar. Coast. Shelf Sci. 47, 21–36.
Geyer, W.R., 1993. The importance of suppression of turbulence by stratification on the estuarine turbidity maximum. Estuaries 16(1), 113–125. Grabemann, I., Krause, G., 2001. On different time scales of suspended matter dynamics in the Weser estuary. Estuaries 24, 688–698. Guezennec, L., Lafite, R., Dupont, J.-P., Meyer, R., Boust, D., 1999. Hydrodynamics of suspended particulate matter in the tidal freshwater zone of a macrotidal estuary (the Seine estuary, France). Estuaries 22, 717–727. Mitchell, S.B., West, J.R., 2002. Particle size distribution in an estuarine turbidity maximum region. In: Winterwerp, J.C., Kranenburg, C. (Eds.), Fine Sediment Dynamics in the Marine Environment. Elsevier, Amsterdam, pp. 251–263. Mitchell, S.B., West, J.R., Arundale, A.M.W., Guymer, I., Couperthwaite, J.S., 1998. Dynamics of the turbidity maxima in the upper Humber estuary system. Mar. Pollut. Bull. 37, 190–205. Nichols, M.M., 1993. Response of coastal plain estuaries to episodic events in the Chesapeake Bay region. In: Mehta, A.J. (Ed.), Nearshore and Estuarine Cohesive Sediment Transport. American Geophysical Union, Washington, DC, pp. 1–20. Uncles, R.J., Barton, M.L., Stephens, J.A., 1994. Seasonal variability of fine-sediment concentrations in the turbidity maximum region of the Tamar estuary. Estuar. Coast. Shelf Sci. 38, 19–39. Uncles, R.J., Barton, M.L., Stephens, J.A., 1996. Seasonal variability of mobile mud deposits in the Tamar estuary. In: Pattiaratchi, C. (Ed.), Mixing in Estuaries and Coastal Seas. American Geophysical Union, Washington, DC, pp. 374–387. Uncles, R.J., Easton, A.E., Griffiths, M.L., Harris, C., Howland, R.J.M., King, R.S., Morris, A.W., Plummer, D.H., 1998. Seasonality of the turbidity maximum in the Humber–Ouse estuary. Mar. Pollut. Bull. 37, 206–215. Vale, C., Sundby, B., 1987. Suspended sediment fluctuations in the Tagus estuary on semi-diurnal and fortnightly time scales. Estuar. Coast. Shelf Sci. 25, 495–508. Wellershaus, S., 1981. Turbidity maximum and mud shoaling in the Weser estuary. Arch. Hydrobiol. 92, 161–198.