QuickAware: A Virtual reality tool for Quick Clay Landslide Hazard Awareness

In recent years, the area of application of Virtual Reality (VR) has grown tremendously from the entertainment industry to the military, to mental health, to hazard identi�cation training, and to our daily lives. VR has been widely employed in hazard identi�cation and prevention, safety training, evacuation, search and rescue, and damage identi�cation of hazards. This paper investigates the application of VR for raising awareness about quick clay landslide hazard. Quick clay landslides are geological hazards that are often triggered suddenly, are di�cult to forecast, and often result in disastrous consequences. Currently, hazard and risk maps are used to communicate quick clay landslide risk to the public. However, these maps are mostly expert speci�c, and they may not appear convenient to communicate natural hazards such as quick clay landslides to the public. VR technology can be used to further enhance the communication of landslide risk to the public by developing simulations that can raise awareness about, among others, landslide initiation signs, preventive measures, safety training, and actions that can increase safety of individuals in a hazard event. To narrow this communication gap, we developed a tool, QuickAware, that can help in improving awareness of quick clay risk using a VR application. The development process of the tool started with a co design approach where stakeholder experts were brought to collaborate in setting up VR scenarios and de�ning the VR environment contents. The contents were then conceptualized and translated into VR experience. The usability of the VR application was examined by conducting a survey with 16 participants. The results of the survey indicated that the tool has a promising value in improving awareness creation for quick clay landslide hazard.


Introduction
In recent years, the area of application of Virtual reality (VR) has grown tremendously from the entertainment industry to the military, to mental health, to hazard training, and to our daily lives.VR can be de ned as a three-dimensional(3D), computer generated environment with which a user can interact, and feel the sense of presence (Steuer 1992).VR enables a user to interact with and be immersed in the environment that simulates reality using immersive technologies such as head-mounted displays (HMDs) and controllers.
VR technologies offer new opportunities to safely and realistically recreate situations that are di cult to investigate in real life such as dangerous situations and areas that are expensive or di cult to explore.Its immersive nature enables users to respond to situations presented in VR as if they were real.Furthermore, its interactive nature enables users to engage within virtual simulation environment and allows them to experience the consequence of their engagements.Both the immersive and interactive capabilities provide VR a purpose to observe how people react to VR environment and to teach how to deal with VR environment components respectively.Many public authorities in different countries invested substantial efforts in generating and improving knowledge of hazardous processes and in collecting and mapping data.For instance, the Norwegian national database for ground investigations (NADAG) has collected almost 370,000 borehole ground investigation results that can assist in landslide hazard zonation and other quick clay issues (NGU 2020).
Similarly, more than 2300 zones have been identi ed and mapped as quick clay zones in Norway (Ryan et al. 2022).However, describing a potential damage that can be caused by a certain quick clay landslide hazard to a lay person is challenging, as there is not an elaborated way of communicating these mappings and other risk assessment results to the public.A survey carried out by (Pedoth et al. 2021) on risk communication of natural hazards supports this statement.According to the survey there is a lack of knowledge about existing hazard maps and the public wants to be better informed and more involved.
Even community leaders and decision makers may have di culties in imaging the real consequences of a natural hazard and this may result in low-risk awareness and poor hazard preparedness in the vulnerable communities.The low-risk awareness of the public may leave the whole burden of hazard risk management process on few experts.
VR is well suited to address such problems as it has an immense potential for conveying valuable information about natural hazards to users who experience its immersive and realistic presentation.In this paper, we developed a tool, QuickAware, that can help in improving risk awareness of quick clay landslide hazards using a Virtual Reality (VR) application.
The rest of this paper is organized as follows: the background and related works section presents some background information about quick clay and VR applications.The Methodology section presents our methodology which entails the requirement gathering, VR planning, materials, and resources we used, system design and development, and the evaluation method.The implementation and evaluation section presents about the VR equipment employed and the usability evaluation tests conducted.Finally, the discussion and conclusion sections discuss the results and summarized the work respectively.

Background And Related Works
In this section we presented the background information about quick clay and quick clay landslides in the Norwegian context and the general application of VR in awareness creation and education.

Quick Clay Landslides
Quick clay soils are sensitive ne-grained soils characterized by a tremendous reduction of undrained shear strength upon disturbance due to loading.They can be so rm in the undisturbed form, but ow as a uid when overloaded.Quick clay behavior is primarily described by a material property known as sensitivity.Sensitivity de nes the ratio between the intact (undisturbed) and remolded (disturbed) shear strengths.Quick clay soils are mostly found in the northern parts of the world which include Scandinavia, Canada, Alaska, and Russia(NGI n.d).In Norway, quick clay is de ned traditionally by a remolded shear strength of less than 0.5 Kpa and sensitivity of greater than 30 (NGF 1974).
Quick clay landslides can develop rapidly when the initially rm clay is exposed to sustained loading that can lead to remolding.After remolding, the clay masse is transformed into a oating liquid in its own pore water.The causes for the remolding can be natural forces or man-made actions.Quick clay landslides can quickly propagate retrogressively backwards, and the debris can travel fast covering a considerable distance.Quick clay landslides can be triggered suddenly and are di cult to forecast.As a result, there is no early warning system.To reduce the risk of quick clay landslides in Norway, mapping of potential areas for quick clay started in the 1980s after the Rissa landslide, and this work is still ongoing.More than 100,000 people live in areas that have been mapped as containing quick clay soil and many more live in areas where quick clay can potentially occur (Ryan et al. 2022).As of today, more than 2,300 known quick clay zones, from which around 1,000 are high-risk zones, have been identi ed.Even though erosion from streams and rivers is the most common natural triggering reason, most large quick clay landslides are still triggered by humans.Excavation, land lling, and construction work plays the biggest role.In the last 70 years, the number of large quick clay landslides in Norway that have been triggered by human activity, of which various construction activities have been the main cause, is more than 50%.A survey by (L' Heureux et al. 2018) showed that between the years 2010 and 2018, about 90% of the incidents have been triggered this way.
Currently, there are no early warning systems designed for quick clay landslides.However, there are several structural mitigation measures such as erosion control, terrain modi cations to improve safety factors, and soil reinforcements.Some non-structural measures such as construction monitoring and banning certain human activities can also be considered mitigation measures.However, the fact that quick clay soil is dominant in densely populated areas makes banning developments unrealistic and unwanted.Therefore, the stakeholders and the public must be aware of the ground conditions on the site one operates, and if there is a potential for quick clays, understand the causes and consequences of possible landslides.VR offers a great opportunity to prepare a virtual eld trip in the domains of geography, geology, geomorphology, and astronomy without visiting the site, in a risk-free, cost-friendly, and time-saving manner.In geology, VR enables trainees and students to experience real life geological outcrops and get a comprehensive understanding of depositional environments or structural features, as shown in (Minocha 2014;Dinis et al. 2017;Pugsley et al. 2022).VR can also be used as a valuable resource for improving the teaching and learning process in geography (Bos et al. 2022;Wright et al. 2022).

Requirement Gathering: A nity Diagram
A co-design approach, similar to the one implemented by Gamberini et al. (2021), was adopted to create an immersive and interactive virtual experience to improve risk awareness of quick clay landslides.The co-design was comprised of different experts (N = 9; Table 1) from several elds of study.They were experts in geotechnical engineering, hydrology, psychology, electronic systems, and VR.The co-design approach includes a nity diagram and brainstorming sessions.The experts were informed about the general requirements of the VR tool prior to the co-design sessions.
A nity diagram is a tool used to generate, organize, and consolidate information related to a complex issue, or problem.The general objective of the a nity diagram session was to gather requirements useful for de ning the contents of the VR tool and planning their translation into VR educational experience.
Two a nity diagrams were organized, one focusing on the moments that occur prior to a quick clay landslide, and the other focusing post landslide period.Both sessions involved all the experts from Table 1 with E3 and E5 serving as conductors and the rest seven as participants.The role of the conductors was facilitating the whole a nity program by creating an amiable environment and introducing the activities the rest of the participants.Five of the participants were gathered in a room and chairs were arranged in an oval shape, while two of the participants joined the session online.On the two narrow sides of the oval, there was a white board on which sticky notes were posted and on the opposite direction a screen on which E4 and E9 joined the session online.Before the start of the two sessions, a short presentation about quick clay landslides, VR and its application in natural hazard risk emergency situations was given by the conductors.
In each of the sessions the conductors raised a very general question on what events or situations should be simulated and represented in VR so that the results will be better in creating awareness about quick clay landslide hazards to the public, politicians, and decision makers.The participants had the task of answering the question and writing it on post-it notes.The answers from E4 and E9 were transferred to the post-it notes by one of the conductors.All the post-it notes were then posted on the whiteboard and read loudly by the conductor (see Fig. 1).The participants were encouraged to discuss the notes, contribute their point of view, and ask questions.The purpose of the discussion was to help the participants to gain an overview of the ideas and achieve common understanding.During the rst a nity diagram session, several ideas related to the time prior to a quick clay landslide were produced.These ideas were placed into four categories namely triggering mechanism, realistic environment, knowledge, and actions.The triggering mechanisms category included processes and activities that could trigger a quick clay landslide.These include erosion process and human-made activities such as lling on the top of a slope and excavation at the toe of a slope.The realistic environment category referred to realistic terrain features, actual infrastructures, and day-to-day activities that people could relate to.
The knowledge category comprised activities related to information and knowledge dissemination aiming at reducing the risk of quick clay landslides.This category included quaternary maps, hazard maps, risk maps, and ground investigation results.Finally, the actions category referred to the activities and situations related to mitigation works that could reduce the risk of quick clay landslides such as river inspection against erosion, erosion protection works, and construction works monitoring.
In the second a nity diagram session, the presented ideas were grouped into ve categories.The categories were dynamics of the landslide, physical consequences of the event, emergency activities, post-emergency activities, and hazard information.The dynamics of the slide category referred to the temporal and spatial changes in the terrain and are mostly related to the numerical simulation of the landslide.It comprised the initiation area, run-out distance, speed, and volume of the slide.The physical consequences category included all visible and audible consequences that a user could experience, for example, the destruction of infrastructure, oating objects, sounds, and vibrations.
The emergency activities category included the contents of an experience that focused on behaviors and actions that could be performed during the emergency, for example, an escaping user, sirens and emergency respondents, actions showing where to escape, and evacuation procedures.Post-emergency activities referred to all activities related to the period after the danger was gone.Events such as rehabilitation of the area, identifying the cause for future learning, and taking responsibility could be included in this category.The hazard information category referred to the dissemination of information and media communication about the event.The dynamics and physical consequences of the landslide received the highest priority.The post-emergency activities category got the least priority and was not used in the VR implementation.

VR Experience Planning: Brainstorming
The results achieved from the two a nity diagram sessions were discussed and analyzed for the realization of the VR scenario proposals.A brainstorming session was done with the VR expert (E7) and one geotechnical engineer (E5) to de ne the virtual experience.The data collected from the two a nity diagrams were analyzed and translated into operational VR scenarios considering the priorities of the categories and the technical capability of VR technologies.
The two a nity sessions led to the creation of two different VR scenarios.The rst scenario was hazard identi cation which focused on the period prior to a quick clay landslide, while the second was an emergency scenario which focused on the post-landslide period.
In the hazard identi cation scenario, the focus was on showing the triggering mechanisms that can lead to quick clay landslide.Out of the four categories listed in section 3. A partially collapsed riverbank with fallen trees and exposed rock was suggested to represent erosion directly.River channel morphology changes such as migrating of river coarse towards the riverbank adjacent to the toe of a given slope could be one indicator of dangerous erosion.In addition, river water level rise could represent erosion indirectly as a river ow increase exacerbates erosion.A pond created behind an over owed culvert was proposed to show a gradual increase in river ow.Construction activities were represented by a simple construction site located on the top of a slope with several trucks and excavators.
High realism VR environment provides a greater sense of presence and immersion in VR experiences (Newman et al. 2022).Referring to the realistic environment category mentioned in section 3.1, it was proposed that the VR environment and its components should re ect an actual environment that a user can relate to.Hence, an actual village covered with quick clay should be considered to represent the scenarios.A realistic digital terrain model (DTM), actual existing infrastructures and terrain features of the actual village were proposed.Furthermore, houses and other man-made features should be selected in a way that they re ect the way of life and culture of the people living in the selected village.
For the virtual environment to be realistic as suggested by Gamberini et al. (2021), environmental sounds related to ora and fauna of the village should be added.Furthermore, some weather elements were proposed to enhance the realism of the virtual environment.
From the discussions, the most relevant information required to increase the knowledge and understanding about quick clay are the spatial extent of the quick clay soil and its relationship with the triggering factors and infrastructures that may be affected when a quick clay happens, and the quick clay's behavior.Hazard maps convey information related the horizontal spatial extent quick clay soil and its relationship with the triggering factors whereas risk maps relate the horizontal spatial extent and its relationship with both the hazards and the possible consequences of a landslide event.Therefore, both maps were proposed to be included in the VR tool.
The vertical extent of the quickly can be obtained from borehole data of the ground investigation.
Borehole data accompanied by some laboratory test results can provide utmost the exact location of clay layer and at least the relative strength of the soil layers along the borehole.Boreholes and a corresponding total sounding chart were proposed to be included in the virtual environment.When a plenty of ground investigation boreholes are available for a given site, it is possible to reconstruct a threedimensional geological model of the site which can show the extent of quick clay layer and its relative location with the other soil layers.Such a model, if appropriately scaled, can be a valuable ingredient to the VR tool.Regarding the behavior of quick clay, there are several eld and laboratory video that demonstrate quick clay behavior when overloaded and/or after salt is added to it.
In the emergency scenario, the focus was on the dynamics and consequences of a quick clay landslide.Numerical simulation was suggested to represent the dynamics of the landslide.Houses, some of them destroyed and others moving intact with the mud, represented the physical consequences of the landslide.The destruction of the houses should be as real as possible.In addition, a 3D audio that could imitate the sound of mud ow was suggested to make the scene more realistic.
Both scenarios should be designed to be used both separately and continuously with chronologically ordered events.This made the whole experience modular for overall or partial use of the simulation, according to the different objectives and contexts.When both the scenarios are designed together, appropriate transition should be provided.
At the end of this session, it was stated that VR offers 360 degrees of freedom to navigate within the virtual environment.However, this brings a considerable risk that the user might miss the relevant elements of the virtual environment because the user's focus may be elsewhere.Integrating spatial audio and visual cues into the VR environment is a powerful way to guide the user and it can also smooth translation between stories within the virtual environment.Hence, audio, and visual cues such as arrows and billboards were suggested to guide the user.

Materials and Resources
To implement the QuickAware VR tool, Gjerdrum landslide, an actual quick clay landslide event was selected to be recreated.Gjerdrum landslide occurred in Ask, the administrative center of Gjerdrum municipality in Viken county, Norway (Penna and Solberg 2021).The landslide led to death of 11 people, evacuation of more than 1600 people and massive material destruction (Ryan et al. 2020).
The VR contents discussed in the brainstorming session were created using Unity version 2021.

Virtual Environment Representation
In this work, the basis for building the 3D virtual environment is the site's terrain.It was provided as a DTM.A custom script based on Unity C#(C-sharp) was used to import the DTM into Unity using the method mentioned in (Alene et al. 2022b).This allowed us to show the actual location and orientation of natural features of topography such as rivers and slopes.The roads and rivers were added carefully, so that they were as close as possible to their actual location.Building and trees were also added.Additional salient objects such as construction site, boreholes, objects representing erosion, etc., as shown in Fig. 3, were added into the terrain.The terrain was also customized with weather effects to enhance the realism of the virtual environment.The maps and video demonstration were represented by a twodimensional(2D) graphical interface.They were put on the walls of a "show room" (see Fig. 3) represent the fact that maps are 2D objects and videos are mostly shown on 2D screen.
In VR, the space and the objects inside become the story (Fraga 2017).The objects and space should be structured such that they could narrate a speci c story at a time to not overwhelm the user (Silva and Brandão 2021).To attain this, the structure was designed in a way that enables the user to interact freely, but also the user must perform certain tasks so that the story would advance.For this purpose, some objects serving as visual cues to guide the user to follow the correct story path were added.Audio narratives were also used as guidance to complement some of the visual cues (see Fig. 3).
On some occasions, the objects inside the space may not be adequate to tell the intended story.In such instances, they were complemented with audio narratives and texts.For instance, the nature and properties of quick clay soil were provided as an audio narrative when the user was introduced to the VR tool and additional information was also added as description to the erosion objects mentioned in the brainstorming session.
As mentioned in section 3.2, the emergency scenario does not necessarily occur just after the hazard identi cation scenario.Therefore, the virtual environments for both scenarios were customized with two different weather effects to present that both did not happen at the same day.A rainy and a dry cloudy weather effects were used for the emergency and hazard identi cation scenarios respectively.

Physics
To realistically simulate the dynamics of the landslide and some of its physical consequences, certain physics have been introduced in the virtual environment.We applied a realistic 3D Computational Fluid Dynamics (CFD) simulation to simulate the dynamics of the landslide and a Unity built-in physics engine for destruction of infra-structures.
A. Quick clay landslide simulation After complete remoulding, the quick clay is considered as a homogenous, viscous, and incompressible uid.The uid ow simulation is carried out using interFoam, a multiphase solver based on the CFD toolkit OpenFOAM® (OpenCFD Ltd 2020).InterFoam solves the 3D continuity and the Navier-Stokes (N-S) governing equations (Versteeg and Malalasekera 2007) for an incompressible uid.The governing equations are given as: where is the density, t is the time, is the velocity in i = x, y, z directions, p is the pressure, gravitational acceleration in i = x, y, z directions and is the effective viscosity.
The solver assumes that the landslide material and air are two incompressible phases, and the surface of the landslide material phase movement is captured with the volume-of-uid (VOF) method ( The effective viscosity, , in Eq. ( 2) is determined locally for each nite volume cell every time step since it varies spatially and temporally and can be computed from Eq. ( 4) as: 6 where is s a maximum viscosity value used for small shear rates when the effective viscosity approaches in nity.
First, the DTM was triangulated and translated into 3D volume by adding side and top walls.Next, the volume was spatially discretized by the OpenFOAM meshing utilities to obtain hexagonal nite-volume cells.The temporal domain was split into time steps and the equations were solved in a time-marching manner.
In this study, mathematical and rheological models were numerically built and implemented using the release of version 2012 of OpenFOAM.The immediate outputs of the simulation are the pressure, velocity, , and effective viscosity for each nite volume cell at each time step.These results were visualized and further processed in ParaView (Ahrens et al. 2005).In ParaView, a contour surface of resulted in a collection of coordinates on the interface of the quick clay mud and air.These coordinates and their interpolated velocities were exported into Unity using customized C# script.The details of the procedure is outlined in (Alene et al. 2022a) and (Alene et al. 2022b).The details of the numerical simulation and preprocessing of the simulation results were given in supplementary information, SI1(online resource).

B. Dynamic Objects Simulation
To enhance the realism of the physical consequences of the landslide, the destruction of houses was simulated.The houses were considered as dynamic objects or made up of dynamic objects.To have convincing physical behavior, the objects in the VR environment accelerate correctly and are affected by collisions, gravity, and other external forces.This was handled by Unity's built-in physics engine.We have two categories of houses; those demolished due to gravity because the ground supporting them slid out and those swept away by the mud.In the rst category, the houses were made up of materials of varying mass and shape such that they fell at different times simulating complete demolition.In the second category, we assumed the houses as monolithic structures and that they move in the direction of the mud.The Unity's built-in physics used two external forces derived from the numerical simulation of the quick clay landslide to simulate the sweeping away of houses.The two forces are buoyancy force and drag force.The buoyancy force used the ow height of the landslide simulation to determine the submerged height of the houses whereas the drag force used velocity from the landslide simulation.The buoyancy and drag forces are given in Eq. ( 7) and Eq. ( 8) respectively.Even though the actual shape of the houses is mostly a combination of rectangular and triangular prisms, we assumed a spherical shape for the Unity engine physics simulation.

8
Where is the density of mud, gravitational acceleration, the volume of the submerged part of the house, drag coe cient, average ow velocity of the mud in the vicinity of the house, and A surface area of the house.

Testing and Evaluation
In addition to the usual tests during the development, the usability of QuickAware was also carried out.
The tool was evaluated by conducting a small-scale survey at the Norwegian University of Science and Technology, Norway.16 participants (aged 18-34 including 12 male and 4 female) were invited to experience and test QuickAware.After the test, a questionnaire was given to each participant.
The questionnaire aimed to get feedback from participants about the effectiveness of the tool in creating awareness about quick clay landslides.The questions were organized into three main groups: (a) general information and computer (IT) background; (b) general questions about prior VR experience; (c) questions about quick clay; (d) speci c questions about this VR tool.The details of the questionnaire are provided in Appendix A. For the VR related questions, two Likert scales were used; one for evaluating the general experience of the participants by scaling the statements provided in Table 2 of the Appendix and the second for evaluating the speci c observations the participants made during their VR experience.In addition, the participants were asked open-ended questions about quick clay triggering factors and as suggestions for improvements.
Table 2 General statements about the general experience of QuickAware.

S1
Interacting with the 3D objects and environment helped me learn about quick clay S2 Seeing the landscape in three dimensions gives me a better appreciation of it than maps and charts ever could.

S3
The voice narratives in the VR experience were helpful in guiding me within the virtual environment.

S4
I feel like I am in a natural environment at all.

S5
The experience was intuitive; it took me few minutes to master the tool by myself S6 I liked that we could visit sites much more directly, without having to walk for kilometers.S7 I will use this and similar tools for the future.

S8
I felt motion sickness such as feeling dizzy or nauseous etc. during the use of this VR tool.

Implementation
We employed a laptop with an NVIDIA GeForce GTX 1070 graphics card and an Intel Core i7-8750H CPU @2.20GHZ, 16 GB RAM six-core processor to create the VR environment.The resulting VR environment has been set up to be used through Oculus Quest 2HMD (resolution:1,832 × 1,920 px per eye; refresh rate: 90Hz; Field of view: 100-degree est.; integrated speakers and microphone).
The tool started with an initial interface as shown in Fig. 4a.The interface was followed by the instruction scene (Fig. 4b).The instruction scene aimed at providing the introduction information about the tool.This scene was supposed to help users make them familiar with the tool.
The knowledge category that included the maps, videos, and some geotechnical investigation results were provided, as shown in Fig. 5. Figure 5a shows the risk map of the area and a user can learn more about risk maps by clicking on the description option.Figure 5b is a representation of a video demonstration about quick clay behavior in the eld and the laboratory.The video was obtained from (NVE 2020).
The ground investigation results were represented by Fig. 6.To the left, there is a cylindrical soil sample comprised of different soil layers.The extent of each layer was determined from the eld and laboratory investigation results in NVE (NVE 2021a).A user could interact with each layer at time by griping and the type of soil could be displayed in front of the user.To the right, there is a total sounding result that the most common ground investigation method in Norway.It shows the relative strength of the soil layer along the depth.Similarly, it was obtained from NVE(NVE 2021a).
Information related to the quick clay triggering mechanisms are presented in Fig. 6.In Fig. 6a, an over owed culvert indicates there has been water level raise in the river and hence river ow increase.
Figure 6b showed a gradual change in the river morphology, i.e., a river course shifting towards the slope.
Both Fig. 6a and Fig. 6b could be an indication for a gradual erosion of the toe of the river.Figure 6c depicted erosion and its subsequent consequence, such as, fallen trees.Figure 6d represented a construction site on a at terrain that is on the top of the slope.
Figure 7 showed the salient events after the occurrence of the quick clay landslide.In Fig. 7a one can see the remolded clay, also called the mud, and the unstable landslide scraps.Figure 7b showed buildings destroyed after the soil mass supporting them slid out.
QuickAware has the capability of engaging users that they could inspect some characteristics of quick clay slide.Figure 8a indicates a user measuring a slope and slope height of a quick clay abundant site to con rm susceptibility of against quick clay landslide.Figure 8b showed a user reading ow characteristics of the quick clay ow landslide, i.e., ow depth and ow velocity.A sample video about a user exploring the entire experience is provided in supplementary information, SI2(online resource).

Usability Evaluation
From the background information of the participants, 92% of them rate their computer and IT skills above average and 63% of them never used VR headset before.Figure 9 shows the box plot of the participants' response to the statements provided in Table 3 of Appendix A. For the rst seven statements(S1-S7), most of the participants held positive attitudes towards the general VR experience.62.5% of the participants showed positive response for the interaction capability (S1) scoring 4 and 5, and 37.5% scored 3 showing moderate (neutral) satisfaction.Perceptions from the responses to the second statement (S2) which indicates the 3D perception of the tool were predominantly positive.87.5% of the participants scored 4 and 5 and 12.5% scored 3. The raw data generated from the usability evaluation was provided in supplementary information, SI3 (online resource).
The effectiveness of voice narratives in tool was evaluated using the responses from statement, S3. 93.75% of the participants responded positively and 6.25% responded with a neutral position.Similarly, the realism of the virtual environment was measured by responses to statement, S4. 81.25% of the participants gave positive feedback, 12.5% gave neutral feedback and 6.25% provided negative feedback.
The responses to statement, S5, indicated that 81.25% respond positively that the tool is easy to use.12.5% and 6.25% provide neutral and negative feedback, respectively.87.5% provided positive perception that the tool is intuitive and easy to follow (statement S5).In addition, 81.25% of them gave positive feedback that they would use such tools in the future.
The last statement, S8, aimed at evaluating the short-term sickness resulted from the VR experience.As compared to the previous statements, results from the responses to S8 showed a relatively normal distribution of the score with 25% scoring 5, 12.5% scoring 4, 25% scoring 3, 18.8% scoring 2 and 18.8% scoring 1.None resulted in an interruption of the experience.We are aware of that there are more detailed, systematic, and validated ways of evaluating the feeling of sickness resulted from the usage of VR technologies (e.g., Chang et al. 2020).Our goal was to indicate that there were some adverse effects perceived by the users and to evaluate the possibility to improve the tool from our end.
Figure 10 shows the level of realism of the mud ow and building destructions visualizations in the VR experience.About 81.25% of them agreed that both the mud ow and the building destructions were realistic enough.62.5% responded "excellent" to the question "How much realistic were the destruction and sweeping away of the houses?", whereas 25% responded "excellent" to the question "How much realistic was the mud ow?".This indicated the destruction of the buildings were more realistic than the mud ow in terms of visualization.
The participants were also asked to rate their understanding of quick clay prior to this experience.31% rate their knowledge of quick clay as excellent, 25% as good and 31% as average showing a generally a positive response.Furthermore, they were asked whether they knew what could trigger quick clay landslide prior to this experiment.69% responded "yes".However, responses to a follow up open ended question by "If your answer is yes, what are the triggering factors "indicated that the responses were biased.This is because the questionnaire survey was conducted after the VR experience, and it was observed much of responses were taken from the experience even though they were asked their knowledge prior the experience.For this reason, we ignored the responses to the questions related to quick clay knowledge prior the experience.
The participants responded to an open-ended question of "What quick clay landslide triggering factors did you learn from this Virtual reality tool?" with the responses provided in Fig. 11.Erosion, construction, and rainfall are the most frequent responses.One important point to note here is that erosion and construction activities were the actual quick clay landslide triggering factors and they were included in the VR environment for the same purpose.However, as discussed in section 3.4.1 the rainfall was included to show both scenarios happened at different time frames.

Discussion
The main objective of QuickAware is to utilize the interactive VR technology to create awareness about quick clay landslide hazard.In Norway, much of the information related to quick clay landslide and other natural hazards is provided in the form of hazard and risk maps, geotechnical investigation, and risk assessment reports.Although such information is public and open-accessed, it is expert-oriented.
The tool brings the realistic environment, situations and realistic data that are already available to the experts into a VR experience so that lay people from diverse backgrounds can have equitable access to information and knowledge.It can also serve as an educational and engaging tool enabling users with average understanding to learn about quick clay via audio narratives and interactions.
As mentioned in section 2. QuickAware applies a VR technology to improve the awareness about quick clay landslide hazards by enabling users to get immersed within a virtually created landslide scene and get engaged with the objects inside the scene.QuickAware can be used as a teaching tool for students by serving as a virtual eld platform where it enables the students to measure slopes and slope heights (see Fig. 8) that are the main topographic characteristics that in uence quick clay landslide.It can also serve as a communication platform between experts and the public.Hazard and risk maps made by experts are not easily understandable for the public.People with limited knowledge of hazard and risk maps may perceive the risk in a certain landslide site when they get experienced it in VR.
Despite having a promising value in creating awareness of quick clay landslide hazards, this study has limitations.Generalizing results from only 16 questionnaires survey can be risky at best.However, these data at least can suggest that the tool can be helpful in raising awareness of quick clay landslides.The responses to the triggering factors, as shown in Fig. 11, showed that many participants could memorize and relate some of the contents of VR tool that were intentionally included as per the discussion in section 3.2 with quick clay.Regarding the responses to the questionnaire, one of the important outputs was that most participants would be glad to use such tool or similar ones in the future.81.25% of participants showed a positive response to this.Some participants suffered from sickness effects, but none stopped the experience.Evaluation of the usability of VR technologies in regard to the sickness effects needs thorough analysis, which is beyond the scope of this work.However, from a technical standpoint, the number of frames per second and navigation mode of a given VR application signi cantly impacts the feeling of sickness.VR applications with the number of frames per second higher than 60 can help reduce sickness effects (Zhang 2020).In QuickAware, it was observed that there are few moments where the frame goes down to around 30.In VR, there are two common ways of navigation within the virtual environment: continuous movement and teleporting.The continuous movement mode is more realistic than the teleporting one.However, a study by (Veličković and Milovanović 2021) indicated that teleporting way of navigation has less possibility of having sickness effects than continuous movement mode.In this VR experience test, we mostly used a continuous move as a navigation mode, but after the test we incorporated the teleporting alternative into the system so that a user can choose one of them at a time.In addition to that, it was suggested by (Vagner 2022) that repeated exposure to HMD-based VR reduces sickness effects.In fact, based on our personal experience, the more frequently the VR systems are used, the more comfortable users become with the VR experience.The fact that 63% of the participants never used VR headset before may contribute to exhibition of the signi cant number of the participants who felt sickness.
One important observation in this VR experience is that care must be taken when creating the VR contents.Rainy weather was used as a transition of the hazard identi cation and emergency scenarios to indicate that both scenarios happened at different times.On the other hand, a substantial number of the participants' response to quick clay landslide triggering factors is rainfall (see Fig. 11).In fact, rainfall is not an immediate triggering factor for quick clay landslide, even though it may exacerbate erosion and overload a slope.

Conclusion
This work has been focused on the development of QuickAware, a VR tool for quick clay landslide hazard awareness.The development of the tool started from collecting data that are useful for de ning its contents.Raw data concerning the contents of the VR tool was collected from experts from geotechnics, hydrology, psychology, and VR technology.The data collected was processed and its translation into VR educational experience conceptualized.
The planning of the VR experience led to the creation of two distinct scenarios: hazard identi cation scenario and emergency scenario.The rst scenario focused on the moments that occurred prior to a quick clay landslide, whereas the second scenario focused on the events after the quick clay started owing or sliding.
For both scenarios, the basis for the 3D virtual environment was the terrain.To demonstrate the tool, we used an actual quick clay zone in Ask, the administrative center of Gjerdrum municipality in Viken county, Norway.All the VR contents derived from the raw data collected from the experts were included to make the tool a full-edged learning platform.Using HMD and hand controllers, a user can get immersed and interact with the virtual environment and its contents.
Usability evaluation was conducted based on a limited number of participants.The analysis of the responses for the questionnaires has enabled us to shed light on how the tool's usefulness in raising awareness of quick clay landslide hazards.81.25% participants responded that they would be glad to use such tool or similar for the future.
The tool can be regarded as a groundbreaking tool to improve the equitable access of information and knowledge as it brings all expert speci c information about quick clay into a way anyone can understand.However, it still has limitations.Firstly, the terrain was textured manually in Unity, which resulted in some loss of realism.Terrain texture processed from photogrammetry would have provided more realism.Secondly, at this stage the VR tool is based on a single case study of Gjerdrum quick clay landslide.
Multiple case studies must be considered in the future so that users have a thorough understanding of quick clay landslide hazards, consequences, and triggering factors that can vary depending on topographic and geotechnical characteristics.Thirdly, the number of participants is limited, and the volunteers were mostly students.In the future, feedback from people from all levels of society should be included.
Answers from the participants posted on the whiteboard and being discussed during the rst a nity diagram session.
Location of the area which the virtual environment based on Geonorge (https://kartkatalog.geonorge.no).
Diagram representing the contents of the virtual environment.
The starting interface of QuickAware.(a) Initial interface; (b) the instruction scene.
Page 28/33 There have been several quick clay landslide incidents in Norway.Some well-known examples are listed below: Verdal, 1893: is the most severe and largest known quick clay slide in Norway.55 million m 3 owed out and ooded the city killing 116 people.Rissa, 1978: is the largest quick clay landslide in Norway in the 20th century.5-6 million m 3 of clay slid out and 1 person was killed.Kattmarka, 2009: 0.3-0.5 million m 3 clay owed taking 10 houses away (NTNU 2009).Byneset, 2012: had a dimension of 100m wide and 400m long.It forced 40 persons to evacuate(NVE 2012).Skjeggestad, 2015: led to the demolition of the Mofjellbekken bridge (NVE 2015).Sørum, 2016: had a dimension of 400m by 300m and killed three people(NGI 2016).Gjerdrum, 2020: mobilized 1.35 million m 3 clay and killed 10 people(Penna and Solberg 2021).

Figure 5 A
Figure 5

Figure 6 A
Figure 6

Figure 7 Post
Figure 7

Figure 10 Histogram
Figure 10 2.2 Virtual Reality Applications in education, safety training and hazard identi cation VR has been widely employed in hazard recognition and prevention, safety training, evacuation, search and rescue, and damage identi cation of hazards (Zhu and Li 2021).Several studies have been done on the application VR in re emergency training (Smith and Ericson 2009; Rüppel and Schatz 2011; Çakiroğlu and Gökoğlu 2019; Cao et al. 2019; Lovreglio et al. 2021), construction safety education and training (Perlman et al. 2014; Le et al. 2015; Zhang et al. 2019, 2022; Getuli et al. 2020; Xu and Zheng 2021), tunneling hazard identi cation (Liang et al. 2019; Isleyen et al. 2021), earthquake emergency training (Gong et al. 2015; Li et al. 2017; Liang et al. 2018; Feng et al. 2020b, a), ooding emergency training (Sermet and Demir 2019; Fujimi and Fujimura 2020) and landslides hazard identi cation (Alene et al. 2022b).The utilization of VR in geohazards is generally limited even though there are several applications for earthquake and a few for ooding.VR can play an important role in teaching process by enabling to present a state of knowledge, to teach practical skills according to previously acquired knowledge, and teach how to use acquired knowledge when faced challenges (Kamińska et al. 2019).Dinis et al. (2017) developed a VR tool to investigate building components such as beams, columns and walls in civil engineering discipline so that students could able to assemble or disassemble those components to understand more about the building system.Janiszewski et al. (2020), the authors demonstrated that a VR system developed for rock engineering, geology and mining education increases the active learning time of students by 50%.Román-Ibáñez et al. ( 2018) integrated immersive VR with robotic simulator to facilitate teaching of robotics.

Table 1
Description of experts involved in the co-design activities.
Ryan et al. 2020) indicate that a given quick clay landslide does not necessarily happen immediately after a triggering activity has occurred.For instance, an erosion that slowly started a decade ago can trigger a quick clay landslide today or a construction ll that was ongoing for a month can be a reason for today's quick clay landslide.On the other hand, a user should not use a VR headset for more than 30 minutes without a break (Meta Platforms Inc.).Therefore, dynamic representation of the triggering mechanisms in VR is inappropriate.
1, the actions category got the least priority and was omitted from the VR scenario.Most Norwegian post-quick clay landslide assessment reports (e.g., 3.11(Unity Technologies 2022).The virtual environment includes natural and man-made features in the vicinity of Ask and the run-out area that debris of the landslide followed.The 3D space of the virtual environment was based on a digital elevation model (DTM) obtained from the Høydedata website (https://hoydedata.no).The DTM is provided as an ESRI ASCII raster format(.asc)whichgives elevation of the topography on a regular two-dimensional grid.Numerical simulation results that represented the dynamics of landslides was fetched from OpenFOAM® (OpenCFD Ltd 2020) simulations carried by the authors and are discussed in the next sections.Hazard and risk maps were collected from the Norwegian water and energy directorate (NVE) web map one quick clay soil sample demonstration (NVE 2020) and another one showing demonstration of quick clay landslide causes and mitigation measures (NVE 2021b).
service (https://temakart.nve.no/tema/kvikkleire). Relevant ground investigation results were taken from a report by NVE (NVE 2021a).A geological model reconstructed in Leapfrog software (SEEQUENT 2022) based on the ground investigation results was used as a 3D geological model to represent the soil layering.Information related to river morphology and erosion was obtained from the Gjerdrum landslide report (Ryan et al. 2020).Four videos; two showing historic quick clay landslides (NGI 2011; ViralHog 2020), (Ryan et al. 2022eral agreement in the literature, whether by practice or qualitative evaluations, that VR technologies are useful in education, safety training, and hazard identi cation.Furthermore, various companies are incorporating VR training for hazard identi cation and accident awareness (e.g.Morgan 2021; Vagner 2022).It has been reported in(Ryan et al. 2022) that much of the quick clay landslides happening recently were triggered by man-made activities.Hence, creating awareness about the nature, cause, and possible consequences of quick clay landslide among the public can reduce the risks associated with the landslide.In addition, in Norway it was noted in most of the postlandslide reports including in(Ryan et al. 2022) that an improved knowledge base of quick clay through research, systematic learning from events, innovation and implementation of new technology is required to manage quick clay landslide.