A Systematic Literature Review of the Hybrid Methodologies in Assessing Flood Indirect Impacts on Transportation

Abstract

As there is a staggering increase in flooding worldwide, many countries have prioritized sustainability of their transportation sector through flood impact prediction to support the transition during flooding. As such, research regarding the flood impacts on transportation has dramatically increased in recent years. Hybrid methods play an important role in simulating the flood situation and its impacts on traffic networks. This article offers a systematic literature review of existing research which employ hybrid methods to assess the indirect impacts of flooding on transportation. In this study, 45 articles are reviewed systematically to answer 8 research questions regarding modeling the indirect impacts of flooding on transportation. The hybrid techniques observed in the existing literature are discussed and along with the main barriers to precise prediction of flooding’s indirect impacts on transportation, future research directions are also suggested.

1. Introduction

The transportation network is a key component in the sustainable functionality of major cities, allowing for the movement of freight, population, and services [1]. In recent decades, climate change has been increasingly impacting the intensity, frequency, and duration of all types of floods, and thus affecting the urban regions’ infrastructures. Such impacts are also projected to increase in the coming decades [2]. Flooding in the transportation network may have serious consequences for human life, directly due to drowning and indirectly by reducing the first responders’ ability to respond to incidents [3]. For example, flooding in Barcelona in 2011 had a direct impact on people’s lives by interrupting the transportation network and had an indirect impact due to flooded road sections and traffic light failures [4]. In the short term, this natural disaster impact transportation by creating chaos in traffic as roads get blocked or become unsafe for travel and consequently the flow of people, goods, and services get adversely affected, and in the long term flooding can involve irreversible losses in and beyond the affected regions [5,6]. Therefore, to increase resilience and reliability, it is important to consider such types of disruptions in transportation network planning.

As flood impacts on transportation networks have garnered attention in the last decade, several literature reviews have been conducted considering different aspects of this issue.

Table 1. Literature review articles regarding flood impacts on transportation networks since 2015.

Article	Title	Covered Period	No. of Papers Reviewed
Kadaverugu et al. [7]	Impacts of Urban Floods on Road Connectivity—A Review and Systematic Bibliometric Analysis	1977–2020	115
Tachaudomdach et al. [8]	A systematic review of the resilience of transportation infrastructures affected by flooding	1998–2018	80
Nazarnia et al. [9]	A Systematic Review of Civil and Environmental Infrastructures for Coastal Adaptation to Sea Level Rise	1999–2019	47
Forero-Ortiz et al. [10]	A review of flood impact assessment approaches for underground infrastructures in urban areas: a focus on transport systems	2004–2019	29
Ahmed and Dey [11]	Resilience modeling concepts in transportation systems: a comprehensive review based on mode, and modeling techniques	2006–2020	Not determined
Rebally et al. [12]	Flood Impact Assessments on Transportation Networks: A Review of Methods and Associated Temporal and Spatial Scales	Up to 2020	30
Johnston et al. [13]	A review of floodwater impacts on the stability of transportation embankments	Not determined	Not determined

summarizes the most recent scholarly literature review articles in this field from 2015 onwards, which are discussed in detail in the following paragraphs.

Kadaverugu et al. [7] focused mainly on the direct impacts of flooding on road networks in their study. Their analysis provides a comprehensive discussion of various flood models applied in flooded road networks, highlighting the progress from classical hydrological models to near real-time crowd-sourced modeling techniques.

Tachaudomdach et al. [8] conducted a systematic literature review (SLR) on the resilience of transportation infrastructures affected by flooding. Their review, similar to Kadaverugu et al. [7], covers studies related to the direct impacts of floods on transportation networks from 1998 to 2018.

Nazarnia et al. [9] reviewed studies on the direct impacts of sea level rise, a specific flood type in coastal regions, on different types of urban infrastructures’ resilience, including transportation, water, and energy. Similarly, Ahmed and Dey [11] reviewed articles on both man-made and natural disasters’ direct impacts on transportation systems’ resilience. Although their concentration is not only on flood impacts, they summarized some modeling techniques of flooded transportation networks from two points of view: (a) transportation mode; and (b) mathematical technique applied for resilience quantification.

Rebally et al. [12] focused on how flooding affects transportation networks by investigating the methods and associated temporal and spatial scales, as well as the transportation models used to formulate flood impacts on the transportation system. Although they examine only research on transportation infrastructures, unlike Nazarnia et al. [9], Rebally et al. [12] denote the less researched area of indirect impacts compared to the direct impacts of flooding on transportation.

Forero-Ortiz et al. [10] and Johnston et al. [13] focused their reviews on the specific transportation assets of metro systems and rail embankments, respectively. Forero-Ortiz et al. [10] presented a review of studies on both the direct and indirect impacts of pluvial floods on metro systems and emphasized the lack of academic research on probable deaths and injuries that may occur in metro systems due to flooding. Johnston et al. [13] considered the direct impact of flooding and reviewed transportation embankment failures due to different types of floods, namely overtopping, offset head, above the slope, and basal floods.

Rebally et al. [12] found that hybrid methodologies, which combine different models or methods to assess flood risks and impacts on transportation networks, are the most commonly used approaches in previous research.

However, the authors suggest that developing an automated tool for conducting flood impact assessments could reduce costs and be applicable to other natural hazards, such as fires. To achieve this, existing studies on the indirect impacts of floods on transportation networks that use hybrid methods are analyzed in the current SLR, while papers that use other methods are excluded. Figure 1 shows the frequency of applied methods in published papers in this field since 2005, based on the same database used in Section 2.3.

The main aim of this paper is to provide an SLR of the hybrid technique, which is the most common technique being used to assess and quantify flood indirect impacts on motorized transportation systems. It means that studies on active modes, including pedestrians and bicycles, are excluded from this SLR, although they may have applied a hybrid method. Additionally, the review will examine the details of both traffic and flood models, as well as their integration techniques, with a focus on the analyzed flood types and transportation modes. This will enable the identification of the optimal options for each level of the development of the automated tool, while also highlighting gaps in the existing literature and paving the way for future research in this field.

Indeed, this SLR could be beneficial to two distinct audiences. Firstly, academic researchers just starting in the subject may find this overview valuable, as it analyzes and compares existing flood impacts assessments on traffic systems from 2005 up to 2023 and thus could be used to identify gaps in existing studies. If a researcher seeks to model a specific part of the flooded networks, this research could reveal which modeling techniques have been applied previously, and thus help in identifying some best modeling practices. It also includes an overview of the data used to parameterize existing studies. Secondly, decision-makers such as municipalities, stakeholders, and city planners might find this article helpful in gaining an overview of the current state-of-the-art on flood indirect impacts on transportation networks. It may offer advice on how to efficiently analyze a specific part or asset or even the whole transportation network against flooding by applying the same methods as the previous studies with the same situation of flood and transportation network.

The remainder of the present paper is developed as follows. Section 2 provides an overview of the SLR method and covers the data-gathering process for this study. Section 3 analyzes and discusses the obtained data to respond to the proposed research questions. It also discusses gaps in the existing field of research to provide additional insight into the outcomes gained from the collected data. Section 4 summarizes the study’s findings and proposes future research opportunities. Finally, Section 5 highlights the concluding remarks.

2. Methodology

The systematic literature review (SLR) technique has several advantages. Firstly, if it is well conducted, it will most likely not miss any relevant article. Secondly, the research questions are stated before reading the selected articles. Therefore, it is certified that the essential information is derived. This SLR was carried out in accordance with the Kitchenham [14] methodology. The guidance offered by Maybury et al. [15] for the domain of transportation planning and traffic management provides a more particular approach to SLRs, and this was also used. Figure 2 shows the instruction of the applied methodology which is adopted from Maybury et al. [15].

2.1. Research Questions

The main goal of this SLR was to provide answers to the research questions listed in

Table 2. Research questions.

NO.	Question
1	What spatial scales and geographical areas are analyzed in assessing the indirect impacts of floods on transportation?
2	What types of floods in transportation networks were modeled?
3	What models were applied for modeling the flood in the network?
4	How are the flood impacts seen in the traffic networks?
5	What transportation modes are evaluated in the indirect impact analysis of floods?
6	What aspects or variables of traffic are modeled?
7	What types of models were used for modeling the traffic situation in flooding?
8	What types and sources of data were used to parametrize the models?

. In this section, each study question is explained briefly.

RQ1 discovers the spatial scales of pilot zones in previous studies that could be divided into three different categories; community, urban and regional. This distinction is clarified in Section 3.2. This question also has the purpose of identifying the countries that have been investigated in this field to pinpoint an area that may require future investigation.

RQ2 aims to identify the type of floods that occurred in the networks that are mainly categorized into three types; pluvial (or surface water), fluvial (or river flood), and coastal (or sea level rise). In parallel, RQ3 examines the models that are applied to different types of floods. This may identify a gap in addition to the most used model for each flood type.

RQ4 aims to find ways of applying the flood model results to the traffic model and determine the techniques used to integrate flood and traffic models. It is crucial to recognize the most efficient way of integrating the flood and traffic models between the existing methods, namely dynamic and static, in different situations. Geographic Information System (GIS) is the most common connector of the flood and traffic models, while dynamic integration programming software, such as Python, can accelerate the analysis by updating the flood situation and its impact on the traffic.

RQ5, by considering the mode of transportation that is analyzed in case of flooding, sets the boundaries of the analysis itself. Rail, road, and waterways are the three motorized modes of transportation that are examined in flood situations. On the other hand, multimodal transport networks have attracted researchers’ attention due to their increasing usage in recent years. Therefore, their situation in case of flooding is a vital issue that needs to be examined.

RQ6 aims to identify which aspects of the flooded network modeling are the focus of used models and may reveal a research gap for future models. Models that aim to analyze flooded networks could consider several factors, such as the increased time and distance required for travel, as well as the additional emissions generated, which can be quantified in monetary terms.

The goal of RQ7 is to explore the various traffic flow models that have developed over time, with the aim of providing a comprehensive understanding of how vehicles interact with the network during flooding scenarios. Macroscopic, microscopic, and mesoscopic are three common traffic models which determine the simulated details in a traffic network. In some cases, a combination of these traffic models could be used as well. Exploring the application and efficiency of each model in different flooded transportation modes lights up the way for further studies and research in this field.

RQ8 provides a comprehensive list of the data types and sources utilized to parameterize existing models. It also could be used as a reference for data collection for future research.

2.2. Search Method and Data Sources

An SLR needs a search of all related databases. To this end, the databases Web of science, Science direct, Scopus, Springer, and Wiley were selected. Using the authors’ domain expertise, it was found that the majority of relevant studies are published on websites listed in these databases. To ensure that all related articles were retrieved, a primary search string was selected as the input of these databases. The search string is composed of three different terms joined together using the Boolean operator ‘AND’. Each phrase is dedicated to a different aspect of flood impacts on traffic modeling. Using the Boolean operator ‘OR’ the authors utilized their domain expertise to merge any relevant terms in each search string. The procedure is explained below.

This SLR is focused on flood hazard-related publications. Hence, the term ‘flood’ must be in the title and/or abstract of each retrieved article. This SLR is related to articles regarding flood impacts. As a result, the first group of terms for the search string is simply composed of two words as follows:

Flood impact OR flood consequence OR flood effect OR flood outcome

More particularly, the indirect impacts of the flood are the main interest of this SLR and the direct impact, such as building damages analysis, will be ignored. Therefore, the next group of terms is defined as follows:

Indirect OR secondary

Finally, since the main purpose of this SLR is to analyze methods of assessing flood impacts on traffic, the third defined group of terms is as follows:

Transportation OR transport OR traffic OR transit

Each group of terms is joined using the Boolean operator ‘AND’ to form the search string as follows:

(Indirect OR secondary)

AND

(Flood impact OR flood contact OR flood consequence OR flood effect OR flood outcome)

AND

(On transportation OR on transport OR on traffic OR on transit)

On 1 January 2023, the search using the given specifications retrieved 114 different articles. These articles were then subjected to the selection procedure.

2.3. Articles Selection Process

The articles which were published from 2005 onwards were analyzed in the current SLR. The procedure for retrieving articles is shown in Figure 3.

Table 3. Exclusion criteria.

ID	Criterion
EX1	The article is not peer-reviewed.
EX2	The article is published after 1 January 2023.
EX3	The article is not focused on traffic disruptions.
EX4	The article does not include a simulation or model involving traffic condition following or during flood.
EX5	The article is not written in English.

lists the exclusion criteria that were used to guarantee that the articles chosen were of efficient quality and that all relevant publications were included. It is worth noting that the phrase peer-reviewed as a criterion for including and excluding the articles is denoting the fact that the organizational reports are not included in the current SLR since it is difficult to verify whether they have been peer-reviewed. Furthermore, because such papers are not indexed by published databases, it is impossible to review them systematically. Finally, articles that applied hybrid methods in their analysis are selected to be analyzed in this SLR. Accordingly, 45 articles from the 158 retrieved articles were selected.

3. Results

3.1. Articles Reviewed

An overview of the reviewed articles in the current SLR and simplified answers to the research questions are provided in

Table A1. Answers to the research questions according to the reviewed articles.

Article	RQ1	RQ2	RQ3	RQ4	RQ5	RQ6	RQ7
Suarez et al. [33]	Boston, USA/City scale	Pluvial and coastal flooding scenarios	Hydrologic and hydraulic modeling	Static	Road	Additional travel time and lost trips	Microscopic
Sohn [34]	Maryland, USA/Regional scale	Hypothetical flood disruption	-	-	Road	Accessibility index compositing of distance and traffic volume	Microscopic
Mitsakis et al. [29]	Athens, Greece/City scale	Pluvial	-	Dynamic	Road	Additional travel time and distance and speed drops	Macroscopic fundamental diagram (MFD)
Suwanno et al. [52]	Sukhumvit district, Bangkok, Thailand/City scale	Pluvial	-	Dynamic and Static	Road	Additional travel time and distance	Macroscopic fundamental diagram (MFD) and Mesoscopic
Q. Li et al. [18]	Shenzhen, China/City scale	Pluvial	-	Static	Road	Additional travel time	Macroscopic
Lu et al. [45]	Hillsborough County, Florida/Regional scale	Coastal	-	Static	Road	Additional travel time	Macroscopic
Pérez-Morales et al. [22]	Murcia, Alcantarilla, Santomera, and Beniel municipalities, in south-eastern Spain/City scale	Fluvial	-	Static	Road	Additional travel time and speed drops	Macroscopic
Wang et al. [46]	mainland China, with Hainan Province excluded/Regional scale	Fluvial	CaMa-Flood model	Dynamic	Road	Additional travel time	Microscopic
Borowska-Stefańska et al. [25]	The Warta Water Region, Poland/Regional scale	Fluvial	-	Static	Road	Speed drops	Macroscopic
Kim et al. [58]	North Gyeongsang Province, South Korea/City scale	Fluvial	Hydrological and hydraulic modeling	Static	Road	Additional travel distance	Macroscopic
Martín et al. [23]	Two Mediterranean regions: Valencia (Spain) and Sardinia (Italy)/City scale	Coastal	-	Static	Road	The decline in the network’s territorial accessibility	Not determined
Yin et al. [38]	Shanghai, China/Community scale	pluvial	Hydrodynamic model (FloodMap-HydroInundation2D)	Dynamic	Road	Additional travel time	Macroscopic
Zhang and Alipour [41]	Iowa/City scale	Pluvial	-	Static	Road	Additional travel time, speed drops, and detour identification	Microscopic
Praharaj et al. [47]	Norfolk, Virginia/City scale	Coastal	-	Dynamic	Road	Additional travel time	Data predictive model
Li et al. [48]	City center of Shanghai/City scale	Pluvial	FloodMap-HydroInundation2D	Dynamic	Road	Additional travel time and vehicle volume redistribution	Microscopic
Zhu et al. [17]	Chinese railway/Regional scale	Fluvial	Monte Carlo sampling method	Dynamic	Rail	Daily cancelled, detoured, and affected trains, and additional travel time	Microscopic
Balijepalli and Oppong [19]	York, UK/City scale	Fluvial	-	Static	Road	Additional travel time	Microscopic
Arrighi et al. [27]	Florence (Tuscany, Italy)/City scale	Fluvial	1D–quasi-2D hydraulic model	Static	Road	Rerouting, additional travel time, and service area reduction	Macroscopic
Bucar and Hayeri [59]	Hoboken, New Jersey/City scale	Pluvial	-	Static	Road	Additional travel time and distance, and trips completed	Macroscopic
Hong et al. [53]	China/Regional scale	Fluvial	Monte Carlo simulation	Static	Rail	Number of interrupted trains and their interruption duration	Macroscopic
Yin et al. [49]	Shanghai, China/City scale	Pluvial	2D hydrodynamic modeling	Dynamic	Road	Additional travel time and function losses from emergency stations to critical locations	Macroscopic
Pregnolato et al. [20]	Newcastle upon Tyne, UK/City scale	Pluvial	2D hydrodynamic model (LISFLOOD-FP)	Static	Road	Additional travel time and distance	Microscopic
Coles et al. [21]	York, UK/City scale	Pluvial and fluvial	Hydrodynamic flood inundation model (1D/2D coupled version of Flood Map)	Static	Road	Changes to service area coverage and response times to vulnerable locations (care homes and sheltered accommodation) by emergency services	Microscopic
Tsang and Scott [32]	Calgary, Alberta /City scale	Fluvial	-	Static	Road	Additional travel time and speed drops	Macroscopic
Duy et al. [66]	Ho Chi Minh City, Vietnam/City scale	Pluvial	Hydraulic 1D classic module	-	Road	Rerouting, network connectivity loss	Macroscopic
M. Li et al. [50]	Central urban area of Shanghai/City scale	Fluvial	FloodMap-HydroInundation2D	Dynamic	Road	Emergency medical service accessibility, additional travel time, and speed drops	Microscopic
Green et al. [3]	Leicester, UK/City scale	Fluvial	Hydrodynamic inundation model (TUFLOW)	Static	Road	Additional travel time	Microscopic
Yin et al. [37]	Manhattan, New York City/City scale	Coastal	Hybrid methodology by applying 2D flood inundation model (Flood Map-Inertial)	Static	Road	Speed drops	Macroscopic
Arrighi et al. [28]	Galluzzo in Florence, Italy/Regional scale	Pluvial	MOBIDIC (MOdello di Bilancio Idrologico DIstribuito e Continuo) hydrological model, Hydraulic model TELEMAC-2D	Static	Road	Additional travel time	Macroscopic
Suh et al. [43]	San Francisco Bay Area/City scale	Coastal	CoSMoS	Static	Road	Additional travel time	Microscopic
Kasmalkar et al. [67]	San Francisco Bay Area/City scale	Coastal	-	Static	Road	Additional travel time	Microscopic
Pyatkova et al. [68]	Marbella, Spain/City scale	Fluvial	MIKE FLOOD model	Static and dynamic	Road	Additional travel time and distance	Microscopic
Borowska-Stefańska et al. [26]	Mazovian Voivodeship (Eastern Poland)/Regional scale	-	-	Static	Road	Additional travel time	Microscopic
Yu et al. [69]	England/Regional scale	Pluvial	A network-based geospatial analysis1 (ArcMap)	Static	Road	Additional travel time and change to service area coverage	Microscopic
Sun et al. [44]	San Francisco Bay Area/Regional scale	Coastal	CoSMoS	Static	Road	Additional travel time	Microscopic
Choo et al. [39]	Sadang-dong area, Seoul, Korea/Community scale	Pluvial	S-RAT (Spatial Runoff Assessment Tool) and Flood Inundation (FLO-2D) model	Static	Road	Travel speed reduction	Microscopic
Liu et al. [60]	Wuhan, China/City scale	Pluvial	InfoWorks Integrated Catchment Modeling, which combined one-dimensional hydraulic and two-dimensional inundation models	Static	Road	Additional travel time and affected population	Microscopic
Evans et al. [4]	Barcelona and Bristol cities, Spain/City scale	Pluvial	1D/2D-coupled flood model	Static	Road	Additional travel time, fuel consumption and pollution levels	For Barcelona: mesoscopic For Bristol: microscopic
Han et al. [40]	Miami-Dade County, FL/City scale	Coastal	-	Dynamic	Road	Additional travel time	Microscopic
Gori et al. [64]	Houston/Regional scale	Fluvial	Hydrologic Engineering Center-Hydrologic Modeling System (HEC-HMS) and hydraulic model Hydrologic Engineering Center-River Analysis System (HEC-RAS)	Static	Road	Additional travel time	Hybrid
Abdulla et al. [35]	Memorial super neighborhood in Houston/City scale	Fluvial	Geospatial model in ArcMap	Dynamic	Road	Speed drops	Macroscopic
He et al. [36]	Kinshasa/City scale	Pluvial and fluvial	Fathom global flood model	Static	Road (public and private) and pedestrian	Additional travel time and distance	Microscopic
Azucena et al. [54]	Mississippi River/Regional scale	Waterways’ disruption including droughts and floods	A hybrid methodology combining statistical analysis and simulation	Dynamic	Waterways and road	Additional travel time and distance	Microscopic
Knight et al. [51]	Harvard Gulch, Denver, Colorado, USA/City scale	Pluvial	A detailed dual-drainage model of green stormwater infrastructure (GSI), surface runoff, and stormwater networks	Static	Road	Additional travel time	Microscopic
Shahdani et al. [30]	Santarem region, Portugal/Regional scale	Fluvial	-	Static and dynamic	Road	Additional travel time and distance, and speed drops	Mesoscopic

and

Table A2. RQ8 answers from reviewed articles.

Article	RQ8: Data Type
	Flood Data	Traffic Data
Suarez et al. [33]	Flood maps, Land use maps, Map of the road network	Road network (road types and speed limits), Traffic demand data
Sohn [34]	-	Highway network, Traffic flow
Mitsakis et al. [29]	Data in 10 min time-steps, providing both the intensity and the total rainfall amounts that were recorded during the storm Traffic data from 219 field detectors (inductive loops) of the Athens traffic management center, measuring vehicle flow, speed and occupancy, Data from detectors located within the same zone for MFDs (vehicle flow vs. vehicle speed vs. occupancy) analysis, Floating Car Data (FCD) from freight vehicle fleets containing GPS coordinates and speed information
Suwanno et al. [52]	Data at 5 min intervals from the road flood monitoring system	Taxi data, The road network for the case study of the Sukhumvit area, Traffic signal input
Q. Li et al. [18]	The meteorological data containing hourly rainfall intensity observed from 44 weather stations Taxi GPS data include records of the date, time, license plate number, longitude, latitude, instantaneous speed, direction, and taxi status taken every 10 s, Road network as a directed graph
Lu et al. [45]	Light detection and ranging digital elevation data	The road network data, The OD data, The zone-based population data of the county
Pérez-Morales et al. [22]	Food polygons including water depth with return period of 100 years	The road network, The OD data, A network service area, Emergency Management Center, Network scenarios
Wang et al. [46]	The runoff from global climate models	The highway network, The OD data, Flood depth of road segments, The major railway stations of the capital cities
Borowska-Stefańska et al. [25]	A 100 year flood-risk map	The road network, The OD data
Kim et al. [58]	Discharge values for the Nakdong River and the Naesung Stream, a tributary of the Nakdong River	A flood hazard map, the annual average daily traffic across each of the six bridges for 2015 and 2027, population growth, change in land use patterns, and development projects near the new city in each time period, The number of trips between cities from a gravity model, The average ratio of trips made by each mode in North Gyeongsang Province
Martín et al. [23]	-	Road network, Traffic flow, Flood scenarios, denoting a network that lacks a certain number of sections
Yin et al. [38]	Topography of urban surface and road network, The rainfall intensities with the duration of one hour and the return period of 1 in 5, 10, 20, 50 and 100 years	GIS-based road network dataset
Zhang and Alipour [41]	Damaged network maps and closure information under flooding scenarios	Graphic network, Traffic features, Geographic features, and Water stage
Praharaj et al. [47]	Flood incident data from the City of Norfolk, Waze flood report data, Rain and tidal gauge data	The roadway network, Traffic volume data, and Streetlight data
Li et al. [48]	Four scenarios concerning rainfall return periods of 10, 20, 50, and 100 years	Road dataset, and Traffic flow
Zhu et al. [17]	National-scale flood return period maps	The geographic information, time table data, and passenger capacity data
Balijepalli and Oppong [19]	-	Road network, and travel demand between OD pairs
Arrighi et al. [27]	The flood map for the worst-case scenario	Road network information, and commuter data
Bucar and Hayeri [59]	Storm magnitude, tide level, inundation maps	Graph network, link attributes, node attributes, OD demand, and driver behavior
Hong et al. [53]	Flood event data for the past 30 years, Flood-induced railway disruption event data	The Chinese rail network, and rail stations locations
Yin et al. [49]	Precipitation data, topography data, road and facility, demographic data	Road network, and travel demand between OD pairs
Pregnolato et al. [20]	Flood depths and flow velocity, bridge geometry and characteristics	Average daily traffic flow, road network, and type of vehicle
Coles et al. [21]	Flood depth, velocity, and duration	Road network, emergency response nodes, facilities (locations of fire stations, ambulance stations and hospitals), and vulnerable population
Tsang and Scott [32]	-	Road network, emergency response nodes, facilities (locations of fire stations, ambulance stations and hospitals), vulnerable population, and Flood maps
Duy et al. [66]	River network associated with cross-sections, Boundary inputs, Processing parameters	-
Li et al. [50]	Flood depth, velocity, and duration	The road dataset, and Traffic data
Green et al. [3]	High-resolution citywide surface water inundation depth data, fluvial inundation data, flood hazard data	Ordnance Survey Integrated Transport Network (ITN) data, The Environment Agency National Receptor Database (NRD) including critical infrastructure nodes and vulnerable locations
Yin et al. [37]	NPCC20 s SLR projection for NYC, NYC floodplain topography	The most recent GIS dataset of city facilities and a single line street base map (i.e., LION), default turn restrictions in ArcGIS10.2.
Arrighi et al. [28]	Rainfall event, Regional hydrologic service report, Digital Terrain Model (DTM) (10 m resolution), Land Use-Land Cover (1:25,000), Pedology (1:25,000), High-resolution DTM (1 m), Road network (1:2000), River network (1:2000), Building footprint (1:2000), Roughness coefficient	Road network (1:2000), Hotspots (1:2000), Population and apartment census, and Building footprint (1:2000)
Suh et al. [43]	The scenario of a 0.5 m sea level rise, the shapefiles containing the land surface and building elevations for the entire San Francisco Bay Area	A population file, A full network file, and A list of daily activity for 500,000 agents
Kasmalkar et al. [67]	-	A road network, OD commuter data, and 1 m resolution flood maps
Pyatkova et al. [68]	Measurements and photos form flooded roads in 2016	Streets with speed reduction or closure, Road network from OpenStreetMap (OSM) including road types and speed limits, an activity-based traffic demand model was set to predict the attributes of trips (purpose, origin, destination, and timing)
Borowska-Stefańska et al. [26]	-	Road network, Traffic flow, The data on the administrative division, Commute data, and Flood hazard maps
Sun et al. [44]	The external tidal forcings at the ocean boundaries, water velocities obtained at multiple points along the boundaries	Full-day activity chains of individual agents, a connected network, transit schedules, and an updated transportation networks based on inundation information
Choo et al. [39]	Geographic data, 3 h rainfall profiles	Rainfall-Vehicle speed curve, Rainfall-Depth curve, Road network, Traffic flow
Liu et al. [60]	Flood hazard maps, Demographic data, Social media data	Point-of-interest (POI), Building footprint, Road network data, and Commute-related data
Evans et al. [4]	For Barcelona: The maximum rainfall intensities for the Baseline and Business As Usual (BAU) scenarios, Foundation for Climate Research (FIC) data, For Bristol: Climate data from UKCP09 predictions	For Barcelona: The road network including a wealth of properties relating both the physical characteristics and imposed rules parameters, such as speed restrictions, number of parking maneuvers per hour, and lane capacity. For Bristol: OpenStreetMap (OSM) data, traffic flows from the National-Receptor-Database (NRD), and the spatial information of land-use points from the NRD
Han et al. [40]	-	Transportation network data, the high resolution of DEM data and storm surge data, the census data and parcel data, and the South Florida household travel survey
Gori et al. [64]	Rainfall data at 4 km2 resolution and 5 min intervals, discharge data, Digital Elevation Models (DEM) data	Houston’s street centerline GIS data, The locations of fire stations and hospitals, and The census block group dataset
Abdulla et al. [35]	The hourly flood depth data, the temporal changes in the flood depth	Road network and other auxiliary information from OpenStreetMap using the OSMnx python package
He et al. [36]	A set of high-resolution global flood maps that capture both the extent and depth of pluvial and fluvial floods	A commuter travel survey with travelers’ socio-economic attributes and Origin-Destination (OD) information of commuters’ trips, An innovative General Transit Feed Specification (GTFS) dataset collected for the purpose of this analysis under normal (dry) and flooded (wet) conditions, and Transportation network vector data from Open Street Maps
Azucena et al. [54]	A map of the United States, Maps of navigable waterways and highways, Vessels, Trucks, Ports along the waterways, Fifteen locks along MKARNS, Twenty-four sites along the MKARNS and Mississippi River, The data used corresponds to Gage Height’s hourly measurements and lock availability data in eighteen different sites
Knight et al. [51]	Imperviousness and land use data from the City and County of Denver Open Data Catalog, Stormwater network data acquired from the City and County of Denver Open Data Catalo	The modified transportation network due to flooding downloaded from OpenStreetMap, OD data of three different periods, i.e., peak hour 1 (17:00–18:00), peak hour 2 (18:00–19:00), and off-peak hours (19:00–23:00)
Shahdani et al. [30]	For static integration: water-depth hazard maps For dynamic integration: water-level measurements from past flood events to characterize the progression of the flood, water levels and annual maximum stream flows measured at six hydrometric stations over the Tagus River	Road network and traffic demand data from Infrastructures de Portugal including road types and speed limits

. Section 3.2 discusses the answers to the research questions in more detail.

3.2. Research Questions’ Answers

More comprehensive answers to this SLR’s questions (given in Table 2) are presented in this section.

RQ1: What spatial scales and geographical areas are analyzed in assessing the indirect impacts of floods on transportation?

3.2.1. Spatial Scales Analyzed

Following Rebally et al. [12], the spatial networks could be classified into three categories, namely community, urban city, and regional. Community-scale assessments focus on networks with a typical length of less than 5 km, urban city scale assessments cover residential networks with a typical length of between 5 and 200 km, and regional scale denotes large-scale networks with a typical length of more than 200 km. The type of network analyzed in the region/area and the impact range are used to assign the mentioned spatial scale ranges. The majority of articles being reviewed in this SLR are focused on urban networks, and any publication that examines highway networks is categorized as being at the regional scale. However, estimating the indirect impacts of a flood on a particular transportation network could be challenging due to the variety of conditions and damages (such as economic and environmental impacts), as well as due to the interdependent nature of transportations connecting different modes, especially in urban contexts [16]. Figure 4 shows the frequency of the different spatial scales analyzed in the literature of indirect impacts of flood on transportation. As it is depicted, the majority of the articles analyzed a spatial scale of the urban city, while only 4% and 30% of articles considered community and regional scales for their analysis, respectively. This could be attributed to the fact that for assessing the impacts of flooding on traffic, the cascading effect of the disruption is non-negligible which leads to considering a larger spatial scale. On the other hand, by considering a larger spatial scale the computational cost of analysis grows too. Therefore, when selecting the appropriate spatial scale for a study, it is important to not only consider the type of network and affected area, but also weigh the pros and cons of the chosen spatial area for the case study.

3.2.2. Continents Analyzed

The geographical areas modeled are analyzed at a continent level and discussed in more detail in this section. There are 17 articles modeling North America, 15 modeling Asia, 14 modeling Europe, and 1 modeling Africa. Figure 5 presents a comparison of the frequency of articles analyzed in the current SLR based on their scale and the analyzed continents. It is evident that Asia has been the primary focus of researchers, with all three types of scales being applied in various analyses. On the other hand, Africa has received the least attention, with only urban city scale being analyzed in this continent. In addition, it should be noted that the urban city scale has been the most commonly used scale in previous research, while the community scale has been the least applied scale.

The vast majority of articles modeling Asia are concentrated on China [17]. This could be influenced by the fact that China has a history of summer flooding, but recent deforestation, reclaiming of wetlands, and storage of water for power generation and irrigation have made it more vulnerable [18].

Of the articles modeling Europe, Balijepalli and Oppong [19], Green et al. [3], Pregnolato et al. [20], and Coles et al. [21] focused their models on the UK; Pérez-Morales et al. [22], Martín et al. [23], and Pyatkova et al. [24] on Spain; Borowska-Stefańska et al. [25] and Borowska-Stefańska et al. [26] on Poland; Arrighi et al. [27,28] on Italy; Mitsakis et al. [29] on Greece; and Shahdani et al. [30] on Portugal. The distribution of articles across Europe highlights the high and frequent levels of disruption in traffic caused by flooding. This may be due to the inadequate drainage systems that are unable to cope with flood runoff and also as mentioned by Britain and Brown, predicted increases in rainfall resulting from climate change could exacerbate this situation [31]. This prediction justifies the growing research in this field worldwide.

Most North America-focused articles modeled the US, except Tsang et al. [32], who modeled Canada. Suarez et al. [33] and Sohn [34] modeled the US more broadly, whereas others had narrowly defined geographic focuses. For example, Abdulla et al. [35] focused on the Memorial super neighborhood in Houston to simulate and assess the effects of floods on traffic. The single article modeling Africa [36] focused on Kinshasa.

Overall, researchers have primarily investigated the indirect effects of flooding on transportation in major metropolitan areas of developed countries with well-established infrastructures. Some examples of these areas include York [19,21], New York [37], Shanghai [38], Shenzhen [18], and Seoul [39]. On the other hand, there has been limited research publications in the literature for municipalities with inadequate or low-quality infrastructure, particularly in developing countries.

3.2.3. RQ2. What Types of Floods in Transportation Networks Were Modeled?

The flood types analyzed in each article are given in Table A1. Generally, floods could be divided into three types, namely pluvial, fluvial, and coastal. A pluvial flood happens when a heavy rainstorm causes flood but there is no overflowing of water bodies. On the contrary, when the water level in a lake, river, or stream rises and overflows onto the banks, shores, and adjacent land, a fluvial or river flood occurs. Coastal flooding or storm surge is the overflowing of coastal areas by seawater. Tsunamis and powerful windstorms are two common sources of coastal flooding. There are 19 articles that worked on the indirect impacts of pluvial, 17 on fluvial, and 9 on coastal floods impacts on transportation networks.

3.2.4. RQ3. What Models Were Applied for Modeling the Flood in the Network?

The majority of articles in this field have employed a methodology based on historical data to indirectly quantify the impact of flooding. This involves the creation of hazard maps for different flood scenarios with varying return periods (e.g., 20, 100, and 1000 years) for a specific pilot zone. These maps show the extent of flooding, water depth, and flow velocity for each scenario. Typically, the water-depth hazard maps are used to identify links in the transportation network that may be closed or require reduced speed due to flooding. This approach is commonly used for the static integration of flood and traffic models. Several studies, including those by Han et al. [40], Shahdani et al. [30], Zhang and Alipour [41], Sohn [34], and Martín et al. [23], have employed this methodology.

On the other hand, some articles have employed more detailed and accurate hydrological and hydraulic models to obtain precise information regarding flood propagation, including its spatial and temporal characteristics. This information is essential for the dynamic integration of flood and traffic models. For instance, Yin et al. [38] utilized the Hydrodynamic inundation model, which is a modified version of Flood Map that utilizes local inertial-based techniques to model flood inundation in topographically complex floodplains [42].

In addition, other studies have utilized advanced 2D hydrodynamic models to obtain more accurate projections of flood conditions. Suh et al. [43] and Sun et al. [44] utilized the CoSMoS (Coastal Storm Modeling System) model, which is a 2D hydrodynamic model that can project local-scale water levels and flood extents resulting from rise in sea level. Similarly, Pregnolato et al. [20] employed the 2D hydrodynamic LISFLOOD-FP model, which can simulate flood depths and flow velocity using gauge data to establish computational fluid dynamics boundary conditions for a flood scenario.

Other flood modeling techniques, such as the CaMa-Flood model, MOBIDIC hydrological model, TELEMAC-2D hydraulic model, MIKE FLOOD model, HEC-HMS, and HEC-RAS model were used in only one article of this SLR.

3.2.5. RQ4. How Are the Flood Impacts Seen in the Traffic Networks?

The integration of flood and traffic models, which can be either static or dynamic, is necessary to capture the relationship between traffic and flood. Dynamic integration takes into account the temporal propagation of floods, whereas static integration assumes a stable flood condition throughout the modeling period. The static integration of the flood and traffic models produces just a binary representation of flooding, which can be a reasonable solution for assessing the effects of the flood on the transport system when the flood propagation speed is as slow as deniable [18,21,22,25,29,30,33,38,40,45,46,47,48,49,50,51]. The dynamic integration applies the likelihood of road closures or car accidents as a result of flooding [25,29,38,46,49,50]. The articles analyzed mainly used flood depth as a criterion to evaluate whether a road is closed due to traffic or requires speed reduction [40,47,48]. Previous research has suggested that a water depth of 0.3 m is the most common criterion for street closures due to flooded vehicle stability thresholds, with a speed reduction to 20 km/h for water depths lower than 0.3 m [30]. However, considering other factors, such as the impact of flood velocity on traffic flow in addition to flood depth, may improve the accuracy of the results [21,50]. Furthermore, floods are complex natural hazards that involve dynamic interactions between floodwater and traffic [20,21,25,32]. As a result, dynamic integration of flood and traffic models may provide more accurate results regarding the indirect impacts of floods on transportation networks.

Of the articles analyzed in this SLR, 30 articles applied a static situation, and 11 articles applied a dynamic situation for flooding. Nonetheless, 3 articles considered and analyzed both static and dynamic integration of flood and traffic in their analysis [24,30,52].

3.2.6. RQ5. What Transportation Modes Are Evaluated in the Indirect Impact Analysis of Floods?

Although this SLR covers all types of transportation vehicles, almost all of the articles examined in the review focused solely on road networks. Only two articles, authored by Zhu et al. [17] and Hong et al. [53], discussed disruptions in rail transportation. Two other articles included considerations of pedestrian and waterway disruptions, in addition to road networks [36,54]. Figure 6 depicts a comparison of the frequency of articles analyzed in the current SLR, based on their scale and the analyzed modes of transportation. The analysis reveals that road networks have garnered more attention than any other type of transportation network at all spatial scales. Rail networks, on the other hand, have only been analyzed in regional scales, while multimodal networks have received the least attention at all spatial scales.

As seen in Figure 6, the articles primarily concentrated on a single mode of transportation when it comes to flooding, with a stronger emphasis on road networks, highlighting their significance in flood situations. The majority of articles focused on the movements of road vehicles, without distinguishing between different categories of vehicles. Nonetheless, a small number of articles did concentrate on specific types of vehicles. For instance, three articles examined emergency service vehicles, such as ambulances and fire trucks [21,49,50], while one article examined private and public road vehicles separately [36].

The limited number of articles focusing on rail networks, or public transportation more broadly, may be due to the absence of critical databases necessary to obtain accurate results, such as traffic control systems, train movement conditions, and passenger behavior during a disruption. However, flooding on railway networks can significantly impact commuter travel and have severe economic consequences for a country, as many businesses rely on rail transportation to move goods and cargo from one location to another.

Conversely, researchers have increasingly turned their attention to disruptions in multimodal transportation networks in recent years, as evidenced by the work of He et al. [36] and Azucena et al. [54]. While these efforts have made significant contributions to the management and mitigation of natural hazards during and after floods, it is believed that further research is needed to assess the impacts of floods on multimodal transportation networks. Although general multimodal traffic is projected to increase in the coming years [12], research into the effects of flooding on multimodal traffic flow and economic analysis remain limited.

It is worth noting that this systematic literature review (SLR) only concentrates on the indirect effects of floods on transportation. However, if a broader range of disasters, such as the spread of epidemics [55], or a wider context of analysis, such as network resilience [56,57], is considered, it will be observed that more modes of transportation are included.

3.2.7. RQ6. What Aspects or Variables of Traffic Are Modeled?

Comparing traffic analysis results under normal and flooded conditions can provide insight into the indirect impacts of floods on transportation networks. Generally, these results are related to travel time, travel distance, and street speeds, although some articles have examined other aspects of traffic. For instance, Yin et al. [49], Coles et al. [21], and Li et al. [50] assessed the impacts of flooding on the capacity of urban emergency responses, taking into account function losses from emergency stations to critical locations (such as care homes and sheltered accommodation) and changes to service area coverage by emergency services. Additionally, Hong et al. [53] analyzed the number of interrupted trains and their duration of interruption as the response variable in their traffic modeling.

Table 4. Number of articles considering the analyzed traffic aspect.

Transportation Type	Analyzed Traffic Aspect	Number of Articles
Private vehicles	Travel time	17
	Streets speed	11
	Travel distance	9
	Accessibility	5
	Detoured trips	5
	Lost trips	3
	Affected population	1
	Fuel consumption	1
	Pollution level	1
Emergency vehicles	Access time	3
	Service area coverage	3
Public vehicles	Number of interrupted trains	2
	Trains interruption duration	1

provides a summary of the traffic aspects analyzed in the articles and the number of articles that used them. Most of the articles focused on changes in private vehicle traffic, with travel time being the most commonly used traffic aspect in the literature concerning flooding situations affecting private transportation. In the case of emergency vehicles, access time and service area coverage are critical aspects that have recently garnered researchers’ attention in flooding situations. The sole article that analyzed traffic in public transportation [53] focused on the train network and examined the number of interrupted trains and the duration of their interruption.

3.2.8. RQ7. What Types of Models Were Used for Modeling the Traffic Situation in Flooding?

Traffic situations in the event of flooding are typically modeled in the literature using three primary methods: macroscopic, microscopic, and mesoscopic. A combination of these models can also be utilized. Macroscopic traffic modeling, a common approach in this field, is capable of representing congestion or diversion in a large area but cannot represent individual vehicles or people on the network. Many articles [18,22,25,27,29,32,35,38,45,49,53,58,59] used macrosimulation in their methods; however, it cannot properly address the dynamics of the transportation system during a flooding event. Nevertheless, the macroscopic modeling method has advanced in the field of traffic flow theory and control in the form of macroscopic fundamental diagrams (MFDs), which have been utilized by Mitsakis et al. [29] and Suwanno et al. [52].

In contrast, microsimulation is a modeling approach that tracks individual vehicles on the network in small time steps, providing very precise results. Despite the higher computational cost and the level of detail required regarding input data, microsimulation is frequently used to analyze the indirect impacts of flooding on transportation. It should be noted that most of the articles that used microsimulation focused on transportation networks in major metropolises of developed countries with a relatively well-recorded database [33,34,36,39,40,51,54,60].

In addition to the macroscopic and microscopic traffic models, mesoscopic simulation and combined models have received attention in most recent articles. In order to apply the temporal and spatial variation of traffic during flood, the mesoscopic simulation can compute vehicle movements with queues which may run up to 100 times faster than a microscopic model [61]. Thereby, a mesoscopic simulation, which falls between these two modeling approaches, can portray a feasible alternative for the assessment of flood impacts in transportation networks; this method was used by Shahdani et al. [30]. Additionally, a number of articles used the mesoscopic method in order to make a comparison between their function and results. For example, Suwanno et al. [52] compared macroscopic and mesoscopic models in their research while Evans et al. [4] applied microscopic and mesoscopic methods for simulating Bristol and Barcelona Road networks.

Combined models have been developed by integrating different traffic flow models, such as micro with macroscopic models, and micro with mesoscopic models [62,63]. The aim of such integration is to obtain a comprehensive understanding of network properties at different levels of complexity. For example, macroscopic modeling is more appropriate for a simplified node representation, while microscopic modeling is better suited to capture vehicle interactions and drivers’ mutual influences along links. Although the combined method has been proposed to address these issues, only one article [64] in the current systematic literature review was found to have used this approach in traffic simulation.

Praharaj et al. [47] used a different approach compared to the previously mentioned combined models. They employed a data predictive model, which utilizes various input factors to generate traffic volume predictions for each roadway link and period.

3.2.9. RQ8. What Types and Sources of Data Were Used to Parametrize the Models?

The main approach utilized in the analyzed articles of the current systematic literature review is the hybrid method, where the flood model output is utilized as one of the inputs for the traffic model. Additionally, the physical characteristics of the network and traffic demand information are crucial to successfully run the traffic models. Depending on the analysis’ objectives and the desired level of accuracy, various transportation information and databases may be necessary.

Figure 7 categorizes the different variables utilized in flood and traffic models, as identified in the literature. These variables are divided into two categories: flood data and traffic data. Various techniques, such as remote sensing, GIS, hydrological models, volunteered geographical information (such as social media posts, pictures and videos), and crowd-sourced data collection using mobile applications, have been used for mapping the flood vulnerability of road networks. However, the analyzed articles in this systematic literature review mainly focus on the indirect impact of flooding, i.e., traffic disruptions, and hence, have mostly selected pilot zones with a complete historical flooding database. Therefore, in addition to demographic data, flood hazard maps that provide information on flood extent, depth, velocity, and duration are usually applied for running the flood model. Some articles, such as Kim et al. [58] and Gori et al. [64], also use discharge data.

4. Discussion and Future Research Directions

The research presented in this paper reviews the current state-of-the-art of hybrid methods in the field of indirect flood impacts on transportation systems. This systematic review is essential in determining the most effective and efficient approach and framework for developing an automated tool to reduce the costs associated with conducting these types of analyses. The specifics of such an automated tool fall outside the scope of this paper. Future publications will delve further into the details of this proposed automated tool, providing a more comprehensive understanding of its implementation and potential benefits. Nevertheless, new research challenges in the field of assessing the indirect flood impacts on transportation systems are emerging and must be explored in order to improve the sustainability of transportation systems and services. Based on this systematic literature review, several key future research challenges and limitations are presented below.

4.1. Future Research Directions

• Considering various types of flooding: Further research is required to develop efficient strategies for transportation planning that can mitigate the impact of various types of flooding. With the expectation of more severe and frequent flood events, particularly intense storm surges resulting from coastal floods, it is imperative that future studies focus on identifying effective response strategies to this type of flood. Given the complexity of flood disaster modeling, utilizing the concept of a stochastic disaster scenario derived from historical flood data can be an effective approach for transportation resilience planning.
• Applying a broader temporal scale: It must be taken into account that the flood time scale in most of the analyzed floods in previous research ranged from one hour to a whole day. Limited studies have considered the temporal scales of flood analysis compared to the spatial scales. This could be also attributed to the lack of necessary data for estimating the intangible and indirect impacts. Considering the frequent prediction of flood increases in the upcoming years in parallel with the urbanization and transportation network growth, the flood-affected links and the overall resilience of the traffic networks are constantly changing. Therefore, applying temporal resolution in the post-flood analysis ranging from weeks to years may potentially recognize both direct and indirect impacts of flooding on transportation networks. The findings can be used to determine what steps should be taken to improve network resilience and sustainability.
• Applying combined methodologies along with the updated data: To achieve precise results in flood modeling and assessing its impacts on transportation, the most crucial requirement is accurate historical data. In local scales, such as in controlled communities, accurate data of many variables, such as precipitation and traffic data could be measured [12]. However, it is worth noting that the reliability of flood maps for indicating indirect impacts and cascading effects of flooding on transportation systems is subject to significant uncertainties. Population growth and rapid urbanization can lead to changes in land use and landscape alterations that compromise the accuracy of historical flood maps. As a result, flood risk assessments based solely on historical flood maps may not provide a reliable indication of the indirect impacts and cascading effects of flooding on transportation systems. Thus, incorporating updated and reliable data is critical to ensure the accuracy of flood risk assessments, particularly in areas experiencing population growth and urbanization. The use of computational and imaging programs, such as GIS and geospatial technologies, in addition to the available spatial and temporal databases, has advanced flood forecasting and modeling for flood risk assessment [65]. Data, such as precipitation, can be converted into hydrograph for flood estimation using physical methods or mathematical equations. Although researchers frequently used statistical analysis for estimating flood frequency in the past, in more recent studies they usually merge this analysis with other methods using the progress of GIS techniques. This combination leads to the development of combined methodologies that are based on integrating several tools, such as remote sensing and GIS, with several simulations, models, data, statistical analyses, and machine learning methods to assess flood impacts [65].
• Applying a whole system approach: Articles mostly have analyzed the traffic of a specific type of vehicle (passenger vehicles) by partially or fully closing roads, while including different vehicle types, such as public, freight, and passenger, can more accurately assess the increased traffic costs and analyze how they interact and affect each other. In other words, a ‘whole system approach’ is required. The whole system approach involves integrating multiple models and approaches to address the problem of how to facilitate the indirect impacts of flooding on transportation as a whole. Subsequently, network modifications could be provided to decrease flood impacts on transportation systems. Additionally, it is still unclear whether strengthening a few critical links can reduce the overall impact of floods on traffic and more research is required on this issue.
• Analyzing the larger spatial scales: The scarcity of data, uncertainties, and assumptions in modeling techniques make flood and traffic modeling at larger spatial scales more challenging. It is not specified whether it is possible to apply regional traffic models to national-level analysis. Because of daily technological advancements (e.g., availability of mobility data) and computational power, it is expected that the indirect flood impacts prediction of large-scale disruption scenarios become comparatively easy. While future transportation disruption modeling will become more complex and challenging due to the deployment of emerging technologies, such as Connected and Automated Vehicles (CAV) in future mobility systems, cybersecurity issues, and the growing frequency and magnitude of floods.
• Addressing the interdependency modeling: The transportation system is closely connected to other critical infrastructures, such as power and communication networks. Interdependency modeling considers the interconnectedness of these different systems. Although interdependency modeling is a new concept in transportation planning, it will become increasingly important in future transportation system analyses. As information technology and related infrastructure become more essential to transportation systems, they must be considered in an integrated and multidisciplinary context, specifically in case of disruptions, such as flooding situation. The vulnerability of energy infrastructure, such as power outages, must also be addressed.
• Evaluation of other specific traffic parameters: This review reveals that the majority of scholarly literature on the indirect impacts of flooding on transportation systems, published since 2005, has focused on improving the resilience of transportation systems and addressing road networks. As described in Section 3, most of the publications examined in this review concentrated on changes in the travel time of passenger vehicles. In addition to the commonly studied traffic factors identified in this systematic literature review, certain other specific parameters have also been explored as indicators of resilience, including reliability, restoration time, and capacity [12]. These parameters can be employed in future research to evaluate the indirect impacts of flooding on transportation.
• Evaluating the results of the studies: In assessing the studies included in this review, the aim was not to provide a summary of their results, as this was not pertinent to the research questions being investigated. Nevertheless, it can be assumed that the analyzed papers meet an acceptable standard of validation and accuracy, given that they are peer-reviewed articles. To properly summarize the findings in this area, the accuracy of the models could be further examined in future research.
• Noting the hybrid methods’ limitations: Regarding hybrid methodologies, it is crucial to carefully select and construct the considered parameters and variables. To achieve this goal, it is essential to consult with as many stakeholders as possible when developing the models. Although hybrid methods have made a valuable contribution in predicting the indirect impacts of flooding on traffic, there are still several challenges to overcome to maximize their impact. For instance, as each model needs to be parameterized, reliable data, specifically travel behavior data, is necessary. However, obtaining this kind of data can be challenging due to its sensitive and private nature. Furthermore, it is essential to consider that no hybrid method can entirely replicate the real world, and they are only adequate guidance tools for predicting flood impacts. Therefore, it is crucial to conduct sensitivity analysis to ensure the modeling results’ robustness in the presence of uncertainty.

4.2. Limitations of the Current SLR

Hybrid methods in the field of flood indirect impacts on transportation refer to the combination of different data sources and methods, such as quantitative and qualitative data, remote sensing, and GIS. The limitations of the current SLR include:

• Limited availability of studies: As with any systematic review, one of the main limitations is the potential for a limited number of studies that meet the inclusion criteria, particularly when focusing on a specific area, such as hybrid methods in flood indirect impacts on transportation.
• Heterogeneity of studies: Hybrid methods can encompass a wide range of data sources and methods, which may be used differently across studies. This can create challenges for comparing and synthesizing findings across studies, and can limit the generalizability of the findings.
• Quality of included studies: The quality of studies included in a systematic review can have a significant impact on the validity of the review findings. With hybrid methods, studies may use different combinations of data sources and methods, which can affect the quality of the studies included in the review. Nevertheless, since the database for the current SLR was selected from peer-reviewed publications, authors assumed that the quality and reliability of the studies included in the review have already been addressed.
• Publication bias: As with any systematic review, publication bias is a potential limitation. This can occur when studies with significant or positive results are more likely to be published than those with negative or inconclusive results.
• Language bias: As with any systematic review, language bias can be a limitation. Studies published in languages other than those included in the review may be missed, which could limit the scope of the review.
• Time constraints: Conducting a comprehensive systematic review of hybrid methods in the field of flood indirect impacts on transportation can be time-consuming. This may limit the depth of the review, potentially resulting in missed studies or incomplete analysis.
• Difficulty in synthesizing qualitative and quantitative data: Hybrid methods often involve the combination of qualitative and quantitative data, which can be challenging to synthesize in a systematic review. This can be due to differences in data types, measurement scales, and analysis methods.

5. Conclusions

This article presents a systematic literature review (SLR) that summarizes previous research on the hybrid method of evaluating the indirect impacts of floods on the transportation network, which is becoming increasingly important due to the rising incidence of floods worldwide. The review’s scope was clearly defined to answer specific research questions. Relevant databases were systematically searched, and the articles within the defined scope were manually curated against the selection criteria. Furthermore, this study highlights the research gaps in this field.

Climate change is causing an increase in the frequency of different types of flooding, leading to more severe negative consequences. Furthermore, the rapid growth of cities in terms of population, complexity, and size is exacerbating the problem. Collaborative efforts are required to address the interdisciplinary challenges associated with the flooding of transportation infrastructures. To enhance resilience and sustainability, it is necessary to develop new structures and upgrade existing ones. In addition, research is needed to make detailed assessments of the vulnerabilities.

The SLR provided in this article highlights the effectiveness of the hybrid method in assessing the impact of floods on transportation networks. To apply this method on a global scale and at various spatial scales, developing a comprehensive tool based could be a potential avenue for future research in this field. Different adaptations and combinations of the hybrid method should be defined for different types of floods and transportation networks to achieve optimal results.

Additionally, the SLR conducted in this article reveals that the impact of flooding on multimodal transportation systems worldwide has not received enough attention in the academic community, despite their crucial role in supporting the daily lives of the population.

In conclusion, strengthening transportation resilience and sustainability following floods is crucial. The findings of this study can aid government agencies and emergency services in analyzing the indirect impacts of flooding on transportation systems and developing effective management and contingency plans. Furthermore, this SLR is useful in identifying the best options for developing an automated tool.