numerical studies OF THE Matter transfer in the Thermaikos GULF

Y. Krestenitis¹, V. Kourafalou² and Y. Savvidis³

^{1, 3} Division of Hydraulics & Environmental Engineering, Aristotle University, GR 54006 Thessaloniki, Greece. tel/fax +30.31.995654, E-mail: ynkrest@civil.auth.gr

² National Centre of Marine Research, GR 10000 Athens, Greece

Abstract: This work aims toward the numerical modelling of the processes that govern the transport of matter in Thermaikos Gulf (NW Aegean Sea, Eastern Mediterranean) across the interface between the low-salinity coastal waters of river origin and the higher salinity basin waters. Of particular importance is the overall basin general circulation, which is influenced by coastal dynamics and the interaction with larger scale Aegean flows; the interaction between land and sea, which takes the form of studies in plume dynamics; the air-sea interaction; and the sedimentation processes that determine the rates and pathways of sediment particles, as well as the location of their final deposition. A comprehensive hydrodynamic model was developed for Thermaikos, which is further used to perform realistic simulations with high frequency forcing. The hydrodynamic model was extended to include sedimentation processes and the physical and sediment transport processes that influence matter transfer on the gulf were studied. The important physical processes, namely the wind stress, buoyancy due to river input, topographic effects and interaction with deeper Aegean Sea flows were elucidated in time scales from a few days to seasonal. The sediment transport processes included advection and diffusion, deposition and erosion, flocculation and settling. The effects of the different processes on the transfer of river-borne low salinity waters and fine grain particles, were examined separately and combined, so that their significance was made clear.

1    INTRODUCTION

Thermaikos Gulf is part of the Northwestern Aegean Sea (Figure 1). Three major rivers discharge in Thermaikos Gulf, Axios, Loudias and Aliakmon. The seasonal and annual mean discharges of the three rivers are shown in table 1. They also supply the coastal basin with a large amount of fine-grained sediments. According to the data of the water discharge and the suspended sediment load collected during 1997-1998, the total mean monthly water discharge is about 158.26 m³/s, while the total mean monthly sediment discharge is about 3756 g/s [8].

    Table 1    Thermaikos Bay River discharge (m³/s) from historical data

Fig. 1    The Thermaikos Gulf

2    Model description

The hydrodynamic numerical simulations are carried out with the three-dimensional, free surface, sigma-coordinate Princeton Ocean Model [1]. The river input is parameterized as a source term in the continuity equation [2], so that the freshwater discharge directly affects the free surface elevation and the resulting vertical velocity near the river mouth. The only parameter that is specified is the amount of river runoff with zero salinity. The horizontal mixing depends on the grid size and the velocity field, as the horizontal eddy diffusivity parameter is given by the Smagorinsky formula. The vertical eddy diffusivity parameter is calculated according to the Mellor and Yamada turbulence closure scheme [6]. The model can successfully represent the dynamics of a freshwater plume [2,3,5].

The transport model is based on the tracer method (Lagrange-Monte Carlo Method or Random Walk Simulation) [4]. According to this method, a large number of particles representing a particular amount of mass are introduced into the flow domain through a source or sources [7]. Their transport and fate is traced with time. The horizontal advection of the particulate matter is controlled by the local fluid velocity, while the vertical advection is controlled by the local fluid velocity and the particle settling velocity. The turbulent diffusion is simulated by the random Brownian motion of the particles due to the turbulence. All the processes affecting the motion and transport of cohesive sediment particles are taken into account [7]: (a) The advection process is determined from the currents. (b) The diffusion process is determined using the diffusion coefficients. (c) The flocculation process due to turbulence only is described from the evolution of the mean floc size (mean diameter) with time and the floc density. (d) The settling process is described by the settling velocity of each particle, derived from the Stokes expression. (e) The deposition and erosion processes are modeled, based on the computation of the bed shear stresses and the critical values of shear stresses for deposition and erosion.

3    The general circulation MODELLING

The seasonal variability of the general circulation in Thermaikos Gulf was examined, as a basis for establishing “mean” hydrographic conditions that could help evaluate the seasonal observational surveys. The limited area domain was employed, but the influence of the larger scale Aegean circulation was included, through one way nesting with a coarser grid model of the entire Aegean Sea [9]. The hourly wind stresses, atmospheric pressure, and relative humidity were derived from the Macedonia Airport Meteorological Center (MAMC) for a full one-year period. The heat flux at the sea surface was calculated interactively from the atmospheric data set. The salinity boundary condition at the surface was calculated through the net evaporation-precipitation fresh water surface mass flux. The evaporation rate is obtained from the latent heat flux and the precipitation from the data set of MAMC. The model was run for a full annual cycle, so that the seasonal variability of the general circulation and of the density structure could be depicted. The model computed temperature, salinity and velocity fields were seasonally averaged; an annual average was also performed [5]. The surface distributions of salinity and velocity fields are shown in figure 2.

Fig. 2    Surface circulation and salinity

The surface flow for the continental margin of Thermaikos Gulf is characterized by southward currents, due to the prevailing northerlies. Near the river area, waters are lower in both temperature and salinity. An intrusion of low salinity waters from the open boundary causes an inflow along part of the eastern boundary. At lower depths there is a northward return flow along the western boundary, while the deeper sea influence causes flow that only affects the southern part of the domain [5]. The open sea signal is still fresher, but also cooler than the interior. The coldest waters are found near the river area.

4    MODELLING THE Buoyancy and wind driven flows

As for most coastal areas, wind stress is a major circulation forcing mechanism. Furthermore, in Thermaikos Gulf river inputs lead to a significant buoyancy forcing. We attempt numerical modeling of these two dominant processes, also taking into account the effect of topography, which is also important in such a semi-enclosed basin. The model initial condition is that of a homogeneous basin at rest. At first, we perform process studies on pure wind-driven (case 1) and buoyancy-driven (case 2) flows i.e., only one of the two forcings is considered for each experiment. Then, we consider the combined effect of buoyancy and wind stress (case 3). Here, we outline the important findings.

Case 1. A forcing of constant magnitude wind at 5 m/s was employed. The wind-driven circulation was dominated by the coastal set-up or set-down, according to the predominant wind direction and its orientation relatively to the west, north and east coast. During southerlies, the north part of the basin has a rise in sea level, while a sea level drop occurs at south. During easterlies, sea elevation contours are parallel to the overall along (north-south) basin, with values dropping from west to east. The wind-driven flow is influenced by the elevation gradients and farther modified by topography. Transport is thus downwind within the nearshore shallow regions and to the right of the wind farther offshore, where Ekman layers develop. Due to the geometry of the basin, an alongshore coastal current develops which brings water toward the inner Gulf during southerlies and westerlies, but flushes inner Gulf waters out during northerlies and easterlies.

Case 2. The buoyancy-driven circulation is dominated by the freshwater input of the rivers. A rate of discharge of 200 m³/s was employed and the river plume was developed within a few days (figure 3). The core of the plume reaches the northern part of the outer Gulf extending from the west coast to the east. A strong anticyclonic circulation is found between the two coasts, within the bulge of the river plume. This type of circulation is typical, due to the offshore movement of low-salinity waters in the vicinity of the river mouth and their subsequent anticyclonic turning due to the Coriolis force. Finally, a coastal current region is found within the river plume and along the west coast. This is due to the piling of low-salinity water at this coast at the bottom of the bulge, carried by the anticyclone. This creates a cross-shore pressure gradient that, balanced by Coriolis in a geostrophic manner, results in a coastal southward current. The preferred pathway of river waters is first eastward, then southward along the east coast, then westward through the interior to the west coast, where strong southward flow develops.

Fig. 3    Model computed near surface (a) salinity and velocity fields at day 20; (b) salinity field at day 40, black is salinity S<34.5 psu, gray is 34.5_S<36.5, outer line is S=37.5

Case 3. Starting with initial condition the density and velocity fields from the previous run, additional forcing of wind at 5 m/s is applied for three wind directions: northerlies, easterlies and southerlies. When winds are downwelling favorable for the west coast, the low-salinity waters are flushed out from the inner Gulf. Conversely, large amounts of low-salinity water enter the inner Gulf during upwelling favorable winds, as the buoyancy driven circulation is reversed. In the lower layers, the signal of the plume is strongest in the inner Gulf during southerlies and along the west coast during northerlies and easterlies. These are the respective areas of strongest vertical velocities and, consequently, the most likely areas for deposition of river associated materials. The velocity fields exhibit the strongest southward coastal current along the west shelf during northerlies; this current has a wind-driven (barotropic) and buoyancy-driven (baroclinic) component. The anticyclonic circulation within the bulge is enhanced. Results are similar in the case of easterlies. During southerlies, flow toward the inner Gulf is dramatically increased, while the southward coastal current is eliminated; eastward transport (i.e. to the right of the wind) in the Gulf interior causes the southward displacement of the plume along the east coast.

5    Sediment transport simulations

The sediment transport model was applied on Thermaikos, using the hydrodynamic data were supplied from the general circulation simulation. The smallest value of mass (»200 g/s) from the sediment load time series, is taken to correspond to mass of a 《parcel》 discharged every six hours from the source river mouth. Each parcel represents a mass of 4320 kg and 28347 parcels used to represent the mass of the sediments discharged to Thermaikos from the rivers, during a year, i.e.122.5´10⁶ kg. The sediment particulate matter recorded on the estuaries of Thermaikos, before the river mouths, consist of very fine silt and clay (diameter of very fine silt » 4μm and of clay < 2μm) [8]. In the simulation the mean size (particle diameter) of the particles entering the marine environment with a mean porosity 0.50 is approximately 6μm (mean density = 1835kg/m³). The respected mean initial size of the primary particles is consequently 3μm with density 2650 kg/m³.

Fig. 4    Thickness of the layer of deposited matter (in mm) after one-year period

The horizontal distribution of the sedimentation - deposition over the Thermaikos seafloor is depicted on figure 4. The largest amounts of the deposited material are restricted to the areas in the vicinity of the delta fronts (river mouths) of the rivers discharging to the west coast of the gulf. The sedimentation rate on these areas is computed to be approximately 0.5 mm/year.

6    Discussion - RESULTS

The physical and sediment transport processes that influence matter transfer on Thermaikos Gulf were studied with the aid of numerical experiments. The effects of the different processes on the transfer of river-borne low salinity waters and fine grain particles were examined separately and combined. This helped develop step by step a comprehensive, three-dimensional hydrodynamic and sediment transport model for Thermaikos that is suitable both for process oriented studies and for realistic simulations with high frequency forcing.

A characteristic plume parameter λ, that relates to the ratio of stratification induced by buoyancy versus the available mixing, may be used to classify river plumes. The parameter λ is defined [4,5] as the ratio of the bulge directly offshore the river mouth over the width of the buoyancy-driven coastal current. For the Thermaikos rivers, λ was found to be 6. The ratio λ effectively relates to a suitably calculated Richardson number that gives the measure of the tendency for stratification due to freshwater discharge over the tendency for homogeneity due to mixing [5]. Therefore λ>1 characterizes plume as supercritical, as the buoyancy driven stratification prevails over the mixing, which, in the absence of wind, is induced by turbulence and bottom friction. The implication of this finding is that conditions are favorable for seaward expansion of the plume near the river delta, as was found during the performed experiments.

When wind forcing is included, the transport of the low-salinity waters that are due to river input depends on the wind components that influence the coastline in the neighborhood of the river. If an upwelling-favorable component prevails, offshore transport is permitted and the southward coastal current is reduced. If a downwelling-favorable component prevails, offshore transport is restricted and the southward coastal current is enhanced. Topographic effects are important as they guide the development of a strong anticyclone in the basin interior, due to the small distance between west and east coasts in the vicinity of the river.

The most important conclusions of the sediment transport model application are: (a) The simulation showed that only a small portion (about 10%) of the total amount of the discharged suspended matter escapes from the southern boundary of the gulf. Transport of the suspended particulate matter takes place mainly along the western coast of the gulf. A small proportion of the particulate matter moves towards the eastern area and an even smaller one seems to reach the northeastern part of the Gulf. (b) Most of the suspended particulate matter (about 80%) is deposited on the seabed of the gulf and mainly near the delta and prodelta regions. This results lead to the conclusion that siltation near the river mouths is not only due to the coarse, non cohesive sediments, but to the contribution of the fine cohesive particulate matter as well. (c) The diffusion process is very important for the distribution pattern of the particulate matter, especially for the very fine particles. The values of the horizontal sediment diffusion coefficient range between approximately 50 and 200m²/s while the values of the vertical sediment diffusion coefficient range between 10^–4 and 10^–6 m²/s. (d) As far as the flocculation process is concerned, we conclude that changes on the floc sizes are observable only for a very small amount of the sediment particles. (e) Erosion due to bed currents does not seem to be taking place and, concentrations of suspended particulate matter are very low in recent years (due to the large reduction of the river discharges, caused by the construction of the hydroelectric and irrigation plants). The sedimentation rates have also been very. The simulation shows that the rate of sedimentation near the river mouths is of the order of 0.50 mm/year.

References

[1]     Blumberg, A.F. & G.L. Mellor, 1983. “Diagnostic and prognostic numerical circulation studies of the South Atlantic Bight”, J. Geophys. Res., 88(C8), 4579-4592.

[2]     Kourafalou, V.H., L.-Y. Oey, J.D. Wang & T. N. Lee, 1996. “The fate of river discharge on the continental shelf. Part I: modeling the river plume and the inner-shelf coastal current”, J. Geophys. Res., 101(C2), 3415-3434.

[3]     Kourafalou, V.H., L.-Y. Oey, T. N. Lee & J. D. Wang, 1996. “The fate of river discharge on the continental shelf. Part II:  transport of coastal low-salinity waters under realistic wind and tidal mixing”, J. Geophys. Res., 101(C2), 3435-3455.

[4]     Koutitas C., 1988. “Mathematicall models in Coastal Enginnering”, Pentech Press Limited, London (UK).

[5]     Krestenitis, Y., K. Barbopoulos & V. Kourafalou, 1998. “Modelling the seasonal variability of the processes that affect matter transfer on Thermaikos Gulf. Part 2: land-sea interaction”, In: Dynamics of Matter Transfer and Biogeochemical Cycles: their modelling in coastal systems of the Mediterranean Sea. EU-MAST-III METRO-MED, 1^st Scientific Report, pp. 209-215.

[6]     Mellor, G.L. & T. Yamada, 1982. “Development of a turbulence closure model for geophysical fluid problems”, Rev.Geophys.SpacePhys.,20(4),851-875.

[7]     Savvidis Y., 2000. “Mathematical Simulation of the Transport of Cohesive Sediments in Coastal Basins. Application to Thermaikos Gulf ”, Ph. D. Thesis, Aristotle University of Thessaloniki., 323 pp.

[8]     Skoulikidis, N. & E. Karamanou, 1997. “The role of the adjacent land area as fresh water, sediment and pollutants supplier”. In: Dynamics of Matter Transfer and Biogeochemical Cycles: their modelling in coastal systems of the Mediterranean Sea. EU-MAST-III METRO-MED, 1^st Scientific Report, pp. 64-75.

[9]     Valioulis, I. & Y. Krestenitis, 1994. “Modelling the Water Mass Circulation in the Aegean Sea. Part I. Wind Stresses, Thermal and Haline Fluxes”, Ann. Geophys., 12, pp 794-807.