Literature Source

The literature search of this paper is limited to international peer-reviewed studies for two reasons. First, the format of non-academic studies (unpublished papers and reports) is not standard enough to be analyzed by bibliometric analysis software. In addition, non-academic studies are more prone to possible distortion, while results published in peer-reviewed studies have been carefully examined, which makes the loss estimations more credible. Data used in this study comes from the Web of Science (WoS) core collection database of Clarivate Analytics. The database contains more than 10,000 multi-disciplinary, high impact, international, authoritative and comprehensive academic journals, and is the most academically authoritative source of citation information in the world. In order to ensure the authenticity and integrity of the research results, it is very important to set the retrieval style in the process of data collection. TCs are classified according to their place of occurrence and intensity. The broadly defined TCs include hurricanes, typhoons and cyclonic storms (Ghosh & Chakravarty, 2018), so they were all set as search terms in this paper. Through repeated discussion of the search terms and comparison of the retrieval style many times, the theme retrieval and title retrieval were both adopted for literature collection.

The identified literature has expanded to a broad range of subjects which were beyond the scope of this review. Therefore, articles on irrelevant subjects needed to be excluded. Irrelevant subject areas are those that do not fall within the scope of atmospheric science and disaster science. The irrelevant subjects were determined by the agreement of at least two research team members. Then, the selected articles went through the process of removing duplicates. Finally, the title, keywords and abstract of searched articles have been checked to determine whether they are related to the research topic. The purpose of this paper is to explore the economic loss assessment of TCs. Hence, research such as factors of increased TC intensity and frequency were excluded from further exploration. 355 articles remained after the total process. The process flows of paper identification and selection are shown in Table 1.

Table 1.

The Process of Paper Identification and Selection.

Research Methods

Research Tools and Processes

In order to clarify the current developments and trends of TCs economic loss assessment, scientific visualization methods were used to plot knowledge domain maps. CiteSpace was used for keyword frequency statistics and high frequency keyword screening and analysis. VOSviewer was adopted for social network analysis and hierarchical clustering to quantitatively and dynamically demonstrate the evolution of TCs economic loss assessment. Moreover, future research direction was also predicted and explored according to the keyword strategy matrix. Research tools and processes are shown in Figure 1.

Figure 1.

Research Tools and Processes of Bibliometric Data Analysis.

Quantitative Analysis of Years and Countries

The development status, knowledge accumulation and maturity of a research field can be measured by its literature publications, which is of great significance to evaluate the development trend and dynamic evolution. First, this paper described the publications and the changing trend in the past 60 years (Figure 2). In general, the volume has generally shown an upward trend, and the issue of TCs economic loss assessment has clearly been noted by scholars.

Figure 2.

Annual Publication Related to Economic Loss Assessment of TCs.

Second, to identify the main countries involved in TCs economic loss assessment and the cooperation among them, CiteSpace was used to draw maps of country distributions, which are shown in Figures 3 and 4. From the time of publications, the USA, Mexico, the UK, Netherlands, Australia are the countries that carried out the economic loss studies of TCs earlier. From the perspective of degree centrality, the USA occupies the first place, indicating that the USA has the greatest academic influence in the field of TCs economic loss assessment, while China is behind.

Figure 3.

Countries/Regions Time Zone View Related to Economic Loss Assessment of TCs. Each dot represents a node (i.e., a country/region studying TCs economic loss assessment), and its node size is directly proportional to the number of publications. The purple ring on the outermost layer of the node represents its degree centrality.

Figure 4.

Countries/Regions Centrality Analysis Related to Economic Loss Assessment of TCs. The sizes of circles and fonts represent different degree centrality.

In terms of the number of publications, the total number of literatures in the top 8 countries is 312, accounting for 87.89% of the total number of literatures. The USA is the most active scholar in this field, with a dominant position in output (183 articles), accounting for about 51.55% of total publications. China ranks second with 47 papers, accounting for 13.24%. Australia ranks third with 16 publications, accounting for 4.51%. Other countries accounting for about 4% of the total literature are Germany, France, Japan and the UK. These countries are economically developed or in a period of rapid economic development, and attach great importance to this research field.

Quantitative Analysis of Institutions

Third, information on the most productive institutions that specialize in a certain theme can be identified via cooperation analysis of institutions, the academic cooperative connections among institutions are shown in Figures 5 and 6 respectively. The top 20 most productive institutions for total publications are shown on Table 2. The major academic contributions, which were concluded in terms of total publications, primarily originated from the MIT (8 articles), Texas A&M Univ (8 articles), Nanjing Univ Informat Sci & Technol (8 articles), Columbia Univ (7 articles), NOAA (7 articles), and Michigan Technol Univ (7 articles). It is followed by Florida State Univ (6 articles), Florida Int Univ (6 articles), Chinese Acad Sci (6 articles), Vrije Univ Amsterdam (6 articles), Beijing Normal Univ (6 articles), Ecole Polytech (6 articles) and Univ Newcastle (6 articles), all of which are important institutions for economic loss assessment of TCs.

Figure 5.

Institution Co-Citation Map Related to Economic Loss Assessment of TCs. The academic cooperative connections among institutions can be displayed by clustering.

Figure 6.

Institution Time Zone View Related to Economic Loss Assessment of TCs. The timeline of institutions’ research can be reflected in the map.

Table 2.

Top 20 Institutions With the Most Publications Related to Economic Loss Assessment of TCs.

From the perspective of cooperative connections among institutions, this paper applied CiteSpace to generate the institution co-citation map, and found that colleges are important frontiers of TCs economic loss assessment research. Currently, three institution networks have mainly been formed. The largest cooperation network is the cooperation network centered on MIT. The cooperation institutions are Columbia Univ, Ecole Polytech, Princeton Univ, Yale Univ, and Purdue Univ. The cooperative network with Nanjing Univ Informat Sci & Technol as the center is the second, and the cooperative institutions mainly included Chinese Acad Sci, China Meteorol Adm and Beijing Normal Univ. The third cooperative network is centered on Texas A&M Univ, and the cooperative institutions mainly include Univ Wisconsin and Harvard Univ. From the perspective of the timeline, NOAA is the core institution for the earliest TCs economic loss assessment research, while Texas A&M Univ, Columbia Univ, Nanjing Univ Informat Sci & Technol and Michigan Technol Univ are frontier institutions for TCs research.

Quantitative Analysis of Authors

Forth, the academic cooperative connections among authors are shown in Figure 7. Tended to cooperate with small groups of collaborators, the authors generated several clusters. However, the research teams of high-yielding authors are scattered, and the research cooperation network has not been formed yet. A total of 678 authors (including all authors of a single article) were counted. The top 16 most productive authors for total publications are shown in Table 3. The major academic contributions, which were concluded in terms of total publications, primarily originated from professor Strobl from University of Bern (12 articles), professor Li from Case Western Reserve University (7 articles) and professor Esteban from Waseda University (6 articles). It indicates that these authors have made outstanding contributions to the research of TCs in different time periods.

Figure 7.

Author Co-Citation Map Related to Economic Loss Assessment of TCs. The academic cooperative connections among authors can be displayed by clustering.

Table 3.

Top 16 Authors With the Most Publications Related to Economic Loss Assessment of TCs.

Quantitative Analysis of Knowledge Bases

Finally, this paper adopted Citespace to plot a reference co-citation knowledge domain map and generated clusters that change over time, as shown in Figure 8. The concept of co-citation was proposed by Henry Small, who is an American intelligence scientist. Co-citation analysis refers to identifying whether two pieces of literature appear together in the references of a third citing piece of literature, in which the two pieces of literature have a co-citation relationship (Small,1973). The knowledge bases of TCs economic loss assessment are composed of citation collection of all literature on this topic. Through the analysis of literature co-citation, foundational knowledge can be revealed from the perspective of the citation trajectory. In the co-citation network, each cluster is identified by a different color, and the deeper the cluster fill color, the earlier the research theme formed. As shown, a total of 43 clusters are obtained related to economic loss assessment of TCs.

Figure 8.

Clustering of the Literature Co-Citation Network Related to Economic Loss Assessment of TCs. The nodes represent different co-cited references, where bigger nodes indicate more co-citation, and each cluster is identified by a different color.

In order to gain a deeper understanding of economic loss assessment of TCs, it is necessary to determine the distribution of the most influential literature, thus this paper collected the top ten articles with the most co-citations. By analyzing literature with high co-citation frequency, this paper can draw hotspots in the field of TCs economic loss assessment. However, the citation relationship of scientific literature is also affected by factors such as the author's social relationship, journal category, and influencing factors, so the analysis only based on the citation frequency will produce bias. By analyzing literature with the highest degree centrality simultaneously, which refers to literature that has important links with many articles, as shown in Table 4, this paper can elaborate key studies more objectively. Through the analysis of these important literature, it can be found that research on TCs economic loss assessment focuses on the application of historical data to carry out the assessment, which has a strong influence.

Table 4.

Reviews of Top 10 Co-Cited Literature With Most Citations and Highest Degree Centrality.

Research Trend Analysis

The research achievements in the field of TCs economic loss assessment are gradually enriched, and the research perspectives and methods are also increasingly diversified. As a highly generalized and research focus of academic topics, keywords are extracted from titles, abstracts and core research perspectives. The co-occurrence network analysis of keywords in academic articles can help to dig out the development trends of the research field more intuitively and clearly.

Using the statistical function of keyword co-occurrence in CiteSpace, this paper sorted out the top 20 high-frequency keywords in the research field of TCs economic loss assessment based on the statistics of key word frequency of the original data. They are climate change, hurricane, impact, tropical cyclone, model, natural disaster, damage, vulnerability, intensity, hurricane katrina, United States, risk, wind, resilience, China, adaptation, risk assessment, typhoon, cyclone and disturbance. According to the annual time zone of TCs economic loss assessment in Figure 9, each keyword node is represented as tree rings, the rings and links are represented in a spectrum of colors corresponding to the years of the keywords’ appearance. Combining the statistical results of keywords with the time zone view of the keyword co-occurrence map, it can be seen that keywords such as climate change, model construction, vulnerability assessment, and risk analysis have long been hot topics.

Figure 9.

Time Zone View of the Keywords Co-Citation Map Related to Economic Loss Assessment of TCs. Each dot represents a keyword, and its node size is proportional to the frequency of the keyword.

Microcosmic Research Topic Analysis

Keywords co-occurrence analysis can describe the core content and structure for certain academic domains, and reveal research frontiers from a large number of articles. This paper used VOSviewer to visualize the results, generated the co-occurrence map of keywords through clustering analysis, and made specific analysis according to the clusters, as shown in Figures 10 and 11. Nodes represent keywords and node size indicates the occurrence frequency of the keyword. Node connection thickness indicates the strength of co-occurrence between keywords. Combined with the keyword co-occurrence network and density clustering, it can be found that the economic loss assessment of TCs can be mainly divided into four main hot clusters.

Figure 10.

Knowledge Domain Map of a Keyword Co-Occurrence Network Related to Economic Loss Assessment of TCs. Each dot represents a keyword, and each cluster is identified by a different color.

Figure 11.

Knowledge Domain Map of a Keyword Co-Occurrence Network With Density-Based Clustering Related to Economic Loss Assessment of TCs. Each cluster is identified by a different color.

Cluster 1 (green): The green cluster is mainly centered on research of TCs on forest ecosystems. TCs disturb forest ecosystems and have the potential to alter forest structure and species composition as well as ecosystem functions (Chapman et al., 2008; Negrón-Juárez et al., 2010). Quantifying these forest disturbances is necessary to evaluate the extent and severity of damage for estimating biomass loss and making management decisions. The main research methods are statistical analysis, remote sensing image analysis, etc. For example, Zeng et al. (2009) synthesized field measurements, satellite image analyses, and empirical models to evaluate forest and carbon cycle impacts for historical TCs from 1851 to 2000. Hochard et al. (2019) used nighttime luminosity to represent temporal trends in coastal economic activity and found that direct cyclone exposure typically results in permanent loss of 5.4–6.7 mo for a community with an average mangrove extent (6.3 m per meter of coastline); whereas, a community with more extensive mangroves (25.6 m per meter of coastline) experiences a loss equivalent to 2.6–5.5 mo.

Cluster 2 (blue): The blue cluster is mainly centered on research of TCs on human health. Research on the influence of TCs on human health includes both the influence on physical impact and mental state. The main methods adopted are statistical analysis, surveys and interviews, etc. For example, Galea et al. (2007) explored the prevalence, severity, and correlations of mental disorders among people exposed to Hurricane Katrina. Kessler et al. (2008) investigated patterns-correlates of recovery from Hurricane Katrina related to post-traumatic stress disorder (PTSD). They found that people were more likely to suffer from mental illness after Hurricane Katrina, such as PTSD (20.9% vs 14.9% at baseline), serious mental illness (14.0% vs 10.9%), suicidal ideation (6.4% vs 2.8%) and suicide plans (2.5% vs 1.0%). Kishore et al. (2018) surveyed 3299 randomly chosen households across Puerto Rico to produce an independent estimate of all-cause mortality after hurricane Maria.

Cluster 3 (yellow): The yellow cluster is mainly centered on research of TCs on social economic systems. The social economic impact assessment of TCs is mainly divided into direct economic losses and indirect economic losses. From the perspective of direct loss research, the impact of TCs on industries and infrastructure (Mitsova et al., 2019; Philpott et al., 2008), quantity and remuneration of labor force (Belasen & Polachek, 2009; Wu et al., 2019), also with a possibility on the overall social economy (Pielke Jr et al., 2003; Wang et al., 2016) is included. The main methods adopted are vulnerability analysis methods, econometric models, along with damage models and artificial intelligence methods respectively. For example, Wen et al. (2018) constructed a damage rate curve based on multiple reference databases and socio-economic data to deduce TCs losses for major economic sectors in China. Deryugina et al. (2018) provided analyses of Hurricane Katrina's long-term economic impact on its victims by using a panel of tax return data. Hallegatte (2007) calibrated the vulnerability of each USA coastal county using data on past hurricanes and their normalized economic losses, and the annual hurricane damage increased by 54% in response to a 10% increase in potential intensity. Shan et al. (2019) collected social media texts sent during and after Typhoon Nepartak and constituted the database for rapid damage evaluations.

In addition, the methods for assessing the indirect economic influence of TCs can be mainly grouped into computable general equilibrium (CGE) models and input-output (IO) models. For example, Chen and Rose (2018) applied a multimodal CGE model to investigate the economic consequences of transportation system failures, and found that the road transportation sector had imposed the largest loss, accounting for 0.6% without resilience by Hurricane Katrina. Lenzen et al. (2019) used a multiregional IO framework to analyze the negative impacts of Hurricane Debbie. The results indicate that the disaster resulted in an additional indirect loss of 3685 jobs and 659 million AUD of value added.

Cluster 4 (red): The red cluster mainly reflects the topic of the influence of TCs under climate change. Recent research suggests that TCs may increase in intensity in the future due to the warming effect of greenhouse gases on surface sea temperature (Kalantari et al., 2019). This will probably lead to changes in the impact of TCs on the economy. Studies that project future losses may give a better indication of the potential impact of climate change on disaster losses and needs for adaptation than the analysis of historical losses (Bjarnadottir et al., 2011; Dorland et al., 1999). The research methods in this field mainly include simulating the running trajectory of TCs and constructing damage models. For example, Esteban et al. (2009) simulated future tropical cyclones by magnifying the intensity of historical tropical cyclones between the years 1978 and 2008 and then used a monte carlo simulation to obtain the expected number of hours that a certain area can expect to be affected by winds of a given strength. It shows the decrease in productivity could result in a loss of up to 0.7% of the annual Taiwanese GDP by 2085. Ye et al. (2019) reconstructed the wind and rainfall fields of historical TCs through an extensive database to extract economic exposure to TC-prone areas in mainland China.

Research Hotspots and Challenges Analysis

Keyword strategy matrix can further reflect the research focus, trend and challenges directly. In this paper, keyword frequency of TCs economic loss assessment was taken as the horizontal axis, and keyword degree centrality was taken as the vertical axis. The mean value of keyword frequency and degree centrality was taken as a dividing basis. Finally, the coordinate system of keyword strategy matrix related to economic loss assessment of TCs was made, as shown in Figure 12.

Figure 12.

Keyword Strategy Keyword Strategy Matrix Related to Economic Loss Assessment of TCs. Keywords in the same quadrant have the same development trend.

Keywords represented by “hurricane”, “impact”, “natural disaster” and “model” are located in the first quadrant and belong to the popular issues, with higher frequency and degree centrality. The research topics represented by these keywords are not only mature in themselves, but also closely related to other research topics, which can be divided into research hotspots of current research. For example, Zhai and Jiang (2014) used a multi-variate least squares regression to construct a hurricane loss model for accurately predicting the impact of super-sized storms.

Keywords such as “damage”, “resilience”, “economic impact” and “recovery” are located in the second quadrant, belonging to the potential hotspots, which are characterized by lower frequency and higher degree centrality. Research represented by these keywords is still in the initial stage of development, but they are related to other topics closely and have the potential to become new hotspots. For example, Ouyang and Duenas-Osorio (2014) introduced a probabilistic modeling approach for quantifying the hurricane resilience of contemporary electric power systems, including a hurricane hazard model, component fragility models, a power system performance model, and a system restoration model.

“Vulnerability”, “variability”, “risk assessment”, “management” and other keywords in the third quadrant, represented by characteristics of lower frequency and lower degree centrality. That is to say, research represented by these keywords belongs to topics that have not been developed yet and have poor connections with other topics. But that also means these topics need more attention, and more scholars are required to keep pushing for breakthroughs in the future. For example, Mitsova et al. (2018) used power outage data to investigate the relationship between the duration of power outage and socioeconomic vulnerability from hurricanes and high-wind events.

Keywords represented by “tropical cyclone” and “climate change” are located in the fourth quadrant, belonging to the marginal zone with higher frequency and lower degree centrality. It suggests that although these topics are well developed, they are less relevant to other topics and may gradually be marginalized. However, research topics represented by these keywords can also be continuously researched in combination with other popular topics or potential issues. For example, Gettelman et al. (2018) adopted a high-resolution climate model and damage model to simulate future economic damage due to TCs in the context of climate change.