A quantitative analysis of research trends in flood hazard assessment

Studies focusing on flood hazard assessment (FHA) reviews are still rare and usually approach mapping, uncertainty, spatial scale, and economic loss. Nevertheless, most of these studies only provide a snapshot of the current development of FHA, showing a partial view, and focusing on a limited number of selected approaches and methods. This study aims to analyze the historical development and emerging fields of FHA-related research and applies a bibliometric analysis method based on the Bibliometrix tool and the Web of Science (WoS) database. We refined the 4135 articles retrieved, established a dataset containing 723 articles, and downloaded the information of all articles, including journal, author, country, keywords, and abstracts. The dataset is entered bibliometric analysis model to study the development of FHA-related research from 2000 to 2020 (data as of 15 February 2021). The results show that the number of articles on FHA-related research has continued to increase and developed rapidly. The top five countries in the number of studies are China, Italy, the United States, Spain, and the United Kingdom. Compared to developed countries, the FHA-related research in developing countries lags significantly behind although it is developing rapidly. Future research is likely to focus on global change, coastal floods, compound disaster, and psycho-social aspects of flooding. Moreover, research using remote sensing and machine learning is becoming increasingly sophisticated. The purpose of this research is to provide direction for future FHA-related research and provide reference and scientific guidance for FHA-related researchers.


Introduction
Globally, floods have become more frequent, leading to substantial economic loss and death, a finding that is consistent with observations of anthropogenic climate change (Stott 2016;Lehner et al. 2006;Luo et al. 2021b). Despite substantial flood prevention efforts, the resultant loss of human life and property persists at high levels, with floods accounting for 34 and 40% of all global natural disasters in quantity and losses, respectively (Lyu et al. 2019;Petit-Boix et al. 2017). Asian conference on disaster reduction (ADRC) defined risk as a function of the hazard, exposure, and vulnerability (Center 2005). Flood hazard assessment (FHA) is considered an important means of decreasing flood disaster losses and ensuring the healthy and sustainable development of human society. Therefore, research in this area is imperative to meet human needs (Wang et al. 2015). FHA, a qualitative or semi-quantitative method, is focused on the combined influence of disasterinducing factors and environments (Hallegatte et al. 2013;Luo et al. 2022a). The current concept of FHA takes into account the probability of flood hazard, the potential negative consequences of floods on human health, the environment, cultural heritage, and economic activities (Solin and Skubincan 2013). In order to monitor possible flood events in the southern Cordoba Province in Argentina, the standardized precipitation index for the FHA was established to facilitate preventive actions to mitigate the impacts of flood events (Seiler et al. 2002). Moreover, fuzzy reasoning, gray fixed weight cluster analysis, and neural network method have been applied in FHA (Nayak et al. 2005;Shao et al. 2014;Zhu 1994).
Several published articles have been reviewed on FHA (Ologunorisa and Abawua 2005;Tsakiris 2014;Winsemius et al. 2013). Thematic reviews were also published on various FHA methods, including mapping (Zhu et al. 2022), uncertainty (Apel et al. 2004), spatial scale (De Moel et al. 2015), economic loss (Merz et al. 2010), and global scale hazard analyses Ward et al. 2020), among others. However, most of these studies only provide a snapshot of the current development of FHA, showing a partial view, and focusing on a limited number of selected approaches and methods (developed mainly by the researchers). Furthermore, previous reviews have not provided comprehensive understanding of trends in FHArelated research or introduce the sources of uncertainty in the field. Bibliometric analysis can help us understand the evolution of global FHA-related research over time.
Since 2006, bibliometric research has been widely used in various fields. In the field of medical science, the research enables us to analyze and understand emerging trends. In the field of economics, it gives us access to valuable information that is not available through other bibliographic methods (Ye et al. 2020). In the field of FHA, the related articles are very abundant (more than 4000 references), but the articles on bibliometric analysis are very few (less than 15 references). Only part of the bibliometric analysis is available, including coastal flooding (Gao and Ruan 2018), urban flood vulnerability (Cho and Chang 2017), China's urban flood risk (Yin et al. 2015), ecological models related to eutrophication (Hu et al. 2019), availability of global datasets for water-related disasters (Lindersson et al. 2020), group decision-making techniques and multi-criteria decision-making methods in risk assessment (Arantes and Carpinetti 2019;Liu et al. 2019).
Here we provide a dataset from all FHA-related articles based on the Web of Science (WoS) database to establish a bibliometric analysis system, revealing cooperation among researchers all over the world. With the combination of cocitation analysis and keyword analysis, our main objective was to systematically discuss the current research status in FHA-related research and visualizes the prospect and development of the field. In addition, we explore the hotspots of FHA-related research based on keyword co-presentation analysis. We explore the leading countries and authors in FHA-related research based on the analysis of high-production countries and high-production authors. We explore excellent examples of FHA-related research based on top journals, g-index articles, h-index articles and m-index articles.
The remainder of the article is structured as follows. First, the methods and databases are described. Second, the investigation results are obtained through descriptive and bibliometric analysis. Finally, limitations and suggestions for future research are provided.

Bibliometric analysis tool
Publications are the record of scientific trends: ''Communication in science is realized through publications. Thus, scientific explanations, and in general scientific knowledge, are contained in written documents constituting scientific literature'' (Riviera 2013). Bibliometric analysis, a quantitative approach that aims to map and evaluate the body of literature to identify potential research gaps and emphasize the boundaries of knowledge, is considered an effective means of grasping critical trends in the scientific investigation (Tranfield et al. 2003).
The study was performed using R 4.0.4 (R Core Team 2022) and ''Bibliometrix'' (Aria and Cuccurullo 2017) package, which provided various routines for both bibliometric and scientometric quantitative research by building data matrices for citation, scientific collaboration analysis, and co-word analysis (Donthu et al. 2021;Zha et al. 2021). Specific codes and commands for bibliometric indices were entered. Subsequently, detailed analysis graphs and tables related to the most productive authors, affiliations, countries, sources, and highly cited articles were produced (Fig. 1). Finally, all the documents were independently coded by the author to recognize the units of analysis in the selected samples to increase the reliability of the research outcomes.

Literature search and refinement
Systematic literature search and extraction were performed using the WoS (http://www.webofknowledge.com) on 15 February 2021, following a reproducible methodology. WoS is the largest comprehensive and multidisciplinary academic database worldwide, comprising more than 8000 worldwide most influential, peer expert reviewed, highquality journals.
Enter the predefined keyword string ''flood hazard assessment OR flood hazard analysis'' in the WoS search and select ''titles, abstracts, and keywords subject'' to search, resulting in a total of 4135 articles. The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow diagram was used to record the process of identification, screening, eligibility assessment, and inclusion in Fig. 2 Moher et al. 2009;Page et al. 2021). Record screening was performed by manually inspecting titles, abstracts, and keywords for the following inclusion criteria: • The publication years were limited from 2000 to 2020.
• The document types were limited to articles, early access, and reviews. • The document languages were limited to English and Chinese.
The year 2021 was excluded because the year had just begun at the time of this research. Before the year 2000 were excluded because there are few records of FHA in the WoS. This constraint may have excluded high-impact and high-quality articles published before 2000, which could have influenced the study's main findings. While language was not used as a filter, it is well known that the database was developed utilizing English and Chinese, which could be understood as a ''quasi filter.'' This language constraint may have excluded high-impact and high-quality other language articles, which could have influenced the study's main findings.
During the screening, documents were excluded if they nonconformance with one or both inclusion criteria. 776 documents were excluded during screening, and 3359 were considered for full-text eligibility assessment. The full-text assessment revealed that 723 records satisfied the criteria (and thus were included in the research), while 2636 records were excluded.
Based on the above process, relevant information (including journal, publication year, author, country, keywords, citations, and abstracts) of 723 articles were downloaded and saved as a plain text file. The file is used as a dataset to enter the Bibliometrix package for analysis.

Documents
As shown in Table 1, which includes the main topics associated with bibliometric analysis, FHA-related research has been published from 222 sources. During the past 20 years, 723 FHA-related scientific productions have been published. Nevertheless, the total number of citations was more than 8000, with an average number of citations per document at 11.16. Regarding authorship, 2596 individual authors have contributed to the field so far. On average, each article is written by three authors (3.6), which indicates that the study requires scholarly collaboration. The collaboration index, which is the ratio of total authors of multi-authored articles (2574) to the total multiauthored articles (698), is at a high level (3.69), and the number of single-authored documents was found to be only 24. Figure 3 shows FHA-related statistical data based on scientific productions published from 2000 to 2020. The trend of annual publications shows that there has been a steadily increasing interest in FHA, and this increase has continued without any significant decrease. Thus, FHA is  an emerging discipline that has not yet peaked. The 2010-2020 period was shown to be an active period with a high yield of documents. This situation may be related to international provisions or programs (such as the European Flood Directive) that provide the impetus for scientific research, although there may be a time lag in the editing and publishing of manuscripts (Díez-Herrero and Garrote 2020). Therefore, we divide the development process of FHA-related research into the initial period (2000)(2001)(2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009) and the development period (2010-present). The top 10 most cited articles on FHA are shown in Table 2. More accurately, the total citations that can be retrieved in the WoS are shown. The results coincide well with the research areas defined in this study. Interestingly, all articles on this list come from different journals. The maximum number of citations reached was 861, which occupies an important position in the FHA-related research. Table 3 shows the top20 frequent keywords of FHArelated research. The most frequent keyword used by authors in their documents was model (98), illustrating that model has played an essential role in FHA. In fact, using the model to simulate the flooding process is the basis of non-engineering measures for flood prevention and control. The second most frequent word is climate-change (74), followed by management (67), vulnerability (65), risk (56) and impact (47), suggesting the FHA-related challenges. Figure 4 shows the conceptual structure map of keywords associated with scientific publications in the FHArelated research. It presents the co-word analysis performed using the word co-occurrences network to map FHA-related articles' keywords in the WoS database (Cobo et al. 2011). It can be seen from the figure that, by multiple correspondence analyses, the articles involved in the study are mainly divided into three clusters of documents that express common concepts, which reveal the intellectual structure of FHA-related research. Thus, it can be observed that the first cluster deals with model themes while cluster two concerns support activities (urbanization).

Sources
According to Source dynamics, which shows the trend of the sources depending on the scientific FHA publications (Fig. 5), the evolution of the most productive journals can be divided into three typologies: • Natural Hazards, which was initiated to publish FHArelated documents in 2010, continued to increase throughout the studied period. • Some journals (Natural Hazards and Earth System Sciences, and Sustainability) characterized by an apparent original production level prior to 2018, but after which conspicuously reduced their FHA-related production with marked intervals of inactivity.
• The rest of the journals, including Journal of Hydrology, Water, Journal of Flood Risk Management, and Science of the Total Environment, began to publish FHA-related documents relatively late but present significant and continuous growth.
The h-index, g-index, m-index, total citations, number of publications, and year of first publication on FHA-related research for the top 15 productive sources contributing to the FHA-related research are shown in Table 4. The h-index was defined as the number of articles of an author that have citation count larger than or equal to the h of all co-authors of each article, as a valuable index to quantify an individual's scientific output, which considers the effect of multiple authorship (Voir Hirsch 2005). The g-index was defined as the largest number g of articles that  together received g 2 or more citations (g C h). Interestingly, the higher the number of citations in the top class (in other words, the skew of the citation distribution), the higher the g-index will be (Egghe 2006). The m-index is the h-index divided by the number of years that have passed since each journal released its first publication in the dataset (Hirsch 2007). The m-index enables researchers to weigh productivity with their career length. Natural Hazards had the highest number of total citations (1199), followed by Journal of Hydrology (578) and Science of the Total Environment (456). However, the average number of citations per article was 17.63, 23.12 and 26.82, respectively, implying that Science of the Total Environment is the most influential, which was confirmed by the research shown in the h-index, g-index and the m-index. Among the most productive sources, Journal of Hydrology is not the journal with the highest impact factor, but it represents the high level of FHA in the industry. As shown in Fig. 6, the above sources account for 33% of the total scientific publications on FHA, forming a core area of FHA research, which confirms Bradford's law -which can be summarized as follows-if sources in a research field are divided into three parts by the number of articles, each area accounting for around one-third of the total articles. The sources in each area will be proportional to 1: n: n 2 , while the first area is usually defined as the essential area (Xiao et al. 2021).

Authors
Regarding the analysis at the Authors level, Fig. 7 shows their distribution over time. The circle area is proportional to the annual number of articles, and the colour depth of the circle is proportional to the total article citations per year. Among the top 20 most productive authors, Wang H (6 articles) published the most articles, and the scholars analyzed a large number of issues in dealing with flood disasters. Based on statistics, Xu Youpeng had the first article on FHA in 2005 and has sustained a steady output ever since. For most researchers, once they initiated the work on FHA, they would constantly conduct research, which implies the importance of FHA in research programs. The documents published by Nicholls, RJ, in 2015 have the greatest number of citations, which implies that, to a large  extent, it has played an essential role in the FHA-related study. Table 5 presents the top 30 productive countries defined by the total FHA-related scientific production counted by the corresponding author's country. Among 30 countries, China makes the top of the list, followed by the USA and the UK, a distant second and third place. Even though flood hazards are among the most severe disasters globally, research indicates that the losses caused by flood hazards are severe in China, which drives significant volumes of FHA-related research.
The proportion of FHA-related scientific production for both multi-country publications (MCP) and singular country publications (SCP), China still tops the list, as shown in Fig. 8. MCP relates to articles where at least one of the co-authors is from various countries, yet SCP refers  to cooperation within a country. Most countries prefer to form national cooperation instead of international cooperation as a rule. Among 20 countries, only four countries, including the Netherlands, Australia, Egypt, and Austria, have a greater proportion of articles published in MCP. The maximum number of citations per article might suggest an inclination toward the formation of MCP (rather than SCP) since they approved of the use of resources and technologies that might be lacking in one country, leading to a more profound citational impact. Table 6 shows the top 15 affiliations publishing the largest number of documents, with many of them having an excellent prestige for scientific research and innovation. Two-thirds of the top 15 affiliations are located in China, which is related to the fact that China has much FHArelated research. Hohai University is the most productive affiliation, with 57 articles published, and its research on FHA has a well-known reputation in Chinese academia. Hohai University is followed by Beijing Normal University (47 articles), Huazhong University of Science and Technology (41 articles), and East China Normal University (34 articles), all top universities in China. Figure 9a shows some essential collaboration among countries around the world, and countries shaded with the darker blue turn out to be the more productive concerning scientific publications. The pink curves represent the collaboration network established among countries worldwide-the more robust the curves, the more frequent collaboration. Obviously, the most international collaboration between China and the USA supports the research shown in Fig. 9b. Meanwhile, it shows a more visible scenery of international collaboration by the visualization of connections among the top 20 contributing countries, which indicates that the more robust collaborations are being established by scholars from various countries, who form connections with their partners in different countries around the world.
According to the analysis of a sample of FHA-related research, the h-index, g-index, m-index, total citations, the number of publications, and year of first publication on FHA-related research for the top 15 authors are shown in Table 7. Ward PJ, Zhang Q, Merz B, Nicholls RJ and Pradhan B have the highest h-indexes in the table, all having an h-index greater than 5, indicating they published at least 5 articles with at least 5 citations each. In terms of productivity, Wang J, Wu ZN and Zhang Q were the best ranked. Among the three authors, Zhang Q has the highest citations and the greatest h-index. Nicholls RJ got the highest number of citations (676). Although Ward PJ and Pradhan B are not the authors with the highest number of citations, the total number of citations is 465 and 489, respectively, but the h-index and g-index are equal, indicating that each of their articles has a high quality. To avoid excluding younger scholars, the m-index was also calculated. Although the total citation of Hong HY is only 41, his articles (4) are published in 2020, and the m-index (2) is the highest in the table, indicating that Hong HY is the author with the most potential.
Finally, a comprehensive analysis of the relationship between keywords, sources and countries is carried out. Figure 10 provides the three-field plot of FHA-related studies from 2000 to 2020, which shows the interactions among the leading journals (left), keywords (middle) and countries (right) within FHA-related research. The visualization results more intuitively reflect the close relationship between the three.

Status of the development of FHA
In the initial period of FHA-related research, a few articles were published, and the research direction focused on the assessment and zoning of flood disasters based on geographical information systems (GIS) and information diffuse estimation (Islam and Sado 2000). Some scholars present a methodology for assessment and zoning of flood disasters in northeast China based on GIS and the technology of natural disaster risk assessment (Zhang et al. 2005). Moreover, the information diffusion theory-based assessment model assessed Tianjin's drought and waterlogging hazards (Du et al. 2009).
In the development period of FHA-related research, the number of published articles and the types of studies has rapidly increased, while scholars from various countries established robust collaborations. Many advanced technologies were applied in FHA-related research, such as remote sensing technology and machine learning technology (Hosseini et al. 2020). Some researchers used remote sensing data such as Shuttle Radar Topography Mission and enhanced Thematic Mapper Plus to assess the flash flood disaster (Youssef et al. 2011). In addition, with limited data, two machine learning models: Genetic Algorithm Rule-Set Production and Quick Unbiased Efficient Statistical Tree, were used to produce a flood risk map (Darabi et al. 2019). The result shows that machine learning models were effective for flood hazard zoning, particularly in areas without detailed hydraulic and hydrographic data.
Finally, the emerging research and development of FHA could be found: (1) the effect of global change (climate change, sea-level rise, and global warming) on the magnitude and frequency of flooding and its associated damage (Hirabayashi et al. 2013); (2) coastal flooding and its economic damage, particularly in major coastal urban areas (Neumann et al. 2015); (3) compound disaster research and Stochastic Environmental Research and Risk Assessment (2023) 37:413-428 423 interactions between the flood disaster with other natural hazards, such as cyclones or other windstorms, earthquakes, and volcanic eruptions (Lindersson et al. 2020), and (4) the psycho-social aspects incorporating flood disaster (Bubeck et al. 2012).

Recommendations for FHA
Compared to developed countries, the FHA-related research in developing countries lags significantly behind although it is developing rapidly. The FHA-related studies carried out on a global scale have received close and longterm attention from researchers (Hirabayashi et al. 2013). These studies focus mainly on the predominance of FHArelated studies, the increased spatial resolution of the recent studies, the number and importance of coastal floods studies, and the direct flood damages assessment. It is not difficult to find that the FHA is an effective tool in managing and reducing flood disasters (Luo et al. 2021a;Wei et al. 2022), but several scholars have pointed out gaps in the FHA-related research (Luo et al. 2022b;Zhu et al. 2020). These research gaps that require improvement in theory and practice are also the focus of future research: (1) Fig. 9a clearly shows a significant gap in FHArelated research in Africa. The reason is not the absence of flooding disasters, but the lack of appropriate data and infrastructure can be considered one of the key factors for the limited FHA in Africa.
The most appropriate choice for most countries in Africa is the 30-m spatial resolution DEM from the ASTER project or the SRTM project (Díez-Herrero and Garrote 2020). Furthermore, attention must be  paid to poor quality or limited flow data availability. Finally, the lack of quality and systematic social data can significantly affect the study of risk exposure and vulnerability components.
(2) As FHA-related research is gradually growing, the collaboration between scholars from various countries is rapidly increasing. Extensive international collaboration would allow scholars to draw on advanced results from each other, thereby enhancing academic research on FHA. However, FHA must also consider the combined effects of local culture, politics, economy, hydrology, climate, and other factors (Bubeck et al. 2012). These factors limit the development and application of FHA to some extent. (3) The FHA-related researches cover numerous fields such as management, engineering, geology, and ecology. Thus, interdisciplinary integration will be becoming increasingly crucial in FHA-related studies (Barron et al. 2017). At present, the absence of specialized talent training systems in FHA has led to a shortage of talents, limiting the integration of FHA and other disciplines. Integrating multidisciplinary methods into a comprehensive platform can provide the impetus for the development and innovation of current disciplines. The future direction of FHA will certainly be interdisciplinary integration and multisectoral construction. (4) With the remote sensing technology applied in FHA, much FHA-related data is bound to be produced (Youssef et al. 2011). A large amount of data laid a solid foundation for FHA but posed significant challenges, such as data storage, data management, and data assimilation from different sources. In addition, the data of uncertain quality can introduce errors to model simulations and experiments. Thus, high-performance distributed storage technologies, data processing technologies, and data assimilation are urgent needs to solve the above problems and are the hotspots of FHA research. (5) With more concentrated humans and assets in urbanized areas, the number of deaths, people affected, and economic losses caused by the flood disaster would be more severe (Bach et al. 2014). Due to the rapid development of the urbanization process and dense population, often with insufficient infrastructure, loss from flooding is expected to increase in urban areas substantially. Thus, it is essential to develop FHA-related research in urban areas.

Limitations of this study
Firstly, the research established a literature dataset to review the development of FHA, but the document collection is not comprehensive enough due to the limitations of database and retrieval tools. Moreover, the document languages were limited to English and Chinese. This language constraint may have excluded high-impact and highquality other language articles and highlighted the contribution of Chinese scholars, which could have influenced the study's main findings. Finally, the names and surnames of Chinese authors are relatively simple, and duplication of names caused by Chinese initials is also a potential cause of the error. In the future study, we will introduce more sources such as the Scopus database, media, meetings, and patents and apply more comprehensive keywords in our search to obtain more detailed results.

Conclusion
This study collected the FHA-related articles published between 2000 and 2020, resulting in a final dataset of 723 articles. A structured review of a bibliometric analysis of the dataset allows scholars to understand the research progress, development trends fully, and focus of FHA and has a certain guiding significance for the current development of FHA. This study found that: (1) the FHA-related research has continued to increase and developed rapidly; (2) China, Italy, and the United States account for about 30% of articles published in FHA; (3) the FHA-related research cover numerous fields such as engineering, ecology and environment, geology, and political economy; (4) The FHA-related research carried out on a global scale has been paid close and long-term attention; (5) There are four future trends in FHA-related research: global change, coastal floods, compound disaster and psycho-social aspects of flooding. Finally, the FHA-related research has relevant past, prominent present, and promising future to solve humanity's problems by reducing natural disasters.
preparation of the manuscript. We also appreciate the reviewers for providing valuable comments.

Conflict of interests
The authors declare no conflicts of interest. The funders had no role in the study design, data collection, and analysis; decision to publish; or preparation of the manuscript. The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.