Flushing efficiency of Augusta Harbour (Italy)

Journal of Coastal Research

SI 56

841 - 845

ICS2009 (Proceedings)

Portugal

ISSN 0749-0258

Flushing efficiency of Augusta Harbour (Italy) I. Lisi†‡, A. Taramelli∞, M. Di Risio‡, S. Cappucci§ and M. Gabellini† † Italian High Institute for Environmental Protection and Research , Rome, 00166, Italy [email protected]

‡ DISAT-LIAM, University of L’Aquila, L’Aquila, 67040, Italy [email protected]

∞ Lamont Doherty Earth Observatory of Columbia University, New York, NY 10964, USA [email protected]

§ Italian Agency for New technologies, Energy and Environment, 00123 Rome, Italy [email protected]

ABSTRACT LISI, I.; TARAMELLI, A.; DI RISIO, M.; CAPPUCCI, S. and GABELLINI, M., 2009. Flushing efficiency of Augusta Harbour (Italy). Journal of Coastal Research, SI 56 (Proceedings of the 10th International Coastal Symposium), 841 – 845. Lisbon, Portugal, ISSN 0749-025. Water quality of semi-enclosed basins crucially depends on flushing efficiency, mainly affected by tidal and wind driven circulation. This paper presents numerical estimation of flushing features of Augusta Harbour (Italy) evaluated by using of 2DH hydrodynamic models. The application presented herein has been carried out in order to investigate the effectiveness of hydrodynamic action in removing any substances from the system. Hydrodynamic finite element model was calibrated in terms of water level and current velocities by comparing numerical results with measurements collected by means of two Acoustic Doppler Current Profilers (ADCPs) placed close to the harbour entrances. On the basis of numerical results, standard hydrodynamic parameters (Tidal Prism Ratio, Flushing time and Exchange coefficient) were estimated in order to evaluate flushing efficiency of the harbour. Finally, numerical estimated Tidal Prism Ratio was compared with the one computed on the basis of field data and good agreement was achieved. ADITIONAL INDEX WORDS: hydrodynamic model, lagrangian model, flushing time, water quality

INTRODUCTION Water quality of semi-enclosed basins crucially depends on flushing efficiency. It is strongly affected by hydrodynamic circulation driven by tidal and wind action. Understanding of water circulation makes it possible to investigate the effectiveness of hydrodynamic action in removing any substance introduced into a water body (FISCHER et al., 1979). Flushing efficiency is thus considered of prime importance in the evaluation of control and mitigation measures and it is frequently used for environmental management purposes (CHOI and LEE, 2004; JOUON et al., 2006). In particular, results of this kind of studies can represent a crucial aspect that has to be considered within the planning strategy. It can be useful to refer to standard hydrodynamic parameters (i.e. Tidal Prism Ratio, Exchange Coefficient, Flushing Time) to carry out a flushing analysis in order to evaluate the water quality within semi-enclosed basins (LUKETINA, 1998; CHOI and LEE, 2004; BILGILI et al., 2005; JOUON et al., 2006). Moreover, lagrangian dispersion model is now considered a very useful tool for a wide variety of application (HUNTER et al., 1993; VISSER, 1997; BILGILI et al., 2005; JOUON et al., 2006; EDWARDS et al., 2006), i.e. to evaluate dispersion of substances (prediction of dredging plume fate, sediment transport, oil spill and pollutants dispersion) and to analyze biological and ecological processes (larval dispersion, nutrient level). This paper describes key features of tidal and wind induced circulation within the Augusta Harbour (Italy) in order to estimate standard hydrodynamic parameters (Tidal Prism Ratio, Exchange Coefficient and Flushing Time) aimed at evaluating flushing efficiency of the basin. In recent years water quality in Augusta Harbour has been deteriorating, due to the growing human activities that take place into the basin (industry, military, urban expansion, etc.). Previous in this context, studies and monitoring

campaigns have been undertaken to assess the human impact (CNR, 1995; SIAP, 2007). The main objective was to investigate the variability of the physical, chemical, biological, sedimentological and circulation patterns. However, increased knowledge of physical process operating throughout the Harbour is needed to obtain a better understanding of the interaction between circulation patterns and their contribution to in situ chemical, biological and sedimentological variability (both spatial and temporal). In this study a 2DH hydrodynamic numerical model and a lagrangian dispersion model were used. Field data were utilized in order to define tidal harmonic amplitudes and to calibrate and validate the hydrodynamic model in terms of water level and current velocities. The paper is organized as follows. Morphological and physical description of the study site is detailed in the next section. Then the methodology is outlined by briefly detailing numerical models used. The succeeding section provides a description of numerical results and a detailed analysis of hydrodynamic parameters.

THE STUDY SITE Augusta Harbour is located in the Augusta Gulf, in the Eastern coast of Sicily Island (Italy, Figure 1). The sheltered area (about 24 km2) extends to about 7.0 km alongshore and 3.5 km crossshore and its average depth is 15.0 m. The Harbour is originated, early 1950, from the closure of most of the natural Augusta Gulf through the construction of three rubble mound breakwaters and it is connected to the Ionian sea by two deep (maximum water depth 40 m) and narrow (300-450 m) entrances, located in the Eastern and Southern part of the Harbour (Figure 1). Furthermore, it communicates with the open sea through the Ponte Rivellino

Journal of Coastal Research, Special Issue 56, 2009 841

Flushing efficiency of Augusta Harbour (Italy)

channel, with extremely limited water depth that does not exceed 1.0 m. Cantera and Marcellino Rivers drain into the harbour with seasonal and discontinuous freshwater inflow. Wind dataset are based on measurements collected by the Catania Station, belonging to the Italian Mareographic Network

Figure 1. Location and bathymetry of Augusta. (ISPRA, 2008), 35 km north from Augusta Harbour. Wind data analysis (temporal interval 1998/10/01-2008/09/30) shows that prevailing surface wind events come from NE (mean annual occurrence of 23%) and from SW (mean annual occurrence of 33%). Maximum wind speed does not exceed 16 m/s (about 30 knots). Wind events from NW and SE are almost absent, except in summer period, when an increasing of winds coming from SE due to baroclinic effects is observed. All the remaining wind directions are almost absent due to the presence of the Etna Volcano, located just North-West to the Catania station. Tidal forcing has been defined on the basis of three datasets collected by the aforementioned Catania Station and by two Acoustic Doppler Current Profilers (ADCPs) placed close to the Easthern and to the Southern harbour entrances (stations 1 and 2, Figure 1). The two ADCPs collected both water levels and velocity profiles during the time interval 2006/10/05-2006/12/06 (SIAP, 2007). Standard harmonic analysis (PAWLOWICZ et al., 2002) shows that semidiurnal components M2 and S2, with amplitude of 0.07 m and 0.04 m respectively, are the more significant. Mean astronomic tide excursion has been estimated to ±0.15 m. Offshore wave climate has been evaluated using measured wave parameters collected by the directional wave buoy placed offshore Catania, belonging to the Italian Buoys Network (ISPRA, 2008). Waves come from NNE and SSE. However, preliminary numerical simulations stressed that short waves do not induces significant circulation within the harbour, as the three breakwaters constitute a screen to waves penetration and any breaking phenomena does not occurs.

METODOLOGY A standard finite element depth-integrated 2DH model solving non linear shallow water equations is used to compute horizontal mean velocity and water level (SVENDSEN, 2006):  (1.1)  h   v   0 t Ts Tb v  v   v  f  v   g     D (1.2)  h     h    t where: t = elapsed time;  = water level; v = mean velocity vector at point (x,y); h = local water depth;  = horizontal gradient operator (  / x ,  / y ); f·v= Coriolis acceleration vector; Ts, Tb = surface and bottom stresses tensors; and D = momentum diffusion terms, that take into account horizontal eddy viscosity. The model manages water level boundary conditions and depth integrated wind stress (Ts) and then it reproduces water circulation induced by tidal and wind forcing. The model provides only the mean velocity along the water column, however it has been demonstrated (CHOI and LEE, 2004) that 2DH numerical models provide higher values of flushing times if compared to 3D stratified model results. Computational domain (about 44,000 elements) covers the whole harbour area along with an offshore area needed to avoid boundaries effects on numerical results within the area of interest. Bathymetry data were obtained from surveys carried out by means of multibeam and side scan sonar technique (SIAP, 2007) and Nautical chart (IIM, 1994). It has to be noted that the two harbour entrances and the Ponte Rivellino channel were numerically reproduced. The model was calibrated in terms of water level and flow velocities by comparing numerical results with field data collected by the two ADCPs (Figure 1). This leaved to define the values of bottom friction and drag coefficient needed to compute the wind induced circulation. Current intensities and directions results within the computational domain have been used to evaluate particles dispersion into the harbour. A lagrangian dispersion model was used (HUNTER et al., 1993; VISSER, 1997; BILGILI et al., 2005; JOUON et al., 2006). Passive particles have been released at specific locations into the sheltered area and material derivative were evaluated in order to compute their trajectories:   6k h r (t  dt ) r (t )  dt  v  dh  (2)   dt   that expresses the vector position r(t+dt) of a particle as a function of the velocity field (v). The model considers a random walk model in order to describe turbulent mixing of water flow. The value of kh parameter (diffusivity parameter) in equation (2) was kept constant and equal to 0.05 m2s-1 (JOUON et al., 2006; RIDDLE and LEWIS, 2000). Three scenarios were considered. The first one is characterized by tidal forcing, defined on the basis of harmonic analysis of level data. The last two are characterized by wind action alone, as the tidal induced currents are very low if compared with wind induced circulation. Wind waves forcing is not considered here since preliminary numerical simulations showed that waves induced circulation is negligible if compared to tidal and wind induced one. Moreover, the system does not include freshwater inputs from rivers, because these sources have short duration and local hydrodynamic effects. Table 1 synthesizes scenarios’ parameters in terms of tide amplitude, wind speed and directions. Scenario A can be viewed as representative of mean tidal excursions characterized by the M2 semidiurnal tidal component (oscillation period equal to about 12


Lisi et al.

hours). Scenarios B and C are characterized by small wind speed. For each of the two angular sectors (NE and SW), wind speed was selected with an exceedance frequency of about 100 days per years, in order to simulate wind induced circulation that frequently occurs. Wind directions were selected equal to the more frequent for the two sectors.

RESULTS AND DISCUSSION As far as tidal induced circulation is concerned (scenario A in Table 1), Figure 2 shows the zonation of the computational area in terms of current intensities. The maximum currents intensities are obtained close to harbour entrances where water volume exchange occurs. Computed water volume exchange is about 5,280,000 m3 and 1,250,000 m3 for the eastern and Southern entrances Table 1: Scenarios’ parameters.

Scenario A B C

Tidal forcing Amplitude Period 0.15 m 12.42 h = = = =

Wind forcing Speed Direction = = 5.2 m/s 35 °N 3.8 m/s 230°N

respectively, and is about 76,000 m3 for Ponte Rivellino channel. The northern part of the harbour is characterized by the highest velocity intensities (about 5 cm/s), where a vertical axis vortex

occurs due to the presence of Ponte Rivellino channel (Figure 2, left panel). Numerical results of wind driven circulation (scenarios B and C, Table 1) show as the whole berthed area is interested by stronger currents if compared with the tidal induced ones (Figure 3). Both for scenario B and C, water volume exchange with open sea depends on the duration of wind event. Wind event duration of 6 hours has been considered in order to make wind and tidal simulation results comparable. Water volume exchange evaluated for scenario B (Figure 3, left panel) is 1,380,000 m3 and 1,620,000 m3 through Eastern and Southern entrances respectively, and 46,000 m3 through Ponte Rivellino channel. For scenario C (Figure 3, right panel) volume exchange is 210,000 m3 and 206,000 m3 for the eastern and Southern entrances respectively, and 26,900 m3 for Ponte Rivellino channel. Both numerical results of scenarios B and C reveal that the largest water exchange with the open sea occurs at the Eastern entrance. In the case of tidal forcing (scenario A) the water volume passing through the Eastern entrance is 80% of the total volume exchange. Also, it has to be highlighted that tidal induced flushing is stronger if compared with the one induced by wind. It is due to the fact that simulated winds induce currents manly directed collinearly to entrances cross section (Figure 3), especially for the eastern entrance. In particular, wind blowing from South-West (scenario C) induces water exchange that is lower than the water exchange induced from North-East wind (scenario B). Furthermore, for the Southern entrance, North-East wind induced water exchange is comparable with the tidal induced one.

Figure 2. Left panel: zonation of Augusta Harbour in terms of current intensities for scenario A (colour scale) and velocity field at the moment of maximum volume exchange (horizontal vector in the bottom left corner of the figure refers to 0.01 m/s current intensity). Right panel: numerical results of lagrangian dispersion model, white markers represent initial position of some of the particles considered, gray markers represent final position of particles after 10 days, and lines indicate the particles trajectories.


Flushing efficiency of Augusta Harbour (Italy)

Numerical results briefly detailed above have been used to compute standard hydrodynamic parameters suitable for synthesising the flushing efficiency of Augusta Harbour: Tidal Prism Ratio, Flushing Time and Exchange Coefficient. Tidal Prism Ratio is defined as the ratio between the Tidal Prism and the total water volume contained into the basin (LUKETINA, 1998). When velocity field at the open boundaries is known, it is possible to estimate the Tidal Prism as the water volume flowing to (or flowing from) the basin. Then, the Tidal Prism Ratio can be viewed as a parameter suitable to describe the exchange phenomena occurring at the harbour open boundaries. In this study the value of Tidal Prism Ratio has been evaluated both on the basis of numerical results and of field data. For the three scenarios (Table 1), computed mean velocity have been interpolated along the cross section of the Augusta Harbour open boundaries (i.e. the two main entrances and the Ponte Rivellino channel) connecting the basin to the open sea. Water discharge value (water flowing to or flowing from the harbour) is estimated for each computational time step. Figure 4 shows water discharge time series flowing through the Eastern entrance for tidal driven circulation (scenario A, Table 1). It should be noted warm up of numerical model up to the second tidal cycle. Water discharge through the Eastern entrance does not exceed the value of 370 m3/s; the Tidal Prism can be directly computed from the water discharge time series by evaluating the discharge time integral (spanning the ebb or flood phase). In the case of wind driven circulation (scenarios B and C, Table 1) only constant wind conditions are considered herein. Hence,

water discharge reaches stationary value and exchange volume can be estimated only if wind duration is defined, as previously pointed out. Tidal Prism Ratio is 1.71·10-2 for scenario A, and 0.79·10-2 and 0.11·10-2 for scenarios B and C respectively, for which a wind event duration of 6 hours have been considered. The same computational method has been employed in order to evaluate the Tidal Prism Ratio on the basis of field data. In this case only the Eastern entrance has been considered and measurements collected along a vertical profile close to the entrance have been extended to the whole cross section. The water discharge was computed by integration of velocity profile along the vertical direction and then over time (considered period 2006/10/06-2006/12/06) in order to compute volume exchange. Tidal Prism Ratio has been evaluated to be equal to 2.26·10-2 for the eastern entrance, being numerical estimate 1.37·10-2 (scenario A). As expected, field data give higher tidal prism ratios if compared with 2DH numerical results (CHOI and LEE, 2004), also due to the simultaneous effects of tidal and wind forcing. Physical meanings of Tidal Prism Ratio can be used to define a global mean Flushing Time (JOUON et al., 2006). Tidal Prism Ratio represents the fraction of harbour water volume exchange with open water during a single tidal cycle. Then it is trivial to infer that the time needed to exchange the whole water body contained in the harbour is related to Tidal Prism Ratio (CHOI and LEE, 2004). Global Flushing Time can be computed as the ratio between the total volume contained into the harbour and the mean discharge

Figure 3. Numerical results for scenario B (left panel, horizontal vector in the bottom left corner of the figure refers to 0.05 m/s current intensity) and scenario C (right panel, horizontal vector in the bottom left corner of the figure refers to 0.01 m/s current intensity) in terms of current intensities (colour scale) and velocity field (vectors). Gray arrows in the upper left corner of each panel indicate wind direction.


Lisi et al.

2.76·10-2, higher if compared with 1.83·10-2 and 2.76·10-2 computed for the middle and northern macro-area respectively. The research presented here is object of further studies carried out by using of 3D numerical models.

LITERATURE CITED

Figure 4. Water discharge flowing through the Eastern entrance for scenario A. flowing to (or flowing from) the basin. It has been estimated to be equal to 12.5 days. Numerical results obtained by using the dispersive model (2) have been used to compute E, the average per cycle exchange coefficient (NECE and RICHEY, 1975), where number of particles substitutes substance concentrations: 1/ i

n  EA 1   i  (3)  n0  Equation (3) relates the average per cycle coefficient E(A) estimated for the macro-area A to n0, the number of particle released at the initial time into the macro-area, and to ni, the number of particles in the area after i tidal cycles. If the value of E parameter reaches the unitary value, all the particles leave the macro-area. The coefficient E has been computed for three subareas covering the whole Augusta Harbour by using model (2) for about 10,000 particles uniformly released into the whole harbour area (Figure 2, right panel). The coefficient E is 0.69·10-2 for the northern portion of the harbour (y > 7000 m in figures 2 and 3), and it is 1.83·10-2 and 2.76·10-2 for the middle (4700 m < y < 7000 m in figures 2 and 3) and the southern part (y < 4700 m in figures 2 and 3) respectively. Hence the southern area of the harbour is characterized by the highest exchange coefficient. This result seems to be in contrast with the Tidal Prism Ratio results for which the Eastern entrance (belonging to the middle area) is the more effective in removing water. The highest value found in the southern part is due to the fact that water is flushed out through the Southern and the Eastern entrances together, whilst for the middle area only the Eastern entrance represent a sink for the fluid flow.

CONCLUSIONS The numerical study described herein tackled the problem of evaluating flushing features of Augusta Harbour (Sicily, Italy). A hydrodynamic 2DH numerical model has been used in order to estimate standard hydrodynamic parameters suitable for synthesising the flushing efficiency of the Harbour. In particular the computed flushing time is equal to about 12.5 days. The Tidal Prism Ratio was estimated equal to about 1.71·10-2 with a lower value of 0.79·10-2 for wind blowing from North-East and of 0.11·10-2 for wind from South-East. Comparison with field data reveals that numerical results substantially agree with field observations. A random walk model was employed in order to estimate flushing efficiency of sub-areas identified within the harbour by defining a lagrangian average per cycle exchange coefficient applied for about 10,000 particles released into the harbour. The southern area is the more effective in flushing water as both the Eastern and Southern entrances act in removing water. The corresponding value of exchange coefficient is equal to

BILGILI, A.; PROEHL, J.A.; LYNCH, D.R.; SMITH, K.W. and SWIFT, M.R., 2005. Estuary/ocean exchange and tidal mixing in a Gulf of Maine Estuary: A Lagrangian modeling study. Estuarine, Coastal and Shelf Science, 65, 607-624. CHOI, K.W. and LEE, J.H.W., 2004. Numerical determination of flushing time for stratified water bodies. Journal of Marine Systems, 50, 263-281. CNR, National Council of Research Staff, 1995. Sistema integrato per il monitoraggio automatico della Rada di Augusta. II Risultati dei rilevamenti dei sistemi di monitoraggio. Italy: CNR Report RAPPORTI 9. CNR - Istituto Sperimentale Talassografico di Messina, 241p., in Italian. EDWARDS K.P.; HARE, J.A.; WERNER, F.E. and BLANTON, B.O. 2006. Lagrangian circulation on the Southeast US Continental Shelf: Implication for larval dispersal and retention. Continental Shelf Research, 26, 1375-1394. FISCHER, H.B.; LIST, E.J.; KOH, R.C.Y.; IMBEROER, J. and BROOKS, N.H., 1979. Mixing in Inland and Coastal Waters. Academic Press, New York, NY, 483 p. HUNTER, J.R.; CRAIG, P.D. and PHILLIPS, H.E., 1993. On the use of random walk models to with spatially variable diffusivity, Journal of Computational Physics, 106(2), 366-376 IIM, Istituto Idrografico Italiano, 1994. Nautical Chart, Da C. Passero a C. Santa Croce, scale 1:100,000, 1 sheet. ISPRA, Italian High Institute for Environmental Protection and Research, 2008. http://www.idromare.it (accessed October 5, 2008). JOUON, A.; DOUILLET, P.; OUILLON, S. and FRAUNIE, P., 2006. Calculations of hydrodynamic time parameters in a semiopened coastal zone using a 3D hydrodynamic model, Continental Shelf Research, 26, 1395-1415. Luketina, D., 1998. Simple Tidal Prism Models Revisited. Estuarine, Coastal and Shelf Science, 46, 77-84. NECE, R.E. and RICHEY, E.P., 1975. Application of Physical Tidal Models in Harbor and Marina Design, Proceedings of the 15th International Conference on Coastal Engineering; 9–13 May, Honolulu, Hawaii, USA. American Society of Civil Engineers, New York, 783-801. PAWLOWICZ, R.; BEARDSLEY, B. and LENTZ, S., 2002. Classical tidal harmonic analysis including error estimates in MATLAB using T_TIDE, Computers & Geosciences, 28, 929-937. RIDDLE, A.M. and LEWIS, R.E., 2000. Dispersion Experiments in UK Coastal Waters, Estuarine, Coastal and Shelf Science, 51, 243-254. SIAP, Sviluppo Italia Aree Produttive Staff, 2007. Attività sperimentali svolte nella Rada di Augusta: presentazione ed interpretazione dei dati raccolti. Italy: Thetis, SIAP, Impresub, IDRA Tech. Report ARMAS-REL_T206.0, 48 p., in Italian. SVENDSEN, Ib. A., 2006. Introduction to nearshore Hydrodynamics. World Scientific Publishing Company, 744 p. Visser, A.J., 1997. Using random walk models to simulate the vertical distribution of particles in a turbulent water column. Marine Ecology Progress Series, 158, 275-281.

ACKNOWLEDGEMENTS The authors wish to thank Pr. Eng. Paolo De Girolamo for the useful discussions.
