Content uploaded by Nan Zhang
Author content
All content in this area was uploaded by Nan Zhang on Dec 12, 2017
Content may be subject to copyright.
Comprehensive analysis of information dissemination in
disasters
N. Zhang1, H. Huang1,*, Boni Su1
1Institute of Public Safety Research, Department of Engineering Physics, Tsinghua University, Beijing,
China
*Corresponding e-mail: hhong@mail.tsinghua.edu.cn
Abstract:
China is a country that experiences a large number of disasters. The number of deaths caused by
large-scale disasters and accidents in past 10 years is around 900,000. More than 92.8 percent of these
deaths could be avoided if there were an effective pre-warning system deployed. Knowledge of the
information dissemination characteristics of different information media taking into consideration
governmental assistance (information published by a government) in disasters in urban areas, plays a
critical role in increasing response time and reducing the number of deaths and economic losses. In this
paper we have developed a comprehensive information dissemination model to optimize efficiency of
pre-warning mechanics. This model also can be used for disseminating information for evacuees
making real-time evacuation plans. We analyzed every single information dissemination models for
pre-warning in disasters by considering 14 media: short message service (SMS), phone, television,
radio, news portals, Wechat, microblogs, email, newspapers, loudspeaker vehicles, loudspeakers, oral
communication, and passive information acquisition via visual and auditory senses. Since
governmental assistance is very useful in a disaster, we calculated the sensitivity of governmental
assistance ratio. The results provide useful references for information dissemination during disasters in
urban areas.
Key words:
Information dissemination; Pre-warning; Natural disaster; Technical accident; Media
1 Introduction
There have been numerous man-made and natural disasters in recent years, resulting in great loss
of life as well as economic loss. For these reasons it is important to give public security greater
attention. When Hurricane Andrew hit Miami on August 24, 1992, 1.4 million residents lost their
power and more than 180,000 people were left homeless [1]. The terrorist attack in New York took
about 3,000 lives on September 11, 2001. On March 11, 2011, more than 20,000 residents died or
disappeared in a 9.0-magnitude earthquake in Japan [2]. According to the data from the international
disaster database of the Centre for Research on the Epidemiology of Disaster (CRED), large-scale
natural disasters and technical accidents caused 896,150 deaths in the past 10 years (2005-2014). In
addition, Fig. 1 shows that economic losses resulting from large-scale natural disasters and technical
accidents have increased sharply since 1965, and that total economic loss for the 4 years from 2011 to
2014 was over 1 trillion dollars. China has unfortunately suffered inordinately from serious natural
disasters, even given the frequent occurrences of different types of disaster and accidents globally.
Since 1900, the death toll and economic loss caused by natural disasters and technical accidents in
China was 16.4 percent and 39.2 percent of the global death and losses, respectively.
Fig. 1. Global and China’s disaster cost trend for the past 50 years.
Fig. 2 shows the statistical data for the frequency and number of deaths caused by the 1,686
serious natural disasters and technical accidents that have occurred in China since 1900. The first five
disaster types resulting in high death tolls are floods (51.7%), droughts (27.5%), epidemics (12.2%),
earthquakes (6.9%), and storms (1.4%). The percentage of deaths caused by other serious disasters
including industrial accidents, landslides, wildfires, extreme temperatures, traffic accidents and
miscellaneous accidents is less than 1%. In these emergencies, a great number of deaths can be
attributed to the lack of, or an undeveloped disaster pre-warning system [3] and a non-existent or poor
mechanism for information dissemination [4]. In disasters including floods, storms, wildfires, droughts,
extreme temperatures, landslides, epidemics are easy to monitor. Some of earthquakes and industrial
accidents can also be monitored, but early forecast is usually useless for the settler in epicenter. For
these disasters, pre-warning information can be spread. Considering, in retrospect, the 1,686 serious
disasters in China since 1900, 39.3 percent of disasters and 92.8 percent of deaths resulting from
natural disasters and technical accidents could have been prevented by improving the publication and
dissemination of pre-warning information.
Fig. 2. Statistical data for deaths caused by natural disasters and technical accidents in China.
Early forecasting of the aftershock activity of the earthquake of M9.0 in japan, 2011 played a
critical role in saving lives [5]. In India in 2004, facing the threat of a serious tsunami, thousands of
lives could have been saved if there had been an effective pre-warning information dissemination [6].
Therefore, in a disaster, an effective information dissemination mechanism is very important. In 2005,
an effective information dissemination mechanism saved a lot of lives and property from hurricane
Rita.
At present, there is a great deal of research focusing on information dissemination before and
during disasters. This information is usually spread via microblogs, short message service (SMS),
phone, email, news portals and television [7-9]. However, there are still some omissions in current
research.
(1) The spread of pre-warning information about disasters is closely related to dynamic population
distribution [10]. Based on an analysis of emergency responses, urban planning and multi-agent, it
has been shown that knowledge of the population is very helpful when preparing for information
dissemination during disasters [11].
(2) Current pre-warning information dissemination is usually based on electronic media. A
disadvantage of this method is that the time when people receive the information depends on when
the information was published. This may reduce the chance of people getting early pre-warning
information on a large scale [12].
(3) There are many studies of information dissemination during a disaster that focus only on the
different social media. But social media requires a good information network, and it will fail when
the network is paralyzed. For example, in the Wenchuan Earthquake in China in 2011, information
could not be spread because of destroyed base stations, and almost 30,000 mobile phones did not
work [13]. Therefore, we should pay more attention to non-network information dissemination
mechanisms, such as oral communication, loudspeaker vehicles, and loudspeakers [14, 15].
(4) Current research usually focuses on single media such as microblogs, SMS, and phones to spread
information about a disaster. However, different people have different preferences for specific
social media, so comprehensive ways that combine several media can easily provide effective
information dissemination during disasters.
(5) From the data, in the majority of disasters, information coverage is not more than 50 percent. Low
information coverage leads to fewer people evacuating [16].
(6) In many situations, the government and disaster-related agencies are not the first places to get
information about a disaster. In some other situations such as in a building, alerts [17],
loudspeakers [18] and inner telecoms [19] cannot work. For these cases interpersonal information
dissemination should be studied.
(7) Current research on information dissemination during disasters focuses mainly on disseminating
pre-warning information before a disaster, but ignores effective information dissemination during
the disaster. Spreading information in real-time during a disaster to assist pedestrian evacuation,
for example, can make a big contribution toward reducing the number of deaths and economic
loss.
In this paper, we establish an information dissemination model under disasters overcoming all
shortcomings listed above. The new multi-media information dissemination model considered dynamic
population flow, information network and building alert paralysis, and effective pedestrian evacuation
to reduce information acquirement time and to improve information coverage and efficiency of
information dissemination.
2 Research plan
Many media can be used during a disaster to provide information. In this paper, we discuss 14
methods for spreading information: television; radio; news portals; newspapers; mobile phones;
Wechat (the most popular communication software in China); microblogs; email; SMS; oral
communication; loudspeaker vehicles; loudspeakers; and passive information acquired by personal
visual and auditory sensing. The last method includes for example a person hearing noises made by
others being evacuated, resulting in their acquiring some evacuation information.
Fig. 3 is the schedule for information dissemination and personal response during a disaster. Of all
the listed information media, television, radio, news portals, newspapers, loudspeaker vehicles, and
loudspeakers can only be used by governments and disaster-related agencies. Oral communication,
passive information acquisition via visual or auditory senses can only be used for people themselves.
The other five media (mobile phones, Wechat, microblogs, SMS, email), can be used by both
governments and residents to spread information during a disaster.
Fig. 3. Schedule for information dissemination and personal response in a disaster
When a disaster occurs, dangers may be monitored or discerned by governments, disaster-related
agencies, or residents. After confirming a danger, pre-warning information is spread via social media or
by physical information dissemination mechanisms by governments or information carriers (who have
already received information about the disaster). t1 is the time between the initiation of monitoring
danger to time when the pre-warning information is disseminated. We call t1 the information generation
time. Once the pre-warning information has started to spread, there is a delay before residents begin
receiving the information, denoted by t2. The duration of t2 is dependent on: the habit of media users;
the time; and the properties of specific media. When residents get the pre-warning information, they
will take some responsive steps, such as evacuation. From the time when information is acquired by
residents at the start of the disaster, a safe response time t3 is experienced during which period the
residents are safe. After a disaster has happened, the victims will take time t4 to reach emergency
shelters. Because victims were exposed to the disaster during this period, we call t4 the risk response
time.
Fig. 4 is a flowchart of information dissemination during a disaster based on some social media
TV
Radio Email
News
portal blog
phone S
M
S
Loudspeaker
vehicle
Loudspeaker Newspaper
Oral
Governments and disaster-related
departments
Information
carrier
Disaster happened
Danger was
monitored
Disaster finished
Pre-warning
information
spread based
on different
media
Residents got the
information
Residents arrived at
emergency shelters
Response time for disaster’s carrier
vision
Auditory
sense
t1t2t3t4
Wechat
and physical dissemination mechanisms without government assistance. The items in red are bad for
information acquisition. In the case of microblogs, Wechat and email, information is obtained only
when users decide to use the media to check for information. These all reduce the speed and probability
of information acquisition on a large scale. In addition, mobile phones may be powered off or are busy,
which lead to failed information spread. Information acquired from oral communication, or passively
via visual and auditory senses have to satisfy the condition that there are information carriers or
evacuees nearby. This condition also overly depends on the ambient environment such as the
distribution of buildings and roads. In addition, the spreading distance for these mechanisms is limited.
The items shown in green in Fig. 4 indicate the better ways to get information. These include a ring
tone or vibration of SMS and mobile phones. These functions can remind a user to check for
information so that the delay time for information acquisition is reduced (t2 in Fig. 3).
Fig. 4. Information dissemination during a disaster without the assistance of governments and
disaster-related agencies.
Through the procedure shown in Fig. 4, information will be spread to people who were previously
SMS Blog, Wechat
and Email Phone Oral Vision and
auditory sense
Get text
Ring tone
Get
information
Check
Media get
information
Ring tone
Power off, busy
line and so on
×
Check
Use
media
Evacuees is
near
Information
carrier is near
Evacuees Can
be seen or
noise can be
heard
SMS
Want to
spread
Blog Wechat Email Phone Oral
Evacuation
Sound Scene
Congestion
High risk
unaware of that information and will in turn make them information carriers. They can then
disseminate the information concerning the disaster to other people using their preferred medium.
However, the speed of interpersonal information dissemination without government and
disaster-related agency assistance is too slow. Evacuations based only on interpersonal information
dissemination has high risk and can easily lead to serious congestion since unfamiliar evacuation
scenarios, tension, and insufficient information about the disaster, lead evacuees to sometimes make
wrong choices without the help of governmental agencies.
Taking into account all the problems in the introduced process mentioned above, we have
developed a comprehensive optimized information dissemination model for disasters. Fig. 5 shows the
optimized information dissemination model considering governmental information published in a
disaster. Information concerning a disaster can be monitored before, during or after the disaster by
governments, disaster-related agencies or individuals. If danger is monitored by governments,
depending on the type of disaster and its characteristics, the situation of the affected area, and the
psychology and dynamic distribution of victims, a government will choose specific information media
to spread information concerning the disaster. In this paper, we refer to this process as the inner loop of
information dissemination, depicted by the blue arrow in Fig. 5. If danger is detected by people not
directly affected by the disaster, relevant information can be injected into the flow of information about
the disaster, and spread following the inner loop. When danger is directly detected by victims,
excluding the information dissemination mechanisms mentioned above, victims can evacuate directly if
they have not enough time to spread information. Other victims might get evacuation information from
other nearby evacuees. Depending on personal psychological factors such as conformity, environmental
influences such as traffic jams and the impact of the disaster, different evacuees need to follow different
evacuation plans. These personal response plans should be disseminated to every evacuee during the
evacuation in real time through different media. In addition, using real-time information from evacuees,
an updated response plan can be fed back to evacuees to optimize their evacuation behavior. This
information dissemination cycle for optimized evacuation [20] is termed the outer loop, depicted by the
yellow arrow in Fig. 5. The outer loop achieves the goal of macro- and micro- control during an
evacuation using information dissemination in a disaster. In this information dissemination process, a
government plays a critical role. Prior to a disaster, the one-to-many style of information dissemination
by governments or disaster-related agencies has a higher efficiency than P-to-P (person to person)
information spread. Real-time information dissemination using a combination of different media during
a disaster can also make a big difference to risk reduction in pedestrian evacuation.
Fig. 5. Optimized information dissemination model considering governmental information publishing
in a disaster
In this paper, we studied an information dissemination model during a disaster considering
different information media and governmental assistance. The goal of this work is to reduce the delay
time for information acquisition (t2) and extend the safe response time (t3).
3 Information dissemination models
We cited 3 models to simulate information dissemination under emergencies in this paper. The
three models include probabilistic model (SMS, wechat, microblog, news portal, TV, oral
communication, phone, email, and newspaper), pre-warning model of loudspeaker vehicles, and
passive information dissemination model (information acquirement by personal vision and auditory
sense).
The probabilistic model confirms the information dissemination rules of different media at the
first and the information is spread step by step according to the rules. The probability of each step
could be calculated. Taking mobile phone as an example, information about disasters is spread by
mobile phones. The mobile phone user A could believe the information when three conditions are
satisfied: (1) The information is spread to the mobile phone of user A; (2) User A picks up the phone
during connection; (3) User A believes the information. When user A received and believed the
information, A is possible to spread the information to others. The detailed calculation of probabilities
of each step and the probabilistic models of other media can refer to Zhang’s research [21].
The pre-warning model of loudspeaker vehicles is applied in the regional information
dissemination by loudspeaker vehicles in the condition of information network paralysis. We developed
an improved path optimization algorithm for loudspeakers combining greedy algorithm and exhaustive
method. The residents in buildings can get the information if they heard the sound from loudspeaker
vehicles. The detailed information dissemination under different population distribution, range of
sound transmission, and number and speed of vehicles can refer to Zhang’s research [14].
The passive information dissemination model defines that victims have to acquire the information
through their visual and auditory sense if there no time to spread disaster’s information. The model
calculates the probability of people finding the evacuees by their vision from horizontal and vertical
angle of view and the probability of people hearing the noise from evacuees by their auditory sense
through the experiments of sound transmission under different conditions. The detailed information
about the passive information dissemination model can refer to Zhang’s research [22].
4 Study area
Fig. 6. Population distribution of study area
Beijing is a metropolis with a high density of buildings, roads, and highly dynamic population flow.
In our study we focused on the Zhongguancun area, which is one of the most prosperous and densely
populated areas of Beijing, with more than 430 buildings and 980 roads.
Fig. 6 shows the study area, blocks represent buildings; and black lines the roads. According to the
data from 2010 Beijing Census [23] and NFPA 101 [24], this area covered 2.3 km2 with a population
density of over 156,000 persons/km2 with approximately 365,000 people in the area during office
hours.
5 Results
5.1 Analysis of information dissemination characteristics of different media
According to the principles of information dissemination, we categorize all media as either social
media (based on information networks), or physical (information dissemination mechanisms not using
any networks). Social media include TV, radio, SMS, phone, news portal, microblogs, Wechat, and
email. Physical information dissemination mechanisms include the traditional methods such as
loudspeaker vehicles, loudspeakers, newspapers, and oral communication. According to the speed of
information dissemination, all media are categorized as: high-, medium- or low-speed. We defined
high-speed media to be media that can spread information to more than 90 percent of users within 1
hour; medium-speed media can reach over 90 percent of people within 1 day; low-speed media within
1 week.
We established information dissemination models for 14 media types: TV, radio, SMS, phone,
email, microblog, Wechat, news portal, loudspeaker vehicle, loudspeakers, oral communication,
passive information acquisition through visual cues, and passive information acquisition through
auditory senses. These information dissemination mechanisms are normally used in disasters [25-26].
(a) (b)
(c)
Fig. 7. Efficiency analysis of information dissemination of 13 media types: (a) high-speed (SMS,
loudspeaker vehicle, visual and auditory sense); (b) medium-speed (Wechat, microblog, news portal,
TV, oral communication, and phones); (c) Low-speed (email, radio, and newspaper)
We cited the method of information dissemination via media provided by Zhang [21], taking into
consideration personal preferences of media usage and the reliability of information provided through
different media, to analyze the characteristics and efficiency of each information dissemination
mechanism. Fig. 7 provides the efficiency analysis of information dissemination of 13 media types
(loudspeakers are not included) in the study area (information generation time t1 is not considered here).
The detailed analysis is given below.
(1) TV and radio
TV is one of the most popular media for disseminating information in a disaster, in contrast to
radio that is no longer so common, and therefore doesn’t have as wide a reach. Both of these media can
provide very rapid information updates [27], ranking in the first three of all media considered in this
paper. TV also has the highest reliability [28] at 79% (TV watchers have probability of 79% would
believe the information about disaster if it come from television) [29], and remains the best medium for
rapid publication of information in a disaster. In the disasters of hurricane, the pre-warning information
is usually forecasted on TV one day before the hurricane. Radio, which is usually used in the warning
of meteorological disasters such as flood and extreme weather, is also very popular in the disasters with
long forecasting time. Fig. 7(b) shows that TV is a medium-speed medium, and can reach 90% of users
within 1 day. TV would have a pretty high speed in information dissemination if informed area is large.
But in our study area which has only 365 thousands of residents, the speed of information
dissemination via TV is not that fast. In addition, the speed of information dissemination via TV
depends on the time at which the information is published, since people tend to watch TV more at peak
times. Peak time for TV is between 6 p.m. and 9 p.m. The average watching time in peak time is 6
times than it of day time. Since radio has a lower usage frequency, the speed of information
dissemination via radio is low. Low utilization ratio is also a defect of radio, and is therefore not very
efficient for information dissemination during a disaster.
(2) SMS and mobile phone
According to Fig. 7(a), SMS is extremely fast for disseminating information, and its curve of
information dissemination has logistic characteristics. Without considering the load on the base station,
SMS ranks first among all media over a large area. However, in a small area, its efficiency is the same
as for a loudspeaker vehicle and passive information acquisition through visual and auditory senses.
For information dissemination during a disaster, SMS has the advantage of providing higher
information coverage with a shorter delay time for information acquisition. Mobile phones have the
advantage of massive and accurate content of information, but some conditions such as a busy line, the
phone being powered off, and missed calls (experiments have shown that 50 percent of people cannot
answer the phone on the first call) resulting in lower efficiency of information dissemination. In
addition, overloading of a base station and a paralyzed network caused by disasters will block
information dissemination by SMS and mobile phones.
(3) News portal
Data shows that the web is the most important social medium used to get information. The
number of Chinese web users reached 0.67 billion in 2015, making China the highest ranked in the
world. Moreover, the average time spent on the internet increased to 3.73 hours per day, based on data
from the “Statistical Report of Internet Development in China” provided by the China Internet Network
Information Center (CNNIC). In emergencies, one advantage of the internet is the ability to post
updated information just as quickly, if not more quickly than for TV and radio [30]. News portals are
therefore very suitable for information dissemination in a disaster. In 2008 the Xinhua News Agency
news portal updated information about the Wenchuan Earthquake 18 minutes after it struck [31]. Five
years later in 2013, the Sohu news portal published the news 17 minutes after the Ya’an Earthquake
(http://www.donews.com/net/201304/1475791.shtm). Fig. 7(b) shows that news portal have the steady
speed in information dissemination, but the speed depends on the time of day. In addition, because of
lower information reliability (69%), the final information coverage rate is less than that of TV, SMS
and phones.
(4) Wechat and microblogs
Wechat and microblogs are social media that have become more and more popular in recent
years. Their model of information dissemination is very similar to Twitter. According to 2015 data,
the number of active Wechat users exceeded 0.55 billion and the number of microblog users was
0.21 billion. Because information can be spread from 1-to-N people, the speed is very quick
especially over a large area. Utilizing nodes with big influence is very helpful in information
dissemination in disasters [32]. Governments, disaster-related agencies, and residents can be
information publishers on Wechat and microblogs. Information dissemination can also dispense
with verification increasing flexibility, reducing time taken for intermediate steps, and accelerating
the speed of information dissemination. In the Ya’an earthquake of 2013, the first microblog
information was spread by the Institute of Care-Life of Chengdu 53 seconds after the quake.
Wechat is more popular than microblogs and has higher utilization ratio (55 percent of users use it
more than 10 times per day and 25 percent of users use it more than 30 times per day). Wechat is
already used for information dissemination in disasters [33]. However, compared with SMS, there
is no reminding function when users receive information from governments or disaster-related
agencies. Moreover, because of the informality of information dissemination, rumors often occur
[34]. The reliability of information on a microblog is about 50%. Therefore, having governments
and disaster-related agencies use these media will increase the reliability of the information.
(5) Email
Email has a longer delay time for information acquisition and lower information reliability
(45%), and is therefore rarely used for disseminating information in a disaster. However, in some
catastrophes where the information network is almost destroyed, information dissemination based
on a small volume of data using email is still feasible [35]. Fig. 7(c) shows that the information
dissemination speed of email is really low but it can be appropriately used for long-term
pre-warning of disasters.
(6) Newspapers
In the 19th and 20th centuries, newspapers were the main means of getting information.
Newspapers were usually used to spread pre-warning information of a disaster that has a long
pre-warning time, such as a storm or extreme temperatures [36]. High information reliability (76%)
and coverage, together with a large amount of information are the advantages of this traditional
media. Of all the media shown in Fig. 7(c), there is no doubt that newspapers are the slowest
media when considering information generation time, especially for a small area. Newspapers are
not good for spreading information about disasters that occur quickly.
(7) Loudspeaker vehicles
A loudspeaker vehicle is the information dissemination medium that can be deployed without
the availability of networks or radio. Information can be steadily spread to residents during a
disaster when all information networks are destroyed. In addition, a loudspeaker vehicle can get
almost everywhere and assure residents acquiring the information for the first time, particularly in
a small area. To meet the demands of acquiring information in time, 2-4 loudspeaker vehicles
should be allocated to a 3.2 km2 urban area with high population density [14]. Since this is very
expensive, loudspeaker vehicle are only feasible for crowded areas in a disaster.
(8) Loudspeaker
A loudspeaker is the fastest information dissemination medium that is usually used for
information dissemination in disasters [37]. All residents can get information about the disaster at
almost the same time if all areas are covered by available loudspeakers. In a pedestrian evacuation,
loudspeakers can also be used to provide real-time guidance to avoid serious congestion and so
effectively avoid stampedes [38]. However, loudspeakers have low flexibility and are also
expensive to deploy broadly. Similar to the loudspeaker vehicle, loudspeakers are suitable for
information dissemination in highly populated areas.
(9) Oral communication
Fig. 7(b) shows the oral communication during a disaster in the study area. Since the study
area is very small, the efficiency of information dissemination via oral communication is very
high. When information needs to be spread in a larger area, the speed of information
dissemination via oral communication is too slow due to the limitation of communicating over
distances. Oral communication is easy to implement, and time consumption is low, so that when
combined with other social media, oral communication can be very useful.
(10) Passive interpersonal information dissemination (visual and auditory senses)
When receiving information about a disaster after the event, people will evacuate without
pausing to spread information through other media. In the case where people receive no
information, they may hear or see others evacuating, and follow suit [22]. Fig. 7(a) indicates that
over a small range, passive interpersonal information dissemination is a quick way to spread
information. However, passive spreading over a large area is very slow. Moreover, people who
obtain information passively may not receive accurate information concerning the disaster, and
may therefore not generate a correct and effective response plan during their evacuation. If
combined with real-time governmental guidance, passive spreading of information can be
effective for disseminating information.
5.2 Information dissemination with different assistance by government
Before or during disasters, pre-warning or information about a disaster might be spread
through governments or disaster-related agencies. Efficient information dissemination by
governments in disasters can promote information acquisition because residents open to receiving
information if it is published by a government. However, information published by governments
cannot always reach everyone in the disaster area. SMS, email, microblogs and Wechat, which are
commonly used by governments and residents (shown in Fig. 3), were chosen to analyze
information dissemination efficiency under the different assistance degree by governments. We
use information coverage ratio (percentage of residents who can be directly informed by
government) to describe governmental assistance.
In Fig. 8, the blue line shows information dissemination without government assistance.
Initially information dissemination is slow, but once a critical mass of people are in possession of
the information relating to the disaster, the speed of information dissemination is high. Disaster
information published by governments can obviously increase the number of people receiving the
information initially. Fig. 8 shows that for these four social media, with the increased
governmental information coverage ratio (100% governmental information coverage ratio means
that all residents can be reached by information about disaster published by governments), the
speed of information dissemination will increase. However, the efficiency of information
dissemination declines with increased governmental information coverage ratio. Therefore, if
residents can communicate adequately, information publish by government which coverage ratio
is not very high is also very useful.
(a) (b)
(c) (d)
Fig. 8. Information dissemination under diferent governmental information coverage ratios:
(a) SMS; (b) Email; (c) Wechat; (d) Microblog
As shown in Fig. 8(a), among the four social media including SMS, email, microblogs, and
Wechat, SMS has the highest sensitivity for governmental information coverage ratio. Because
SMS has the highest utilization ratio, the least delay time and a poorer ability to forward
information compared with microblog and Wechat, improving information coverage ratio by
governments can help accelerate the speed of information acquisition. On the contrary, for email
(Fig. 8(b)), we could see that with the continuous improvement of governmental information, the
increase of informed population becomes slow because email has a very strong ability in
information forwarding.
From Fig. 8, the final number of informed population through SMS, microblogs and Wechat
almost keep the same, except emails. Many email users would miss the information about disaster
for a long time if they do not use email frequently. Therefore, when information disseminated by
emails, governmental information coverage has a strong influence to increase the final number of
informed population.
5.3 Comprehensive analysis of information dissemination ability
In order to analyze efficiency and ability of information dissemination of different media, we
set five indices including information coverage (C), time to notice half of users (H), reliability
(probability that people believe the information through media) (R), usage frequency (F), and
amount of information (A). In this part, six media including TV, SMS, loudspeaker vehicle, news
portal, microblog and email are chosen as examples to analyze comprehensive efficiency of
information dissemination.
Fig. 9 is the radar maps of information dissemination efficiency of six media based on five
assessment indices. The area of green polygon partially describes the information dissemination
ability. In our study area, loudspeaker vehicles, TV and news portals ranked top three in
high-efficiency information dissemination, while microblog and email ranked the last two.
Because of limited range of loudspeaker vehicles, it should be used in a small area with high
population density and high risk of disasters. TV has large amount of information but low usage
frequency. It fits for spreading information at peak time such as 18:00 to 24:00. News portal has a
steady ability in information dissemination and improving its reliability is an efficient way to
accelerate speed of information dissemination. The amount of information of SMS is too little,
and people are difficult to get whole information. The information dissemination ability of
microblog and email is low. Therefore they should not be used alone during disasters. All of these
results listed above are obtained based on the scale of study area. The efficiency of information
dissemination would be changed if simulation scale is varied.
Fig. 9. The radar maps of information dissemination efficiency of six media: TV, SMS, loudspeaker
vehicle, news portal, microblog and email
5.4 Comprehensive information dissemination using a combination of different media
A single information medium cannot meet all the demands for timeliness of information
dissemination in disasters. It is therefore important to determine how to combine different media to
increase efficiency of information dissemination. We do not consider the newspaper and passive
information acquisition by visual and auditory senses. Examining the characteristics of information
dissemination for each medium shown in Fig. 10, SMS, phone, Wechat, microblogs, and email are
initially slow for disseminating information. There is no initial limitation with TV, radio, and news
portals, but they are overly dependent on the time of day. Compared with these media, loudspeaker
vehicles and oral communication provide a steady speed when disseminating information.
Overusing information media during disasters will spend unnecessary time and money. We
assumed information is spread at 6 p.m. Fig. 10 shows four combinations of different media which
have: (a) fast and slow information dissemination speed (SMS & Radio); (b) fast information
dissemination speed (SMS & Loudspeaker vehicle); (c) similar characteristics of information
dissemination (Wechat & Microblog); (d) different characteristics of information dissemination
(Wechat & News portal).
(a) (b)
(c) (d)
Fig. 10. Information dissemination for different combinations of media:
(a) SMS & radio; (b) SMS & loudspeaker vehicle; (c) Wechat & microblog; (d) Wechat & News portal
In Fig. 10(a), although the information dissemination speed of radio (purple line) is far less than that of
SMS (red line), the efficiency of information dissemination in combination (green line) is greatly
improved. The reason for this is that information dissemination by radio initially fast, thus covering for
the initial slowness of SMS. Two media which both have fast speed in information dissemination are
combined in Fig. 10(b), the speed is improved about 20%. For media which have the same
characteristics in information dissemination (Wechat and microblog have the speed limitation of initial
information dissemination), speed can be improved but the problem of limited original speed still exists
(as shown in Fig. 10(c)). On the contrary, in Fig. 10(d), combine media with different characteristics of
information dissemination, efficiency can be obviously improved. Therefore, carefully combining
different media with different information dissemination characteristics can improve efficiency in a
large scale.
6 Conclusions
A comprehensive model of information dissemination in disasters is very helpful in increasing
personal response time, guiding evacuation plans, and reducing risk. The inner information
dissemination loop guides pre-warning information dissemination and the outer information
dissemination loop assures the efficiency by real-time macroscopic governmental information
publication. According to analysis of the characteristics of the 14 information dissemination
mechanisms: television, radio, SMS, phone, news portals, Wechat, microblogs, email, newspapers,
loudspeaker vehicles, loudspeaker, oral communication, and passive information acquisition via visual
and auditory senses, the different media demonstrate strong differences. All media can be categorized
according to the speed of information dissemination: high-, medium-, and low-speed media. From
characteristics of information dissemination curve view, there are three different shapes including
logistic, linear, and logarithmic. In addition, according to mechanisms of information dissemination,
media can be categorized to social media and traditional media. Social media are based on information
network and traditional media are based on physical methods. By analyzing different combinations of
media, we found that the efficiency of information dissemination could be greatly improved if the
media have different characteristics. A low-speed medium such as radio can also assist information
dissemination by a high-speed medium such as SMS to overcome the limitation of a slow initial speed.
Moreover, sensitivity of governmental information coverage ratio is studied. We found that SMS has
the highest sensitivity and email has the lowest sensitivity in governmental information coverage ratio.
In a disaster, considering time and economy consumption in information dissemination, government
need not to spread information covering every people. Under the condition of a good interpersonal
communication, governmental information coverage ratio could be appropriately reduced. The model
and results provide useful references for information dissemination under disasters in urban areas.
Acknowledgements
This work was supported by the National Natural Science Foundation of China (Grant No. 71473146)
and the Ministry of Science and Technology of the People´s Republic of China under Grant No.
2015BAK12B01.
References
[1] S. Zahran, D. Tavani, S. Weiler, Daily variation in natural disaster casualties: information flows,
safety, and opportunity costs in tornado versus hurricane strikes. Risk Anal. 33 (2013) 1265-1280.
[2] S. Kobayashi, M. Hanagama, S. Yamanda, M. Yanai, Home oxygen therapy during natural
disasters: lessons from the great East Japan earthquake. Eur. Resp. J. 39 (2012) 1047-1048.
[3] D.H.A.A. Khudhairy, A. Alessandro, Tsunami: time for models to be tested in warning centres.
Nat. 464 (2010) 350-350.
[4] D.H. Dale, G. James, Tsunami: unexpected blow foils flawless warning system. Nat. 464 (2010)
350-350.
[5] T. Omi, Y. Ogata, Y. Hirata, K. Aihara, Forecasting large aftershocks within one day after the
main shock. Sci. Rep. 3 (2013) 2218.
[6] B.G. Mcadoo, L. Dengler, G. Prasetya, V. Titov, Smong: How an Oral History Saved Thousands
on Indonesia’s Simeulue Island during the December 2004 and March 2005 Tsunamis. Earthq.
Spectra. 22 (2006) S661-S669.
[7] W. Macias, K. Hilyard, V. Freimuth, Blog functions as risk and crisis communication during
Hurricane Katrina. J. Comput-Mediated Commun. 15 (2009) 1-31.
[8] P. Meier, R. Munro, The unprecedented role of SMS in disaster response: Learning from Haiti.
SAIS Rev. Int. Affair 30 (2010) 91-103.
[9] X. Lu, C. Brelsford, Network Structure and Community Evolution on Twitter: Human Behavior
Change in Response to the 2011 Japanese Earthquake and Tsunami. Sci. Rep. 4 (2014) 6773.
[10] V. Palchykov,M. Mitrović, H.H. Jo, J. Saramäki, R.K. Pan, Inferring human mobility using
communication patterns. Sci. Rep. 4 (2014) 6174.
[11] M.C. Gonzalez, C.A. Hidalgo, A.L. Barabasi, Understanding individual human mobility patterns.
Nat. 453 (2008) 779-782.
[12] J.H. Sorensen, Hazard warning systems: Review of 20 years of progress. Nat. Hazards Rev. 1
(2000) 119-125.
[13] Y. Ran, Considerations and suggestions on improvement of communication network disaster
countermeasures after the Wenchuan earthquake. IEEE Commun. Mag. 49 (2011) 44-47.
[14] N. Zhang, H. Huang, B. Su, H. Zhang, Population evacuation analysis: considering dynamic
population vulnerability distribution and disaster information dissemination. Nat. Hazards 69
(2013) 1629-1646.
[15] N. Uchida, K. Takahata, Y. Shibata, N. Shiratori, Never Die Network Extended with Cognitive
Wireless Network for Disaster Information System. Complex Intell. Softw. Intens. Syst. (IEEE,
2011) 24-31.
[16] M. Nakatani, D. Suzuki, N. Sakata, S. Nashida, A study of a sense of crisis from auditory warning
signals. Proc. World Congress Eng. Comput. Sci. 1 (2009).
[17] M. Kobes, I. Helsloot, B. de Vries, J.G. Post, N. Oberijé, K. Groenewegen, Way finding during
fire evacuation; an analysis of unannounced fire drills in a hotel at night. Build Environ. 45 (2010)
537-548.
[18] V.C. Poekoel, K. Hira, Y. Chisaki, T. Usagawa, Unidirectional Sound Signage for Speech
Frequency Range Using Multiple-Loudspeaker Reproduction System. Open J. Acoust. 3 (2013)
120-126.
[19] W. Lee, M. Cheon, C.H. Hyun, M. Park, Development of building fire safety system with
automatic security firm monitoring capability. Fire Safety J. 58 (2013) 65-73.
[20] N. Zhang, H. Huang, B. Su, J.L. Zhao, Analysis of dynamic road risk for pedestrian evacuation.
Physica A 430 (2015) 171-183.
[21] N. Zhang, H. Huang, B. Su, J.L. Zhao, B. Zhang, Information Dissemination Analysis of
Different Media towards the Application for Disaster Pre-Warning. PloS One 9 (2014) e98649.
[22] N. Zhang, X.Y. Ni, H. Huang, J.L. Zhao, M. Duarte, J. Zhang, The impact of interpersonal
pre-warning information dissemination on regional emergency evacuation. Nat. Hazards 80 (2016)
2081-2103.
[23] The 5th census information office and statistical bureau of Haidian district. Census information of
Haidian district of Beijing, China Statistics Press (2012).
[24] ISO/TC 131, National Fluid Power Association.
[25] M. Grabowski, K. Roberts, High reliability virtual organizations: Co-adaptive technology and
organizational structures in tsunami warning systems. ACM T. Comput-Hum. Interac. 18 (2011)
19.
[26] Y. Fujinawa, Y. Noda, Japan's Earthquake Early Warning System on 11 March 2011:
Performance, Shortcomings, and Changes. Earthq. Spectra. 29 (2013) S341-S368.
[27] A. Tanner, D.B. Friedman, A. Koskan, D. Barr, Disaster communication on the Internet: A focus
on mobilizing information. J. Health Commun. 14 (2009) 741-755.
[28] D. Walton, S. Lamb, V. Dravitzki, An experimental investigation of the influence of media type
on individual perceptions of the severity of earthquake events. Int. J. Emerg. Manag. 4 (2007)
630-648.
[29] N. Zhang, H. Huang, B. Su, J.L. Zhao, B. Zhang, Dynamic 8-state ICSAR rumor propagation
model considering official rumor refutation. Physica A 415 (2014) 333-346.
[30] D.A. Troy, A. Carson, J. Vanderbeek, A. Hutton, Enhancing community‐based disaster
preparedness with information technology. Disasters 32 (2008) 149-165.
[31] P.J. Xu, The analysis of the reactions of internet media to the unexpected events in 2008. Dalian
University of Technology (2009) (In Chinese)
[32] R. Narayanam, Y. Narahari, A shapley value-based approach to discover influential nodes in
social networks. IEEE T. Autom. Sci. Eng. 8 (2011) 130-147.
[33] Y.J. Zheng, Q.Z. Chen, H.F. Ling, J.Y. Xue, Rescue Wings: Mobile Computing and Active
Services Support for Disaster Rescue. IEEE T. Serv. Comput. 1-1 (2015).
[34] L. Zhao, Q. Wang, J. Cheng, Y. Chen, J. Wang, W. Huang, Rumor spreading model with
consideration of forgetting mechanism: A case of online blogging LiveJournal. Physica A 390
(2011) 2619-2625.
[35] N. Uchida, K. Takahata, Y. Shibata, Disaster information system from communication traffic
analysis and connectivity (quick report from Japan Earthquake and Tsunami on March 11th,
2011). Network-Based Inform. Syst. 2011 14th Intern. Conf. (2011) 279-285.
[36] T.T. Huynh, Y.J. Park, A.Y. Chen, Faces of China: new Chinese migrants in South Africa, 1980s
to present. Afr. Asian Stud. 9 (2010) 286-306.
[37] I.M. Susmayadi, Sudibyakto, H. Kanagae, W. Adiyoso, E.D. Suryanti, Sustainable Disaster Risk
Reduction through Effective Risk Communication Media in Parangtritis Tourism Area,
Yogyakarta. Procedia Environ. Sci. 20 (2014) 684-692.
[38] D. Helbing, P. Mukerji, Crowd disasters as systemic failures: analysis of the Love Parade disaster.
EPJ Data Sci. 1 (2012) 1-40.
