Discussion

In the present study, we illuminated that approximately 66% of those who died due to the impact of Hagibis were aged 65 years or above. Overall, this finding is in line with the previous evidence. For example, it was previously reported that in Italy, in the 50-year period from 1965 to 2014, the percentage of flood-related fatalities among those aged 60 years and above was 27.3%, and 42.3% among males and females, respectively. ^{Reference Salvati, Petrucci, Rossi, Bianchi, Pasqua and Gender10} Additionally, in the US, during 1959 to 2005, people older than 60 years were found to be highly vulnerable to flooding. ^{Reference Ashley and Ashley11} Furthermore, during Hurricane Katrina, 60% of the total deaths occurred among those aged 65 years or above in New Orleans, ^{Reference Jonkman, Maaskant, Boyd and Levitan3} while 49% of total deaths were among those aged 75 years or older in Louisiana. ^{Reference Brunkard, Namulanda and Ratard12} Vulnerabilities to flooding among the elderly reportedly roots from normalcy bias, and deteriorating physical and cognitive functions. ^{Reference Brunkard, Namulanda and Ratard12} The present study reinforces previous findings by demonstrating that a timely evacuation of the elderly population should be prioritized to help reduce injuries and deaths during floods.

There were several important findings with respect to the existing evacuation strategies currently used in Japanese municipalities. Notably, the residents who drowned indoors, specifically those whose locations were identified, died in their homes in locations that are at high risk of flooding. This finding suggests a need to prioritize and enforce evacuation measures among those who live in or around such high-risk areas. It also suggests that evacuation may not be absolutely necessary among those who live in locations that are not considered high-risk areas. Flooding is relatively easier to predict than other disasters, such as earthquakes, owing to historical records, previous flood experiences, and geographical and environmental characteristics. Analyses of past records, as well as current anti-flooding measures, are used to develop flood hazard maps published at the local municipality level. ^{Reference Artigas, Bosits, Kojak, Elefante and Pechmann13} Also, past experiences have led to important improvements in evacuation strategies in Japan. For example, based on experiences during Japan’s 2011 triple disaster (earthquake, tsunami, and nuclear disaster), the Basic Act on Disaster Management was revised in 2014, and presently, in the context of disasters, municipalities are legally required to list all residents who have special care needs. Consequent to this revision, 98.9% (1720/1740) of the municipalities that were surveyed responded that they had already prepared the relevant lists by June 1, 2019, ¹⁴ accounting for approximately 6840000 vulnerable people. ¹⁵ However, the current legal frameworks continue to have some shortcomings. Specifically, the law currently does not specifically require the preparation of evacuation plans for these vulnerable populations at the individual level. ¹⁵ Thus, further improvement is needed in the prevailing evacuation strategies so that municipalities and residents can fully comprehend, in advance, the necessity of evacuating when faced with the prospect of severe flooding, and can provide particular care for those whose mobility and possibility of escape to higher floors is restricted.

Additionally, our study confirmed the importance of reevaluating the timing of evacuation orders issued by municipalities, given the finding that 54 people drowned outdoors. A warning regarding the potential landing of Typhoon Hagibis and subsequent flooding was widely disseminated a week before its actual landfall via numerous media outlets and central government announcements. Evacuation is an effective safety strategy if residents reach shelters before the possible dangers manifest. ^{Reference Lane, Charles-Guzman, Wheeler, Abid, Graber and Matte16} However, in this case, a majority of the local municipalities issued evacuation orders to their residents on the day before the typhoon landed and, in some instances, on the actual day, exposing residents to the risk of drowning during evacuation. ^{Reference French, Ing, Von Allmen and Wood17,Reference Du, FitzGerald, Clark and Hou18} It is clear that relevant stakeholders should revisit evacuation guidelines, as proposed by either national or local governments, and consider how these guidelines should be communicated and enforced in order to minimize damage and loss of life. ^{Reference Pfister19} Additionally, local residents should independently prepare for evacuation, seek information, check hazard maps when necessary, and devise optimal escape routes, irrespective of instructions from local municipalities

In the case of the 18 residents who drowned indoors, details on the locations of their deaths were available, which indicated that their bodies were found on the first floor of their residences. In other words, no death occurred among those evacuated to the second floor or above, a finding which is consistent with the observations from Hurricane Sandy. ²⁰ This finding also confirms the obvious fact that evacuation to higher locations in a building may be a practical strategy to avoid rising flood waters, when an escape from the building is no longer an option. Further, this can be considered a reasonable option, given the finding that more than half of the deceased residents drowned outdoors.

However, in this context, it is meaningful to consider the case of the 10 people who reportedly died on the first floor of their residences, despite the presence of a second floor in the buildings. Of these 10, evacuation of the 6 who had physical disabilities would not have been straightforward. It would have required action from their family members, neighbors, and other volunteers to either assist them in evacuation or to move them to higher floors. Similarly, those living in single-story houses, and in areas where the floodwater could exceed the second floor of a residence, should consider evacuating immediately after being advised to, and the shelters are opened, preferably before flood waters start to rise. Thus, it is reasonable to say that evacuation to a higher location is only possible if (1) the building has 2 stories or more, (2) residents can go up or be transferred to higher floors, and (3) the final floodwater depth is not greater than the height of the highest floor or roof of the building. This once again demonstrates the importance of individual evacuation strategies for those with special care needs and even those with intact physical and mental functions. Hazard maps are useful in this respect since they generally show potential areas that would be submerged by floods with a scale of once in 1000 years. ^{Reference Ozaki, Kanemoto and Wada21} Those living in areas at high risk of flooding should heed warnings and evacuate to safety early. Those unable to do so should put in place essential safety measures.

The remaining 4 victims did not seek higher locations or evacuate outdoors despite their intact physical functions. Plausible explanations according to available information, are lack of awareness of actual dangers, normalcy bias, and other personal priorities as shown in previous research. ^{Reference Ozaki, Kanemoto and Wada21–Reference Ozaki, Leppold and Tsubokura24} It must be noted that 2 residents (ID 4 and ID 14) refused to evacuate despite requests from others. Whatever the reasons, such attitudes would put not just themselves, but others, such as neighbors, rescuers, and first responders, at risk. Also, 1 resident living by themselves (ID 8) may have benefited from the support of neighbors or other persons. Based on the lessons learned from the Brisbane flood, ^{Reference Brockie and Miller25} it is essential to build strong and healthy communities; an individual approach to successfully manage evacuation, and ideally, determine community responses beforehand, as much as possible. ^{Reference Burger, Gochfeld and Lacy26}