Analysis of the spatio-temporal evolution of dredging from satellite images: a case study in the Principality of Asturias (Spain)
Abstract
This work was funded by the Science, Technology and Innovation Plan of the Principality of Asturias (Spain) Ref: FC-GRUPIN-IDI/2018/000225, which is partly funded by the European Regional Development Fund (ERDF).
Full text
Journal of Marine Science and Engineering Article Analysis of the Spatio-Temporal Evolution of Dredging from Satellite Images: A Case Study in the Principality of Asturias (Spain) Vanesa Mateo-Pérez 1, Marina Corral-Bobadilla 2,* , Francisco Ortega-Fernández 1 and Vicente Rodríguez-Montequín1 Citation: Mateo-Pérez, V.; Corral-Bobadilla, M.; Ortega-Fernández, F.; Rodríguez-Montequín, V. Analysis of the Spatio-Temporal Evolution of Dredging from Satellite Images: A Case Study in the Principality of Asturias (Spain). J. Mar. Sci. Eng. 2021,9, 267. https://doi.org/10.3390/jmse9030267 Academic Editor: Rodger Tomlinson Received: 5 February 2021 Accepted: 25 February 2021 Published: 2 March 2021 Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations. Copyright: © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/). 1Project Engineering Department, University of Oviedo, 33004 Oviedo, Principality of Asturias, Spain; [email protected] (V.M.-P.); [email protected] (F.O.-F.); [email protected] (V.R.-M.) 2Department of Mechanical Engineering, University of La Rioja, 26004 Logroño, La Rioja, Spain *Correspondence: [email protected]; Tel.: +34-941-299-274 Abstract: One of the fundamental tasks in the maintenance of port operations is periodic dredging. These dredging operations facilitate the elimination of sediments that the coastal dynamics introduce. Dredging operations are increasingly restrictive and costly due to environmental requirements. Understanding the condition of the seabed before and after dredging is essential. In addition, determining how the seabed has behaved in recent years is important to consider when planning future dredging operations. In order to analyze the behavior of sediment transport and the changes to the seabed due to sedimentation, studies of littoral dynamics are conducted to model the deposition of sediments. Another methodology that could be used to analyze the real behavior of sediments would be to study and compare port bathymetries collected periodically. The problem with this methodology is that it requires numerous bathymetric surveys to produce a sufficiently significant analysis. This study provides an effective solution for obtaining a dense time series of bathymetry mapping using satellite data, and enables the past behavior of the seabed to be examined. The methodology proposed in this work uses Sentinel-2A (10 m resolution) satellite images to obtain historical bathymetric series by the development of a random forest algorithm. From these historical bathymetric series, it is possible to determine how the seabed has behaved and how the entry of sediments into the study area occurs. This methodology is applied in the Port of Luarca (Principality of Asturias), obtaining satellite images and extracting successive bathymetry mapping utilizing the random forest algorithm. This work reveals how once the dock was dredged, the sediments were redeposited and the seabed recovered its level prior to dredging in less than 2 months. Keywords: Luarca port; random forest; dredging activities; sediment transport; satellite images 1. Introduction Dredging activities are commonly used in coastal areas to maintain the designed depth of navigation channels or basins and to remove deposited sediments. These activities include not only the processes of removing sediment from the bottom, but also their subsequent transport to another location. Dredging of ports improves access and exit conditions for cargo and passenger ships [ 1 , 2 ]. This is important for marine safety, removing sediment that has accumulated in channels in order to maintain the designed depths of the existing facilities [ 3 , 4 ]. However, despite its importance in maritime works and its effect on economic and social development, and the current advances in dredging techniques, the siltation of ports continues to be one of the least understood branches of coastal engineering. It is of great importance to ports, in maintaining and improving its depth and, also, in developing new facilities or creating new ports [ 5 ]. Thus, it is particularly important to study the sedimentation that takes place in the basins and entrance channels of ports. Excessive sedimentation can impair the functioning of the port and lead to a decrease in J. Mar. Sci. Eng. 2021,9, 267. https://doi.org/10.3390/jmse9030267 https://www.mdpi.com/journal/jmse
J. Mar. Sci. Eng. 2021,9, 267 2 of 18 economic activities [ 6 – 8 ]. The construction of dikes and other coastal protection works have a great influence on hydrodynamic forces and movement of sediment in the offshore area [ 9 ]. Hydrodynamic and sedimentary conditions can cause sedimentation of the port channel, which must be removed by maintenance dredging [10]. The Port of Luarca is located in the Principality of Asturias and is one of the most important ports in northern Spain. In recent years there have been significant changes in the marine dynamics of the port, as well as in the erosion–sedimentation processes that operate there. To maintain navigation, dredging is undertaken once a year. Technical specifications include the estimation of siltation volume, location, duration and phases of dredging works to be performed, the dredging method and the location of suitable sediment dumping locations that will not affect the environment adversely [ 11 ]. At Luarca, sandy bottoms prevail, although several areas are presently covered by mud [ 12 ]. The purpose of dredging the dock area is to guarantee the functionality of the port, which is conditioned by the depth of the dock. The location of the Port of Luarca at the mouth of the Rio Negro favors the accumulation of sedimentary materials of fluvial origin. The dock’s location forestalls the tidal dynamics that are necessary to wash away the deposited sediments. For this reason, periodic dredging is necessary to maintain the dock. This enables optimal levels of use of the different port areas to be achieved. Throughout its operational life, the port has faced significant sedimentation problems and a consequent decrease in the depth of the draft in its entrance channel. The problem of sedimentation is increasing due to the environmental problems that are associated with dredging and the restrictive regulations imposed as a result. This sedimentation problem may be related to the positioning and layout of the port basin and its entry location [ 13 , 14 ]. Due to high sedimentation rates, the basin and entrance channel of the Port of Luarca must be dredged regularly for the port to continue its operation. Many different methodologies have been used in an effort to understand the behavior of sediments. These have ranged from the use of probabilistic design methods [ 15 – 18 ] to direct numerical simulations [ 19 – 21 ]. Direct numerical simulations provide a rough understanding of the behavior of sediments. However, it is a costly methodology, because it requires the survey of the entire area in order to understand its 3D geometry. Another technique that could be used to study the behavior of sediments is to conduct an analysis of the bathymetries that were obtained in the study area. If the evolution of the seabed is known, it is possible to know where the sediment deposits are produced and their sedimentation rate. This technique would also help to determine the need to undertake dredging, as well as its periodicity. The main limitation when analyzing the behavior of the seabed is the availability of bathymetries. Bathymetries are normally conducted before and after dredging, but not frequently enough to understand the evolution of the seabed, due to their high cost and the difficulty of carrying them out. This clearly limits their use in analyzing the behavior of the seabed. Therefore, for this purpose, bathymetries are obtained from satellite images using the simplest techniques that offer a high level of precision. In addition, the techniques must be applicable in areas of turbid and shallow water and provide rapidity and flexibility of use. Conventional bathymetric methods usually provide depth profiles or point measurements of acceptable accuracy. However, their use is costly and inefficient. Bathymetric estimation can be improved by combining echo sounding and satellite data. Remote sensing technology is used in many topographical studies. This is very useful for situations in which the depths of water are required on temporal and spatial scales, but are difficult to obtain [ 22 – 25 ]. Simple methods that use optical images in estimating depths are used by some investigator methods, as well as linear regression algorithms to estimate the depth of water [ 26 ]. In recent years, the use of machine learning techniques and optical sensors to estimate bathymetries has gained popularity. This can be attributed to advances in the development of algorithms, as well as greater availability of data and new satellites. Many authors have sought to determine the depths of water depths from optical sensors and the use of regression models that were derived from machine learning techniques [ 27 – 32 ]. The
J. Mar. Sci. Eng. 2021,9, 267 3 of 18 random forest algorithm was used in this study to obtain the bathymetries. This algorithm has a number of advantages, including, low-cost, simplicity, time-effectiveness, and its wide-coverage for shallow water. Thus, the random forest algorithm has been found to be applicable to construct regression models that rely on bathymetry data from satellite images [33–41]. Many studies have concentrated on migration of sediments over the long term and how this affects water quality and the ecosystem condition. They model movement of material that is removed during the dredging operation [ 42 ]. However, optimizing the dredging operation has not been proposed. The main objective of this work is to determine if the behavior of sedimentation differs with the depth of dredging. In other words, is there a stable mean depth outside of which erosion and sedimentation occur more quickly? This is possible due to the capacity of the methodology, which enables data from the past with a frequency of 1–2 months to be analyzed. This provides great information of the seabed’s behavior. 2. Study Site The Port of Luarca is situated in northwestern Spain on the coast of the Cantabrian Sea at a longitude of 6 ◦ 32’1” W and a latitude of 43 ◦ 32’45” N (Figure 1a). Luarca has been associated with maritime activity since it began as a fishing enclave in the 10th century. Several changes have occurred throughout past years in Luarca Port. In 1910, the Canouco dike (Figure 1b), 40 m in length, was built to shelter the outer basin of the port. In 1940 a new dike, the La Encoronada dike, was built (Figure 1b), 124 m in length on the western side [ 43 ]. It is necessary to conduct dredging of the dock annually. The lack of draft means that the boats cannot enter the port, as many currents are generated in the entrance channel. Thus, navigation in this area is endangered by the risk that the boats will be stranded, especially during storms. The port is used generally by small boats. Thus, the minimum drafts in the navigation channel range from 2 to 3 m, with an average tidal run of 2 m. The draft in the docking area is somewhat less—2 m. Although dredging is conducted in the inlet channels, it also affects the inner basin. Of the two areas, it is the inner basin that receives the most interventions and is most limited due to draft. Nalona, a dredger boat (Suction Hopper Dredger, IMO 9047453) [ 44 ], has been used to dredge the bed surrounding the port. After the sand and mud have been collected from the bottom of the Port of Luarca, the sand is deposited in the area of Punta Muyeres, in front of the third beach of the village, in an effort to renew this area. However, it is not very clear whether this discharge location benefits the port, since a large amount of the sand that is dumped there will be moved by the currents and re-enter the dikes. The mud, on the other hand, is deposited at another location that is farther from the beaches. The dates on which the dredging is carried out in the inner basin are provided by the Port Service of the Principality of Asturias. Dredging operations were carried out in 2017 and 2018. In 2017, the dredging began in October, but was stopped because of a problem with the dredge. It resumed in November and ended in December, 2017. In 2018, dredging began on November 15 and continued until the end of the year. In 2019, the inlet channel was dredged, but little work was done in the inner basin.
J. Mar. Sci. Eng. 2021,9, 267 4 of 18 J. Mar. Sci. Eng. 2021, 9, x FOR PEER REVIEW 4 of 18 Figure 1. The Port of Luarca: (a) position of the port on the Cantabrian Sea coast; (b) the study site location. 3. Data and Methodology 3.1. Data In Situ Measurements and Satellite Data Field measurements are generally necessary to study the sedimentation in the port, including the hydrodynamic parameters of the area, in this case only bathymetries will be used as field measurements, and these bathymetries are provided annually by the Port Service of the Principality Asturias. The bathymetries were conducted with the use of a Navisound 210 single beam echo sounder, with a 1 cm vertical resolution and dual frequency (Reson, Inc.: Slangerup, Denmark). The specifications of echo sounder Navisound 210 are shown in Table 1. Table 1. Characteristics of echo sounder Navisound 210. Navisound 210/400 Specifications Frequency 190–235 kHz Potency of transmission 300 W Impedance 100 Ohm Echo approval length 100 µs–210 kHz Depth range 0.5–100/400/1200 m depending on the frequency Resolution 1 cm Accuracy 1 cm at 210 kHz (1 σ) assuming correct sound velocity, transducer depth etc. The study’s echo sounding data for Luarca were determined, on 28 June 2016, 10 May 2018 and 28 May 2019. For the present work, XY positions from the survey data were Figure 1. The Port of Luarca: ( a ) position of the port on the Cantabrian Sea coast; ( b ) the study site location. 3. Data and Methodology 3.1. Data In Situ Measurements and Satellite Data Field measurements are generally necessary to study the sedimentation in the port, including the hydrodynamic parameters of the area, in this case only bathymetries will be used as field measurements, and these bathymetries are provided annually by the Port Service of the Principality Asturias. The bathymetries were conducted with the use of a Navisound 210 single beam echo sounder, with a 1 cm vertical resolution and dual frequency (Reson, Inc.: Slangerup, Denmark). The specifications of echo sounder Navisound 210 are shown in Table 1. Table 1. Characteristics of echo sounder Navisound 210. Navisound 210/400 Specifications Frequency 190–235 kHz Potency of transmission 300 W Impedance 100 Ohm Echo approval length 100 µs–210 kHz Depth range 0.5–100/400/1200 m depending on the frequency Resolution 1 cm Accuracy 1 cm at 210 kHz (1 σ) assuming correct sound velocity, transducer depth etc. The study’s echo sounding data for Luarca were determined, on 28 June 2016, 10 May 2018 and 28 May 2019. For the present work, XY positions from the survey data were
J. Mar. Sci. Eng. 2021,9, 267 5 of 18 projected using UTM Zone 30N. As they are usually costly, they are normally used only for specific dates. The satellite Sentinel-2 provided the data with which to predict the bathymetry of the water depth at the port (Table 2). Table 2. Dates of acquisition of bathymetries. Year 2017 2018 2019 2020 05/01/2019 25/01/2020 24/02/2018 24/02/2019 19/02/2020 16/03/2018 10/03/2020 10/04/2018 20/04/2019 30/05/2019 24/06/2018 14/06/2019 04/07/2017 29/07/2018 24/07/2019 13/08/2017 18/08/2018 23/08/2019 02/09/2017 02/09/2018 12/09/2019 02/10/2017 02/10/2018 22/10/2019 21/11/2017 16/11/2018 21/11/2019 21/12/2017 31/12/2018 3.2. Methodology The process to obtain the bathymetries is shown schematically in Figure 2. J. Mar. Sci. Eng. 2021, 9, x FOR PEER REVIEW 5 of 18 projected using UTM Zone 30N. As they are usually costly, they are normally used only for specific dates. The satellite Sentinel-2 provided the data with which to predict the bathymetry of the water depth at the port (Table 2). Table 2. Dates of acquisition of bathymetries. Year 2017 2018 2019 2020 05/01/2019 25/01/2020 24/02/2018 24/02/2019 19/02/2020 16/03/2018 10/03/2020 10/04/2018 20/04/2019 30/05/2019 24/06/2018 14/06/2019 04/07/2017 29/07/2018 24/07/2019 13/08/2017 18/08/2018 23/08/2019 02/09/2017 02/09/2018 12/09/2019 02/10/2017 02/10/2018 22/10/2019 21/11/2017 16/11/2018 21/11/2019 21/12/2017 31/12/2018 3.2. Methodology The process to obtain the bathymetries is shown schematically in Figure 2. Figure 2. Methodological workflow for development of bathymetric maps. 3.2.1. Processing Satellite Images The processing step concerns using Sentinel-2A data as vector of variables for training and fitting of the random forest model (Figure 2). Sentinel-2A satellite data was retrieved from ESA Copernicus Open Access Hub. SNAP (Sentinel Application Platform) software was used to visualize and preprocess Sentinel-2A data (10 m resolution) [45]. The data from the Sentinel-2A satellite reflectance bands (B1, B2, B3, B4, B5, B6, B7, B8, Figure 2. Methodological workflow for development of bathymetric maps. 3.2.1. Processing Satellite Images The processing step concerns using Sentinel-2A data as vector of variables for training and fitting of the random forest model (Figure 2). Sentinel-2A satellite data was retrieved from ESA Copernicus Open Access Hub. SNAP (Sentinel Application Platform) software was used to visualize and preprocess Sentinel-2A data (10 m resolution) [ 45 ]. The data from the Sentinel-2A satellite reflectance bands (B1, B2, B3, B4, B5, B6, B7, B8, B8A, B9, B11, and B12) were used to predict the depth of the water at the study port. All spectral bands in the Sentinel-2A images were resampled to achieve a resolution of 10 m using the SNAP S2 Resampling Processor [ 15 , 46 , 47 ]. As a result, a dataset without georeferencing
J. Mar. Sci. Eng. 2021,9, 267 6 of 18 was obtained, and to determine the positioning of the reference points, the geographical location of each point was defined by its longitude and latitude using the SNAP program. Then, using the WGS84 ellipsoid, a coordinate projection was created in order to obtain the coordinates in ETRS89 [ 48 ]. This system was also used to project the positions that the echo sounder provides. The ellipsoid projections had an average position error of 1 cm. The data that was obtained was compared to the bathymetry that had been projected. To accomplish this, a geodesic calculator was used to project the coordinates. This enabled the authors to obtain the data for bands that are associated with UTM x-y coordinates. The random forest algorithm was used to assign the z coordinate. From these points and using the QGIS software (version 3.14), the surface was obtained by digital models of the TIN (Triangulated Irregular Network) terrain type. Triangulation was undertaken using linear interpolation [ 49 ]. The error that this process involved was small due to the absence of any great irregularities in the smooth surfaces of the seabed. The port’s zero serves as the reference point for the z coordinates. The former is the minimum measured level of the surface of the water during the highest low tide in the last 15 years. Pixels were assigned dimensions on the basis of their x-y locations. 3.2.2. The Random Forest Algorithm The random forest algorithm [ 50 ] facilitates classification and regression by use of a randomized subset of predictors that enable it to create a variety of classification trees [ 51 ]. Many of the trees start as bootstrapped training data samples. A random subset of predictor variables (z coordinate) is used at each fork in the process. This causes each tree to be different. Although each tree is a poor predictor, each pair of them provides a different response. This aggregates the predictions of uncorrelated trees, reducing prediction variance and increasing accuracy [ 52 – 54 ]. This work involved 100 trees and 13 variables (Sentinel-2A satellite bands B1, B2, B3, B4, B5, B6, B7, B8, B8A, B9, B11, and B12, and tidal). The default value was adopted as the minimum size of nodes. The random forest algorithm for prediction of bathymetry was provided by the Random Forest (v 4.6-2) R package [ 55 ]. 3.2.3. Training and Testing Dataset In order to evaluate the accuracy of the model, the dataset that was obtained from Sentinel-2 was divided into two groups. Training the random forest algorithm required 80% of the dataset (1593 data points), and testing the model used the remaining 20% ( 388 data points ). Data from echo sounding measurements were used to calibrate the model without any kind of elaboration [ 56 ]. To test the model, the data obtained with the model were compared with field measurements for 20% of the points. As the tide affected the depth measurement results, the measured depths were the mean sea levels (MSL) that served as references. This was accomplished by deducting the measured depth during high tide from the MSL. The tidal data were provided by the nearest tidal station [34]. Finally, in order to determine the accuracy of the model’s estimates of depth, satellite derived bathymetry maps were compared with that of the field measurements obtained by using the echo sounder. The residual statistic between the satellite derived depth (Zsatellite) and the echo sounding measurements (Zecho-sounder) were reported along three metrics. They were the mean absolute error (MAE), the root mean squared error (RMSE) and the correlation coefficient (adjusted R2). They are calculated by the equations below. MAE =1 n n ∑ i=1 |ZSatellite −Zecho−sounder| RMSE =s1 n n ∑ i=1 (ZSatellite −Zecho−sounder)2 R2=∑n i=1Zecho−sounder −Zecho−sounderZSatellite −ZSatellite q∑n i=1Zecho−sounder −Zecho−sounder2∑n i=1Zecho−sounder −Zecho−sounder2
J. Mar. Sci. Eng. 2021,9, 267 7 of 18 where Z Satellite are the depths that the random forest methodology predicted from Sentinel-2 images. Zecho-sounder denote the in situ echo sounding depths and nis the number of data. 4. Results The random forest algorithm achieved a good predictive performance, with an MAE of 0.37, an RMSE of 0.47 and an R2of 0.974. The testing data set results appear in Table 3. Table 3. Error statistics reported in meters produced by the random forest algorithm. Algorithm MAE (m) RMSE (m) R2 Random forest 0.37 0.47 0.974 Figure 3provides a histogram of the relationship between mean absolute error (MAE) and the depth of the port of Luarca that was obtained by the random forest algorithm. As can be seen in Figure 3, the maximum error (60 cm) occurred between − 6 and − 8 m and the minimum (24 cm) occurred at − 10 m. This error is acceptable for the purpose of this study, which is to analyze the behavior of the seabed with respect to time. J. Mar. Sci. Eng. 2021, 9, x FOR PEER REVIEW 7 of 18 RMSE=1𝑛(𝑍−𝑍) R=∑(𝑍−𝑍 )(𝑍−𝑍 ) ∑( 𝑍−𝑍 )∑( 𝑍−𝑍 ) where ZSatellite are the depths that the random forest methodology predicted from Sentinel2 images. Zecho-sounder denote the in situ echo sounding depths and n is the number of data. 4. Results The random forest algorithm achieved a good predictive performance, with an MAE of 0.37, an RMSE of 0.47 and an R² of 0.974. The testing data set results appear in Table 3. Table 3. Error statistics reported in meters produced by the random forest algorithm. Algorithm MAE (m) RMSE (m) R2 Random forest 0.37 0.47 0.974 Figure 3 provides a histogram of the relationship between mean absolute error (MAE) and the depth of the port of Luarca that was obtained by the random forest algorithm. As can be seen in Figure 3, the maximum error (60 cm) occurred between −6 and −8 m and the minimum (24 cm) occurred at −10 m. This error is acceptable for the purpose of this study, which is to analyze the behavior of the seabed with respect to time. Figure 3. Variation of MAE errors versus depth (m). As an example of the results that were obtained, the bathymetries before and after the 2017 dredging are represented in 3D and 2D (Figures 4 and 5). Figure 4 represents the 3D evolution of the seabed for 21/11/17 (Figure 4a), 21/12/17 (Figure 4b) and 24/02/18 (Figure 4c). These show that the elevation of the seabed was maintained in Figure 4a,b, but had declined in Figure 4c due to dredging. Figure 3. Variation of MAE errors versus depth (m). As an example of the results that were obtained, the bathymetries before and after the 2017 dredging are represented in 3D and 2D (Figures 4and 5). Figure 4represents the 3D evolution of the seabed for 21/11/17 (Figure 4a), 21/12/17 (Figure 4b) and 24/02/18 (Figure 4c). These show that the elevation of the seabed was maintained in Figure 4a,b, but had declined in Figure 4c due to dredging. Figure 5represents one of the dredging episodes by a series of bathymetries that were analyzed in the plane and profile view. These indicate the effect of the dredging that was undertaken. First, when dredging is carried out, there is a decrease in the elevation of the basin and the inlet channel (Figure 5a,c). The elevation recovers when the dredging stops (Figure 5b,d). This effect is seen in both the inner dock (section AA’) and in the inlet channel (section BB’). In both representations (Figures 4and 5), the areas of greatest variation are the inner basin and the inlet channel. These are the areas where dredging takes place. Additionally, in Figures 4and 5, the areas in which the greatest changes occur are in the lower right corner (inner dock area and inlet channel) and in the upper left corner that coincides with the beach area. Both are areas that undergo many changes in topography due to coastal dynamics.
J. Mar. Sci. Eng. 2021,9, 267 8 of 18 J. Mar. Sci. Eng. 2021, 9, x FOR PEER REVIEW 8 of 18 Figure 4. 3D bathymetries. Dates: (a) 21/11/17, (b) 21/12/17 and (c) 24/02/18. Figure 4. 3D bathymetries. Dates: (a) 21/11/17, (b) 21/12/17 and (c) 24/02/18.
J. Mar. Sci. Eng. 2021,9, 267 9 of 18 J. Mar. Sci. Eng. 2021, 9, x FOR PEER REVIEW 9 of 18 Figure 5. A representation of the behavior of the interior basin the dredging on: (a) 02/10/17, (b) 21/11/17, (c) 21/12/17 and (d) 24/02/18; and cross-sectional profile (e) AA’ of the inner dock, and (f) BB’ of the inlet channel. Figure 5. A representation of the behavior of the interior basin the dredging on: ( a ) 02/10/17, ( b ) 21/11/17, ( c ) 21/12/17 and (d) 24/02/18; and cross-sectional profile (e) AA’ of the inner dock, and (f) BB’ of the inlet channel.
J. Mar. Sci. Eng. 2021,9, 267 16 of 18 recovered almost immediately (in the dredging of 2018 in 5 days) with filling rates of 589.90 and 3556.35 m 3 /day, for 2017 and 2018, respectively. It also can be stated that the sporadic arrival of sediment above the mean depth also disappeared without external intervention, keeping the mean depth between − 3 and − 4 m. As a result, it can be concluded that lowering the elevation by dredging does not seem adequate, unless the dredging depth is close to the average elevation that is maintained naturally. For future work, it might be useful to extend this investigation to other ports. Before determining the need for dredging, it would be useful to know if there is a depth beyond which it is not appropriate to conduct further dredging due to the rate at which sediment returns. The analysis of the behavior of the bottom of the ports in recent years provides valuable information for the planning of maintenance tasks and the knowledge of their behavior in the face of littoral dynamics. Author Contributions: Conceptualization, V.M.-P. and F.O.-F.; software and validation, V.M.-P.; methodology, M.C.-B.; data curation, V.M.-P.; writing—original draft preparation, M.C.-B.; writing— review and editing, F.O.-F., and V.R.-M. All authors have read and agreed to the published version of the manuscript. Funding: This work was funded by the Science, Technology and Innovation Plan of the Principality of Asturias (Spain) Ref: FC-GRUPIN-IDI/2018/000225, which is partly funded by the European Regional Development Fund (ERDF). Institutional Review Board Statement: Not applicable. Informed Consent Statement: Not applicable. Acknowledgments: The authors would like to thank the Port Service and Transport Infrastructures of the Principality of Asturias for their collaboration in this work. Conflicts of Interest: The authors declare there are no conflict of interest. References 1. Quang Tri, D.; Kandasamy, J.; Cao Don, N. Quantitative Assessment of the Environmental Impacts of Dredging and Dumping Activities at Sea. Appl. Sci. 2019,9, 1703. [CrossRef] 2. Bolam, S.G.; Rees, H.L. Minimizing Impacts of Maintenance Dredged Material Disposal in the Coastal Environment: A Habitat Approach. Environ. Manag. 2003,32, 171–188. [CrossRef] 3. Norén, A.; Fedje, K.K.; Strömvall, A.-M.; Rauch, S.; Andersson-Sköld, Y. Integrated Assessment of Management Strategies for Metal-Contaminated Dredged Sediments—What Are the Best Approaches for Ports, Marinas and Waterways? Sci. Total Environ. 2020,716, 135510. [CrossRef] 4. Wang, W.; Men, C.; Lu, W. Online Prediction Model Based on Support Vector Machine. Neurocomputing 2008 ,71, 550–558. [CrossRef] 5. Khorram, S.; Khalegh, M.A. A Novel Hybrid MCDM Approach to Evaluate Ports’ Dredging Project Criteria Based on Intuitionistic Fuzzy DEMATEL and GOWPA. WMU J. Marit. Aff. 2020,19, 95–124. [CrossRef] 6. Cáceres, R.A.; Zyserman, J.A.; Perillo, G.M.E. Analysis of Sedimentation Problems at the Entrance to Mar Del Plata Harbor. J. Coast. Res. 2016,32, 301–314. [CrossRef] 7. Feola, A.; Lisi, I.; Salmeri, A.; Venti, F.; Pedroncini, A.; Gabellini, M.; Romano, E. Platform of Integrated Tools to Support Environmental Studies and Management of Dredging Activities. J. Environ. Manag. 2016,166, 357–373. [CrossRef] 8. Mahmoodi, A.; Lashteh Neshaei, M.A.; Mansouri, A.; Shafai Bejestan, M. Study of Currentand Wave-Induced Sediment Transport in the Nowshahr Port Entrance Channel by Using Numerical Modeling and Field Measurements. J. Mar. Sci. Eng. 2020 , 8, 284. [CrossRef] 9. Chen, B.; Wang, K. Suspended Sediment Transport in the Offshore near Yangtze Estuary* *Project Supported by the National Natural Science Foundation of China (Grant No.40576017), the National Basic Research Program of China (973, Program, Grant No. 2007CB411804). J. Hydrodyn. Ser. B 2008,20, 373–381. [CrossRef] 10. Zuo, S.; Xie, H.; Ying, X.; Cui, C.; Huang, Y.; Li, H.; Xie, M. Seabed Deposition and Erosion Change and Influence Factors in the Yangshan Deepwater Port over the Years. Acta Oceanol. Sin. 2019,38, 96–106. [CrossRef] 11. Erftemeijer, P.L.A.; Robin Lewis, R.R. Environmental Impacts of Dredging on Seagrasses: A Review. Mar. Pollut. Bull. 2006 ,52, 1553–1572. [CrossRef] 12. Flor, G.; del Busto, J.A.; Blanco, G.F. Morphological and Sedimentary Patterns of Ports of the Asturian Region (NW Spain). J. Coast. Res. 2006,48, 35–40.
J. Mar. Sci. Eng. 2021,9, 267 17 of 18 13. Stock, F.; Knipping, M.; Pint, A.; Ladstätter, S.; Delile, H.; Heiss, A.G.; Laermanns, H.; Mitchell, P.D.; Ployer, R.; Steskal, M.; et al. Human Impact on Holocene Sediment Dynamics in the Eastern Mediterranean—the Example of the Roman Harbour of Ephesus. Earth Surf. Process. Landf. 2016,41, 980–996. [CrossRef] 14. Sharaan, M.; Ibrahim, M.G.; Iskander, M.; Masria, A.; Nadaoka, K. Analysis of Sedimentation at the Fishing Harbor Entrance: Case Study of El-Burullus, Egypt. J. Coast. Conserv. 2018,22, 1143–1156. [CrossRef] 15. Poursanidis, D.; Traganos, D.; Reinartz, P.; Chrysoulakis, N. On the Use of Sentinel-2 for Coastal Habitat Mapping and SatelliteDerived Bathymetry Estimation Using Downscaled Coastal Aerosol Band. Int. J. Appl. Earth Obs. Geoinf. 2019 ,80, 58–70. [CrossRef] 16. Vittori, G.; Blondeaux, P.; Mazzuoli, M.; Simeonov, J.; Calantoni, J. Sediment Transport under Oscillatory Flows. Int. J. Multiph. Flow 2020,133, 103454. [CrossRef] 17. Finn, J.R.; Li, M.; Apte, S.V. Particle Based Modelling and Simulation of Natural Sand Dynamics in the Wave Bottom Boundary Layer. J. Fluid Mech. 2016,796, 340–385. [CrossRef] 18. Finn, J.R.; Li, M. Regimes of Sediment-Turbulence Interaction and Guidelines for Simulating the Multiphase Bottom Boundary Layer. Int. J. Multiph. Flow 2016,85, 278–283. [CrossRef] 19. Kidanemariam, A.G.; Uhlmann, M. Direct Numerical Simulation of Pattern Formation in Subaqueous Sediment. J. Fluid Mech. 2014,750, 1–13. [CrossRef] 20. Leont’yev, I.O.; Akivis, T.M. Modeling of Coastal Dynamics of the Anapa Bay-Bar. Oceanology 2020,60, 279–285. [CrossRef] 21. Armanini, A.; Cavedon, V.; Righetti, M. A Probabilistic/Deterministic Approach for the Prediction of the Sediment Transport Rate. Adv. Water Resour. 2015,81, 10–18. [CrossRef] 22. Flener, C.; Lotsari, E.; Alho, P.; Käyhkö, J. Comparison of Empirical and Theoretical Remote Sensing Based Bathymetry Models in River Environments. River Res. Appl. 2012,28, 118–133. [CrossRef] 23. Giardino, C.; Bresciani, M.; Matta, E.; Brando, V.E. Imaging Spectrometry of Inland Water Quality in Italy Using MIVIS: An Overview . In Advances in Watershed Science and Assessment; Younos, T., Parece, T.E., Eds.; Springer International Publishing: Cham, Switzerland, 2015; Volume 33, pp. 61–83. ISBN 978-3-319-14211-1. 24. Jawak, S.D.; Vadlamani, S.S.; Luis, A.J. A Synoptic Review on Deriving Bathymetry Information Using Remote Sensing Technologies: Models, Methods and Comparisons. Adv. Remote Sens. 2015,4, 147–162. [CrossRef] 25. Hedley, J.; Roelfsema, C.; Chollett, I.; Harborne, A.; Heron, S.; Weeks, S.; Skirving, W.; Strong, A.; Eakin, C.; Christensen, T.; et al. Remote Sensing of Coral Reefs for Monitoring and Management: A Review. Remote Sens. 2016,8, 118. [CrossRef] 26. Lyzenga, D.R. Passive Remote Sensing Techniques for Mapping Water Depth and Bottom Features. Appl. Opt. 1978 ,17, 379. [CrossRef] 27. Cheng, N.-S.; Chiew, Y.-M. Pickup Probability for Sediment Entrainment. J. Hydraul. Eng. 1998,124, 232–235. [CrossRef] 28. Liu, S.; Gao, Y.; Zheng, W.; Li, X. Performance of Two Neural Network Models in Bathymetry. Remote Sens. Lett. 2015 ,6, 321–330. [CrossRef] 29. El-Mewafi, M.; Salah, M.; Fawzi, B. Assessment of Optical Satellite Images for Bathymetry Estimation in Shallow Areas Using Artificial Neural Network Model. Am. J. Geogr. Inf. Syst. 2018,7, 99–106. 30. Obelcz, J.; Wood, W.T.; Phrampus, B.J.; Lee, T.R. Machine Learning Augmented Time-Lapse Bathymetric Surveys: A Case Study from the Mississippi River Delta Front. Geophys. Res. Lett. 2020,47, e2020GL087857. [CrossRef] 31. Tonion, F.; Pirotti, F.; Faina, G.; Paltrinieri, D. A Machine Learning Approach to Multispectral Satellite Derived Bathymetry. ISPRS Ann. Photogramm. Remote Sens. Spat. Inf. Sci. 2020,3, 565–570. [CrossRef] 32. Mavraeidopoulos, A.K.; Oikonomou, E.; Palikaris, A.; Poulos, S. A Hybrid Bio-Optical Transformation for Satellite Bathymetry Modeling Using Sentinel-2 Imagery. Remote Sens. 2019,11, 2746. [CrossRef] 33. Sagawa, T.; Yamashita, Y.; Okumura, T.; Yamanokuchi, T. Satellite Derived Bathymetry Using Machine Learning and MultiTemporal Satellite Images. Remote Sens. 2019,11, 1155. [CrossRef] 34. Manessa, M.D.M.; Kanno, A.; Sekine, M.; Haidar, M.; Yamamoto, K.; Imai, T.; Higuchi, T. Satellite-Derived Bathymetry Using Random Forest Algorithm and Worldview-2 Imagery. Geoplanning J. Geomat. Plan. 2016,3, 117–126. [CrossRef] 35. Kogut, T.; Weistock, M. Classifying Airborne Bathymetry Data Using the Random Forest Algorithm. Remote Sens. Lett. 2019,10, 874–882. [CrossRef] 36. Yunus, A.P.; Dou, J.; Song, X.; Avtar, R. Improved Bathymetric Mapping of Coastal and Lake Environments Using Sentinel-2 and Landsat-8 Images. Sensors 2019,19, 2788. [CrossRef] [PubMed] 37. Bures, L.; Sychova, P.; Maca, P.; Roub, R.; Marval, S. River Bathymetry Model Based on Floodplain Topography. Water 2019 ,11, 1287. [CrossRef] 38. Setiawan, K.T.; Suwargana, N.; Ginting, D.N.B.; Manessa, M.D.M.; Anggraini, N.; Adawiah, S.W.; Julzarika, A.; Surahman, S.; Rosid, S.; Supardjo, A.H. Bathymetry extraction from spot 7 satellite imagery using random forest methods. Int. J. Remote Sens. Earth Sci. IJReSES 2019,16, 23–30. [CrossRef] 39. Moeinkhah, A.; Shakiba, A.; Azarakhsh, Z. Assessment of Regression and Classification Methods Using Remote Sensing Technology for Detection of Coastal Depth (Case Study of Bushehr Port and Kharg Island). J. Indian Soc. Remote Sens. 2019 ,47, 1019–1029. [CrossRef]
J. Mar. Sci. Eng. 2021,9, 267 18 of 18 40. Ha, N.T.; Manley-Harris, M.; Pham, T.D.; Hawes, I. A Comparative Assessment of Ensemble-Based Machine Learning and Maximum Likelihood Methods for Mapping Seagrass Using Sentinel-2 Imagery in Tauranga Harbor, New Zealand. Remote Sens. 2020,12, 355. [CrossRef] 41. Misra, A.; Ramakrishnan, B. Assessment of Coastal Geomorphological Changes Using Multi-Temporal Satellite-Derived Bathymetry. Cont. Shelf Res. 2020,207, 104213. [CrossRef] 42. Huber, M.E.; Zigic, S.; Gilbert, R.; Smith, D.; Edison, K.; Goudkamp, K.; Langtry, S.; Burling, M. Improved Dredge Material Management for the Great Barrier Reef Region. In Proceedings of the Australasian Port and Harbour Conference; Engineers Australia: Barton, Australia, 2013; p. 400. 43. Lara, J.L.; Lucio, D.; Tomas, A.; Di Paolo, B.; Losada, I.J. High-Resolution Time-Dependent Probabilistic Assessment of the Hydraulic Performance for Historic Coastal Structures: Application to Luarca Breakwater. Philos. Trans. R. Soc. Math. Phys. Eng. Sci. 2019,377, 20190016. [CrossRef] 44. Nalona (Suction Dredger) Registered in—Vessel Details, Current Position and Voyage Information—IMO 9047453|AIS Marine Traffic. Available online: https://www.marinetraffic.com (accessed on 18 September 2019). 45. Sentinel Application Platform. ESA Toolboxes, 2009. SNAP. Available online: http://step.esa.int/main/toolboxes/snap (accessed on 12 January 2020). 46. Lanaras, C.; Bioucas-Dias, J.; Galliani, S.; Baltsavias, E.; Schindler, K. Super-Resolution of Sentinel-2 Images: Learning a Globally Applicable Deep Neural Network. ISPRS J. Photogramm. Remote Sens. 2018,146, 305–319. [CrossRef] 47. Rumora, L.; Miler, M.; Medak, D. Impact of Various Atmospheric Corrections on Sentinel-2 Land Cover Classification Accuracy Using Machine Learning Classifiers. ISPRS Int. J. Geo-Inf. 2020,9, 277. [CrossRef] 48. Son, N.-T.; Chen, C.-F.; Chen, C.-R.; Guo, H.-Y. Classification of Multitemporal Sentinel-2 Data for Field-Level Monitoring of Rice Cropping Practices in Taiwan. Adv. Space Res. 2020,65, 1910–1921. [CrossRef] 49. Alwhaely, U.; Hussein, M.A.; AL-Kaaby, L.F. Using GIS and Remote Sensing Satellite Data to Mapping and Monitoring Shatt Al-Arab Estuary (out Bar Area) and Nearby Coastline Southern Iraq. Al-Qadisiyah J. Pure Sci. 2020,25, 1–21. [CrossRef] 50. Breiman, L. Random Forests. Mach. Learn. 2001,45, 5–32. [CrossRef] 51. Reiss, H.; Cunze, S.; König, K.; Neumann, H.; Kröncke, I. Species Distribution Modelling of Marine Benthos: A North Sea Case Study. Mar. Ecol. Prog. Ser. 2011,442, 71–86. [CrossRef] 52. James, G.; Witten, D.; Hastie, T.; Tibshirani, R. An Introduction to Statistical Learning; Springer: Cham, Switzerland, 2013; Volume 112. 53. Prasad, A.M.; Iverson, L.R.; Liaw, A. Newer Classification and Regression Tree Techniques: Bagging and Random Forests for Ecological Prediction. Ecosystems 2006,9, 181–199. [CrossRef] 54. Peters, J.; Baets, B.D.; Verhoest, N.E.C.; Samson, R.; Degroeve, S.; Becker, P.D.; Huybrechts, W. Random Forests as a Tool for Ecohydrological Distribution Modelling. Ecol. Model. 2007,207, 304–318. [CrossRef] 55. Kuhn, M. Caret: Classification and Regression Training. R Package Version. Available online: https://www.R-project.org (accessed on 5 February 2021). 56. Mateo-Pérez, V.; Corral-Bobadilla, M.; Ortega-Fernández, F.; Vergara-González, E.P. Port Bathymetry Mapping Using Support Vector Machine Technique and Sentinel-2 Satellite Imagery. Remote Sens. 2020,12, 2069. [CrossRef] 57. Confederación Hidrográfica del Cantábrico. Available online: https://www.chcantabrico.es/las-cuencas-cantabricas/marcofisico/hidrologia/rios/negro (accessed on 12 November 2019).