Changes in Human Vulnerability to Flood and Landslide: Evidences from Historical Data

Abstract

Human impact by floods and landslides (FLs) is a significant concern, necessitating a deeper understanding to implement effective reduction measures, in line with the Sendai Framework for Disaster Risk Reduction’s goal to reduce disaster mortality between 2020 and 2030. This study examines the evolution of human interaction with FLs over the past 70 years in Calabria, Italy. By systematically analyzing regional newspapers and historical archives from 1951–1960 and 2011–2020, a database was created documenting fatalities, injuries, and the involvement of people in FL incidents. For each victim, the database includes demographic details, accident time and place, circumstances of death or injury, and whether the victim’s behavior was hazardous or protective. Results indicate a drastic reduction in both the total number of fatalities (196 versus 20) and high mortality events from 1951–1960 to 2011–2020 (6 versus 1). However, the number of people involved in incidents has increased (202 versus 1102), although this may be partly due to improved dissemination of information. Changes in population habits and the construction of more robust houses have significantly reduced high-fatality events, enhancing security. The study highlights the importance of data collection for developing locally tailored risk reduction strategies, increasing community resilience by addressing specific vulnerabilities and strengths.

1. Introduction

People’s vulnerability to hazards such as floods and landslides (FLs) is a pressing problem for both developed and developing countries. This vulnerability is expected to increase due to the rising frequency of extreme meteorological events triggered by climate change. Many processes, including climate change and globalized economic development, are creating new interconnected risks, which during major events can lead to dangerous cascading effects. Some papers in the literature face the impact on people of the whole of phenomena occurring during bad weather periods [1], and many others focus on the risk for people related to a specific hazard source, as for example floods [2,3,4], landslides [5,6,7,8], lightning [9], wind [10], and storm surges [11]. Articles mainly investigate the main factors and circumstances leading to the occurrence of fatalities. Unevenly distributed around the earth, FLs affect different areas but the pattern of disaster risk reflects the social construction of exposure and vulnerability according to the different countries.

Besides the numerous articles on this topic, insufficient efforts are conducted by local governments to promote the application of knowledge from the scientific literature to establish policies for the mitigation, adaptation, prevention, and reduction in the risk to people, addressed by initiatives sporadic in space and time. Despite international institutions offering didactic materials to prepare for emergencies (redcross.org (accessed on 14 June 2024)), in practice, people are commonly unaware of risks related to natural hazards, often thinking that they cannot affect just him/her.

Communications media contribute to keeping the level of public awareness low, commenting on the FL impact on people as “avoidable if someone else did something”, instead of promoting self-protective behavior, and demanding the safeguard of individuals only to institutions. Moreover, in the aftermath of flood events, TV news and newspapers focus on “someone to blame”. In contrast, out of respect for victims, journalists do not report any “hazardous behaviors” of fatalities, or do not report those behaviors as dangerous. For this reason, people always repeat same mistakes, such as for example trying to save goods and cars in flooded underground and ground floors [4].

Oppositely, the United Nation Office for Disaster Risk Reduction stated that “hazards do not have to turn into disasters”, meaning that we can reduce existing risk and support the resilience of societies in the face of residual risk that cannot be further reduced (PreventionWeb.net (accessed on 23 May 2024)). Actually, even in advanced societies committed to constantly improving disaster management and warning systems [12], people can be treated, injured, or killed, because risk can be mitigated but it cannot be completely eliminated, and effective preparedness measures can help minimize (but not eliminate) the loss of lives.

An important gap on the side of scientific research is represented by the absence of specific investigations focusing, besides fatalities, on people injured or involved, who are persons miraculously escaped from flood or landslide phenomena. Actually, descriptive semi-quantitative data on the circumstances of these interactions can reveal clues useful to plan informative campaigns for people, discouraging hazardous behaviors and promoting self-protective actions improving people–hazard interaction.

Research on these topics has been started in the past on the region of Calabria (Italy) for the period of 2000–2014 and afterwards extended to the period of 1980–2016, focusing on people vulnerability to FLs but also to lightning and wind impact. The present paper is based on that same approach, focusing on the FL impact on people during two different historical periods in order to: (a) perform a FL mortality comparison between past and present; (b) perform a comparison of the frequency of circumstances in which people were injured or not directly hurt by FLs; and (c) find the reasons responsible for changes between past and present, if any. The general aim of the research goes in the direction of the Sendai Framework for Disaster Risk Reduction target for reducing disaster mortality between 2020–2030 (PreventionWeb.net (accessed on 23 May 2024)), because it put the light on information that can be proficiently used in both making disaster management informed decisions and realize educational campaigns for disaster risk reduction, because the importance of education [13] is basically aiming to decrease the impact of (inevitable) future FLs. The severity of flood-related fatalities, for example, is not equally distributed worldwide, with specific geographical patterns identified, such as the Southern, Eastern, and South-Eastern regions of Asia having the deadliest floods [14]. Similarly, for landslide fatalities, Asia represents the dominant geographical area [6].

Then, considering that “disasters do not discriminate but their impact does, they disproportionately affect the poorest and most vulnerable because they exacerbate structural inequalities” (PreventionWeb.net (accessed on 23 May 2024)), it is necessary to investigate the impact of genders, ages, and motivations on a person’s likelihood to be affected by a FL event in a certain circumstance or in another, as we tried to do in this research.

Section 2 presents the methodology to select, collect and organize data on people–hazard interaction in a specific database. In Section 3, the application of the methodology to the Calabria (Italy) study area is presented, and the results obtained for this region are illustrated. The Discussion, presented in Section 4, illustrates limitations and the inherent uncertainty of the data collection and elaboration. The main conclusions are finally presented in Section 5.

2. Materials and Methods

The methodology is based on implementing a damage database (DD) obtained through the systematic analysis of documentary data sources, focusing exclusively on those FL events impacting people in a study area across two distinct periods. International damage databases, such as EM-DAT (https://www.emdat.be/ (accessed on 23 May 2024)), report the impact of major disasters on people, goods, and the environment. These databases can serve as an initial guide to identify the most severe events, typically those causing more than ten fatalities and/or necessitating international aid. However, smaller-scale events that do not generate significant impacts at a national or international level are often unreported in these international databases. Such events must be documented using original documentary data sources [15], because not only the threat of high-impact events, but also the frequent, low-impact events that are often hidden must be considered PreventionWeb.net (accessed on 23 May 2024), because they are very frequent especially in developed countries [3,15].

In contrast, the proposed methodological approach involves analyzing all instances of FL events that affect people. To collect data at a sufficiently detailed level, sources such as newspapers and news media are particularly relevant, as they are reliable sources of societal information. The literature has shown that these sources are frequently used to build specific catalogs detailing the impacts of one or more types of damaging phenomena at regional [16] national [17,18,19,20,21], or supranational scale [2].

The systematic analysis of multiple editions of various newspapers and news websites enables the identification of FL events impacting people during the study periods. The level of detail characterizing the articles is variable according to several factors, such as the general confusion characterizing the aftermath of the event or the sensitivity of the reporter, to name a few.

Considering the data sources used, in this paper the acronym FL encompasses both river floods and pluvial floods (F), as well as all types of rain-triggered landslides (L), regardless of their size or type. This broad inclusion is necessary, because non-scientific data sources often provide vague descriptions and use nontechnical terminology, making it difficult to distinguish among the various types of landslides described in the literature. The main steps of the methodology are described as follows.

2.1. Data Gathering

(a)

Identification of the documentary data sources.

The first step of the methodological approach is the identification of suitable documentary sources, particularly a primary local newspaper with a complete archive of past editions. Ancillary data sources, such as additional local newspapers or national ones, should also be identified and used to confirm data and/or clarify missing details on victims or accidents. It is important to note that national newspapers usually provide fewer details than local media, especially for low-severity and/or older events. Additionally, data availability can vary by country, due to factors such as privacy laws, which in some countries may restrict the reporting of victims’ names.

(b)

Selection of study decades.

Two decades sufficiently distant in time should be selected as study periods to detect significant local changes in society that occurred between them. For each decade, the primary and ancillary documentary sources must be identified and their availability checked, taking into account that, for older periods, historical research may be incomplete due to external factors affecting the accessibility of archives and data. After identifying the sources, the systematic consultation of all editions published during the study decades can be planned.

(c)

Data extraction.

What are we looking for in the consulted newspapers? Starting from the narrative of the accident reported in the selected articles, a series of information can be identified, disaggregated, and systematized in the fields of the DD. The description of the impact on people is classified according to three severity levels: fatalities, injured, and involved people.

Fatalities: The number of persons whose deaths were directly caused by FL short-term impacts, either immediately or in the hospital. Missing people are not included because a careful reading of all editions published during the event will clarify their fate; they are either eventually found or officially declared dead, after sufficient time has passed.

Injured: The number of persons whose health or physical integrity is affected as a direct result of the FL. This figure does not include people suffering from long-term consequences of the disaster (e.g., epidemics).

Involved: These are people, distinct from fatalities and injured, who were present at the place of the accident but survived without physical harm. They may have exhibited hazardous behavior or were able to protect themselves. Analyzing their behavior helps on one hand to investigate the causes of injuries or fatalities and on the other to identify behaviors that saved people during past events.

The number of fatalities is certain for all events because, unlike in developing countries, the number of fatalities per event in developed countries is generally small. The number of injured people is certain for most events. For involved people, especially in larger events, reporters often used collective measures, limiting the ability to determine the exact number of involved people. Therefore, these data should be considered more qualitative than quantitative, providing useful insights into the average behavior of people facing FLs and escaping unscathed.

2.2. Damage Database (DD) Creation

The DD is a spreadsheet where each row represents a person (fatality, injured, or involved) and comprises four data categories, each containing a series of fields to be filled with data extracted from documentary sources. The structure largely follows the one implemented in previous research at the Euro-Mediterranean scale [3,4]. The compilation is facilitated by fixed menus, pre-arranged in previous research to minimize typing errors and expedite data entry. For this research, the menus have been updated by adding new items (mainly from older events) and eliminating items not encountered in the data collected for the studied decades.

(a)

Event identification

Time: year, month, and day when the event occurred.

Localization: defined in the study case by NUTS 3, LAU name, and LAU code according to European nomenclature (https://ec.europa.eu/eurostat/ (accessed on 10 June 2024).

Hazard type: includes landslide, river flood, and pluvial flood. Pluvial floods occur when heavy rainfall creates flooding independently of an overflowing water body, typically in urban environments where the local drainage system cannot handle the surface runoff.

(b)

Victim profile

Name and Surname: when available, allowing exact identification and avoiding double counts.

Gender: can be available even in cases where name and surname are not provided.

Residency, age, occupation, and vulnerabilities. Vulnerabilities related to physical, mental, or social conditions are included, although this information is reported only in a few cases and we cannot be sure if this means that victim had no particular vulnerabilities or simply this information was not reported.

(c)

Victim–event interaction

Place characteristic and Place type, that is the place where the accident occurred.

Victim condition and Victim activity at the moment of the accident.

Accident dynamic causing fatalities, injuries, or involvement without physical harm.

Human response

Protective behaviors and Hazardous behaviors

Clinical cause of death, when available

Types of wounds, identified based on the analyzed data.

Clearly, not for all cases do we have enough data to fill all the DD fields; for this reason, in the menu of possible choices of each field, there is the possibility of selecting “unknown”. As conducted in previous studies [22], data analysis focuses on collecting and interpreting all available information from the sample data, rather than testing an existing hypothesis. This qualitative approach is akin to the grounded theory approach, a research method widely accepted in the social sciences and nursing, described as the “discovery of emerging patterns in data”. It aims to generate theory from the research situation in the field, as it is. See

Table 1. Data categories and fields of the damage database on fatalities, injured and involved people in flood and landslide events (NUT3 and LAU from: https://ec.europa.eu/eurostat/ (accessed on 10 June 2024)).

Data Categories	Fields
Event Identification	➢ Generality ● Name ● Surname ● Unknown ➢ Time ● Year ● Month ● Day ➢ Localization ● NUT3 ● LAU name ● LAU code ➢ Hazard Type ● Landslide ● River flood ● Pluvial flood
Victim Profile	➢ Gender ● Female ● Male ● Unknown ➢ Age ● 0–5 child ● 6–14 Boy-girl ● 15–29 Young adult ● 30–49 Adult ● 50–64 Middle age ● >64 Elderly ● Unknown ➢ Residency ● Resident ● Not resident ● Tourist ● Unknown ➢ Vulnerabilities ● Illnesses ● Mental disability ● Physical disability ● Pregnancy ● Homeless ● Unknown ➢ Occupation ● Driver ● Employee ● Entrepreneur ● Farmer ● Fireman ● Hiking guide ● Housewife ● Policeman ● Priest ● Retired ● Shepherd ● Store clerk ● Student ● Trader ● Teacher ● Train driver ● Volunteer ● Workman ● Unknown
Victim-Event Interaction	➢ Place Characteristic ● Indoor ● Outdoor ● Unknown ➢ Place Type ● Basement ● Bridge ● Camping ● Countryside ● Factory ● Ford ● Haystack ● Home ● Hotel ● Railway ● Riverbed ● River mouth ● Riverside ● Road ● School ● Seaside ● Shop ● Underpass ● Unknown ➢ Victim Condition ● By bus ● By car ● By mule/horse ● By tractor/excavator ● By train ● By truck ● By van ● Laying ● Standing ● Unknown ➢ Victim Activity ● Driving ● Finding snails ● Fishing ● Recreational activities ● Rescuing someone ● Shovelling mud ● Sleeping ● Staying home ● Travelling ● Walking ● Working ● Unknown ➢ Accident Dynamic ● Blocked in a flooded room ● Buried by landslide debris ● Caught in a derailment ● Caught in a house buried by landslide ● Caught in a house dropped by flood ● Caught in a road collapse ● Caught in building collapse ● Dragged by water/mud ● Fallen in a slope ● Fallen in a river ● Hit by debris carried by water ● Hit by landslide ● In a house hit by a landslide ● On a road blocked by a landslide ● Surrounded by water/mud ● Unknown
Human Response	➢ Protective Behavior ● Ask for help by mobile ● Climbing trees ● Driving to avoid danger ● Getting on roof/upper floor ● Getting on the car roof ● Getting out of building ● Getting out of vehicle ● Grabbing on to someone ● Grabbing on to bushes ● Moving to safer place ● Rescuing someone ● Swimming ● To warn people ● Unknown ➢ Hazardous Behavior ● Driving in a flooded underpass ● Fording rivers ● Refuse evacuation ● Refuse warnings ● Staying on river banks ● Take shelter under a bridge ● Unknown ➢ Cause of Death ● Chest trauma ● Collapse ● Drowning ● Heart attack ● Poly-trauma ● Poly-trauma and suffocation ● Suffocation ● Unknown ➢ Types of Wounds ● Abrasion ● Contusion ● Contusion and abrasion ● Contusion and shock ● Cranial trauma ● Fractured limbs ● Panic attack ● Principle of hypothermia ● Poly fractures ● Shock ● Thoracic trauma ● Wounded limbs and cranial trauma ● Drowning principle ● Unknown

.

3. The Study Area: Calabria Region, Italy

To demonstrate the potential outcomes of the methodological approach, we applied it to Calabria study region. Calabria is the southernmost peninsular region of Italy, covering an area of 15,080 km². According to the National Institute of Statistics (ISTATa, http://www.istat.it/it/ (accessed on 10 June 2024)), currently this region has 1,970,521 inhabitants, with 49% males and 51% females, residing in 404 municipalities, with an average of 4790 inhabitants per municipality. Due to its peninsular position by the Mediterranean Sea, Calabria experiences annual precipitation that varies according to the altitude above sea level, characterized by autumn–winter seasons punctuated by short and intense downpours that frequently trigger damaging FL events, on either a local or regional scale.

3.1. Data Gathering for the Case Study

For this research, we utilized a comprehensive and nearly exhaustive damage database concerning FL events that have affected Calabria since the late 19th century. This database has been continuously updated by researchers from the CNR-IRPI of Cosenza (Research Institute for Geo-Hydrological Protection) since 1996, when the first data collection was published. The master database of Calabrian FL impacts is derived from the systematic analysis of documentary sources such as newspapers and regional/national news websites, and reports from risk mitigation and management agencies, including civil protection and fire departments. It has served as the foundation for research on damage susceptibility zoning, high-impact events, analysis of meteorological and climatic triggering conditions, impact on people along road networks, and identification of descriptive rainfall indexes, among other studies.

From this database, we selected only the records reporting FL impacts on people during two study decades: D1 (1951–1960) and D2 (2011–2020), distant enough to witness changes in society and event management; then, we planned and carried out further investigation on these events. This involved historical research in state archives, local newspaper archives, and online newspaper collections, in order to collect the information required to fill the DD fields (Figure 1) for each person who faced the hazards.

For D1, we consulted national newspapers (Il Tempo, Il Corriere della Sera, Il Mattino, and La Gazzetta del Mezzogiorno) and the only Calabrian newspaper continuously publishing since 1952 (La Gazzetta del Sud). To verify the names of fatalities from major events, we searched for commemorative epigraphs dedicated to the victims, finding one in the Reggio Calabria hamlet of Valanidi and another in the Careri municipality. Additionally, accurate lists of fatalities from the deadliest events were obtained from the Reggio Calabria National Archive, including a list of families who lost members in the 1951 disaster, and for this received economic state subsidies (Figure 2).

For D2, we consulted all the digital daily editions of the Calabrian newspaper Il Quotidiano della Calabria and several local and national news websites (e.g., Adnkronos, Il Crotonese, La C News, Strettoweb.com, and Corrieredellacalabria.it), as well as meteorological amateur websites (e.g., Centro Meteo, Meteoweb, and Inmeteo.net).

3.2. Results

By cross-referencing all the mentioned data sources, we obtained the DD, resulting in a total of 1625 records. Of these, 13.3% concerned fatalities (FAT), 6.5% were injured (INJ), and 80.2% involved people (INV).

In both decades, river floods represent the most frequent phenomenon impacting people. From D1 to D2, a significant decrease in river flood FAT (from 196 to 20) and a substantial increase in the number of INV in landslides (from 81 to 207) and river floods (from 121 to 501) are evident (

Table 2. Number of fatalities caused by flood and landslides in D1 and D2 and sorted by type of hazard, gender, and age of victims. The total columns are highlighted in gray.

Fatalities		D1: 1951–1960		D2: 2011–2020
	Total	Landslide	River Flood	Total	Landslide	River Flood
Total	196	49	147	20	2	18
Males	106	22	84	13	2	11
Females	62	17	45	7	-	7
Unknown	28	10	18	-	-	-
0–5 years	18	3	15	1	-	1
Males	6	2	4	1	-	1
Females	6	1	5	-	-	-
Unknown	6	-	6	-	-	-
6–14 years	13	6	7	1	-	1
Males	9	4	5	-	-	1
Females	4	2	2	-	-	-
15–29 years	23	6	17	2	-	2
Males	15	4	11	1	-	1
Females	8	2	6	1	-	1
30–49 years	23			11	1	10
Males	17	3	14	6	1	5
Females	16	5	1	5	-	5
50–64 years	11	4	7	2	1	1
Males	6	1	5	1	1	-
Females	5	3	2	1	-	1
>64 years	13	2	11	3	-	3
Males	4	1	3	3	-	3
Females	9	1	8	-	-	-
Age unknown	95	20	75	-	-	-
Males	49	7	42	-	-	-
Females	24	3	21	-	-	-

). Conversely, the number of INJ (

Table 3. Number of injured people by flood and landslides in D1 and D2 and sorted by type of hazard, gender, and age. The total columns are highlighted in gray.

Injured		D1: 1951–1960		D2: 2011–2020
	Total	Landslide	River Flood	Total	Landslide	River Flood	Pluvial Flood
Total	52	19	33	52	19	33	-
Males	36	13	23	36	13	23	-
Females	6	3	3	6	3	3	-
Unknown	10	3	7	10	3	7	-
0–5 years	2	-	2	-	-	-	-
Males	-	-	-	-	-	-	-
Females	-	-	-	-	-	-	-
Unknown	2	-	2	-	-	-	-
6–14 years	6	3	3	4	1	3	-
Males	5	3	2	1		1	-
Females	1	-	1	3	1	2	-
15–29 years	9	4	5	3	-	3	-
Males	8	3	5	1	-	1	-
Females	1	1	-	2	-	2	-
30–49 years	15	1	14	9	3	6	-
Males	14	-	14	9	3	6	-
Females	1	1	-	-	-	-	-
50–64 years	2	1	1	3	-	2	1
Males	1	1	-	3	-	2	1
Females	1	-	1	-	-	-	-
>64 years	-	-	-	4	2	-	2
Males	-	-	-	2	1	-	1
Females	-	-	-	1	1	-	1
Age unknown	18	10	8	30	8	12	10
Males	8	6	2	16	6	7	3
Females	2	1	1	2	1	-	1
Unknown	8	3	5	12	1	5	6

) remains quite similar (D1: 52 and D2: 53) (Figure 2). Additionally, in D2, we detected 394 INV in pluvial floods, while in D1, we found no data on the occurrence of this phenomenon (

Table 4. Number of involved people in flood and landslides in D1 and D2 sorted by type of hazard, gender, and age. The total columns are highlighted in gray.

Involved		D1: 1951–1960		D2: 2011–2020
	Total	Landslide	River Flood	Total	Landslide	River Flood	Pluvial Flood
Total	202	81	121	1102	207	501	394
Males	67	10	57	206	62	91	53
Females	24	8	16	90	12	60	18
Unknown	111	63	48	806	133	350	323
0–5 years	2	-	2	19	-	1	18
Males	1	-	1	3	-	3	-
Females	1	-	1	3	-	2	1
Unknown	-	-	-	13	-	13	-
6–14 years	31	20	11	47	13	33	1
Males	7	1	6	4	-	3	1
Females	1	-	1	2	-	2	-
Unknown	23	19	4	41	13	28	-
15–29 years	15	1	14	28	7	17	4
Males	12	1	11	17	6	10	1
Females	2	-	2	9	1	7	1
Unknown	1	-	1	2	-	-	2
30–49 years	29	11	18	21	9	11	1
Males	19	6	13	13	6	6	1
Females	8	5	3	7	3	4	-
Unknown	2	-	2	1	-	1	-
50–64 years	1	-	1	2	-	2	-
Males	1	-	-	1	-	1	-
Females	-	-	-	1	-	1	-
>64 years	2	-	2	28	5	18	5
Males	1	-	1	10	3	4	3
Females	1	-	1	13	2	9	2
Unknown	-	-	-	5	-	5	-
Age unknown	122	49	73	957	173	402	382
Males	26	2	24	158	47	64	47
Females	11	3	8	55	6	35	14
Unknown	85	44	41	744	120	303	321

). The following sections describe the main factors differentiating D1 and D2, with possible explanations for the observed differences summarized in the Discussion section.

Tables summarizing data about fatalities are available in Appendix A and Appendix B.

3.2.1. D1 1951–1960: Fatalities

In D1 we recorded 196 persons died because of FLs. This decade is characterized by high levels of mortality caused by FLs (196 FAT, 75% by river flood and 25% by landslide). Except for 1955, FAT occurred in all the years of D1, with an average of 19.6 deaths per year. Two high-mortality events occurred in the autumn seasons of 1951 and 1953, respectively, killing a total of 141 people. In the October 1951 event, 76 people died due to both river floods (53) and landslides (23), while in the October 1953 event, fatalities were primarily due to river floods (64), with only one fatality from landslides. According to international databases like EM-DAT, which consider severe events those killing more than 10 people, in D1, six severe events occurred in D1 over 10 years. The severest events of 1951 and 1953 both affected large regional sectors, but FAT only occurred in the southernmost Calabrian province of Reggio Calabria, significantly impacting the regional distribution of FAT in the entire decade D1 (Figure 3).

Of the 196 FAT, the surnames and names of the victims were unknown in only 30 cases (15% of the total). This is notable, considering the events occurred more than 70 years ago, and this information was obtained by cross-checking different newspaper editions with data from other sources, such as the National Archive of Reggio Calabria (Figure 2). It must be taken into account that data about FAT and INJ are usually greatly detailed in terms of circumstances and number of victims, while data about people involved and not physically hurt are usually well defined in terms of circumstances but less precise in terms of the number of people affected.

Regarding the demographics of victims, we were unable to identify the gender of 14% of FAT, while the remaining 86% were largely males (54%). The age of FAT was obtained only for 52% of cases, with the highest percentages being Young adults (12%) and Adults (12%) (see the list of abbreviations). A percentage of 87% of FAT in D1 were residents in the place where they died. We did not find information about the vulnerabilities of those who died in D1 (Appendix A).

The occupation of victims is known for only 55 cases (28%), with farmers and shepherds being predominant, as can be expected in the typical peasant society characterizing Calabria in those years. Regarding the place, 46% of the fatalities occurred at home. Particularly in the most affected province (Reggio Calabria), 51% of the 149 fatalities occurred at home, especially because some events (like the one in 1953) occurred at night, surprising people in their sleep. It must be considered that, in D1, several people were accommodated in rudimentary dwellings largely characterizing the housing stock of the innermost villages of southern Italy in the post-World War II period. Particularly, in Reggio Calabria province, a series of “temporary homes” were realized by the government after the disastrous earthquake-tsunami which occurred on 28 December 1908 (https://cat.ingv.it/it/media-e-documenti/news-it (accessed on 6 June 2024)), and after almost half a century, in October 1951, people still lived in them. These shacks, weak and often located just near the rivers, were the places where floods found people, killing entire families. Moreover, after this event, further shacks were prepared for people who had escaped from the 1951 event, and in October 1953 the story repeated itself. Sixty-five people died, 64 of whom were swept away by floods, and 41 (63%) died at home, even if the concept of home was different from today’s. The scientific literature highlights that most injuries, damages, and deaths from disasters can be prevented, and disaster preparedness measures such as housing adjustments against risks can reduce the damage caused by disasters and accordingly improve recovery [15]. Considering victim conditions, no cases involving cars were recorded, mainly due to the scarcity of cars circulating at that time; ISTAT assessed that in Italy in 1951, there were 9 cars per 1000 inhabitants (https://www.istat.it/it/files/2011/03/Italia-in-cifre.pdf (accessed on 13 June 2024)).

Moreover, the road network of that time consisted of mule tracks and dirt roads, which often took advantage of the riverbeds, devoid of vegetation, to reach innermost hamlets. For this reason, fording rivers, which nowadays represents hazardous behavior, was often a forced choice in D1, because official roads to reach the innermost villages and to move from the hills to the plains passed through one or more riverbeds. Even if fording rivers was dangerous, it was necessary to accomplish daily activities.

3.2.2. D1 1951–1960: Injured and Involved

In D1, 52 persons were injured, with 63% by river flood and 37% by landslide. They were mainly males (69%) and, regardless the gender, they were resident in the accident place (54%). The age was unknown for 35% of cases, and in the other cases the highest percentage pertained to Adults (29%). The place of the accident was mainly home (29%), where people chiefly suffered contusions (58%), and the most frequent accident dynamics were the following: hit by landslide (17%), and caught in a house dropped by flood (13%). Details about hazardous and protective behaviors, as well as on victims’ vulnerabilities were scarce or absent.

In D1, 202 persons were involved: 59% faced river flood and 40%, landslide. The gender of the INV was mainly unknown (54%), 33% were males, and regardless of gender, they were mainly residents (75%). The age was unknown for 60% of cases and, in the remaining cases, the most frequent age group was Boy/Girl (15%). In the majority of cases, people were at home (61%) and were involved in similar percentages in landslides (31%) and river floods (30%). Only in 5% of the cases was hazardous behavior reported, but due to the small number of cases, we do not consider it significant.

3.2.3. D2 2011–2020: Fatalities

In D2, 20 persons died due to FLs, with 90% by river floods and 10% by landslides, respectively (Table 1). This decade shows a strong reduction in both total FL mortality and high-mortality events. During three years of the decade, no fatalities were detected; the average number of FAT per year is two. Except for the two 2018 river floods (causing 10 FAT in the August event and 3 in the October event), and the one that occurred in 2016 (causing 2 FAT), all remaining events caused a single FAT. The majority of FAT occurred in the Cosenza and Catanzaro provinces, although the number of FAT was considerably lower compared to the D1 decade (Figure 3). The gender of FAT is known for all the victims, with males being more numerous than females (65% versus 35%). A percentage of 50% of FAT were tourists, killed in the northern sector of the region, during the excursion in the gorges of the Raganello river, a touristic summer day trip. The remaining percentage were residents (40%) in the accident place, or with whom residence information was not available (10%). As far as the other variables of the DD, despite data being available, the small number of FAT does not allow for significant conclusions.

3.2.4. D2 2011–2020: Injured and Involved

In D2, 53 persons were injured, with 49% by river flood, 26% by landslide and 25% by pluvial flood. The gender of INJ was unknown for 23% of cases and the majority of remaining cases affected males (60%), mainly by river flood (32%) and secondly by landslide (19%). The age was unknown for 57% of INJ, and among the remaining cases, the majority were Adults (17%). Residency was unknown for 40% of INJ while the majority of remaining cases were tourists (34%), mainly involved in the mentioned Raganello accident. The occupation is largely unknown (72%). The accident place is unknown for 17% of INJ, and the majority of remaining cases occurred outdoors (72%), mainly on the road (30%) and in the riverbed (21%). The victim condition is unknown for only 4%, while in the remaining cases, INJ were standing (53%) or in a motor vehicle (34%), especially a car (26%). The accident dynamic is almost known (unknown cases: 4%): INJ were mainly dragged by water/mud (26%) or hit by landslide (21%), while the most common injuries were contusions and abrasions (30%), contusions (21%), and panic attack (13%). Data available about INJ behaviors are very scarce and not numerically significant for discussion.

In D2, 1102 persons were involved, a significantly higher number compared to the 202 cases recorded for D1. This increase is primarily due to the large availability of newspapers and web news services characterizing D2, allowing for more information by cross-checking multiple data sources. Moreover, there is greater public attention to signals of climate change, sometimes leading reporters and website editors to engage in a frantic hunt for damaging effects of rain (and consequent FLs), emphasizing these impacts in a sort of sensationalism.

People were involved primary in river flood (45%), followed by pluvial flood (36%) and landslides (19%). Information collected about gender, age, and occupation is not numerically significant. In contrast, the place where INV faced the hazard is unknown for only 2% of cases, while the remaining cases occurred in a series of places which are not mentioned in D1 records. Places were either outdoors (76%), mainly on the road (57%) and in underpasses (10%), or indoors, at home (18%) and in the basement (3%). Regarding the condition, unknown in 3% of cases, INV were generally on board of a motor vehicle (65%), mainly a car (60%), either driving (56%) or travelling in a vehicle driven by someone else (12%). In fact, the most common dynamic was being in a vehicle surrounded by water/mud (69%), a circumstance occurring due to pluvial floods (35%), river flood (31%) and landslide (4%), with differences only in proportions between water and mud, according to the kind of phenomena. By cross-checking dynamics and protective behaviors (which were effective if we consider that people were not hurt), regardless of the total number of cases, we note that, for example, in situations where people were surrounded by water/mud, a frequent self-protective behavior was getting on the roof/upper floor or moving to a safer place. Similarly, in cases of roads blocked by landslides, a frequent protective behavior was to warn people passing on the same road, while a self-protective behavior was to ask for help via mobile phone. Information on hazardous behaviors was gathered in a few cases and included driving in a flooded underpass, fording rivers, and trying to save belongings or vehicles.

4. Discussion

To understand the significance of the results, the advantages and disadvantages of the methodology must be compared. Starting with the limitations of the methodology, we have to consider the following points:

The DD consists of non-instrumental records obtained from documentary data, which may suffer from incompleteness that is difficult to quantify. It is virtually impossible to “validate” data in such a database because independent ancillary information is not available. Typically, all the available information is just sufficient to compile the catalog in a sort of artisanal work, where all existing information is considered and crosschecked. If new information becomes available, it is crosschecked against the previous data, and the catalog is confirmed, updated, or modified accordingly. Therefore, despite the uncertainty that can affect documentary data, they remain the only tool for constructing databases of hazardous events and their human consequences.

Data about FAT and INJ are usually detailed in terms of the number of victims, while data about people involved are usually well-defined in terms of circumstances and less precise in terms of the number of people. Thus, they must be considered as qualitative, especially in the severest events where reporters do not provide the exact number of INV and often use collective measures (e.g., “tens of people”).

Both the cause of death and the type of injuries are clinically defined for those cases in which it was reported in the data source, while often it can be inferred from descriptions. Privacy laws prevent the direct collection of clinical data from hospitals unless clinical studies have permission to access such data. In Calabria, no such studies are available, so the information brought to light by this research is the only publicly available information in the checked documentary data sources.

Regarding the advantages of the methodology, it is important to emphasize the following points:

The methodological approach can be applied to other regions as long as original data are locally collected. Despite similar societies behaving similarly, the reliability of results can be obtained only based on the analysis of data concerning the local scenarios of interactions between FL and people. This is crucial, especially considering that the majority of the literature studies used as references on this topic have been conducted in developing countries, where lifestyles and risk management can be very different from developed regions, and this can lead to incorrect deductions.

The selection of two study periods so distant in time is definitely successful because it allows for the detection of a series of obsolete dangerous circumstances and the onset of new ones. Knowledge of actual events and circumstances in a region with specific environmental, anthropogenic, and cultural characteristics can be effectively used to customize information campaigns to the local context.

By focusing on people’s behaviors (both right and wrong) and their evolution over time, it is possible to identify valuable insights for planning customized informational campaigns aimed at enhancing awareness and improving precautionary behaviors and self-rescue strategies to improve prompt reactions in coping with FL events.

Besides the pros and cons, considering the results, what have we learned by comparing FL events that occurred in Calabria in two periods so distant in time?

Firstly, in Calabria, mortality related to FLs decreased, mostly because of the improvement in housing strength and the intensive governmental reforestation and river embankment program carried out in the 1950s by the Cassa per il Mezzogiorno, a public body created to improve the wellbeing and development of southern Italy. While, in D1, because of the absence of safe paved roads, people were forced to use muleteers running near or inside riverbeds, putting themselves in dangerous situations, in D2, we detected cases of people fording rivers only to shorten their daily route, sometimes driving a SUV (Sport Utility Vehicle), as detected also in different geographical frameworks [23].

Secondly, while the number of FAT decreased from D1 to D2, the number of INV increased, mainly because of greater people mobility, as can be detected by the number of INV in vehicles in D2 (69%) compared to D1 (20%). A large number of drivers have to face pluvial floods. An initiative to increase their resilience should be, for example, the introduction of guidelines “on how to drive in heavy rain” into driving school programs, in order to equip newly licensed drivers with knowledge that they would otherwise have to learn directly the first time they face the dangerous situation of triggering FLs. Despite data about INV being difficult to collect, they can be useful to understand the framework in which people–event interaction develops. INV can be either persons that were lucky and/or able to escape from danger, or simply reckless people, putting themselves in danger and often others (i.e., people traveling with them by car). Thus, despite information on INV not being particularly detailed, it provides a sampling of current hazardous behaviors useful for shaping educational campaigns.

Despite the current large data availability, after current fatal events, TV news and newspapers do not report details about the hazardous behaviors of fatalities or do not report those behaviors as dangerous and mainly focus on “someone to blame”. For this reason, people repeat the same mistakes, such as trying to save goods and cars in flooded underground and ground floors [4]. Working on this point could strengthen people’s awareness of the risks related to FLs and the knowledge of dangerous behaviors leading to fatalities.

Significant points of this research include the potential use of the gathered data in future scientific studies on human vulnerability, as well as providing valuable information on the most prevalent dangerous habits today. Additionally, from a civil protection perspective, bringing old events to light can enhance awareness and risk perception among the new generations, who may be unaware of past events affecting their place of residence.

5. Conclusions

The research on flood and landslide events affecting humans in a region located in the Mediterranean basin (Calabria, Italy), carried out by comparing people–events interaction in the decades 1951–1960 and 2011–2020 highlighted a strong decrease in the number of fatalities, from 196 people in the oldest decade versus 20 fatalities in the most recent decade. This is mainly related to the improvement in life conditions in the region throughout the years, in terms of housing stock, road safety, and thanks to an intensive governmental program of structural works carried out in the 1950s, which mitigated flood and landslide risk in the most affected areas.

The gender of victims confirmed the percentages highlighted by previous similar studies, with a predominance of males over females among fatalities, injured, and involved. River floods remain, as in the past, the most frequent cause of fatalities in the region, causing 75% of fatalities in 1951–1960 and 90% of fatalities in the 2011–2020 decade. However, in recent years, there has been a significant increase in the number of people involved in pluvial floods (reaching 36% of cases, versus 45% related to rivers floods and 19% to landslides). While in the past, citizens were predominantly affected indoors, they now face hazards mostly outdoors, particularly when traveling by motor vehicles. Drivers display some hazardous behaviors, such as fording rivers or entering an underpass flooded by pluvial floods, but they also exhibit protective behaviors, such as warning other drivers in case of landslides on the road.

These outcomes can be proficiently used to inform disaster management decisions and enhance public awareness by realizing educational campaigns for disaster risk reduction. The results of this data collection (or similar data collections carried out in other regions using the same methodology) reflect the behaviors of people living in the area under study. Thus, they can be used to develop locally tailored strategies for human risk reduction, specifically targeting the weaknesses identified from the collected data. The ultimate purpose is to increase citizens’ resilience and decrease the impact of (inevitable) future floods and landslides.