**About the Editors**

#### **Juan J. Munoz-Perez**

Dr. Juan J. Munoz-Perez has worked in private companies both in consulting and public ˜ works, as well as in the ports of Barcelona and Cadiz. As Chief Engineer of the Atlantic Coastal Department of Andalusia (Spain) has planned, supervised, and directed nearly 400 maritime works. He has combined the former tasks with teaching as Professor at the University of Cadiz since 1991 for the grades of Sea Sciences and Civil Engineering. He is the author of numerous contributions in congresses and articles in prestigious journals as a result of the projects he has directed through the Coastal Engineering research group that he runs (scholar.google.es/citations?user=xyE3rnoAAAAJ&hl=es). He currently manages the Campus of International Excellence of the Sea or CEIMAR (https://campusdelmar.com/ )

#### **Luis Moreno**

Dr. Luis J. Moreno has more than 30 years of professional experience in the field of Coastal and Oceanographic Engineering consultancy in a variety of roles, not only in the public sector but also in the private sector. Has successfully managed a wide range of projects in the Middle East, Europe, Latin America, and Africa.

Dr. Moreno is a recognized professional who regularly participates in technical conferences and meetings, as well as in national and international postgraduate and master's courses. He served as a national delegate at the Committee of Advisory Nature/Marine Science and Technology of the European Union and has been a project evaluator of R+D+i projects for the European Union.

As a part-time Associate Professor at Technical University of Madrid (UPM) since 2003, he has educated hundreds of students in harbor and coastal engineering as well as in port operation and infrastructure management.

### *Editorial* **Beach Nourishment: A 21st Century Review**

**Luis J. Moreno 1,\* and Juan J. Muñoz-Perez 2,\***


Long-term erosion is experienced by most of the coastlines worldwide, and it is usually attributed not only to sea level rise but also to the retention of sand in dams, the occupation of dry beaches by urbanized areas, the disturbance of the natural patterns of longshore drift, the mining of sand as building material for construction, and so on. Beach nourishment has evolved as the favored erosion-mitigation strategy in many areas of the world. The increasing number of people living on the coast, the safety of those people, and the high values of coastal property [1] are all factors that have made beach nourishment a cost-effective strategy for managing erosion in many locations. However, a new scenario of sand scarcity and environmental care has arisen in recent decades [2]. There have been a number of different and interesting cases of various aspects of beach nourishment in the last years. The purpose of this Special Issue has been to publish the different experiences and research related to this topic.

After a careful review process, nine papers were included. Their thematic contributions include the use of field methods such as the use of remotely piloted aircraft systems (RPAS) or un-manned aerial vehicles (UAV) for faster and automated mapping of the coastal area or the acquisition of geomagnetic data in marine environments; the use of multi-approach methodologies to assess the interaction between coastal structures and beaches and in particular of submerged pipelines; the need to adopt a plan for the optimal use of limited resources of available sediment from a regional perspective and the assessment of the effectiveness of beach nourishments; the understanding of the role of submerged geological control of beach profiles together with the implementation of innovative beach nourishment strategies while facing the non-trivial challenge of visualizing and communicating mesoscale modeling assumptions, uncertainties and outcomes to both coastal specialists and decision makers; and the influence of sea-level rise and erosion on diminution of beach habitats.

The contributions are commented upon in order of appearance in this Special Issue. Although an effort has been made to compile contributions that cover an update in the state-of-the-art of innovative techniques in beach nourishment, by no means should they be limited to the topics presented hereby.

To begin with, the size and great dynamism of coastal systems require faster and more automated mapping methods, such as the use of a remotely piloted aircraft system (RPAS) or unmanned aerial vehicle (UAV). However, the main problem for surveying using lowaltitude digital photogrammetry in beach areas is their visual homogeneity. Obviously, the fewer the homologous points defined by the software, the lower the accuracy. Contreras-de-Villar et al. [3] have addressed the error performed in photogrammetric techniques, such as flight height, flight time, percentage of frame overlap (side and forward), and the number of ground control points (GCPs). Among their conclusions, it should be highlighted that the error for noon flights is almost double that for early morning flights. Moreover, a minimum value of 7 GCP per hectare should be taken into account when designing a beach leveling campaign using RPAS.

**Citation:** Moreno, L.J.; Muñoz-Perez, J.J. Beach Nourishment: A 21st Century Review. *J. Mar. Sci. Eng.* **2021**, *9*, 499. https://doi.org/ 10.3390/jmse9050499

Received: 21 April 2021 Accepted: 29 April 2021 Published: 5 May 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/).

Coastal areas are usually very impacted because of demographic and industrial pressure, which leads to an interaction between anthropic infrastructures and littoral dynamics. One of the public works that can most influence the sediment transport is a submerged pipeline. Lersundi-Kanpistegi et al. [4] studied the extension of the wastewater pipeline in Vigo (Northwest Spain) crossing the most important urban beach of the city by using a multi-tool strategy based on high resolution bathymetry data, seabed physical characterization, a grain size study of the superficial sediment, and a numerical simulation of the tide, wave climate, and sediment transport in low and high energy conditions using the open source Delft3D software. The results indicate characteristics that the design must follow in order to ensure that the future structure would not alter the global sediment dynamics of the beach. The multi-approach methodology presented can be applied to other studies of the interaction between coastal structures and beaches.

Beach nourishment is generally seen as the preferred means of rectifying coastal erosion, due to its low environmental impact and natural evolution. Martell et al. [5] present a study regarding the effectiveness of beach nourishments in Cancun (Mexico), but its conclusions regarding the erosion tendency directly linked to the incidence of extreme hydrodynamic conditions and the scarcity of natural sediment sources can be applied to beaches with similar characteristics in any other area. Furthermore, the need for improving long term predictions of the wave climate under global warming scenarios must be highlighted.

Submerged geological control of beach profiles, e.g., through the existence of reef flats or submerged sills, is a topic that has been widely studied over the last years. Moreover, fringing reefs have significant impacts on beach dynamics, yet there is little research on how they should be considered in beach nourishment design, monitoring, and conservation work. Thus, the behavior and characteristics of nourishment projects at two reef protected beaches, in Hawaii (USA) and in Cadiz (Spain), are compared [6] to provide transferable information for future nourishment projects and monitoring in this type of environment. Several differences were detected related to the nourishment cost, distance to the borrow site, post-nourishment monitoring frequency and assessment of accuracy, measurement of the beach volume increase after nourishment, etc.

Innovative beach nourishment strategies have been developed in the last decade, driven by the increased worldwide interest in environmentally friendly coastal protection measures. In this context, the massive nourishment project of the Netherlands (known as Sand Engine [7]) has been hailed as a successful means of beach protection. Adapting this idea, a very small and bell-shaped Sand Engine was designed to protect the beachfront at a tourist resort near Puerto Morelos, Mexico [8]. This micro Sand Engine is seen as a sustainable and eco-friendly coastal protection measure, especially applicable when large nourishment projects are not viable. Maintenance work for this type of nourishment is cost- and time-effective, and any negative impacts on sensitive ecosystems nearby can be detected and controlled quickly.

Coastal geomorphologists and engineers worldwide are increasingly facing the nontrivial challenge of visualizing and communicating mesoscale modeling assumptions, uncertainties and outcomes to both coastal specialists and decision makers. Payo et al. [9] show how the risk of simulation model outcomes can be minimized by using the Coastal Modeling Environment (CoastalME). CoastalME is a modeling framework for coastal mesoscale morphological modeling that can achieve close linkages between the scientific model abstractions and the 3D representation of topographic and bathymetric surfaces. A transparent methodology to merge the required variety of data types and formats into a 3D-thickness model is presented through the case study of Happisburgh (eastern England, UK). Finally, some of the barriers to the adoption of this methodology are analyzed.

Sometimes, the limited resources of available sediment make it necessary to adopt a plan for their optimal use from a regional perspective. This is the case presented by Pranzini et al. [10], who present a study carried out to support the Region of Tuscany Coastal Sediment Management Plan along the 215 km-long continental sandy coast of this Italian region. Sand stability and color compatibility were determined in order to assess the possibility of using the available sediment in accreting sectors to nourish the beach in eroding areas. This kind of study is of great interest for the proposal of sound management actions to counteract the increasing erosion processes linked to climate change phenomena and human effects on rivers and coastal systems.

A method for the acquisition of geomagnetic data in marine environments, developed by the Oceanographic and Hydrographic Research Center of Colombia, is presented by Oviedo et al. [11]. Leaving sub-bottom profiling and side-scan sonar techniques aside, the most representative uses of the geomagnetic method are the location of pipelines and metal plates, detection of buried ordnance, identification of sites of archaeological interest, and the characterization of geological structures. To test the method, a grid of geomagnetic data was surveyed in an area close to the island of San Andrés (Northwest Colombian maritime territory) and compared with survey data obtained from National Oceanic and Atmospheric Administration (NOAA) magnetic data. Despite the long time interval between the two surveys, almost 50 years, no significant differences were observed in terms of the analyzed variables.

Finally, the influence of sea-level rise and erosion (along with shoreline hardening and reduced sediment inputs) on diminution of beach habitats is shown by Martin et al. [12]. Their study shows that increasing sandy beach habitat can be beneficial to wildlife, but the method of placement, timing of the project, and fate of the beach afterward can modulate or prevent beneficial effects. Frequent repetition of sand placement may accumulate impacts without allowing sufficient time for the ecosystem to recover.

Closing this editorial, the guest editors consider that this Special Issue will provide benefits to technicians, engineers, researchers and managers in the area of beach nourishment.

**Author Contributions:** L.J.M. and J.J.M.-P. wrote and reviewed this editorial, and both authors have agreed to the published version of the manuscript. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Acknowledgments:** The Guest Editors wish to express their deepest appreciation to the authors for their cooperation, to the anonymous reviewers for their valuable comments, and to the Editor-in-Chief for the valuable comments, as well as to Esme Wang for her guidance, continuous support, and constructive advice throughout the publication process.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


## *Article* **Beach Leveling Using a Remotely Piloted Aircraft System (RPAS): Problems and Solutions**

**Francisco Contreras-de-Villar 1,\*, Francisco J. García 2 , Juan J. Muñoz-Perez <sup>1</sup> , Antonio Contreras-de-Villar <sup>1</sup> , Veronica Ruiz-Ortiz <sup>1</sup> , Patricia Lopez <sup>1</sup> , Santiago Garcia-López <sup>3</sup> and Bismarck Jigena <sup>4</sup>**


**Abstract:** The size and great dynamism of coastal systems require faster and more automated mapping methods like the use of a remotely piloted aircraft system (RPAS) or unmanned aerial vehicle (UAV). This method allows for shorter intervals between surveys. The main problem for surveying using low-altitude digital photogrammetry in beach areas is their visual homogeneity. Obviously, the fewer the homologous points defined by the program, the lower the accuracy. Moreover, some factors influence the error performed in photogrammetric techniques, such as flight height, flight time, percentage of frame overlap (side and forward), and the number of ground control points (GCPs). A total of 72 different cases were conducted varying these factors, and the results were analyzed. Among the conclusions, it should be highlighted that the error for noon flights is almost double that for the early morning flights. Secondly, there is no appreciable difference regarding the side overlap. But, on the other side, RMSE increased to three times (from 0.05 to 0.15 m) when forward overlap decreased from 85% to 70%. Moreover, relative accuracy is 0.05% of the flying height which means a significant increase in error (66%) between flights performed at 60 and 100 m height). Furthermore, the median of the error for noon flights (0.12 m) is almost double that for the early morning flights (0.07 m) because of the higher percentage of grids with data for early flights. Therefore, beach levelings must never be performed at noon when carried out by RPAS. Eventually, a new parameter has been considered: the relationship between the number of GCPs and the surface to be monitored. A minimum value of 7 GCP/Ha should be taken into account when designing a beach leveling campaign using RPAS.

**Keywords:** UAV; RPAS; littoral systems; aerial photogrammetry; DTM; monitoring; SfM; GCPs

#### **1. Introduction**

Coastal erosion has become one of the most important concerns of different countries [1]. Coastal areas are the focal points of tourist attractions, which translates into an important source of economic income [2–4]. Moreover, the study of coastal behavior helps us understand the complex processes that occur in these areas [5,6]. Their understanding leads us to the prevention of coastal erosion, and thus monitoring the evolution of our beaches is essential [6]. Thus, a methodology for carrying out measurements of some oceanographic phenomena using Unmanned Aerial Vehicles (UAV also known as remotely piloted aircraft system or RPAS) have already been presented by other researchers (e.g., [7]). Nevertheless, some aspects can still be taken into account as we will show later.

Correct coastal modeling needs a three-dimensional reconstruction of the study area [8]. Coastal modeling is represented by digital terrain models (DTMs) of high spatial

**Citation:** Contreras-de-Villar, F.; García, F.J.; Muñoz-Perez, J.J.; Contreras-de-Villar, A.; Ruiz-Ortiz, V.; Lopez, P.; Garcia-López, S.; Jigena, B. Beach Leveling Using a Remotely Piloted Aircraft System (RPAS): Problems and Solutions. *J. Mar. Sci. Eng.* **2021**, *9*, 19. https://doi.org/ 10.3390/jmse9010019

Received: 23 November 2020 Accepted: 21 December 2020 Published: 26 December 2020

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2020 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/).

resolution. Geomorphological state can be defined as a multitemporal surface [9,10]. Depending on the beach area to be mapped (dry zone, intertidal zone, or submerged zone), various techniques and methodologies can be used [11,12]. The dry beach and the intertidal zone have been mapped using direct topography techniques [12]. Initially, a tachymeter was used, being later replaced by the total station, an electronic transit theodolite integrated with electronic distance measurement (EDM), and an on-board computer to collect data and perform triangulation calculations. This task is currently done with GPS techniques. This type of point-to-point data collection is cheaper than the previous ones because it only requires one technician. However, GPS surveying is limited to small and easily accessible areas.

The size and great dynamism of these coastal systems require faster and more automated mapping methods [13]. Thus, the synchronous nature of the data is not lost [14]. Photogrammetry has evolved to the technique called structure from motion (SfM) [13–17] based on algorithms that allows one to obtain excellent cartographic results from a set of frames that cover an area. The emergence of RPAS systems as well as the high definition of today's digital cameras have induced new cartographic systems. The use of these systems significantly reduces costs and execution times, providing excellent accuracy [18]. This technology, tested in multiple applications, appears as a serious competitor against other cartographic techniques (e.g., LIght Detection and Ranging or LIDAR) [19]. Hugengoltz et al. [20], for instance, stated that the vertical RMSE of an RPAS data set was equivalent to the RMSE of a bare earth LiDAR DTM for the same site.

The work procedure involves the definition of a series of parameters such as flight height (Figure 1a), covering area, percentage of overlap between adjacent frames (Figure 1b,c), and a different number of ground control points (GCPs) [21] (Figure 1d).

**Figure 1.** Sketch showing basic concepts of the remotely piloted aircraft system (RPAS) system: frame and flight height (**a**), overlap between two adjacent frames (**b**,**c**) and distribution of the different number of ground control points (GCPs) (**d**).

A particular case of the problems of low-altitude digital photogrammetry is the identification of common points in contiguous frames over poorly differentiated visual areas. When performing a low-altitude flight over highly homogeneous surfaces (beach sand, snow, agricultural areas of the same crop) is difficult to find common points [22]. The fewer the homologous points defined by the program, the lower the accuracy. This fact is common in the photogrammetry of beach areas and the accuracy of the DTM will be the result of the concatenation of the errors in different stages [23].

Thus, this paper aims to compare the vertical accuracy of a beach leveling, by using an RPAS, performed with different parameters of flight (height, time, side and forward overlap) and number of GCPs. Eventually, some guidelines will be presented to minimize the error of a photogrammetric survey.

#### **2. Study Area**

The chosen beach is Los Lances beach, in Tarifa (SW Spain). This area is considered as a bird special protection area due to its privileged situation that gives it a relevant role in air and marine migration processes (Figure 2).

**Figure 2.** Location of the study area.

This space has a good state of conservation of ecosystems and a high-quality landscape. The beach is 3854 m long and covers an area of 280,000 m<sup>2</sup> . The study area, a 178 m by 84 m rectangle, is also shown in Figure 2.

Its degree of urbanization is low. It has fine golden-colored sand, composed of medium-coarse unconsolidated sediments. The D<sup>50</sup> of the emerged sand is 0.34 mm. It is a dissipative beach and has waves of medium-moderate degree [24]. The maximum tidal range is about 1.40 m, and the significant wave height Hs is about 3.7 m [21]. It is a type of semi-urban beach widely used by windsurfers and kitesurfers due to the abundant windy days of the year that occur in this area. The prevailing winds in the area are eastwards and westwards [25]. The dry beach area before the dune area is over 100 m wide, which is an optimal area for conducting the study.

#### **3. Methods**

The factors that influence the error performed in digital photogrammetric techniques are flight height, overlap (side and forward), and GCP number. Different tests were conducted varying these factors, and the results were analyzed. The main problem for surveying using photogrammetric methods in beach areas is their visual homogeneity. This effect reduces the number of homologous points among neighboring frames and thus prevents optimal correlation.

In normal conditions, it is recommended to make the flights in hours close to noon since the sun is in the highest position, generating few shadows [26]. However, in addition to the noon flights, other flights were carried out early in the morning when the sun was at a low altitude. These supplementary flights were performed trying to find out if the shadows produce enough differentiation in the terrain as to decrease the margin of error (Figure 3).

**Figure 3.** Frame detail: (**a**) early morning; (**b**) at noon. Frame details showing how shadows produce more homologous points in the early morning than at noon.

One difficulty is the short flight time of RPAS. Thus, the usual flight time (20–25 min) must be balanced against each other parameter: the surface to be flown, the flight height, and the overlap in the images we want to obtain.

The parameters that varied on each flight were the following:


#### *3.1. Data Collection*

As we previously specified, the area chosen for the study is 178 m alongshore and 84 m cross-shore, and therefore its surface area is about 15,000 m<sup>2</sup> , although the overflight area was obviously taken of a larger surface.

Data collection was planned with Phantom 4 Pro, based on the following three stages: (a) Distribution on the beach of prefabricated landmarks to improve precision and calibration of the camera. Though some authors [27] state that direct georeferencing with high camera location accuracy and GNSS receivers can limit the need for GCPs, these landmarks were used as GCPs and georeferenced. The GCPs were plastic, measuring 24 × 30 cm and about 5 mm high (Figure 4), and had a hollow that helped fix their position in the sand. The reverse was painted, creating an alternating white-and-black grid. The positions were chosen to try to get an optimal placement according to the literature (covering all four corners of the site, the highest and lowest elevations, and with sufficient cross-shore and alongshore coverage). Moreover, their location was maintained during the two flight campaigns so that the results were not affected by any movement.

**Figure 4.** Dimensions of the GCP used and an aerial view of one of them already placed on the beach.

(b) Performing a topographic survey of the area using direct topography with GPS in RTK (Real-Time Kinematic) mode that provides precision around 3 cm. To ensure this accuracy, each topographic reading was repeated by taking three consecutive shots, which were checked and validated only if their difference was less than 1 cm in planimetry and 2 cm in altimetry. Moreover, the GPS rod man distinguished all the locations where pronounced changes in the beach topography appeared, due to his/her training and experience. Therefore, the density of GPS points was increased in these areas. A total of 657 survey points were taken, with an average distance among points of five meters approximately. This density of points is very high for the characteristics of the terrain.

The aims were twofold: first, calculating a topographic surface to compare with the photogrammetric data and, second, determining the coordinates of the GCPs for the RPAS postprocessing. The points were defined in European coordinates UTM ETRS89, and the levelings referred to the Spanish Datum (mean sea level in Alicante) by using the EGM2008 geoid provided by the National Geographic Institute [28].

(c) Introduction of the flight parameters into the RPAS software and realization of the photogrammetric flights. Six flights were made on the same day. In this way, the weather conditions and the situation of the terrain would be the same and therefore would not influence the results of the study. Flight planning requires the establishment of the limits of the area to be photographed, camera characteristics, flight height, flight direction, and frame overlap in side and forward directions. Thus, three flights were made at 8 am (when the sun angle is still low), with flight heights of 60, 80, and 100 m. The flights were taken with a side and forward overlap of 85%. However, also 75% of overlap in both directions was considered afterward by using the software. There were three more flights at noon (when the sun is at its highest position), repeating the same operation.

Therefore, six flights were made, but a total of 72 cases were studied (Table 1) by combining the parameters of two times of the day with different heights of flight (3), different side and forward overlaps and different number of GCPs (5, 7, and 10).


**Table 1.** Values of the different flight parameters and number of studied cases.

Flight mission planning was previously done. For this, the Pix4D Capture program was used. This program uses the aerial images of Google Earth® as a base on which the area to be flown is defined, the flight height and the side and forward coverings were described, and the flight course was marked to optimize the times. The program calculated the flight speed and shooting interval among photographs. Figure 5 shows the flight plan scheme.

The camera technical data are presented in Table 2.


**Table 2.** Data of camera.

Number of frames and duration of flight are shown in Table 3.


**Table 3.** Flight data, frame characteristics, frame number, and duration of flight.

The second column displays the Ground Sampling Distance (GSD), which is directly related to the flight altitude and camera parameters. The GSD is defined as the distance between two consecutive pixel centers measured on the ground. The greater the GSD value, the lower the spatial resolution of the image, and the less visible the details [29].

**Figure 5.** Image of the flight plan indicating the area of interest and frame overlap. The overflight area is bigger than the study area.

#### *3.2. Method of Obtaining DTM by Photogrammetry and DTM Checking*

The methodology for obtaining the DTM is based on the structure from motion (SfM) algorithm. The software used is Agisoft-Metashape Professional Educational ® . Unstructured aerial images using fast, inexpensive, and highly automated image processing produces three-dimensional information. This RPAS-SfM pairing gives good results in cartographic production [27,30,31]

Firstly, once a set of frames was loaded into the software, an approximate orientation of the frames, based on the EXIF data of each photograph, was performed. EXIF is short for exchangeable image file, a format that is a standard for storing interchange information in digital photography image files using JPEG compression. It relied mainly on the focal length of the camera used, the time of taking the picture, and GPS coordinates.

Once the complete block was ordered, the program searched for tie points among adjacent frames. At this point, we could define the degree of precision that we require, as well as the key points and maximum tie points to be used in each frame to perform the operation.

The result of this process was a global point cloud that collected all the tie points of the flight frameset. At this time, the program had already created a three-dimensional point cloud. These point clouds were adjusted, georeferenced, and corrected for the lens distortion by using the GCPs. This procedure required entering the coordinates (X, Y, Z) of the GCPs and identifying them graphically in each of the frames in which they appeared. Since the GCPs points were defined in coordinates in the UTM-ETRS89 system, the adjusted point cloud would be in that same system.

At this point, we had the points of the topographic survey performed by GPS on the ground and the 72 DTMs obtained by processing the former set of point clouds. To facilitate and simplify the statistical reading of the DTMs, the size of each basic element of the DTM (tile size) was defined as a square of 1 m on the side. The Z value of each tile was defined as the average of the specific values it contained. Once the DTM was obtained, we cut it to the area of interest. By forming the DTM with all the points and cutting it later, we avoided the loss of data and extrapolation in the boundary areas.

These DTMs have been widely used, and much research on their error and uncertainty has already been investigated [14]. The quality of these models depends on several factors, such as the method used to attain the altimetric data, the density of the starting data, the resolution of the mesh, or the interpolation algorithm used, among others.

#### *3.3. Calculation of the Error*

To check the final quality of each flight, the error of each of the 72 DTMs (generated from the cloud of points obtained with the RPAS) was calculated by comparing to the DTM defined from the topographic data taken with GPS in RTK as the reference (Equation (1)). All the DTMs had the same dimensions, and a grid size of 1 × 1 m was chosen to facilitate comparison. Moreover, the percentage of grids that contained at least one datum was calculated, and its value was used as another reliability parameter.

The result of the comparison is another DTM whose characteristic is the difference between the altitudes of the flight DTM and the topographic DTM (GPS on the ground), that is, the vertical error (ε) in every grid.

$$
\varepsilon = Z\_{flight} - Z\_{ground} \tag{1}
$$

Figure 6 shows an example of this error (ε) calculated as the difference between the altitudes of the flight DTM and the topographic DTM (GPS on the ground)

**Figure 6.** Map of the vertical error in every grid(ε). Example of the difference between the altitudes of the flight digital terrain models (DTM) and the topographic DTM (GPS on the ground).

Given the higher precision in the horizontal plane (approximately twice that in the vertical plane) and the very gentle slope of the beach profile (<2%), we will assume that the influence of the possible location error of a point on the vertical precision is negligible.

However, this average of the vertical errors suffers from that positives and negatives can cancel each other out and give a false sense of accuracy. That is the reason why another statistic, the RMSE (Equation (2)) was calculated.

The National Standard for Spatial Data Accuracy (NSSDA) is a recent standard proposed by the Federal Geographic Data Committee (1998) [32] and can be used for both analog and digital cartographic data [33]. This standard assumes a normal distribution of ε and uses the root-mean-square error (RMSE) as the most common and valid statistic for the evaluation of products obtained by photogrammetry and remote sensing.

$$RMSE\_Z = \sqrt{\frac{1}{n} \sum\_{i=1}^{n} \left( Z\_{f\text{light}} - Z\_{\text{ground}} \right)^2} \tag{2}$$

The 95% confidence interval (Equation (3)) for the vertical accuracy reached in each of the grids was determined according to the NSSDA as

$$P\_{Z,95\%} = 1.96 \cdot RMSE\_Z \tag{3}$$

Thus, Equation (4) shows the range of values that do not exceed the established accuracy.

$$\left\{ \begin{array}{l} \overline{\mathfrak{x}} + 1.96 \cdot RMSE\_{\overline{Z}}\\ \overline{\mathfrak{x}} - 1.96 \cdot RMSE\_{\overline{Z}} \end{array} \right\} \tag{4}$$

#### **4. Results and Discussion**

#### *4.1. Error for Each of the 72 Cases*

As previously mentioned, the average of the vertical errors (ε in Equation (1)) results in a number not too helpful because positives and negatives can cancel each other out. That is the reason why another statistic, the RMSE (Equation (2)) was calculated. From these data, the vertical accuracy (Equation (3)) for each of the cases was determined. The results of these two values for each of the 72 cases are shown in Table 4. Moreover, another error parameter defined in the methodology is also presented in Table 4, the percentage of non-empty grids (1 × 1 m<sup>2</sup> ), i.e., with at least one homogeneous point inside.

#### *4.2. Influence of Number of GCPs*

The first variable to consider is the number of GCPs. A box-and-whisker plot of their RMSE error is shown in Figure 7. Note that a boxplot is a standardized way of displaying the dataset based on a five-number summary: the minimum, the maximum, the sample median, and the first and third quartiles. You can see from the graph that there is a large distance between the lower (25%) and upper quartiles (75%) (IQR-Interquartile range), which are 0.10, 0.08, and 0.03 m for 5, 7, and 10 GCPs, respectively. Note that the whiskers (the two lines outside the box that extend to the highest and lowest observations) are similar in the three cases. The high value for the 10 GCP case is due to the existence of outliers for the noon survey. The variation of these results is far away from the results presented by other authors as James et al. [34] whose RMSE had a negligible deviation because of the number of GCPs, obtaining 3.12, 3.57, and 3.59 cm for 5, 10, and 15 GCPs, respectively.

*J. Mar. Sci. Eng.* **2021**,*9*, 19



**Figure 7.** Box-and-whisker plot of the average error based on the GCP number.

Thus, though other authors such as Zimmerman et al. [30] found that (7 to 9) wellplaced GCPs in the optimal configuration produced the same magnitude of error as using more (15) poorly placed GCPs, the only acceptable values in our case are those collected by using 10 GCPs, with an IQR less than 3 cm.

Moreover, 10 GCPs is just the maximum number of points for this particular case. To generalize this value and its use in any other case, the number of GCPs has been divided by the surface in hectares (Ha), a unit frequently used in topographic surveys. Thus, if we divide 10 GCPs by 1.5 Ha (15,000 m<sup>2</sup> ), we get a rounded value for the density of GCPs (7 GCP/Ha), a new starting parameter when designing a beach leveling campaign using RPAS. Regrettably, a limitation of this study is that we did not check whether the accuracy might even increase more by using more than 10 GCP, and, therefore, the trend of the inclusion of more GCPs remains unknown.

#### *4.3. Influence of Flight Time*

As previously noted, visual homogeneity of beach areas is one of the main problems for surveying using photogrammetric methods because of the reduction of homologous points among adjacent frames. Therefore, two different times for the flights were chosen (8 a.m. and 12 a.m.) to find out if shadows in the early morning (Figure 8) produce more homologous points than at noon and, thus, a decrease in the error committed. The number of tie points are presented in Table 5.

Based on the results of the previous subsection, the values obtained for 5 and 7 GCPs were discarded to avoid distorting the statistical results of the rest of the variables. Figure 7 shows the values of vertical RMSE for both aforementioned times of the flight and all the flight heights

As can be seen in Figure 8 the IQR ranges from 5 to 14 cm with a median of 0.09 m at noon while there are just a mean of 3.5 cm, no outliers, and a negligible dispersion in the early morning. Moreover, the percentage of grids with data is almost 60% higher for early flights than for noon flights. Therefore, it can be established that RPAS for beach leveling must be performed early in the morning. Or, in other words, RPAS surveys must be banned at noon.

**Figure 8.** Box-and-whisker plot of RMSE vs. the flight time (8 a.m. and 12 a.m.) for all the flight heights.


#### *4.4. Influence of Frame Overlap and Flight Height*

Analyzing the percentage (70% vs. 85%) of side and forward overlap, four different cases were considered. Moreover, flights were carried out at three different heights (60, 80, and 100 m). The results presented in Table 4 are now shown in Figure 9, where only the 10 GCP experiments have been taken into account.

Starting with the noon flights (dashed lines), it can be seen that there is not too much difference between the results of 60 and 80 m flight heights. Regarding the forward overlap, there is no appreciable difference between both (70 and 85) percentages, i.e., forward overlap change from 85 to 70 did not influence final results. However, RMSE decreased enormously (from 0.15 to 0.05 m) when the side or transverse overlap changed from 70% to 85%.

On the other side, when results from 8 am flights are analyzed, RMSE remained constant and, therefore, independent from both side and forward overlap percentages. Furthermore, there is a small but still significant difference for RMSE as a function of the flight height. RMSE was 3, 4, and 5 cm for 60, 80, and 100 m heights, respectively. Following Gonçalves and Henriques [35], a relative accuracy of the flying height can be calculated. This relative accuracy was 0.35‰ in their case (from 0.046 m to 131 m flying height). A numerical value very similar to the results presented here where relative accuracy was 0.5‰ for each of the flights performed early in the morning but lesser than the values found for the flights performed at noon which can reach up to 1.5‰.

**Figure 9.** RMSE vs. different side (S) and forward (F) overlap for different flight heights and flight times (dashed lines are used for noon flights while solid lines are for 8 a.m. flights). Note that the number of GCPs is not a variable because only experiments performed with 10 GCPs were considered.

Thus, in brief, it can be established again that early morning flights minimize vertical error. Moreover, side overlap should not be less than 85% while forward overlap percentage is not a decisive factor. Finally, the decision about the flight height (when designing an RPAS for a beach leveling) must take into account that variation of vertical RMSE, though small in absolute value (5 cm vs. 3 cm), can be relatively significant (about 66%) when a 100 m height is chosen instead of a 60 m height.

#### **5. Conclusions**

A common fact in the photogrammetry of beaches (poorly differentiated visual areas) is the difficulty in the identification of common points in contiguous frames [22]. And, obviously, the fewer the homologous points defined by the program, the lower the accuracy. Thus, the main objective of this work is to determine the parameters of flight (height, time, frame overlap) and number of GCPs to optimize the accuracy of photogrammetric surveys when using RPAS in cases of visually homogeneous areas.

The following variables have been taken into account: flight height (60, 80, and 100 m), flight time (8 a.m. and 12 p.m.), side and forward overlap (70% vs. 85%), and the number of ground control points or GCPs. The combination of these variables results in 72 cases.

Firstly, one of the main conclusions is related to the density of GCPs. A minimum value of 7 GCPs/Ha has been found for this new parameter when designing a beach leveling campaign using RPAS. However, the trend of the inclusion of more GCPs remains unknown. This aspect is pending for future research.

Secondly, there is no appreciable difference regarding the forward overlap. But, on the other side, RMSE increased to three times (from 0.05 to 0.15 m) when side overlap decreased from 85% to 70%.

Moreover, the median of the error for noon flights (7 cm) is double that for the early morning flights (3.5 cm) because of the higher (almost 60%) percentage of grids with data for early flights. Therefore, beach levelings must never be performed at noon when carried out by RPAS.

Finally, there is a significant difference (till 66%) for RMSE as a function of the flight height. RMSE was 3, 4, and 5 cm for 60, 80, and 100 m heights, respectively, when only results from the 8 a.m. flights are analyzed. Furthermore, in this case, RMSE remains constant, and therefore independent, for the different side and forward overlap percentages. **Author Contributions:** Conceptualisation, F.C.-d.-V.; F.J.G. and J.J.M.-P.; investigation and writing original draft preparation, F.C.-d.-V.; F.J.G.; J.J.M.-P.; A.C.-d.-V.; V.R.-O.; P.L.; S.G.-L.; B.J.; review and editing, F.C.-d.-V.; J.J.M.-P. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was partially funded by Fundacion Campus Tecnologico de Algeciras. And the APC was funded by the Coastal Engineering Research group (University of Cadiz).

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Data sharing not applicable.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


## *Technical Note* **A New Method for the Collection of Marine Geomagnetic Information: Survey Application in the Colombian Caribbean**

**Karem Oviedo Prada 1,2,\*, Bismarck Jigena Antelo 1,\* , Nathalia Otálora Murillo <sup>2</sup> , Jeanette Romero Cózar <sup>1</sup> , Francisco Contreras-de-Villar <sup>1</sup> and Juan José Muñoz-Pérez 1,\***

juanjose.munoz@uca.es (J.J.M.-P.)


**Abstract:** In recent years, the Oceanographic and Hydrographic Research Center (part of the General Maritime Directorate of Colombia (DIMAR) has made important efforts to advance research in the field of marine geophysics, in particular, the techniques of geomagnetism, sub-bottom profiling, and side-scan sonar, the first being the most developed at the present time. A method is presented for the acquisition of geomagnetic data in marine environments, as used by DIMAR in the Colombian maritime territory. The development of the geomagnetic method not only offers the opportunity to advance basic scientific knowledge, but it is also of great importance in support of national sovereignty issues. Among other applications, the most representative uses of the geomagnetic method are the location of pipelines and metal plates, detection of buried ordnance, identification of sites of archaeological interest, and the identification and characterization of geological structures. As a result of testing the method, a grid of geomagnetic data was surveyed in an area close to the Island of San Andrés in the north-west of the Colombian maritime territory. The survey was prepared with a regional geometric arrangement, the result of which was compared with survey data obtained from the National Oceanic and Atmospheric Administration (NOAA) magnetic data repository and carried out in the same study area. Despite the long time interval between the two surveys, almost 50 years, no significant differences were observed in terms of the analyzed variables. Finally, results show negligible differences between the magnetic data obtained for the years 1970 and 2018 for all the variables measured, such as the inclination, declination, and total magnetic field. These differences may be attributable to a geological component or also to the acquisition and processing methods used in the 1970s.

**Keywords:** marine geophysics; magnetic method; Colombian Caribbean; DIMAR; CIOH

#### **1. Introduction**

The increase in marine geophysical activity in recent years has provided essential data for evaluating theories about the origin of oceans and continents. Of the different methods used to explore the sea floor and underlying mantle, the magnetic field and its measurements have proven to be one of the most powerful tools for discovering and delineating structural and geological patterns [1].

According to Ewing et al. [2], before World War II, almost all marine magnetic observations had been made by the research ship "Carnegie" (1909–1929), which was specially built to work along widely spaced lines in the Atlantic, Pacific and Indian oceans. After the war, the fluxgate magnetometer, originally developed as an airborne instrument for detecting submarines, was adapted for marine applications by the Lamont Geological Research Observatory. These were the first measurements made with a magnetometer towed by a ship. Later, for work in the maritime field, the fluxgate magnetometer was replaced by

**Citation:** Oviedo Prada, K.; Jigena Antelo, B.; Otálora Murillo, N.; Romero Cózar, J.; Contreras-de-Villar, F.; Muñoz-Pérez, J.J. A New Method for the Collection of Marine Geomagnetic Information: Survey Application in the Colombian Caribbean. *J. Mar. Sci. Eng.* **2021**, *9*, 10. https://dx.doi.org/10.3390/jmse901 0010

Received: 20 November 2020 Accepted: 16 December 2020 Published: 23 December 2020

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2020 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/).

the proton magnetometer, having the advantages of absolute field measurement and not requiring orientation of the head [3].

The use of the geomagnetic method is widely known globally, for its various local and regional applications [4]. Due to its high costs, which involve equipment and logistic development and multiple applications, this geophysical method is generally undertaken by foreign private companies with commercial aims, as the high costs prove to be a disadvantage for state organizations dedicated to science and marine research, which have limited budgetary allocations [5]. Therefore, it can be summarized that established foreign companies, which compose the entire market, dominate geophysical exploration, including those services in limited supply, such as gravimetry and magnetometry and its applications [6–8].

The General Maritime Directorate (DIMAR) is located at Cartagena de Indias, in the Colombian Caribbean Sea. DIMAR started the project "Geomagnetismo Marino" in 2015 with the purpose of recovering research capacity through the use of the G-882 marine magnetometer from geometrics. One of the recovery activities included training on the handling of the magnetic sensor and data acquisition. For the former, a document was produced [9] in which a vast database and manuals were compiled, which served as a base for the production of the following geophysical work methodology.

The need to propose a work methodology was pressing, as there was no record in Colombia of any other public entity carrying out this type of scientific research. Therefore, the efforts of the Caribbean Oceanographic and Hydrographic Research Centre (CIOH) were aimed at the standardization of the guidelines and the parameters required for the optimization of the marine geomagnetic method. After much effort and field tests in the Colombian Caribbean, a methodology has been obtained that offers high-quality marine geomagnetic data collection.

Thus, the following method aims to sequentially show the planning and acquisition of geomagnetic information in deep marine environments in Colombian territory on board the oceanographic research vessel ARC (Navy of the Republic of Colombia) Providence. The guide has become a tool that provides an effective and efficient response for geophysical research at the service of the nation. For this work, a bibliographic compilation was carried out taking into account aspects such as the verification of magnetic sensors and operators that can be powerful sources of magnetic noise. [10]. A fundamental aspect in the survey was to determine the distance at which the magnetometer sensor must be towed to reduce the magnetic effects of the vessel. Finally, the optimal lateral spacing between the lines also had to be considered, which is directly related to the depth of the water [11].

Moreover, geomagnetism is a geophysical prospecting method, applicable to the oil industry, and also mining and archaeological artefact explorations [11–13]. In mineral exploration, magnetometry is widely used to directly prospect for magnetic minerals, such as magnetite and other ferromagnetic minerals, and the method stands out for its speed and low cost. This method is the most widely used in geophysical surveys, at local and regional scales, and it is based on the study of the Earth's magnetic field and its variations, as a consequence of additional magnetic fields produced by magnetized rocky bodies positioned on the surface and close subsoil [14,15].

The magnitude measured in the magnetic method is the Geomagnetic Field, which is related to the magnetization of the environment and which, in the majority of materials, appears when a magnetic field is applied to a body [16]. In the magnetic method, the objective is to investigate the geology of the subsoil, from the variations in this geomagnetic field, resulting from the magnetic properties of the underlying rocks [17,18]. Not all the rock-forming minerals are magnetic, but certain types of rock contain sufficient magnetic minerals to be able to produce significant magnetic anomalies, such as iron and magnetite, among others. The influence of the total magnetic field can be measured anywhere on earth, with a certain direction and intensity, subject to periodic variations and non-periodic disturbances, the magnitude of which on the planet's surface can vary from point to point from 25,000 to 65,000 nT [19,20].

When a magnetic material is placed in a magnetic field, the material is magnetized, and the external field of magnetization is reinforced with the induced magnetic field in the material. This is known as induced magnetization, and it is based on the magnetic susceptibility of the materials, (understood as the degree of magnetization of a material in response to a magnetic field), and the magnitude and direction of the magnetic field [21]. When the external field disappears, the induced magnetization disappears immediately, but some materials retain a residual magnetism, and its direction will be fixed in the direction of the inductive field [22]. The residual magnetism reflects the history of the material. Thus, there is a contrast of magnetism between an anomalous source and the adjacent lateral formations. These two types of magnetization are due to spontaneous magnetization, which is a property of the ferromagnetic minerals in the Earth's crust [23].

To calibrate the data, check its reliability, and study the variation of the new data, it was necessary to have a reference work. To do this, we took into account previous work carried out between 1970 and 1971, obtained from the repository of the National Oceanic and Atmospheric Administration (NOAA). More specifically, from the Marine Geology and Geophysics data from the National Centers for Environmental Information (NCEI), formerly the National Geophysical Data Center or NGDC [24].

The objective of this work is to present a method to carry out magnetic surveys that is compatible with other techniques used in different areas of engineering and science (hydrographic surveys, side-scan sonar, search for magnetized bodies, search for archaeological remains, etc.). To evaluate the quality of the work, the results obtained in a recent campaign were compared with those from previous ones, and the differences were analyzed.

#### **2. Materials and Methods**

#### *2.1. Study Area*

The study area is located in the Colombian Caribbean Sea and more specifically to the south of the archipelago of San Andrés, Providencia and Santa Catalina (SAPSC). The geophysical survey was carried out in the area located between San Andrés Island, Cayos de Albuquerque Island and Cayos de Este-Sudeste Island, within the polygon marked in yellow, as can be seen in Figure 1. The study area covers an area of approximately 2040 km<sup>2</sup> .

**Figure 1.** Polygon of acquisition in study area.

A magnetic survey measures the local magnetic field characteristics of a certain region. This type of technology only detects minerals and/or materials that respond to magnetic fields. For this reason, its applications are mainly aimed at mineral exploration, but it can also be useful for the exploration of coal, oil, and gas and in the detection of shipwrecks. [2,6,25,26]. A geophysical survey consists of different phases.

#### *2.2. PHASE 1. Planning of the Acquisition Campaign*

The form of the geophysical survey is established in this phase, and the times, the necessary inputs, and the possible unforeseen events that may occur at sea, are estimated.

Before planning the data acquisition, the study objective and the scale of the work (local or regional) should initially be taken into account. The configuration and length of the lines to be acquired will depend on these.

The generation of the acquisition grid is made based on the sought objectives. It is important to consider whether it is required to determine the regional magnetic field (e.g., changes of magnetic polarity reflected in the marine magnetic anomalies, regional guidelines, etc.), or to determine local geologic anomalies (e.g., geologic bodies and structures), or to identify the anomalies due to metallic objects produced by humans [27]. This is related to the fact that the geometric arrangement of the acquisition must take into account the spatial resolution of the body to be characterized—that is, the smaller the object, the denser and less spaced the survey grid must be. —

It is also important to take into account the sensitivity of the sensors, since, to recognize an anomaly, this must be several times greater than the sensitivity (resolution) of the magnetometer and the external noise level. It is important to define this parameter to know if the object is detectable on the surface and, in such a case, how much the readings in the profile, and the distance between adjacent profiles, would have to be spaced (spacing of the grid). Ideally, a grid should be shaped to cover the whole area in such a way that the anomaly can be always detected by a profile. This means that there must be some overlapping between profiles [9].

Additional magnetic information is required, whether from magnetic observatories or from a Base Station near the survey area, with the purpose of improving the quality of the data. In this case, a Geometrics G-862 RBS Base Station (Figure 2) was used, which was acquired by the General Maritime Directorate in 2015. This was positioned at a minimum radius of 60 m from any source of electromagnetic interference. This reduces the errors that can occur in the data, due to fortuitous cases, such as electromagnetic interference from the solar field.

**Figure 2.** Installation of the Geometrics G-862 RBS Base Station.

Starting the planning activity, it is essential to have high-resolution bathymetric data, in order to support the identification of the geological structure that is required to be recorded with magnetometry [28]. As the objective of the project was to determine the magnetic anomalies, generated by the volcanic bodies and geological structures (faults)

located to the south of the SAPSC, in the vicinity of San Andrés Island, Cayos de Alburquerque Island and Cayos de Este- Sudeste Island (Figure 1). The survey lines were carried out, taking into account the geoforms displayed in the bathymetry. For this reason, the lines are established perpendicular to faults or other structures, in regular meshes, where it is ensured that the separation between lines is equal to the estimated minimum distance between the sensor and the magnetic object or target. This is why it is recommended to take into account the depths at which the survey will be carried out [29].

The area included in the geophysical research polygon, in which it was planned to undertake five main lines of acquisition, with a NW-SE direction (azimuth of 300 ◦ ), with lengths between 70 and 57 km, and a separation of 7 km. The six control lines, oriented perpendicularly to the main lines, are distributed with a spacing of 23.50 km, and they have of an average length of 32 km (Figure 3). In order to calculate the days needed for the survey, the total length of the lines at the optimal survey speed in linear nautical miles was considered, assuming 24 working hours per day [9]. This calculation is shown in Tables 1 and 2.

**Figure 3.** Lines of survey.

**Table 1.** Estimation of duration of the survey in days from linear nautical miles.


Once the configuration of the survey was established, the times for the voyages and duration of the acquisition were estimated. As mentioned previously, it is important to maintain good data density that can adequately represent the objective; that is to say, the optimum survey speed was 5 knots, assuring that the intensity of the signal was stable, and 10 samples per second were obtained.

Finally, a meticulous control was made to ensure that the acquisition of the data was carried out successfully. In the case of consumable equipment, such as RS232-USB converters, it is ideal to have spare parts, in case of unexpected events.

**Table 2.** Estimation of time of the operation, including the displacement.


#### *2.3. PHASE 2. Data Acquisition*

The oceanographic research vessels, ARC Malpelo and ARC Providencia, were enabled to operate with the Geometrics G-882 marine magnetometer [30], property of the DIMAR (Figure 4). This apparatus has a broad range of detection for ferrous materials of various sizes and a sensitivity of <0.004 nT/πHz rms, which increases the probability of detection. It has a hydrodynamic design that helps reduce the probability of rock incrustation, and it operates to a depth of approximately 2750 m, and at temperatures from −35 ◦C to 50 ◦C. The cesium–vapor sensor is at the rear of the "fish" in the cylinder that forms a T with the longest axis, where the direction of the sensor can be modified; this was vertical, as the work was to be carried out in equatorial latitudes. Finally, the sampling interval ranged from one sample every three seconds, to twenty samples per second, with an absolute precision of <2 nT. The acquisition of field data was carried out with MagLog software from Geometrics Inc. πHz rms, which increases the probability of − – sensor is at the rear of the "fish" in the cylinder that forms a T with the

**Figure 4.** Geometrics G-882 marine magnetometer. Source Karem Oviedo Prada, 2020.

#### *2.4. PHASE 3. Office and Data Processing*

In this office phase, the data were analyzed and filtered and subsequently processed. Oasis Montaj version 8.5 software from Geosof was used for this, and ArcGIS version 10.7 software from ESRI was used for charting.

For this work, the data obtained from the NOAA repository of magnetic surveys carried out in the study area between 1970 and 1971 were used as a reference. A hydrographic and bathymetric survey was performed according to technical specifications of the International Hydrographic Organization (IHO), S-44 publication for Order 2 requirements [31,32]. These regulations guarantee the quality and standardization of the results. This geophysical working method is applicable in all the deep waters of the Colombian marine territory, which are considered to be from the isobath of 100 m, to the maximum registered depth of 4600 m. This specification is also stated by Standard S44 of the IHO [31,32], which recommends that Order 2 surveys are limited to areas deeper than 100 m [33]. Nevertheless, those reference data were about 50 years old, and there were no other modern data available for the study area until the current survey. However, nautical charts 1624 and 004 edited by the CIOH in 1998 nd 2018, respectively [34,35], were taken into account to study the temporal and local variation of the geomagnetic field in the area. Thus, the declination, due to the annual variation effect, was corrected and was 4º18 ′ (W) in 2020. Likewise,

′

the magnetic declination was compared with the data published in the AIP COLOMBIA Report of the Gustavo Rojas Pinilla Airport on San Andres Island, for 16 July 2020, which was 02º48 ′ W [36]. The difference in declinations is due to the separation between the San Andres Airport (North of the island) and the area where the magnetic declination is defined in the nautical chart 004, which is located about 90 km NE of the airport. For the analysis of magnetic declination, the procedure specified by Udias and Mezcua [23] was followed.

02º 48′

In this work, the Minimum Curvature Gridding or Splines method and Geographic coordinate system were used. The Gridding method refers to the process of interpolating data onto an equally spaced grid of "cells" in a specific coordinate system. This interpolation method estimates values using a mathematical function that minimizes the curvature of the surface, resulting in a smooth surface that passes exactly through the input points [37–39]. –

#### *2.5. Components*

Going into the field, some indispensable elements must be taken into account to carry out an optimal acquisition. For example, the magnetometry sensor and the portable winch with 300 m of telemetry cable were specifically adapted to collect the geophysical information.

The vessel ARC Providencia (Figure 5) has special adaptations, such as a winch with 2800 m of telemetry cable (Figure 6), a wet laboratory aboard the ship, and the computer center where the magnetic data are visualized and stored in real time.

**Figure 5.** Research vessel ARC Providencia.

**Figure 6.** 2800 m geophysical research winch on board the ARC Providencia.

at a "junction box" where the magnetic data are related to those of the

The assembly used in the vessels is shown in Figure 7. The magnetic data are communicated from the sensor and submerged in the water, passing through the winch, the on-board cable, and finally, arriving at a "junction box" where the magnetic data are related to those of the positioning obtained by the Global Navigation Satellite System [33,40]. From there, they are transmitted to and visualized in the computer by means of MagLog software [41], as shown in Figure 8. at a "junction box" where the magnetic data are related to those of the

**Figure 7.** Assembly for the magnetic data acquisition.


**Figure 8.** Visualization of geomagnetic information in MagLog software in real time.

The acquisition is made with the help of the ship's personnel, tak-

m. It is important that the "fish"

In Figure 8, showing the visualization of data in real time, the red box to the left is the navigation window, where the position of the vessel and its course are shown. In addition, the lines that the helmsman must follow, according to the planning instructions, are indicated in this window [42]. The blue box, to the right, shows the curves of the data expected, or calculated, by the International Geomagnetic Reference Field (IGRF) model in that precise geographic position and the actual collected data. The small green indicators show the intensity of the signal, the magnetic data, and the positioning. All these indicators must be green, so that the data are correctly acquired. The red rectangle, in the extreme lower right, indicates that it is not recording, and it must be pressed to begin to record the data during the acquisition [37].

The acquisition is made with the help of the ship's personnel, taking into account certain guidelines, such as a maximum velocity of 5 knots, and a separation from the sensor of at least three times the length of the vessel, which in this case was 150 m. It is important that the "fish" is towed from the stern, as indicated in Figure 9. The acquisition is made with the help of the ship's personnel, takm. It is important that the "fish"

**Figure 9.** System of stern towing of the magnetometry equipment.

Magnetic recording is only useful in straight transects, whereas data from the turns between profile and profile are not considered, as the recorded values are affected by the magnetic field induced by the boat approaching each time a turn is made [43].

The different stages involved in the acquisition of marine geomagnetic data are subjected to a series of decisions that can radically affect the final result of the research [44]. Several usual errors exist that can be committed throughout the process, and which can be classified, according to the development stage of the study. For example, there are frequent errors related to planning that involve poor design of the lines to acquire, which could make it difficult to discern the exact form and size of the anomaly; a measurement is only of interest if the margin of error of that measurement is known. What is interpreted is a collection of data, which is why the sampling must be in accordance with the dimension of the objective to be reached. Other types of errors are associated with the measuring equipment, which can lead to mistaken readings and affect the quality of the data, operator errors, sampling errors, and errors related to environmental noise, among others. Some examples of error handling in geophysical and gravimetric data processing are shown in [45,46].

#### **3. Results and Discussion**

In the application of the method, some setbacks were presented in terms of what was planned. These were due to logistical issues with the vessel. It is also worth mentioning that the data collection was carried out on board the ship ARC Roncador, with a 300 m portable winch. The geometric arrangement had to be slightly less extensive than originally planned, as shown in Figure 10.

**Figure 10.** Geometric adjustment of the geomagnetic acquisition.

− The geophysical study comprises the data collected between 20 June and 1 July 2018, in an area south of San Andrés Island, comprising four lines perpendicular to the general direction of geological structures, with a maximum length of 70.67 km, and four lines parallel to these formations, with a maximum length of 31 km. A grid-shaped geometric arrangement was preserved to provide good resolution for a regional geological study. The general direction of geological structures and geoforms has a northeast direction [47,48].

The chart of the total field of collected data appears in Figure 11, and it shows the magnetic surface of the collected data, after processing for corrections of diurnal variation, delay, direction in degrees, and of the IGRF mode [37]. A significantly positive anomaly was observed in this area, above the Nutibara Depression. The variations were in a range of −170.48 to +159.37 nT. <sup>−</sup>

**Figure 11.** Geomagnetic surface of the total field with corrections.

In order to compare the obtained data with pre-existing data, a review was made of the bibliographical material and the data available from possible geomagnetic surveys carried out in the area. As a result, two research cruises were identified, giving free access to marine geomagnetic data from the National Oceanic and Atmospheric Administration [24], which are represented in two survey lines related to geophysical data that contain seismic, side-scan sonar, and magnetometry information. The first downloaded file of the zone, identified by code CH100L12, was collected by the Woods Hole Oceanographic Institution of the United States (WHOI) in 1971. The second file of the zone, identified by the code V2808, was collected by the Lamont–Doherty Earth Observatory of the United States between 1970 and 1971. The two compiled data lines appear in Figure 12. –

**Figure 12.** Tracking line chart of oceanographic cruises that acquired geomagnetic data in 1970 and 1971.

' Taking into account that the magnetic field is dynamic and presents significant annual variations, an attempt was made to find geomagnetic information for the area, finding the only free access data to be those previously described, with the possible source of error of an elapsed time of around 50 years between both surveys. More current geomagnetic data on the study area have been found; however, these are global data obtained through NCEIs (National Center Environmental Information) global aeromagnetic project pertaining to the NOAA. They are aeromagnetic data obtained for the study and modeling of the Earth's magnetic field and have a much higher scale of resolution. In addition, their correction processes are different from the data obtained in situ, and specifically from those obtained in the geomagnetic surveys presented in this work. For those reasons, these aeromagnetic data were discarded for comparison, as they were incompatible in terms of resolution and processing [24]. On the other hand, the variations presented by the magnetic data during the half century between the two surveys have also been taken into account in this work. These variations have an important component of anomalies due to geological sources that have persisted in the study area during that time.

The magnetic information downloaded from NOAA [49] has an extension MGD77T and contains the positions of the tracking line and the data with corrections for diurnal variation and the IGRF. To be viewed on a common surface, the two geodatabases were joined and charted to WGS 1984 UTM Zone 17N, corresponding to the projection of the study area [39,50]. With the positional and magnetic intensity data, a magnetic surface was generated (Figure 13) showing positive and negative anomalies that vary from −328.99 to +40.48 nT.

−

**Figure 13.** Geomagnetic surface of the total field, corresponding to the National Oceanic and Atmospheric Administration (NOAA) data.

In order to visualize the surface of Figure 13, a magnetic grid was generated by the Minimum Curvature Method [38], with a cell size corresponding to 1000. In this image, magenta colors are observed that are associated with positive magnetic peaks, which are located toward the northwest and over the Nutibara Depression and the Wayuu Spur; the geoforms are mentioned in Figure 10.

The layouts of the two magnetic grids are shown in Figure 14, where the positive anomalies are similarly identified on the Wayuu stimulus and the Nutibara depression, to the east and northwest, at the low Nicaraguan elevation. As for the negative anomalies, the magnetic bass located on the areas near the island of Cayos de Albuquerque stands out.

**Figure 14.** Geomagnetic surfaces of the total field. (**A**) Surveyed magnetic field. (**B**) Magnetic field from data downloaded from the NOAA.

It is important to mention that the color scales are not associated exactly with the same ranges on the two surfaces, but they are very close, remember that the color blue is always

−

′ ′′ ′ ′′

−

associated with low magnetic and pink with high magnetic [51]. These differences are due to the main factor that it is a time difference between the two surveys of around 50 years, and from which other factors that influence the acquired data can be derived, such as the accuracy of the magnetic sensor, the disposition of the field magnetic model in 1970, which presents variations with respect to that of 2018. This due to the displacement of the field and the geometric arrangement used for each case.

Comparatively speaking, in Figure 14A, the magnetic peaks are within the range of −170.48 to 159.37 nT, while in Figure 14B, the magnetic lows and highs range from −328.98 to 45.08 nT, indicating a variation of 329.85 nT in Figure 14A and 374.06 nT in Figure 14B.

Using the NOAA magnetic calculator [52] and working with the magnetic data observed for the years 1970 and 2018, small but significant differences are identified in all the measurement variables, such as the inclination with 1º46′47′′, the declination with 5º2 ′41′′, and the magnetic field represented by 4415.7 nT. These results show a variation of the magnetic field across the timeline. It is important to highlight that these anomalies identified in the study area are geological in nature, which infers that they will be present as sources of magnetic anomalies for a very long period of time.

The significant magnetic anomalies, which can be seen in Figure 11, marked in fuchsia/magenta, correspond to a magnetic high. In Figure 10, it can be seen that this anomaly corresponds to the geomorphology of the Nutibara depression, and that it could be generated by some type of mineral deposition that can be found, associated with ferrous materials, or also with volcanic material with a high iron content compared to its geological environment. On the other hand, magnetic lows are observed north of the Cayos de Alburquerque Island, which would seem to be a contradiction, since its morphology is typical of a seamount, and, therefore, its magnetic response should be high [48,53,54].

The results of this work are very important for the scientific community, as this is an area where there is an immense lack of data. The data obtained in the survey, carried out in 2017 by the CIOH, 50 years later, are very important data taken in situ, with a high resolution that make them very reliable and precise, and they are unique in the area of the Archipelago of San Andres, Providencia and Santa Catalina.

We have carried out an exhaustive search for data and geomagnetic surveys in the study area and, although no new works have been found, similar or related works have been carried out in the environment of the study area: magnetic mapping of the northern Caribbean region using marine magnetic data from GEODAS [55], works about gravity and magnetic field referring to hydrocarbon prospects at the Tobago Basin [56], geological description and interpretation in Providencia and Santa Catalina Islands [57], and many others related to tectonics and volcanism [58,59].

#### **4. Conclusions**

Surveying for the acquisition of geomagnetic data in marine environments is a method that is gaining ground worldwide, and it offers great opportunities for development and advances in new lines of research and scientific knowledge. In addition, it is a technology that supports different research and engineering projects, such as the detection of the location of pipelines and covers, buried ordnance, shipwrecks, identification of sites of archaeological interest, and the characterization of geological structures, among other applications, and also in projects related to national sovereignty and the study of a country's natural resources.

The methodology for marine geomagnetic acquisition has become the prime standard for marine geophysical research for the study of national resources and Colombian sovereignty.

Although geophysical exploration is dominated by established foreign companies, DIMAR now has the capacity to offer geophysical magnetometry services to different countries within its sphere, with excellent technical and human resources, and research equipment and vessels.

After much effort and field tests in the Colombian Caribbean, the geomagnetic acquisition procedure has been standardized as a methodology that can obtain high-quality marine geomagnetic information.

The acquisition of the G-882 marine magnetometer, the application of this methodology to a survey in the Colombian Caribbean, and the development of the magnetometry method have responded to the need of the DIMAR to recover the capacity for scientific research at national level, and scientific leadership in the region, by having an efficient tool in geological and archaeological prospecting supported by two modern, well-equipped scientific research platforms, namely, the ARC Malpelo and ARC Providencia research vessels.

Small differences have been identified between the magnetic data obtained for the years 1970 and 2018, being negligible in the variables measured, such as the inclination, declination, and total magnetic field. These results show a variation of the magnetic field across the very long timeline, so it can be inferred that these anomalies in the study area have an important geological component and will be present for a long time. These differences may also be attributable to the acquisition and processing methods used in the 1970s.

The results of this work are very important for the scientific community, because this is an area where there is a great lack of magnetic data. The data from the survey carried out in 2017 by the CIOH are very important due to the survey resolution reached, having achieved 91,285 data taken in the field at a rate of 20 data/second with an average ship speed of 7 knots, which managed to obtain a datum every 2 cm (0.02 m/datum).

**Author Contributions:** Conceptualisation, B.J.A.; J.J.M.-P. and K.O.P.; investigation and writing original draft preparation, K.O.P., B.J.A., J.J.M.-P., J.R.C., N.O.M., F.C.-d.-V.; review and editing B.J.A., J.J.M.-P., F.C.-d.-V. All authors have read and agreed to the published version of the manuscript.

**Funding:** The research for the acquisition and capture of geomagnetic data was supported by the Caribbean Oceanographic and Hydrographic Research Center (CIOH), attached to the General Maritime Directorate within the framework of the "Marine Geomagnetism Project". The APC was funded by CIOH and RNM912 Coastal Engineering Research Group of the University of Cadiz.

**Data Availability Statement:** The data presented in this study are available on request from the Centro de Investigaciones Oceanograficas e Hidrograficas de Colombia. The data are not publicly available due to military restrictions.

**Acknowledgments:** This work was possible thanks to the support of the "Marine Geomagnetism" project, financed by the General Maritime Directorate. The authors thank the crew of the ARC Roncador Oceanographic Research Vessel and the staff of the survey area of the Colombian Hydrographic and Oceanographic Research Center (CIOH) for their collaboration during the survey campaigns. We also thank the Captain of the Navy, Germán Augusto Escobar Olaya, Director General of CIOH for his support in the fieldwork and authorisation for the use of CIOH data for the preparation of this document. The authors thank Javier Idárraga García, the editors and the two anonymous reviewers for their comments and suggestions which greatly improved the manuscript.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


*Article*
