International Journal of Disaster Risk Reduction 95 (2023) 103842

Contents lists available at ScienceDirect

International Journal of Disaster Risk Reduction

journal homepage: www.elsevier.com/locate/ijdrr

A dynamic emergency decision support model for emergencies in

urban areas
Linghan Meng, Wuyi Cheng *
School of Engineering and Technology, China University of Geosciences (Beijing), Beijing, 100083, China

ARTICLE INFO ABSTRACT

Keywords: Given the dynamic and cascading nature of urban emergencies, decision-makers are often faced
Emergency decision-making with challenging emergency decisions. In this study, a dynamic emergency decision-support
Scenario evolution model considering the cascading effect is proposed. First of all, this paper divides the whole
Mobile phone location data process of urban emergency development into several time slices, and each time slice describes a
Scenario cascade network
cascading process. Then, it uses the population distribution heat map to obtain the distribution of
Urban emergencies
affected people in each scenario area during the cascading process to assess severity. At the same
time, using the network index, a method of analysing the risk of each scenario from the perspec-
tive of network structure is proposed to determine the degree of risk. Finally, combine these two
indicators to determine the scenarios that need to be dealt with urgently at this stage. In addition,
this study takes the Tianjin Port's hazardous cargo explosion accident as an example to illustrate
the application of the proposed method. The results show that this method can identify the key
emergency areas and targets in each time slice.

1. Introduction
Emergencies are always unpredictable and challenging, especially in densely populated and industrialised urban areas. High-
density buildings, complex transportation networks, and dense distribution of people, such as complex urban systems, a vast disaster-
bearing body, make emergencies more prone to cascading effects [1]. This phenomenon, expands the impact of accidents, such as the
Qingdao oil pipeline explosion in 2013 [2], the Tianjin explosions in 2015 [3], and the Japan earthquake and tsunami in 2011 [4].
Complex urban emergencies stress the emergency response capacity of governments and challenge sustainable urban development
[5].
The emergency management capability of a city is crucial for ensuring its safety and sustainable development [6]. Emergency de-
cision-making (EDM) is one of the core elements and critical aspects of emergency management and determines its effectiveness. Cur-
rently, researches on EDM methods have attracted the attention of scholars, and these studies have made breakthroughs in medical
scheduling [7,8], resource allocation [9,10], and public health [11,12]. To solve many practical problems in the process of EDM,
many scholars have used different methods to establish relevant decision-making models in the stages of event prediction and re-
sponse [13–16]. Undoubtedly, these studies provide positive guidance for making EDM. However, emergencies have the characteris-
tics of multi-stage, dynamic and information updating [17]. During the emergency response to Hurricane Katrina in 2005, various
problems have a risen, such as information lag, delayed decision-making and lack of dynamic decision-making, resulting in heavy
losses [18]. To deal with this situation, a novel decision-support method is needed to support the emergency decision. Therefore,
some researchers turn their attention to dynamic EDM.

* Corresponding author.
E-mail address: [email protected] (W. Cheng).

https://doi.org/10.1016/j.ijdrr.2023.103842
Received 29 November 2022; Received in revised form 30 June 2023; Accepted 3 July 2023
Available online 8 July 2023
2212-4209/© 2023 Elsevier Ltd. All rights reserved.
L. Meng and W. Cheng International Journal of Disaster Risk Reduction 95 (2023) 103842

Considering that the preferences of decision-making groups may change during the EDM process, Xu et al. [19] proposed a new
method for dynamic EDM for large-group risk based on cumulative prospect theory. Fang et al. [20] proposed a decision support
method based on case similarity matching, which matched the similarity between changing emergency scenarios and cases, and then
provided decision support information for decision-makers to deal with current emergencies. Keneally et al. [9] developed a Markov
decision process model for the optimal dispatch of military medical evacuation assets. Considering the psychological behaviour of hu-
man beings, Wang et al. [21] established an emergency decision-making method during beforehand-ongoing two stages based on the
regret theory. Zheng et al. [22] proposed a dynamic method based on case retrieval and group decision-making, which can adjust the
emergency alternative according to changing emergencies. Considering the evolution of the emergency, Liu [17] proposed a dynamic
programming model based on the psychological reference satisfactory degree, to effectively generate and adjust the alternative of
each stage. Wan et al. [23] proposed a multi-period dynamic emergency material distribution model under uncertain demand by con-
sidering the disaster degree and demand of different disaster areas. The above literature accounts for the dynamics inherent to a real
EDM process from the perspectives of information updates, decision-maker psychological changes, and situational changes during the
decision-making process. In the selection of emergency decision-making schemes, the deployment of emergency supplies, the distrib-
ution of rescue forces, and other aspects of dynamic decision-making model construction research. These studies have obtained some
research results. However, in the existing dynamic decision analysis methods, the cascading effects of urban emergencies are rarely
considered, and few studies have considered the impact of cascading effects on emergency decisions.
Previous accident cases show that emergencies will have a cascading effect under the influence of the complex urban environ-
ment, forming a dynamic evolution process [24]. Therefore, what decision-makers are faced with is not just a single hazard scenario
with significant impact but multiple hazard scenarios, and these event scenarios may occur simultaneously or sequentially. At the
same time, some chain events may be more serious and complicated than the trigger and need urgent response [25]. This cascading
effect often leads to the inability of decision-makers to effectively judge the current emergency situation and determine which cascad-
ing scenarios need to be dealt with urgently. Emergency cascading effects are complex and multidimensional, and evolve over time
[25]. Different cascading processes may appear in each time stage of emergency development, so we need to consider the cascading
process of emergencies in the process of making dynamic EDM. Based on the above discussion, in this research, we propose a dynamic
emergency decision support model considering the cascading effects of emergencies, which divides the entire development process of
urban emergencies into various stages from the perspective of time dynamics, and then analyses the emergency urgency of each sce-
nario in the cascading process to support EDM.
Decision-making requires decision-makers to judge the current emergency situation. The timely and accurate estimation of im-
pacts and damages is essential to better understand emergency conditions and to support emergency response operations [26]. In
some previous studies, scholars judged the situation by quantitatively evaluating event scenarios to make accurate decisions [27,28].
So, this study proposes two indicators (severity and risk) to measure the emergency urgency of each event scenario in the emergency
cascade process, so as to identify the event scenarios that need to be focused on in each period of the emergency process, and provide
recommendations for decision-makers to develop decision options at each stage. The severity of an event plays a crucial role in the
EDM process, and decision-makers often give priority to events with higher severity. In this study, combined with the regional situa-
tion of the actual accident scene, this research obtains information through location data to determine the severity of the scenario
area at different times. The development of the information age has facilitated acquiring disaster information. The texts, images, map
positioning, and mobile phone signals obtained by social media platforms provide a large amount of information and data. Real-time
data can assist decision-makers in emergency decisions [29–31]. Compared with traditional methods, these location data can assist
decision-makers in obtaining real-time disaster information quickly, with high accuracy, and on a large scale, effectively solving the
problem of lagging information in the dynamic decision-making process [32].
Currently, Knowledge-Unit [33], Bayesian network methods [34,35], Petri net [36,37], and other methods are used to analyse the
evolutionary process of events and help decision-makers determine the relative danger to adjust emergency decisions. However, these
methods focus on exploring the development process of events to assist decision-making. They rarely consider the threat of a situation
from the perspective of cascading effects. Networks theory plays a vital role in studying the evolution of disasters in recent years [38].
Since complex networks can be used to present the complex connections of real-world systems, some metrics based on node and edge
creation can better understand the topology of complex networks. Complex network theory is often used to analyse the cascading ef-
fects. Tang et al. [39] analysed cascading effects of earthquakes in urban areas using complex networks for providing insights to de-
sign disaster mitigation strategies from system thinking perspectives. Chen et al. [40] combined various accident cascading effects in
urban transit systems with complex networks to judge the vulnerability of accidents. Hu et al. [41] adopted a complex network ap-
proach to examine the cascading effects of typhoon disasters and to identify the types of hazard events and vulnerability paths for ur-
ban areas. Based on this background, combined with the network index and topology index of the complex network model, the risk of
the scenarios in the cascading process of emergencies can be effectively analysed.
Based on the above discussion, the main purpose of this study is to propose a dynamic emergency decision-support model that
considers the cascading effects of urban emergencies, and determines the phased decision-making objectives by evaluating the emer-
gency urgency of each event scenario in the cascading process. To provide reference suggestions for decision-makers to make emer-
gency decisions. Compared with traditional decision-making methods, this paper considers cascading effects in the dynamic decision-
making process, which aligns more with the cascade characteristics of urban emergencies. The remainder of this paper is organised as
follows. Section 2 introduces the evolution mechanism of urban emergencies. Section 3 presents the dynamic EDM model. Section 4
presents the results and discusses a case application. Finally, Section 5 concludes the study.

2
L. Meng and W. Cheng International Journal of Disaster Risk Reduction 95 (2023) 103842

2. Conceptual framework of urban emergencies scenario evolution

After an emergency, due to the influence of various factors, other scenarios are derived from the initial scenario. These scenarios
change in different stages, generating a dynamic scenario evolution process. To investigate the cascading process of emergencies in
different periods, based on the chronological nature of emergencies, time slices were used to divide the entire duration of the develop-
ment of emergencies into smaller time intervals [42]. The length of time interval can be determined according to the actual situation
over time. Each time slice represents a scenario cascading process at different periods, as shown in Fig. 1. From the occurrence of an
emergency to its end, the scenario not only has its state change, but also may set off other scenarios. The emergency is a multi-level
and dynamic system, and the corresponding EDM is multi-level and multi-objective [43].
Fig. 1 shows that scenario S1 continues to develop between time t1 and t2, and a cascade phenomenon occurs at t2, expanding the
accident. However, scenario S2 is controlled after the occurrence of t2 and does not last until t3. The scenarios at each stage change
due to the intervention of relevant emergency measures or develop and spread rapidly due to the lack of timely or adequate inter-
vention [44,45]. Therefore, decision-making is not a one-time single decision but a multi-stage decision-making process, which con-
tinuously and dynamically adjusts the emergency response plan [46]. After the emergency, emergency decision-makers need to con-
tinuously observe the development of the emergency, determine the current direction of emergency actions from a complex process,
and develop multi-level sequential decisions in response to the cascading effect of scenarios at different stages.

3. Construction of dynamic EDM model

3.1. Framework
The increasing complexity of urban areas allows the impact of urban emergencies to spread rapidly. Hence, the time left for deci-
sion-makers is limited, and government emergency decision-makers must quickly judge and make choices based on the actual situa-
tion [47]. As mentioned above, each time slice is associated with a different scenario cascade process, and the cascade process is very
complex, so it is very important to scientifically analyse the emergency urgency of each stage of the scenario for making emergency
decisions. In the EDM process, decision-makers focus on key scenarios at each stage [48]. In the so-called key scenario, the situation
has to be dealt with urgently—that is, it is a very serious scenario [15]. Additionally, a scenario can have a huge impact on the devel-
opment of the accident—that is, it has a high degree of risk [49]. Based on this idea, according to the actual situation of the cascading
process, this paper selects the two indicators of scenario severity and risk to determine the emergency urgency of each scenario.
The severity of emergencies plays a crucial role in the formulation of emergency decisions. Population, buildings, infrastructure
are the main influencing factors for severity assessment [50,51]. In the emergency response process, it is first necessary to determine
the population distribution information of the scenario area and carry out emergency rescue [52]. Therefore, this study considers the
distribution of affected people as the key factor in determining the urgency of emergency response to an accident. The complex sys-
tem of cities increases the number of people affected by accidents, causing an escalation in the severity of accidents [53], and such sit-
uations require emergency rescue and evacuation to mitigate accident losses. The more people affected by an accident, the greater the
severity of the accident and the greater the urgency of the emergency. In the past, the number of casualties has been used to classify
an accident's severity and initiate the corresponding emergency action plan [54]. However, the number of casualties is a statistical in-

Fig. 1. The schematic representation of time slices.

3
L. Meng and W. Cheng International Journal of Disaster Risk Reduction 95 (2023) 103842

dicator gathered after the event, which does not meet the ‘urgency’ and ‘timeliness’ requirements of the emergency response [55].
Urban emergency decisions are made based on the situation of on-site accidents. Given the high positive correlation between the dis-
tribution of the population and the number of potential casualties [56], using the number of people currently affected to determine
the severity of the scenario is reasonable.
Emergencies have a high risk, and the more dangerous the scenario, the more urgently it should be dealt with. For the determina-
tion of the risk level of emergencies can be determined through methods such as historical case inference [57] or establishing an index
system to assess [58,59]. In addition, the degree of risk can be determined through joint modelling with the degree of severity. Con-
sidering that multiple event scenarios are involved in the cascading process, it is often difficult to use this method for analysis [60,61].
In the past, these methods mostly focused on a single event, but as the cascading effect of urban emergencies has become increasingly
prominent, the combined effect of the cascading phenomenon can strongly amplify the risk caused by emergencies [25]. Therefore, in
addition to focusing on localised problems, systematically understanding the overall situation when making emergency decisions is
crucial [62]. As mentioned above, after an emergency occurs, the event often forms a complex cascade network due to the influence
of many factors. To address the difficulties in emergency decision-making caused by the cascade effect of the accident, it is necessary
to consider the macro-evolutionary system to minimise the loss and impact caused by the accident. During the formation of the sce-
nario cascade network, specific scenarios play a crucial role in the evolution process, and the emergence of these scenarios makes the
disaster damage accumulate and expand. These scenarios are the focus of emergency response [25]. So, we introduce a complex net-
work model to analyse the cascade network structure of scenarios, evaluate the degree of risk of each scenario through relevant indi-
cators. Based on the above analysis, the emergency urgency of each scenario is evaluated based on its degree of risk and severity, so as
to identify the scenarios that need to be dealt with urgently during the evolution of emergency scenarios.

3.2. Scenario severity assessment

Information collection is the primary function of government EDM, and the timeliness and accuracy of the information directly af-
fect its effectiveness. The number of people affected is essential information for EDM; hence, obtaining the distribution of population
data efficiently and accurately is crucial in the information collection stage. Two traditional ways exist to ascertain the number of
people affected by accidents: one is to use the population census data and the city statistical yearbook data to calculate the number of
trapped people in the corresponding areas, and the other is to obtain it through on-site personnel communication. However, very lit-
tle time is left for information collectors once an emergency occurs. Predictions based on census data cannot accurately capture the
real-time distribution data of the affected population, and obtaining dynamic data of the trapped population in all areas based on
scene surveys may be challenging. The uncertainty of the spatial scope and the event generate difficulties of estimation. Therefore,
obtaining distribution data of the people in the area affected by the accident scenario is a problem that urban emergency managers
must solve.
In recent years, with the advent of Big Data, the mining and analysis of location data may provide a real-time and objective per-
spective for information collection, compensating for the lack of timeliness and dynamics of traditional data surveys [63]. In the 2010
Earthquake in Haiti, information collected from social media and open street maps became crucial for disaster relief due to the lack of
data from official sources [64]. With the popularisation of location equipment, a large amount of geographic and human social infor-
mation data containing spatiotemporal information is being produced continuously [31]. In 2011, the Baidu company developed the
Baidu heatmap, a Big Data visualisation product to intuitively reflect the spatial distribution of the population in various regions [65].
The product analyses the spatial distribution of users' real-time location data by clustering, uses the ‘heat index’ to indicate the degree
of crowd aggregation in each area and uses different colours to represent the heat value on Baidu Maps. In addition, the heat index is
time-sensitive and is updated every 15 minutes to ensure real-time information. This product makes it possible to obtain the dynamic
distribution and flow characteristics of the population, and the use of the product has gradually matured in the fields of urban plan-
ning [66,67] and epidemic control [68], and other areas.
Therefore, based on this feature of the Baidu heatmap, this study obtains a real-time heat map of the distribution of people in each
scenario-affected area at each time slice. It uses ArcGIS software to convert the heat map from colours to a 0–7 value tiff raster [65],
obtain the heatmap raster data of the study area, and calculate the population heat value of each scenario area through Eq. (1), to
judge the severity of each area in the scenario development and evolution process. The population heat value of each scenario area is
calculated using the following equation:
m
(1)
∑
H= Cn Vn
n=1

where H is the population heat value of each scenario area, m is the total number of colours (m = 8 in this study), Cn is the heat value
corresponding to the n th colour, and Vn is the number of pixels corresponding to the n th colour (n = 0,1,2,3,4,5,6,7).

3.3. Scenario risk assessment

A complex network model is an approach for abstracting complex systems based on graph theory [69]. A ‘point’ of arbitrary shape
and size is used to describe an entity, and an ‘edge’ is employed to describe a relationship. Such displays of points and lines are topo-
logical graphs, which allow investigating the relationships between entities. The accident scenario cascade network addressed in this
study has the topology characteristics of a complex network. The accident scenario can be regarded as a node in the network, and the
cascading effect between accident scenarios is an edge. Therefore, this study combines the theoretical knowledge of complex net-

4
L. Meng and W. Cheng International Journal of Disaster Risk Reduction 95 (2023) 103842

works and the structural characteristics of scenario cascade networks, using typical indicators of network topology properties to mea-
sure the risk degree of events.
It is well known that cascading is highly affected by a tiny fraction of key nodes [70]. Considering the topological structure, the
more important the scenario node is in the whole network structure, the higher the risk level of the scenario. Commonly used indica-
tors to evaluate the importance of nodes include degree, closeness centrality, and betweenness centrality [71–74]. Among them, out-
degree is the simplest measure in quantifying node importance in the directed network, and it has been widely used in identifying im-
portant nodes [74]. In the directed network, the higher value of the out-degree for a given node indicates that the more nodes the
node points to, the more influential the node is in the network. In the scenario cascade network, the out-degree indicates the number
of other scenarios directly triggered by the accident scenario. The higher the value of the out-degree of the scenario, the higher the
risk of the scenario. However, the risk degree of the scenario should also consider the scenario's impact on the whole cascaded net-
work. In the scenario cascade network, the energy released by the unexpected accident evolves and expands with the cascading effect.
Hence, the edges in the network can be understood as the event energy transmission channels, which contribute to the formation of
the cascade network. In 2002, Girvan and Newman have proposed the concept of edge betweenness to reflect the strength of energy
transfer capability [75]. Pescaroli [76] has demonstrated that identifying the interactions orienting the escalation of secondary emer-
gencies through paths is crucial. Therefore, the importance of a node in the network is also related to the ‘knock-on’ effect edge of the
node. The correlation between scenarios also affects decision-making. During the formation of the scenario cascade network, the
larger the edge betweenness in the evolution of the scenario node, the more times the edge acts as a ‘bridge’ in forming the scenario
cascade network. The greater the role played by this edge in the dissemination and diffusion of disaster impacts, the higher the risk in
the network. Therefore, edge betweenness reflects the degree of correlation between scenarios as shown in Eq. (2):

( ) i,j,k,l
( )
∑ Nkl eij
CB eij = (2)
∈V
Nkl

where Nkl(eij) is the number of the shortest paths going across the edge eij between nodes k and l. Nkl is the number of the shortest paths
between nodes k and l.
The scenario cascade network unit analysis in Fig. 2 indicates that the risk of the node also depends on the importance of its sec-
ondary node in the network. The greater the impact of its secondary node Sj on the entire network, the higher the risk of node Si.
Among the topological indicators of complex networks, betweenness centrality is a global attribute of the network [39]. It reflects the
importance of the accident scenario in the scenario cascade network, as follows:

j≠i
Njk (i)
(3)
∑
BCi =
≠k∈V
Njk

where Njk(i) is the number of shortest paths from node j to k that pass through node i and Njk is the total number of shortest paths from
node j to k.
Considering the local and global effects of nodes, the three indicators of edge betweenness, out-degree, and betweenness centrality
are combined to analyse the risk level of each scenario and identify the critical scenario assessment in the scenario cascade network.
The proposed model reads as follows:
∑ ( )( )
I i = Li + CB eij 1 + BCj (4)
i≠j∈V

where Ii is the risk degree of node i, Li represents the out-degree of node i, CB(eij) indicates the edge betweenness of the edge from node
i to node j, and BCj represents the betweenness centrality of node j.
To verify the validity of the developed model, a network consisting of 24 nodes and 50 directed links is constructed, and uses Eq.
(4) for calculation. Fig. 3 (b) shows the linear relationship between the model proposed in this paper and the measure based on node
out-degree, where the adjusted R-squared of the linear fit is 0.9557. Obviously, there is a high similarity between the proposed
method and the importance assessment by using the traditional method, indicating that the rationality of the proposed approach is
verified to some extent. However, compared with the traditional measure of judging the importance of nodes based on the out-degree

Fig. 2. The scenario cascade unit.

5
L. Meng and W. Cheng International Journal of Disaster Risk Reduction 95 (2023) 103842

Fig. 3. Verify the validity of the developed model: (a) The built complex network. (b) A linear fit between the node out-degree and the model proposed in this paper. (c)
Comparison between the two methods.

of nodes, the proposed method has higher discrimination. From Fig. 3 (c), it can be found that the calculation using Eq. (4) can more
accurately distinguish the importance of nodes with the same node out-degree value. From the perspective of actual scenario cascade
analysis, even if the triggering scenarios have the same out-degree, but the risk levels of the induced scenario cascaded by the scenar-
ios are different, the risk levels of the triggering scenarios are also different. Through verification, it can be found that the formula
constructed in this paper provides a more realistic measure of the degree of risk to assist decision-makers in sorting out the risk of
each scenario in the cascading process.

3.4. Scenario emergency urgency assessment

By analysing the above indicators, we relate the urgency to the severity of the accident scenario and the cascading risk. From the
definition of each indicator, the higher the severity of the scenario area, the higher the urgency of the emergency. The higher risks the
accident scenario, the higher the urgency of the emergency. Based on the logical relationship between the two factors, this paper uses
the product of the severity value and the risk value to measure the emergency urgency of the scenario, and constructs an evaluation
model of the emergency urgency E, as follows:

Ei = Hi Ii (5)

Where Ei is the emergency urgency of the accident scenario Si. Hi is the normalised calculated population heat value of the scenario,
namely, the severity of scenario Si, Ii is the risk value of scenario Si.

6
L. Meng and W. Cheng International Journal of Disaster Risk Reduction 95 (2023) 103842

4. Case application
In this section, a case study is conducted to illustrate the use of the proposed method. This section uses the Tianjin Port 8.12 fire
and explosion accident as a background to conduct research. We investigate how to determine the emergency urgency of each sce-
nario in a specific time slice by implementing the above framework. In doing so, we provide recommendations for decision-makers to
make staged emergency decisions.

4.1. Building the scenario cascade network

On August 12, 2015, a devastating warehouse fire and explosion occurred in the hazardous goods warehouse of Ruihai Interna-
tional Logistics Company Limited (hereafter, Ruihai Company) in Tianjin Port, in the Tianjin Binhai New Area [77]. The accident
caused 165 deaths, severely damaged surrounding buildings, and affected areas as far as 13.3 km from the explosion. Under these cir-
cumstances, timely and effective emergency response decisions were particularly important.
As scenarios change or produce new scenarios when different factors intervene at different times, it is necessary to constantly ob-
serve and update the accident information. After the accident, with the change in the accident environment and the implementation
of emergency measures, the situation experiences new changes. Therefore, this study addresses the T2 period after the explosion at t2
and considers the scenario evolution process at this stage as the main research object to determine the emergency action direction at
T2. Fig. 4 exhibits the scenario cascade network diagram at T2, compiled according to interviews with event experiences, experts, and
official news materials of Xinhuanet.
There are seven nodes in this scenario cascade network, each representing an accident scenario. The red node S1 has no incoming
relationship and represents the large area combustion of hazardous chemicals in the warehouse, which is the original scenario in the
evolution process. The blue nodes represent cascading scenarios, including the occurrence of an explosion in Ruihai Company (S2), as
surrounding companies were damaged (S3) and roads were damaged (S4), surrounding communities were affected (S5), and subway
stations were damaged (S6). Green node S7 is the end node of the evolution process and represents the evolution result of ‘personal in-
jury or death’.

4.2. Scenario severity assessment

This section is based on the Baidu heat map to obtain the heat value of population aggregation in the area affected by the scenario
to analyse the severity of the scenario. Due to the limitations of the software, it is impossible to obtain the historical accident-affected
area heat map. Therefore, based on the background of the accident, this paper uses the existing data to extrapolate a heat map of the

Fig. 4. Scenario cascade network at T2.

7
L. Meng and W. Cheng International Journal of Disaster Risk Reduction 95 (2023) 103842

accident site area after the explosion for severity analysis. Although, after the accident, some areas changed, the regional street struc-
ture did not, as shown in Fig. 5. Considering that the explosion occurred at 23:35, to obtain the area heat map after the explosion for
the buildings that survived the accident, such as the Vanke residential complex and the Qihang residential complex, this study collects
the Baidu heat maps from 1 July 2022, to 5 July 2022, at 23:35 each day. These five heat maps are used as the data source for extrapo-
lating the population distribution heatmap at the time of the accident in the area. The Fuzzy Or model in ArcGIS software is used to
select the state with the highest heat value of population aggregation in the area, which is a pessimistic estimate of the distribution of
people in the study area. Therefore, the heat map of population distribution in the accident-affected area is obtained, as shown in Fig.
6, and then the population aggregation heat value of each scenario area is determined by Eq. (1). The same approach is used for other
areas, such as the Donghailu subway station.
The occupancy rate of the residential complex in this area was approximately 80% during the accident, including nearly 8000 resi-
dents in Vanke residential complex and 905 households with approximately 2200 people in Qihang residential complex. We substi-
tute the population density values represented by each colour in the literature [78] into Eq. (6) to estimate the number of affected

Fig. 5. Tianjin Port 8.12 fire and explosion accident area.

Fig. 6. Heatmap of the study area at T2.

8
L. Meng and W. Cheng International Journal of Disaster Risk Reduction 95 (2023) 103842

people in this area. The number of people in the area was 125,91. As the occupancy rate was 80% at the time, the final number of oc-
cupants was 10,073, with an error of 1.2% from the historical setting. Therefore, the extrapolated heat map can be used to describe
the population distribution at the time of the accident:
m
(6)
∑
P= Cn 𝛿dn
n=1

where P is the number of people in the study area, δ is the unit pixel area (the actual area represented by each pixel in this study is ap-
proximately 1.26 m2), and dn is the population density corresponding to the n th colour.
For the areas affected by the accident that no longer exist, this study judges their urban development land classifications by the
secondary indexes in the code for classification of urban land use and planning standards of development land. It selects the same de-
velopment land categories in the adjacent areas as the study object for heatmap extraction, used to determine the heat value informa-
tion of the scenario-affected area. For example, Binhai Park was a logistics and warehouse company before the explosion. This study
selects the heatmap of the nearby logistics and warehouse companies from 1 July 2022, to 5 July 2022, at 23:35 every day. The same
method is adopted to calculate the regional heat value. Then, we infer the heat value of the area of Ruihai Company at the time of the
explosion based on the area ratio. Table 1 shows the population heat value of the scenario-affected area.

4.3. Scenario risk assessment

We calculate the risk of each scenario using Eq. (4), combined with the Tianjin explosion scenario evolution network. Table 2 re-
ports the results. The risk of some cascade scenarios is even higher than that of the original scenario. S2 is the most dangerous in the
cascade process in T2, and the presence of this scenario has a vast impact on the overall accident evolution. This result indicates that
S2 plays a crucial role in escalating the cascading effect, suggesting that the explosion scenario exerts a greater influence on all other
scenarios it can trigger and should therefore be considered in determining the emergency response objectives at this stage.

4.4. Scenario emergency urgency assessment

According to the scenarios' severity and risk degree, we calculate each scenario's emergency urgency using Eq. (5), as shown in
Table 3. In this accident, after scenario S1 occurred at time t1, the emergency response did not effectively control the scenario. Hence,
scenario S1 lasted until t2, and the cascading effect of the accident scenarios occurred. The development and evolution of scenarios
made the emergency task change. Hence, we need to determine the direction of the emergency action in T2 and make new emergency
decisions. Table 3 shows that the emergency urgency of the incident scenario S5 (surrounding communities were affected) is the high-

Table 1
Scenario influenced regional heat values.

Scenario Heat value Normalised value

S1 52 0.0002
S2 1188 0.0047
S3 32219 0.1265
S4 11987 0.0471
S5 204982 0.8051
S6 4175 0.0164

Table 2
Estimates of the risk of each scenario.

Scenario Risk value

S1 10.6
S2 14.4
S3 2
S4 2
S5 2
S6 2

Table 3
Estimates of the emergency urgency of each scenario.

Scenario Emergency urgent value

S1 0.002
S2 0.067
S3 0.253
S4 0.094
S5 1.610
S6 0.033

9
L. Meng and W. Cheng International Journal of Disaster Risk Reduction 95 (2023) 103842

est in time slice T2. Therefore, during the emergency process in this period, priority should be given to implementing the correspond-
ing emergency measures for this area, such as immediately organizing personnel to rescue injured people and evacuating other people
to safe areas. In addition, the calculation results indicate that several accident scenarios with extreme urgency, such as ‘surrounding
companies were damaged’ and ‘roads were damaged’, are cascading scenarios caused by explosions, which is consistent with the re-
sults we obtained from structurally analysing the hazard level of the scenarios. Therefore, the decision-makers at the T2 stage should
pay more attention to rescuing and evacuating the people in the surrounding area affected by the explosion to avoid damage caused
by another explosion.

4.5. Analysis results

Combined with the analysis of the actual situation of the explosion accident in Tianjin, the main task of the emergency personnel
after the fire in the arrival area at t1 was to prevent the fire from expanding. However, due to the obstructed fire engine access and the
unknown nature of the materials, the fire was not extinguished promptly and appropriately before the explosions [77]. The high-
temperature environment caused the explosion of the surrounding hazardous chemicals at t2, resulting in many casualties of rescuers.
In order to ensure the safety of rescuers and personnel in the surrounding area after the emergence of the cascade scenario at T2, the
importance of fire-extinguishing tasks was reduced, and the emergency decision-makers changed the focus of emergency response
from fire extinguishing to emergency rescue and evacuation of personnel in the surrounding area. This demonstrates that the deci-
sion-making objectives of the T2 stage proposed in this paper by analysing the emergency urgency of each scenario are in line with the
actual situation. We can see from the accident case that in the process of urban EDM, decision-makers often face not a single hazard,
so we need to consider the cascading effects to determine the key scenarios that need to be dealt with urgently in EDM. By analysing
the emergency urgency of each scenario, this study can assist decision-makers in studying and judging the cascading situation of
emergencies at each stage, to determine the appropriate emergency entry point. In summary, the framework proposed in this study
recognises the time-based dynamics of the EDM process and considers the cascading effects of emergencies.

5. Conclusions
In the context of accelerated urbanisation, the cascading effect of emergencies has become a relatively common phenomenon,
which increases the difficulty of EDM. When making emergency decisions, decision-makers are faced with constantly changing sce-
narios [79]. To avoid the confusion in the emergency decision process caused by the cascading process of emergencies, this study pro-
poses a dynamic EDM model considering the cascading effect of emergencies from the perspective of time dynamics. First, the emer-
gency development process is cut into different time slices in a time-dynamic manner, to clarify the cascading process of scenarios in
each time slice. Then, in view of the different emergency urgency of each scenario area in time slice, this study obtains the distribu-
tion of the affected people in different scenarios through the heat map to analyse the severity of the areas affected by each scenario.
Furthermore, this paper proposes a risk measuring method that analyse the risk degree for different scenarios under the perspective of
cascade structure. This measurement method is verified to be more in line with the actual situation of scenario cascade than the tradi-
tional measure based on the out-degree of nodes. Finally, the two indicators are combined to determine the emergency urgency of dif-
ferent scenarios, to provide a reference for decision-makers in formulating emergency decisions.
This study provides a new attempt to study dynamic EDM. The proposed method fully considers the characteristics of urban emer-
gency cascades in the decision-making process, and uses a case study to explain how to use this method to determine the emergency
urgency of each scenario in the cascade process. At the same time, it also demonstrates that the decision-making should be dynami-
cally adjusted according to the situation at each stage. Therefore, considering the cascading effect, the multi-stage dynamic EDM
aligns with the actual situation. This model can help decision-makers clarify emergency goals in complex cascade situations. Since it
is a relatively new exploration to incorporate the analysis of cascading effects into the dynamic EDM process, the model constructed
in this paper is preliminary, so there are inevitably research deficiencies that need further exploration and improvement in follow-up
studies. First, in this study, as it is currently impossible to track an ongoing case in real-time for analysis, this paper uses a historical
case to illustrate the application of the model. Therefore, it is the next step of this research to further verify and improve the model
through real-time cases of different scales in more urban areas. In addition, this paper divides the emergency development process
into time slices to determine the emergency operations plans in different periods but lacks compatibility between emergency plans in
multiple stages. Subsequently, an additional next step in the research is determining how to connect the emergency operations plans
of different stages effectively.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to
influence the work reported in this paper.

Data availability
Data will be made available on request.

10
L. Meng and W. Cheng International Journal of Disaster Risk Reduction 95 (2023) 103842

References
[1] X. Chen, M.-P. Kwan, Q. Li, J. Chen, A model for evacuation risk assessment with consideration of pre- and post-disaster factors, Comput. Environ. Urban Syst.
36 (3) (2012) 207–217, https://doi.org/10.1016/j.compenvurbsys.2011.11.002.
[2] B. Zhao, Qingdao pipeline explosion: introductions and reflections, Nat. Hazards 74 (2) (2014) 1299–1305, https://doi.org/10.1007/s11069-014-1232-2.
[3] B. Zhao, Facts and lessons related to the explosion accident in Tianjin Port, China, Nat. Hazards 84 (1) (2016) 707–713, https://doi.org/10.1007/s11069-016-
2403-0.
[4] T. Ohnishi, The disaster at Japan’s fukushima-daiichi nuclear power plant after the march 11, 2011 earthquake and tsunami, and the resulting spread of
radioisotope Contamination1, Radiat. Res. 177 (1) (2011) 1–14, https://doi.org/10.1667/RR2830.1.
[5] W. Su, L. Chen, X. Gao, Emergency decision making: a literature review and future directions, Sustainability 14 (17) (2022) https://doi.org/10.3390/
su141710925, Article 17.
[6] M. Su, Study on the method of selecting and dynamic adjusting alternatives in multi-stage emergency decision, Mastre’s dissertation (2011) (Northeastern
University).
[7] K. Bohm, L. Kurland, The accuracy of medical dispatch—a systematic review, Scand. J. Trauma Resuscitation Emerg. Med. 26 (1) (2018) 94, https://doi.org/
10.1186/s13049-018-0528-8.
[8] J. Xie, J. Ding, M. Wang, Y. Cheng, Construction of emergency command and decision system for military civilian medical rescue based on gis, J. Phys. Conf.
1533 (2) (2020) 022124, https://doi.org/10.1088/1742-6596/1533/2/022124.
[9] S.K. Keneally, M.J. Robbins, B.J. Lunday, A markov decision process model for the optimal dispatch of military medical evacuation assets, Health Care Manag.
Sci. 19 (2) (2016) 111–129, https://doi.org/10.1007/s10729-014-9297-8.
[10] C. Qi, J. Fang, L. Sun, Implementation of emergency logistics distribution decision support system based on GIS, Cluster Comput. 22 (4) (2019) 8859–8867,
https://doi.org/10.1007/s10586-018-1983-8.
[11] O.M. Araz, M. Jehn, Improving public health emergency preparedness through enhanced decision-making environments: a simulation and survey based
evaluation, Technol. Forecast. Soc. Change 80 (9) (2013) 1775–1781, https://doi.org/10.1016/j.techfore.2012.09.018.
[12] T. Xie, M. Ni, Z. Zhang, Y. Wei, Parallel simulation decision-making method for a response to unconventional public health emergencies based on the
scenario–response paradigm and discrete event system theory, Disaster Med. Public Health Prep. 13 (5–6) (2019) 1017–1027, https://doi.org/10.1017/
dmp.2019.30.
[13] M.-Y. Li, P.-P. Cao, Extended TODIM method for multi-attribute risk decision making problems in emergency response, Comput. Ind. Eng. 135 (2019)
1286–1293, https://doi.org/10.1016/j.cie.2018.06.027.
[14] A. Mukhopadhyay, Z. Wang, Y. Vorobeychik, A decision theoretic framework for emergency responder dispatch, Proceedings of the 17th International
Conference on Autonomous Agents and MultiAgent Systems (2018) 588–596.
[15] A. Mukhopadhyay, G. Pettet, S.M. Vazirizade, D. Lu, A. Jaimes, S.E. Said, H. Baroud, Y. Vorobeychik, M. Kochenderfer, A. Dubey, A review of incident
prediction, resource allocation, and dispatch models for emergency management, Accid. Anal. Prev. 165 (2022) 106501, https://doi.org/10.1016/
j.aap.2021.106501.
[16] P.T. Savolainen, F.L. Mannering, D. Lord, M.A. Quddus, The statistical analysis of highway crash-injury severities: a review and assessment of methodological
alternatives, Accid. Anal. Prev. 43 (5) (2011) 1666–1676, https://doi.org/10.1016/j.aap.2011.03.025.
[17] Y. Liu, A multistage dynamic emergency decision-making method considering the satisfaction under uncertainty information, J. Adv. Transport. 2021 (2021)
e5535925, https://doi.org/10.1155/2021/5535925.
[18] N. Kapucu, M. Van Wart, Making matters worse: an anatomy of leadership failures in managing catastrophic events, Adm. Soc. 40 (7) (2008) 711–740, https://
doi.org/10.1177/0095399708323143.
[19] X. Xu, B. Pan, Y. Yang, Large-group risk dynamic emergency decision method based on the dual influence of preference transfer and risk preference, Soft
Comput. 22 (22) (2018) 7479–7490, https://doi.org/10.1007/s00500-018-3387-3.
[20] C. Fang, L. Zhu, Z. Liu, Y. Li, Y. Huang, Research on rail transit dispatcher emergency decision support based on case similarity matching, Urban Rail Transit 8
(2) (2022) 146–156, https://doi.org/10.1007/s40864-022-00170-1.
[21] Y. Wang, Y. Liang, H. Sun, A regret theory-based decision-making method for urban rail transit in emergency response of rainstorm disaster, J. Adv. Transport.
2020 (2020) e3235429, https://doi.org/10.1155/2020/3235429.
[22] J. Zheng, Y. Wang, K. Zhang, J. Liang, A dynamic emergency decision-making method based on group decision making with uncertainty information,
International Journal of Disaster Risk Science 11 (5) (2020) 667–679, https://doi.org/10.1007/s13753-020-00308-4.
[23] M. Wan, C. Ye, D. Peng, Multi-period dynamic multi-objective emergency material distribution model under uncertain demand, Eng. Appl. Artif. Intell. 117
(2023) 105530, https://doi.org/10.1016/j.engappai.2022.105530.
[24] F. Yu, B. Fan, X. Li, Improving emergency preparedness to cascading disasters: a case-driven risk ontology modelling, J. Contingencies Crisis Manag. 28 (3)
(2020) 194–214, https://doi.org/10.1111/1468-5973.12314.
[25] G. Zuccaro, D. De Gregorio, M.F. Leone, Theoretical model for cascading effects analyses, Int. J. Disaster Risk Reduc. 30 (2018) 199–215, https://doi.org/
10.1016/j.ijdrr.2018.04.019.
[26] S. Li, Y. Cui, M. Liu, H. He, S. Ravan, Integrating global open geo-information for major disaster assessment: a case study of the Myanmar flood, ISPRS Int. J.
Geo-Inf. 6 (7) (2017) https://doi.org/10.3390/ijgi6070201, Article 7.
[27] X. Ji, W. Weng, W. Fan, Cellular automata-based systematic risk analysis approach for emergency response, Risk Anal. 28 (5) (2008) 1247–1260, https://
doi.org/10.1111/j.1539-6924.2008.01104.x.
[28] C. Liu, J. Qian, D. Guo, Y. Liu, A spatio-temporal scenario model for emergency decision, GeoInformatica 22 (2) (2018) 411–433, https://doi.org/10.1007/
s10707-017-0313-2.
[29] X. Chaoxu, N. Gaozhong, F. Xiwei, Z. Junxue, P. Xiaoke, Research on the application of mobile phone location signal data in earthquake emergency work: a
case study of Jiuzhaigou earthquake, PLoS One 14 (4) (2019) e0215361, https://doi.org/10.1371/journal.pone.0215361.
[30] Z. Li, C. Wang, C.T. Emrich, D. Guo, A novel approach to leveraging social media for rapid flood mapping: a case study of the 2015 South Carolina floods,
Cartogr. Geogr. Inf. Sci. 45 (2) (2018) 97–110, https://doi.org/10.1080/15230406.2016.1271356.
[31] Z. Xing, X. Zhang, X. Zan, C. Xiao, B. Li, K. Han, Z. Liu, J. Liu, Crowdsourced social media and mobile phone signaling data for disaster impact assessment: a
case study of the 8.8 Jiuzhaigou earthquake, Int. J. Disaster Risk Reduc. 58 (2021) 102200, https://doi.org/10.1016/j.ijdrr.2021.102200.
[32] T. Yabe, N.K.W. Jones, P.S.C. Rao, M.C. Gonzalez, S.V. Ukkusuri, Mobile phone location data for disasters: a review from natural hazards and epidemics,
Comput. Environ. Urban Syst. 94 (2022) 101777, https://doi.org/10.1016/j.compenvurbsys.2022.101777.
[33] Y. Wang, L. Zhang, Research on scenario deduction of unconventional emergency based on knowledge-unit, 2015 Ieee/Wic/Acm International Conference on
Web Intelligence and Intelligent Agent Technology (Wi-Iat) 3 (2015) 144–147, https://doi.org/10.1109/WI-IAT.2015.177.
[34] Z. Hao, Z. Xu, H. Zhao, H. Fujita, A dynamic weight determination approach based on the intuitionistic fuzzy bayesian network and its application to
emergency decision making, IEEE Trans. Fuzzy Syst. 26 (4) (2018) 1893–1907, https://doi.org/10.1109/TFUZZ.2017.2755001.
[35] L. Shi, X. Yang, J. Li, J. Wu, H. Sun, Scenario construction and deduction for railway emergency response decision-making based on network models, Inf. Sci.
588 (2022) 331–349, https://doi.org/10.1016/j.ins.2021.12.071.
[36] C. Chen, Y. Yang, M. Wang, X. Zhang, Characterization and evolution of emergency scenarios using hybrid Petri net, Process Saf. Environ. Protect. 114 (2018)
133–142, https://doi.org/10.1016/j.psep.2017.12.016.
[37] O. Nyvlt, S. Haugen, L. Ferkl, Complex accident scenarios modelled and analysed by Stochastic Petri Nets, Reliab. Eng. Syst. Saf. 142 (2015) 539–555, https://
doi.org/10.1016/j.ress.2015.06.015.
[38] J. Li, C. Chen, Modeling the dynamics of disaster evolution along causality networks with cycle chains, Phys. Stat. Mech. Appl. 401 (2014) 251–264, https://
doi.org/10.1016/j.physa.2014.01.052.
[39] P. Tang, Q. Xia, Y. Wang, Addressing cascading effects of earthquakes in urban areas from network perspective to improve disaster mitigation, Int. J. Disaster

11
L. Meng and W. Cheng International Journal of Disaster Risk Reduction 95 (2023) 103842

Risk Reduc. 35 (2019) 101065, https://doi.org/10.1016/j.ijdrr.2019.101065.

[40] H. Chen, L. Zhang, L. Ran, Vulnerability modeling and assessment in urban transit systems considering disaster chains: a weighted complex network approach,
Int. J. Disaster Risk Reduc. 54 (2021) 102033, https://doi.org/10.1016/j.ijdrr.2020.102033.
[41] Z. Hu, G. Wu, H. Wu, L. Zhang, Cross-sectoral preparedness and mitigation for networked typhoon disasters with cascading effects, Urban Clim. 42 (2022)
101140, https://doi.org/10.1016/j.uclim.2022.101140.
[42] L. Du, Y. Feng, L. Tang, W. Lu, W. Kang, Time dynamics of emergency response network for hazardous chemical accidents: a case study in China, J. Clean. Prod.
248 (2020) 119239, https://doi.org/10.1016/j.jclepro.2019.119239.
[43] Y. Sheng, L.-X. Sun, Y.-M. Wang, Emergency scenario evolution and extraction Method of key elements, Journal of Safety Science and Technology (1) (2015)
17–21.
[44] H. Shen, W.W. Huang, J. Zhao, T. Jin, A sequential group decision process for emergency response, in: 2008 IEEE International Conference on Systems, Man
and Cybernetics, 2008, pp. 867–871, https://doi.org/10.1109/ICSMC.2008.4811389.
[45] Y. Wang, Y. Liang, H. Sun, Y. Yang, Decision-making for fire emergency of urban rail transit based on prospect theory, Math. Probl Eng. 2021 (2021)
e3414589, https://doi.org/10.1155/2021/3414589.
[46] J. Ding, J. Cai, G. Guo, C. Chen, An emergency decision-making method for urban rainstorm water-logging: a China study, Sustainability 10 (10) (2018)
https://doi.org/10.3390/su10103453, Article 10.
[47] L.-X. Hou, L.-X. Mao, H.-C. Liu, L. Zhang, Decades on emergency decision-making: a bibliometric analysis and literature review, Complex & Intelligent Systems
7 (6) (2021) 2819–2832, https://doi.org/10.1007/s40747-021-00451-5.
[48] H. Jiang, J. Huang, The study on the issues of scenario evolvement in real-time decision making of infrequent fatal emergencies, Journal of Huazhong
University of Science and Technology(Social Science Edition) (1) (2009) 104–108, https://doi.org/10.19648/j.cnki.jhustss1980.2009.01.022.
[49] Y. Ren, G. Tian, Emergency scheduling for forest fires subject to limited rescue team resources and priority disaster areas, IEEJ Trans. Electr. Electron. Eng. 11
(6) (2016) 753–759, https://doi.org/10.1002/tee.22300.
[50] D. Chen, Y. Fang, Urgency evaluation of emergency demand in emergency logistics operation based on FAHP, IEEE Conference Anthology (2013) 1–4, https://
doi.org/10.1109/ANTHOLOGY.2013.6785075.
[51] L. Huang, P. Shi, H. Zhu, An integrated urgency evaluation approach of relief demands for disasters based on social media data, Int. J. Disaster Risk Reduc. 80
(2022) 103208, https://doi.org/10.1016/j.ijdrr.2022.103208.
[52] Q.-L. Shu, Resource allocation and scheduling for unconventional emergency based on “scenario-response”, University of Science and Technology of China
(2012) PHD thesis.
[53] M.-P. Kwan, J. Lee, Emergency response after 9/11: the potential of real-time 3D GIS for quick emergency response in micro-spatial environments, Comput.
Environ. Urban Syst. 29 (2) (2005) 93–113, https://doi.org/10.1016/j.compenvurbsys.2003.08.002.
[54] Y. He, Grade of traffic emergency of coupling of multiple accidents, Huazhong University of Science and Technology (2013) PHD thesis.
[55] X.-P. Jiang, Z.-S. Xu, Research on grading of highway tunnel accidents and the emergency response, Journal of Catastrophology (4) (2008) 91–95.
[56] B. Wei, G. Nie, G. Su, L. Sun, X. Bai, W. Qi, Risk assessment of people trapped in earthquake based on km grid: a case study of the 2014 Ludian earthquake,
China, Geomatics, Nat. Hazards Risk 8 (2) (2017) 1289–1305, https://doi.org/10.1080/19475705.2017.1318795.
[57] G. Liu, E. Dai, X. Xu, W. Wu, A. Xiang, Quantitative assessment of regional debris-flow risk: a case study in southwest China, Sustainability 10 (7) (2018)
https://doi.org/10.3390/su10072223, Article 7.
[58] H.-M. Lyu, W.-J. Sun, S.-L. Shen, A. Arulrajah, Flood risk assessment in metro systems of mega-cities using a GIS-based modeling approach, Sci. Total Environ.
626 (2018) 1012–1025, https://doi.org/10.1016/j.scitotenv.2018.01.138.
[59] H. Qi, W. Tian, F. Zhao, Risk assessment of snow disaster for trunk highway transportation in Shaanxi, China, Nat. Hazards 85 (1) (2017) 523–536, https://
doi.org/10.1007/s11069-016-2584-6.
[60] W. Gong, J. Jiang, L. Yang, Dynamic risk assessment of compound hazards based on VFS–IEM–IDM: a case study of typhoon–rainstorm hazards in Shenzhen,
China, Nat. Hazards Earth Syst. Sci. 22 (10) (2022) 3271–3283, https://doi.org/10.5194/nhess-22-3271-2022.
[61] C. Huang, Y. Huang, An information diffusion technique to assess integrated hazard risks, Environ. Res. 161 (2018) 104–113, https://doi.org/10.1016/
j.envres.2017.10.037.
[62] Y. Cai, M.W. Golay, A dynamic Bayesian network-based emergency decision-making framework highlighting emergency propagations: illustrated using the
Fukushima nuclear accidents and the Covid-19 pandemic, Risk Anal. (2022), 13928 https://doi.org/10.1111/risa.13928, risa.
[63] L. Palen, S. Vieweg, S.B. Liu, A.L. Hughes, Crisis in a networked world: features of computer-mediated communication in the april 16, 2007, Virginia tech
event, Soc. Sci. Comput. Rev. 27 (4) (2009) 467–480, https://doi.org/10.1177/0894439309332302.
[64] J.P. de Albuquerque, B. Herfort, A. Brenning, A. Zipf, A geographic approach for combining social media and authoritative data towards identifying useful
information for disaster management, Int. J. Geogr. Inf. Sci. 29 (4) (2015) 667–689, https://doi.org/10.1080/13658816.2014.996567.
[65] Z. Zhang, Y. Xiao, X. Luo, M. Zhou, Urban human activity density spatiotemporal variations and the relationship with geographical factors: an exploratory
Baidu heatmaps-based analysis of Wuhan, China, Growth Change 51 (1) (2020) 505–529, https://doi.org/10.1111/grow.12341.
[66] S. He, H. Dang, M. Zhang, Research on dynamic changes of urban square space in spatial and temporal based on Baidu thermal diagram A case study on the
wuyi square of changsha, in: 2017 Ieee 2nd International Conference on Big Data Analysis (Icbda), 2017, pp. 517–522. https://www.webofscience.com/wos/
woscc/summary/c18bc034-bdfa-4004-9ae5-eacf270ea66b-5810cca6/relevance/1.
[67] W. Wang, L. Liu, Y. Yang, Spatial matching analysis and development strategies of county night-time economy: a case of anning county, yunnan province,
Sustainability 14 (9) (2022) 4891, https://doi.org/10.3390/su14094891.
[68] X. Zhao, Y. Zhou, W. Chen, X. Li, X. Li, D. Li, Mapping hourly population dynamics using remotely sensed and geospatial data: a case study in Beijing, China,
GIScience Remote Sens. 58 (5) (2021) 717–732, https://doi.org/10.1080/15481603.2021.1935128.
[69] H. Si, Research and implementation of Chinese automatic word segmentation system based on complex network features, Wireless Commun. Mobile Comput.
2022 (2022) e3812643, https://doi.org/10.1155/2022/3812643.
[70] J. Liu, Q. Xiong, W. Shi, X. Shi, K. Wang, Evaluating the importance of nodes in complex networks, Phys. Stat. Mech. Appl. 452 (2016) 209–219, https://
doi.org/10.1016/j.physa.2016.02.049.
[71] C. Gao, D. Wei, Y. Hu, S. Mahadevan, Y. Deng, A modified evidential methodology of identifying influential nodes in weighted networks, Phys. Stat. Mech.
Appl. 392 (21) (2013) 5490–5500, https://doi.org/10.1016/j.physa.2013.06.059.
[72] P. Tang, S. Lai, A framework for managing public security risks with complex interactions in cities and its application evidenced from Shenzhen City in China,
Cities 95 (2019) 102390, https://doi.org/10.1016/j.cities.2019.102390.
[73] O. Ugurlu, Comparative analysis of centrality measures for identifying critical nodes in complex networks, Journal of Computational Science 62 (2022)
101738, https://doi.org/10.1016/j.jocs.2022.101738.
[74] Q. Zeng, Y. Liu, M. Tang, J. Gong, Identifying super-spreaders in information–epidemic coevolving dynamics on multiplex networks, Knowl. Base Syst. 229
(2021) 107365, https://doi.org/10.1016/j.knosys.2021.107365.
[75] M. Girvan, M.E.J. Newman, Community structure in social and biological networks, Proceedings of the National Academy of Sciences of the United States of
America 99 (12) (2002) 7821–7826, https://doi.org/10.1073/pnas.122653799.
[76] G. Pescaroli, M. Nones, L. Galbusera, D. Alexander, Understanding and mitigating cascading crises in the global interconnected system, Int. J. Disaster Risk
Reduc. 30 (2018) 159–163, https://doi.org/10.1016/j.ijdrr.2018.07.004.
[77] G. Fu, J. Wang, M. Yan, Anatomy of Tianjin Port fire and explosion: process and causes, Process Saf. Prog. 35 (3) (2016) 216–220, https://doi.org/10.1002/
prs.11837.
[78] Y. Zhang, Analysis on the Space of the Zijingang Campus of Zhejiang University by the Big Data Method and the Space Syntax. Mastre’s Dissertation, Zhejiang
University, 2016.
[79] J.B. Cuartas, T. Frazier, E. Wood, The application of cascading consequences for emergency management operations, Nat. Hazards 108 (3) (2021) 2919–2938,
https://doi.org/10.1007/s11069-021-04807-5.

12