• Berg, L. P., and J. M. Vance, 2017: Industry use of virtual reality in product design and manufacturing: A survey. Virtual Real., 21, 117, https://doi.org/10.1007/s10055-016-0293-9.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Bostrom, A., R. Morss, J. K. Lazo, J. Demuth, and H. Lazrus, 2018: Eyeing the storm: How residents of coastal Florida see hurricane forecasts and warnings. Int. J. Disaster Risk Reduct., 30 ,105119, https://doi.org/10.1016/j.ijdrr.2018.02.027.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Huang, Q., and Y. Xiao, 2015: Geographic situational awareness: Mining tweets for disaster preparedness, emergency response, impact, and recovery. ISPRS Int. Geo-Inf., 4, 15491568.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • R Development Core Team, 2018: R: A language and environment for statistical computing. R Foundation for Statistical Computing, www.r-project.org.

  • Regnier, E., 2008: Public evacuation decisions and hurricane track uncertainty. Manage. Sci., 54, 1628, https://doi.org/10.1287/mnsc.1070.0764.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Sundar, S. S., J. Kang, and D. Oprean, 2017: Being there in the midst of the story: How immersive journalism affects our perceptions and cognitions. Cyberpsychol. Behav. Soc. Netw., 20, 672682, https://doi.org/10.1089/cyber.2017.0271.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • View in gallery

    Screenshots of various stages of category 3 tropical cyclone landfall virtual reality simulation.

  • View in gallery

    A mock tropical cyclone advisory, typical of a National Hurricane Center product, used in the survey.

  • View in gallery

    Survey questions on behavioral intentions and descriptive statistics of the two response groups, where 1 is very unlikely and 5 is very likely. Bold indicates statistically significant differences between the two groups.

  • View in gallery

    Behavioral intentions, in response to Likert-style questions, where 1 is very unlikely and 5 is very likely, of 124 individuals surveyed. One group of 62 individuals saw both the mock tropical cyclone advisory products and virtual reality simulation before responding (dark blue bars), while another group of 62 individuals saw just the mock tropical cyclone advisory products before responding (light blue bars).

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 102 102 53
PDF Downloads 55 55 25

Communicating Hurricane Risk with Virtual Reality: A Pilot Project

View More View Less
  • 1 Department of Geology, Environment and Sustainability, Hofstra University, Hempstead, New York
  • 2 Maurice A. Deane School of Law, Hofstra University, Hempstead, New York
  • 3 Department of Educational And Research Technology Services, Hofstra University, Hempstead, New York
  • 4 Department of Geology, Environment And Sustainability, Hofstra University, Hempstead, New York
© Get Permissions
Free access

Abstract

Landfalling hurricanes in the United States can inflict extreme damage and loss of life. The latter, particularly, can be caused by a host of socioeconomic factors, including insufficient understanding of risk by individuals expected to be impacted by the storm. Thus, we test the use of an emerging technology, virtual reality (VR), to enhance the communication of real-time risk from a hurricane forecast to make landfall. In this pilot study, individuals are presented with a hypothetical scenario where a major hurricane is forecast to impact their community within 48–72 h. The survey includes two different types of warning products related to the hypothetical hurricane: static text and maps emulating those traditionally used by media outlets and local officials to communicate risk, and a VR video simulating a hurricane landfall in a residential neighborhood. We survey two groups of equal size (each n = 62), one viewing both the VR simulation and traditional products, and the other only the latter. Each group was then asked a series of Likert-scale and open-ended questions to assess the effectiveness of both products. We determine that participants viewing both the VR and traditional products are significantly more likely to take action in preparation for the hypothetical landfall than those being exposed to just the traditional products. These results demonstrate that VR can be a useful component of hurricane warning products, and further work can be done to improve the effectiveness of such products and assess how broader segments of the population can access this information.

© 2019 American Meteorological Society. For information regarding reuse of this content and general copyright information, consult the AMS Copyright Policy (www.ametsoc.org/PUBSReuseLicenses).

CORRESPONDING AUTHOR: Jase Bernhardt, jase.e.bernhardt@hofstra.edu

Abstract

Landfalling hurricanes in the United States can inflict extreme damage and loss of life. The latter, particularly, can be caused by a host of socioeconomic factors, including insufficient understanding of risk by individuals expected to be impacted by the storm. Thus, we test the use of an emerging technology, virtual reality (VR), to enhance the communication of real-time risk from a hurricane forecast to make landfall. In this pilot study, individuals are presented with a hypothetical scenario where a major hurricane is forecast to impact their community within 48–72 h. The survey includes two different types of warning products related to the hypothetical hurricane: static text and maps emulating those traditionally used by media outlets and local officials to communicate risk, and a VR video simulating a hurricane landfall in a residential neighborhood. We survey two groups of equal size (each n = 62), one viewing both the VR simulation and traditional products, and the other only the latter. Each group was then asked a series of Likert-scale and open-ended questions to assess the effectiveness of both products. We determine that participants viewing both the VR and traditional products are significantly more likely to take action in preparation for the hypothetical landfall than those being exposed to just the traditional products. These results demonstrate that VR can be a useful component of hurricane warning products, and further work can be done to improve the effectiveness of such products and assess how broader segments of the population can access this information.

© 2019 American Meteorological Society. For information regarding reuse of this content and general copyright information, consult the AMS Copyright Policy (www.ametsoc.org/PUBSReuseLicenses).

CORRESPONDING AUTHOR: Jase Bernhardt, jase.e.bernhardt@hofstra.edu

Virtual reality (VR) has gained some acceptance as a technology for visualization in various fields. As a technology in an industrial setting, for example, it is considered “mature, stable, and, most importantly, usable” and “helped inform the scientific inquiry process” (Berg and Vance 2017). Immersive experiences such as VR can trigger emotional responses beyond the typical reaction to static text. Recent work by Sundar et al. (2017) showed that survey participants reading New York Times stories including a VR supplement recalled and believed the credibility of an article more readily than their counterparts who only read it in text form. Might VR also be useful in stimulating people to seek safety or otherwise respond to hurricane warnings? We are seeking to answer that question by comparing a text warning with a VR simulation of a hurricane landfall to see if the VR compels individuals to take a predicted tropical cyclone (TC) landfall more seriously.

METHODOLOGY.

The VR simulation.

We developed a VR simulation of a category 3 TC landfall using Unity. Unity is a professional development platform used to create three-dimensional video games, virtual environments, and immersive simulations. The application allows inclusion of 3D models representing real objects such as houses, streets, cars, and people. These models can be modified by programming code and instructed to behave in specific ways, either in an automated fashion or based on interaction with the user. The completed simulation is exported to a format that permits playback on a VR headset known as a head-mounted display (HMD); users wearing the headset are fully immersed in the simulated environment, complete with a 360° view and surround sound audio.

The VR simulation (a 360° YouTube video1 is available at www.youtube.com/watch?v=XWZCXdkVPZQ) lasts approximately 90 s, capturing the essence of a major TC landfall, and allowing for user interaction with surround sound. When first beginning the experience, the participant finds him or herself on the first floor of a typical single-family home, looking out the window onto a tree-lined, suburban-style street, assumed to be in a low-lying area adjacent to the coast. A large area of text appears, displaying the Saffir–Simpson category (3) of the TC being simulated, along with the winds typical of such a storm, in customary units (Fig. 1a). After 10 s, the text disappears, and the user can observe the extreme conditions outside, highlighted by heavy rain, strong winds, and several large pieces of airborne debris—one of which shatters the window in front of them.

Fig. 1.
Fig. 1.

Screenshots of various stages of category 3 tropical cyclone landfall virtual reality simulation.

Citation: Bulletin of the American Meteorological Society 100, 10; 10.1175/BAMS-D-17-0326.1

Although the viewer must remain at a fixed location within the home, they are free to look in any direction (i.e., up, down, left, right). When looking back into the home, for example, the viewer notes that the strong winds blow through the window and knock over common household items such as a television stand (Fig. 1b). As the viewer turns back to look outside, the conditions turn more severe (Fig. 1c). Storm surge flooding begins, with water rising into the street, transporting large objects such as parked cars. Moreover, the continued strong winds blow off shingles, initiate partial roof failure, and knock down tree branches. Eventually, the storm surge reaches the home, with water pouring through the open window, surrounding the viewer, and reaching chest level (Fig. 1d). At this point, the VR simulation ends.

Special care has been taken to ensure scientific accuracy of the VR simulation. Category 3 TCs can cause major damage to well-built homes, such as those shown in the simulation. Moreover, storm surge is dictated by factors beyond Saffir–Simpson category, such as TC size, central pressure, and speed of motion. We select a surge of several feet, which reaches chest-level of a human in a home near the coast.

The survey.

To make a preliminary assessment of the influence of our VR simulation on behavior, we disseminated a survey from October through December 2017 on campus at Hofstra University, on Long Island, New York. We presented individuals with a hypothetical scenario where a category 3 TC was forecast to make landfall on Long Island in 48–72 h. We selected that time frame because at that point individuals are typically considering a potential evacuation (e.g., Regnier 2008) and making final reinforcements to their home (such as covering windows).

To elucidate the forecast, we created a mock TC forecast track map, a rough version of the frequently viewed product often conveyed on television, in social media, or other means (e.g., word of mouth; Fig. 2). Further, a brief text message with key details on the storm (i.e., wind speed and location) and potential impacts accompanied the map. That text was meant to briefly encapsulate some of the most popular information a typical person consumes 2–3 days prior to a potential landfall (e.g., as identified in Bostrom et al. 2018).

Fig. 2.
Fig. 2.

A mock tropical cyclone advisory, typical of a National Hurricane Center product, used in the survey.

Citation: Bulletin of the American Meteorological Society 100, 10; 10.1175/BAMS-D-17-0326.1

Within our survey, after being asked a set of demographic questions, each participant was shown the contextual information contained in Fig. 2. Then, the participants were randomly split into two equal groups. One group viewed the VR simulation before proceeding to a mixture of Likert-style (i.e., scale of 1 to 5, where 1 is very unlikely and 5 is very likely), open-ended, and dichotomous questions meant to assess the individual’s likelihood to prepare for the hurricane, such as stocking up on supplies, notifying elderly neighbors, and considering evacuation (Fig. 3). The second group moved on directly to those questions without viewing the VR simulation. At the end of the survey, each group was also asked to identify the products that were most and least useful in informing their decisions. Last, the second group was shown the VR simulation at the conclusion of the survey, and asked an additional question on whether their previous answers would have changed after viewing the VR.

Fig. 3.
Fig. 3.

Survey questions on behavioral intentions and descriptive statistics of the two response groups, where 1 is very unlikely and 5 is very likely. Bold indicates statistically significant differences between the two groups.

Citation: Bulletin of the American Meteorological Society 100, 10; 10.1175/BAMS-D-17-0326.1

A total of 124 individuals were surveyed, 62 in each group. Given that we distributed the survey on a college campus, the demographics reflected that location. The vast majority of individuals surveyed were under the age of 25 (80.6%), while 46.0% lived on campus at Hofstra University. Moreover, given the recent impact from Superstorm Sandy on Long Island in October 2012, 71.8% of respondents reported having previously experienced a hurricane, although Sandy’s winds were primarily of tropical storm force over land. Further, most (66.1%) respondents identified themselves as not living in a category 3 flood zone, based on a Federal Emergency Management Agency (FEMA) map provided.

RESULTS.

Viewing the VR simulation had a considerable impact on the behavioral intentions of survey participants (Fig. 4). Given the relatively small sample size of each survey group, a Mann–Whitney U test was run using R software (R Development Core Team 2018) to evaluate the hypothesis of a statistically significant difference between the two groups (VR in the middle of survey, VR at the end of survey) in response to the five Likert-scale questions.2 Based on that test, significant differences were found in the responses of the two groups to the Likert-scale questions asking for the likelihood of making alterations to one’s home, notifying elderly neighbors, evacuating to a path farther away, and evacuating to a storm shelter. The only Likert-scale question without a significant difference between the groups was the likelihood of stocking up on supplies, while the greatest numerical differences in the mean responses were for the two evacuation questions. Further, when asked the dichotomous “yes or no” question, 55 of 62 participants who viewed the VR simulation stated they would evacuate, while only 45 of 62 participants who did not see it responded affirmatively.

Fig. 4.
Fig. 4.

Behavioral intentions, in response to Likert-style questions, where 1 is very unlikely and 5 is very likely, of 124 individuals surveyed. One group of 62 individuals saw both the mock tropical cyclone advisory products and virtual reality simulation before responding (dark blue bars), while another group of 62 individuals saw just the mock tropical cyclone advisory products before responding (light blue bars).

Citation: Bulletin of the American Meteorological Society 100, 10; 10.1175/BAMS-D-17-0326.1

Moreover, important trends emerged when we performed content analysis on qualitative responses to the open-ended questions. Specifically, we asked which aspects of the products shown were most and least useful in informing their hypothetical decisions. Of the respondents who viewed the VR, 28 indicated that the simulation was most useful, while only 17 mentioned the more traditional products. Some participants also identified specific aspects of each set of products as being most helpful. For example, 18 individuals listed the flooding and surge in the VR simulation as being most useful, while only 3 mentioned the winds as shown in the VR. Several responses enthusiastically endorsed the utility of the simulation, including one observation that “the VR experience was a cool aspect that would catch peoples’ eyes. Most people wouldn’t read a pamphlet or look up what to do during a hurricane until after they live through one. The VR experience can create the sense of fear and make people think about what they actually would do.” Meanwhile, just 17 participants described the mock traditional products as being most useful, 10 indicating the best-track map in general, and 7 the hurricane strength as depicted in the map. Moreover, for the non-VR group, half of the participants (31 of 62) identified the mock traditional products as being most useful, 18 indicating the best track map, and 13 the strength.

Additionally, an analysis of the follow-up responses of the group who viewed the VR simulation at the end of the survey further revealed the important impact it had on respondents’ anticipated behavioral response. Of the 17 individuals in that group who stated they would not evacuate, 8 responded that viewing the VR would change their minds, while 3 stated it would not or they were not sure, and the rest did not submit a response. Even those who already confirmed they would evacuate without viewing the VR were further influenced by seeing the simulation; 25 of the 47 indicated they would change their response (i.e., take warnings even more seriously), for example, one respondent noted, “I would pay more attention to flood projection and take further measures in regards to flood control.” Such a response further reinforced the conclusion that the major surge shown in the simulation was impactful for survey participants. Last, although further division of the survey population into demographic groups resulted in generally insufficient sample sizes, initial results indicated that some demographic factors might influence behavioral intentions, particularly gender, where significant differences were present. For example, across both survey groups, male responses to the Likert-scale questions were lower than that of females.

DISCUSSION AND CONCLUSIONS.

The results of this pilot study indicate—in a preliminary way—that viewing a VR simulation of a category 3 TC landfall in addition to more traditional products encourages survey respondents to take warnings more seriously when placed in a hypothetical scenario. This finding is supported by significant differences between the two survey groups in four of the five Likert-style questions on behavioral intentions contained in the survey. The one question without a significant difference, on likelihood to stock up on supplies, is found to be reasonable given the well-known tendency of individuals to procure essentials from grocery stores in advance of forecast severe weather, such as prior to Sandy in October 2012 (e.g., Huang and Xiao 2015). However, individuals are far more hesitant to take more extreme action, such as evacuation, which requires substantially more time and resources, and may be intractable for some due to concerns such as health issues or financial instability.

Results suggest there is value to the VR simulation we developed. There may be future utility to using VR as a supplement to traditional hurricane warning products. VR simulation, especially the sensation of being surrounded in your own home by rapidly rising storm surge, can produce a visceral reaction in its viewers—perhaps enough to convince them to evacuate even when other personal factors would make such an action difficult.

Moreover, as VR technologies improve and become ubiquitous, we expect continued application of these tools to enhance weather warning communication. We are currently expanding the survey described in this short paper to a nearby community vulnerable to TC landfalls—Long Beach, a barrier island on the Atlantic coast—in order to survey a more diverse audience. Older people may be less familiar with immersive technologies than the relatively young subjects of this pilot study, and they may trust it less as well. VR can be integrated with other emerging technologies, such as augmented reality, to create a broader suite of interactive hurricane warning products, and can also be applied to other forms of severe weather, including tornadoes.

VR simulation may show promise as a supplement to the general awareness social media posts and preparedness literature disseminated, in the off-season, for example, by National Weather Service Forecast Offices and other federal agencies (e.g., National Hurricane Center). VR is viewable not only using expensive equipment, but also with inexpensive, portable, and easily available technology such as Google Cardboard. VR simulations might someday complement other products that portray risk from each hurricane hazard (e.g., wind, storm surge, flooding, and tornado) separately. Great care must be taken, however, to ensure that VR does not also mislead populations less vulnerable to a particular storm and convince them to evacuate when they should not. Thus, additional research into the effectiveness of VR is necessary, including further tests of real-world applicability with broader survey samples and close evaluation of the “false alarm” rate of such products.

ACKNOWLEDGMENTS

The work was supported by funds from Hofstra University which allowed for robust collaboration between the coauthors, comprising undergraduate students, a faculty member, and administrators in the university’s Education Technology department. Moreover, we thank Art DeGaetano for his review of a draft manuscript. Last, we are grateful to those who chose to participate in the survey.

FOR FURTHER READING

  • Berg, L. P., and J. M. Vance, 2017: Industry use of virtual reality in product design and manufacturing: A survey. Virtual Real., 21, 117, https://doi.org/10.1007/s10055-016-0293-9.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Bostrom, A., R. Morss, J. K. Lazo, J. Demuth, and H. Lazrus, 2018: Eyeing the storm: How residents of coastal Florida see hurricane forecasts and warnings. Int. J. Disaster Risk Reduct., 30 ,105119, https://doi.org/10.1016/j.ijdrr.2018.02.027.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Huang, Q., and Y. Xiao, 2015: Geographic situational awareness: Mining tweets for disaster preparedness, emergency response, impact, and recovery. ISPRS Int. Geo-Inf., 4, 15491568.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • R Development Core Team, 2018: R: A language and environment for statistical computing. R Foundation for Statistical Computing, www.r-project.org.

  • Regnier, E., 2008: Public evacuation decisions and hurricane track uncertainty. Manage. Sci., 54, 1628, https://doi.org/10.1287/mnsc.1070.0764.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Sundar, S. S., J. Kang, and D. Oprean, 2017: Being there in the midst of the story: How immersive journalism affects our perceptions and cognitions. Cyberpsychol. Behav. Soc. Netw., 20, 672682, https://doi.org/10.1089/cyber.2017.0271.

    • Crossref
    • Search Google Scholar
    • Export Citation
1

The video can be viewed on a traditional two-dimensional display with the ability to pan or as a virtual reality experience by utilizing technology such as Google Cardboard and the YouTube application on a smartphone.

2

Stock up on supplies, p = 0.0511, U = 2281, Cliff’s delta = 0.187; make alterations to home, p < 0.001, U = 2657, Cliff’s delta = 0.382; notify elderly neighbors, p = 0.00796, U = 2432, Cliff’s delta = 0.265; evacuate to storm shelter, p < 0.001, U = 2887, Cliff’s delta = 0.502; evacuate to path farther away, p < 0.001, U = 2972, Cliff’s delta = 0.546.

Save