Can the visualization of rip currents prevent drowning accidents? Consideration of the effect of optimism bias

Drowning accidents at beach in Japan are caused by rip currents. To reduce these accidents, a new technology that can detect rip currents and notify beachgoers by using the Internet of Things (IoT) and Artificial Intelligence (AI) was proposed. However, studies on the effect of visualizing rip currents or considering the effect of optimism bias have not been conducted. This study investigates if visualization of rip currents might help in preventing drowning accidents, while considering the effect of optimism bias. The participants were 90 Japanese beachgoers. They were asked to answer questions based on their knowledge of the beach and rip currents, their optimism bias regarding rip currents, and awareness with or without visualization. The results of the analyses suggest that despite optimism bias, the visualization of rip currents increases the tendency of beachgoers to perceive and avoid rip currents. As described above, it was found that by visualizing the rip current, beachgoers were able to perceive and avoid rip currents. In addition, an understanding of rip currents is positively related to the intent to avoid rip currents even when rip currents are visualized. Therefore, it is necessary not only to enhance the avoidance tendency by visualizing rip currents, but also to further enhance knowledge of beachgoers to deepen the understanding of rip currents including the danger associated and methods to avoid them.


Drowning accidents caused across the world by rip currents
Japan has approximately 35,000 km of ocean coastline. More than 20 million people visit the 1250 beaches across the nation and enjoy swimming in the summer season. However, many drowning accidents occur every year, and precious lives are lost. There are 2000-3000 rescues involving unconscious drowning each season, and 45% of all drowning accidents are caused by rip currents in Japan (Ishikawa et al., 2014). Rip currents (often 1 3 called "rips" or "rip tides") are strong, narrow seaward flows arising from alongshore variations in wave setup landward of the breaker zone. Due to their dependence on wave breaking, rips can develop in any beach environment in oceanic, sea, and lacustrine environments (Houser et al., 2017). There are many drowning accidents caused by rip currents in foreign countries. More than 50% of drowning accidents reported in Australia, the USA, and the UK, were caused by rip currents (Brighton et al., 2013). Therefore, it is necessary to prevent these accidents on a global scale.

Efforts to prevent drowning accidents caused by rip currents
To prevent accidents caused by rip currents, beachgoers must be able to recognize and avoid rip currents . However, identification is complicated by the fact that different rip current types or forcing conditions can create different visual signatures. For example, channel rips that form in deeper channels between sand bars (Castelle et al., 2016) typically present, optically, as an area of darker and calmer water because of a relative lack of wave breaking when compared to the shallow sandbars on either side. In contrast, flash rips (Castelle et al., 2016) are not channelized but are instead generated by transient surf zone eddies resulting from vertical motions associated with shortcrested breaking waves and are typically characterized by sediment-laden plumes of water extending offshore and a turbulent water surface. In addition, the wave-current interaction between incoming waves and the offshore rip current flow can present visually itself as a rippled and bumpy water surface (Ménard et al., 2018). Therefore, previous studies have shown that many beachgoers are unable to identify rip currents (Pitman et al., 2021). Morgan et al. (2009) found that gender, age, alcohol consumption, overconfidence in swimming ability, and lack of knowledge about rip currents, are associated with drowning accidents by rip currents. Brannstrom et al. (2015) and Caldwel et al. (2013) showed beachgoers pictures of rips and asking them to identify the area of the currents, reported that most of them misidentified the area. In addition, Pitman et al. (2021) compared the photographic case with the actual viewing case and suggested that the beachgoer's ability to identify rip currents is lower than that reported in previous studies using photographs.
Warning signs are put up at many beaches to inform people of rip currents (Ménard et al., 2018). In addition to warning signs, in many jurisdictions flags are used to indicate lifesaving surveillance areas, safe swimming areas, or rip currents and other hazards. However, previous studies evaluating the effectiveness of warning signs, indicated that more than half of the beachgoers did not notice signs posted on the beach (Kaminski et al., 2017;Matthews et al., 2014). It has also been reported that many people did not prepare for or avoid rip currents, even if they were aware of the warning signs (Hall & Slothower, 2009;Karanci et al., 2005;Siegrist & Gutscher, 2006). Furthermore, there are several problems associated with the usage of beach flags. As the color of the flag and its intended meaning differs from country to country, beachgoers from other countries may not be sure of the intended meaning (Ménard et al., 2018). Although lifesavers need to understand topographical features before finding rip currents, it is difficult to indicate the exact area through beach flags . Furthermore, not all beaches are managed by lifesavers or beach flags, and beach flags cannot be set up outside the managed areas. Consequently, there is a possibility that beachgoers may believe that an area is safe and get caught in rip currents that are in areas that are not being managed.
Previous studies (Hatfield et al., 2012;Houser et al., 2017;Ishikawa et al., 2019) have suggested that, when beachgoers are given sufficient visual information about where rip currents occur, they can identify the area and are more likely to avoid these areas based on their observations. In the case of signs only, it is difficult for beachgoers to be alerted because they usually cannot identify the rip current area, whereas showing where the rip current occurs making it easier for them to be alerted because they can identify the area. In Australia, lifesavers check the area for rip currents and other dangers, and inform beachgoers by sharing information through a smartphone app (MashableAsia, 2016). However, as the information is fed in at a predetermined time, it is not possible to track the constantly changing situation. In Japan, a new system using the Internet of Things (IoT) has been developed that automatically detects the occurrence of rip currents through Artificial Intelligence (AI), and the information is displayed on a digital signage installed at the beach to alert beachgoers . The advantage of this approach is the ability to display real-time alerts of geographically and temporally changing rip currents regardless of the area and time of the day. Endo et al. (2019) examined the awareness of beachgoers with the use of this approach. Based on a survey of 142 beachgoers, it was reported that more than 90% of them identified rip currents and tried to avoid the area.
Based on the above, it can be deduced that beachgoers can perceive the danger and avoid rip currents by visualizing them.

The effect of cognitive bias
As reported in a previous study (Endo et al., 2019), visualization of rip currents through AI is an effective method to prevent drowning accidents. However, the study did not consider the effect of beachgoer's cognition of rip currents. Ménard et al. (2018) pointed out that all individuals are subject to cognitive biases in decision-making and these biases may have severe consequences when applied to swimming decisions. Confirmation bias is the most common bias, which is the cognitive tendency to focus on evidence that supports one's beliefs or decisions and to ignore evidence that disproves them. This bias can make people look for evidence that it is safe to swim (for example, there are other people in the sea, there are no waves) and ignore evidence that it is not safe to swim (for example, there are red flags and warning signs). Scaman (2017) asked participants to evaluate their decision to swim in the sea by showing them photographs of beaches with different waves and different number of people. She reported that although there were rip currents, participants were more likely to enter the sea after seeing photographs where people were in the sea than when they saw photographs of the same beach without people. The results indicated the impact of confirmation bias, as people make decisions based on the presence and behavior of others, and not on wave conditions. Additionally, if the perception of fear is inadequate, people assume that "It won't happen to me" and therefore they are safe and can take the risk (Slovic, 1987;Slovic et al., 1981). This tendency to interpret and predict things according to their advantage and to estimate that their risk is lower than that of others is called optimism bias (Armor & Taylor, 2002;Klein & Weinstein, 1997). Previous studies have reported that optimism bias occurs in crime, traffic accidents, natural disasters such as earthquakes, and health problems such as illness (Burger & Palmer, 1992;Green et al., 2003;McKenna et al., 1991). Optimism bias has also been studied in Japan. Oikawa and Oikawa (2010) examined the risk of infection by themselves and others based on the outbreak of swine flu in 2009 among university students. As a result, they found that college students rated their own risk of infection to be lower than that of others. Sasatake 1 3 (2014) examined the risk of sexual victimization among female university students and reported that many of them rated the risk of others being victimized higher than themselves. Furthermore, previous studies have indicated that when optimism bias occurs, riskaverse behavior is less likely to occur. Sasatake (2014) found that when optimism bias was high, crime prevention awareness was low.
As described above, optimism bias has been reported to occur in a variety of domains and decreases the probability of risk-averse behavior occurring. Most people are not afraid of drowning as swimming is considered to be a low fear activity with known risks (Sandman, 1989;Slovic, 1987;Slovic et al., 1981). In addition, Matthews et al. (2014) suggested that beachgoers have formed their own beliefs similar to optimism bias (the beach was relatively safe, or that the potential hazards would not happen to them). Therefore, even if the danger of drowning by rip currents is mentioned, the danger is likely to be underestimated due to optimism bias.

Purpose of the study
Various efforts have been made to prevent drowning accidents due to rip currents. However, there are has been no research on the effect of visualizing rip currents or the impact of optimism bias. Ménard et al. (2018) elaborated on the need for psychological studies to develop strategies to ensure that beachgoers avoid rip currents and prevent drowning accidents. This study investigates if the visualization of rip currents might assist in preventing drowning accidents, while taking into consideration the impact of optimism bias. In addition, this study focuses on flash rip which is difficult to visually understand due to its intermittent occurrence.

Place (Figs. 1 and 2)
This study was conducted at Aoshima beach in Miyazaki Prefecture, Japan, which has rip currents due to the proximity to the sea (Nishi et al., 2005). The study beach faces the Pacific Ocean and has a 300 m long swimming area with a foreshore zone that has a very gentle slope of 1/100 composed mainly of fine sand. The wave condition is relatively calm due to the influence of the cape on the south side, but high waves often hit the beach caused by Typhoons. The energy mean wave height and period at offshore area are 1.78 m, 8.64 s during summer season. On this beach, flash rips are generated by wave conditions in the swimming area. Also, this beach is used by up to 22,000 people per day.

Participants
The survey was conducted with one hundred Japanese beachgoers (43 men, 55 women, and 2 unknown) at Aoshima beach. As a result of using list-wise case deletion for missing value processing, data from a total of 90 participants (41 men, 49 women; M age = 36.1 years, SD age = 16.9 years) were analyzed. The survey was conducted on August 10 and 11 (both 8:30-16:30), 2019. Two authors asked the participants to cooperate at the entrance of the beach (two locations). After obtaining informed consent, they were shown the image before the visualization of the rip current and asked to answer the items described below. Subsequently, they were shown the screen after the visualization and asked to answer the items described below. There were no incentives for them, and the number of participants who refused to answer was approximately fifty. Before conducting the survey, the purpose and contents of the questionnaire were explained to the head of the beach and permission was taken. Additionally, at the time of the survey, participants were informed that their responses were voluntary and there was no disadvantage of non-participation. Their consent was taken before they responded to the questionnaire.

Measurements
Participants were asked to answer question associated with the following three survey items. In addition, they were asked the possibility of damage by rip currents, the evaluation of a system for visualizing rip currents, and the intent to return to beaches after visualizing rip currents. However, the details have been omitted as they were not relevant for the purpose of this study.

Knowledge of the beach and rip currents
The following questions were prepared based on a previous study (Houser et al., 2017): frequency of going to the beach (infrequent: fewer than ten times in my life, once every year typically on vacation, multiple times every year, several times every year, frequently: weekly or daily), experience of a drowning accident (no experience, experienced), prior information gathering such as weather, strength of the wind (no, yes), swimming ability such as swimming fast or long distances (unable to swim, weak swimmer, competent swimmer, highly competent swimmer), and understanding of rip currents (do not understand, understand, understand well). The participants who stated "understand" or "understand well" were asked to describe rip currents in detail. Whether the answer was correct was determined by consultation between two contributors who were familiar with rip currents. We referred to the description of rip currents (rip currents are strong, narrow seaward flows arising from alongshore variations in wave setup landward of the breaker zone) provided in a previous study (Houser et al., 2017). After the participants responded to these questions, to unify their understanding, rip currents were explained in detail (above description of rip currents by Houser et al. (2017)) through this image (Fig. 3).

Optimism bias for rip currents
Studies have not measured optimism bias regarding rip currents. Based on previous studies and approaches that have been used to measure optimism bias (Oikawa & Oikawa, 2010;Sasatake, 2014;Shepperd et al., 2013), the participants were asked to judge the possibility of them being caught in rip currents and the possibility of others being caught in rip currents on that day. They were asked to rate the chances for themselves and others from 0% (never get caught in) to 100% (definitely get caught in).

Rating of beachgoers in the situation of visualization rip currents
Considering previous studies (Endo et al., 2019), we conducted the survey as follows. First, we asked the participants to think of a normal situation in the sea (without visualization), and to answer whether they could cognize the area of a rip current (cognition), consider it dangerous (danger), and play without it (avoidance). They rated based on an 11 point scale, from 0% (cannot cognize at all, not at all dangerous, cannot avoid) to 100% (can cognize, extremely dangerous, can avoid). Thereafter, they were shown a visualization of a rip current ( Fig. 4; rip current occurred, with visualization situation) and asked whether they could recognize the area of the rip current, they considered the rip current was dangerous, and they could avoid the rip current. The answer choices were similar to those in the without visualization situation.

Statistical analyses
First, a descriptive analysis was conducted to summarize all the indicators. Second, we examined whether an optimism bias was observed in participants. Previous studies have shown that optimism bias is found when one rates one's own risk as lower than that of others (Sasatake, 2014). In addition, when assessing one's own risk, one compares oneself with others of the same generation and gender (Sasatake, 2014). Moreover, those who were found to have an optimism bias rate their risk itself lower. Therefore, in this study, those who rated that the possibility of being getting caught in rip currents is less than 50%, or their possibility of getting caught is less than 50%, but rated that the possibility of others getting caught is higher than theirs, were judged to have an optimism bias. Third, paired t-tests were carried out for the difference between the awareness of rip currents without visualization and the awareness with visualization for people with optimism bias. Fourth, Pearson's product moment correlation coefficients were calculated to examine the relationship between each variable. For the analysis, the following variables were dummy coded: gender (0: female, 1: male), frequency of going to the beach (0: infrequent and once every year typically on vacation, 1: multiple times every year, several times every year, frequent), experience of a drowning accident (0: no experience, 1: experienced), prior information gathering (0: no, 1: yes), swimming ability (0: unable to swim and weak swimmer, 1: competent swimmer and highly competent swimmer), and understanding of rip currents (0: do not understand, 1: understand Fig. 3 The image that we used to explain rip currents and understand well). Finally, a hierarchical multiple regression analysis was performed with the intent to avoid rip currents as a dependent variable. While analyzing, based on Cohen (1992), we eliminated the relationships with a low correlation coefficient (r < 0.20). Demographic data were entered in the first step as control variables. The scores of awareness of beachgoers without visualization were entered in the second step. Finally, the scores of awareness of beachgoers after visualization were entered in the third step. Statistical analyses were performed using SPSS (Version 25.0) with the level of significance set at 5%.

Sample characteristics
The results of the sample characteristics are presented in Table 1. "Once every year typically on vacation" was mentioned as the frequency of going to the beach by 37 participants, which is the highest (41.1%). "No experience" for experience of a drowning accident was mentioned by 85 participants (94.4%), which implied that most participants did not have any experience of drowning accidents. "No" for prior information gathering was mentioned by 46 (51.1%) participants, which implied that approximately half of the participants gathered information about the beach. "Competent swimmer" and "highly competent swimmer" for swimming ability was stated by 35 (38.9%) and 27 (30.0%) participants, respectively, which indicated that more than half of them could swim. "Not understand" for understanding of rip currents was stated by 53 participants (58.9%), which implied that more than half of them did not understand the phenomenon.

Optimism bias for rip currents among participants
Based on the set criteria, we examined if the participants are affected by optimism bias about rip currents. It was observed that 63 of 93 participants (33 men and 30 women) experienced optimism bias. Table 1 summarizes the characteristics of those who were experiencing optimism. Moreover, the results showed that the high percentage of participants gathered prior information. However, no differences were found for the other variables. Table 2 summarizes the descriptive statistics of cognition, danger, and avoidance of rip Table 1 Sample characteristics in terms of frequency, experience of a drowning accident, prior information gathering, swimming ability, and understanding of the rip current a All participants (n = 90), b participants with optimism bias (n = 63)

Ratings of rip currents when visualized in individuals with optimism bias
Paired t-tests were carried out to determine the difference between the cognition, danger, and avoidance with and without visualization rip currents in people with optimism bias (Table 3). It was observed that participants recognize the area of rip currents better after visualization than without visualization. Similarly, they had a stronger intent to avoid rip currents than those without visualization. There was no significant difference in the degree of perceived danger.

Correlation coefficients between each variable
Pearson's product moment correlation coefficients were calculated to examine the relationship between each variable (Table 4). As most people had never experienced drowning accidents, this variable was excluded from the analyses. The intention to avoid rip currents showed significant positive correlation with the understanding of rip currents, the recognition of the area of rip currents, and the degree of danger with visualization. Additionally, gender showed marginally significant positive correlations with the recognition of the area of rip currents and the intent to avoid rip currents without visualization. 1 3

The result of hierarchical multiple regression analysis predicting the intent to avoid rip currents
Hierarchical multiple regression analysis was performed to predict the intent to avoid rip currents ( Table 5). The following predictor variables were subsequently entered into the regression model: control variables (i.e., gender and understanding of the rip currents) in .55 ** .47 ** 11. Danger (V) 34 ** 12. Avoidance (V) Table 5 Hierarchical multiple regression analysis predicting avoidance (V) among participants with optimism bias (n = 63) ***p < .001, ** p < .01, *p < .05; WV: without visualization; V: with visualization Independent value Avoidance (V) Step the first step, recognition of the area of rip current, and the intent to avoid rip currents without visualization in the second step, recognition of the area of rip currents, the degree of danger, and the intent to avoid rip currents with visualization in the third step. In the third step, a positive relationship was observed between the understanding of rip currents, the recognition of the area of rip currents with visualization, and the intent to avoid rip currents with visualization. The coefficients of determination (R 2 s) and their increase (ΔR 2 ) except ΔR 2 in the second step, were significant in each step (ps < 0.01), and no multicollinearity was found between any predictor variables (variance inflation factor (VIF) = 1.00-1.55).

Knowledge of the beach and rip currents
Previous studies have reported that most beachgoers misidentified the area of rip currents when they were shown pictures and asked to identify the area of rips (Brannstrom et al., 2015;Caldwell et al., 2013;Houser et al., 2017). In this study, participants were asked if they knew what rip currents are, and if they knew about them, they were asked to offer an explanation through a free description. Fifty-three participants (58.9%) answered, "not understand" regarding their understanding of rip currents. This implied that more than half of them did not understand the phenomenon. Although there are differences in survey methods, the results of this study are similar to those of previous studies, which implies that more than half of beachgoers do not understand rip currents. Ménard et al. (2018) pointed out that all individuals are subject to cognitive biases in decision-making, and these biases may have severe consequences when applied to swimming decisions. Previous study has pointed out that people may be caught up in rip currents due to confirmation bias, which is one of the cognitive biases (Scaman, 2017). In addition, people generally assume that "It won't happen to me," when fear is inadequate, and that they are safe from a particular risk (Slovic, 1987;Slovic et al., 1981). This tendency to interpret and predict things according to their advantage and to estimate that their risk is lower than that of others is called optimism bias (Armor & Taylor, 2002;Klein & Weinstein, 1997). Most people are not afraid of drowning as swimming is considered to be a low fear activity with known risks (Sandman, 1989;Slovic, 1987;Slovic et al., 1981). In addition, Matthews et al. (2014) suggested that beachgoers have formed their own beliefs similar to optimism bias (the beach was relatively safe, or that the potential hazards would not happen to them). Therefore, even if the danger of drowning by rip currents is mentioned, the danger is likely to be underestimated due to optimism bias. However, previous studies have not examined the extent to which optimism bias toward rip currents occurs in beachgoers. We investigated whether participants displayed an optimism bias for rip currents based on a set criterion. It was observed that more than half of the participants rated the possibility of others getting caught in rip currents as higher than their own, which implied the existence of optimism bias. This result supports previous studies' suggestion that cognitive biases occur in swimming situations, which may have severe consequences. This study prepared items for optimism bias regarding rip currents based on previous studies, but the reliability and validity of the items have not been sufficiently verified. In the future, it is necessary to validate a method that can accurately measure optimism bias.

Avoidance tendency after visualization of rip currents
Many beaches provide warnings by placing warning signs of rip currents, surveillance by lifesavers, and beach flags indicating the area of safe swimming areas (Ménard et al., 2018). However, signs may be ineffective because beachgoers cannot often identify the area of rip currents. Meanwhile, for beach flags, as the color of the flag and its intended meaning differ per country, beachgoers from other countries may not be sure of the intended meaning. Further, although lifesavers need to understand topographical features before finding rip currents, it is difficult for them to point out the exact area . Previous studies (Hatfield et al., 2012;Houser et al., 2017;Ishikawa et al., 2019) suggest that when beachgoers are given sufficient visual information about where rip currents occur, they can identify the area. Endo et al. (2019) reported that more than 90% of beachgoers who had visual information tried to avoid the areas of the rip currents. This study compares the perception of the area of rip currents, the degree of danger, and the intent to avoid rip currents with and without visualization rip currents in participants with optimism bias. The results indicated that the visualized phenomenon was significantly more likely to perceive and avoid rip currents than the non-visualized phenomenon. Therefore, the results of this study support previous studies and it is established that visualization of rip currents increases the intent to avoid them even if optimism bias exists. Morgan et al. (2009) reported that gender, age, alcohol consumption, overconfidence in swimming ability, and lack of knowledge about rip currents, were associated with drowning accidents due to the phenomenon. In this study, gender and frequency of going to the beach showed significant positive correlation with the recognition of the area of rip currents and the intent to avoid rip currents without visualization. Therefore, in the case of non-visualization, attributes such as gender and frequency are related to the intent to avoid rip currents as has been established in previous studies, and it is considered that this can lead to drowning accidents. On the other hand, the intent to avoid rip currents with visualization is not related to the above-mentioned variables, and the intent to avoid rip currents with visualization is positively related to the understanding of rip currents, the area of rip currents in the visualization, and the recognition of the danger. It is assumed that the results of this study reflect that everyone can recognize the area of the rip current after visualization, regardless of gender and frequency.
To prevent drowning accidents due to rip currents, previous studies have focused on placing warning signs, lifesaving surveillance, and beach flags, to indicate the safe swimming areas, or rip currents and other hazards (Ménard et al., 2018). However, studies that have evaluated the effectiveness of signs suggest that less than half of the beachgoers are not aware of signs posted on the beach (Brannstrom et al., 2015;Kaminski et al., 2017;Matthewset al., 2014). It has also been suggested that many people did not prepare for or avoid rip currents, even if they were aware of the warning signs (Hall & Slothower, 2009;Karanci et al., 2005;Siegrist & Gutscher, 2006). The results of analyses suggest that even those with optimism bias could avoid rip currents significantly by visualizing them. A hierarchical multiple regression analysis using the intention to avoid rip currents as a dependent variable showed a significant positive relationship with the recognition of the area of rip current after visualization. Therefore, the results of this study suggest that visualizing the rip currents may increase the intent to avoid the area and thus plays an important role.
Previous studies have emphasized on educating people to deepen the understanding of rip currents (Endo et al., 2019;Hatfield et al., 2012). As has been mentioned above, half of the participants did not understand the phenomenon. The analysis suggests that the understanding of rip currents is positively related to the intent to avoid rip currents after visualization. Therefore, it is necessary not only to enhance the avoidance tendency by simply visualizing the rip currents, but also to further enhance education to deepen the understanding of the phenomenon including the danger and avoidance strategies.

Limitation
This study is limited to data from only one beach, and conventional measures to prevent drowning by due to rip currents have not been compared. In the future, similar surveys should be conducted at various beaches, and conventional measures in these surveys against drowning must be compared. In addition, this study did not investigate whether people actually avoided rip currents because they only answered their intention to avoid them. Therefore, whether bathers actually avoid rip currents by observing signage should be examined. Furthermore, in human decision making, the influence of various variables is possible. It is also necessary to visualize the rip current, extract factors that cause differences between those who avoid it and those who do not, using qualitative research, and examine the relationship between these variables and avoidance behavior.

Conclusion
This study investigates if visualization of rip currents could assist in preventing drowning accidents, considering the effects of optimism bias. The results of the analyses suggest that even if optimism bias exists, the visualization of rip currents increases the tendency of beachgoers to recognize and avoid rip currents. As described above, it has been observed that after visualizing rip currents, beachgoers can avoid them. A system that uses AI to detect rip currents and display them on signage has already been introduced in Japan (e.g., Onjuku, Chiba Prefecture), and it is being actively considered for other beaches. Based on the results of this study, it is expected that the system will function as a useful tool for preventing drowning accidents. In the future, based on the limitations of this study clarifying the effectiveness of the visualization of rip currents and developing strategies to prevent drowning accidents are necessary.