Interactive Landslide Simulator: A Tool for Landslide Risk Assessment and Communication

Abstract

Understanding landslide risks is important for people living in hilly areas in India. A promising way of communicating landslide risks is via simulation tools, where these tools integrate both human factors (e.g., public investments to mitigate landslides) and environmental factors (e.g., spatial geology and rainfall). In this paper, we develop an interactive simulation model on landslide risks and use it to design a web-based Interactive Landslide Simulator (ILS) microworld. The ILS microworld is based on the assumption that landslides occur due to both environmental factors (spatial geology and rainfall) as well as human factors (lack of monetary investments to mitigate landslides). We run a lab-based experiment involving human participants performing in ILS and we show that the ILS performance helps improve public understanding of landslide risks. Overall, we propose ILS to be an effective tool for doing what-if analyses by policymakers and for educating public about landslide risks.

Full text

Interactive Landslide Simulator: A Tool
for Landslide Risk Assessment
and Communication
Pratik Chaturvedi, Akshit Arora and Varun Dutt
Abstract Understanding landslide risks is important for people living in hilly areas
in India. A promising way of communicating landslide risks is via simulation tools,
where these tools integrate both human factors (e.g., public investments to mitigate
landslides) and environmental factors (e.g., spatial geology and rainfall). In this
paper, we develop an interactive simulation model on landslide risks and use it to
design a web-based Interactive Landslide Simulator (ILS) microworld. The ILS
microworld is based on the assumption that landslides occur due to both environ-
mental factors (spatial geology and rainfall) as well as human factors (lack of
monetary investments to mitigate landslides). We run a lab-based experiment
involving human participants performing in ILS and we show that the ILS per-
formance helps improve public understanding of landslide risks. Overall, we pro-
pose ILS to be an effective tool for doing what-if analyses by policymakers and for
educating public about landslide risks.
Keywords Early warning systems Interactive landslide simulator (ILS) 
Landslide risk communication Feedback Learning
P. Chaturvedi (&)
Defence Terrain Research Laboratory (DTRL), Metcalfe House,
Delhi 110054, India
e-mail: prateek@dtrl.drdo.in
P. Chaturvedi V. Dutt
Applied Cognitive Science Laboratory, Indian Institute of Technology (IIT) Mandi,
Mandi 175001, H.P, India
e-mail: varun@iitmandi.ac.in
A. Arora
Thapar University, Patiala 147004, Punjab, India
e-mail: akshit.arora1995@gmail.com
©Springer International Publishing Switzerland 2017
V.G. Duffy (ed.), Advances in Applied Digital Human Modeling
and Simulation, Advances in Intelligent Systems and Computing 481,
DOI 10.1007/978-3-319-41627-4_21
231

1 Introduction
Over the past few decades, catastrophic and disastrous effects of landslides have
caused extensive damage to life, property, and public utility services world over.
Thus, ensuring effective Early Warning Systems (EWSs) for landslides is essential
for the survivability of people in case of occurrence of a disastrous event. To be
effective, EWSs need to have not only a sound scientific and technical basis, but
also a strong focus on people, who are actually exposed to risk. Unfortunately, such
risk communication systems only address part of the existing challenge; the other
important part being related to the properties of human perceptual-cognitive factors
(cooperation; attitude; and effects of economic and educational background) [1–3].
Moreover, recent surveys in developing countries (like India) show only mediocre
knowledge and awareness about causes and consequences of landslide disasters
among the general public [4–6]. For example, a recent survey conducted in Mandi
town of Himachal Pradesh, India, showed a big gap between experts and general
public on understanding of hazard zonation maps and probability of landslides [4].
The existence of this gap is problematic because zonation maps, developed by
landslide experts, are currently the common medium to communicate the suscep-
tibility of a region to landslides.
An important aspect of EWSs is related to the development, evaluation, and
improvement of risk communication, which helps in transferring risk related
knowledge (like causes, consequences and what to do in case a disaster event takes
place) and warnings in a manner easily understandable to the local community.
A promising way of improving existing risk communication among EWSs is via
simulation tools (also called microworlds), which are able to integrate human factors
in landslide risk mitigation in addition to physical factors. Such simulation tools, and
the models they are built upon, could help risk managers since personal experience
and the visibility of processes are the two main influencing factors for improving
people’s mental models about natural disasters. Promising recent research has shown
that regular feedback from a system likely provides an effective tool for people to
improve their understanding about the system (Dutt and Gonzalez 2011, 2012). For
example, research has documented some benefits of repeated feedback in computer-
based microworlds in reducing people’s misconceptions about Earth’s climate [7].
Dutt and Gonzalez (2012) developed Dynamic Climate Change Simulator (DCCS)
microworld and used it as an intervention to help participants understand basic
characteristics of the climate system [8]. DCCS helped provide feedback to people
about their decisions and enabled them to reduce their misconceptions compared to
no DCCS intervention. As DCCS-like tools seem to be effective in improving
people’s understanding on problems, there is a need to develop simulation models
that are able to integrate human factors in landslide risk mitigation in addition to
physical factors. Such simulation models could be helpful for the risk managers since
personal experience and the visibility of processes are the two main influencing
factors explaining the content of people’smentalmodels[9,10].
232 P. Chaturvedi et al.

Furthermore, affect or emotional response to stimuli is seen to influence risk
perception and decision making [11,12]. For example, Finucane et al. [12] have
provided the “affect heuristic,”where this heuristic allows people to make decisions
and solve problems quickly and efficiently, in which current emotions of fear,
pleasure, and surprise influences decisions [12]. According to Finucane et al. [12],
the orientation of one’s feelings (negative or positive) could be an effective tool for
risk communication [12].
In the present work, an interactive simulation model of landslide is developed for
understanding the influence of monetary contributions for landslide risk mitigation.
Furthermore, the interactive simulation model is used to design a web-based
Interactive Landslide Simulator (ILS) too. The ILS tool is based on the assumption
that landslides occur due to the presence of both physical factors (spatial geology
and rainfall) as well as human factors (monetary investments made for landslide
risk mitigation). Thus, even in the presence of physical factors (which are outside of
one’s control), the landslide risk could be reduced by increasing community
investments towards landslide mitigation. Beyond considering human factors in the
landslide problem, the ILS also models the damages due to the occurrence of
landslide events in terms of fatality, injury, and loss of property. It considers how
such damages might impact one’s daily income as well as property wealth. In
summary, the ILS tool allows participants to make decisions on the landslide risk
mitigation, observe the consequences of their decisions (via real-time feedback),
and enable participants to try new decisions.
In this paper, we highlight the use of the interactive landslide model as well as
the ILS tool in educating the general public about landslide risks. Specifically, we
use affect heuristic in the ILS by creating affect-rich feedback to enable people to
perceive risks and benefits for investments made against landslides. This will also
enable them to develop a deeper causal understanding about landslide disasters and
their consequences. Our hypothesis is that monetary contributions against land-
slides (which is an indicator of improved understanding) will be larger when
affective feedback about monetary losses is high compared to low. Based on results
of a lab-based experiment involving human participants, we propose a number of
benefits of the ILS tool for educating people and for policymaking in terms of
generating “what-if”analyses. As part of our outreach activities, we plan to pop-
ularize the use of the ILS tool among students in K-12 schools and colleges in
mountain areas in India that are prone to landslide risks.
2 Interactive Simulation Model of Landslides
2.1 Interactive Landslide Simulator (ILS) Model
The ILS model focuses on calculation of total probability of landslides (due to
natural factors and due to anthropogenic factors, i.e., investments made by people
Interactive Landslide Simulator: A Tool for Landslide …233

against landslides). The model is also capable of simulating types of damages
caused by landslides and their effects on people’s earnings.
Figure 1shows the model of ILS proposed in the present research work. In this
model, the probability of landslide is calculated as a weighted sum of probability of
landslide due to environmental factors and probability of landslide due to people’s
investments. Probability of landslide due to environmental (natural) factors is a
combination of probability of landslide due to rainfall and probability of landslide
due to slope and soil properties. Model also simulated the losses caused due to
occurrence of landslide events.
The Calculation of Total Probability of Landslides.
Total probability of landslide = W * P(I) + 1 - WðÞ* P(E)ðÞð1Þ
where W is the weight factor, which is between [0, 1]. In the model, W have been
assigned a value = 0.8, which indicates that investments against landslides will
cause the system to respond rapidly and reduce the probability of landslide.
Fig. 1 Probabilistic model of the interactive landslide simulator microworld
234 P. Chaturvedi et al.

The total probability formula involves calculation of two probabilities, P(I) and
P(E), which is described below:
Probability of Landslide Due to Investment P(I). The calculation used here is
based on expected payoff equation used in Hasson (2009) [13], i.e.,
PðIÞ¼1MPn
i¼1xi
nBð2Þ
where,
B Budget available towards addressing landslide (if a person earns a daily
income or salary, then B is the same as this daily income or salary).
n Number of time periods (days). In the default formulation of the game, n = 60
simulated days, i.e., the game is played for 60 simulated days.
x
i
Investments made by a person at each day i to mitigate landslides; x
i
≤B,
M Return to Mitigation, which captures the lower bound probability of P(I)
when P
n
i¼1
xi¼nB, i.e., people invest their entire daily income in mitigating
landslides.
P(I) Probability of landslide after an investment is made.
Probability of Landslide Due to Natural Factors P(E). Natural factors include
rainfall, soil type, slope profile, etc. These can be categorized into two parts:
•Probability of landslide due to rainfall (P(T))
•Probability of landslide due to soil type, slope profile etc. [spatial probability,
P(S)]
The approach used to calculate both of them is based on a research paper [14].
Equation used for calculation of probability of landslide due to rainfall (P(T)):
z¼3:817 + DR * 0:077 + 3 DCR * 0:058 + 30 DAR * 0:009 ð3Þ
fðzÞ=1
1+ezð4Þ
z: 1to þ1;P:0to1
The logistic regression retains the daily (DR), 3-day cumulative (3 DCR) and
30-day antecedent rainfall (30 DAR) as significant predictors influencing slope
failure. P(T) = f(z), that is the temporal probability of landslide. The rainfall data
was collected as raw data from NASA’s TRMM project, from January 1, 2004 to
April 30, 2013.
Now,
P(E) = P(T) * P(S) = f (z) * P(S) ð5Þ
Interactive Landslide Simulator: A Tool for Landslide …235

Damage Modeling. The damage caused can be classified into 3 categories:
(a) Property Loss
(b) Injury
(c) Fatality
All 3 of them have different kinds of effects on the player’s wealth and income in
the simulator. The data used for calculating probabilities of the above damage has
been obtained from Parkash [15]. The stochastic nature of landslide occurrence and
damages caused by it have thus also been considered. The exact assumptions about
damages are detailed ahead in this manuscript.
2.2 Interactive Landslide Simulator (ILS) Microworld
Computer-based decision-making tasks have spread across disciplines and different
levels of education [16]. Furthermore, these decision-making tasks have been long
used in the study of dynamic decision making behaviour (also called Microworld,
see Gonzalez et al. [17]), and many more specialized tasks have been created to
provide decision makers with practice and training in organizational system’s
control; also called Management Flight Simulators [18,19].
ILS microworld is a computer-based task, where a decision maker’sgoalisto
maximize one’s economic level. The economic level (defined by wealth due to
income and property in ILS microworld) is influenced by exogenous environment
circumstances (spatial and temporal conditions) and the past decisions made by
humans. The economic level may decrease (by damages caused due to landslides,
like injury, death or property damage) or increase (due to daily income and property
wealth). However, the exact functional form governing these increases or decreases
was unknown to decision makers. Decision makers could only observe the values
that occurred in the previous time period. The level of wealth at time t depends
upon the previous time period t −1, a characteristic of dynamic systems called
interdependency [20]. Also inherent in dynamic systems are feedback loops, where
one observes the effect that one variable has on itself and others. Feedback loops
can be positive or negative, “self-reinforcing”or “self-correcting”[21]. Both types
of loops are present in ILS because decision makers make repeated investments so
as to increase or decrease their economic level.
Figure 2represents graphical user interface of ILS, which requires a decision
maker controlling her economic level and keep it up as much as possible. The
economic level is represented graphically as curve of ‘Property wealth’and ‘Total
income not invested in landslides’versus number of times investment decision has
been made (since the decision made is one per day the graph is plotted against
number of days passed in the simulator). The plot of ‘Total probability of landslide’
236 P. Chaturvedi et al.

versus number of days describes the cumulative effect of this variable on probability
of landslide. The ‘Game Parameters’table on the right hand side describes specific
values like daily income, property wealth, probability of landslide, and damages
due to landslide.
A decision-maker must enter the investment input in the text field specified on
top left of the screen. The investment can only be made between zero (minimum) to
the player’s current daily income (maximum). Once investment is made, the
decision-maker can observe changes in the daily income, property wealth, and
damages caused due to landslide [loss of daily income (due to death and injury) and
loss of property wealth].
After a decision-maker enters the investment decision and clicks on the ‘Invest’
button, the system provides feedback on whether a landslide occurred or not. If a
landslide occurs, then the system decides what kind of damage the landslide has
caused and the resulting economic level is shown as a loss via a negative feedback
screen (see Fig. 3). If landslide did not occur, however, a positive feedback screen
is shown to the decision maker (see Fig. 4). The user can get back to investment
decision screen by clicking on ‘Return To Game’button.
Fig. 2 ILS microworld graphical user interface [game] (source http://pratik.acslab.org)
Interactive Landslide Simulator: A Tool for Landslide …237

Returning to game causes the player to come back to the main graphical user
interface of ILS (see Fig. 2). Once a player has played multiple days in ILS (where
the end-point is not known), the interface shows the amount of income and property
wealth left at the end of the game. Although ILS is made to capture the dynamics of
landslides, the tool can actually be deployed for other natural calamities so long as
the geological data related to those calamities is available. Lastly, we have setup
ILS as a web-application; therefore, it is accessible anywhere in the world at any
time and on any web-browser compatible computing device.
Fig. 3 ILS microworld’s negative feedback screen where a landslide has occurred (source http://
pratik.acslab.org)
Fig. 4 ILS microworld’s positive feedback screen where a landslide did not occur (source http://
pratik.acslab.org)
238 P. Chaturvedi et al.

3 ILS Experiment: Testing Affective Feedback in ILS
In order to showcase the effectiveness of the ILS tool, we performed a lab-based
experiment where we used ILS with human participants. In this experiment, we
manipulated the feedback, i.e., the effect of landslides on a person’s income and
property wealth using two different conditions: high-affect condition (i.e., high
probability of death, injury, and property due to a landslide) and low-affect con-
dition (low probability of death, injury, and property due to a landslide). The
expectation was that participants will invest more and improve their understanding
about landslides in the high-affect condition compared to the low-affect condition.
These conditions and results are explained in greater detail below.
3.1 Methods
Experimental Design. Participants were randomly assigned to one of the two
between-subjects conditions: high-affect and low-affect. In both conditions, par-
ticipants were given daily income and were as asked to make daily investment
decisions. In high-affect condition, the probability of property damage, fatality and
injury were set as 10, 3, and 30 %, respectively. In low-affect condition, the
probability of property damage, fatality and injury were 3, 1, and 10 %, respec-
tively. The goal was to maximize the net wealth (coming from property wealth and
daily income combined) over multiple rounds of ILS (where the end-point was not
known to participants). The nature of functional forms used in ILS were unknown
to participants, and participants simply observed the values of the probability of
human factors and natural factors and all the damages occurring in an event of a
landslide.
The amount of damage (in terms of daily income and property wealth) that
occurs in an event of fatality, injury and property damage was kept constant in both
the affect conditions. The property wealth decreased to ½of its value every time
property damage occurred in an event of a landslide. The daily income was reduced
by 10 % of its latest value in case of injury and 20 % of its latest value in case of
fatality loss. The initial property wealth was fixed to INR 2 million, which is the
expected property wealth in Mandi district of Himachal Pradesh. The initial daily
income of the person was kept 292 INR (taking into account the GDP and
per-capita income of Himachal Pradesh, India where the study was carried out). The
time duration of the simulation was 30 days (this duration was not known to
participants). Weight of human factors in probability of landslide (W) was fixed
to 0.8 and that of natural factors (1 −W) was fixed to 0.2. The W value was known
to participants on the graphical user interface.
Interactive Landslide Simulator: A Tool for Landslide …239

We used decision maker’s average investment ratio as a dependent variable for
the purpose of data analysis. The average investment ratio was defined as the ratio
of investment made to total investment possible averaged across all participants and
days. On account of Affect heuristic [12], we expected the average investment ratio
to be greater in the high-affect condition compared to in the low-affect condition.
Participants. Forty-three participants at Indian Institute of Technology Mandi
from diverse fields of study participated in the experiment. There were 20 partic-
ipants in high-affect condition and 23 participants in low-affect condition to yield a
medium to large effect size (= 0.5) in our results (for Alpha = 0.05 and a
Power = 0.80). All participants were students from Science, Technology,
Engineering, and Mathematics (STEM) backgrounds and their ages ranged in
between 21 and 28 years (Mean = 23.54; Standard Deviation = 4.08).
Twenty-eight participants were Master’s students, 5 were Ph.D. students, and 10
were B. Tech. students. When asked about their previous knowledge about land-
slides, 20 participants mentioned having a basic understanding, 16 having little
understanding, 5 being knowledgeable, and 3 having no idea. All participants
received a base payment of INR 50 and an additional bonus according to their
performance in the task.
Procedure. Participants were recruited via an online advertisement circulated
via an email at Indian Institute of Technology Mandi. Experimental sessions about
30 min long per participant. Participants were randomly assigned to one of the two
conditions, and they were given instructions on the computer before entering the
ILS microworld. Participants were encouraged to ask questions after reading
instructions. Participants were not given any information concerning the nature of
environment or conditions in the microworld. They were told that their goal was to
maximize their final income and they were then asked to play ILS for 30 days.
3.2 Results
Data were analyzed for all participants in terms of their average investment ratio in
both the high-affect as well as low-affect condition. The result was as per our
expectation: Average investment ratios were significantly higher in high-affect
condition compared to those in the low-affect condition (see Fig. 5).
As shown in Fig. 5, participants had lesser investment ratios in low-affect
condition(M = 0.38, SE = 0.05) compared to those in high-affect condition
(M = 0.67, SE = 0.045) [t(41) = −4.1, p< 0.05, r= 0.54]. Thus, our hypothesis
related to affect heuristic was satisfied with these results from ILS.
240 P. Chaturvedi et al.

4 Discussion and Conclusions
One way of improving existing risk communication practices for landslides is by
training people about these risks via simulation tools. That is because personal
experience and the visibility of processes are the two main factors for improving
people’s understanding and seriousness about natural disasters. Interactive
Landslide Simulator (ILS) is an interactive simulation model, which could be used
by policymakers to do what-if analyses. ILS can also be used as an educational tool
for the general public to increase their understanding and awareness about
landslides.
In order to showcase the performance of ILS in improving public’s perception
towards landslide risk, we conducted an experiment involving 43 students of dif-
ferent educational backgrounds and made them invest against landslides for
30 days in different affective conditions. As expected, the result from the experi-
ment suggested that participants making investment decisions in high-affect con-
dition invested much higher than the participants doing the same task in low-affect
condition. This result can be explained by previous lab-based research on use of
repeated feedback or experience (Cronin et al. 2009; Dutt and Gonzalez 2011) and
affect heuristic (Fischoff 2001; Finucane et al. [12]. Affect heuristic allows people to
make decisions and solve problems quickly and efficiently, in which current
emotions of fear, pleasure, and surprise influences decisions. As the emotional
feedback is higher in high-affect condition, participants have made higher monetary
contributions in this case. Thus, ILS has exhibited success in terms of improving
public’s seriousness and awareness towards landslide risk. In future, various other
system-response parameters (e.g. w or M), feedback (e.g. numbers, text messages
and images for damage) will be varied to study their effect on public’s decision
making. Here, we would like to evaluate affect and its ability to increase public
contributions in the face of other system-response parameters.
Other uses of ILS include packaged education material for classroom and
workshops that accounts for factors like feedback, affect, and social norms. We
believe that such material, when tailored to specific individuals, will help improve
decision making of individuals against landslides. ILS tool can be used for
Fig. 5 Average investment
ratio in low- and high-affect
conditions
Interactive Landslide Simulator: A Tool for Landslide …241

communicating landslide risk in other landslide prone states by customizing the
spatial probability (based on geology, soil properties etc.) and temporal probability
(based on rainfall) of landslides in such areas. In future, we also plan to use ILS to
understand the effects of social norms on people’s investment decisions towards
mitigation of landslide risk.
Acknowledgments This research was partially supported by Thapar University, Patiala and
Indian Institute of Technology, Mandi, India. The authors thank Akanksha Jain and Sushmita
Negi, Centre for Converging Technologies, University of Rajasthan for their contribution in
collection of human data.
References
1. Basher, R.: Global early warning systems for natural hazards: systematic and people-centred.
Philos. Trans. R. Soc. London A Math. Phys. Eng. Sci. 364(1845), 2167–2182 (2006)
2. Meissen, U., Voisard, A.: Increasing the effectiveness of early warning via context-aware
alerting. In: Proceedings of the 5th International Conference, on Information Systems for
Crisis Response and Management (ISCRAM), pp. 431–440 (2008)
3. Villagran de Leon, J.C., Pruessner, I., Breedlove, H.: Alert and Warning Frameworks in the
Context of Early Warning Systems (2013)
4. Chaturvedi, P., Dutt V.: Evaluating the public perceptions of landslide risks in the Himalayan
Mandi town. Accepted for Presentation in the 2015 Human Factor & Ergonomics Society
(HFES) Annual Meeting, L.A (2015)
5. Oven, K.: Landscape, livelihoods and risk: community vulnerability to landslides in Nepal.
Doctoral Dissertation, Durham University (2009)
6. Wanasolo, I.: Assessing and Mapping People’s Perceptions of Vulnerability to Landslides in
Bududa, Uganda (2012)
7. Dutt, V., Gonzalez, C.: Why do we want to delay actions on climate change? Effects of
probability and timing of climate consequences. J. Behav. Decis. Mak. 25(2), 154–164 (2012)
8. Dutt, V., Gonzalez, C.: Decisions from experience reduce misconceptions about climate
change. J. Environ. Psychol. 32(1), 19–29 (2012). doi:10.1016/j.jenvp.2011.10.003
9. Knutti, R., Joos, F., Müller, S.A., Plattner, G.K., Stocker, T.F.: Probabilistic climate change
projections for CO
2
stabilization profiles. Geophys. Res. Lett. 32(20) (2005)
10. Wagner, K.: Mental models of flash floods and landslides. Risk Anal. 27(3), 671–682 (2007)
11. Baumeister, R.F., Vohs, K.D., Tice, D.M.: The strength model of self-control. Curr. Dir.
Psychol. Sci. 16(6), 351–355 (2007)
12. Finucane, M.L., Alhakami, A., Slovic, P., Johnson, S.M.: The affect heuristic in judgments of
risks and benefits. J. Behav. Decis. Mak. 13(1), 1–17 (2000)
13. Hasson, R., Löfgren, Å., Visser, M.: Climate Change Disaster Management: Mitigation and
Adaptation in a Public Goods Framework, No. 178 (2010)
14. Geosciences Group: Experimental Landslide Early Warning System for Rainfall Triggered
Landslides along Rishikesh-Badrinath, Rishikesh-Uttarkashi-Gaumukh, Chamoli-Okhimath,
Rudraprayag-Kedarnath and Pithoragarh-Malpa route corridors, Uttarakhand: Approach
document. http://bhuvan-noeda.nrsc.gov.in/disaster/disaster/tools/landslide/doc/landslide_
warning.pdf (2015). Accessed 10 Mar 2016
15. Parkash, S.: Historical records of socio-economically significant landslides in India. J. South
Asia Disaster Stud. 4(2), 177–204 (2011)
16. Foss, B.A., Eikaas, T.I.: Game play in engineering education: concept and experimental
results. Int. J. Eng. Educ. 22(5), 1043–1052 (2006)
242 P. Chaturvedi et al.

17. Gonzalez, C., Vanyukov, P., Martin, M.K.: The use of microworlds to study dynamic decision
making. Comput. Hum. Behav. 21, 273–286 (2005)
18. Paich, M., Sterman, J.D.: Boom, bust, and failures to learn in experimental markets. Manage.
Sci. 39(12), 1439–1458 (1993)
19. Sterman, J.D.: Teaching takes off, flight simulators for management education: ‘‘ The Beer
Game’’.http://web.mit.edu/jsterman/www/SDG/beergame.html (2011). Accessed 18 Jan 2011
20. Edwards, W.: Dynamic decision theory and probabilistic information processing. Hum.
Factors 4,59–73 (1962)
21. Sterman, J.D.: Business dynamics: systems thinking and modeling for a complex world.
McGraw Hill, Cambridge (2000)
Interactive Landslide Simulator: A Tool for Landslide …243