3.1. Respondents’ Profile
In total, 105 completed questionnaires were returned out of the 413 that were originally sent (response rate of 25.4%). The participants were from municipal (49), regional (17) and national authorities (39). Although the resulting sample size is deemed adequate compared with other surveys addressed to experts (e.g., Flyn et al. [
53]; Darko et al. [
54]), it should be noted that there are limitations to the survey, that is, we can capture certain views and perceptions relating to certain respondents’ attributes and characteristics, such as experience, age and qualifications. Respondents were between 31 and 64 years of age (almost 63% were between 45 and 54 years of age), and almost 70% of them were males. Information on educational level shows that half (50.5%) of the respondents hold a master’s or doctoral degree, 36.2% are university graduates, 12.4% have at least secondary education, and the rest (less than 1%) have elementary education. Further, a significant portion of the respondents (i.e., 46.8%) has a relevant scientific background (e.g., geology, civil and/or environmental engineering, forestry, and meteorology), although the majority does not hold a pertinent qualification.
Regarding their professional experience in positions related to civil protection, almost three-quarters (i.e., 74.2%) of the participants have served for less than 10 years, i.e., about 32% between ten and five years, 30% between five and two years, and 12% less than two years. The rest, namely, 25.8%, stated that they have a service record of more than ten years. Out of the 29 respondents who have a service track for longer than 10 years, only 18 have a relevant education indicating an even lower percentage that combines experience and educational background (17%). Moreover, based on their answers, more than half of the respondents (i.e., 55.2%) lack hands-on experience. More specifically, about 37% said that they had experienced one or two flooding incidents, and 18% stated that they have no experience in such events. The rest, i.e., 44.8%, declared that they had faced at least three or more flooding incidents in their area of responsibility.
3.2. Assessment of Various Everyday Flood Risk Management Elements/Aspects
To monitor views on everyday flood risk management practices and aspects, participants were first asked to assess the levels of satisfaction on several critical resources and processes, including funding, workforce, supplies and materials, infrastructure, political will, cooperation and coordination with competent organizations, warning processes and protocols, cooperation with volunteer teams, and public response to authority warnings and guidelines. The ranking was carried out on a 1 (‘completely dissatisfied’) to 4 (‘completely satisfied’) scale.
Respondents were found to be mostly dissatisfied by most of the elements assessed (
Table 1), with the mean rating being lower than average (2.500). Specifically, “funding” came last (2.019), followed by “political will” (2.067) and others in ascending order of satisfaction. Only “warning processes and protocols” gathered a marginally positive view (2.638) in terms of satisfaction, since 61.0% of the respondents stated that they are mostly or completely satisfied.
It is interesting to note, that laypeople’s views on early warning processes show a lack of trust with three-quarters of them perceiving them as inadequate or very inadequate [
55]. This contrast between the overall satisfaction of officials and laypeople’s lack of trust, although no specific evidence is currently available, maybe due to shortcomings in the communication of risk towards the general public and should be further researched.
Chi-square tests were conducted to establish if any relationship exists between satisfaction levels shown above and key socio-demographic (age, experience, qualifications) and other variables (frequency of cooperation with other authorities, respondent’s position, whether he/she is a uniformed or civilian official). Amongst all tests carried out, significant differences were only found when comparing the responses between local and central authority officials, as well as between civilian and uniformed personnel. In both cases, the differences were identified concerning their level of satisfaction regarding human resources, equipment and infrastructure (
Table 2).
Using an open-ended question, participants were also asked to indicate the most important tangible or intangible resource/asset in terms of usefulness in flood protection and risk mitigation. Then their responses were grouped in five main categories. The most frequent responses included various types of specific equipment and tools (40%), followed by various operational protocols, processes and communication procedures (29.4%), human resources, (i.e., skilled/trained personnel or personnel with appropriate qualifications) (16.5%), various regulations (10.6%), and various types of structural flood protection measures (3.5%).
The statistical tests showed no statistical differences, indicating that the years of service, the experience of previous disasters, qualifications, self-assessed knowledge and education level do not have a statistical association with the answers. Certain differences were found between local and central authority personnel, but with no statistical significance overall. Local officials mentioned some forms of regulation as the most important resource/asset in a percentage of 2.5%, as opposed to a percentage of 17.8% recorded by officials of central authorities. Similarly, various types of structural flood protection measures were also different, as they were mentioned only by local officials at a percentage of 7.5%, whereas officials of central authorities did not report this type of measures at all (
Figure 1). Between these two groups, other answers presented similar percentages.
Non-statistically significant differences in responses were identified between uniformed and civilian personnel, as well. For instance, uniformed personnel considered various types of equipment as the most important resource/asset in terms of usefulness. On the contrary, civilians considered various types of physical infrastructure and structural flood protection measures as the most useful asset (
Figure 1).
Further, using another open-ended question, respondents were required to identify the most important obstacle they face in flood risk mitigation. The most popular responses included: Poor prevention processes and planning (38%), lack or insufficiency of structural flood protection measures (19.4%), inadequate funding (16.5%), problems arising from regulations, bureaucracy, and political will (11.6%), or related to early warning (5.8%), to human resources (4.9%), and connected with the public (2.9%) or other (1.0%).
Statistically important differences were identified only between local authority officials and officials of central authorities, with the former mentioning “inadequate funding” in 28.9% of the responses, in a much higher percentage compared to the latter (5.3%) (
Figure 2). On the other hand, officials of central authorities mentioned obstacles “related to the early warning”, “arising from regulations” and “connected with the public in higher percentages than local ones (10.5%, 7.9% and 7.9% as opposed to 2.6%, 2.6%, and 0%, respectively). The differences were statistically significant at 5% level (χ
2 = 12.563, df = 4,
p = 0.044).
In general, many officials mentioned poor prevention, lack of infrastructure and funding as the most important obstacles in everyday practice along with certain factors that are connected to them, including equipment and human resources. Political will and coordination between competent authorities were also identified as problematic. Climate change or any other climate-related factor was very rarely mentioned indicating that they are low on the priority list of problems. This is in agreement with literature findings on laypeople perceptions [
15] suggesting that floods are perceived more like a problem caused by human interference, rather than a natured-induced or climate-induced phenomenon. This tendency has been documented among laypeople in Greece as well [
55].
Survey participants ranked regional and municipal authorities (22.6% and 12.8%, respectively), the fire department (16.9%), and the police (13.3%) among the most important authorities in terms of cooperation regarding flood risk management, followed by the General Secretariat for Civil Protection (9.2%), volunteer groups (4.1%), and others (
Table 3). The frequency of cooperation of competent authorities was investigated by asking respondents to indicate how frequently they cooperate with other authorities, and which authorities they work with. More than 60% of them responded that the cooperation with other authorities is rare (once per six months or even less), 25% said that they cooperate once or twice a month, and the rest (i.e., approximately 13.5%) stated that they engage in cooperation with other authorities every week or more frequently. Local officials indicated a lower frequency of cooperation in comparison to officials of central authorities with a statistical significance at 10% (
p = 0.084). Further, more experienced personnel showed a higher frequency of cooperation compared to less experienced individuals (
p = 0.010).
3.3. Knowledge
When asked to assess their knowledge on floods and flood risk management, 37% of respondents stated that they perceive their knowledge to be high, 49% to be moderate, and 14% to be low. Apart from the self-assessment, respondents were asked to respond on the availability of specific flood-protection measures in their area of jurisdiction as an objective indication of their knowledge.
As shown in
Figure 3, almost half of them (46.7%) replied that structural measures had been implemented, followed by flood risk maps (42.9%), integrated flood risk plans (37.1%), public awareness and training (27.6%), and other non-structural measures (20%). Negative responses ranged approximately between 30% and 50%, indicating a significant lack of protection measures in some cases. Further, the data indicated a relatively high level of uncertainty considering the proportion of ‘not sure’ responses (between 23% and 33%), as well as wrong answers.
Indicatively, in the case of “flood risk maps”, in which information on nationwide availability is provided by a Ministry of the Environment and Energy of Greece database, participant responses proved to be wrong in 18.1% of cases. If we add the 26.7% of “not sure” answers to the 18.1% wrong ones, then we have approximately half of the respondents (44.8%) giving a wrong answer, as far as the availability of flood risk maps in their area of jurisdiction is concerned. Nevertheless, those who have both an experience (over five years), and a relevant background present a ratio of right to wrong answers twice as high as that of the rest of the participants (1.34 and 0.65, respectively). In other words, less trained or less experienced personnel were found to provide more wrong answers than their experienced and trained colleagues.
Further indications of poor knowledge come from the answers on knowledge of structural measures for which information is stored in the archives of the Technical Chamber of Greece. Out of 26 local officials that responded to the question regarding the presence of structural measures in their area of jurisdiction, only 12 (46.1%) gave the right answer, whereas 14 provided a wrong or a “not sure” answer.
Chi-Square tests of independence showed that respondents’ perceived knowledge on floods is associated with the knowledge on non-structural measures (χ2 = 9.394, df = 2, p = 0.052), flood risk maps (χ2 = 12.153, df = 4, p = 0.016), and integrated flood risk master plans (χ2 = 9.957, df = 4, p = 0.041). No other statistically significant associations were observed.
3.4. Risk Perception
The vast majority of respondents rated the risk from floods, in general, as “very important” (49.5%) or “important” (36.2%). Further, almost one-tenth of them rated it as of “average importance”, and a minor percentage (i.e., less than 3%) as of “low” or “negligible importance”.
Compared to other natural hazards, namely, earthquakes, forest fires, landslides, and tornados, floods were ranked third, behind earthquakes and forest fires. On a 5-point scale, with 1 indicating “not important” and 5 indicating “very important”, earthquakes ranked first with an average of 4.74, followed by forest fires (4.34), floods (4.33), landslides (3.49) and tornados (3.20). Multiple pairwise comparisons were conducted using the nonparametric two-sample Wilcoxon rank-sum test (a.k.a. Mann-Whitney test) to assess the differences in importance between the natural hazards. The results show that the null hypothesis (i.e., that the populations have equal medians) is rejected when comparing earthquakes with every other natural disaster, as well as when comparing floods and forest fires with landslides and tornados. The null hypothesis, however, cannot be rejected when comparing floods and forest fires (p = 0.988).
As regards the perception on the likelihood of future floods in their area of jurisdiction, almost two-thirds of the respondents (63.5%) believe that such an event is likely or extremely likely, 21.5% that is neither likely nor unlikely, 12.5% that is unlikely, and less than 3% that is extremely unlikely.
Besides the probability of occurrence, the participants were also asked to assess the severity of the consequences of future floods in their area of responsibility. About 22% of them characterized the potential consequences as very severe, 46.6% as severe, 21.4% as moderate, 8% as slight and 2% as negligible.
In order to explore demographic and other factors affecting the views of participants concerning the likelihood of occurrence and the severity of impacts of future flooding events, multivariate linear regression models were fitted to the dataset. The results are presented in
Table 4 and
Table 5.
According to
Table 4, the perceived likelihood of future floods (which was stated using a scale of 1 to 5) increases for respondents who provide higher rating to the risk from floods; have experienced flooding events in the past; serve more years in positions related to civil protection; have higher education, and are men. Similarly, based on the sign of the coefficients of
Table 5, it seems that the severity of consequences increases for those who provide higher rating to the risk from floods; have experienced flooding events in the past; are more dissatisfied with existing infrastructure for flood risk prevention; serve more years in positions related to civil protection, and are younger.
3.5. Early Warning Processes Assessment
Another important aspect examined in this survey has been the early warning processes and protocols followed. To this end, both input and output channels of information were sought. The vast majority of the respondents (almost 90%) said that their organization monitors warning information constantly, and more than 80% of those who replied positively said that they take specific measures upon warning. The comparison of the responses of 27 experienced and trained officials with the rest of the participants show a statistical difference (at p < 0.1 level) in the ratio of positive to negative answers with regard to taking action upon warning (4.4 and 2.75, respectively). That is, more experienced and trained respondents are more likely to act.
Concerning the input of warnings, about 46% stated that they receive warning information from official channels, 18.1% from the media, 15.2% from other organizations through non-official channels (e.g., personal communication), 12.4% from automatic early warning systems, and the rest declared other sources of information (e.g., social media, online community or forum, etc.). Only one-third of the participants claimed that their department warns the public for upcoming flood threats. Further, respondents were asked to express their opinion on the reliability of weather warnings about flooding events. As shown in
Figure 3, the majority of respondents (80.7%) gave a positive answer, whereas approximately 1 in 5 (19.3%) gave a negative one.
An ordered logit model was used to identify the factors that affect respondents’ opinion about the reliability of weather warnings for potential flooding, the results of which are reported in
Table 6.
Among the respondents’ characteristics analyzed, education is a statistically significant factor. Based on the coefficient, for a one-unit increase in education (i.e., going from secondary to tertiary education), the odds of believing that warnings are more reliable are 1.5 times higher, given all of the other variables in the model are held constant. Similarly, if a respondent provides a higher rating to the risk from floods and is more satisfied with existing warning processes, the odds of warnings reliability between two consecutive scales of reliability are 1.83 and 3.38 times higher, accordingly. It has to be noted that respondents from local authorities provided fewer positive answers (64.3% against 80.7% of the total), although no statistical significance was observed.
Finally, participants were asked to suggest potential measures or initiatives towards improving early warning processes and protocols through an open-ended question. Among the measures suggested, the most important ones, in order of popularity, were: Improvement of spatial and temporal accuracy of predictions, adoption and use of additional technologies (e.g., SMS and other electronic messages), overall improvement of authorities’ preparedness as well as public awareness campaigns and improved communication with the public. Statistically significant differences were found by comparing local to non-local actors and uniformed to civilian personnel. In detail, uniformed personnel prioritized public awareness and improved communication with the public at a higher rate than civilian personnel (χ2 = 15.449, df = 3, p = 0.001). In the same fashion, local personnel provided answers showing a higher percentage than non-local in improving prediction accuracy and adoption of additional new technologies (χ2 = 9.598, df = 3, p = 0.022).