*4.1. Synthesis: Virtual Approach Opportinities and Limitations*

The four examples given above illustrate the application of the Google Earth Engine to the identification and even preliminary description of coastal megaclast deposits. The approach seems to be promising. In all four cases, the deposits were identified and delineated without any difficulty and irrespective to the different resolution of satellite images. All study areas were new to megaclast research, and these were found on the basis of the only virtual exploration of the coastline, without consideration of any preliminary evidence. This approach can be used for visual surveying of the coasts and identification of localities suitable for subsequent field or virtual investigations (or both). Moreover, some basic characteristics of coastal megaclast deposits can be examined with the Google Earth Engine taken alone, although such a study cannot 'replace' in-depth field investigations to be undertaken further. These characteristics include the dominant particle size, package density, and spatial homogeneity versus heterogeneity of deposits. Individual, group, or lenticular occurrence of megaclasts can be also registered. As for the particle size, in all cases, it was not only possible to make a distinction of blocks from megablocks, but also to note the presence of smaller and larger blocks, i.e., to deal with subcategories of the main grain-size divisions. Some notions on the possible origin of

the deposits are also possible. Importantly, the use of the Google Earth Engine permitted us a very quick survey of coasts. Apparently, one researcher can check up to a thousand kilometers of coastal zone for the presence of megaclast accumulations per 5–7 working hours. No field investigation can allow such a speed of work and so massive information gathering. Moreover, many areas will remain 'untouched' by megaclast research due to the relatively circle of the involved researchers and logistic restrictions. If so, it is better to know the position of the localities and to have the only preliminary characteristics of the deposits than to concentrate on only well-known and well-accessible localities studied for years.

As implied by the examples, virtual studies of coastal megaclast deposits with the Google Earth Engine face limitations of three kinds. First, the resolution of satellite images did not permit us to deal with boulders and other finer components of the deposits, and the presence of these components could not be proven in some cases. Second, the resolution of the satellite images available for the Hondarribia and Ponza Island localities allowed the size and shape documentation of all megaclasts, whereas the resolution of the satellite images of the Wetar Island and Humboldt o Coredo Bay localities permitted doing this for only huge blocks and megablocks, i.e., for the only part of the deposit. In the two latter cases, there were also difficulties with distinction between true megaclasts, semi-detached megaclasts, and exposures of parent rocks. This means that detailed investigations of coastal megaclast deposits and their quantitative descriptions with the Google Earth Engine are possible in only some cases. Third, there are local limitations of the approach like tall cliff shadows (the Ponza Island) or dense vegetation cover (Wetar Island and Humboldt o Coredo Bay). Anyway, all these limitations do not preclude from the documentation of the spatial occurrence of megaclast deposits along coasts of seas and oceans. When necessary, field studies can be undertaken later, and these studies can be coordinated effectively with the information obtained with the Google Earth Engine. In other words, the limitations are less important than the opportunities. Moreover, it is expected that the resolution of satellite images employed by the noted software will continue to increase in the nearest future.

#### *4.2. Tourism Information as a Source for Geographical Justification of Virtual Surveys*

The total world coastline is too lengthy taking into account continental and all island coasts, and even the Google Earth Engine does not make efficient its visual surveying for the purposes of the identification and description of megaclast deposits. The latter occurs locally, and it is sensible to focus on those areas, which seem to be promising for finding megaclasts. Some previous descriptions ('occasional' descriptions of megaclast deposits can be found in the literature) and geological, geomorphological, and geographical knowledge (for instance, the selection of coasts with tall cliffs and prone to severe storms, super-typhoons, and tsunamis) is helpful in many cases for the geographical justification of virtual surveys. However, some other, non-scientific evidence can be also considered.

The first opportunity is linked to photos provided by the users (first of all, tourists) of the Google Earth Engine, the geographical attachment of which is displayed directly on satellite images. Megaclasts and their accumulations are notable features, and these can be recognized easily on photos. Moreover, the latter can help in satellite image interpretation, i.e., for the correct description of the particle shape and package density, as well as for the distinction between true megaclasts, semi-detached megaclasts, and parent rock exposures; the content of boulders and finer components of deposits can be also registered. In the present study, the photos provided by the Google Earth Engine were helpful in the cases of the Hondarribia and Ponza Island localities. However, the number of the proper photos was limited for some remote, infrequently visited places, or these were not available at all.

The second opportunity that can be recommended for the geographical inventory of coastal boulder deposits is also relevant to tourism-linked information available online. Among various sport and tourism activities, bouldering has gained importance [101–105]. A lot of information on individual 'boulders' (these may be either isolated rocks, but also true megaclasts) has been accumulated. This information is easily accessible with the available on-line catalogues [106–108], and it permits finding coastal megaclasts and their groups, as well as their preliminary visual examination with the available images. For instance, there are typical coastal boulder deposits on the Ao Sane Beach in Phuket, Thailand [106]. These deposits consist of mixed angular blocks and boulders, and there are also angular blocks lying separately on a sandy substrate. In regard to the position of the later, resedimentation cannot be excluded. Such information can be of the utmost importance for finding promising areas to focus virtual surveys of coastal megaclast deposits.

#### *4.3. Do Satellite Images Provide Data for the Quantitative Analysis of Coastal Megaclast Deposits?*

Finding new localities of coastal megaclast deposits with the Google Earth Engine poses a question of more detailed analysis providing sedimentological information. The examples given above imply that some descriptions of such deposits are possible, but these are preliminary and too qualitative. Actually, in some cases, quantitative interpretations seem to be possible. For instance, grain-size distribution of megaclasts can be analyzed with statistical tools. There are four conditions that allow such an investigation. First, the resolution of satellite images of the study area should allow unequivocal recognition of particles with the size of 1–2 m (minimal size of megaclasts). Second, there should be not any 'masking' effects of shadows, vegetation, etc. Third, the majority of megaclasts should be oriented subhorizontally because vertical orientation does not permit the measurement of the maximum size. Fourth, megaclasts should form a relatively thin layer' alternatively, the lowermost particles covered by the other blocks cannot be measured.

From the four examples considered in this study, the only Hondarribia locality satisfied all four conditions outlined above. To demonstrate the efficacy of satellite image analysis, the western plot of the locality (Figure 3) was considered. The Google Earth Engine provides the option of the measurement of objects directly on a satellite image. The maximum size of 100 megaclasts was measured with this tool on the chosen plot. Apparently, the measured particles constituted about 70% of all megaclasts visible on this plot. The size of megaclasts varied between 1.0 and 26.2 m. The mean size was 7.8 m, and the median size was 6.3 m. These values corresponded to coarse blocks, according to the classification of Bruno and Ruban [16]. The grain-size distribution of megaclasts is shown on a histogram (Figure 10). The majority of particles had the size of 2.5–5 m, i.e., these were medium blocks. However, coarse blocks (5–10 m) and fine megablocks (10–25 m) were the most common. Generally, these coastal boulder deposits included 37% of coarse blocks, 26% of medium blocks, 24% of fine megablocks, 12% of fine blocks, and 1% of medium megablocks, i.e., it was dominated by the particles with the size of 2.5–25 m, which indicates on a restricted sorting. Undoubtedly, this deposit also bore some finer components like boulders, gravels, and even sand, but these were invisible on the satellite image. Anyway, the dominance of the noted megaclasts in these deposits was indisputable.

**Figure 10.** Grain-size distribution of megaclasts on the western plot of the Hondarribia locality.

### *4.4. Satellite Evidence of a Decade-Long Stability of Coastal Megaclast Deposits*

The modern studies of megaclast focus much (even over-emphasize) on their origin and, particularly, their relevance to extreme events such as storms and tsunamis [4–13,19–27]. Evidently, these studies deal with the dynamics of the coastline and megaclasts themselves. The Google Earth Engine gives exceptional opportunity to contribute to such investigations, as a time series of satellite images is available. This means that direct comparison of the views of some areas at different time slices is possible. Three limitations are as follows. First, 'old' satellite images are not available for all areas and the date of these images differs for different areas. Second, the resolution of 'old' images can be lower than that of the current images. Third, due to a different time of image making, shadows, vegetation, and, importantly, water level may look differently. However, the third problem can be overcome in many cases.

In order to understand the importance of the analysis of the image time series for the understanding of the dynamics of coastal boulder deposits, the new localities identified in the present work were taken as examples. 'Old' satellite images (captured 10–20 years ago) of appropriate resolution are available for three of them (Hondarribia, Ponza Island, and Wetar Island—one plot was selected for analysis in each case). However, in the only case of the Ponza Island, the maximum appropriate resolution of the 'old' image was the same as that of the 'new' image. In two other cases (Hondarribia and Wetar Island), the resolution of the 'old' image was lower, but it allowed recognition of the principal features, i.e., megaclasts, via comparison to the higher-resolution 'new' image. Megaclasts from the western plot of the Hondarribia locality occupied the same position in 2001 (Figure 11) as in 2018 when the currently available image was captured (Figure 3). Apparently, the number of blocks and megablocks did not change, although some of them had worse visibility on the 'old' image because of the higher sea level (apparently, the time of tide). No changes were also found between 2007 (Figure 12) and 2017 (Figure 5) on the northwestern plot of the Ponza Island. Moreover, the 'old' image was of better quality in this case. The shadow from a tall cliff did not 'mask' a part of the deposits, and the better seawater transparency permitted us to document that up to a quarter of the deposits stretch to the submarine environment. Finally, nothing related to megaclasts changed on the eastern plot of the Wetar Island between 2009 (Figure 13) and 2019 (Figure 6).

**Figure 11.** Satellite view of the western plot of the Hondarribia locality in 2001 (the view provided by the Google Earth Engine).

**Figure 12.** Satellite view of the northwestern plot of the Ponza Island locality in 2009 (the view provided by the Google Earth Engine).

**Figure 13.** Satellite view of the eastern plot of the Wetar Island locality in 2007 (the view provided by the Google Earth Engine).

The evidence presented above implies a decade-long stability of the coastal megaclast deposits in the new localities. Megaclasts were not destroyed, replaced, or over-turned despite that the regions are susceptible to severe weather conditions [62], and tsunamis cannot be excluded there [109,110]. Even if the periodicity of the high-energy events that can be responsible for changes of coastal megaclast deposits seems to be longer than a decade, the registered stability is notable. This is especially the case of the Hondarribia locality located on a high-energy coast of the Biscay Bay, for which a two-decade long comparison (Figures 3 and 11) is possible. Apparently, severe storms of the 2000s and the 2010s [62] did not affect megaclasts of this locality.

It is sensible to add that the Google Earth Engine gives opportunity to get images with a high frequency (Sentinel 2 or SPOT 7 images). If so, some localities can be specially monitored in the case of any future severe storms or tsunamis to document the dynamics of coastal megaclast deposits.

#### *4.5. Automatic Detection of Coastal Megaclast Deposits: the Current State of the Problem*

The development of the Google Earth Engine has permitted to pose the question of the invention of advanced digital tools for automatic detection of particular features on satellite images. For instance, such tools are helpful in urban studies [111] and archaeology [112]. In geomorphological studies, two successfully tested approaches are notable. The first has been proposed by Luijendijk et al. [113]

for the global-scale mapping of sandy beaches and establishing their dynamics. The second approach has been employed by Vos et al. [114] to document shoreline dynamics (some examples also deal with sandy beaches). The validity and the importance of the both above-mentioned studies are indisputable, and the availability of any similar approach would help in the global mapping of coastal megaclast deposits. It is not the purpose of the present work to develop such an advanced interpretative technique, but some limitations of the latter can be discussed in the light of the present findings.

First of all, the main differences of coastal megaclast deposits from sandy beaches should be noted. These deposits are more localized and often narrow; these rarely form lengthy 'belts' like sandy beaches. The latter are often distinguished on satellite images by a white or yellow color, whereas coastal megaclast deposits may look very differently in each case. Importantly, these deposits are often not continuous particle packages, but individual stones or stone groups dispersed along the coastline (this is especially well visible at the Weater Island locality—Figures 6 and 7). To make a clear distinction between true megaclasts and semi-detached megaclasts or rock exposures is not always easy, especially when the surface of a given area is not flat. When sandy beaches are detected, the objects of study are beaches themselves, and the researchers do not to pay attention to sand particles that cannot be determined from the space. In contract, studies of coastal megaclast deposits require attention to giant particles. That is why such studies depend stronger on image resolution. The experience with the Ponza Island (Figures 5 and 12) implies water transparency influences on the detection of megaclasts that are fully or partly drowned, whereas the evidence from the Hondarribia locality (Figures 3 and 11) indicates on the importance of the sea level, i.e., tides. As a result, visibility of coastal megaclast deposits decreases under certain conditions (Figure 14).

**Figure 14.** Dependence of coastal megaclast deposit visibility on satellite images on the sea-level position.

The above-said does not suggest against the necessity of techniques for automated detection of coastal megaclast deposits, but reveals serious barriers for the invention of such techniques. Moreover, it should be stressed that the studies of Luijendijk et al. [113] are aimed at making global- or regional-scale conclusions. In contrast, what do the megaclast researchers actually need are particular localities, especially providing representative examples of coastal megaclast deposits from different parts of the world. In regard to this, simple visual surveys of shorelines with the Google Earth Engine provide reasonable information.

#### **5. Conclusions**

The use of the Google Earth Engine has permitted to us find four new localities of coastal megaclast deposits in Atlantic Europe, the Mediterranean, Southeast Asia, and Central America. The dominance of blocks and subordinate number of megablocks, as well as their chiefly angular shape are visible on satellite images. More detailed, quantitative investigations are possible on the basis of the both field and virtual studies of these localities. Despite some limitations, the Google Earth Engine seems to be an almost ideal instrument for quick tracing the geographical distribution of megaclast deposits along the coasts of seas and oceans, which is important for the coordination of further research. The growth of tourism and voluminous online representation of the relevant photos facilitate finding areas promising for identification of megaclast localities. The novelty of this study is triple, i.e., the potential of virtual search for coastal megaclast deposits was proven, four new localities found, and the knowledge of megaclast spatial occurrence in several large regions was summarized (it was also explained how geographical gaps in this knowledge could be filled with virtual finding of new localities). More generally, this work did not argue for 'replacing' field studies by virtual surveys. In contrast, it demonstrated how to find efficiently new localities for further field studies. In the other words, finding localities as a research task is separated from in-depth sedimentological investigations.

This paper dealt with four, almost randomly selected localities. Further virtual surveys of the coastal zones will permit to extend the relevant knowledge, as well as to improve the methods of the satellite image analysis. One of the central problems is the precise megaclast size and shape description, as satellite images show the only 2D projection of these features, which can be oriented differently. It is important problem to think how some tools proposed earlier [115–117] can be coupled with the use of the Google Earth Engine in coastal megaclast studies. It cannot be excluded that satellite images can be used for automatic detection of megaclast-promising areas in the future.

**Author Contributions:** D.A.R. is the only author of this work, and he is fully responsible for its content. The Author has read and agree to the published version of the manuscript.

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

**Acknowledgments:** The author gratefully thanks M.E. Johnson (USA) for his kind invitation to contribute to this special issue and various support, as well as all reviewers for their helpful suggestions.

**Conflicts of Interest:** The author declares no conflict of interest.

#### **References**


© 2020 by the author. 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 (http://creativecommons.org/licenses/by/4.0/).

MDPI St. Alban-Anlage 66 4052 Basel Switzerland Tel. +41 61 683 77 34 Fax +41 61 302 89 18 www.mdpi.com

*Journal of Marine Science and Engineering* Editorial Office E-mail: jmse@mdpi.com www.mdpi.com/journal/jmse

MDPI St. Alban-Anlage 66 4052 Basel Switzerland

Tel: +41 61 683 77 34 Fax: +41 61 302 89 18

www.mdpi.com

ISBN 978-3-0365-2579-2