Ocean Dynamics (2013) 63:1321–1340 DOI 10.1007/s10236-013-0659-4
Numerical modelling of circulation and dispersion processes in Boulogne-sur-Mer harbour (Eastern English Channel): sensitivity to physical forcing and harbour design Nicolas Jouanneau & Alexei Sentchev & Franck Dumas
Received: 30 November 2012 / Accepted: 6 October 2013 / Published online: 17 November 2013 # Springer-Verlag Berlin Heidelberg 2013
Abstract The MARS-3D model in conjunction with the particle tracking module Ichthyop is used to study circulation and tracer dynamics under a variety of forcing conditions in the eastern English Channel, and in the Boulogne-sur-Mer harbour (referred to hereafter as BLH). Results of hydrodynamic modelling are validated against the tidal gauge data, VHF radar surface velocities and ADCP measurements. Lagrangian tracking experiments are performed with passive particles to study tracer dispersal along the northern French coast, with special emphasis on the BLH. Simulations revealed an anticyclonic eddy generated in the harbour at rising tide. Tracers, released during flood tide at the Liane river mouth, move northward with powerful clockwise rotating current. After the high water, the current direction changes to westward, and tracers leave the harbour through the open boundary. During ebb tide, currents convergence along the western open boundary but no eddy is formed, surface currents inside the harbour are much weaker and the tracer excursion length is small. After the current reversal at low water, particles are advected shoreward resulting in a significant increase of the residence time of tracers released during ebb tide. The effect of wind on particle dispersion was
Responsible Editor: Martin Verlaan This article is part of the Topical Collection on the 16th biennial workshop of the Joint Numerical Sea Modelling Group (JONSMOD) in Brest, France 21-23 May 2012 N. Jouanneau (*) : A. Sentchev Laboratoire d’Océanologie et de Géosciences, UMR8187, Univ. du Littoral Côte d’Opale, Wimereux, France e-mail:
[email protected] N. Jouanneau e-mail:
[email protected] F. Dumas Dyneco/Physed, IFREMER, Plouzané, France
found to be particularly strong. Under strong SW wind, the residence time of particles released during flood tide increases from 1.5 to 6 days. For release during ebb tide, SW wind weakens the southward tidally induced drift and thus the residence time decreases. Similar effects are observed when the freshwater inflow to the harbour is increased from 2 to 10 m3/s during the ebb tide flow. For flood tide conditions, the effect of freshwater inflow is less significant. We also demonstrate an example of innovative coastal management targeted at the reduction of the residence time of the pathogenic material accidentally released in the harbour. Keywords Tidal currents . Numerical modelling . Residence time . Lagrangian tracking . Water quality
1 Introduction Predicting the transport, dispersal and major pathways of pollutants is of primary importance for coastal zones subject to complex forcing regime. Water flow in the eastern English Channel (EEC) is a result of the interaction between tides, river discharge, meteorological forcing and non-tidal sea-level changes. Transport patterns and dispersion processes in the English Channel have been studied by means of numerical modelling (Sentchev and Korotenko 2004, 2005; Bailly du Bois et al. 2012) and field investigations (Prandle et al. 1996; Lafitte et al. 2000). These studies demonstrated the effect of coastline, bottom bathymetry and freshwater discharge on tracer dispersal, and highlighted the spatial variability of the transport in the EEC. To our knowledge, similar investigations have never been performed in the EEC at subregional scale, i.e. at a scale of a harbour or estuary. Hydrodynamic modelling and water quality investigations of the Boulogne-sur-Mer harbour (hereafter BLH) have attracted the attention of local authorities since initiating the
1322
European Water Quality Plan in 2000. This was prompted in part by concern of water quality management conducting by regional authorities (Water Agency, VEOLIA-Environment), and in part by the desire for a better understanding of the local circulation and mixing processes in the EEC, along the northwest French coast. Knowledge of water motion, dispersal and mixing is of general interest in studies of fisheries, water quality, sediment transport, marine biology and chemistry. Understanding the key processes influencing the water quality is an important aspect of harbour design, and numerical modelling offers an efficient tool for assessment of such processes. Many modelling studies have been carried out to investigate engineering and environmental problems at a harbour scale. Sánchez-Arcilla et al. (2002) determined that a harbour's water renovation and capacity to flush are controlled by hydrodynamics. Flow pattern favourable for flushing can minimize water quality degradation (i.e. avoid problems related to anoxia, eutrophication, etc.). Therefore, hydrodynamics is closely related to the evolution of water quality problems. Hartnett and Nash (2004) correlated flow patterns with the water quality by modelling the chlorophyll-a production as a function of nutrient fluxes. Other studies have been performed to assess human impact on the marine environment and to predict oil dispersion (Comerma et al. 2002) and eutrophication (Lee et al. 1999) in enclosed water bodies. Zhow and Li (2005) have shown that the water circulation is also essential for estimating sediment transport and morphodynamic changes. Qin et al. (2006) used a twodimensional modelling to assess erosion and deposition zones of a harbour bassin. Thus, understanding the hydrodynamic regime is important for harbour development and determining planning measures and policies for a sustainable environmental protection (Montaño-Ley et al. 2007). In the present study, we focus on the processes governing local circulation and dispersal in the BLH (Fig. 1). Boulogne harbour is the largest fishing port in France and one of the largest centres of seafood industry in Europe. A unique combination of historical heritage, tourism, aquatic and fishing industry makes this region of considerable commercial and ecological interest, and requires new concepts of ecologically sustainable development. Nevertheless, experimental or modelling studies of the marine environment in this part of the EEC are rare. With the exception of earlier modelling studies of tidal and residual circulation in the EEC performed by Orbi and Salomon (1988), Bailly du Bois and Dumas (2005), and Sentchev and Korotenko (2004, 2005), up to now, there were no relevant results in literature on the water dynamics along the north-eastern French coast. Small-scale features of circulation, their magnitude and dependence on various forcing factors remain relatively uncertain. To fill this gap, we performed comprehensive numerical investigation of the role that various physical forcing play in dispersal or retention of materials in the region, using a random walk
Ocean Dynamics (2013) 63:1321–1340
Lagrangian particle-tracking model, in conjunction with the three-dimensional hydrodynamic model MARS. We also analysed a quantity T90 defined as the time it takes for 90 % of tracers to leave the harbour forever through the open boundary. This definition is very close to the definition of residence time proposed by Dronkers and Zimmerman (1982) or Monsen et al. (2002). The study contributes to our understanding of marine system behaviour in a macro-tidal harbour. The paper is organized as follows. In the next two sections we briefly describe the domain and the methods used for investigation of dispersion processes and transport patterns in the EEC and at smaller scale, in the BLH. In Section 4, modeling results are presented and particle-tracking experiments are analyzed. Field experiments, performed to validate the results of numerical simulations, are also discussed. Conclusions complete the paper.
2 Domain and environmental conditions 2.1 Study area The study area is located along the Opal coast of France and at the entry to the Strait of Dover (Fig. 1, right panel). The coastline is meridionally oriented with a large embayment in the central part (Boulogne harbour) and a number of inlets and small river estuaries. The water depth is less than 65 m throughout the domain. In the middle, there are sandbanks oriented in the alongshore direction. Above the main sand bank (Bassure de Baas), the water depth does not exceed 2 m at spring low tide. 2.2 Environmental conditions Tidal waves, arriving into the EEC from the western Channel and from the North Sea, generate currents characterized by clockwise rotation in response to the combined effect of the Coriolis force and pressure gradients related to tidal wave propagation. However, in the shallow nearshore region, along the Opal coast, the tidal ellipse polarization changes to anticlockwise under the effect of bottom friction, producing a convergence–divergence zone several kilometres offshore (Sentchev and Yaremchuk, 2007). The interaction of tidal waves is the dominant factor that determines variability of the sea surface height (SSH) and currents in the region. The SSH in the BLH shows strong (4 to 9 m) variations (Fig. 2), with the predominant semi-diurnal period, small diurnal inequality and pronounced fortnightly modulation due to the interference of the major semi-diurnal (M2, S2, N2) constituents. Tidal velocities and transports also show fortnightly variability in response to the spring–neap cycle.
Ocean Dynamics (2013) 63:1321–1340
Fig. 1 Left panel, domain of the local model (eastern English Channel) and the bathymetry (colour shading). Tidal gauge locations for 20 ports used for model validation are shown by black circles. The name of all ports is given. Also shown are geographic location and name of the rivers contributing to the freshwater buoyancy input to the domain. The VHF
1323
radar measurement zone is represented by hatched area. Right panel, domain of higher resolution modelling (off the Opal coast) with the bottom topography (colour shading) and the location of Boulogne harbour (BLH). ADCP location is shown by black circle. The Liane River provides fresh water input to the BLH
Fig. 2 Sea surface height from tidal gauge record in Boulogne (red) and modelling (blue) for the 8-day period in 2003. The difference in sea level (green) yields the relative error of 5.4 % for the period shown
1324
Strong tidal forcing induces vertical mixing throughout the water column in the majority of the domain. In the nearshore zone, river runoff has a significant impact on the local circulation. The majority of the freshwater comes to the EEC from the Seine, Somme, Canche, Authie and other rivers on the north-eastern coast of France and occurs in winter. The Seine River provides 80 % of the freshwater input to the EEC (Fig. 1). Annual cycle of the river runoff demonstrates strong seasonal variability with the secondary peaks occurring in late winter–early spring. As the consequence, a profound haline front, separating offshore saline waters of Atlantic origin from freshened near-shore waters, is a typical feature of water dynamics along the Opal coast. In the BLH, the Liane River discharges at the mean annual rate of 2 m3/s. This transport is regulated by a number of tidal gates. When the gates are opened, the peak discharge may reach 10 m3/s, which is quite significant given the small water volume in the harbour. Statistics of year-long wind observations (not shown) at the Boulogne light house in 2009 shows two dominant wind regimes: south-western winds, more frequently observed in autumn and winter, and north-eastern winds predominantly blowing in spring. Both wind regimes will be used in numerical simulations of circulation and dispersal in the BLH.
3 Model set-up 3.1 Regional model configuration We used the sigma-coordinate hydrodynamic model MARS3D (Lazure and Dumas 2007) to simulate tidal circulation in the domain shown in Fig. 1 (left panel). The model was also forced by realistic freshwater runoff and winds. MARS solves finite-difference analogues of the primitive equations in the three spatial dimensions with fully prognostic temperature and salinity fields and the free surface. The entire region of the EEC, including the Strait of Dover, is represented on a horizontal grid with homogeneous spacing of 1 km. This model configuration will be referred hereafter as the “regional model”. The bottom topography shows complicated geometry with numerous shallow banks separated by a series of 50-m deep basins, both oriented in the along-shore direction (Fig. 1). The “Arakawa C” differencing scheme for the momentum equations is used in the horizontal. In the vertical, there are 20 sigma levels, distributed such as to provide enhanced resolution in proximity to the surface and seabed. The horizontal diffusion coefficient, K H, is calculated using Smagorinsky formula with the Smagorinsky constant C s = 0.15. The vertical diffusivity, K V, is obtained from the level 2.5 turbulence model of Mellor and Yamada (1982). A bottom-friction approximation is used with a variable (in time and space) drag coefficient, C d , set to be inversely
Ocean Dynamics (2013) 63:1321–1340
proportional to the local depth and the bed roughness (Lazure and Dumas 2007). A time-difference scheme employed in the tracer conservation equation is diffusion free and its performance was evaluated by Kerr and Blumberg (1979)). An innovative mode splitting technique (Lazure and Dumas 2007) adopted for computational efficiency allows to use the time step of 80 s for both external and internal modes. 3.2 Boundary conditions, initial conditions and model runs Tidal forcing was specified by prescribing sea surface elevation at the open boundaries using 60 tidal constituents extracted from a tidal database of the French Navy (Service Hydrographique et Océanographique de la Marine, SHOM). The database covers the entire English Channel and the shallow water part of the Bay of Biscay (Leroy and Simon 2003). Numerical experiments have shown that this type of BC provided better accuracy than the BC derived from the core (5 km) resolution, large-scale circulation model. We also used the mean sea level, heat and salt fluxes prescribed at the open boundaries from the core model. Wind stress, freshwater and heat fluxes at the sea surface extracted from the French Meteorological Office archive (3 h resolution) were linearly interpolated in time and in space onto model grid points. Realistic daily freshwater discharge of the rivers on the French and English coasts was also included. The model spin-up runs were conducted starting from homogeneous temperature (12° and 15°, mean values according to measurements in the region), salinity (34.5) and zero velocity fields. The model was run for two 35-day-long periods, in May 2003 and June 2009. For each simulation, after the initial ramp up over one tidal cycle, the model was run over nine additional spin-up cycles. The next 30 days of the model run were used for analysis of tidal and wind-driven circulation. For the analysis of the coupled effect of tides and river discharge, the spin-up period was prolonged to 30 days. 3.3 Local circulation model and studied cases The regional model was used to force a higher resolution “local” model (Fig. 1, right panel). The local computational domain (Fig. 1) extends from 12 km south of the BLH up to Cape Gris Nez in the north and 14 km offshore to the west. The domain is covered by geographically oriented rectangular grid with uniform horizontal resolution of 140 m and 20 unevenly spaced sigma levels. The local model employs a time-varying grid technique in the vicinity of the coastline thus taking into account the drying beach phenomenon (Plus et al. 2009). The model was run with the time step of 25 s, for both external and internal modes. A complete set of boundary conditions was provided by the regional model runs: SSH, currents, fluxes on three model open boundaries, wind stress,
Ocean Dynamics (2013) 63:1321–1340
heat fluxes on the sea surface and fresh water input by the Liane River. All boundary conditions were linearly interpolated in time and space. The duration of the spin-up period was set equal to that of the regional model (i.e. 5 or 30 days). Identical initial conditions were used for both models. Numerical experiments, each 30 days long after the spinup, were designed to investigate the circulation and the effect of various forcings (tide, wind, river runoff) on displacement, dispersal and residence time of passive tracers in the BLH. We also focused on assessment of the harbour design and its impact on circulation and tracer dispersal. Four sets of simulations were performed, using different boundary conditions. For circulation studies, we used tidal forcing covering three characteristic periods of the tidal cycle (spring, neap and mean tide). In the second case, we included freshwater discharge of the Liane River. Two values of river discharge were considered: 2 m3/s (mean summer discharge) and 10 m3/s (high discharge when tidal gates are open). The third case included a combined effect of tidal and wind forcing: the model was forced by constant wind corresponding to two dominant wind regimes (south-western and northeastern winds of 6 m/s). We have also investigated the effect of stronger winds of 12 m/s. Finally, the effect of seawall modification/removal on local hydrodynamics and the residence time of tracers was assessed. Additional simulations, covering shorter periods in March and June 2012, were also performed in order to validate the local circulation model and to investigate some specific features of circulation in the BLH. 3.4 Lagrangian tracking approach To study dispersion processes in the BLH, we used a Lagrangian tool, Ichthyop, described by Lett et al. (2008). In all simulations, a thousand particles were released at the Liane river mouth as the river was assumed to be the major source of potentially polluting material. At the release, the spatial distribution of neutrally buoyant particles had a circular shape of 300 m in diameter, occupying approximately 2×2 model grid cells (see Fig. 11 for patch location). Particles were evenly spread in the surface layer 1 m thick. Using current velocities generated by the MARS-3D model with 30 min time step, particles are firstly advected by currents in three-dimensional space. Next, dispersion is modelled as a random walk at 10 s time step with position increments proportional to horizontal and vertical eddy diffusivity. The coordinates of particles were we recorded every 6 min, thus providing information on the tracer concentration in threedimensional space. Recall that projection of these concentrations onto a horizontal plane, in some cases, might produce a (visual) effect of accumulation of tracers. However, three-dimensional concentrations never increase.
1325
A number of simulations were performed with particles released at different stages of the tidal cycle: release during the ebb tide 1 h before low water (LW−1 h) and during the flood tide 1 h before high water (HW−1 h). These simulations were done separately for spring, mean and neap tide conditions. In order to assess the residence time of particles inside the BLH, we have defined a domain of interest (rectangular area in Fig. 7c) and recorded the number of particles remaining inside the domain at every time step (6 min). We have also estimated two parameters, “T90” and “T50”, which show how much time 90 % (50 %) of the total number of particles released at a specified location within a waterbody will remain in the waterbody before leaving it forever.
4 Model validation 4.1 Regional model To assess the performance of the regional model, sea surface elevation, surface current velocities and horizontal velocity profiles predicted by the model were compared with observations for two particular periods of time: May 2003 and June 2009. Figure 2 shows the modelled and observed tidal elevation in BLH. The simulated SSH agrees well with observations, replicates the spring to neap tidal variation and the very small diurnal inequality of the tide. We used the normalized root mean square to calculate the relative error (ε) of SSH simulation in BLH. vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi uX u ðxobs −xmodel Þ2 t X ε ð xÞ ¼ ð1Þ ðxobs Þ2 This error was found to be close to 5 %. Estimation of ε for other ports used in model validation (9 ports on the French coast and 11 ports on the UK coast, see Fig. 1 for location) provided similar numbers (Table 1). A comparison of the model-derived and experimental (derived from tidal gauge data) amplitudes for principal tidal constituents in Boulogne is given in Fig. 3. The agreement is satisfactory for all semi-diurnal and quarter-diurnal tidal waves which in total account for 75 % of SSH variability. Higher discrepancy is found for diurnal constituents with a general tendency of the model to overestimate the observed amplitudes by about 50 %. However, the amplitude of diurnal constituents varies from 1 to 5 cm and their contribution to SSH variability does not exceed 2 %. Table 2 summarizes the results of harmonic analysis for all the ports used for model validation. We focus on the major semi-diurnal (M2, S2) and quarter-diurnal (M4, MS4) constituents which comprise 75 % of the tidal variations. The data values for comparison were extracted from the
1326 Table 1 Relative error of simulated sea surface height compared to tidal gauge data at 20 ports located on French and English coast
Ocean Dynamics (2013) 63:1321–1340
Ports
ε (%)
Calais Wissant Boulogne Le Touquet Berck
2.3 3.5 5.4 6.2 4.6
Cayeux Le Treport Dieppe Le Havre Average FR ports Selsey Bognor Little Hampton Shore Hampton Brighton New Heaven Eastbourne Hastings Rye Harbor Folkstone Dover average GB ports
6.1 5.6 2.4 6.6 5.0 5.8 6.5 2.5 3 2.4 6.5 7.8 9.6 7.4 6.3 5.2 5.7
International Hydrographic data Bank. The results show that the contribution of the principal M2 constituent is reproduced
with high accuracy, with the mean error for French and UK ports less than 4 and 11 %, respectively. The phase error is negligible (
