Proposal of a method to analyze children’s flood risk exposure and risk perception using GPS tracking data and questionnaire survey

Background Climate change has made the relationship between disaster and sustainable development more critical. Especially in developing countries, disasters frequently devastate local communities and hinder progress. Children are at a higher risk during catastrophes; however, their knowledge of disaster risk is limited. Education for disaster risk reduction can bring the necessary changes for a safe society. School disaster prevention education can raise awareness through lectures and games, but interactive dialogues between students and teachers can be more effective. Also, GPS tracking can be a valuable tool for understanding people’s behavior during disasters. Methodology This study proposes a method to analyze the relationship between students’ daily behavior and their experiences of flood damage and preparedness for disaster risk. It is a simple method using questionnaires and GPS loggers and can be applied in mountainous areas in developing countries where equipment is inadequate. The proposed method will be used to understand the current state of disaster risk among children and identify disaster risk reduction issues. The study area lies in Paluntar Municipality, Gorkha district, Nepal. Forty-five students from two schools in the region participated in the study. Results The results showed that students who commute to school in areas inundated by heavy rainfall in recent years were not adequately prepared for disasters, even though they had experienced flood damage in the past and were significantly more concerned about future harm. Our field observations and interviews of teacher-student relationships showed that students were very attentive to their teachers’ instruction. Teachers understand the dangers of heavy rainfall and how to prepare for it, but students need to learn how to deal with it. In other words, the school staff’s disaster education to the students effectively reduces disasters. Early warning is needed when severe floods are expected during the rainy season.


Introduction
Climate change has made the relationship between disaster and sustainable development more critical.Especially in developing countries, frequent disasters devastate local communities and impede sustainable development (Clerveaux and Spence 2009).The Sendai Framework for Disaster Risk Reduction targets a reduction in the number of people affected by disasters and economic losses worldwide by 2030 (United and Nations 2015).Reducing disaster risk damage in developing countries is a challenging topic.
Children are often more likely than adults to be victims of disasters because their knowledge of disaster risk is less than that of adults (Yildiz et al. 2023).High-risk areas have made efforts to increase children's knowledge of disasters (Victoria et al. 2014).Education for disaster risk reduction has the potential to bring about the changes necessary for a safe society through education appropriate to schools and communities (Petal 2009).
School disaster prevention education centered on lectures can raise children's risk awareness.Children participating in hazard education programs will have a sound knowledge of disasters (Ronan and Johnston 2001).Acquiring knowledge also leads to improved responses to warnings and lessened fear of disasters (Powell et al., 2009).Disaster prevention games can influence children's cognition and behavior and promote awareness (Nouchi and Sugiura 2014;Zaini et al. 2020).Soffer et al. (2010) showed that a combination of lectures and drills is likely best to prepare students concerning their behavior during a disaster.Evacuations occur when a person has sufficient motivation, has acquired the ability to perform the evacuation, and finally has the impetus to act (Fogg 2009).However, learning the importance of pre-disaster countermeasures and taking action to prevent disasters is challenging.
Disaster education with student interaction raises students' awareness of the importance of countermeasures (Shiwaku et al. 2007).Participatory mapping is an effective tool for raising awareness of disaster risk among young people, as it can make disaster-related concepts tangible for everyone (Cadag and Gaillard 2012).For example, making children aware of the routes they usually take to school and how a disaster would affect their location could catalyze two-way conversation.Understanding whether children are aware of such risks may lead to developing measures to protect them from disaster.Yasuda et al. (2012) analyzed elementary school students and their parents who commute to and from school in areas at risk of a tsunami in Wakayama Prefecture, Japan.They found that the local community has fostered a sense of community-wide protection against tsunami disaster prevention and regularly participates in tsunami evacuation drills every year.
GPS is an excellent tool for understanding people's behavior in detail.In the tourism field, Hardy et al. (2017) and Muthiah et al. (2018) suggested tracking the movements of tourists to reveal the extent to which they stay at local tourist attractions and sites.Kikuchi et al. (2012) proposed visualizing crime hot spots by tracking children's movements in crime prevention.Leelawat et al. (2018) tried to follow the trajectory of participants to evacuation sites during evacuation drills.
This study proposes a method to analyze the relationship between students' daily behavior and their experiences of flood damage and preparedness for disaster risk.It is a simple method using questionnaires and GPS loggers and can be applied in mountainous areas in developing countries where equipment is inadequate.In a questionnaire, we ask children in mountainous regions of a developing country about their past flood experiences and flood preparedness.At the same time, we give them GPS loggers to obtain data on their daily commuting behavior to and from school.Finally, we discuss the future of disaster education in this region based on the obtained results.
Previous studies on disaster risk education for flood risk in developing countries include the following.The study focusing on schools in Pakistan pointed out that students anticipate that big floods will damage their houses in the future, although they have not prepared for the countermeasures (Shah et al. 2020).A study of children in Bangladesh identified the impact of floods on children and developed measures to reduce flood damage through traditional methods developed by the community (Martin 2010).Azmi et al. (2020) studied the effectiveness of flood disaster education for Malaysian primary school children.Their results showed that using a learning kit they developed, which incorporates elements to arouse children's curiosity, may lead to successful early disaster education.As can be seen from this, various learning programs have been developed worldwide to address flood prevention for children.
In disaster prevention education, it is essential to visualize and share dangerous areas.Identifying hazardous areas is usually done by participant comments in workshops or field surveys (Aini et al. 2011).Participants' statements sometimes point to the wrong places, and field surveys require a great deal of preparation, including the scope of the survey and the selection of participants.The novelty of this study lies in its ability to accurately obtain GPS data on children's usual behavior and its relationship to their perceptions of disasters.It is easy to get data by simply handing a GPS logger to a child and taking a questionnaire.It is a preliminary step in disaster education and contributes to providing more concrete educational materials.

Methodology
We apply the proposed method in a mountainous area in Nepal.Our study consists of three sub-studies: the first analyses the distribution of children's behavior using GPS loggers (Sub-study 1), and the second elucidates the relationship between children's behavior and flood risk by overlaying it on the distribution of children's behavior (Sub-study 2).Finally, we conducted a questionnaire survey to elucidate children's awareness of floods (Sub-study 3) and recommend local disaster prevention planning.
The reason for using GPS to track children's behavior was to examine their map legibility, accuracy, and their behavior.Before this study, we surveyed local children to determine their map comprehension.Local children were shown a map of the study area and asked in a questionnaire, "Do you know where this area is?"As a result, the percentage of children who answered that they knew this location was 75%.In addition, behavioral surveys such as interviews have the common problem of accuracy declines depending on the time of day; GPS loggers have the advantage of eliminating these problems in a simplified manner.

Case study area
The case study area lies in ward No. 9 of Paluntar Municipality, Gorkha District Nepal, shown in Fig. 1.This is a small area of Paluntar Municipality, which is about 130 km far from Kathmandu in the west direction.The population of Paluntar Municipality as a whole is 37,409, of which ward no. 9 has 4,185 inhabitants, according to Fig. 1 Case study area and location of two schools the Government of Nepal national planning commission central bureau of statistics (2021).The area lies on the mid-hill of Nepal with fragile geologic features, and the main occupation is agriculture.
Many small rivers flow down towards the southern part of the area.During the monsoon season, heavy rainfall occurs in the upstream region of the river.It causes damage to agricultural land, infrastructures, and crops in the downstream part.The heavy rain also damages bridges, roads, and sometimes houses nearby.Once in twenty years, an enormous flood occurs in the river, damaging lots of agricultural land properties.This large disaster happened in recent July 2021.The community people have learned to cope with the frequent floods in the area.However, they were helpless against the devastating floods of 1996 and 2021.The floods divided the community and caused them economic hardship.
The broad catchment area of the river at the site location is 28.9 km 2 .The rainfall in the site was 1986 mm in 2022, according to our investigation.Based on our hydrological calculation, there may be a maximum flow of 561.4m 3 water in the river in 100 years return period, similarly considering 50 years and 30 years return period full flow in the river will be 280.7 m 3 and 168.4 m 3 , respectively.Considering the width of the river is 20 m, the expected flow height in the river for 100 years, 50 years, and 30 years of return period was 1.8 m, 2.6 m, and 4.25 m, respectively.
The ward has two schools: Dharmodaya Higher Secondary School and Shree Madane Danda Secondary School shown in Fig. 1.Dharmodaya Higher Secondary School was established in 1995, and currently, there are 296 students.Among these, 156 students are females, and 130 students are males.The age of the students varies from five years to 18 years.Shree Madane Danda Secondary School was established in 1914, and currently, there are 371 students.Among these, 181 students are females, and 190 students are males.The age of the students varies from five years to 18 years.Some students walk to school from nearby areas, but many commute by bus from further afield outside the region.The buses are full of students in the morning and evening, providing a convenient way to get to school.There are several routes to Dharmodaya Higher Secondary School, but they almost always pass through small rivers.It can be assumed from this that at least most of the students attending Dharmodaya Higher Secondary School were usually concerned about flood hazards.
Teachers and parents take a coordinated approach to flood management.In a severe flood, teachers instruct students to stay inside the school for their safety, call parents, and request transportation if necessary.If the flood is small enough to allow students to cross the river, the teacher accompanies them and helps them cross.
However, rainfall measurement systems are absent in the area.As there is no meteorological information from government agencies, schools must judge the likelihood of flooding based on their experience.Due to this situation, they occasionally encountered a significant flood during a school lesson.When the river level did not fall when they returned home, and their commuting route was missing, they had to sleep at school.In the meantime, teachers sometimes bought food for their students with their own pocket money.Developing an accurate flood forecasting system will support the schools' flood management.

Understanding disaster preparedness
Before this study, we held a small workshop for two male and two female students and a teacher at Dharmodaya Higher Secondary School in September 2022 to Fig. 2 Small workshop with school students and a teacher understand the current status of their disaster education (Fig. 2).
The following description describes the entire school's disaster preparedness atmosphere from an interview with one teacher and four students who participated in the small workshop.Teachers had taught students what to do during floods in this school.Many students understood the dangers of flooding but had often been in trouble.For example, when flooding occurred on the way to school, some children crossed the flooded route to school together while older children rescued others.They want innovative evacuation drills to protect themselves from disasters.
These facts indicate that this school may have educated students about what areas are in danger during a flood but has yet to focus on individual situations, such as who may be more at risk.It is necessary to understand the actual position from the individual's perspective.

Distribution of children's behavior by GPS logger
To determine the concentration of children at risk of flooding, we hired 45 students: 22 students attending Dharmodaya Higher Secondary School and 23 students attending Shree Madane Danda Secondary School.
Students participating in the study will bring their GPS loggers and go about their daily routines from the afternoon of March 15, 2023, to the morning of March 16, 2023.
The used GPS loggers are the model number SMG-PSFCA, from Suncor company (about 46 USD/piece).They are extremely small and lightweight, measuring 14 × 40 × 22 mm and weighing only 14 g (Fig. 3).Since the device is turned on and given to the students, data can be automatically obtained simply by placing it in their bags or pockets as they go about their daily lives.
As mentioned earlier, during the flood the older children seemed to help the youngest that crossed the flooded road to go to school.It is conceivable that many children still commute the same route both in normal times and during floods.Therefore, in this study the routes taken by GPS are those in normal times, but the analysis is based on the assumption that these routes remain the same even during heavy rain.

Illustration of flood risk
As previously described, significant floods occur in this area about once every 20 years.This study defined flood risk as the extent of inundation caused by the heavy rainfall in 2021.
In September 2022, we interviewed residents to determine the extent of inundation caused by the July 2021 flood and using a printout of aerial photographs taken by a drone, we asked the residents to fill in the inundation area.Figure 4 shows the pictures taken at the time of the interviews.Then we converted the information to digital data using GIS.
According to the interviews, Fig. 5 shows the estimated inundation area, and the photograph taken by a resident shows that most low-lying areas are submerged.

Questionnaire survey
Along with the survey with GPS loggers, we also conducted a questionnaire survey with the same participants.
The survey questions relevant to this study are related to personal attributes such as age, school, family structure, and disaster risk preparedness.We numbered the GPS logger subjects to match them with the survey respondents.Table 1 shows the contents of the questions.For risk perceptions, we refer to the study of Tuladhar et al. (2014), who surveyed Nepal.

Results and discussion
We administered GPS and questionnaires to 45 students but could not collect one student's data from Dharmodaya Higher Secondary School.Therefore, we analyze and discuss using the data of 44 students, 21 from Dharmodaya Higher Secondary School and 23, from Shree Madane Danda school.
Table 2 shows the gender and age of the respondents by the two schools.There are two reasons why the age of some students is above the school's target age.One is that students do not decide on a course of study after graduation and stay in school; the other is that they stay in school during their studies.The tendency of attribution indicates that respondents at Dharmodaya School have a higher number of female students and a higher age range than those at Shree Madane Danda School.Figure 8 shows the total distance children commute to school.Using a GPS logger, we identified the location of the child's home based on the child's movements on March 15.The figure shows that most children commute to school at distances of 1 to 2 km, while some commute at distances of 5 km or more.Based on the GPS tracking intervals, some students may travel by bus, etc. Table 3 shows the student survey responses.It indicates that more than 80% of the students have experienced flood damage in the past and believe that disasters will occur again, exposing the population to risk.On the other hand, they lack awareness of disaster risk preparedness.For example, only 23% of the students have an evacuation plan for their families and it indicates that despite their experience with floods and concern about future floods, their preparedness needs to be improved.
Figure 9 shows the relationship between the behavior of 44 students for one day (from the afternoon of March 15 to the morning of March 16) and the estimated inundation area in 2021.The mesh size of the footprints in the figure is 10 m.The footprints are the recordings from GPS loggers carried by students.The total number of GPS loggers in each mesh is calculated using the spatial computing function of GIS.It clarifies that the number of footprints is higher (red color) in the vicinity of the two schools, indicating that students are most likely to congregate there during the daily activity survey.In addition, the intersections between several paths are obviously redder, suggesting that the students are dispersing and assembling at the crossroads.The problem is commuting to and from Dharmodaya Higher Secondary School.As shown in the figure, there is only one way to get to this school except through a flood-prone area, meaning that students may be exposed to extreme danger during the monsoon.
Figure 10 shows the relationship between temporal changes in children's moving behavior and flood risk, particularly in hazardous situations.The points in the figure indicate individual students; different colors mean different students.Each time map illustrates the past 30 min footprint using ArcGIS Tracking Analyst.The figure explains that children pass through high flood-risk areas at 15:30 when they return home and at 10:00 when they go to school.If a similar flooding occurs around 9:00 a.m., before many students are on their way to school, or around 2:00 p.m., before they return home from school, they must decide to change their route to school or abandon their commute.Also, the number of students commuting to school through the flood area is 15, and all of these students were from Dharmodaya Higher Secondary School, which is located near the flood zone.
Table 4 compares the results of the questionnaire responses with and without using flooded areas as a commuting route.The number of students using the area is 15, categorized in the former paragraph.This table tabulates each of the questions in Table 1.Non-responses were omitted from the total.The right column in the table shows the p-values calculated by Fisher's exact test.It indicates that students who go through flooded areas as a route to school tend to experience damage from the past flood, such as hunger, and are concerned about future flood damage (Q3, Q 11, Q 12, Q 13).Despite the awareness of such experiences, the responses to the  Have you ever experienced flooding at your home site (including the condominium site, same as below)?7 Have you ever experienced flooding on your home site?8 Have you seen flood damage?9 Have you ever experienced flood damage from a relative or someone close to you? 10 Have you ever crossed a flooded river?11 Have you slept on an empty stomach due to flooding?12 Have you reduced the number of meals per day due to flooding?13 Have you gone to school on an empty stomach due to flooding?14 Have you benefited from school-based feeding schemes?15 Have you made a family emergency plan?16 Have you ever practiced for an emergency at your own home?providing disaster prevention education to avoid injury, such as giving detour routes and cancelling commuting to school during floods.Also, among the students who commute to school in the flood zone, three are in the 2 km range, and two are over 5 km.In the case of students commuting more than 5 km, they may be traveling by car, etc.Therefore, it is necessary to take preventive measures such as suspending the operation of buses on days when flooding occurs.
In addition, it clarified that four students (two at each school) must pass through the river to get to school due to the bridge collapse shown in Fig. 7.It is hazardous, especially during the rainy season, unless the bridge is restored.Schools and the local community should teach students about the possibility of suffering damage on the way due to a sudden rise in river water levels.
When we held a debriefing of the survey results in the area a few months later, many residents were very interested in the risk mapping that we had done in combination with the questionnaire and GPS survey.In a questionnaire survey conducted after the meeting, approximately 60% of the residents indicated that the survey would be helpful for disaster prevention planning.The method proposed in this study to elucidate the relationship between disaster risk hazard and disaster awareness based on children's daily behavior using GPS and questionnaires is very simple.By collecting data from a more significant number of students, it may be possible to propose customized disaster education for each student.

Conclusion
Increasing hazard awareness of disaster risk promotes disaster mitigation and contributes to realizing a sustainable society.Since children are less knowledgeable about disasters and more vulnerable to them, the best way to protect them from catastrophes is to provide the appropriate information.This study proposed a method to analyze the relationship between students' daily behavior and their experience of flood damage and preparedness for disaster risk, using a questionnaire and GPS loggers.Results revealed that some students who commute to schools near flooded areas use the areas as a morning and evening route to school.Our field observations and interviews of teacher-student relationships showed that students were very attentive to their teachers' instruction.Teachers understand the dangers of heavy rainfall and how to prepare for it, but students need to learn how to deal with it.In other words, the school staff 's disaster education to the students effectively reduces disasters.Early warning is needed when severe floods are expected during the rainy season.Since there are no water level gauges in the area, one possible approach would be to rely on teachers to assess the rainfall situation early.In addition, it found that such students have been affected by flood hunger in the past, and despite their concern about future flood damage, they have not been able to take adequate countermeasures.We could analyze the data in such detail using a simple method such as the one used in this study.
Visualization of local disaster risks through street walks and inspections of school routes conducted with students is effective for children's community disaster prevention.However, the number of participants in such events is limited, and it is impossible to plot the movements of all students.In addition, as mentioned above, the comprehension rate of maps among children in this area is 75%, which poses a challenge for checking indoor risks using maps.The GPS survey proposed in this study can accurately determine children's routes simply by lending them Another countermeasure is for schools to make simple water level gauges and have students take turns measuring the water level.Over an extended period, this approach will improve students' understanding of the flooding risk and provide them with early warning.Erediano et al. (2021) reported that students' awareness of flood preparedness has increased due to monitoring simple water level gauges in the case study in the Philippines.
Note that there are constraints on the student's behavior.From our observations of teacher-student relationships in the field, most students followed their teachers.Even if advanced disaster management information can be established, such as water level gauges, if the school does not inform the students to refrain from commuting, students are likely to go to school as usual.Therefore, when teachers expect flooding, they should inform the students to avoid going to school.Fortunately, some students have smartphones and a good communication environment.For example, one idea is to use smartphones to provide information from the school to some students and to share it among them.
The limitation of this study is as follows.First is the refinement of flooded areas.In this study, the extent of flooded areas was visualized by interviewing residents who lived near flood-damaged regions in 2021.However, all roads in the area were unpaved, which suggests that the site may have been damaged by the inundation even if it was not flooded.These may have affected the students' commuting to school, but this study has not clarified the problem.Therefore, collecting more information on flood damage and analyzing the relationship between students' disaster experience, disaster exposure, and risk perception is necessary.The second is the feedback on the results of this study to the local communities and suggestions for improving their behavior.This study is limited to analyzing students' exposure and perception of disaster risk based on their daily conduct and does not provide feedback on these results to the local community.Understanding their behavior may motivate them to learn about disaster risk reduction and discuss ways to improve their behavior.It is essential to improve the method by installing panel surveys and other forms to make it more useful as a resource for disaster risk reduction education.
Fig. 3 Size of GPS logger

Fig. 7
Fig. 7 Effects and damage caused by the flood in 2021

Fig. 8
Fig. 8 Distribution of respondents' commuting distances to school

Fig. 10
Fig. 10 Footprints and time of the day of respondents passing through flooded areas

Fig. 9
Fig. 9 Relationship between flooded area and number of footprints

Table 1 Questions
No.Question 1 Do you think flood will hit the around your school again? 2 Do you think flood will damage the houses around your school? 3 Do you think people will be killed by a major flood around your school? 4 Do you think communities living in the proximity of the main river is at great risk? 5 Do you think basic supplies (water, electricity, etc.) will be interrupted during disasters?6

Table 2
Personal attributes of participants

Table 3
Respondents' risk experience and awareness

Table 4
Comparative results of risk experience and awareness map, and with time, it is possible to create a risk map based on a complete survey. a