Cities 81 (2018) 91–100

Contents lists available at ScienceDirect

Cities
journal homepage: www.elsevier.com/locate/cities

An Agent-Based Modeling approach for sustainable urban planning from T

land use and public transit perspectives
⁎
Hamid Motieyan , Mohammad Saadi Mesgari
Department of Geo-Spatial Information System, Faculty of Geodesy and Geomatics, K.N. Toosi University of Technology, Tehran, Iran

A R T I C LE I N FO A B S T R A C T

Keywords: Urbanism is a challenging topic in the world, which has resulted in several land-use and transportation issues in
Urban planning urban environments. To address these issues, urban planners follow integrated planning approaches that are
Sustainable development more compatible with sustainable development objectives. Transit-Oriented Development (TOD) is widely re-
Transit-Oriented Development cognized as one of the most feasible and comprehensive sustainable planning approaches. In this research, a
Public transit infrastructure
three-step TOD-based method was developed for sustainable urban planning in the central region of Tehran.
Geospatial Information System
Agent-Based Modeling
First, a measurable index was developed to assess public transit infrastructure (PTI) and TOD levels in the study
area. At the second step, which was the focus of the research, an Agent-Based Modeling (ABM) approach was
used to make a balance between TOD and PTI levels. ABM is a bottom-up approach that can solve spatial
problems and can be integrated with top-down policies and spatial analysis tools. Finally, the performance of the
model was evaluated by conducting several statistical, visual and empirical assessments using Tehran munici-
pality's reference data. These assessments confirmed the efficiency and feasibility of the model.

1. Introduction lead to the creation of lively, healthy, and pedestrian-friendly com-

Nowadays, a majority of the world's population lives in cities
(Chawla, 2016); therefore, there is a substantial need for the effective
management of limited resources in cities. A comprehensive planning
approach plays an underpinning role in the efficient and smart man-
agement of limited resources in urban environments. Without having
such a comprehensive urban planning strategy, cities will face with
critical issues such as urban sprawl, inappropriate distribution of fa-
cilities, environmental pollutions, traffic congestion, and insufficient
urban infrastructures such as public transit system (Ewing, Tian, Lyons,
& Terzano, 2017; Shirzadi Babakan, Alimohammadi, & Taleai, 2015;
Wey, 2015; Wey, Zhang, & Chang, 2016; Widyahari & Indradjati,
2015). To tackle these issues, the urban planning strategy should cover
both the land use and transportation parts. These parts have a mutual
relationship; therefore, many researchers have developed Integrated
Land-Use and Transportation (ILUT) models (Shirzadi Babakan &
Alimohammadi, 2016; Waddell, Ulfarsson, Franklin, & Lobb, 2007;
Wegener, 2011; Yim, Wong, Chen, Wong, & Lam, 2011).
A successful ILUT model from the perspective of developing public
transit system is Transit-Oriented Development (TOD) model. This
model encourages non-motorized transportation modes of walking and
cycling for short trips and public transport modes including bus and
subway for longer trips. Thus, the implementation of this model will
munities and neighborhoods (Binglei & Chuan, 2013; Cervero, 2004;
Cervero, Guerra, & Al, 2017; Evans, Pratt, Stryker, & Kuzmyak, 2007;
Galelo, Ribeiro, & Martinez, 2014; Singh, Fard, Zuidgeest, Brussel, &
van Maarseveen, 2014; Wey, 2015; Wey et al., 2016).
In recent years, different definitions of TOD have been proposed in
the literature (Nasri & Zhang, 2014). Nevertheless, there are some
common factors in most of these definitions such as developing mixed-
use area with high density and diversity, and being convenient for
walking and cycling in 2000 ft around the public transit stations (Belzer
& Autler, 2002; Calthorpe, 1993; Cervero, Ferrell, & Murphy, 2002;
Kwon, 2015; Ratner & Goetz, 2013; Wey, 2015; Wey et al., 2016). In
addition, researchers have adopted the TOD concept in the urban
planning. The TOD planning approach covers all the areas, and not just
the areas around the public transit stations (Cervero, 2013; Cervero &
Dai, 2014; Evans et al., 2007; Motieyan & Mesgari, 2017a, 2017b; Singh
et al., 2014). In this study, we developed a novel agent-based modeling
approach based on the concept of TOD planning to manage land-use
activities as well as the public transit system. This planning requires a
robust method that can handle top-down and bottom-up procedures.
1.1. Literature review
ILUT models have been widely used in various research and

⁎
Corresponding author.
E-mail addresses: H.motieyan@gmail.com, H.motieyan@sina.kntu.ac.ir (H. Motieyan).

https://doi.org/10.1016/j.cities.2018.03.018
Received 18 August 2017; Received in revised form 26 March 2018; Accepted 28 March 2018
Available online 03 April 2018
0264-2751/ © 2018 Elsevier Ltd. All rights reserved.
H. Motieyan, M.S. Mesgari
development investigations since the 1960s. Earlier versions were not
successful because of their theoretical and technical limitations.
However, their recent versions have been significantly improved by
using disaggregated micro-simulation methods (Babakan & Taleai,
2015; Shirzadi Babakan & Alimohammadi, 2016; Wegener, 2011). One
of the most practical and comprehensive ILUT models is UrbanSim
(Waddell et al., 2003). The focus of this model is on tracking the re-
sidential location of individual households, distributing employments,
and identifying travel patterns according to households' socioeconomic
attributes, environmental characteristics, and accessibility levels
(Waddell et al., 2007). However, the UrbanSim model has the following
drawbacks: 1) Allocating a household or a job in a location depends on
many attributes that the model cannot certainly consider; 2) It cannot
create the sustainable lively and pedestrian-friendly neighborhoods; 3)
The recent concerns of urban planning, including urban sprawl and
smart growth, are not considered; 4) This model has some technical
problems in terms of huge data need, computing time, and stochastic
variation. These limitations have persuaded researchers to build their
models regarding conceptual, spatial, and temporal requirements
(Kakaraparthi & Kockelman, 2010; Wegener, 2011). On the other hand,
TOD planning is a sustainable ILUT planning approach that considers
many factors such as smart growth, urban sprawl, land-use planning,
public transport planning, attracting private investment, optimizing
landscape and urban design, and developing all the areas without
considering who wants to live there. In TOD planning, these factors are
considered to ensure sustainable urban planning and social justice in
cities.
In many studies, the TOD concept has been employed to select the
best scenarios for enhancing public transit services (Wey, 2015; Wey
et al., 2016). However, there are limited studies investigating the TOD
planning throughout an entire urban area using a spatial index
(Motieyan & Mesgari, 2017a, 2017b; Singh et al., 2014). Singh et al.
(2014), first developed a spatial TOD index to evaluate TOD levels in
the entire urban area; then they used a clustering method to determine
areas with high TOD levels for developing public transit stations. Al-
though their study was a novel contribution to the TOD planning, their
proposed approach had the following weaknesses: 1) They did not
consider comprehensive and efficient indicators to develop a TOD
index. 2) An effective aggregation method was not used to aggregate
indicators and criteria. 3) They did not suggest any solution for im-
proving TOD levels in the region, especially in the areas with potentials
for developing public transit services. These weaknesses can be ad-
dressed by several ameliorative actions such as increasing density, di-
versity, and accessibility. These actions have different consequences
and thereby it is essential to develop a method to assess their con-
sequences and to select the optimum actions.
1.2. Research contribution and objectives
The main contribution of this study is the development of a novel
agent-based method that helps in making ameliorative decisions with
regard to the TOD level, public transit infrastructure (PTI) level, and
upstream policies. The proposed method uses both bottom-up and top-
down procedures to consider all criteria and policies required for the
amelioration. According to the authors' knowledge, such a compre-
hensive method for TOD planning has not yet been presented. This
method would potentially help urban planners to make smarter, better,
and more comprehensive decisions in an urban area.
In this study, we pursued several objectives. First, to develop a
comprehensive, efficient and practical method, it was necessary to ex-
tend the indicators considered in existing TOD models. In comparison
with Singh et al. (2014), we considered new indicators such as ad-
ministrative density, accessibility, street connectivity, street design,
population categories, household size, educational level, facility level,
trip generation, and route and service performance. Based on our study,
these indicators play important roles in urban planning. The second
92
Cities 81 (2018) 91–100
objective was to use an effective aggregation method to aggregate the
indicators, as the previous models (e.g., Singh et al. (2014)) did not use
an appropriate aggregation method. The third objective was to develop
an agent-based model for evaluating urban areas from different per-
spectives, and for making suitable decisions based on the evaluations
and upstream policies. Finally, to verify the viability of the proposed
method, it was essential to assess the consequences of the decisions
using some reference data.
2. Materials
2.1. Study area and data
In this study, the proposed method is implemented in the 7th mu-
nicipal region of Tehran, the capital of Iran. There are several reasons
for selecting this region as the case study: 1) This region is a high-
density area with different land uses; therefore, land use planning is
essential for the region; 2) Trip generation and trip attraction levels are
high in this region; thus, the public transportation plays a key role in
the region. Crossing all subway lines and two major BRT lines within
the region is a strong witness for this key role. The 7th region is divided
into five districts and twenty-two neighborhoods. Additionally, total
population and area of the region are around 300,000 individuals and
1536 ha, respectively. Fig. 1 shows the 7th region and its neighbor-
hoods.
In this study, to develop the proposed method, data sets with the
following descriptions are used.
1) Maps of regions, neighborhoods, and parcels of Tehran with their
characteristics such as area and land use; and street map with the
attributes of street type and pavement type
2) Demographic data at the neighborhood scale, such as educational
level, professional status, age, and household size.
3) Public transit map including the location of routes and stations with
their attributes such as capacity, average speed, and frequency of
services.
2.2. TOD and PTI concepts
TOD is a fundamental concept in urban planning and smart growth.
Therefore, different studies have investigated this concept from dif-
ferent points of view. Calthorpe (1993), a pioneer in this research area,
defined TOD as “mixed-use community within an average 2000-foot
walking distance of a transit stop and core commercial area”. He found
that TOD needs high-density areas with various residential, commer-
cial, administrative, and recreational land uses. Additionally, these
areas should provide an appropriate walkable environment and en-
courage people to make their trips by bicycle and public transit systems.
Adequate access to public transit services reduces traffic congestion,
facilitates peoples travels, and improves urban quality of life (Wei et al.,
2017; Zheng, Zhang, Sun, & Wang, 2017). In other words, insufficient
supply of public transit results in higher levels of traffic congestion,
economic loss, and environmental degradation. To improve the public
transit systems, it is necessary to model, monitor, and evaluate public
transport systems using effective approaches, such as the TOD planning
method (Niyonsenga, 2012).
2.3. Agent-based modeling
ABM is considered a method to describe and simulate a system
composed of real-world entities (Patel, Crooks, & Koizumi, 2018). In
ABM, the system is composed of discrete agents or individuals inter-
acting with each other and their environment (Shirzadi Babakan &
Alimohammadi, 2016). Each agent is an identifiable unit of a computer
program with autonomous and goal-directed behavior (Prudkov, 2010).
The interactions among agents provide aggregated attributes, which
H. Motieyan, M.S. Mesgari
Fig. 1. (a): The 7th municipal region of Tehran; (b) neighborhoods of the 7th region.
can facilitate decision-making procedures. These interactions are per-
formed at the bottom level by agents and create a structure at the top
level. As a bottom-up approach, ABM provides a framework for
studying and managing urban systems affected by various economic,
social, and environmental factors which are notoriously difficult to si-
mulate (Alghais & Pullar, n.d.; Babakan & Taleai, 2015; Ligmann-
Zielinska & Jankowski, 2010; Shirzadi Babakan et al., 2015; Shirzadi
Babakan & Alimohammadi, 2016; Torrens, 2000). However, according
to (O'Sullivan & Haklay, 2000), a purely bottom-up or top-down mod-
eling is not sufficient for a complex task as urban planning. An alter-
native way to address this deficiency is to use a combination of both
perspectives (Ligmann-Zielinska & Jankowski, 2010; Verburg &
Overmars, 2009; Ward, Murray, & Phinn, 2003).
In this research, we used ABM due to four reasons. First, each
neighborhood has its own characteristics and its own TOD and PTI le-
vels. Based on these differences, each neighborhood needs a different
approach to improve its levels. Second, there are some constraints in
the study area that should be satisfied with the collection of neigh-
borhoods. Therefore, neighborhoods not only compete with each other
to improve their levels, but also interact with each other to satisfy those
constraints. Obviously, some priorities need to be defined in these
competitions. Third, ABM has the capability to couple with top-down
methods and policies, which are useful for TOD planning. Finally, a
disaggregated method like ABM is required to consider experts
knowledge related to neighborhoods spatial conditions.
3. Methodology
The main purpose of the proposed method is to make a balance
between TOD and PTI levels by making optimum decisions based on the
constraints of the study area and its neighborhoods. This method en-
ables decision-makers to develop a region from the perspectives of land
use and public transit planning. For this purpose, 22 neighborhoods in
the study area are considered as agents. The agents have their own
attributes such as spatial location, area, population, parcels, streets, and
public transit data, which differentiate their TOD and PTI levels. First,
the agents compute a number of land-use and transportation indicators
93
Cities 81 (2018) 91–100
based on appropriate models in GIS (Geospatial Information System).
Then, these indicators are aggregated using the AHP method to calcu-
late the TOD and PTI levels of agents. The TOD and PTI levels along
with upstream policies determine the behaviors and actions of agents.
The agents directly interact with each other and adjust their behavior
and actions to satisfy the constraints of the region. Finally, these in-
teractions result in making optimum decision to balance the TOD and
PTI levels.
The agent-based model proposed in this study is based on three
main steps. In the first step, the model evaluates PTI level on the basis
of a number of criteria, including Density, Diversity, Design, and
Economic development. In addition, the model evaluates TOD level on
the basis of route and service performances. These criteria are measured
by calculating a comprehensive set of spatial indicators defined by
appropriate mathematical formulas. Additionally, GIS is used to ana-
lyze and calculate the spatial indicators. In the second step, the model
compares TOD and PTI levels and provides proper decisions for de-
veloping and balancing these levels. The decisions are based on char-
acteristics of neighborhoods and upstream policies. The upstream po-
licies are derived from the Tehran municipality's policies for the study
area. Finally, a validation process is performed at the third step. In this
step, outcomes of the decisions resulted from ABM are assessed by
conducting several statistical analyses, and by comparing them with
municipal decisions and the reference data in the study area. The fra-
mework of the proposed method is illustrated in Fig. 2.
3.1. First step: evaluating TOD and PTI levels in neighborhoods
In this research, the TOD and PTI levels are evaluated by agents at
the neighborhood scale. The TOD indicators considered in earlier stu-
dies are not comprehensive and efficient enough (Evans et al., 2007;
Polzin, Pendyala, & Navari, 2002; Singh et al., 2014; Wey, 2015).
Therefore, we complete these indicators by considering other important
indicators such as street connectivity and street design, defined on the
basis of experts' views and earlier studies in land use planning. In ad-
dition, we use PTI indicators proposed in (Niyonsenga, 2012). The list
of the criteria and indicators used in this study is presented in Table 1.
H. Motieyan, M.S. Mesgari Cities 81 (2018) 91–100

Population,
Parcels,
Streets,
Public transit
Sensors Evaluation Environment
data

Methods, GIS, Attributes TOD level PTI level

AHP

Comparison

Other agents’ Which feasible actions can be performed

decisions based on Agents actions and constraints

Utility
Which actions are optimum?
functions

Constraints Changing
and upstream Actuators density, land use,
policies streets, public
transit services

Agent

Fig. 2. The framework of the proposed methodology.

Table 1 Table 2
The criteria and indicators used for calculating TOD and PTI levels in neigh- List of agents' actions and constraints at the neighborhood scale.
borhoods.
Agents' actions Agents' constraints
Criteria Indicators
Adding residential land use Maximum addible area is 50,000 square
TOD level Density Residential density meters (about 500 parcels)
Commercial density Adding commercial land use Maximum addible area is 5000 square
Administrative density meters (about 100 parcels)
Diversity Entropy Index Adding administrative land use Maximum addible area is 5000 square
Design Accessibility meters
Street connectivity Improving land use mixed-ness Maximum improvement is 5%
Street design changing streets and their Maximum improvements is 10%
Land use mixed-ness characteristics
Economic development Population Changing facilities Maximum addible area is 5000 square
Number of employees meters
Percentage of young people Adding a new public transit
Average family members infrastructure
Educational level Changing public transit Maximum improvement is 25%
Facilities level characteristics
Trip generation
PTI level Route performance Service coverage
Number of stations ∑ Ai
accessibility to the service Di =
S (1)
interchangeability to other services
Service performance Passenger load
Frequency where Di is the residential, commercial, or administrative density, Ai is
Average speed and comfortableness the total area of the land use i, and S is the total area of all land uses.
Entropy: entropy is one of the most efficient diversity indicators.
This indicator provides the results between 0 (homogeneity, wherein all
All of these indicators are measured and calculated using GIS. In the land uses are of a single type) and 1 (heterogeneity, wherein the de-
rest of this section, the calculations of some indicators are explained in veloped area is evenly distributed among all land-use categories).
more detail. Entropy is calculated by Eq. (2) (Cervero & Kockelman, 1997).
Density: three indicators of residential, commercial, and adminis-
trative densities are computed by Eq. (1), for each neighborhood. n
Entropy = ∑ pi 1n pi /1n n
i=1 (2)

94
H. Motieyan, M.S. Mesgari Cities 81 (2018) 91–100

Table 3
The list of agents' actions and constraints at the regional scale.
Agents' actions Agents' constraints

Adding residential land use Maximum addible area in the region is 400,000 square meters (about 4000 parcels)
Adding commercial land use Maximum addible area in the region is 30,000 square meters (about 600 parcels)
Adding administrative land use Maximum addible area in the region is 30,000 square meters
Adding a new public transit infrastructure Adding a new subway station in a neighborhood requires that its neighbor also has the same need.
Adding a new BRT station in a neighborhood requires that its neighbor also has the same need and there is an appropriate street
connecting them.
The public station should be shared with the neighborhood, requiring the same public transit infrastructure.

Table 4 where pi is the proportion of the area assigned to the land use i, and n is
The weight of indicators used in the agent's utility function. the number of land use categories (Frenken, Van Oort, & Verburg,
Indicators Weights
2007).
Accessibility: Accessibility is defined as the potential for reaching
Residential density 0.102 spatially distributed opportunities such as employment, recreation, and
Commercial density 0.203 shopping. This indicator is computed by Eq. (3) (Babakan & Taleai,
Administrative density 0.169
Entropy Index 0.314
2015; Geurs & Van Wee, 2004; Páez, Scott, & Morency, 2012; Shirzadi
Accessibility 0.045 Babakan et al., 2015; Van Eck & de Jong, 1999).
Street connectivity 0.045
Street design 0.022 A ijm= ∑ Oj f (Cijm)
Land use mixed-ness 0.033 j (3)
Population 0.007
Number of employees 0.013 where Aijm is the accessibility at point i to the opportunity at point j
Percentage of young people 0.007 using transport mode m, Oj is the opportunity at point j, and ƒ(Cijm) is
Average household size 0.007
Educational level 0.007
the impedance or cost function to travel between i and j using mode m.
Facilities level 0.013 Street connectivity: one of the most effective street connectivity in-
Trip generation 0.013 dicators is the alpha index (α). The alpha index is estimated by the ratio
Transit service coverage 0.242 of the number of fundamental circuits to the maximum possible number
Number of stations 0.121
of circuits. The alpha index is achieved by Eq. (4) (Theall et al., 2009).
Accessibility to the transit service 0.182
Interchangeability to other transit services 0.151
(e − v + 1)
Passenger load 0.106 α=
Frequency of transit service 0.121 2v − 5 (4)
Average speed and comfortableness of the transit service 0.077
where e is the number of edges and v is the number of nodes in the
street network.
Land use mixed-ness: To improve TOD level in an area, some

Fig. 3. (a) OTOD level in neighborhoods (b): OPTI level in neighborhoods.

95
H. Motieyan, M.S. Mesgari Cities 81 (2018) 91–100

Fig. 4. Comparison between OTOD and OPTI levels in neighborhoods before decision making.

Table 5
The optimum decisions made by agents in neighborhoods.
Neighborhood no. Agents' decisions

1 Construction of a subway station

2 Developing old texture level, increasing administrative density and land-use diversity, constructing parks and facility centers
3 Development of the current transit fleet and its service performance
4 increasing administrative and commercial densities and land use diversity, developing design of streets and old texture, constructing parks and facility
centers
5 developing design of streets and old texture
6 Development of the current transit fleet and its service performance
7 Increasing residential, commercial, and administrative densities
8 Construction of a subway station
9 there is no decision
10 Construction of a subway station
11 Construction of a subway station
12 Development of the current transit fleet and its service performance
13 there is no decision
14 Development of the current transit fleet and its service performance
15 Construction of a subway station
16 there is no decision
17 Construction of a subway station and Development of the current transit fleet and its service performance
18 Construction of a subway station and Development of the current transit fleet and its service performance
19 Construction of a subway station and a BRT service
20 Construction of two subway stations and a BRT service
21 Construction of a subway station and a BRT service
22 Increasing residential, commercial, and administrative densities

important characteristics such as developing various activities, ex- As

panding different land uses, and designing mixed-use activities should
be fulfilled. These characteristics can be evaluated by the land use
mixed-ness index (Eq. (5)) (Zhang & Guindon, 2006).
∑∩i Sc
MI =
∑∩i (Sc + Sr )
where MI is the Land use mixed-ness Index, Sc is the total area of the
commercial and administrative land uses, and Sr is the total of re-
sidential land uses.
Service coverage: This indicator has prominent effects on the PTI's
accessibility, efficiency, and level of service. Therefore, it shows the
suitability level of the public transit system in a neighborhood. This
indicator is calculated by Eq. (6) (Niyonsenga, 2012).
P
SC = B
P (6)
where PB is the population served in a certain buffer around a public
transit stop, and P is the total population of a neighborhood.
Accessibility to the service: This indicator reflects the ease of acces-
sibility of the population resided in a certain buffer area around a public
transit stop to the transit service. Accessibility to the service is com-
puted by Eq. (7).
LAS =
A (7)
where AS is the total area of buffer zones around the streets connected
to public transit stops, and A is the total area of all buffer zones.
After calculation of the indicators, they are aggregated by the
agent's utility function to evaluate TOD and PTI levels. The agent's
(5) utility function is defined according to urban planners' viewpoints.
Furthermore, this function specifies some behaviors of the agent.
Therefore, two utility functions are defined based on a simple ag-
gregation model, Eq. (8), to evaluate TOD and PTI levels.
n
∑ Ij W j
j=1 (8)
where, Ij is the jth indicator, Wj is the weight of jth indicator, and n is
the number of indicators.
In this aggregation model, the indicators are those defined in
Table 1. The weights of indicators are determined by experts' points of
view and Analytic hierarchy process (AHP) (Saaty, 1988). In AHP, ex-
perts perform several pairwise comparisons between the criteria and
indicators using a numerical scale in the range of 1 to 9, where the
number 1 represents the equal priority, and the number 9 shows the
extremely priority between two criteria or indicators. Then, these
numbers are converted to the weights of criteria and indicators (for
more details see (Saaty, 1988)).

96
H. Motieyan, M.S. Mesgari
Fig. 5. (a): NTOD level in neighborhoods (b): NPTI level in neighborhoods.
Fig. 6. The comparison between NTOD and NPTI levels.
3.2. Second step: decision-making process by agents
After evaluation of TOD and PTI levels, agents try to maximize the
efficiency of infrastructures and land uses in the study area. For this
purpose, they try to balance the TOD with the PTI levels through some
ameliorative decisions. The lack of balance between TOD and PTI levels
is the result of any of these two situations in a neighborhood. 1) There
are some public transit stations with lower travel demand in the
neighborhood because of inappropriate land-use planning. 2) Although
the demand for public transit is high in the neighborhood, there are no
suitable public transit infrastructures in the neighborhood. Therefore,
for making the balance between demand and supply, ABM involves
three main steps. At the first step, each agent evaluates its TOD and PTI
levels. According to this comparison, the neighborhood is assigned to
one of the following classes.
1. The TOD level is significantly lower than the PTI level in the
neighborhood.
2. The TOD level is significantly higher than the PTI level in the
neighborhood.
97
Cities 81 (2018) 91–100
3. The TOD and PTI levels are approximately the same in the neigh-
borhood.
In the second step, the agent determines what actions should be
performed based on its neighborhood's class. The TOD level should be
improved for the first class of neighborhoods, while the PTI level should
be enhanced for the second class. The third class does not require any
improvement. Table 2 shows the agents' actions and constraints defined
to improve the TOD and PTI levels. By improving the indicators, these
actions have various effects on the TOD or PTI levels. However, the
agent should select the actions with higher levels of improvement. This
selection is based on the utility function and upstream policies. The
former is used to determine the effect of each indicator on the TOD or
PTI level. The latter is used to know whether the improvement of an
indicator is in line with the municipality's policies in the neighborhood.
In this research, we model the effects of the upstream policies, which
are top-down procedures, by defining particular weights in the utility
function. Therefore, the structure of the utility function is transformed
into Eq. (9).
H. Motieyan, M.S. Mesgari Cities 81 (2018) 91–100

Fig. 7. The constructed and scheduled subway stations in the study area.

Table 6 4. Implementation and results

The statistical analysis of TOD and PTI levels before and after ABM.
OTOD and OPTI NTOD and NPTI
In this research, the proposed agent-based model is implemented
using the data of the 7th municipal region of Tehran as a case study.
Mean and standard Mean(|OTOD- Mean (|NTOD- The study area includes 22 neighborhoods which are considered as the
deviation (Std) OPTI|) = 0.2086 NPTI|) = 0.0670 agents. Each agent senses the data of its surrounding and calculates the
Std (|OTOD- Std (|NTOD-
OPTI|) = 0.2559 NPTI|) = 0.0956
indicators using GIS tools. The calculated indicators are aggregated by
RMSE RMSE(OTOD- RMSE(NTOD- the agent's utility function, (Eq. (8)), to evaluate the TOD and PTI levels
OPTI) = 0.264 NPTI) = 0.0941 in the neighborhood. The weights used in the utility function are de-
fined on the basis of experts' opinions and AHP method (Table 4).
n
We express the results of aggregation by Old TOD (OTOD) and Old
PTI (OPTI) levels since this evaluation is performed prior to the deci-
∑ I j Wj Pjk
j=1 (9) sion-making process. These levels are normalized using Eq. (10) and
assigned to the neighborhoods of the study area (Fig. 3).
where Pjk is the accordance of the jth indicator with upstream policies xi
x norm =
in the kth neighborhood. The accordance degree is indicated by a max(x ) (10)
number between 0 and 1, where the value of 0 shows that the indicator
is extremely against the policies, and the value of 1 shows that the where xi is the value of OTOD or OPTI level in the ith neighborhood,
indicator is thoroughly compatible with the policies. and xnorm is the normalized value for that neighborhood.
In the third step, first, some constraints at the level of region are In the next step, each agent compares the OTOD and OPTI levels to
defined, on the basis of municipalitys policies, to control the maximum make ameliorative decisions (Fig. 4). Afterwards, the agents make their
amount of area that can be changed into any land use type (Table 3). optimum decisions on the basis of the utility function and upstream
Agents directly interact with each other and share their decisions to policies (Eq. (9)), constraints, and their communications. The output
satisfy these regional level constraints. For instance, an agent may de- decisions are expressed in Table 5. According to these decisions, the
cide to add 50,000 square meters residential area. It shares this decision TOD and PTI levels are computed once again. Fig. 5 shows the New
with other agents to satisfy the constraint on the maximum amount of TOD (NTOD) and the New PTI (NPTI) levels resulted from agents' de-
the residential area that can be added to the region. If a regional-level cisions in the study area.
constraint is disobeyed, as the result of decisions of some agents, the
agent with the lowest difference between the TOD and PTI levels will 5. Discussion and assessment
discard its decision. Therefore, the interactions between agents lead to
some competitions among them. The winners are the agents with As shown in Fig. 5, only 6 neighborhoods have different classes with
stronger needs for change. respect to NTOD and NPTI levels, whereas before running the model, 17
neighborhoods were assigned to different classes (see Fig. 3). In other
words, by implementing the proposed agent-based model, the

98
H. Motieyan, M.S. Mesgari
agreement of the TOD and PTI levels improved from 23% to 73%.
Therefore, the imbalance between the TOD and PTI levels in neigh-
borhoods is significantly decreased as the result of agents' actions
(Fig. 6).
Comparing the agents decisions, as presented in Table 5, with the
decisions made by the municipality shows considerable similarities. For
example, the municipality has decided to increase accessibility, per-
meability, and street design in the neighborhoods of 4, 5, 7, 8, 11 and
12. Additionally, the municipality has planned to improve the TOD
level in the neighborhood 4, which has a very high PTI level, by in-
creasing commercial and administrative densities, land use diversity,
and street design. Furthermore, in the northern and southern neigh-
borhoods, which have high TOD levels, the municipality has planned to
upgrade the conventional bus services to the BRT services. In addition,
in the municipal plan, the transit services are to be developed in the
neighborhoods 8, 10, 11, 15, 18, 19, 20, and 21 by constructing subway
stations. Fig. 7 shows the current subway stations along with the ones
scheduled to be constructed.
Finally, the statistical analysis demonstrates noticeable improve-
ments in TOD and PTI levels in neighborhoods (Table 6). As it is per-
ceived from Table 6, the difference between TOD and PTI levels gen-
erally experiences 64% improvement. Consequently, the model can be
used for making practical decisions regarding various land-use and
public transit issues.
6. Conclusion
The complexity and mutual relations of transportation and urban
resource management necessitate researchers to develop more effective
planning methods that cover both transportation and land use dimen-
sions of an urban area. One of the most efficient planning methods
enabling the sustainable development of the urban area is TOD plan-
ning. In this study, we proposed an effective method from the TOD
perspective to help urban decision makers improve their decisions. For
this purpose, a novel agent-based model was developed to represent the
interactions between land use and transportation in the 7th municipal
region of Tehran, Iran. In the proposed model, the neighborhoods were
considered as agents that evaluated the TOD and PTI levels. The agents
make several decisions to balance the TOD and PTI levels, according to
the neighborhoods' conditions, municipal policies, and the decisions
made by other agents. The results of the agents' decisions were assessed
by comparing them with the detailed plan of Tehran municipality.
These assessments indicated that the use of the proposed model de-
creases the difference between the TOD and PTI levels by 64%.
In future research, it would be helpful to use a multi-agent simu-
lation system to enhance the efficiency and feasibility of decision-
making process. In a multi-agent model, there are different types of
agents such as citizen, landowner, and planner who play a variety of
roles in the land-use and transportation planning. These agents have
different attributes and behaviors and compete with each other to sa-
tisfy the objectives and constraints of the planning.
References
Alghais, N., & Pullar, D. (2017). Modelling future impacts of urban development in
Kuwait with the use of ABM and GIS. https://doi.org/10.1111/tgis.12293.
Babakan, A. S., & Taleai, M. (2015). Impacts of transport development on residence
choice of renter households: An agent-based evaluation. Habitat International, 49,
275–285.
Belzer, D., & Autler, G. (2002). Transit oriented development: Moving from rhetoric to reality.
Washington, DC: Brookings Institution Center on Urban and Metropolitan Policy.
Binglei, X., & Chuan, D. (2013). An evaluation on coordinated relationship between urban
rail transit and land-use under TOD mode. Journal of Transportation Systems
Engineering and Information Technology, 13(2), 9–13.
Calthorpe, P. (1993). The next American metropolis: Ecology, community, and the American
dream. Princeton Architectural Press.
Cervero, R. (2004). Transit-oriented development in the United States: Experiences, challenges,
and prospects. Vol. 102. Transportation Research Board.
Cervero, R. (2013). Transit-oriented development and land use. Transportation
99
Cities 81 (2018) 91–100
technologies for sustainability (pp. 947–958). Springer.
Cervero, R., & Dai, D. (2014). BRT TOD: Leveraging transit oriented development with
bus rapid transit investments. Transport Policy, 36, 127–138.
Cervero, R., Ferrell, C., & Murphy, S. (2002). Transit-oriented development and joint
development in the United States: A literature review. TCRP research results di-
gest (52).
Cervero, R., Guerra, E., & Al, S. (2017). Beyond mobility: Planning cities for people and
places. Island Press.
Cervero, R., & Kockelman, K. (1997). Travel demand and the 3Ds: Density, diversity, and
design. Transportation Research Part D: Transport and Environment, 2(3), 199–219.
Chawla, L. (2016). Growing up in an urbanizing world. Routledge.
Evans, I., Pratt, R. H., Stryker, A., & Kuzmyak, J. R. (2007). Transit-oriented
development—Traveler response to transportation system changes.
Ewing, R., Tian, G., Lyons, T., & Terzano, K. (2017). Does compact development increase or
reduce traffic congestion? (Retrieved from).
Frenken, K., Van Oort, F., & Verburg, T. (2007). Related variety, unrelated variety and
regional economic growth. Regional Studies, 41(5), 685–697.
Galelo, A., Ribeiro, A., & Martinez, L. M. (2014). Measuring and evaluating the impacts of
TOD measures–Searching for evidence of TOD characteristics in Azambuja train line.
Procedia - Social and Behavioral Sciences, 111, 899–908.
Geurs, K. T., & Van Wee, B. (2004). Accessibility evaluation of land-use and transport
strategies: Review and research directions. Journal of Transport Geography, 12(2),
127–140.
Kakaraparthi, S. K., & Kockelman, K. M. (2010). Application of UrbanSim to the Austin,
Texas, region: Integrated-model forecasts for the year 2030. Journal of Urban Planning
and Development, 137(3), 238–247.
Kwon, Y. (2015). Sejong Si (City): Are TOD and TND models effective in planning Korea's
new capital? Cities, 42, 242–257.
Ligmann-Zielinska, A., & Jankowski, P. (2010). Exploring normative scenarios of land use
development decisions with an agent-based simulation laboratory. Computers,
Environment and Urban Systems, 34(5), 409–423.
Motieyan, H., & Mesgari, M. S. (2017a). A novel spatial index using spatial analyses and
hierarchical fuzzy expert system for obtaining green TOD: A case study in Tehran
city. Geocarto International, 1–22.
Motieyan, H., & Mesgari, M. S. (2017b). Towards sustainable urban planning through
transit-oriented development (a case study: Tehran). ISPRS International Journal of
Geo-Information, 6(12), 402.
Nasri, A., & Zhang, L. (2014). The analysis of transit-oriented development (TOD) in
Washington, DC and Baltimore metropolitan areas. Transport Policy, 32, 172–179.
Niyonsenga, D. (2012). Assessing public transport supply for Kigali. Rwanda: University of
Twente Faculty of Geo-Information and Earth Observation (ITC).
O'Sullivan, D., & Haklay, M. (2000). Agent-based models and individualism: Is the world
agent-based? Environment and Planning A, 32(8), 1409–1425.
Páez, A., Scott, D. M., & Morency, C. (2012). Measuring accessibility: Positive and nor-
mative implementations of various accessibility indicators. Journal of Transport
Geography, 25, 141–153.
Patel, A., Crooks, A., & Koizumi, N. (2018). Spatial agent-based modeling to explore slum
formation dynamics in Ahmedabad, India. Geocomputational analysis and modeling of
regional systems (pp. 121–141). Springer.
Polzin, S., Pendyala, R., & Navari, S. (2002). Development of time-of-day-based transit
accessibility analysis tool. Transportation Research Record: Journal of the
Transportation Research Board, 1799, 35–41.
Prudkov, P. N. (2010). A view on human goal-directed activity and the construction of
artificial intelligence. Minds and Machines, 20(3), 363–383.
Ratner, K. A., & Goetz, A. R. (2013). The reshaping of land use and urban form in Denver
through transit-oriented development. Cities, 30, 31–46.
Saaty, T. L. (1988). What is the analytic hierarchy process? Mathematical models for de-
cision support (pp. 109–121). Springer.
Shirzadi Babakan, A., & Alimohammadi, A. (2016). An agent-based simulation of re-
sidential location choice of tenants in Tehran, Iran. Transactions in GIS, 20(1),
101–125.
Shirzadi Babakan, A., Alimohammadi, A., & Taleai, M. (2015). An agent-based evaluation
of impacts of transport developments on the modal shift in Tehran, Iran. Journal of
Development Effectiveness, 7(2), 230–251.
Singh, Y. J., Fard, P., Zuidgeest, M., Brussel, M., & van Maarseveen, M. (2014). Measuring
transit oriented development: A spatial multi criteria assessment approach for the
City Region Arnhem and Nijmegen. Journal of Transport Geography, 35, 130–143.
Theall, K. P., Scribner, R., Cohen, D., Bluthenthal, R. N., Schonlau, M., & Farley, T. A.
(2009). Social capital and the neighborhood alcohol environment. Health & Place,
15(1), 323–332.
Torrens, P. M. (2000). How land-use-transportation models work.
Van Eck, J. R., & de Jong, T. (1999). Accessibility analysis and spatial competition effects
in the context of GIS-supported service location planning. Computers, Environment and
Urban Systems, 23(2), 75–89.
Verburg, P. H., & Overmars, K. P. (2009). Combining top-down and bottom-up dynamics
in land use modeling: Exploring the future of abandoned farmlands in Europe with
the dyna-CLUE model. Landscape Ecology, 24(9), 1167.
Waddell, P., Borning, A., Noth, M., Freier, N., Becke, M., & Ulfarsson, G. (2003).
Microsimulation of urban development and location choices: Design and im-
plementation of UrbanSim. Networks and Spatial Economics, 3(1), 43–67.
Waddell, P., Ulfarsson, G. F., Franklin, J. P., & Lobb, J. (2007). Incorporating land use in
metropolitan transportation planning. Transportation Research Part A: Policy and
Practice, 41(5), 382–410.
Ward, D. P., Murray, A. T., & Phinn, S. R. (2003). Integrating spatial optimization and
cellular automata for evaluating urban change. The Annals of Regional Science, 37(1),
131–148.
H. Motieyan, M.S. Mesgari
Wegener, M. (2011). From macro to micro—How much micro is too much? Transport
Reviews, 31(2), 161–177.
Wei, R., Liu, X., Mu, Y., Wang, L., Golub, A., & Farber, S. (2017). Evaluating public transit
services for operational efficiency and access equity. Journal of Transport Geography,
65, 70–79.
Wey, W.-M. (2015). Smart growth and transit-oriented development planning in site se-
lection for a new metro transit station in Taipei, Taiwan. Habitat International, 47,
158–168.
Wey, W.-M., Zhang, H., & Chang, Y.-J. (2016). Alternative transit-oriented development
evaluation in sustainable built environment planning. Habitat International, 55,
109–123.
Widyahari, N. L. A., & Indradjati, P. N. (2015). The potential of transit-oriented
100
Cities 81 (2018) 91–100
development (TOD) and its opportunity in Bandung Metropolitan Area. Procedia
Environmental Sciences, 28, 474–482.
Yim, K. K., Wong, S., Chen, A., Wong, C. K., & Lam, W. H. (2011). A reliability-based land
use and transportation optimization model. Transportation Research Part C: Emerging
Technologies, 19(2), 351–362.
Zhang, Y., & Guindon, B. (2006). Using satellite remote sensing to survey transport-re-
lated urban sustainability: Part 1: Methodologies for indicator quantification.
International Journal of Applied Earth Observation and Geoinformation, 8(3), 149–164.
Zheng, S., Zhang, X., Sun, W., & Wang, J. (2017). The effect of a new subway line on local
air quality: A case study in Changsha. Transportation Research Part D: Transport and
Environment.. http://dx.doi.org/10.1016/j.trd.2017.10.004.