Impacts of Storm Surges on a Changing Climate: a Global Bibliometric Analysis of the Coastal Zone

An increase in the global mean sea is predicted during the 21st century as a consequence of global average temperature projections. In addition, changes in the strength of atmospheric cyclonic storms may alter the development of storm surges, exacerbating the risks to coastal communities. Based on the fact that the interest and range of papers are growing on this topic, this study aims to present the global scientic production status of studies that have correlated climate change and the impact of storm surges on the coastal zone leading to erosion and ooding (inundation) via a bibliometric analysis. We analyzed 429 papers published in journals between 1991 and February 2021 from the Scopus database. Through the VOSviewer and Bibliometrix R package, we describe the most relevant countries, aliations, journals, authors, and keywords. Our results demonstrate that there has been an exponential growth in the research topic, and that authors from the United States and the United Kingdom are the most prolic. Among the 1454 authors found, 10 researchers published at least 5 papers on the topic and obtained at least 453 citations in the period. The most represented journals were the Journal of Coastal Research, Climatic Change, and Natural Hazards. We also found, and discuss, the lack of standardization in the choice of keywords, of which climate change, storm surge, and sea level rise are the most frequent. Finally, we have written a guide to facilitate the authors' bibliographic review.


Introduction
Global predictions project an increase of 4ºC to 6ºC in the global average temperature by 2100 on the most extreme greenhouse gas representative concentration pathway (Collins et al. 2013). This may lead to a hydrological cycle acceleration, consequently increasing the intensity and frequency of extreme events, such as coastal storms and storm surges (Coco and Ciavola 2017; Oppenheimer et al. 2019). In this extreme scenario (RCP8.5), it is estimated that global mean sea level (GMSL) may increase by approximately 1.10 m in the 21st century (Oppenheimer et al. 2019). A recent study conducted by Nicholls et al. (2021) shows that, in the last two decades, the GMSL has risen 2.5 mm yr − 1 and that the impact in subsiding coastal areas is four times faster with an average relative sea-level rise varying from 7.8 mm to 9.9 mm yr − 1 . Thus, as a result, we know that coastal systems will increasingly ood and experience erosion during the 21st century, if adaptation or mitigation measures to sea level rise and extreme events are not taken (Nicholls 2002 A fundamental aspect of coastal storms that distinguishes them from other extreme events is their genetic process, which is driven by atmospheric disturbances over the open ocean. Tropical cyclones and extra-tropical cyclones are the main synoptic systems responsible for the vast majority of coastal storms worldwide. Depending on several factors, related to both the cyclonic system as well as the coastal setting in which it occurs, these systems may also generate storm surges (Harley 2017). Storm surges refer to the abnormal rise in seawater level caused by low atmospheric pressure and the force exerted on the sea surface by strong winds. By de nition, it is measured as the height of the water above the normal predicted astronomical tide ( During a changing climate, changes in the number, path, and strength of atmospheric cyclonic storms may alter the formation and development of storm surges (Lowe and Gregory 2005). The amplitude and impact at any given location depend on the intensity, size, and speed of the storm, coastline orientation, and the local topobathymetric characteristics ( Leal et al. 2020;NOAA 2021). In addition, hazards and disasters associated with this abnormal rise in seawater level, are highly correlated with storm frequency and characteristics (Harley 2017; Lin et al. 2019). Severe coastal ooding usually happens when wind-induced waves and storm surges coincide with high tides (Resio and Westerink 2008;Kumbier et al. 2017; Chen et al. 2021). In addition, these storms are the main driver for coastal ooding, and responsible for extensive coastal erosion (Resio and Westerink 2008;Leal et al. 2020). Their impacts can be different and more destructive on densely urbanized coasts, in terms of human and economic losses (Neumann et al. 2015). Although the impacts of sea-level rise and, consequently, of storm surges, are potentially strong, the application and success of adaptation are large uncertainties that require more assessment and consideration (Nicholls and Cazenave 2010). Therefore, the scienti c community's interest in studies on those impacts and possible adaptations to climate change is growing.
Based on the fact that the range of published papers is growing exponentially on this topic, we see a bibliometric analysis as an important tool, which can facilitate and expand the capacity of researchers in a given subject. For instance, in some cases, the bibliographic review process can take a long time, since it is essential to select relevant papers for the scienti c community, thus the results of the bibliometric study can be a quick solution. In addition to the accelerated production, control, and dissemination of information, the development of computer programs for bibliometrics and the creation of databases facilitates the gathering of data in the same place, streamlining information processing and access to new papers. Moreover, synthesizing past research ndings is one of the most important tasks for advancing a particular research line. Quantitative analysis for given scienti c research, through bibliometric studies, determines the scienti c knowledge derived from publications and represents the current research trends, through which the direction of science and related institutions is de ned. Besides, bibliometrics can connect published papers, authors, or journals; identify research substreams; and produce published research maps (Davyt and Velho 2000; Zupic and Čater 2015).
Thus, this study aims to present the global scienti c production status of studies that have correlated climate change and the impact of storm surges on the coastal zone leading to erosion and ooding (inundation). Through a bibliometric analysis, we resolve the following fundamental topics: 1) the most relevant countries, a liations, authors, and journals; 2) the most cited papers; 3) direct scientists to the research area in their bibliographic reviews. To the best of our knowledge, the present study combines, for the rst time, climate change and the impacts of storm surges in the coastal zones to conduct a bibliometric analysis.

Page 4/30
The compilation between systematic literature review and bibliometric analysis provides a rigorous and formal methodical procedure, which seeks to minimize bias and possible errors when selecting studies to characterize an area of knowledge (Denyer and  . Four main steps were followed to address the research objective (Fig. 1).
In the rst step, researchers need to de ne their research aim (Zupic and Čater 2015). The main intention of the bibliometric analysis was to present the global scienti c production status of studies that have correlated climate change and the impact of storm surges on the coastal zone leading to erosion and ooding (inundation), as well as direct scientists to the research area in their bibliographic reviews.
A bibliographic database was selected and compiled during the second step. First of all, considering the three main databases (Google Scholar, Web of Science, and Scopus), we found out that Scopus is the most relevant database, indexing 25,751 journal titles, and it also has the largest number of publications related to the topic, outperforming the other databases. Therefore, the review was based on the Scopus database. According to Lima and Bonetti (2020), the selection of a single database is the actual best alternative to perform a bibliometric analysis, although it must be highlighted that it will never cover all the existing information and limits will always exist on using this approach.
It is necessary to clarify that the present study has only considered the papers that have correlated climate change and the impact of storm surges on the coastal zone leading to erosion and ooding (inundation). Studies that did not consider such disasters (i.e., erosion and ooding) were not included in this paper. Four criteria were de ned for choosing the terms to be included in the systematic review: 1) the study area should be impacted by storm surge and ooding (inundation) or erosion; 2) the area must be located along a coastal zone; 3) the impact must be associated with climate change; 4) papers with biological bias were disregarded. Thus, the query expression applied to the Advanced paper search in the Scopus database was: The terms were searched in the titles, abstracts, and keywords (TITLE-ABS-KEY) of each indexed paper. We select only papers (LIMIT-TO (DOCTYPE, "ar")) published in journals (LIMIT-TO (SRCTYPE, "j")) and written in English (LIMIT-TO (LANGUAGE, "English"). No periods (dates) were determined. Thus, the search has considered all papers in the historical series. In the same step, all abstracts were read practically, using the Show all abstracts option. The intention was to select only the papers of interest, based on the four pre-established criteria. Finally, the le with selected papers was exported in two extensions, .csv e .bib, which are automatically recognized by the bibliometric analysis software.
The third step consisted of data re nement (i.e., data cleaning) and analysis. Based on the studies already carried out (Lima and Bonetti 2020; Majeed and Ainin 2021), we opted for a more robust analysis with results generated by both the Bibliometrix package for R version 4.0.4 and VOSviewer version 1.6.16 software. The Bibliometrix package provides various routines for importing bibliographic data from Scopus and others, performing bibliometric analysis, and building data matrices for co-citation, coupling, scienti c collaboration analysis, and co-word analysis. It is a unique tool, developed in the statistical computing and graphic R language, according to a logical bibliometric work ow (Aria and Cuccurullo 2017). In the same sense, VOSviewer is a software tool for creating maps based on network data and visualizing and exploring these maps (Jan van Eck and Waltman 2020).
In addition, during the third step, the appropriate bibliometric methods were chosen to respond to the rst stage. Table 1 presents the relational and evaluative methods that were generated by using the Bibliometrix package for R version 4.0.4 and VOSviewer version 1.6.16. The choice of methods was based on the proposed objective of presenting the global scienti c production status on the topic.
Moreover, Zupic and Čater (2015) discuss the pros and cons of presenting each method in the results, which also contributed to the decision on the nal results of the present study.
The co-authorship method is established when two authors co-publish a paper. It can provide evidence of collaboration and produce the social structure of the eld. The citation method is used as a measure of in uence. If an paper or journal is heavily cited, it is considered important. In addition, it can quickly help to nd important works in the research eld. The co-citation method by the author is de ned as the frequency in which two authors are cited together. This analysis uses co-citation counts to construct measures of similarity between authors or journals. A fundamental assumption of co-citation analysis is that the more two items are cited together, the more their contents are related. The co-occurrence method by author keywords connects words when they appear in the same keyword list. The idea underlying the method is that when words frequently co-occur in papers, it means that the concepts behind those words are closely related ( Finally, in the fourth step, all the results generated for scienti c production were interpreted by the following units of analysis: countries, a liations, journals, authors, and keywords. In addition, a table was attached as supplementary material to facilitate the literature review of other researchers. For each paper analyzed, the following information was tabulated: authors, title, year of publication, journal, and DOI.

Global Scienti c Production Status
We found 486 papers written in English and published in journals between 1991 and February 2021 in the Scopus database. However, 429 papers were selected according to the abstracts and pre-established criteria in the second step of the methodology. The analysis resulted in papers produced in 57 countries, 182 journals, 660 a liations, by 1454 authors, who used a total of 1075 keywords. The rst and most relevant results for each unit of analysis were listed to make the data presentation clarify.
As shown in Fig. 2, during the period between 1991 and February 2021, there was an increase in annual scienti c publications according to the Scopus database. The rst paper was written in 1991, and the second in 1992, with an interval without publication in 1993. Since 1996, more than one paper has been

Analysis by country
Among 57 countries, at least 10 have 18 papers or more published.  (14) than Spain (18) but adds 45 more citations. Figure 3 shows the co-authorship network by country using a cluster approach. Colors separate countries into different clusters by collaborations; the size of the circle represents the number of publications, and the line thickness represents how strong the relationship between countries is. For example, lines are thicker between countries with more co-authorships. The results of co-authorship by countries are represented by 10 different clusters, in which 43 authors have at least two co-authorships (Fig. 3). A larger network among researchers was observed in 39 countries. The United States researchers were the most productive (132 papers) in the period and have established co-authorship with 22 different countries. Among them, China, the United Kingdom, Italy, Australia, and India presented a high relationship.
Researchers from the United Kingdom have 54 published papers and at least two co-authorships with 24 other countries. Among them, Ghana, China, Germany, and the Netherlands. Authors from the Netherlands, Germany, and Italy have 36, 35, and 33 papers (respectively) and at least two co-authorships with researchers from 17 countries. Authors from these three countries have a strong relationship. Both Australia and China had strong relationship, co-authorinig with 11 countries. In Canada, two coauthorships occur with 8 countries, and in India, with 6, which are represented in the same cluster as Iran.
Despite being in tenth place in the publication ranking, Spain has at least two co-authorships with 18 countries. Brazil, which has 4 published papers, has a strong relation with Portugal and the United States.

Analysis by a liations
Authors' addresses recorded a total of 660 a liations. It was found that 490 a liations (74.24%) published only 1 paper. Figure 4 lists the 12 organizations with which authors were a liated in terms of the most papers published in the period between 1991 and February. Pennsylvania State University is the most proli c a liation with 20 papers, followed by Deltares and Princeton University, which published 15 papers each. The 12 a liations represent 29.84%, with 128 papers out of a total of 429 publications on the topic. The other a liations published 6 papers or less. Regarding countries, 7 a liations are located in the United States. The Netherlands is the second country that most appeared in the rank, with 2 a liations. The United Kingdom, Greece, and Taiwan are host countries for one a liation.

Analysis by journals
The 429 selected papers were published in 182 indidivual journals. Table 3  The rank of the journals that have been published more is different, regarding the most cited journals. also do not appear in the ranking of those that have published more on the topic, but they total 763, 648, and 619 citations, respectively. The other journals, which are not in the rank of the rst 10, were cited 412 times or less. Among the 13,548 citations, the rst 10 journals have 6,778 (50.03% of the total). The results also demonstrate that 16 journals did not obtain citations, 14 had 1 citation, and 152 were cited 2 times or more.

Analysis by authors
The results indicate a total of 1,454 authors. The 10 most productive authors were highlighted, according to the number of publications on the topic. In addition, the number of citations, a liations, country, and gender of the author is presented in Table 5. In our analysis, we have only selected the authors who have correlated climate change and the impact of storm surges on the coastal zone leading to erosion and ooding (inundation) in their papers. In the current times, in which gender equality is sought, it is necessary to present and show that women still do not dominate ranks, such as those presented in Table 5 and Table 6. Concerning the largest number of publications, Lin (female) occupies the second position. In the same sense, Torresan occupies the ninth position of the rank. However, the results point out that women are the minority compared to men, who account for 80% of the 10 authors on the top ranking. Figure 7 shows the clusters of researchers who are co-authors in at least two papers. Among the 1454 authors, 183 are co-authors in the 429 papers analyzed. However, the largest set of connecting coauthorship consists of 49 authors. In this case, the cluster analysis allows the identi cation of the niches of co-authors. In the light blue cluster, Nicholls was the author who published the most, with 13 papers, and he is co-author with 10 other researchers. Among his strongest relationships are Brown, Hanson, and Vafeidis. Lin published 9 papers, thus being the author who has most published in the yellow cluster, coauthoring with 8 researchers. In the green cluster, Lowe, who published 7 papers, was the researcher with the highest co-authorship number (15). In the dark blue cluster, Weisse was the most proli c author with 5 published papers, and is a co-author with 10 researchers. Gonnert published 3 papers, hence is the most productive author in the red cluster, being the co-author with 9 other researchers. In the purple cluster, Kirshen, which published 4 papers, is co-authored with 5 researchers. Sterl was the most proli c publisher in the brown cluster with 3 published papers, and is co-author with 9 other authors. In the orange cluster, Dawson, which has 2 publications, is co-author with 4 researchers. Finally, in the pink cluster, Xian was the author who has written the most number of papers (3) and is co-author with 3 researchers. the United Kingdom, with two researchers. France appears only once, in the seventh position in the ranks. As in the analysis of the number of papers published by the author (Table 5), the results related to gender in Table 6 also indicate that women occupy only 20% of the rank of most cited researchers. It is also noteworthy that Neumann and Zimmermann were co-authors of the same work, with Neumann being the rst author. Again, it is shown that women are not the majority among the 10 authors on the top ranking. The co-citation patterns indicate the history of recognition and academic impact of publications since a publication can be relevant for future research and can, thus, potentially be cited (Hjørland 2013). The cocitation analysis by the author presents the group of researchers who are cited by a group of authors. Therefore, it becomes important to demonstrate to the authors that it is also related to the topic. The analysis of co-citation by the author demonstrates, from four clusters, the set of authors that were most cited by the 1454 authors of the 429 papers analyzed. In the selected papers, 24,623 researchers were cited, and the 36 researchers represented in Fig. 9 were cited at least 65 times. The authors' proximity to clusters refers to the list of their proposed themes within the general topic of storm surges and climate changes.
Nicholls appears as the most cited author, with 552 citations. In the same papers, authors, such as Woodroffe, Hinkel (the fth most cited, with 144), Tol, Vafeidis, Hallegatte, and Corfee-Morlot are also cited. It does not exclude the fact that the authors of other clusters are also cited in the same paper, in which Nicholls appears. Emanuel is in the second position, with 247 citations. The authors most cited by the same paper are represented in the green cluster, in which Lin, fourth-most cited is part, with 146.
Church is the third in rank, with 147 citations by base papers. The author is represented by the yellow cluster, in which also appears Cazenave, the sixth most cited author, with 141 papers. The other researchers were cited 121 times or less. January and April 2, 2021, the paper had already been cited 49 times, as indicated by Scopus.

Keywords' analysis
The evolution of the rst 5 keywords between 1991 and 2021 is shown in Fig. 11. The keywords plus (i.e., most commonly used words in titles, abstracts, and keywords list) were represented according to their annual and noncumulative occurrence. The main keywords are highlighted: Climate Change, Storm Surge, Floods, Sea Level, and Sea Level Change. In general, the terms appeared considerably since 2007. The most used keyword per year was climate change. Between the period of 2019 to February 2021, the curve is stable, with an annual occurrence of approximately 35. It shows that its use is constant and more elevated, concerning the other highlighted keywords.
Storm surge is the second most used keyword since 2007. Its usage peak occurred between 2017 and 2019. However, it must be considered that only papers from January and February of the year 2021 were selected. Studies related to the topic are growing exponentially until the year 2020, as seen in Fig. 2. In addition, only two months in 2021 account for more publications than several other years, such as 2009. The growth occurs due to the current situation of climate change and environmental impacts. Sea level change was the most used keyword in the papers since 2007. However, between 2015 and 2018, it remained stable, with a drop in its employment in 2019. After 2016, oods started to have a higher annual occurrence in keywords. Its peak was in the rst two months of 2021. The use of the term sea level in keywords was stable between 2017 and 2019, slightly decreasing in 2020.
Among the 1075 keywords established by the authors, 17 occurred in at least 10 papers out of 429 analyzed. The co-occurrence network based on keywords occurrences in papers is shown in Fig. 12. The size of the node is proportional to the frequency of occurrence of the keyword, and the thickness of the line represents the intensity of co-occurrence between individual keywords (Mishra et al. 2020). Zupic and Čater (2015) point out the negative aspect of presenting co-word analysis. The authors show that words can appear in different forms, as well as they can have different meanings.
The keywords determined by the authors did not follow a pattern. It can occur in different research elds since different words can have the same interpretation meaning. In the case of the present study, words, such as sea level rise and sea-level rise, were found among the authors' the 17 most used keywords. The researchers used it in 64 and 53 papers, respectively. Another example of duplicity is storm surge and storm surges, which appeared 82 and 26 times, respectively. Moreover, words like coastal ooding (24) and ooding (22) were used. Considering that the research only delimited studies related to the coastal area, the two terms can be classi ed as synonyms. The keyword inundation (13) may or may not give rise to the same interpretation as ooding. There is no consensus, among authors in the research area. In the same sense, erosion and coastal erosion appeared (both 13 times), which probably refer to the same topic. It should be noticed that, logically, when one of these keywords is used, the other one will not be used. It justi es the fact that they have the same interpretation.
An analysis of some papers by the three authors who published more on the topic revealed divergences in the keywords pattern. Nicholls (2002)  It is understood that the use of words in the singular or plural, as storm surge, may be related to the number of events that were analyzed by the authors. The three keywords most used by the researchers were climate change (168 times), sea level rise (sea-level rise), and storm surge (storm surges). The highlighting of these keywords is justi ed, as they are directly related to the keywords used in the query expression determined in the present study. Regarding keywords that did not show duplicity, there is vulnerability and adaptation, which appeared in 18 papers each, next to the sea level keyword, which was used 14 times. Risk assessment appeared in 11 papers, while the words coastal hazards, coastal management, and ood risk appeared in 10 papers. The other keywords were used 9 times or less.
Four clusters are represented in Fig. 12. In papers that the authors opted for the keyword climate change, the other 16 keywords presented were also used at least once. In the same green cluster, terms, such as sea level rise, coastal ooding, storm surge, sea level, and ood risk, appear. Storm surges appear related to the words represented in the blue cluster, considering its strong relationship with climate change and sea level rise, as well as sea-level rise, being more used in conjunction with words belonging to the red cluster, in addition to having a strong relationship with storm surge. Erosion and coastal hazards appear in the yellow cluster. Terms considered as synonyms appear in different clusters, which shows the fact that they are not used together in the keyword lists.

Guide For Authors: Bibliographic Review
For researchers who intend to make a signi cant bibliographic review or even publish on the topic, it is worth looking for authors, papers and journals with signi cant recognition among the scienti c community. Two countries occur as the most productive and cited: the United States and the United Kingdom. The United States is the country with the largest number of authors who publish on the topic. In addition, the country hosting important universities is highlighted in Fig. 4

Final Considerations
The bibliometric analysis was presented as an alternative approach to demonstrate the global scienti c production status. Both software used, the Bibliometrix package for R and VOSviewer generated important results that complemented each other and, thus, have supported a better understanding of the scienti c knowledge in the research eld. The results showed that there is an exponential growth in the publication of papers written and published in journals since 1991. The peak in the publication of papers related to the topic was observed in 2020, but in the rst two months of 2021, it has been published more papers than the entire period before 2007.
According to the query expression applied in the Scopus database, which was based on four preestablished criteria, 429 papers were selected.            Keywords usage growth from 1991 to February 2021 Figure 12