Management of Dredging Activities in a Highly Vulnerable Site: Simulation Modelling and Monitoring Activity

: Unfortunately, more and more contaminants, such as heavy metals and other organic micro-pollutants, degrade the good ecological status of marine systems. The removal of contaminated sediments from harbours through dredging activities may cause harmful changes in the environment. This present work shows how monitoring the activity and validated numerical models can be of great help to dredging activities that can cause environmental impacts due to the increase of the suspended solid concentration (SSC) and their dispersion and deposition far from the dredging point. This study is applied to a hypothetical dredging project in a very vulnerable coastal site in Southern Italy, the Mar Piccolo Basin. A statistical analysis of the simulated parameter SSC was carried out to numerically estimate its spatial (vertical and horizontal) variability, thereby allowing an evaluation of the potential environmental e ﬀ ects on the coastal area.


Introduction
In the last years, the problem of contaminated marine sediments has become a worldwide environmental issue, because it threatens the marine systems and human health. However, contamination often involves large volumes of sediment but with low contamination levels; in this case, "no action" may be a preferred alternative in cases in which the remedy may be worse than the disease [1].
However, when dredging is a fundamental element of the economic/environmental performance of a basin, it is very important to define an environmental monitor strategy to assess the state of the marine ecosystem in order to avoid any adverse effects to the environment and prepare also the eventual adoption of mitigation measures.
All types of dredging operations create turbidity in the water column, depending on the type of dredges (hydraulic or mechanical), of the sediment bed and hydrodynamic conditions.
The most important environmental problems on marine organisms are reductions in dissolved oxygen and the decrease of sunlight in surface water due to turbidity.
In the last years, increasing attention has been paid to environmental impacts during dredging activities, and therefore, mitigating measures to reduce these effects have been adopted [2].
This has led to the development of new types of ecological dredges in order to reduce suspended sediment and turbidity, especially to remove and to relocate contaminated materials. In particular, special watertight buckets called "environmental" buckets are used to reduce the turbidity of dredging operations in the presence of contaminated sediments [3][4][5][6].
as supporting environmental studies and the subsequent indirect monitoring and modelling in a management context during a dredging activity. In this paper, we focused our attention on a coastal site considered one of the most polluted marine ecosystems in Europe.
More recent research paid attention to the hydrodynamic, hydrology, geology and topography characterisations of the area in order to estimate the potential environmental impacts induced by dredging activities [13][14][15][16].
This paper is organised as follows: firstly, a thorough calibration of a 3D hydrodynamic flow model is performed using a set of measured data; after this, a transport module is coupled to the hydrodynamic model to study the response of the basin to a hypothetical dredging activity.

Study Site
The Mar Piccolo of Taranto, located in Southern Italy (Ionian Sea), is a complex marine ecosystem model important in terms of ecological, social and economic activities for the presence also of extensive mussel farms [17] (Figure 1). It is composed of two bays and joined to the external basin named Mar Grande by means of two channels, i.e., the Navigable Channel and the Porta Napoli Channel. The two bays of the Mar Piccolo are considered as two different ecosystems influencing each other, and the Mar Piccolo basin was assimilated into an estuary by many authors [18][19][20].
The area of the Mar Piccolo Basin is roughly 20.72 km 2 and is characterised by the presence of a large number of submarine springs and of two small rivers, called Galeso into Bay I and Ajedda into Bay II. The maximum depth is about 12 m in Bay I and about 9 m in Bay II.
Mar Piccolo, with typical lagoon features and with a suffering scarce circulation, is extremely vulnerable, and unfortunately, this area is characterised to continue the release and diffusion of contaminants with strong chemical-ecological risks towards the marine ecosystem and human health [21][22][23][24][25][26]; therefore, it is important to test potential remediation strategies for contaminated sediments.
The lagoon features of the Mar Piccolo are mainly due to the presence of 34 submarine freshwater springs (locally called "citri"), of which 20 are in the first inlet and 14 in the second inlet.  The area of the Mar Piccolo Basin is roughly 20.72 km 2 and is characterised by the presence of a large number of submarine springs and of two small rivers, called Galeso into Bay I and Ajedda into Bay II. The maximum depth is about 12 m in Bay I and about 9 m in Bay II.
Mar Piccolo, with typical lagoon features and with a suffering scarce circulation, is extremely vulnerable, and unfortunately, this area is characterised to continue the release and diffusion of contaminants with strong chemical-ecological risks towards the marine ecosystem and human health [21][22][23][24][25][26]; therefore, it is important to test potential remediation strategies for contaminated sediments.
The lagoon features of the Mar Piccolo are mainly due to the presence of 34 submarine freshwater springs (locally called "citri"), of which 20 are in the first inlet and 14 in the second inlet.

Environmental Monitoring Action
Environmental monitoring in the Mar Grande and Mar Piccolo of Taranto includes two fixed stations briefly described below (Figure 2).

Environmental Monitoring Action
Environmental monitoring in the Mar Grande and Mar Piccolo of Taranto includes two fixed stations briefly described below (Figure 2).
In the Mar Grande Basin (Figure 3), a bottom-mounted acoustic doppler current profile (ADCP); a multidirectional wave array; a weather station and a CTD (measuring water conductivity, water temperature and depth) were installed. In the Navigable Channel, a second bottom-mounted ADCP and a wave array were installed ( Figure 2). More details on the stations can be found in De Serio and Mossa [27].  Moreover, some field data were collected by using a Nortek AWAC vessel-mounted acoustic doppler current profiler (VM-ADCP) on 26 November 2014 between 9:30 a.m. and 13:00 p.m. (GMT) by the research group of the Department of Civil, Environmental, Land, Building Engineering and Chemistry of the Polytechnic University of Bari (Italy) in the frame of the RITMARE Project. Figure  4 shows the points where the measurements were assessed. In the Mar Grande Basin (Figure 3), a bottom-mounted acoustic doppler current profile (ADCP); a multidirectional wave array; a weather station and a CTD (measuring water conductivity, water temperature and depth) were installed. In the Navigable Channel, a second bottom-mounted ADCP and a wave array were installed ( Figure 2). More details on the stations can be found in De Serio and Mossa [27].

Environmental Monitoring Action
Environmental monitoring in the Mar Grande and Mar Piccolo of Taranto includes two fixed stations briefly described below (Figure 2).
In the Mar Grande Basin (Figure 3), a bottom-mounted acoustic doppler current profile (ADCP); a multidirectional wave array; a weather station and a CTD (measuring water conductivity, water temperature and depth) were installed. In the Navigable Channel, a second bottom-mounted ADCP and a wave array were installed ( Figure 2). More details on the stations can be found in De Serio and Mossa [27].   Moreover, some field data were collected by using a Nortek AWAC vessel-mounted acoustic doppler current profiler (VM-ADCP) on 26 November 2014 between 9:30 a.m. and 13:00 p.m. (GMT) by the research group of the Department of Civil, Environmental, Land, Building Engineering and Chemistry of the Polytechnic University of Bari (Italy) in the frame of the RITMARE Project. Figure 4 shows the points where the measurements were assessed. The data recorded during the monitoring survey were used to validate the hydrodynamic model.

Numerical Simulation
Hydrodynamic and transport modules are used to evaluate how the sea circulation affects sediment transport for different dredging techniques within the target area. In particular, hydrodynamic (HD) and mud transport module (MT) modules of MIKE 3 FM, produced by the Danish Hydraulic Institute (DHI) [28], were used to model the sea current field and the sediment plume dynamics.

Modelling of Marine Currents and Calibration
The basic characteristics, numerical formulation and process equations of the model MIKE 3 FM are provided by DHI [29]. A finite mesh of 7235 triangular elements with ten vertical layers was used. The simulation was forced at the sea open boundary by the Temperature and Salinity vertical profiles and the u − v velocity vertical profiles extracted by the Mediterranean Sea physics reanalysis model [29]. The atmospheric data (u and v components of wind (m/s), atmosphere pressure (Pa), total cloud cover (%), solar radiation (J/m −2 ) and air temperature (°C )) were extracted by ERA5 developed through the Copernicus Climate Change Service (C3S). The precipitation data (mm/d) was extracted by CPC Merged Analysis of Precipitation (CMAP) [30].
A k-ε formulation for the vertical direction [31] and on the Smagorinsky formulation for the horizontal direction [32] were used. Figure 5 shows the comparison between the computed motion field (grey vectors) and the measurements (red vectors). For the sake of brevity, only two selected depths are shown, i.e., −4 m and −7m.

Mud Transport Modelling
The mud transport module (MT) coupled with the validated hydrodynamic model (HD) were used. It is able to reproduce the SSC spatial and temporal evolution during a hypothetical dredging activity in an area of 50 ha with a bed thickness equal to 50 cm ( Figure 6). For more details on this module (MT), the reader can refer to [33,34].
The Mar Piccolo is mainly characterised by silty bottom sediments (particle diameter d < 0.004 mm) with medium consolidation; consequently, a critical shear stress τce = 0.2 N/m 2 was derived for The data recorded during the monitoring survey were used to validate the hydrodynamic model.

Numerical Simulation
Hydrodynamic and transport modules are used to evaluate how the sea circulation affects sediment transport for different dredging techniques within the target area. In particular, hydrodynamic (HD) and mud transport module (MT) modules of MIKE 3 FM, produced by the Danish Hydraulic Institute (DHI) [28], were used to model the sea current field and the sediment plume dynamics.

Modelling of Marine Currents and Calibration
The basic characteristics, numerical formulation and process equations of the model MIKE 3 FM are provided by DHI [29]. A finite mesh of 7235 triangular elements with ten vertical layers was used. The simulation was forced at the sea open boundary by the Temperature and Salinity vertical profiles and the u − v velocity vertical profiles extracted by the Mediterranean Sea physics reanalysis model [29]. The atmospheric data (u and v components of wind (m/s), atmosphere pressure (Pa), total cloud cover (%), solar radiation (J/m −2 ) and air temperature ( • C)) were extracted by ERA5 developed through the Copernicus Climate Change Service (C3S). The precipitation data (mm/d) was extracted by CPC Merged Analysis of Precipitation (CMAP) [30].
A k-ε formulation for the vertical direction [31] and on the Smagorinsky formulation for the horizontal direction [32] were used. Figure 5 shows the comparison between the computed motion field (grey vectors) and the measurements (red vectors). For the sake of brevity, only two selected depths are shown, i.e., −4 m and −7 m. these sediments [28]. In the present work, we used a value of the coefficient of erodibility E equal to 1 × 10 −4 kg/s/m 2 , following [28]. Therefore, two simulation runs, denoted as T1 and T2, were carried out with the aim of comparing the effects due to a conventional hydraulic dredge and an environmental mechanical dredge.
The dredging activity was modelled over a period of 12 days, starting on 2 July 2014 at 8:00 a.m. for two tests. The hydraulic and mechanical dredges work 8 h a day with a production capacity of dry mass Qs = 0.1388 m 3 /h. Numerous researchers have developed approaches for estimating resuspension rates associated with the typical operation of hydraulic and mechanical dredges, showing a very large range of possible values.
The mean resuspension factor R provided by the USEPA [35] is equal to 2.5% for the hydraulic dredge cutter head (T1); for test T2, a conservative mean resuspension factor R was taken to be 1% based on studies by [36] for releases from an environmental bucket dredge.
Moreover, following the literature data [37], the imposed mass flow rate of suspended solids was 5.56 kg/s for the hydraulic dredge (T1) and 3.44 kg/s for the mechanical dredge (T2). Table 1 summarises the model simulations.

Mud Transport Modelling
The mud transport module (MT) coupled with the validated hydrodynamic model (HD) were used. It is able to reproduce the SSC spatial and temporal evolution during a hypothetical dredging activity in an area of 50 ha with a bed thickness equal to 50 cm ( Figure 6). For more details on this module (MT), the reader can refer to [33,34].
The Mar Piccolo is mainly characterised by silty bottom sediments (particle diameter d < 0.004 mm) with medium consolidation; consequently, a critical shear stress τ ce = 0.2 N/m 2 was derived for these sediments [28]. In the present work, we used a value of the coefficient of erodibility E equal to 1 × 10 −4 kg/s/m 2 , following [28].
Therefore, two simulation runs, denoted as T1 and T2, were carried out with the aim of comparing the effects due to a conventional hydraulic dredge and an environmental mechanical dredge.
The dredging activity was modelled over a period of 12 days, starting on 2 July 2014 at 8:00 a.m. for two tests. The hydraulic and mechanical dredges work 8 h a day with a production capacity of dry mass Q s = 0.1388 m 3 /h. The mean resuspension factor R provided by the USEPA [35] is equal to 2.5% for the hydraulic dredge cutter head (T1); for test T2, a conservative mean resuspension factor R was taken to be 1% based on studies by [36] for releases from an environmental bucket dredge.
Moreover, following the literature data [37], the imposed mass flow rate of suspended solids was 5.56 kg/s for the hydraulic dredge (T1) and 3.44 kg/s for the mechanical dredge (T2). Table 1

Sea Circulation and Sediment Plume Dynamics
During a dredging activity, it is necessary to calculate the spatial distribution of SSC and to characterise sensitive resources (receptors). Therefore, a statistical analysis of the simulated parameters SSC is carried out to numerically estimate their spatial (vertical and horizontal) variability in order to have an evaluation of the potential environmental effects on the coastal area. For both type of dredges, the SSCmax distribution was studied for two different layers during the dredging activity. Figures 7 and 8 represent the maps of SSCmax elaborated for the surface and bottom layers, respectively. The typical patterns, already observed by [38,39], are still evident, with a prevailing mean surface current outflowing from the basin during the dredging activity (Figure 7a). The dredged plume is transported from the dredged area towards the navigable channel, directed towards the Mar Grande (Figure 7b,c). Focusing on the bottom, the average current during the dredging activity inflows towards the Mar Piccolo Basin, as shown in Figure 8a. Generally, the reproduced velocity values are in the range 0.02-0.1 m/s, with some peaks (0.3 m/s) along the navigable channel of the Mar Piccolo. The simulation output confirms the double-flux along the navigable channel and, further, proves the hydrodynamic mechanism of the water-mass exchange between the Mar Grande and the Mar Piccolo.
Consequently, the Mar Piccolo Basin is seriously exposed to the plume pollution load. In fact, the plume is confined in Mar Piccolo Basin and reaches also Inlet II (Figure 8b,c).

Sea Circulation and Sediment Plume Dynamics
During a dredging activity, it is necessary to calculate the spatial distribution of SSC and to characterise sensitive resources (receptors). Therefore, a statistical analysis of the simulated parameters SSC is carried out to numerically estimate their spatial (vertical and horizontal) variability in order to have an evaluation of the potential environmental effects on the coastal area. For both type of dredges, the SSC max distribution was studied for two different layers during the dredging activity. Figures 7 and 8 represent the maps of SSC max elaborated for the surface and bottom layers, respectively. The typical patterns, already observed by [38,39], are still evident, with a prevailing mean surface current outflowing from the basin during the dredging activity (Figure 7a). The dredged plume is transported from the dredged area towards the navigable channel, directed towards the Mar Grande (Figure 7b,c). Focusing on the bottom, the average current during the dredging activity inflows towards the Mar Piccolo Basin, as shown in Figure 8a   Consequently, the Mar Piccolo Basin is seriously exposed to the plume pollution load. In fact, the plume is confined in Mar Piccolo Basin and reaches also Inlet II (Figure 8b,c).
Moreover, in the case of the hydraulic dredge, a higher intensity of the SSC max in the immediate vicinity of the dredging point is shown at the surface and bottom layers (Figures 7a and 8a).

Detecting Sensitive Areas around the Dredging Site
During the monitoring of a dredging activity, quantitative impact thresholds for the suspended solid concentration (SSC) must be established. In this study, a threshold of SSC was arbitrarily considered equal to 5 mg/L [40], and a regular grid of checkpoints (Figure 9) in the zone with a high impact was considered [41].

Detecting Sensitive Areas around the Dredging Site
During the monitoring of a dredging activity, quantitative impact thresholds for the suspended solid concentration (SSC) must be established. In this study, a threshold of SSC was arbitrarily considered equal to 5 mg/L [40], and a regular grid of checkpoints (Figure 9) in the zone with a high impact was considered [41]. For all checkpoints, starting from the SSC time series, the mean and maximum values (SSCmean and SSCmax) were calculated (Figure 9).
According to Equation (1) by Feola et al. [40], the SSCnum-SSC number (mg s/L) was evaluated as the sum of the products of intensity and duration of the single events.
For the sake of brevity, only the checkpoint A, highlighted with a black circle in Figure 9b, is shown (Figures 10 and 11). Figure 12a,b shows the distribution of SSCnum for hydraulic and mechanical dredging plumes at the surface and bottom layers for all checkpoints (Figure 9). For all checkpoints, starting from the SSC time series, the mean and maximum values (SSC mean and SSC max ) were calculated (Figure 9).
According to Equation (1) by Feola et al. [40], the SSC num -SSC number (mg s/L) was evaluated as the sum of the products of intensity and duration of the single events.
For the sake of brevity, only the checkpoint A, highlighted with a black circle in Figure 9b, is shown (Figures 10 and 11). Figure 12a,b shows the distribution of SSC num for hydraulic and mechanical dredging plumes at the surface and bottom layers for all checkpoints (Figure 9).    Maps of the SSCnum distribution are shown for the hydraulic and mechanical dredgings at the surface and bottom layers (Figures 13 and 14). Figures 13a and 14a highlight a greater "affected" area in terms of the SSCnum variation at the surface and bottom layers, and therefore, the hydraulic dredging produces the worst conditions. Maps of the SSC num distribution are shown for the hydraulic and mechanical dredgings at the surface and bottom layers (Figures 13 and 14). In order to study the effects of the plume dynamic as a function of distance from the dredging zone, the variogram function ɣ(h) [42] was derived (Figure 9). If the variogram reaches a limit value (sill), it means that there is a distance (range) beyond which the variable does not have spatial dependence (Figure 15). The variograms of the SSCnum presented in Figure 16 were obtained, for both hydraulic and mechanical dredges, at the surface and bottom layers.
By analysing Figure 16, the experimental variogram starts out straight, then bends over sharply and levels out; therefore, the spherical model matches in a better way the shape of the empirical variogram ɣ(h).
Considering the relative differences for contiguous values lower than 5 × 10 −1 , the spatial extent of the bottom layer SSCnum correlation can be identified around 1300 m and 700 m for hydraulic and mechanical dredges, respectively (Table 2).
Using the same criterion, the spatial extent of the surface layer SSCnum correlation can be identified around 850 m and 380 m for hydraulic and mechanical dredges, respectively (Table 2).
Coherently with the maps of the SSCnum distribution (Figure 13a,b), this result highlighted that the most dangerous conditions are observed for hydraulic dredge at the bottom layer.
Choosing a SSCnum meaningful threshold equal to 100 mg/L × h, circles that can be considered as the limit between "not affected" and "affected" areas are introduced in Figures 13a and 14b. Figures 13a and 14a highlight a greater "affected" area in terms of the SSC num variation at the surface and bottom layers, and therefore, the hydraulic dredging produces the worst conditions.
In order to study the effects of the plume dynamic as a function of distance from the dredging zone, the variogram function γ(h) [42] was derived (Figure 9). If the variogram reaches a limit value (sill), it means that there is a distance (range) beyond which the variable does not have spatial dependence (Figure 15). In order to study the effects of the plume dynamic as a function of distance from the dredging zone, the variogram function ɣ(h) [42] was derived (Figure 9). If the variogram reaches a limit value (sill), it means that there is a distance (range) beyond which the variable does not have spatial dependence (Figure 15). The variograms of the SSCnum presented in Figure 16 were obtained, for both hydraulic and mechanical dredges, at the surface and bottom layers.
By analysing Figure 16, the experimental variogram starts out straight, then bends over sharply and levels out; therefore, the spherical model matches in a better way the shape of the empirical variogram ɣ(h).
Considering the relative differences for contiguous values lower than 5 × 10 −1 , the spatial extent of the bottom layer SSCnum correlation can be identified around 1300 m and 700 m for hydraulic and mechanical dredges, respectively (Table 2).
Using the same criterion, the spatial extent of the surface layer SSCnum correlation can be identified around 850 m and 380 m for hydraulic and mechanical dredges, respectively (Table 2).
Coherently with the maps of the SSCnum distribution (Figure 13a,b), this result highlighted that the most dangerous conditions are observed for hydraulic dredge at the bottom layer.
Choosing a SSCnum meaningful threshold equal to 100 mg/L × h, circles that can be considered as the limit between "not affected" and "affected" areas are introduced in Figures 13a and 14b. The variograms of the SSC num presented in Figure 16 were obtained, for both hydraulic and mechanical dredges, at the surface and bottom layers.

Conclusions
The numerical model MIKE 3D FM MT was implemented to evaluate the response of the Mar Piccolo Basin to a hypothetical dredging activity during the spring season. To reproduce the sea circulation structure in the target area, a calibrated MIKE 3D FM model was used.
Environmental improvements to dredging techniques have become increasingly important over the last 25 years. Therefore, two simulation runs, denoted as T1 and T2, were carried out with the aim of comparing the effects due to two different dredges (i.e., hydraulic dredge cutter head and environmental mechanical bucket dredge).
A statistical analysis of the simulated parameter SSC was carried out to numerically estimate its spatial (vertical and horizontal) variability, thereby allowing an evaluation of the potential environmental effects on the coastal area.
For both types of dredges, the SSCmax variability distribution was evaluated at the surface and bottom.
In the case of a hydraulic dredge, a higher intensity of SSCmax in the immediate vicinity of the dredging point is shown at the surface and bottom layers. For both tests, T1 and T2, at the surface level, the dredged plume is transported from the dredged area towards the navigable channel of the By analysing Figure 16, the experimental variogram starts out straight, then bends over sharply and levels out; therefore, the spherical model matches in a better way the shape of the empirical variogram γ(h).
Considering the relative differences for contiguous values lower than 5 × 10 −1 , the spatial extent of the bottom layer SSC num correlation can be identified around 1300 m and 700 m for hydraulic and mechanical dredges, respectively (Table 2). Using the same criterion, the spatial extent of the surface layer SSC num correlation can be identified around 850 m and 380 m for hydraulic and mechanical dredges, respectively (Table 2).
Coherently with the maps of the SSC num distribution (Figure 13a,b), this result highlighted that the most dangerous conditions are observed for hydraulic dredge at the bottom layer.
Choosing a SSC num meaningful threshold equal to 100 mg/L × h, circles that can be considered as the limit between "not affected" and "affected" areas are introduced in Figures 13a and 14b.
The diameters of the circles are about 1300 m and about 700 m at the bottom layer and about 850 m and about 380 m at the surface layer for hydraulic and mechanical dredges, corresponding to the range shown by the variograms analysis.

Conclusions
The numerical model MIKE 3D FM MT was implemented to evaluate the response of the Mar Piccolo Basin to a hypothetical dredging activity during the spring season. To reproduce the sea circulation structure in the target area, a calibrated MIKE 3D FM model was used.
Environmental improvements to dredging techniques have become increasingly important over the last 25 years. Therefore, two simulation runs, denoted as T1 and T2, were carried out with the aim of comparing the effects due to two different dredges (i.e., hydraulic dredge cutter head and environmental mechanical bucket dredge).
A statistical analysis of the simulated parameter SSC was carried out to numerically estimate its spatial (vertical and horizontal) variability, thereby allowing an evaluation of the potential environmental effects on the coastal area.
For both types of dredges, the SSC max variability distribution was evaluated at the surface and bottom.
In the case of a hydraulic dredge, a higher intensity of SSC max in the immediate vicinity of the dredging point is shown at the surface and bottom layers. For both tests, T1 and T2, at the surface level, the dredged plume is transported from the dredged area towards the navigable channel of the Mar Piccolo, directed towards the Mar Grande Sea. Confined at the bottom the plume is the Mar Piccolo Basin, reaching also Inlet II.
Although, for careful management during a dredging activity, site-specific evaluations must be carried out, in this case study, a preliminary SSC threshold and an integrated index, SSC num , were proposed.
Moreover, to study the effects of the plume dynamics due to two different dredges as a function of distance from the dredging zone, a geostatistical analysis of function SSC num was carried out.
In this way, the variogram upper-bound value (sill) was used to estimate the severity of the effects. Table 2 shows that the most dangerous conditions are shown for hydraulic dredge at the bottom layer. It is evident that the correct prediction of contaminant transport depends on the reliability of the input data, as well as on the calibration made.