Using virtual reality to study human response to flood risk across controlled experiments

Flood consequences are related to human risk perception and behavior during emergencies. However, the lack of real-world data makes it difficult to study how people response. Taking advantage of virtual reality (VR), this study designs VR-based flood scenarios by simulating rising water in the city to study how human responses vary when humans are on the street or in the car under different conditions of rain level, visibility level and warning modality. Also, learning from protection motivation theory (PMT), this study investigates how the contextual information, psychological and socioeconomic factors affect human response during flood emergency. Results from 50 participants show that: (1) Females exhibit longer response times and greater tolerance for flood risk compared to males; (2) Risk recognition takes less time when people are in the car than on the street; (3) ANOVA results show significant differences in response times across controlled experiments; (4) Linear mixed-effects models suggest that human responses are associated with contextual, psychological and socioeconomic factors, for example, low visibility significantly correlate with short response time, self-efficiency positively correlate with earlier risk recognition, higher education and income may be associated with faster responses. The findings highlight the variability in human perception across contexts and the value of human–VR interaction for studying individual-level risk perception and behavior. • People are more sensitive to flood risk when they are in a car than on the street, especially at night. • Alerts can encourage people to take earlier action, but alerts cannot guarantee evacuation behavior. • Higher self-efficacy may be positively associated with earlier response in the event of flooding. • Income and education levels are significantly associated with people’s reaction time to Flood risk.