Communication in Disaster—The Contribution of the Press to Highlighting Vulnerabilities: The Case of Rio Grande Do Sul State, Brazil

Abstract

In this article, we explore the role of the media in highlighting an important yet often underestimated aspect of disasters: vulnerability. We use coverage of a disaster that occurred in Brazil’s southernmost state to demonstrate that the effects of a disaster extend far beyond the intensity of the extreme event itself. The concept of vulnerability is increasingly recognized in disaster literature, but the communication factors influencing this concept have not been thoroughly examined. We employ a modern conceptual framework that suggests disasters, such as the one that occurred in Brazil in 2024, stem from two interconnected threats: one natural and one anthropogenic. This second component, often not explicitly mentioned in reports, becomes evident when viewed through the lens of disaster risk. This understanding is beneficial for researchers, policy makers, and disaster experts in systematically identifying the socio-structural factors that affect the impact of extreme natural events. Furthermore, effective disaster reporting can transform how individuals and crisis managers understand hazards and respond to disasters.

1. Introduction

The term “disaster” has many definitions. According to the 2017 UNDRR terminology (UNISDR 2017), a disaster is “a serious disruption in the functioning of a community or society that causes a large scale of human, material, economic or environmental loss and exceeds the capacity of the affected community or society to cope with the situation using its own resources”. While this is the terminology adopted by the United Nations, with which the Sendai Framework is closely associated, it is not the only one. Individual countries and even other international organizations define disasters in different ways. For example, according to Federal Law 14.750 of December 2023, a disaster in Brazil is the “result of an adverse event of natural origin or triggered by human action that affects ecosystems and vulnerable populations, causing significant human, material or environmental damage, as well as economic and social losses” (Brasil 2012).

Another important concept for risk management is the so-called “Early Warning System—EWS”. An EWS is a set of measures, procedures, and protocols that form the core of what, since the beginning of this century, according to the Hyogo framework (Zhou et al. 2014), constitutes the field of “Disaster Risk Reduction” (DRR). EWSs are fundamentally based on four pillars: risk knowledge, monitoring and forecasting, response capacity, and dissemination and communication (Moraes 2023; Trogrlić et al. 2022; Kelman and Glantz 2014).

Risk knowledge refers to the community cognizance of disaster risk (de Silva et al. 2021). It encompasses the awareness of the magnitude of risks, local hazards, and exposure, as well as the susceptibility and the capacity of elements at risk of local hazards, in addition to providing the needed impetus for community resilience and a psychological boost to community participation in DRR and mitigation. Essential aspects of risk knowledge and community resilience are hazard analysis and vulnerability assessment (Odiase et al. 2020), because they enable the community to design adaptive responses to a potential disaster (Bogardi and Birkmann 2004). Aside from adaptive capacity, risk knowledge improves adaptive responses to local risk through community participation in hazard mapping (Cutter et al. 2008; Gaillard and Pangilinan 2010). Risk monitoring and forecasting is about collecting data and information to identify potential hazards and imminent risk situations in order to provide early warnings of disasters. This data includes physical aspects of meteorology, such as precipitation totals and lightning strikes. In terms of response capacity, the forms of local organization and strategies for dealing with announced risks are taken into account (Aven 2016). These capacities are generally linked to the economic, social, cultural, and institutional conditions of the community and are directly related to different types of vulnerabilities, such as economic, political, scientific, and institutional ones (Biswas and Nautiyal 2023; Cutter 2020). Last but not least, there is the communication and dissemination axis; risk communication is about informing and notifying different societal actors, including public agencies at different levels of government and sectors, such as emergency, health, and transportation, as well as vulnerable communities and voluntary organizations. These notifications extend to different spatial and temporal levels about potential threats and vulnerabilities that may have natural and/or technological causes (Hansson et al. 2020; Stewart 2024).

In any EWS, communication plays an important role in the exchange of information. In practical terms and according to Stewart (2024), the science communication lens focuses attention on a trio of practices for DRR: one-way dissemination of risk information to a broad public; two-way dialogues that identify, engage, and consult with specific stakeholders in the risk management process; and three-way participation initiatives that enable informed conversations between communities and decision makers and within communities themselves to motivate action. Communication is also important for raising public awareness about disasters. This includes basic knowledge, types of disasters, disaster-prone areas, disaster preparedness and mitigation, the post-disaster period, including rehabilitation and reconstruction, and contingency plans or sustainable plans to minimize various potential risks of future disasters (Kuran et al. 2020).

In recent decades, the study of environmental disasters has become increasingly important in academic research. Different research perspectives have been adopted and a range of methods have been developed to evaluate different aspects of tragedy (Huang et al. 2011). Reis et al. (2017) have examined the panorama of scientific productions on media and disasters for the period between 1996 and 2016. According to them, coverage plays a role in the construction and contextualization of disasters and climates in social, cultural, and political contexts (Houston et al. 2012). There is now a large body of academic work that seeks to understand how the mass media, in its various forms, is involved in the construction and representation of meanings, narratives, and discourses about disasters and climate change (Antunes et al. 2022; Manatsa and Sakala 2023). Analyzing the media’s treatment of major disasters and extreme events is a highly effective approach to understanding the key links between communication and disaster risk reduction (Houston et al. 2012; Lin et al. 2020; Boholm 2019; Chacowry 2016; Biggs et al. 2018; Saha and James 2017). However, the academic focus has mainly been on mass media from developed countries, and there is a dearth of examples specifically seen in countries of the Global South (Rautela 2016). This article is therefore a contribution to filling this gap by examining how the Brazilian media reported on a major disaster in Brazil. In particular, we focus on identifying the underlying factors that were important to the impact of this disaster. We evaluate the involvement of the authors, as studies have generally shown that the media tend to focus on the dramatic and descriptive qualities of events, i.e., the impacts of disasters on people and the built or natural environment, rather than on their causal explanations (Brüggemann and Engesser 2017). Antunes et al. (2022) examined something similar for the period from 1865 to 2015, but with a focus on Portuguese media for disasters in that country. They concluded that discursive patterns in news production contribute to the naturalization of disasters and to the gap in the public’s understanding of risk by presenting an approach that focuses on relief efforts and ignores the social issues, vulnerabilities, and resilience of the population, and reduces the discourse on preparedness for future disasters. In addition, it is important to keep in mind the media statement presented at the Global Platform for Disaster Risk Reduction (Parida et al. 2021) in Bali, Indonesia, in May 2022, which explicitly calls on the media to “prioritize solutions proposed by disaster risk reduction strategies and programmes and devote more time, space and resources to reporting on the causes of disasters and what can be done to prevent disasters”. Other recent work has looked at the role of press media in a major disaster that occurred in May 2024 in the Brazilian state of Rio Grande do Sul, considering a modern definition of risk of disaster.

In the literature, the risk of a catastrophe is represented by various risk equations that show similarities and differences between the respective definitions of risk (De León and Carlos 2006). In a strict mathematical sense, however, they are meaningless, since in all these equations the only thing that can be expressed probabilistically under all circumstances is the occurrence of a hazardous event. Recently, however, Moraes (2023) has shown that the risk of a catastrophe can be represented by a mathematical representation in which the probability of a catastrophe is the product of two other probabilities. One is the probability that an extreme natural event such as an earthquake, hurricane, cyclone, etc., will occur, and the other is the probability that this event will occur in a place where there is anthropogenic vulnerability, which includes so-called social, cultural, technological, political, and ethnic vulnerabilities, as well as so-called exposures, response capacities, education, etc. In other words, the entire range of vulnerabilities for which humans are responsible.Vulnerability, with the immense range of possibilities described by Hansson et al. (2020) and the possibilities included in the United Nations’ conceptualization, is therefore represented by the idea that it is people who make decisions about how to increase or decrease the potential impact of an event over which they have no control.

This work explores risk not just as a general concept, but through a specific operational definition: the combination of an atypical natural event with pre-existing vulnerabilities. The disaster in Brazil in May 2024 was a combination of a completely atypical natural event that occurred in a region where several vulnerabilities were present. In other words, the realization of a natural hazard in the face of anthropogenic hazards that had built up over the years. The focus of the work is on how the media, through its usual form of communication, described the presence of these anthropogenic hazards in this region. Consequently, a secondary aim of the work is to contribute to the development of a resilient environment by exposing vulnerabilities and thus guiding action to reduce the risks of future disasters.

This paper is organized as follows: First, there is a brief contextualization of the importance of the communication carried out by various media, not only for the coverage of the event but also with regard to the relevance of this component for risk management. This is followed by a description of the disaster that occurred between late April and early May 2024 in the state of Rio Grande do Sul, which includes the meteorological phenomenon and its effects. The analysis method is included in the following section, which also presents the results. Before the comments and conclusions, there is a section in which several vulnerabilities, taken directly from the press, are exemplified.

2. Communication in Disaster

According to Spence et al. (2007), communication in times of crisis aims to prevent or reduce the negative consequences of a particular event and fulfills two main functions: informative and persuasive. First, the messages should create a rational understanding of the risk, and, second, encourage the public to take action to avoid a potential threat or mitigate the consequences of such events. Information, or a lack thereof, can positively or negatively influence all phases of the disaster. In this sense, the media plays a crucial role in communicating and understanding disasters as well as their impacts (Pantti et al. 2012). According to Quarantelli (1991), most of what people know about disasters is what they learn through the media. In this sense, the role of the media in disasters can be understood not only as a tool for communicating and describing an event and informing the public. Guion et al. (2007) argue that the media is not only one of the most important means of informing people about certain risks and dangers, but that it is also used extensively during the different phases of a tragedy. In this sense, Leitch and Bohensky (2014) point out that the media should also contribute to individual and community preparedness, help identify potential threats, allow communities to tap into local potential and experience to adapt to crises, disasters, and other challenges, and provide a forum for community planning for post-disaster recovery. Media coverage can promote disaster prevention and risk-reduction policies on the public and political agendas. This view is based on the conclusions of several previous studies that the media is the most important risk-reduction tool available to authorities during a disaster, as its actions influence public perceptions of the risks of the event (Miles and Morse 2007; Pérez-Lugo 2001). Furthermore, Mattedi and Ludwig (2016) found that the occurrence and intensity of disasters depend more on the degree of vulnerability of the disaster scenarios and the affected communities than on the magnitude of the negative events. In principle, the perspective adopted here is consistent with Mattedi and Ludwig’s (2016) postulate, but also that of Baker (2009), that tragedies are socially constructed events and that vulnerability to risk is configured as a dynamic process that depends on a range of contextual factors. The media dynamic leads to an approximation of disaster by relating the news to the public’s experiences and concerns. Susmayadi (2014) also believes that behind the journalistic imperative to cover the tragedy, the media must have an interest in creating and defining an agenda that places disaster risk reduction at the center of local attention, as the media serves to disseminate information, direct, educate, persuade, and reduce people’s concerns.

3. The Rio Grande Do Sul State Disaster

In just over a week, at the beginning of May 2024, in Brazil’s southernmost state, entire neighborhoods in more than 400 municipalities, equivalent to around 90% of all municipalities in the state, were swallowed up by unstoppable rainfall. The biggest climate tragedy in the history of that state, it caused at least 170 deaths and affected more than 2 million people, more than 35 thousand of whom became refugees. The companies located in these municipalities accounted for 87.2% of industrial jobs in the region. A study carried out by the IDB (Inter-American Development Bank), the World Bank, and ECLAC (Economic Commission for Latin America and the Caribbean) calculated that the economic impact of the floods in Rio Grande do Sul this year was USD 15 billion, representing an impact of 15% and 1.8% of the state and Brazil’s GDP in 2024, respectively. In the Brazilian case, the impact at the national level will be much greater than the impact of Katrina in the United States, because the economy of the state of Rio Grande do Sul represents 6.5% of Brazil’s GDP, while Louisiana represents 1% of the American economy.

The extent of this catastrophe can be seen from some shocking figures: The international airport in the state capital resumed operations in October 2024, six months after its closure. Rail transport, which connects the same capital with the municipalities in the region, was not expected to resume normal operations until December. According to the state government, 13.7 thousand kilometers of roads were affected and, more than a hundred days after the climate tragedy began, the disruptions on federal and state roads continued to cause logistical difficulties for businesses and the state’s population.

The Natural Hazard

According to Marengo et al. (2024b, 2025), a total of 29 billion cubic meters of water fell into the Jacui, Pardo, Caí, Sinos, Vacacaí, Gravatai, and Taquari river basins areas between 27 April and 7 May (Figure 1). This amount was caused by the combination of several factors that normally influence rainfall in the region, which in this case acted simultaneously to create a “perfect storm”. The state is located in the southernmost region of Brazil and is often influenced by dynamic weather systems due to its geographical location. At certain times of the year, such as austral autumn and austral winter, atmospheric troughs can form over the region, bringing intense winds and atmospheric instability (Schumacher and Silva 2016). The state can also be affected by the influence of moisture corridors that extend from the Amazon to southern Brazil. This phenomenon, often referred to as the South American Low-Level Jet (SALLJ), transports warm and humid air from the Amazon basin to the southern regions of Brazil, contributing to increased precipitation and extreme weather events in these areas (Marengo et al. 2024a). Another system that increases precipitation in the region is the El Niño phenomenon (Moraes and Marengo 2023). In Rio Grande do Sul, heat waves characterized by high temperatures and stable weather conditions can occur during the summer months. In such situations, an atmospheric blockade can form over the state, preventing the passage of cold fronts and weather systems that would bring thermal relief and regular rainfall. As a result, the remote areas of the state may receive more intense and concentrated precipitation, while the central region may remain relatively dry and hot for long periods of time (Stefanello et al. 2022). However, the precipitation systems in this region are also strongly influenced by the typical biome of southern Brazil, Uruguay, and northwestern Argentina (Moraes 2000; Souza et al. 2019). According to the National Center for Monitoring and Alerting of Natural Disasters, CEMADEN, the combination of these four factors led to heavy rainfall in the state of Rio Grande do Sul.

4. Press Coverage of the Disaster: Uncovering the Anthropogenic Hazard

According to Barnes et al. (2008), the ability of the mass media to set the agenda for public discussion is known as agenda-setting. Agenda-setting influences the public agenda and policy through the targeted coverage of events and issues, with the media prompting policy makers to take action and satisfy the public interest or demand for responses. Compared to the response to an earthquake, for example, the response to a hurricane often “lacks a well-organized community policy and therefore consists largely of ad hoc disaster relief episodes” (Barnes et al. 2008). In the specific case of disasters, the media generally tends to prioritize differently depending on the nature of the disaster or the people affected (Birkland 1997).

A study by Ball-Rokeach and Loges (2000) found that most media coverage of Hurricane Katrina focused on the government’s response and less often addressed the level of preparedness or the responsibility of individuals and communities. Thus, more articles reported on response and recovery than on mitigation and preparedness. Based on the newspapers studied, the authors concluded that the news agenda focused significantly more on the government response than on key disaster management functions. In some ways, this underscores the thinking of Susmayadi (2014), that the media must be interested in creating and defining an agenda that places disaster risk reduction at the center of local attention beyond the journalistic imperative of reporting on the tragedy.

4.1. Methodology

A key challenge for the large-scale content analysis of disasters in the Brazilian context is the lack of a validated computerized lexicon. To fill this gap and ensure a transparent and rigorous methodological process, we developed our own lexicon of keywords. This process was not arbitrary, but rather a systematic attempt to translate disaster research theory into an operational analysis tool. Initially, the selection of terms was strictly anchored in official and recognized sources. We selected words and expressions taken directly from the terminology manuals of the United Nations Office for Disaster Risk Reduction (UNDRR)—a global standardization—and from the Brazilian Classification and Codification of Disasters (COBRADE), which adapts these concepts to the national context. This has ensured that our lexicon reflects the technical and scientific consensus in the field. Secondly, the terms have been organized into a hierarchical and conceptual structure. Instead of a flat list, the keywords have been classified according to the phases of the disaster, which form the theoretical pillar of this article. Following this procedure, the content was categorized into six themes: The first refers to the precursors of the disaster, such as rainfall. The next is the consequences of the hazardous event forecasted, which means the alerts, communication, and preparation. Next comes the disaster itself. Here we included keywords to recognize that the disaster has occurred. The characterization of the disaster, which takes the form of flooding, inundation, flash floods, and landslides, is included at this stage. The magnitude of these events depends on the anthropogenic hazards, i.e., the vulnerabilities. For this reason, we have included this topic in this timeline. Finally, we have included information on disaster impacts. This structured dictionary was used as a conceptual “sensor” with a dual function: to search and identify relevant messages and to systematically categorize mentions of each topic within the message corpus and quantify frequency/repetition. In this way, we transformed unstructured textual data into quantitative and classified data that allowed us to objectively map and measure the focus of media coverage. This approach ensures that our analysis is not subjective, but a reproducible process that is firmly rooted in disaster research theory.

From a practical point of view, any message was cataloged into one of these six groups. The result was a table listing the message, the date it was published, and the source from which it originated.

4.1.1. AI Analysis

The first approach had two steps: search and identification of relevant news, and extraction and processing of news content.

To identify relevant news, the free version of GNEWS API was used. It is easy to use and based on artificial intelligence, providing real-time access to news from around the world (GNews API 2024). The search was conducted using a list of 124 keywords related to the disaster and developed with an exploratory bias, including “Rio Grande do Sul 2024 disaster”, “RS 2024 floods”, “Rio Grande do Sul public disaster”, etc. The inclusion criteria included news published in Portuguese from May to September 2024.

The choice of GNEWS is related to the possibility of filtering news by language, publication period, and keywords, which allows the targeted collection of news relevant to the disaster in question. To interact with the GNEWS API, we used the Python (V3.12.0) programming language, which provides a simplified interface and a rich collection of libraries for manipulating, processing, cleaning, and visualizing data (McKinney 2010). The collected messages were stored in a DataFrame using the Pandas library. The search process resulted in the identification of approximately 1500 potentially relevant messages, of which 1309 were used for processing after an initial visual assessment and search for inconsistencies. This structure allowed for the efficient organization and manipulation of the collected data, including title, description, URL, publication date, and media company.

The Diffbot API, an artificial intelligence-based web scraping and content analysis tool, was used to extract the content of the identified news items. Diffbot was chosen because of its ability to extract structured content from webpages with high precision, even with complex and diverse layouts, using natural language processing (NLP) techniques. The extraction process was implemented through a Python script that uses the requests libraries to make HTTP calls to the Diffbot API (Diffbot 2024), Pandas to manipulate data and store it in table format (McKinney 2010), and tqdm, short for “taqaddum”, which means “progress” in Arabic, to visualize the progress of the extraction (Casper and Dane 2024).

Throughout the data collection and extraction, the terms of use of the platforms used and the robots.txt guidelines of the news sites were observed, and the best ethical practices of digital research were followed (Zimmer 2010). The extracted data was saved in a file in XLSX format (Excel) to facilitate subsequent analysis and processing. Each row of the file corresponds to a news story, with the columns representing different attributes.

The 1445 articles automatically identified by the search engine are spread across almost 200 news sites, ranging from sites of large communication networks to sites of microenterprises. All identified and extracted articles required an additional process of classification by topics of interest related to the natural event, warnings, disaster, vulnerabilities, and impacts. In other words, the classification followed the logic of the normal sequence of events of a disaster.

The process of identifying and automatically extracting 1309 items of news from almost 200 websites required a refinement of these elements, as many of the websites not only could not be identified in terms of their media channel (printed newspapers, online newspapers, news sites, radio stations, etc.) but also belonged to official institutions that used their communication channels for official information. These channels were eliminated, leaving 992 messages that were cataloged between the months of May and September. These 992 items are distributed across 106 essentially journalistic news channels. All are edited and published in Portuguese and in Brazil. Geographically, they are spread across the entire national territory.

From the texts of these 992 reports, we were able to assess how some risk management processes were treated in the press based on the occurrence of keywords. For example, the weather system that triggered the disaster appears 1018 times. To be included in this group, the following words were considered: rain, precipitation, storm, severe weather, extreme weather event, etc. A summary of this procedure is presented in Figure 2.

Of the total 6050 words cataloged in these six groups, the group of “IMPACTS” is the one that contains the largest number, accounting for 36% of the total. In second place is the typology cluster (floods, inundations, flashfloods, and landslides), with 1362 keywords. The natural hazards, i.e., the triggers that caused the catastrophe, account for 17% of the total, while the anthropogenic risks, resulting from the failure of the flood forecasting system and lack of urban planning, account for another 9%. If we consider that this 9% can be added to the impact group, it becomes clear how much human activity determines the extent of the impact, and you realize that it is based on four pillars.

The evaluation of the messages cataloged with this method is very simple and fast, but may not correspond to the reporting of various media outlets on the disaster. As mentioned above, the 992 news items are located on 106 websites. Considering that these news items were cataloged through a search that spanned 150 days, which is an average of 6 news items per day, the statistical representativeness of the analysis is not significant.

This search by the IA method cataloged 1309 news stories published by both the mainstream press and small media outlets between May and September 2024. There were 730 news stories in May, 499 in June, 37 in July, 24 in August, and 19 in September. A simple visual inspection of the file in XLSX format revealed that this method was not robust enough to be used for the purposes of this work. For example, the AI only cataloged two reports in these 5 months by the local newspaper O SUL, and the authors are aware that this newspaper publishes a much larger number of reports. The authors suspect that the free version of the AI application has some limitations that prevent a better overview of how the press is dealing with this disaster. This was the main motivation for using the “manual” methodology. It was the limitations of this search method and thus the impossibility of a more-objective analysis that motivated the use of the human and traditional search method.

4.1.2. Direct Analysis

Considering that the process described above, although agile, simple, accessible, easy to use, and allowing for the verification of the consistency of the objective of this work, is limited for a more-robust assessment, the process described in this section has been adopted. This process is of course more time-consuming and laborious, as it required hours of manual work. It consisted of analyzing the news published in four daily newspapers. Two of these newspapers have a national circulation. One of them is based in the city of Rio de Janeiro and belongs to the largest newspaper group in Brazil: O GLOBO, owned by the Globo organization. The other, Folha de São Paulo, is based in São Paulo and also has a large national circulation. In addition, two daily newspapers in the capital of the state of Rio Grande do Sul were also included: ZH and O SUL.

The two national newspapers analyzed, Folha de São Paulo, based in the city of São Paulo, and O Globo, based in the city of Rio de Janeiro, carried 205 and 148 headlines about the disaster in the state of Rio Grande do Sul, respectively. In Folha de São Paulo, they were on the front page for 20 days. In O Globo, they were on the front page for 19 days. These figures give an idea of how this disaster was covered by two of Brazil’s largest traditional newspapers. The above figures do not include the articles for which the regular columnists are responsible.

As expected, the coverage of the disaster in the two newspapers based in Porto Alegre, the state capital, was more extensive. In the ZH newspaper, the number of headlines was 704, while in the O Sul newspaper it was 1223. It is worth noting that the ZH newspaper does not have a Sunday edition, and the O SUL newspaper only has a digital version. It is also worth noting that the disaster was on the front page of all May editions of both newspapers.

The number of reports published in these four newspapers on each day in May is shown in Figure 3. Each color represents one of the newspapers. As mentioned earlier, the number of articles in the local newspapers is much higher than in the national newspapers. While on some days the local newspapers focused almost exclusively on the disaster, on some days the national newspapers were not as focused on the disaster. Note, however, that on some days only two national newspapers contained more than or close to 20 stories, suggesting that coverage of the event was much greater than that captured by the previous method.

A total of 2280 published messages were analyzed individually and divided into the six groups listed in

Table 1. Keyword dictionary created for searching with the AI method. This dictionary was created based on the UNDRR terminology and the Brazilian risk management system. The keywords used for the AI method were also used for the manual method.

Disaster Phases	Cluster	Keywords
Natural hazards	Hazard, Hazardous event, Risk, Risk management, Risk reduction	Rain, precipitation, storm, cyclone, meteorology, climate, forecast
Early warning and preparation	Early Warning, Structural and Non-Structural Measures	Alert, Warning, Communication, Prevention, Planning, Management
Disaster	Disaster	Tragedy, catastrophe, disaster, civil defense, volunteer, response, rescue, recovery, cleanup, help, crisis, situation, blockade, interdiction, interruption
Events	Typologies	Flood, Flooding, landslide, Flooding, flashflood, Rivers, Streams, Lagoons, Level, Quota, Flow, Volume, Overflow, Taquari, Guaíba, Lagoa dos Patos, Caraá, Maquiné, Caí, Sinos
Antropogenic hazards	Vulnerabilites	Pump houses, dikes, piers, dams, reservoirs, manholes, public buildings and works, urban planning, slums, sanitation, utilities, water, sewage, waste, infrastructure, highways, roads, streets, bridges, transportation, busses, airport, hospitals, energy, electricity, communications, telephony, internet, politicians, governor, mayor, mistakes, guilt, denial, lies, fake news, access, mobility, traffic, home, commerce, business, stores, land, city government
Impacts	Human and Economic Damages	Affected, victims, people, families, merchants, animals, dead, missing, injured, sick, homeless, displaced, losses, damages, destruction, agricultural losses, production, farming, livestock, health, diseases, epidemics, contaminated water, food, security, theft, robbery, crime, assaults, wealth

. However, 219 of the 2280 messages did not fit into any of these six groups. Figure 4 below shows the distribution of the remaining news coverage in the four newspapers. As we can see from this figure, of the 2061 remaining news items, the stage that was quoted the most in all newspapers was related to the impact of the disaster. However, another important aspect is that the second-most-quoted stage is related to the anthropogenic hazards. This was not apparent when the analysis was carried out by artificial intelligence. In that case, anthropogenic hazards were cited less frequently than the impacts, typologies, and natural hazards. It is astonishing that the national and regional newspapers similarly highlight weaknesses that are certainly recognized, such as the drainage system, the lack of urban planning, and the absence of integrated disaster risk management.

The manual process enabled a refinement in which the keywords were carefully extracted from the texts. The results of this process can be found in

Table 2. The 2280 news items published in the four newspapers were classified into the 6 groups considered and 19 clusters. Numbers in the columns represent how many times each of the groups was covered in a report by that group by each newspaper.

Cluster	Keywords	ZH	FSP	SUL	GLO
NAT HAZARD	Rain, Storm, Cyclone, Meteorology	31	6	60	6
NAT HAZARDS	Climate change	10	5	20	4
EW & PREPARATION	Planing Managment	29	8	32	2
EW & PREPARATION	Warnings and Alerts	17	4	35	1
DISASTER	Disaster, catastrophe	6	4	12	5
DISASTER	Response, rescue, civil defense volunteer, help	105	20	188	16
DISASTER	Government aid	35	10	48	5
TYPOLOGIES	Floods, inundation, landslides	34	16	79	5
TYPOLOGIES	Rivers, Lagoons, Overflow, Taquari, Guaíba, Lagoa dos Patos, Caraá, Maquiné, Caí, Sinos	20	6	47	5
ANT HAZARD	highways, roads, streets, bridges, transportation, busses, airport, train	72	18	99	15
ANT HAZARD	Pump houses, dikes, piers, dams, reservoirs	52	10	60	10
ANT HAZARD	Unplanned urban, mobility	29	12	39	4
ANT HAZARD	Lack of institutional linkage	16	7	26	6
IMPACTS	Animals, livestock, farming	15	5	20	4
IMPACTS	Public security system, theft, robbery, crime, assaults, wealth	20	5	58	4
IMPACTS	health, diseases, epidemics, hospitals	15	3	30	3
IMPACTS	Deaths, missing, displaced, injured, homeless	52	15	101	15
IMPACTS	Water, sewage, waste, energy, food	31	13	56	8
IMPACTS	Economic impacts, agricultural losses, industrial losses, commercial losses	52	19	67	15
OTHER	ST and education	21	6	45	7
OTHER	Other	42	13	101	8

and Figure 5.

Of the total 2061 messages, 475 are clearly related to the vulnerabilities that determined the extent of the impact. These are related to the failure of the drainage system, the lack of urban planning, and the lack of institutional coordination for integrated risk management. The impact category contains 626 messages. The largest vulnerabilities were in the transportation sector, particularly at the international airport, on highways, bridges, and the urban train system. Highways that pass through areas prone to landslides or flooding, bridges that were built but are not resistant to strong river currents, and even railroad tracks laid on land without proper soil treatment are impacted because they are vulnerable.

In the impact category, the cluster related to fatalities, missing persons, and homeless as well as displaced persons was the most frequently reported cluster, followed by the economic impact and water, food, and energy security clusters.

The disaster typologies, i.e., floods, inundations, rainfall, and landslides, accounted for 212 news items, while the trigger, the rain events, accounted for 142 news items. These figures alone show that the traditional press has recognized that it is more so the vulnerabilities than the extreme event that determine the impact of the event.

4.2. Vulnerabilities Uncovered by the Press: Some Illuminating Examples

The vulnerabilities of the disaster in the state of Rio Grande do Sul have been exposed by the press and will be illustrated in this session with examples. As all the news was written in Portuguese, the translation into English was carried out with an appropriate and careful process.

4.2.1. Technological Vulnerabilities

Perhaps the most glaring and well-known weakness was the case of the flood protection systems in Porto Alegre, the state capital, and in the surrounding municipalities. This system failed for a number of reasons beyond simple maintenance. This vulnerability was one of the most investigated and publicized in the press. Below are some examples of articles published by some of the major news organizations reporting on this disaster. These examples are some excerpts from the total of 132 reports contained in the analyzed collection and relate to the failure of the flood protection system.

An important part of the flood protection system is the 23 rainwater pumping stations (Ebaps), better known as pump houses, which are located in Porto Alegre. They are spread across various points in the city and operate with motors that allow the water collected at certain points in the capital to be drained away. On 13 May, during the first days of the long period in which the city was flooded, the newspaper ZH published the following article: “Why the POA was flooded: Maintenance and construction errors and neglect of the basic waterproofing system raised the capital’s flood level” (Gonzato 2024).

The need to maintain this system, which is designed to drain water and prevent flooding, was not unknown to the city administration. In its May 27 edition, the newspaper ZH also published the following report: “DMAE’s actions will be investigated: The city government has ordered an investigation after it was revealed that there were warnings in 2018 and 2023 about failures in the flood protection system” (Moreira and Mathias 2024).

Fifteen days after the floods began in the state capital, the website G1, linked to the Globo Organization, published a report entitled “Why the flood protection system in Porto Alegre didn’t work. The system, which was supposed to withstand floods of up to six meters, was not properly maintained for decades” (Pontes 2024).

On 26 August, CNN Brasil published “Flooding in the RS: A report points to structural deficiencies in dikes, sluices and pump houses”. This article also showed severe weather conditions never before recorded in Rio Grande do Sul, such as excessive rainfall and the presence of winds that dammed the water (Netto 2024).

The floods also affected other municipalities in the capital’s metropolitan region. In some of these municipalities, it was not the failure of the pumping stations that exacerbated the effects of the disaster, but the rupture of a system that is just as important as the sewage system: the sluice system. On 7 June, a month after the flooding began and water levels had returned to normal, a GZH report showed another side of the system that cannot be ignored. “VIDEO: Aerial footage shows broken dykes in the metropolitan region”. Furthermore, the text of the report warns how this vulnerability was exposed: “The images taken on May 22 show how the structures in Canoas and São Leopoldo were unable to contain the floods and how far the water reached, inundating entire neighborhoods” (Rocha 2024).

4.2.2. Infrastructure Vulnerabilities

Nothing could be more fitting than to begin this session with the headline on the front page of the May 26 edition of the newspaper O Globo: “After the RS, Brazil needs a comprehensive review of infrastructure risks”. The details of this article can be found on page 19 of this day’s edition, with the title “Off the radar. Extreme weather conditions force a review of contracts and risk maps for infrastructure concessions” (Freitas 2024).

Infrastructures such as highways, railways, bridges, and airports represent not only vulnerabilities but also sectors that are affected by the disaster. Poorly planned highways, railways, and bridges built or passing through risk areas represent typical examples of well-known technological vulnerabilities. However, in the case of the state of Rio Grande do Sul, another strategic infrastructure, the airport, was significantly affected. The airport was partially reopened on 21 October, approximately 6 months after the anomalous precipitation. Regarding this transportation infrastructure, a total of 94 reports were published by the four newspapers and cataloged for this work.

Below are more examples describing how the newspapers described failures in basic infrastructures. In other words, if you look at all the news in the four journals related to the impact on the transport sector, about 45% was dedicated to the airport.

The vulnerability of the land transportation sector is also illustrated by some revealing reports. For example, on May 26, about 3 weeks after the heavy rains, the newspaper O Sul published the following article: “In the RS there are 115 points where highways are completely or partially blocked” (O Sul 2024b). On the front page of the newspaper Zero Hora on May 25, there is a very revealing headline: “Floods have affected at least 2,700 kilometers of roads” (Aires 2024).

But rail transport was also affected. The train that connects the state capital with the neighboring municipalities, the so-called Trensurb, was not fully operational again until December 2024, 8 months after the heavy rains. On 22 May, the newspaper Folha de São Paulo published an example of how not only the Trensurb was impaired; it was a weakness that brought the system to a standstill: “The mud softens the ground under the tracks and the trains have no date for resuming service” (Vilela 2024).

Finally, the fragility of the bus transport infrastructure is illustrated by the press when we look at the news published by the newspaper Zero Hora on May 22: “The bus station still has no forecast for reopening” (Costa 2024). The main part of the article reads as follows: “The site is covered in mud, the surrounding roads are flooded and the electrical and hydraulic systems are destroyed”. In other words, three weeks after the tragedy began, there was no bus service in the state capital to any other city in the state, country, or abroad.

4.2.3. Unplanned Urban Development as Vulnerability

Among the most-glaring problems, perhaps a reality in most Brazilian municipalities, is the lack of urban planning that makes cities resilient and prepared or points to adaptation policies, albeit minimal ones. The Folha de São Paulo newspaper includes a report in its May 25 edition that highlights this vulnerability: “Documentary shows tragedy and lack of planning after floods in RS” (Goulart and Witzel 2024). No less impressive is the blog by journalist Rodrigo Lopes in GZH, which was published almost a year before the disaster. On 16 June 2023, Lopes wrote a text whose title read “Rain doesn’t kill. What kills is the lack of urban planning” (Lopes 2024).

On 8 May, the newspaper O Globo published the news “City with 40,000 inhabitants must be evacuated” (Azevedo and Ribeiro 2024). This news refers to the municipality of Eldorado do Sul. The article is illustrated with an excerpt from an interview with the mayor of the town, in which he explains that reconstruction will take a year. In this context, however, a brutal statement is in order. It is not enough to rebuild the city. You have to rebuild properly: rebuild better. This municipality was built on the banks of the Guaíba River, in an area that was used for irrigated rice plantations. In other words, the average height of the community in relation to the level of the river into which all the catchment areas flow (see Figure 1) is less than one meter. In addition, it crosses the BR116 highway on the opposite side, which is about 5 meters high. In other words, the highway serves as a dike. The city was built on a basin that is just waiting for the water to come.

The lack of urban planning, which is now seen as a hazard for unplanned urban development, is illustrated very well in the article in the newspaper O Sul from May 29: “Companies create manual for the reconstruction of buildings in RS” (O Sul 2024a). Twelve companies have joined forces to create a guide that provides guidance for safe and resilient construction in similar events.

Finally, we can use the article in the Folha de São Paulo newspaper of 19 May as an example in this section: “In almost two centuries, Porto Alegre has filled the Guaíba River and tripled the size of the city center” (Zylberkan et al. 2024). In other words, the expansion of the city center occurred with the advance over the natural riverbed.

4.2.4. Disaster Risk Management Vulnerabilities

To conclude the session with examples of vulnerabilities that represent the boundary conditions—limited or unlimited—for the extent of the impact, the lack of a clear risk management policy should be considered. The lack of a coordinated and articulated system where all decision-making bodies know exactly what they have to do, what their responsibilities, protocols, goals, and objectives are, can exacerbate not only the delay in recovery but also in preparedness. It is not uncommon for idiosyncrasies to emerge during a tragedy in the vacuum of a missing risk management plan. Three examples will serve to illustrate this.

The newspaper O Globo published an article in its May 15 edition titled “Disagreements on the front lines” (Gularte 2024). This article was subtitled “Task Force to Aid RS causes discomfort in the Planalto and provokes confusion among Ministers”. It should be noted that Planalto in this text refers to the presidential palace.

Another informative article was published by the newspaper Folha de São Paulo on 9 May, entitled “Lack of authority hinders disaster relief” (Petrocilo 2024). According to experts, Brazil lacks an institution that plays a role in risk management, and civil defense measures are insufficient. It is also mentioned that the available resources are not only scarce, but also scattered.

Finally, there was also an article in the Folha de São Paulo on the 13th entitled “Congress controls the budget of the ministry working on a disaster” (Marchesini and Vargas 2024). Importantly, the article highlights the way in which the Ministry’s budget is determined according to the interests of politicians who are not involved in the administration. Funds are allocated for tractors and paving work in voter strongholds, i.e., places that can bring votes at election times.

5. Comments and Conclusions

The aim of this paper was to emphasize that disaster risk is a combination of natural and anthropogenic hazards. Indeed, many disasters are caused exclusively by anthropogenic actions, as demonstrated by the well-known disasters of Mariana and Brumadinho, also in Brazil, in 2015 and 2019. However, in the specific case of hydrometeorological disasters, it is not only the intensity of the meteorological event that determines the severity of the impact. This was one of the objectives of this paper. Considering that there are multiple and possible vulnerabilities caused by humans, this article also tried to emphasize the importance of the traditional press as a tool to promote accountability for, albeit indirectly, the implementation of risk management measures. In other words, public officials should be reminded that it is not just a matter of monitoring the weather and issuing severe weather warnings, but that it is necessary to take action to reduce vulnerabilities, because only humans can act on human constructions. Nothing could be more illustrative of an anthropogenic hazard in disaster risk than the state governor’s interview in which he said “Studies have warned, but the government has other plans, too.” The state’s highest-ranking official would hardly have made such a statement had he not been asked by the press. This article, which was published on page 14 of the Folha de São Paulo newspaper on 20 May, shows how important communication is for the implementation of disaster risk-reduction measures.

It is still unclear to what extent decision makers have internalized the academic findings. On the other hand, there is a wealth of research showing that the dissemination of research knowledge in practice is often incomplete or confusing. The gap between academic studies and political practice is widening. Policy decisions require that the right information is disseminated to the right people at the right time, and these tasks are often not considered part of the role of academia. In knowledge transfer in particular, it is expected that the results are concrete, that the amount of evidence is high, that the statistical methods are easy to understand, and that there are clear practical implications. Above all, it is important that the practical implications are clear. Herein lies another strength of this article: part of the academic activity related to risk management is to identify existing weaknesses and vulnerabilities in order to guide the actions of policy makers. When the press exposes such weaknesses, academia can provide the right guidance.

Successful disaster risk reduction has for many years demonstrated the need for an improved interface between science and policy. This depends on recognizing science as a process to provide a basis for decision-making and to identify optimal strategies and necessary countermeasures. Furthermore, scientific knowledge enables more-targeted investment in disaster risk reduction. For academics, therefore, there is nothing more confusing and frustrating than the fact that decision makers often fail to learn from experience. Implementing or adopting measures that significantly reduce the risk of future losses seems to be at odds with the scientific evidence. With rare exceptions, the result is that the same houses and businesses are rebuilt in the same places that were flooded or affected by landslides. The Brazilian news media has recognized this weakness.