Floods (pluvial, fluvial, or coastal) are the response of river basins to heavy rainstorms normally accompanied by a range of devastations, with economic, social, ecological, and environmental impact. Flood damage worldwide has increased considerably in recent decades, mainly due to the steady growth of populations and economic activities in flood-prone areas [1
]. These extreme events affect not only the local population but also the land’s infrastructure and its geomorphology. Several studies focusing on direct flood losses and population risk on a global scale for different levels of warming indicate that flooding will increase in intensity and frequency worldwide [2
]. It is expected that flood consequences in Latin America and the Caribbean (hereafter named LAC) will be more intense due to the exponential, unregulated urbanization of the floodplains, catchment degradation caused by anthropogenic activity, lack of preparedness and resilience for emergency response, the persistence of poverty, inefficient public policies, and infrastructural problems [6
An additional disadvantage for LAC is El Niño and La Niña that, together with the anomalies of the Intertropical Convergence Zone (ITCZ), strongly impact the temporal and spatial distribution of precipitation. El Niño-Southern Oscillation (ENSO) induces strong arid conditions in the northeast of South America and the north of Brazil, promoting the frequency and intensity of forest fires; while torrential rainfalls may hit the coastal areas of Ecuador, Colombia, and Peru, the north of Argentina, Uruguay, Paraguay, and southern Brazil [7
]. Upper-latitude, extreme weather events such as cold fronts, ENSO, trade winds, and tropical cyclones are the most common phenomena that cause flooding in Central America and the Caribbean every year [9
]. Furthermore, landfalling atmospheric rivers, defined as “river” bands of intense moisture transport in the atmosphere, are important systems for delivering heavy precipitation over the coastal regions and are, therefore, precursors of flooding [13
According to the International Disaster Database (hereafter called EM-DAT) of the Centre for Research on the Epidemiology of Disasters (CRED), Université Catholique de Louvain (UCL), Belgium (Available online: https://www.emdat.be/
(accessed on 9 July 2021)), flooding constitutes a major natural hazard, responsible for 45% of the recorded natural disasters in LAC since the beginning of the 21st century. It is not surprising that scholars from all over LAC have studied the risk and hazards of floods from different perspectives. Recently, the scientific community aimed to unravel and understand the dynamics and trends of flooding using advanced methods and technologies. Unfortunately, developed knowledge in LAC is very fragmented, and a bibliometric database of the research related to flooding does not exist, even though a bibliometric analysis would be extremely helpful for the scientists, politicians, and water managers. Such an analysis would offer relevant conclusions regarding the recent evolution of flood knowledge. This would help scientists focus on gaps and support politicians and water authorities in the creation of adequate policies and measures. In this way, it would be possible to prevent and mitigate the effects of floods.
2. Materials and Methods
We performed a bibliometric review of research papers that deal with the risk and hazard of flooding in LAC. The Web of Science (WoS) database was used to identify peer-reviewed articles published during the last two decades (period 2000–2020), using the following selection criteria: [((TI = (flood *)) AND TS = (country_name)) AND PY = (2000–2020)) AND LA = (English)]. Individual searches for each LAC country were performed. In this study we decided to include Puerto Rico in the country list because, although it is not recognized as part of LAC according to the United Nations list, it represents an important location in the Caribbean region. The search for the LAC countries yielded over 1887 references published in the 2000–2020 period. It is worth mentioning that the literature search with the flexible keyword term “flood” yielded a higher number of studies than this review attempts to address. Therefore, the searches were filtered by language, “English”, and by type of document, “Article”, narrowing the search space. Initially, the title, abstract, and keywords were screened to exclude articles that were not useful for the purpose of this study. The following exclusion criteria were applied: (i) papers not covering flood hazard or flood risk; (ii) papers that used the word “flood” in the title, keywords or abstract, but did not include a reference to flood hazard or flood risk in the article; (iii) book chapters, book reviews and book synopses; (iv) conference reports and readings; and (v) editorials and forewords. Differences in opinion during the identification process of relevant papers were reconciled by consensus, after which the text of the preselected articles was investigated.
The selected articles were organized and classified in Microsoft Excel 2019, mainly using crosstabs. From each document, the information was collected according to temporal, geographical, methodological, aim and journal information. When an article covered different methods and aims, each one was accounted for independently. The percentage of publications with each function was quantified with a respective number of studies (n).
Applying the above-mentioned criteria to 302 flood-related research papers, 16% of the originally identified articles were selected for further analysis. Figure 1
depicts the geographical distribution of the selected articles. The analysis revealed that 21 LAC countries feature at least one peer-reviewed publication in English, implying that the remaining 12 countries did not publish any flood-related articles in any journal registered in the WoS database. In proportion, 71% of the countries feature two or more publications, and 29% feature only one publication. The countries with the largest number of published research papers are Mexico, Brazil, Chile, Peru, and Argentina (Table 1
). Puerto Rico, on the other hand, boasts a record of 14 publications. As expected, Mexico and Brazil feature the most significant academic contributions to flood-related studies because of their greater territory and population (47% of the total). The top five countries collectively published 213 articles, representing 70% of the screened articles. The frequency distribution of the articles according to LAC country and region is shown in Figure A1
In this study, publication growth represents the relative increase or decrease in the available statistics over a period. Unlike frequency, which only considered one aggregated period, the total time considered for the data collection was sliced into one-year time windows to calculate the growth of LAC’s flood-related publications. This approach was used to detect sudden bursts or declines by country, journal, and flood-related publications, since this could indicate major milestones or the discovery or failure of a research topic. The country publication growth indicated that most LAC countries intensified their flood-related research activities from 2010 onwards (e.g., Brazil, Chile and Peru, see Table 1
The annual frequency distribution of LAC flood-related articles published in WoS-registered journals is shown in Figure 2
. The analyses of the temporal distribution of annual published articles highlight the degree of interest and development trends in flood hazard and risk assessment research and can serve as a baseline for evaluating the key topics of future research. Figure 2
shows that the number of articles published annually is rising, and the growth rate is increasing. The trend line for the period 2000–2020 fits an exponential curve and corresponds to the following equation: ln(Y) = 0.1774 × X−354.4870, with R2
equal to 0.52 (p
-value ≤ 0.0001). The variable X in the equation represents the “n” year since 2000. If the current publication rate continues, the publication record of flood-related articles might reach 120 and ~300 in 2025 and 2030, respectively. Furthermore, the nonhomogeneous difference equation was used to characterize the development process of flood-related research papers published in WoS-registered journals in detail.
Based on the results, the period of interest (2000 to 2020) can be divided in the following three stages:
An initial period (2000–2005) with a variation between 1.13 and 3.00 (a = Nt/Nt-1 where t is the year, and Nt the cumulative number of published articles), accompanied by a large fluctuation, which was expected for the initial stage in this study.
A unimodal period (2006–2011) in which the fluctuation range was significantly reduced. It starts with a slow growth period from 1.18 to 1.38 and then decreases to 1.13.
A stable development period stretching from 2012 to 2020. In this period, the value of a is stable at about 1.19, and the variance is 0.002. This stage implies that research entered a stable development period, with scholars paying increased attention to this field. According to the identified trend, flood-related studies will remain a relevant topic, and it seems to have been triggered by the years major ENSO events are recorded (2002–2003, 2004–2005, 2009–2010, 2015–2016, and 2019 [15
Various criteria were used for the classification of the consulted articles. First, the journals in which LAC researchers prefer to publish were identified (Table 2
): Natural Hazards, Journal of Flood Risk Management, and Natural Hazards and Earth System Sciences, respectively. These journals comprise 18.5% of the total number of flood-related articles (302) published by LAC researchers in the 2000–2020 period. The top 10 journals in this domain represent 36%, and the remaining 64% are published in the 152 flood-related journals registered in WoS. The latter clearly indicates that the flood-related papers of LAC researchers are scattered across many scientific journals. However, the increase in publications in the top 10 journals over the past ten years reveals an increasing interest on the part of the LAC research community in publishing results in the most prestigious journals in the field.
LAC’s published flood-related articles also were classified according to their altitudinal range based on the Stadel [16
] vertical zonification, considering four altitudinal classes, below 1000 m a.s.l. (A = hot zone), from 1000 to 2000 m a.s.l. (B = temperate zone), from 2000 to 4000 m a.s.l. (C = cold zone), and above 4000 m a.s.l. (D = glacial zone), respectively (Table 3
). The latter combines the frost and snow zones, as specified in [16
]. The survey revealed that 77% of the published articles consist of flood studies in the hot zone, coastal, and lowland or foothill areas (see Figure 1
). As the elevation increases to 1000 m a.s.l. and beyond, a marked decrease in the number of articles is observed. Temperate zones correspond to areas of intermediate mountain slopes with altitudes below 2000 m a.sl., which represent 9% of the analysed articles, and ~11% of the articles deal with research in the high mountain areas in the cold zone. A lower number of flood-related articles are related to areas above 4000 m a.s.l. (~2% of total studies) and correspond to the highest mountain peaks and snow-capped mountain areas.
Keywords constitute an important feature of document retrieval, classification, topic search, and trend analysis, and they provide a glimpse of the article’s content. Several journals do not employ the keyword approach in their articles (e.g., Natural Hazards and Earth System Sciences and Journal of Hydrologic Engineering); as a result, 25 articles do not include keywords. In total, 865 different keywords were retrieved from 277 selected articles, among which 155 keywords (18%) possessed a frequency larger than 1. The highest number of identified keywords and the low frequency of most keywords reflect a high variety in the objectives and methods of the consulted articles. Table 4
lists the 10 most used keywords and, as expected, the keywords “flood/s” and “flooding” featured the highest frequency, but together accounted only for 4.3% of the total number of used keywords. Mexico was the third most frequently used keyword, which is in line with the highest number of Mexican flood-related articles encountered among the total of consulted articles. Overall, the most frequently used keywords were related to flood hazard, risk, and vulnerability in relation to climate change. Surprisingly, the seventh most frequently used keyword was related to “GLOF”, which stands for Glacial Lake Outburst Floods, which shows that a considerable number of articles are related to outburst floods caused by the dam failure of a glacial lake in the Andes mountain range. It must be noted that the 10 top keywords by frequency only represent 12.5% of the total number of keywords. In addition, the 4 most frequently used keywords, together with their frequency, after the top 10 keywords listed in Table 4
, were Haiti (9), flood hazard/s (8), flood management (8), GIS (8; the acronym of Geographic Information Systems), and risk (8).
and Table 6
depict the ranking of the consulted articles in terms of “the aim of the study” and “the method used”. Due to the variety in aims and methods, only the main objective and methodological classes were delineated using the diversity approach. Under this approach, the main aims and methods appearing in the scholarly articles of flood-related research domain are categorized into distinct groups using relevant categorization characteristics; i.e., they are classified from specific to general. A brief description of each class listed in the tables appears in the Annex of this article. Regarding the aim and the method of the study, the articles were classified into five “aim” classes (Table 5
), and six “method” classes (Table 6
). The objectives of most of the consulted articles (class 2 in Table 5
) are related to flood hazards and risk assessment (67.7%). Of this class, 45% of the articles focus on flood risk assessment, 26% on flood hazard mapping, and 9% on historical flood reconstruction. The second most common aim of the surveyed articles relates to the social aspects of flooding (12.4%) with vulnerability assessment being the most frequently cited social aspect (44%). The major “aim” of the third class (class 5) relates to the statistical analysis of floods, representing 10.2% of the surveyed articles. In total, 66% of the articles of this class focus on the development and use of statistical flood forecasting methods. Regarding the methodology used, most published studies applied flood hazard modeling (class 2 in Table 6
), from which 61% of the articles used hydraulic/hydrodynamic models. The next most frequent method class, class 6 in Table 6
, indicates that 23% of the articles included the use of geographical information systems and remote sensing, where 71% used different remote sensing techniques. In the third method class, class 5, the focus of the articles on the use of statistical methods (11.6%) can be observed, 43% are related to multicriteria analysis and decision making.
Based on the conducted bibliographic analysis, it is expected that the number of flood risk studies and research papers will further increase, and that this process will be accelerated by the imperative need for climate change adaptation. according to our understanding of the publication evolution and progress in the period 2000–2020, countries such as Mexico, Brazil, Chile, Peru, and Argentina will continue to lead research around flood risk assessment and the delineation of policies and management strategies to temper the destructive effects of flooding. In addition, the research community should encourage the development of flood risk studies in emerging flood areas such as Costa Rica, Dominican Republic, Haiti, Puerto Rico, Colombia, and Ecuador. It is compulsory to promote and increase flood risk scientific research in the entire LAC region, with specific attention to less developed countries. Moreover, the academic community of the Latin American and Caribbean countries might keep publishing flood-related research findings in a limited number of specialized journals, such as Natural Hazards, the Journal of Flood Risk Management Natural Hazards and Earth System Sciences, Water, the Journal of Hydrology, Hydrological Sciences Journal, Sustainability, Geomorphology, Hydrological Processes, and the International Journal of Disaster Risk Reduction. Parallel to this, a growing acceptance of open-access initiatives sponsored by universities, public, private, and funding institutions will take place.
The presented bibliographic analysis clearly reveals the prevailing trends in LAC’s flood-related research, highlighting the predominant ongoing research based on keywords, aims and methods, altitude, and reviews of the pattern of publication by country. Although the majority of publications in LAC are mostly associated with hydrometeorological analysis, flood hazard/risk assessments, physiographic/geomorphological/and ecosystem approaches, vulnerability and resilience studies, statistical analyses, and GIS/remote sensing methods, we noticed some important gaps as well as new, emerging topics that need to be addressed or expanded in the coming years. In conclusion, we suggest five emerging directions for future flood-related research in LAC: (1) Intensification of the use of machine learning approaches and new satellite imagery products to improve flood prediction and flood early warning systems; (2) Research with a focus on the standardization of post-flood data collection for model validation; (3) Identification of the role vegetation plays in flood episodes; (4) Search for adequate and cost-benefit structural and non-structural flood protection policies and measures; and (5) Analysis of the interaction and effects when flooding occurs at different locations in river networks simultaneously. Based on past and present research, it is to be expected that the flood-related research community in LAC will develop flood risk reduction solutions in a timely manner.