Flood modelling and flood decision making: a scoping review of the progress of flood technologies and applications

Abstract

Flood intensity and frequency are expected to continue rising due to climate change, necessitating improved prevention measures. Despite the growing body of research, communities and economies remain severely affected by the consequences of floods, underscoring the need to examine how flood models are understood and translated into disaster preparedness and mitigation strategies. Following the Preferred Reporting Items for Systematic Reviews and Meta-Analysis extension for Scoping Reviews (PRISMA-ScR), a total of 824 articles published between 1980 and March 2026 were selected from the Web of Science, Scopus, and IEEEXplore. The review maps advance across several technologies, including hydrological modelling, geospatial analysis, multi-criteria decision analysis (MCDA), and artificial intelligence (AI), and explores how heuristic and metaheuristic approaches couple flood management and analysis. An exponential increase in the adoption of AI and MCDA was identified, driving an exponential increase in flood susceptibility and flood risk mapping across the reviewed literature. This review indicates that modern technologies play a decisive role in advancing flood applications; however, key methodological limitations persist. These limitations include a limited integration of urban infrastructure, stormwater drainage, and groundwater variability into flood prediction and assessment. A lack of coherent methodologies connecting discharge, surface runoff, urban flash floods, and prevention measures remains a significant gap. The translation of rainfall-runoff and discharge prediction into flood analysis is an ongoing challenge. Future research should therefore prioritise integrating runoff potential, discharge modelling, surface water drainage, and groundwater variability into flood applications. A decentralised methodological approach is recommended to empower state and local governments to implement adaptive, context-specific flood solutions, thereby enhancing early warning systems, urban planning, and emergency response. Furthermore, this review surfaces critical questions warranting focused investigation by global research communities: (i) relative importance of surface runoff as a flood-generating mechanism across different hydrological settings, and it weighting in flood susceptibility and risk mapping; (ii) hydrological distinctions and predictive possibilities between rainfall-runoff and observed streamflow modelling, and under what conditions do they best explain observed flooding, and (iii) do runoff calibration reliably explain flood extents and severity in urban settings and what methodological framework can make this operationally viable in data scarce environment?. Addressing these questions can advance the integration of hydrological processes and practical flood prediction and management.

1 Introduction

Climate change has intensified extreme weather events (EWEs) such as heat waves, droughts, floods, cyclones, and wildfires (IPCC, 2014, 2023). Among these EWEs, flooding is one of the most destructive EWEs, resulting in severe loss of life, economic damage and destruction of infrastructure, both direct and indirect (Kumar R. et al., 2023). Flooding is considered a meteorological component (Wedajo et al., 2024) because of excessive precipitation and river discharge (Munyai et al., 2021). Flooding is extensively characterised by a high runoff or excessive water levels more than what a catchment can contain, with its intensity and frequency expected to rise under the influence of climate change, population growth, and urban development (Ding et al., 2024). Climate change increases the frequency of heavy rainfall, which can lead to flooding (Munyai et al., 2021), where unmonitored floods can harm the ecosystem, infrastructure and population (Lapietra et al., 2024) and can adversely cause water pollution (Wedajo et al., 2024).

Floods have a profound impact on both the economy and public health. The economic impact of floods includes the destruction of infrastructure and buildings, the failure of powerlines, and the interruption of the transportation of goods and services. The health impact includes drinking contaminated water; industrial and hazardous waste exposure; injuries and trauma caused by destruction of the buildings, falling trees, falling powerlines and debris (Agonafir et al., 2023). In addition, the impact of flooding can be both direct and indirect. Where indirect impact accounts for the interruption of transportation of services from one point of the market to another across the directly affected region (Kumar R. et al., 2023). Most economic impacts can last for several years after an event, including business interruption and reduced tourism. Among other impacts, flooding can be linked to other natural calamities such as landslides, mudflows, and soil loss (Grab and Nash, 2023). Although flood events cannot be entirely prevented, their impact can be mitigated through a proactive strategic planning (Munyai et al., 2021), such as identifying vulnerable populations and infrastructure.

Whilst flooding can be viewed as the overflow of a confined stream, water or other body of water or the accumulation and overflow of water over an area that is usually not submerged (Membele et al., 2022; Eslamian and Eslamian, 2022). There are various types of floods: “river (fluvial) flood,” “flash (pluvial) floods,” “coastal floods,” “inundation,” “urban floods,” “dam, lake outbursts” (IPCC, 2014) and “groundwater floods.” For instance, a flash flood is a sudden, short-lived torrent that develops in areas without streams, often after several hours of rainfall or following a dam or levee failure. In contrast, inundation events slowly develop over hours or days, whereas flash floods occur suddenly (Eslamian and Eslamian, 2022). These types of floods commonly manifest as river, runoff, coastal, estuary, outburst, and urban flooding in basements. Floods have a complex underlying physical processes constrain data and models leading to significant uncertainty (Kumar R. et al., 2023). As a result, many communities remain vulnerable, lacking the resources and preparedness needed to effectively reduce flood risk and recover from flood damage and disease outbreaks after an event (Govender et al., 2025). However, important and significant strides have been made in flood management, with researchers developing strategies focused on prevention, mitigation, warnings, response, and recovery (Amitrano et al., 2024). This approaches relies heavily on various technologies such as hydrological, artificial intelligence (AI), rainfall-runoff modelling, and geospatial tools.

At present, there are numerous reviews on flood modelling and simulations. Among these reviews, Devia et al. (2015) attentively reviewed hydrological models, detailing hydrological approaches including empirical, conceptual and physical models. Their study discussed briefly about variable infiltration capacity, soil and water assessment tool (SWAT), topographic-based hydrological model (TOPMODEL), hydrologiska Byrans Vattenavdelning model (HBV), and MIKE system hydrologique European (MIKESHE).

Tariq et al. (2020) reviewed flood risk and the selection of suitable measures and attentively demonstrated the significance of understanding and controlling flood impact through risk, vulnerability, susceptibility, hazard, and exposure. Their study posed critical questions about flood risk assessment and its place in modern concepts, theory, and practice. Giving a detailed analysis of flood risk perception, risk strategies, risk reduction, risk-based concepts and designing an effective scheme for the floodplain. Their study draws an adverse conclusion about the characteristics of three rainfall-runoff (hydrological) models based on empirical, conceptual, and physical models, without thoroughly examining other methods and flood applications.

Kumar R. et al. (2023), comprehensively reviewed flood modelling approaches of recent advanced techniques and attentively described the workflow and process of each technology, ranging from hydrological-hydraulic modelling, numerical modelling, rainfall-runoff modelling, remote sensing and GIS-based flood modelling, application of AI in flood modelling, multi-criteria decision-based flood management, heuristic, and metaheuristic flood management models. Kumar R. et al. (2023) strongly focused on detailing methods, algorithms, their advantages, and disadvantages.

However, despite contributions to the growing body of knowledge, there is a lack of a clear synthesis of the connections and linkages between methods and flood applications. Our review is positioned to determine the linkage between methods and applications.

This review aimed to answer the following:

• (i) Which methods are used in each flood application?

• (ii) How is flood discipline fragmented?

• (iii) Where is integration missing?

• (iv) Which parameters are used for each application?

• (v) What is missing in the holistic applications approach that can enhance a better understanding of the field?

By that, the study examines various flood applications, including flood detection and inundation mapping, susceptibility mapping, risk mapping, vulnerability, future forecasting, rainfall-runoff modelling, and flood simulation, with emphasis on methods like AI, geographical information systems (GIS), remote sensing, heuristic, metaheuristic, multi-criteria decision analysis (MCDA), and hydrological tools.

To answer aim (v), we used the screened document and a literature search to find the key impacts of flood modelling that are often not accounted for in many studies. This includes groundwater impact on flooding, surface water and misalignment of urban stormwater drainage. The study also identified that changing environment is a key factor to runoff, while rainfall remains a key driver for flooding. Despite methodological advancements and study framework, this coping review remains limited in addressing how surface runoff, rainfall-runoff modelling, and charge calibration relate and explain observed flooding. There is a need for future studies to address the critical role of surface runoff in flood generation; an explicit relationship between runoff depth and streamflow; and how calibrated runoff and discharge best explain flooding in urban settings.

2 Overview of technologies

2.1 Hydrological and hydraulic

The hydrological and hydraulic model described by Devia et al. (2015) serves the purpose of simulating stormwater and surface water runoff (Agonafir et al., 2023), and is concerned with the occurrence, movements, distribution, chemical and physical properties of the earth’s water, mainly the interaction of water, environment, and living organisms in particular (Jehanzaib et al., 2022). These models simulate water movement in the watershed, including infiltration, runoff, and evapotranspiration (Jyrkama and Sykes, 2007). Water flow can be modelled in 1D, 2D, or 3D, depending on the river channel and floodplain (Pinos and Timbe, 2019). The hydrological technique aims to investigate each stage of the hydrological cycle to understand the interconnections among components, for example, the association between river flow and precipitation (Jehanzaib et al., 2022). To understand the hydrological system, hydrological modelling employs mathematical-based flood simulation to predict water during a floods (Herrera et al., 2022). Hydrological modelling employs several factors, including topographical and meteorological factors such as elevation, land use, precipitation, evapotranspiration, and soil moisture content (Kumar R. et al., 2023).

Hydrological models are divided into three main categories: empirical, conceptual, and physically based models, each with its own benefits and disadvantages. The empirical models are mathematical equations derived from input and output time series and are observation-oriented. They are considered data-driven and do not account for prevailing hydrological processes; they are typically developed using statistical principles that utilise regression and correlation to determine relationships between input and output variables. Empirical approaches include ML techniques, whereas conceptual models or parametric models describe all components of the hydrological system. These models bring to attention the interconnection of physical elements of a catchment, in which they are recharged by rainfall, infiltration and percolation and discharged or emptied by evaporation, runoff and drainage (Kumar et al., 2023b). These models are semi-empirical equations in which model parameters are assessed and augmented using field data. Examples of such models include the Stanford watershed model (SWM).

The physical-based approach is a mathematical idealised model that is often regarded as a mechanistic technique that employs principles of physical processes. Unlike conceptual models, the physically based approach does not necessarily require extensive hydrological and meteorological data for calibration; instead, it evaluates parameters that describe the catchment’s physical characteristics.

2.2 Geospatial technique

Geospatial techniques are a unique problem-solving technology with a remarkable influence in flood mapping, detection, monitoring and management and are concerned with spatial analysis and mapping, which have proven essential for identifying and prioritising flood-affected areas (Ahmed et al., 2024), using mathematical and statistical principles. The geospatial approach is the backbone of hydrological simulations and is used across several phases to understand hydrological systems.

For effective flood modelling, GIS and remote sensing have been coupled with other technologies such as AI, MCDA, and numerical models. The geospatial technique significantly serves in various cycles of the flood management approaches, for instance, in mapping the affected regions, identifying the areas at risk of flooding before disastrous events, remote sensing of the affected areas, and planning the mitigation routes and strategies, given that GIS and remote sensing have the potential to handle, analyse and visualise geospatial data using various software’s (Kabenge et al., 2017). Comparable GIS and remote sensing are effective; for instance, satellite images of affected regions can be valuable for mitigation and recovery and can provide timely, cost-effective information for mapping and monitoring flooded areas. The evolution of RS has enabled near-real-time monitoring through diverse satellite constellations orbiting Earth.

Remote sensing satellite constellations provide detailed information about Earth’s surface, which can be comprehensively analysed to identify key aspects of an event (Sajjad et al., 2022). These satellite constellations are equipped with passive or active sensors with different functional capabilities. For instance, active sensors provide their own illumination; passive sensors depend on external sources to illuminate targets. Passive optical multispectral sensors have limited night-time observation, unlike active sensors, such as microwave SAR radar, which are renowned for penetrating through cloud cover and providing all-weather observations (Tripathi et al., 2021). Besides its ability to penetrate clouds and provide all-weather observation, the radar can effectively distinguish between water and non-water pixels based on differences in backscatter. Although satellite systems offer unique data collection techniques, they are limited to spatial, temporal, and radiometric resolutions. Using low-temporal-resolution sensors results in the omission of short-duration flood events. Very high-resolution satellites are mainly for commercial use, with their spatial resolution that can effectively distinguish flooded areas. Furthermore, satellite images captured by optical multispectral sensors are often cloud-covered during flood events, because optical sensors and have limited ability to penetrate clouds (Tripathy and Malladi, 2022).

2.3 Multi-criteria decision analysis

Multi-criteria decision analysis (MCDA) is a widely used approach that has proven its effectiveness in flood mapping, together with various flood conditional factors, assigning the weight and rank of each conditional factor (Seydi et al., 2022). When confronted with a complex and conflicting decisions, MCDA can offer effective and well-informed solutions and consider both qualitative and quantitative approaches (Kumar R. et al., 2023). MCDA can efficiently evaluate and compare various flood management alternatives, including structural and non-structural measures, based on criteria, to minimise the risk of flood damage or reduce the effect of flood events (de Brito et al., 2019).

There are two broad categories of MCDA, which are: multi-attribute decision-making (MADM), and multiple-objective decision-making (MODM; Gebre et al., 2021). MCDA are classified into several different classes (1) value/utility function methods like multi-attribute value theory, and multi-attribute utility theory; (2) pairwise comparison methods like analytic hierarchical process (AHP); analytic network process (ANP); (3) outranking techniques like (ELECTRE, PROMETHEE); (4) distance-based methods such as technique for order preference by similarity to ideal solution (TOPSIS), weighted aggregated sum product assessment (WASPAS); and (5) fuzzy decision-making methods like multi analysis decision making (MADM; Kumar R. et al., 2023).

2.4 Artificial intelligence

The evolution of artificial intelligence (AI) involves machine and deep learning techniques that utilise mathematics and statistical methods, enabling computers and machines to simulate human intelligence and problem-solving capabilities through learning from past events and experiences. The AI is key in addressing complex tasks and solving complex situations through learning curves, without explicit programming. This technology excels in various spatial analyses, prediction, forecasting and flood mapping. It is widely used for flood simulation, flood future forecasting, data analysis, risk assessment, and mitigation planning. They are implemented as a complement to physics-based models (Agonafir et al., 2023).

ML algorithms are widely employed to analyse large amounts of meteorological, hydrological, and topographical data. ML algorithms are given learning tasks and learn to accomplish them through past experiences (Liakos et al., 2018). There are four categories of ML approaches: supervised, unsupervised, semi-supervised, and reinforcement learning (Kumar et al., 2023a). ML can be viewed as a set of functional algorithms, which typically include regression, classification, clustering, and time series analysis (Kumar R. et al., 2023).

However, ML algorithms are still in the early stages of development and are limited by the need for data from multiple sources, including meteorological, topographical, and hydrological data. Integrating different formatted data from various sources can be challenging, often resulting in errors and misinterpretations (Kumar R. et al., 2023). These data-driven models analyse the relationships between prior occurrences of hazardous events and environmental variables to understand their correlations.

2.5 Heuristic and metaheuristic

Heuristic and metaheuristic models were developed in the early 1970s (Kumar and Yadav, 2022), mainly to overcome the shortcomings of the traditional optimisation method, including linear (LP), non-linear (NLP), and dynamic programming (DP; Kumar and Yadav, 2018). The sequential evolution of conventional optimisation techniques from LP is one of the simplest and most popular optimisation approaches. LP is limited to linear equations and constraints. Following LP, NLP was developed to overcome LP’s major drawback: its inability to solve nonlinear problems. The NLP was found to have a drawback when the problem dimensionality increased, and DLP was developed to address stochastic and nonlinear problems. Early optimisation methods were applied to issues related to flood control, flood warning, and risk management, and their limitations, including the inability to quickly solve problems, were addressed. These brought the evolution of heuristics and meta-heuristics to promptly solve problems that traditional optimisation methods would delay in solving (Kumar and Yadav, 2022).

The renounced heuristic techniques are aimed at solving problems quickly when a traditional approach is slow. Metaheuristic techniques are more advanced for choosing or deriving heuristics that can address optimisation problems (Bellos and Tsakiris, 2016). Broadly, heuristics and metaheuristics are classified into population-based and neighbourhood-based algorithms. Population-based algorithms are adaptable to swarm intelligence and evolutionary algorithms (Kumar and Yadav, 2022) and can offer a global solution (Kumar R. et al., 2023).

Nonetheless, the increasing potential of open-source packages such as R, SciKit-Learn, TensorFlow and cloud computing like Google Earth Engine and Google Colab has intensified the implementation of AI (Mehmood et al., 2022). Open-source packages facilitate data sharing and build communities focused on solving various phenomena. Along with free availability and open data access, researchers are extensively involved in investigating multiple problems more efficiently than ever before.

3 Materials and methods

3.1 Scoping review

A scoping review is considered an ideal tool for determining the scope and coverage of the literature on a given topic. It attentively gives a clear indication of the volume of literature and its broad overview (Munn et al., 2018). Munn et al. (2018), acknowledged that scoping reviews are built on the purpose of (i) identifying available evidence, (ii) clarifying concepts, (iii) examining research frameworks in the field, (iv) identifying and analysing research gaps, and (v) as a “precursor to a systematic review.” A scoping review followed a Preferred Reporting Items for Systematic review and Meta-Analysis extension for Scoping Reviews (PRISMA-ScR) framework, which focuses on synthesising the scope of evidence rather than assessing the quality of studies (Tricco et al., 2018). Traditionally, reviews are fundamental because they can be used to analyse the entire intellectual structure of studies (Morante-Carballo et al., 2022). This is done by collecting all relevant and possible studies on the topic, analysing and presenting the findings (Ahn and Kang, 2018) in a structured and transparent manner.

This study follows a scoping review methodology to identify knowledge gaps, clarify concepts, and map flood models and their applications. Three databases were searched: Elsevier’s Scopus, Clarivate Analytics’ Web of Science (WoS), and IEEE Xplore. Web of Science and Scopus databases have proven to be valuable components of modern search ecosystems (Irbık, 2022; Madzivanyika et al., 2026). However, they do not cover as wide a range of studies as Google Scholar.

3.2 Search criteria and strategies

The study defined a biased search within databases (Table 1). The review searched peer-reviewed journals published between 1980 and 30 March 2026. This captures the changing trends in methodologies from the historical context of the 1890s, the development of computational models between the 1980s and 2000s, and the present-day advanced technologies from the post-2000s.

We applied the search strategy for title (TI) and abstract (AB) [TI = (“flood susceptibility” OR “flood risk” OR “flood risks assessment” OR “flood mapping” OR “flood modelling” OR “flood simulation” OR “flood prediction” OR “flood forecasting” OR “flood delineation” OR “flood vulnerability” OR “flood detection” OR “flood change detection” OR runoff)] AND [AB = (machine learning” OR “geographical information system” OR GIS OR “remote sensing”)] AND [AB = (“multi-criteria” OR “land use and land cover” OR groundwater OR “surface water” OR runoff OR slope OR DEM OR topography OR rainfall OR precipitation OR drainage OR “climate change”)] (Supplementary Table A1).

All studies that did not meet the criteria were excluded from the first screening (title and abstract); the same approach was applied in the second screening (full document), which screened all the documents included in the first screening. All documents without a DOI were excluded from the screening process. Duplicate studies were removed in EndNote by comparing DOIs and authors. The non-duplicate articles list was exported to Excel for screening and data collection. The first screening was performed in Excel, and the secondary screening was performed in EndNote, where the documents were searched. Articles that were not found in EndNote were searched using the DOI and title.

We compiled the data from all relevant studies in Excel for data extraction. The authors, the year of publication, title, DOI, study area, method, section, sensors and approach were extracted. Additionally, Excel, R-Studio (v4.4.1) and Flourish (https://app.flourish.studio) were used for data visualisations.

3.3 Study selection

This review methodology followed the PRISMA-ScR framework to map the concepts, methods, and gaps in flood management research rather than to assess the quality of studies. This scoping review did not follow a fully compliant PRISMA, which is often used for systematic reviews. The PRISMA-ScR was used to collect studies for analysis, following the framework (Figure 1).

Figure 1

The study followed the exclusion and inclusion criteria, defined in Table 1. This involves screening the title, abstract and full document. While the screening was conducted by a single author, the decision regarding the methodological approach and selection criteria was rigorously reviewed by other authors.

This was done to account for studies in the flood management domain by looking at the method’s key words (AI, hydrological models, remote sensing and GIS), together with aspects of flood modelling (susceptibility, detection, flood forecasting, rainfall-runoff, vulnerability and flood risk). After selection of paper based on the title and abstract, a full document screening was done following questions such as (i) which methods are followed, (ii) which discipline the paper fragment, and (iii) which methods are integrated? However, these methods introduced a potential bias in selection by a single reviewer. This paper did not assess any agreement measure.

After data collection, duplicate removal, and screening, a total of 824 studies were eligible for final analysis (Figure 1). In addition, we extracted information and parameters (Supplementary B), and tried to identify the gap by looking at what has been identified as a key impact to flooding, but not widely explored (Section 5.1).

3.4 Article classification

Since this study aimed to map a relationship between methods and applications. Classification and categorisation of articles were critical. Firstly, the study identified methods, linked them to applications, and assessed their integration (Figure 2). Following those classifications, we further separate integrated methods and applications to investigate how each discipline is fragmented. For instance, other studies applied geospatial technology and a hydrological model for flood detection and rainfall runoff modelling. Such a study was categorised into two methods: the geospatial and hydrological models. This case shows that the geospatial technique was applied for flood detection, and the hydrological technique was applied for rainfall-runoff application. In summary, in this case, we counted a single study as two different applications.

Figure 2

We reviewed the flood application based on these principles:

• Flood detection: This function identifies and maps the extent of flooded areas.

• Flood simulation: Function of predicting, modelling, simulating rainfall, and delineating flood models (recreation of flood scenarios).

• Flood risk: Evaluate the vulnerability, exposure, and hazards, assessing potential impacts on human life and infrastructure at risk.

• Flood vulnerability: Measuring the likelihood of harm to infrastructure or populations, considering economic, social, and physical factors. It is a function of susceptibility, exposure, and resilience.

• Flood susceptibility: Assessing an area’s likelihood of being flooded based on historical inundation and conditional parameters.

• Rainfall-runoff: Scrutinising the rainfall-runoff relationship by assessing the transfer of rainfall to surface runoff.

• Flood forecasting: Generating quantitative information in time and space to predict future scenarios. Forecasting rainfall, simulation, susceptibility and risk.

For this scoping review, data were summarised based on the application of technology to flood analysis. After extracting the data, we identify the parameters used for each flood analysis and the gaps in what has not been widely explored and is critical to flood analysis.

4 Results and analysis

4.1 Research and publication trend

Based on the the search and screening criteria, this review did not identify any studies published before 1991. However, this review highlights a gradual increase in research focus on flood management approaches since 1991, with the topic gaining attention in 2012 (7 articles) and experiencing a significant increase from 2015 (12 articles; Figure 3). From 2015, there has been an exponential increase, with greater attention and relevance in 2022 (86), 2023 (98), 2024 (144), and 2025 (191). This pattern is also expected in 2026, for instance, the 2026 search included January to March, but there have been 65 publications, which is a record never achieved before 2021. This trend indicates the increase of relevant articles based on the review scope. Also, Diaconu et al. (2021) noted an increase in publications from 2000. The growing attention to flood scope is driven by more frequent flooding events and the resulting visible destruction worldwide, which are prompting greater scrutiny and requiring additional evidence to mitigate flood impacts.

Figure 3

Historically, geospatial and hydrological contributed 100% in the studies, without integration to other technologies. However, evidence of this review shows that researchers are coupling various technologies to enhance the novelty of flood assessment studies. Figure 4 illustrates that while geospatial technology remains applied, MCDA and AI are taking the space in flood susceptibility, risk and forecasting, while hydrological models remain applied for flood modelling, rainfall-runoff and runoff simulation. Figure 4, indicate that from 2014, AI studies have accounted from 14 to 51% of the overall publications, and further integrated with heuristic, hydrological, geospatial and MCDA. Geospatial technology remains applied for flood detection, where researchers rely on its function to find flooded areas before assessing flood risk, susceptibility and other applications, showing that geospatial technology is now widely applied solely.

Figure 4

Figure 5 simplifies the contribution of each technology in flood applications by scrutinising the methodological applications in Figure 4. There is a shift from traditional geospatial and hydrological models toward advanced MCDA and AI-based approaches, with heuristic methods gaining attention; this shift is evident from 2012 to 2015 to the current date. The dynamic shift from computationally intensive hydrological and hydraulic models is associated with the evolution of technologies, which bring the availability of free datasets and sophisticated methodologies. Not only had the flood analysis approach shifted between technologies, but this sophisticated transformation also led to the development of multifaceted approaches and multifactor analyses to understand flood dynamics. As a result, MCDA and AI-based approaches yield practical approaches such as risk analysis, vulnerability, and susceptibility, which assume that flooding is influenced by specific parameters, or “conditional factors.”

Figure 5

4.2 Global trends

Since floods have no political, geographical, economic or social boundaries, their occurrence and impact have been declared in several nations and studied globally (Garrote, 2022). As demonstrated in Figure 6, Asian countries such as India (131 articles), China (120), Iran (60), Pakistan (31), the United States of America (24), Bangladesh (23), and Vietnam (22) are leading in flood management and research publications. On the Asian continent, flood severity and occurrence have been devastating and destructive (Chen et al., 2022). Global nations such as Canada and African countries such as Ethiopia, Egypt, Nigeria, and Algeria are contributing to flood research due to the increasing incidence (Danhassan et al., 2023). The frequency of flooding has heightened researchers’ attention to the issue. It is essential to note that countries have different priority levels for flooding, which is reflected in the amount of research on ongoing flooding. Although countries like Mozambique and South Africa have historically experienced several floods, the topic has not yet attracted much attention.

Figure 6

Nonetheless, the inclusion criteria for English-language publications can distort these results. This accounts for publication in other languages rather than English.

4.3 Application of flood technology to flood analysis

Among these flood applications, flood susceptibility has received the most attention, followed by flood forecasting, flood detection, flood modelling, flood vulnerability, and rainfall-runoff modelling. Flood susceptibility has been explored mainly by AI and MCDA; flood risk by MCDA and AI; flood forecasting by AI; flood detection by geospatial methods alone; flood modelling and rainfall-runoff by hydrological and hydraulic methods; and vulnerability by MCDA, as demonstrated in Figure 7.

Figure 7

4.3.1 Flood detection

Flood detection delineates flood-inundated areas, depicting the extent and severity of flooded regions. Flood inundation maps give discrimination between flooded and non-flooded areas and are useful for flood mitigation strategies, such as land use planning, infrastructural design, emergency response and planning (Sanders et al., 2020). Charting of flood is essential for other disciplines of flood mapping, including susceptibility, vulnerability, risk and forecasting, especially for MCDA and machine learning models. Remote-sensing satellite images are an affordable means of analysing data to detect flooded regions. This approach is achieved through methods such as comparing the non-flood and flood image (pre- and post-event) and image segmentation, which can classify all water pixels in an image, including those from unmanned aerial vehicle (UAV) and digital camera imagery.

To map flood inundation extent, the research literature reveals that numerous methods and techniques have been used. Among these methods, flood change detection is the most widely used for flood inundation mapping. Pre-processed data are applied to image analysis methods such as thresholding and image segmentation, to identify affected areas (Kumar R. et al., 2023). The thresholding approach defines the area as either flooded or non-flooded based on a threshold value. A wide range of thresholding approaches has been investigated, including Otsu methods, Kittle and Illingworth, and the fixed method, which involves setting a value and examining the change in backscattering coefficient or sourcing a threshold value from literature (Amitrano et al., 2024). The main aim of thresholding is to distinguish between water and non-water pixels, and these thresholding methods aid in making those decisions. Other methods for detecting flooded areas include assessing changes in water indices between pre- and post-event images or using ML and DL for classification and segmentation. Such methods include the likelihood ratio, normalised band ratio and multi-temporal change estimators. As proposed by Amitrano et al. (2024), normalised band ratios include the temporal water index (TWI) and absolute change violet estimator (ACE).

Existing knowledge highlights that attention has been given to AI due to its ability to delineate flooded zones. Among AI algorithms, support vector machine (SVM) and random forest (RF) have been the focus, with little attention to neural networks (NN) and numerous other ML methods. Among the geospatial approaches, the threshold technique is frequently applied (Figure 8).

Figure 8

The evidence reveals that synthetic aperture radar SAR sensors such as ALOS-2 PALSAR-2, COSMO-SkyMed, European Remote Sensing Satellite-1 (ERS-1) and ERS-2, RADARSAT, TerraSAR-X, Sentinel-1, and Gaofen-3 have been used for flood detection (Figure 9), with Sentinel-1 attracting particular attention due to its availability and coverage. For instance, the Sentinel-1 group of Sentinel-1A launched in 2014, and Sentinel-1B in 2016, with a spatial resolution of 10 m, and global coverage in 12 days, which was split into 6 days for Sentinel-1A and 6 days for Sentinel-1B (Tripathy and Malladi, 2022), lifted the drawback of ALOS PALSAR and RADARSAT-1, which had a temporal resolutions of 35, 46, and 24 days, respectively; however, Sentinel-1B retired in 2022, making it challenging to achieve improved revisit.

Figure 9

Together with SAR, ground-based sensors such as a camera, and numerous satellite-based optical sensors such as Landsat Thematic Mapper (TM), Enhanced Thematic Mapper (ETM) and Sentinel-2, remain popular for flood mapping during cloud-free images. Optical sensors are renowned for land use and land cover monitoring and vegetation mapping, and systems such as IKONOS, Pleaides-1A, and PlanetScope are considered high-resolution imaging systems. When choosing a satellite image to map flood inundation, it is important to select the image based on its value such as temporal resolution, where images close to the peak of an event are more reliable in deriving flood extent.

4.3.2 Flood susceptibility

Flood susceptibility is the likelihood of an area being flooded under given conditions, based on comparisons of flood-related parameters, including geological, meteorological, and digital elevation model data, as well as historical and hydrological flood occurrences. Flood conditional factors are chosen based on expert knowledge and published literature. Flood susceptibility maps (FSMs) highlight the environmental drivers of flooding and provide decision-makers involved in flood mitigation and prevention planning with informative guidance. FSM can be grouped into several approaches, including ML, DL, qualitative-based MCDA, hydrological-based, and statistical techniques (Seydi et al., 2022), allowing areas prone to flooding to be identified by probability or qualitatively, based on measurement levels. Flood susceptibility modelling often does not consider the date of an event (Fell et al., 2008).

Methods such as RF, SVM, frequency ratio (FR), ANN, and NB have gained popularity. MCDA methods such as AHP, fuzzy AHP (FAHP), ANP, VIKOR, and TOPSIS have been adopted. There has been a growing number of studies utilising the heuristic method Dingo Optimisation Algorithm (DOA), Weighted Chimp Optimisation Algorithm (WCOA), Particle Swarm Optimisation (PSO), and Hybrid Black Widow Optimisation (HBWO). Results reveal that flood susceptibility can be modelled using a wide range of expert knowledge, MCDA, and AI technologies, each with distinct aspects and numerous flood-causative factors, as demonstrated in Figure 10.

Figure 10

Numerous conditional parameters (Supplementary Table B1) have been coupled with labelled flood data to determine the extent and spatial patterns of flood-prone areas. When applying a model, the significance of each parameter is usually assessed using techniques such as variable importance.

4.3.3 Flood vulnerability

Although “susceptibility” and “vulnerability” are used interchangeably, they hold different meanings. Where susceptibility generally refers to the likelihood or probability of a flood happening at a specific location (Pugliese Viloria et al., 2024). Vulnerability can be viewed as a response to vulnerable values, such as population or infrastructure. Vulnerability refers to the probability that infrastructure, physical, and capital assets, and human beings will suffer harm and loss when a flood event hits (Abdrabo et al., 2023). In other terms, it can be viewed as the degree of susceptibility to damage in an area or population and mainly depends on human resistance ability (Eini et al., 2020), considering the conditions that make the location vulnerable to floods, like social, physical, and economic (Abdrabo et al., 2023), and environmental factors (Corominas et al., 2014), highlighting different impacts (Munyai et al., 2021). For example, people living near rivers on floodplains, or by bridges, can be vulnerable to flooding.

Flood vulnerability can be computed by integrating flood exposure, susceptibility, and resilience. As illustrated in Equation 1 (Tariq et al., 2020). Nonetheless, where the zone is not susceptible to flood, and there is no exposed infrastructure or population, then vulnerability is considered very low or zero.

Flood resilience is defined as a system’s ability to return to equilibrium after certain disturbances (Tariq et al., 2020), or the ability to face and recover from hazards (Asian Development Bank, 2013). Three main components, namely exposure, susceptibility, and resilience, have different values regarding vulnerability. Integrating exposure enables to determine how flood events can disrupt a system based on its location. The exposure assesses the extent to which assets, such as goods, infrastructure, cultural heritage, and agricultural fields, are exposed to flooding. The resilience principle applies to locations that have previously experienced flood events (UNESCO-IHE, 2024).

Highly vulnerable areas are highlighted by a combination of a location’s susceptibility or sensitivity to flood hazards and a lack of capacity to respond and adapt. The focus of economic vulnerability due to flooding is the destruction of buildings. Some buildings are more vulnerable than others. Whereas social vulnerability draws attention to communities’ reactions and resistance to flood events. The social vulnerability criteria include demographic characteristics demonstrating citizens’ resilience to flooding. The setting of social factors depends on the demographic characteristics of the city, the degree of urban development and the available dataset (Eini et al., 2020).

Therefore, understanding the difference between the terms “susceptibility” and “vulnerability” is crucial for flood management analysis. A highly susceptible region can also represent a very low vulnerability, for instance, due to the absence of people and infrastructure. Hence, vulnerability modelling becomes essential.

Among MCDA-based methods, AHP has been attentively used, and ANP is gaining momentum, while other algorithms, such as Fuzzy and decision-making trial and evaluation laboratory, have not been widely used. Various geospatial techniques have been adopted, for example, morphometric analysis and normalising the index. AI, principal component analysis (PCA) and FR have been used in vulnerability analysis. While hydrological models have only attempted the soil conservation service curve number (SCS-CN) model, in Figure 11. Results reveal a need for AI and MCDA modelling of flood vulnerability, given their applicability to flood susceptibility, and that numerous algorithms can be employed.

Figure 11

Along with vulnerability methods, causative parameters are significant in deciding the classification of vulnerable zones. Among these parameters, attention has been given to measures of exposure, such as drainage density, slope, elevation, LULC, rainfall, soil texture, and sensitivity measures, such as population density and average income (Supplementary Table B2). A fundamental understanding of vulnerability concepts enables the use of diverse measures to assess exposure, sensitivity, and adaptive capacity. Other studies have highlighted additional parameters, including exposure parameters such as the groundwater table and aquifer type; flow accumulation; sensitivity parameters such as building conditions, building typology, children’s population health care facilities, and low-income households; and adaptive capacity parameters such as early warning information, emergency care, evacuation drills and training. Most of these parameters can be used for susceptibility and vulnerability, since vulnerability is a function of susceptibility, and some terms can be used synonymously across studies, such as health care facilities and hospitals.

4.3.4 Flood risk

Flood risk is a function of flood hazards, exposure, and vulnerability. Assesses the potential for flood occurrence and the resulting impact on human lives and the environment (Kron, 2005). Flood risk helps simulate the expected degree of loss during flooding (Ali et al., 2016) and plays a decisive role in reducing flood damage (Eini et al., 2020) by assessing the environment and the inhabitants at risk of flooding. A flood risk model is widely used to alert decision-makers and inhabitants in the risk zone to the dangers and impacts of flooding. The risk principle can be used for urban planning and design to plan and develop infrastructure to control floodwater and prevent severe impacts. Flood risk emerges from “flood hazard,” a convolution of the probability of flood inundation, and “vulnerability,” a likelihood of associated negative consequences of floods (Ali et al., 2016; Ibrahim et al., 2024) (Equation 2). While risk is exacerbated by a high degree of social vulnerability and a limited coping capacity (Ikram et al., 2024). In flooding, risk is often conceptualised as a product of a hazard and its consequences (Kron, 2005) or as a degree of potential consequence associated with a hazard (Tariq et al., 2020). Flood risk assessment comprises four main steps: describing the area, determining hazard levels and intensity, and assessing vulnerability and risk.

When determining a flood risk map, three fundamental principles are derived from hazard, exposure and vulnerability before comprehensive flood risk management (Hirabayashi et al., 2013). Flood risk is considered zero where there are no hazards and susceptibilities.

Where FR is flood risk in the given instance, the probability of loss from flooding depends on the exposed elements. The flood risk map is derived from hazard, exposure and vulnerability maps (Pugliese Viloria et al., 2024). Flood hazard (FH) maps express a function of probability and intensity (Tariq et al., 2020), illustrated in Equation 3, or alternatively, the magnitude and frequency of such events (Schmidt et al., 2011). Considering that “probability” refers to likelihood or chance, “intensity” is the severity or force of an event, “magnitude” is size or extent, and “frequency” defines how often an event of magnitude occurs. These maps provide crucial information about an event’s water depth, extent and return period (Bentivoglio et al., 2022).

Since the consequences are people or infrastructure, and probability is the likelihood of a flood event, different regions can be categorised by varying levels of risk; if the probability is high and the consequences are high, this suggests a high-risk zone. Therefore, when classifying flood risk zones, it is significant to fully assess the degree of factors like exposure, which generally refer to the presence of infrastructure or population at the location (Corominas et al., 2014) and hazard, which is defined as the risk of harm, loss, or damage from flood occurrence (Ikram et al., 2024), at different magnitudes (Crichton, 2002). Flood depth is significant in determining flood risk indices (Ghent, 2013). Potential flood damages depend on the flood depth, duration, velocity, and impulse, which are the products of water level and velocity, the rise of water levels, and frequency of occurrence (Dang et al., 2011). Like other approaches, risk is hypothetical; for instance, a population located near a major river experienced a high flood risk (Ikram et al., 2024).

Among ML algorithms, random forest has been widely applied, followed by support vector machines and Naïve Bayes (Figure 12). Various ML algorithms, such as gradient boosting machines (GBMs) and categorical boosting, are gaining momentum. AHP-based MCDA has been widely explored for flood risk mapping. Following AHP, Fuzzy AHP, and TOPSIS have also been widely recognised, as demonstrated in Figure 12. Hydrological and Hydraulic modelling has gained attention for flood risk, with HEC-RAS and HEC-HMS widely utilised. Attention has shifted to deriving flood risk using geospatial and statistical methods coupled with AI. The overall results point out that the evolution of technology shows a growing focus and interest in AI and MCDA; these technologies are applicable for identifying flood risk zones (Rafiei-Sardooi et al., 2021).

Figure 12

Studies have incorporated various vulnerability, exposure and hazard parameters to assess the flood risks. Study findings highlight exposure parameters, including elevation, slope, LULC, rainfall, soil texture, and distance from the river (Supplementary Table B3).

4.3.5 Flood simulation (rainfall-runoff simulation)

The concept of “flood simulation” is sometimes used interchangeably with “rainfall-runoff simulation,” which simulates the behaviour of rainfall at Earth’s surface and is separable because of its functionality. While simulation considers water depth, flood extent, velocity, prediction and inundation map, rainfall-runoff focuses on converting rainfall to runoff, although it can be extended as a function of discharge. For this study, we separated flood simulation from rainfall-runoff. While we referred to rainfall-runoff as a model predicting runoff, we extended it to discharge, particularly hydrograph analysis. The fundamental results of flood simulation include flood extent, depth, velocity, hazard, risk, and inundation maps; this extends beyond the extent of runoff.

Rainfall-runoff models and flood-simulations are described as a series of mathematical principles that help to quantify the amount of rainfall that can be converted into runoff as a function of several parameters used to characterise watersheds, like soil properties, vegetation cover, topography, soil moisture content and characteristics of the aquifer (Amitrano et al., 2024). The classification of these models is based on the inputs parameters, as well as the extent to which physical principles are applied. Consequently, they can be classified as lumped or distributed models based on model parameters that depend on space and time, whereas deterministic and stochastic models are based on criteria (Devia et al., 2015).

Flood simulations have been widely analysed using hydrological techniques to determine the extent of inundation. Flood simulation has long been used to understand the spread and impact of flooding, long before detection, susceptibility and risk. The simulation derivation is computer-intensive and primarily uses a DEM, land-use and land-cover hydrological soil groups (HSG), curve numbers, and precipitation data. Hydrological technologies such as HEC-HMS, HEC-RAS, LISFLOOD-FP, MIKE-11, and SWAT are widely used for flood simulation. Normally, HEC-HMS is coupled with HEC-RAS, where HEC-HMS can produce a hydrograph or rainfall-runoff prediction, and HEC-RAS provides estimates of 1D/2D flood maps.

Among hydrological techniques, BG-FLOOD, GeoFabrick, PARAM, and ANUGA have not received attention, as shown in Figure 13. Flood simulations provide an overview of flooding volume, depth, velocity, hazard, and inundation maps. Hydrological technologies have been proven to use complex mathematical and hydrological principles to give the best model of hydrological systems (Devia et al., 2015).

Figure 13

4.3.6 Rainfall-runoff (RR) prediction

Rainfall is the primary cause of flooding (Liu et al., 2023). Since high rainfall intensity, duration, and spatial distribution have been highlighted as correlates of flood disasters. Therefore, investigating and fully understanding rainfall characteristics is significant for flood risk mitigation (Noori et al., 2019). Flooding can result from excessive runoff from prolonged high precipitation intensity, a dam break, or failure. Surface runoff forms part of the hydrological cycle and refers to the portion of rainwater that does not infiltrate due to soil mechanisms that hinder infiltration and increase surface water flow. For instance, water moves from high-elevation to low-elevation regions into streams, returning surplus water to the ocean and increasing in the absence of vegetation cover and increasing asphalt and impermeable surfaces. Therefore, it is agreed that runoff is controlled by the interaction of several parameters, including climate, vegetation, soil texture, and hillslope (Gabr et al., 2021).

Runoff models are based on technologies that describe rainfall-runoff relations, and models that simulate the dynamics of rainfall and runoff relationships (flood simulation models) use a numerical approach. Hydrological simulations can depict what happens in the water system due to land cover modification from vegetation to impervious surface (Amitrano et al., 2024).

The significance of runoff models is highlighted by their potential to derive the conversion of rainfall to runoff with means of discharge and the ability to facilitate the design and development of various hydraulic infrastructures, operations, and management of reservoirs (Masoud et al., 2024). Therefore, the behaviour or runoff or water flow dynamics is predicted with rainfall-runoff simulation (flood simulation), which tends to delineate vulnerable zones from runoff. Therefore, these simulations extend the contents of rainfall-runoff models.

To derive and simulate the surface runoff model, field rainfall data and corresponding flood data are required. Since in-situ monitoring is not always possible, physiographic characteristics, soil formation, and land use become essential. Remote sensing data plays a significant role in addressing the challenges of in situ data collection.

Runoff models primarily address how much rainfall is converted to runoff, when the river reaches peak flow after rainfall, and infiltration, percolation, evapotranspiration, baseflow, and scenario analysis. The outcomes of this approach include the hydrograph, runoff depth, peak discharge, and catchment water balance. Further interpretation of rainfall-runoff modelling has been reviewed in the literature (Jehanzaib et al., 2022; Kumar R. et al., 2023).

Hydrological and AI technologies have been widely applied to derive runoff and discharge potential. Including models like HEC-HMS and HEC-RAS, GRHUM, MIKE 11, MARINE, Liuxihe Model, SCS CN, ArcSWAT, among other models (Figure 14). Other models that have received more attention are K-Means and a novel group method of handling (GMDH), which is considered a generalised structure of the GMDH (GSGGMDH) model. Rainfall-runoff can be associated with questions such as how much rainfall is converted to runoff, when the river will peak after rainfall, the impact of land-use change, soil type, slope, and climate variability, infiltration, percolation, evapotranspiration, baseflow, and river discharge.

Figure 14

4.3.7 Flood forecasting

Since floods have become an increasing problem due to EWEs and continuous climate change, there has been a growing concern about future flooding. Changes in land use and land cover can impact flooding. Predicting flooding events before they occur is now part of the global agenda. The computational power and data availability facilitate the strategic future flooding prediction (Zanchetta and Coulibaly, 2020).

Methods for flood forecasting are continually advancing, enabling more accurate predictions of environmental and climate changes that alter susceptibility and flood risk. The operational perspective of these models lies in their potential to delineate and predict flood occurrence with its magnitude (Zanchetta and Coulibaly, 2020). Forecasting is an activity that generates quantitative information over time and space, including “short-term” and “long-term” forecasting. These two categories are differentiable in their aims, while the “short-term” technique is used to forecast up to a window of about 6 h: the “long-term” technique forecasts into a long-term window (Akbari Asanjan et al., 2018).

Flood forecast results are often manifested as timely, early warning alerts, which allow more time for safety preparations (Agonafir et al., 2023). Early warning is important for reducing property and life losses during floods.

Curiosity about flood analysis and planning scenarios has elevated the future flood management. The flood forecasting approach helps plan for what is to come. Since floods are difficult to prevent, adequate planning for evacuation and mitigation can significantly reduce their impact. Although early evacuation cannot prevent a flood, some damage can be prevented through early warning systems. For instance, loss of lives can be prevented through early evacuation of vulnerable people, but vulnerable infrastructure, such as buildings, cannot be moved.

This review indicates that hydrological and AI have also been used to derive future flood analyses. ML algorithms such as RF, SVM, ANN, LSTM, CNN, MLP, and KNN have been widely employed (Figure 15). The hydrological model has highlighted the potential to simulate flood events for different return periods.

Figure 15

While flood susceptibility, vulnerability, and risk analysis have been extensively applied to historical events, a growing number of research studies are focusing on the importance of projecting these indices into future forecasting scenarios. Studies such as Xu et al. (2022) have used global climate scenarios, including representative concentration pathways (RCPs), to predict risk intensity and frequency. Similarly, Nguyen et al. (2022), Rashidiyan and Rahimzadegan (2024) and Banjara et al. (2024) demonstrated how LULC transitions can amplify and shift susceptibility zones over time, and shift in hydrological processes.

4.4 Comparison of flood technologies

Each flood methodological approach (Supplementary Table C1) comes with inherent limitations. Hydrological and hydraulic models face challenges in calibration, since they rely on physical interpretation. Hydrological software like HEC-HMS uses the principles of the Soil Conservation Service (SCS) curve number (SCS-CN), which derives runoff from hydrological soil groups, land use and land cover, soil moisture, and rainfall. While AI methods are superior and are improving exponentially, they require large datasets and have limited transferability. On the other hand, MCDA are computationally easy but remains subjective in its weighting of criteria. While AHP-based MCDA is frequently used, it is not fixed to criterion weighting (Table 2). Therefore, recognising these limitations is crucial for ensuring transparency in model application and for guiding future research toward hybrid approaches that combine the strengths of multiple methodologies.

Modelling runoff potential in urban settings can be complex, especially when it intersects with multiple catchments. Rainfall-runoff and flood analysis have only partially accounted for groundwater variability, and artificial stormwater drainage that removes water from the catchment.

5 Key factors and implications to flood modelling

5.1 Key impact of factors on flood dynamics

This review found that rainfall remains a key driver of flooding and is applied in all flood analyses (Supplementary B). In addition, while land use and land cover change alteration have been viewed as the main driver exacerbating surface runoff. Studies have applied a wide range of parameters to investigate flood susceptibility, vulnerability, risk, rainfall-runoff and flood modelling. There are remaining parameters that are not widely applied, such as surface water, urban stormwater drainage, and groundwater variability.

5.1.1 Impact of surface water on flooding

While the evidence from the literature reveals that numerous studies have effectively derived flood analysis using various parameters by taking rainfall as a primary flood conditional factor, and analysing the rainfall-runoff relationship, there is a need to explore other factors that might induce surface runoff, which have not been fully explored, for example, surface water-flood and groundwater-flood.

Various surface-water indices, such as NDWI, MDWI, NDVI, WRI, and soil moisture, help understand the dynamics of surface-water change. For instance, Serban et al. (2022) investigated the effectiveness of NDWI, NDVI, and MNDWI, as well as the weighted normalised difference water index (WNDWI) and water ratio index (WRI) for predicting changes in surface water in Tuzla Lake, Romania. The volume of surface water can significantly increase the rainfall runoff. Since flooding is defined as an overflow of water, high rainfall in regions with high surface water increases the risk of flooding. Therefore, flood occurrence can be hypothesised to be seasonally impacted. There is a need to investigate the volume of surface water before flooding.

In addition, wetlands and regions with predominantly surface water are highly susceptible to flooding, particularly from prolonged light rainfall or short-duration high-intensity rainfall. When comparing the magnitude of those events, it might not have resulted in flooding in dry regions.

5.1.2 Impact of groundwater on flooding

The interaction of groundwater and surface water plays a major role in the hydrological cycle (Joo and Tian, 2021). Groundwater is mainly recharged by surface waters and discharged artificially or naturally as it manifests in the form of a spring or flows in the low-lying area. Understanding groundwater levels and soil moisture is crucial for assessing the rate of surface water infiltration. For instance, high groundwater levels and high soil moisture content reduce infiltration into the aquifer and increase the prevalence of surface water. However, it is complex to understand groundwater dynamics during flood events, and most studies often ignore groundwater when analysing flood risk (Kreibich et al., 2009). As a result, the groundwater domain in surface flood is assumed to be saturated, and most flood peaks are primarily caused by precipitation.

However, other studies have investigated the role of groundwater dynamics in flooding. For instance, Joo and Tian (2021) assess the impact of stream-groundwater interaction on flood peaks based on the hydrological-ecological integrated watershed-scale flow model (HEIFLOW) and highlighted that the stream runoff at the Miho catchment in South Korea is highly affected by groundwater flow during dry and flood seasons, indicating that most baseflow downstream of Miho Catchment is sourced from groundwater and the peak flow during flow events is highly influenced by groundwater flow to the stream. The researchers noted that groundwater-surface water interactions should be considered to mitigate the water hazard in the catchment area. Traditional methods of flood risk mitigation using land-based hydraulic structures generally focus on surface water, considering dams, rivers, and reservoirs (Joo et al., 2014).

In addition, Padilla et al. (2016) reveal a growing need for integrated surface water and groundwater models. A wide range of techniques can be used to investigate groundwater interaction with floods. Joo and Tian (2021) explored HEIFLOW, whose forerunner is groundwater and surface water FLOW (GSFLOW), which simulates the hydrological process, integrating a precipitation-runoff modelling system (PRMS) with the modular groundwater flow model (MODFLOW-2005).

5.1.3 Impact of changing environment on flooding

Land use and land cover alteration by anthropogenic activities, such as urbanisation and deforestation, has been identified as a significant factor contributing to the alteration of the hydrological cycle in the watershed, which significantly influences stream flow and flood volume (Kabanda and Palamuleni, 2013). The change in LULC is attributed to changes in the hydrological mainstream, such as infiltration, evapotranspiration, and precipitation (Yin et al., 2017). The changes in the environment are mainly due to multi-purpose development. The rapid growth of the population drives urban expansion and the transformation of forests, vegetated areas, or green spaces into urban or built-up areas, driven by development such as concrete and impervious surfaces (Banjara et al., 2024). The effect of LULC change on the hydrological cycle can be examined through hydrological modelling, statistics and comparative analysis. Banjara et al. (2024) highlighted the watershed’s transition to an urbanised landscape, which appears to replace the region dominated by agriculture and forests. By this indication, they suggested an increase in surface runoff volume by 2050 and 2080. The LULC transition, therefore, promotes a rise in flood-related perspectives.

5.1.4 Impact of misalignment of urban stormwater drainage

Urban stormwater drainage systems are critical components of the urban infrastructure, primarily designed to transport, store and regulate surface water flow. When these systems are misaligned and fail, the cities become susceptible to flooding, environmental degradation, sanitation issues, health risks, and disruption of essential services (Miguez et al., 2012). Urban drainage plays a fundamental role in stormwater management, particularly as natural land cover is replaced by impervious surfaces, which inhibit infiltration and increase surface water accumulation and runoff. Properly designed and aligned urban stormwater drainage helps mitigate floods by reducing the excessive accumulation of surface water and providing a channel for rainwater to flow. However, when the drainage system is poorly maintained, misaligned, or unable to handle stormwater variability, the likelihood of flooding increases.

This issue is prevalent, particularly in underdeveloped nations, where urban drainage infrastructure often suffers from poor planning, lack of maintenance, and sediment and debris accumulation. Such obstructions and gradual variability of precipitation exacerbate flooding. Addressing these challenges requires comprehensive urban planning, regularly maintained, and sustainable drainage solutions to minimise urban impact on stormwater-related disasters.

6 Discussion

In recent years, flooding has become a prominent topic in scientific research, with substantial advancement in modelling techniques (Agonafir et al., 2023). As flood frequency increases, studies are increasingly focusing on developing more sophisticated models to improve flood assessments. Various methods, such as AI, MCDA, geospatial techniques, hydrologic and hydraulic techniques, and heuristic techniques, offer distinct strengths, efficacy, and contributions to various aspects of food analysis. These technologies have been effectively applied to investigate flood phenomena such as flood detection, flood susceptibility, flood vulnerability, flood risk, flood forecasting, and rainfall runoff in flood simulation.

Flood detection is often considered the initial phase of flood assessment and has been extensively investigated through geospatial and AI techniques. Satellite-based remote sensing, particularly SAR sensors such as Sentinel-1, has proven highly effective for flood detection and analysis. This review indicates that flood applications are exponentially increasing, largely assessed using AI and MCDA.

Our reviews argue that the increase in the application of AI in flood forecasting, susceptibility, vulnerability, and risk is often driven by its predictive power, data availability in developed regions, and the greater outputs it offers compared to the subjective nature of MCDA. However, there is an increasing reliance on MCDA in data-scarce regions. Given the different dynamics and capabilities of each technology, we have identified growing interest in hybrid technologies for flood applications. A hybrid approach leverages the strengths of multiple models to improve predictive performance.

Additionally, a wide range of parameters has been applied to flood susceptibility, vulnerability, and risk, but few studies have incorporated infiltration and runoff parameters, such as the SCS-CN runoff potential and groundwater variability. Among other hydraulic factors, urban stormwater infrastructure remains a critical flood-control measure that must be acknowledged in flood applications.

6.1 Practice implications

This scoping review identified a limited incorporation of stormwater drainage into flood susceptibility assessments, and modelling, suggesting that future assessments should better represent urban drainage systems, particularly in flood-prone cities where drainage failure contributes substantially to flood occurrence. The limited incorporation of stormwater drainage suggests that flood assessment in urban areas relies solely on topographic, hydrological, and land-use factors, potentially leading to underestimation or overestimation of flooding. Incorporating stormwater drainage infrastructure and planning can help reduce flood risk, aligning with Priority 3 of the Sendai Framework for Disaster Risk Reduction 2015–2030, United Nations Office for Disaster Risk Reduction (UNDRR), which calls for flood risk reduction (UNDRR, 2015). Therefore, studies should prioritise flood stormwater drainage features to support more realistic risk assessment. The frequent application of MCDA technology in data-constrained regions highlights its practical utility for preliminary flood risk modelling, where hydrological observations and historical flood data remain limited. These patterns suggest an important opportunity for the government to implement a policy or invest in data generation for constrained data settings, including Southern Africa and Southeast Asia, where data remains limited. In that context, MCDA can support flood susceptibility mapping and spatial prioritisation for flood risk preparedness, infrastructure planning, and land-use regulation. Provided that MCDA methods and assumptions are transparent.

This scoping review also indicates that LULC change is given high consideration in flood mapping due to its influence on runoff generation and flood exposure. The strong influence of LULC dynamics on flood risk assessment reinforced the need for close integration between urban policies and flood risk management. This problem is particularly relevant to underdeveloped nations, where urban settlement extends beyond flood-prone zones.

The findings indicate a need for an integrated approach that accounts for hydrological, infrastructural, and land-use transformation dynamics to improve flood-mapping reliability.

6.2 Research and methodology directions

Beyond practice and implications (section 6.1), identified in this scoping review, several methodological and research gaps warrant further attention.

The review demonstrates that global nations have demonstrated the capabilities of flood management and mitigation technologies, which have helped identify susceptible and vulnerable flood-risk zones and forecast future flood events. The continuous advancements in technology are paving the way for more sophisticated methods to understand and manage natural disasters, such as floods. Innovations in remote sensing, machine learning, and big data analytics are particularly promising. The review indicates an increasing trend toward the application of sophisticated models across different stages of flood management. The increasing frequency of floods calls for more efficient flood modelling approaches.

Despite technological advancements in flood management, a notable research gap remains. Data scarcity, particularly in underdeveloped nations, limits the applicability of high-resolution models, while real-time adaptive systems remain underdeveloped. This challenge might have led to increasing adoption of technologies such as MCDA in a data-constrained environment. Therefore, future research should prioritise improving the transferability assessment of these methods to ensure robust decision-making under constrained data environments.

There is limited evidence translating surface runoff to flooding. While there are numerous applicable standards for hydrological models and AI in flood forecasting, rainfall-runoff is often applied in isolation. The remaining question is the ability to use AI to forecast runoff without considering discharge. This study lays a hypothesis that needs further improvement. Empirical rainfall-runoff is a major indicator and often translates into flooding compared to streamflow (river discharge). This scoping review argues that rainfall-runoff forecasting is needed to understand future flood dynamics, especially in regions where surface runoff is dominant rather than discharge.

Integrating rainfall, LULC, drainage infrastructure and runoff generation may improve the understanding of future flood dynamics to support more context-specific flood assessment frameworks.

6.3 Our study’s significance and limitations

The existing body of evidence on flood analysis reflects an interdisciplinary approach that integrates innovation in data analytics, geospatial technology, and predictive modelling. This is evidenced by Tariq et al. (2020) who focused on risk-based flood management, while Kumar R. et al., (2023) focused on the art of DL and ML for flood forecasting and management stages. Geospatial techniques, particularly GIS and remote sensing, are essential in all flood management stages, mainly for pre-flood and post-flood stages and demonstrate common concepts in flood management (Ghent, 2013). The screening was based on the title and abstract, followed by a full document screening that answered questions such as (i) which methods are used in the paper? (ii) Which discipline does the paper fragment belong to? and (iii) which methods are integrated?

This review provides a comprehensive synthesis of the state of the art in the linkage between flood methods and their applications. This study reviews the available evidence, highlights current trends and modelling techniques, and identifies areas that need further integration. Importantly, this review encourages adopting a multidisciplinary approach, integrating insights from diverse technologies to enhance flood management strategies. However, this review has limitations due to the limited number of database engines used: Scopus, Web of Science, and IEEE Xplore. Although there are other databases, such as ProQuest, EBSCO, and Google Scholar, as well as regional journals such as the African Journal, that provide valuable peer-reviewed content. This review also acknowledges the limitations arising from the selection of studies (language restrictions, database selection, and exclusion of grey literature), all of which introduce bias. Limiting the evidence to articles published in English excludes evidence from regions that publish in other languages.

The review followed the PRISMA-ScR framework rather than the full PRISMA, was limited to a scoping review, and did not follow a systematic review process. This scoping review, focused on breadth rather than depth, did not exhaustively assess the risk of bias, the quality of study assessment, or reliance on individual authors for screening. Since this study screening was done by a single author, a constructive selection criterion and information selection were critical; this was done to account for studies in the flood management domain by looking at the method’s key words (AI, hydrological models, remote sensing and GIS), together with aspect of flood modelling (susceptibility, detection, flood forecasting, rainfall-runoff, vulnerability and flood risk).

Acknowledged, one of the major limitations of scoping reviews is the risk of bias in the evidence, which is not mandatory to perform (Peters et al., 2015).

While flood forecasting can widely depend on time horizon and purpose, such as return periods, decades-scale mapping (susceptibility and risk), near-real-time and short-term forecasting, this study’s scope did not cover fine details on the purpose and temporal scale of prediction.

We advise further improvement and analysis of flood management policies, and an investigation into how models such as rainfall-runoff and discharge are translated into flooding. Such analysis can provide evidence that discharge and runoff can be used interchangeably.

7 Conclusion

This paper has comprehensively examined the current state of technologies and applications for flood analysis, highlighting the significant evolution of AI, MCDA and geospatial tools. These techniques have advanced flood management analysis, including flood susceptibility, risk, vulnerability, and forecasting. Whereas hydrological models remain widely applied for rainfall-runoff and flood modelling, including flood simulation. However, heuristic and metaheuristic techniques have not been popular in various flood analysis approaches. Our review highlights that studies are applying and exploring various algorithms across different study regions, while a few are exploring new flood modelling approaches. The new evidence facilitates the integration of parameters, such as the curve number for runoff potential, in flood applications, including susceptibility and risk.

Flood-prone regions across Asia, including India, China, Iran, Bangladesh, and Vietnam, have made substantial progress in flood research. However, knowledge remains scarce in other vulnerable regions, such as South Africa, where floods are frequently declared natural disasters, for instance, recent events in 2017, 2019, 2022 and 2024 (Grab and Nash, 2023). Technologies such as remote sensing, GIS, AI (ML and DL), and MCDA have been widely used across various stages of flood management. While science has identified the potential risk of flooding, there is limited evidence linking surface runoff to flooding. In addition, there is limited knowledge of groundwater integration in surface runoff modelling. We found that discharge is often used to forecast flood events, especially with DL. While hydrological software such as HEC-HMS with the SCS-CN is widely used to predict runoff and is calibrated against observed discharge. The limited quantification of runoff makes it challenging to accurately predict and forecast runoff without accounting for streamflow. This study argues that in other regions, using streamflow to forecast floods may be appropriate for the river, but it is not necessary for the overall study region.

This study calls for future studies to compare the efficiency of rainfall-runoff modelling with streamflow modelling for forecasting flooding in urban settings. Nonetheless, we found that stormwater drainage is not often accounted for in flood risk and susceptibility modelling.

This review offers insight into global advancements in flood modelling methodologies, providing a foundation for the scientific community to integrate and explore new literature. By adopting recognised methodologies, flood policies and improvements on flood resilience and preparedness can be developed. Our study can serve as a strategic starting point for developing a localised framework in high-risk zones, ensuring effective policy implementation and response. The presented methodologies can help government agencies enhance disaster preparedness and response by strengthening early warning systems, urban planning and infrastructure development, climate and adaptation strategies, and disaster relief and recovery. For instance, Satellite data coupled with ML and DL can help to enhance early warning systems and generate near-instantaneous flood extent maps, improving emergency responses. Land use and land cover change analysis can be used to enhance the mapping of flood-prone, susceptible, risk and vulnerable zones, ensuring better zoning regulations and resilience infrastructure planning. Despite the findings, future research communities should address the role of surface runoff in flood generation in urban settings, and the predictive distinction between uncalibrated runoff depth and streamflow (discharge) in explaining flooding in urban areas.

References

• 1

  AbdraboK. I.KantoushS. A.EsmaielA.SaberM.SumiT.AlmamariM.et al. (2023). An integrated indicator-based approach for constructing an urban flood vulnerability index as an urban decision-making tool using the PCA and AHP techniques: a case study of Alexandria, Egypt. Urban Clim.48:101426. doi: 10.1016/j.uclim.2023.101426

  • CrossRef
  • Google Scholar
  • View reference in article

• 2

  AgonafirC.LakhankarT.KhanbilvardiR.KrakauerN.RadellD.DevineniN. (2023). A review of recent advances in urban flood research. Water Secur.19:100141. doi: 10.1016/j.wasec.2023.100141

  • CrossRef
  • Google Scholar
  • View reference in article

• 3

  AhmedI.DebnathJ.BhowmikM.BhattacharjeeS. (2024). Flood hazard zonation using GIS-based multi-parametric analytical hierarchy process. Geosyst. Geoenviron.3:100250. doi: 10.1016/j.geogeo.2023.100250

  • CrossRef
  • Google Scholar
  • View reference in article

• 4

  AhnE.KangH. (2018). Introduction to systematic review and meta-analysis. Korean Journal of Anesthesiology71, 103–112. doi: 10.4097/kjae.2018.71.2.103

  • CrossRef
  • Google Scholar

• 5

  Akbari AsanjanA.YangT.HsuK.SorooshianS.LinJ.PengQ. (2018). Short-term precipitation forecast based on the PERSIANN system and LSTM recurrent neural networks. J. Geophys. Res. Atmos.123, 12543–12563. doi: 10.1029/2018JD028375

  • CrossRef
  • Google Scholar
  • View reference in article

• 6

  AliK.BajracharyaR. M.KoiralaH. (2016). A review of flood risk assessment. Int. J. Environ. Agric. Biotechnol.1, 1065–1077. doi: 10.22161/ijeab/1.4.62

  • CrossRef
  • Google Scholar
  • View reference in article

• 7

  AmitranoD.Di MartinoG.Di SimoneA.ImperatoreP. (2024). Flood detection with SAR: a review of techniques and datasets. Remote Sens.16:656. doi: 10.3390/rs16040656

  • CrossRef
  • Google Scholar
  • View reference in article

• 8

  Asian Development Bank (2013). Moving From Risk to Resilience: Sustainable Urban Development in the Pacific. Mandaluyong, Philippines: Asian Development Bank. Available online at: https://www.adb.org/publications/moving-risk-resilience-sustainable-urban-development-pacific

  • Google Scholar
  • View reference in article

• 9

  BanjaraM.BhusalA.GhimireA. B.KalraA. (2024). Impact of land use and land cover change on hydrological processes in urban watersheds: analysis and forecasting for flood risk management. Geosciences14:40. doi: 10.3390/geosciences14020040

  • CrossRef
  • Google Scholar
  • View reference in article

• 10

  BellosV.TsakirisG. (2016). A hybrid method for flood simulation in small catchments combining hydrodynamic and hydrological techniques. J. Hydrol.540, 331–339. doi: 10.1016/j.jhydrol.2016.06.040

  • CrossRef
  • Google Scholar
  • View reference in article

• 11

  BentivoglioR.IsufiE.JonkmanS. N.TaorminaR. (2022). Deep learning methods for flood mapping: a review of existing applications and future research directions. Hydrol. Earth Syst. Sci.26, 4345–4378. doi: 10.5194/hess-26-4345-2022

  • CrossRef
  • Google Scholar
  • View reference in article

• 12

  ChenC.JiangJ.LiaoZ.ZhouY.WangH.PeiQ. (2022). A short-term flood prediction based on spatial deep learning network: a case study for Xi County, China. J. Hydrol.607:127535. doi: 10.1016/j.jhydrol.2022.127535

  • CrossRef
  • Google Scholar
  • View reference in article

• 13

  CorominasJ.van WestenC.FrattiniP.CasciniL.MaletJ. P.FotopoulouS.et al. (2014). Recommendations for the quantitative analysis of landslide risk. Bull. Eng. Geol. Environ.73, 209–263. doi: 10.1007/s10064-013-0538-8

  • CrossRef
  • Google Scholar
  • View reference in article

• 14

  CrichtonD. (2002). UK and global insurance responses to flood hazard. Water Int.27, 119–131. doi: 10.1080/02508060208686984

  • CrossRef
  • Google Scholar
  • View reference in article

• 15

  DangN. M.BabelM. S.LuongH. T. (2011). Evaluation of food risk parameters in the Day River Flood Diversion Area, Red River Delta, Vietnam. Nat. Hazards56, 169–194. doi: 10.1007/s11069-010-9558-x

  • CrossRef
  • Google Scholar
  • View reference in article

• 16

  DanhassanS. S.AbubakarA.ZanginaA. S.AhmadM. H.HazaeaS. A.IshakM. Y.et al. (2023). Flood policy and governance: a pathway for policy coherence in Nigeria. Sustainability15:2392. doi: 10.3390/su15032392

  • CrossRef
  • Google Scholar
  • View reference in article

• 17

  de BritoM. M.AlmoradieA.EversM. (2019). Spatially-explicit sensitivity and uncertainty analysis in a MCDA-based flood vulnerability model. Int. J. Geogr. Inf. Sci.33, 1788–1806. doi: 10.1080/13658816.2019.1599125

  • CrossRef
  • Google Scholar
  • View reference in article

• 18

  DeviaG. K.GanasriB. P.DwarakishG. S. (2015). A review on hydrological models. Aquat. Procedia4, 1001–1007. doi: 10.1016/j.aqpro.2015.02.126

  • CrossRef
  • Google Scholar
  • View reference in article

• 19

  DiaconuD. C.CostacheR.PopaM. C. (2021). An overview of flood risk analysis methods. Water13:474. doi: 10.3390/w13040474

  • CrossRef
  • Google Scholar
  • View reference in article

• 20

  DingJ.WangY.LiC. (2024). A dual-layer complex network-based quantitative food vulnerability assessment method of transportation systems. Land13:753. doi: 10.3390/land13060753

  • CrossRef
  • Google Scholar
  • View reference in article

• 21

  EiniM.KaboliH. S.RashidianM.HedayatH. (2020). Hazard and vulnerability in urban flood risk mapping: machine learning techniques and considering the role of urban districts. Int. J. Disaster Risk Reduct.50:101687. doi: 10.1016/j.ijdrr.2020.101687

  • CrossRef
  • Google Scholar
  • View reference in article

• 22

  EslamianS.EslamianF. (Eds.). (2022). Flood Handbook: Principles and Applications (1st ed.). (Chennai, India: Deanta Global Publishing Services). doi: 10.1201/9781003262640

  • CrossRef
  • Google Scholar

• 23

  FellR.CorominasJ.BonnardC.CasciniL.LeroiE.SavageW. Z. (2008). Guidelines for landslide susceptibility, hazard and risk zoning for land use planning. Eng. Geol.102, 85–98. doi: 10.1016/J.ENGGEO.2008.03.022

  • CrossRef
  • Google Scholar
  • View reference in article

• 24

  GabrS. S.AlkhaldyI. A.El-SaoudW. A.HabeebullahT. M. (2021). Flash flood modeling and mitigation of Al-Hussainiyah area, Makkah, Saudi Arabia. Arab. J. Geosci.14:2044. doi: 10.1007/s12517-021-08400-9,

  • Pubmed Abstract
  • CrossRef
  • Google Scholar
  • View reference in article

• 25

  GarroteJ. (2022). Free global DEMs and flood modelling—a comparison analysis for the January 2015 flooding event in Mocuba City (Mozambique). Water14:176. doi: 10.3390/w14020176

  • CrossRef
  • Google Scholar
  • View reference in article

• 26

  GebreS. L.CattrysseD.Van OrshovenJ. (2021). Multi-criteria decision-making methods to address water allocation problems: a systematic review. Water13:125. doi: 10.3390/w13020125

  • CrossRef
  • Google Scholar
  • View reference in article

• 27

  GhentE. O. (2013). Application of remote sensing and geographical information Systems in Flood Management: a review. Res. J. Appl. Sci. Eng. Technol.6, 1884–1894. doi: 10.19026/rjaset.6.3920

  • CrossRef
  • Google Scholar
  • View reference in article

• 28

  GovenderI. H.ReddyM.PillayR. P.GovenderI. H.ReddyM.PillayR. P. (2025). A Review of Residual Flood Risks in South African-Vulnerable Coastal Communities: Opportunities to Influence Policy. Yohe, G. & Smith, J. (Eds.). London, United Kingdom: IntechOpen.

  • Google Scholar
  • View reference in article

• 29

  GrabS. W.NashD. J. (2023). A new flood chronology for KwaZulu-Natal (1836–2022): the April 2022 Durban floods in historical context. S. Afr. Geogr. J.106, 476–497. doi: 10.1080/03736245.2023.2193758

  • CrossRef
  • Google Scholar
  • View reference in article

• 30

  HerreraP. A.MarazuelaM. A.HofmannT. (2022). Parameter estimation and uncertainty analysis in hydrological modeling. John Wiley and Sons Inc9:1569. doi: 10.1002/wat2.1569

  • CrossRef
  • Google Scholar
  • View reference in article

• 31

  HirabayashiY.MahendranR.KoiralaS.KonoshimaL.YamazakiD.WatanabeS.et al. (2013). Global flood risk under climate change. Nat. Clim. Chang.3, 816–821. doi: 10.1038/nclimate1911

  • CrossRef
  • Google Scholar
  • View reference in article

• 32

  IbrahimM.HuoA.UllahW.UllahS.AhmadA.ZhongF. (2024). Flood vulnerability assessment in the flood prone area of Khyber Pakhtunkhwa, Pakistan. Front. Environ. Sci.12:1303976. doi: 10.3389/fenvs.2024.1303976

  • CrossRef
  • Google Scholar
  • View reference in article

• 33

  IkramQ. D.JamalziA. R.HamidiA. R.UllahI.ShahabM. (2024). Flood risk assessment of the population in Afghanistan: a spatial analysis of hazard, exposure, and vulnerability. Nat. Hazards Res.4, 46–55. doi: 10.1016/j.nhres.2023.09.006

  • CrossRef
  • Google Scholar
  • View reference in article

• 34

  IPCC (2014). in Climate Change 2014: Synthesis Report. Contribution of Working Groups I, II and III to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change, eds. PachauriR. K.MeyerL. A. (Geneva, Switzerland: Intergovernmental Panel on Climate Change).

  • Google Scholar
  • View reference in article

• 35

  IPCC (2023). Sections. In: Climate Change 2023: Synthesis Report. Contribution of Working Groups I, II and III to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change (H. Lee and J. Romero, Eds.). (Geneva, Switzerland: IPCC) 35–115. doi: 10.59327/IPCC/AR6-9789291691647

  • CrossRef
  • Google Scholar
  • View reference in article

• 36

  IrbıkD. E. (2022). A systematic literature review of water-migration-gender Nexus toward integrated governance strategies for (non) migrants. Front. Water4:921459. doi: 10.3389/FRWA.2022.921459/TEXT

  • CrossRef
  • Google Scholar
  • View reference in article

• 37

  JehanzaibM.AjmalM.AchiteM.KimT. W. (2022). Comprehensive review: advancements in rainfall-runoff modelling for flood mitigation. Climate10:147. doi: 10.3390/cli10100147

  • CrossRef
  • Google Scholar
  • View reference in article

• 38

  JooJ.KjeldsenT.KimH. J.LeeH. (2014). A comparison of two event-based flood models (ReFH-rainfall runoff model and HEC-HMS) at two Korean catchments, Bukil and Jeungpyeong. KSCE J. Civ. Eng.18, 330–343. doi: 10.1007/s12205-013-0348-3

  • CrossRef
  • Google Scholar
  • View reference in article

• 39

  JooJ.TianY. (2021). Impact of stream-groundwater interactions on peak streamflow in the floods. Hydrology8:141. doi: 10.3390/hydrology8030141

  • CrossRef
  • Google Scholar
  • View reference in article

• 40

  JyrkamaM. I.SykesJ. F. (2007). The impact of climate change on spatially varying groundwater recharge in the Grand River watershed (Ontario). J. Hydrol.338, 237–250. doi: 10.1016/j.jhydrol.2007.02.036,

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 41

  KabandaT. H.PalamuleniL. G. (2013). Land use/cover changes and vulnerability to flooding in the harts catchment, South Africa. S. Afr. Geogr. J.95, 105–116. doi: 10.1080/03736245.2013.806165

  • CrossRef
  • Google Scholar

• 42

  KabengeM.ElaruJ.WangH.LiF. (2017). Characterizing flood hazard risk in data-scarce areas, using a remote sensing and GIS-based flood hazard index. Nat. Hazards89, 1369–1387. doi: 10.1007/s11069-017-3024-y

  • CrossRef
  • Google Scholar

• 43

  KreibichH.ThiekenA. H.GrunenbergH.UllrichK.SommerT. (2009). Natural hazards and earth system sciences extent, perception and mitigation of damage due to high groundwater levels in the city of Dresden, Germany. Hazards Earth Syst. Sci9, 1247–1258. www.nat-hazards-earth-syst-sci.net/9/1247/2009/

  • Google Scholar

• 44

  KronW. (2005). Flood risk = hazard • values • vulnerability. Water Int.30, 58–68. doi: 10.1080/02508060508691837

  • CrossRef
  • Google Scholar

• 45

  KumarV.AzamathullaH. M.SharmaK. V.MehtaD. J.MaharajK. T. (2023b). The state of the art in deep learning applications, challenges, and future prospects: a comprehensive review of flood forecasting and management. Sustainability15:3918. doi: 10.3390/su151310543

  • CrossRef
  • Google Scholar

• 46

  KumarR.KumarM.TiwariA.MajidS. I.BhadwalS.SahuN.et al. (2023). Assessment and mapping of riverine flood susceptibility (RFS) in India through coupled multicriteria decision making models and geospatial techniques. Water15:10543. doi: 10.3390/w15223918

  • CrossRef
  • Google Scholar

• 47

  KumarV.SharmaK. V.CaloieroT.MehtaD. J.SinghK. (2023a). Comprehensive overview of flood modeling approaches: a review of recent advances. Hydrology10:141. doi: 10.3390/hydrology10070141

  • CrossRef
  • Google Scholar

• 48

  KumarV.YadavS. M. (2018). Optimization of reservoir operation with a new approach in evolutionary computation using TLBO algorithm and Jaya algorithm. Water Resour. Manag.32, 4375–4391. doi: 10.1007/s11269-018-2067-5

  • CrossRef
  • Google Scholar

• 49

  KumarV.YadavS. M. (2022). A state-of-the-art review of heuristic and metaheuristic optimization techniques for the management of water resources. Water Supply22, 3702–3728. doi: 10.2166/ws.2022.010

  • CrossRef
  • Google Scholar

• 50

  LapietraI.ColaciccoR.RizzoA.CapolongoD. (2024). Mapping social vulnerability to multi-hazard scenarios: a GIS-based approach at the census tract level. Appl. Sci.14:4503. doi: 10.3390/app14114503

  • CrossRef
  • Google Scholar

• 51

  LiakosK. G.BusatoP.MoshouD.PearsonS.BochtisD. (2018). Machine learning in agriculture: a review. Sensors18:2674. doi: 10.3390/s18082674,

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 52

  LiuC.LiW.ZhaoC.XieT.JianS.WuQ.et al. (2023). BK-SWMM flood simulation framework is being proposed for urban storm flood modeling based on uncertainty parameter crowdsourcing data from a single functional region. J. Environ. Manag.344:118482. doi: 10.1016/j.jenvman.2023.118482,

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 53

  MadzivanyikaC.UteteB.MabvureT. J.SangoI. (2026). Fiscal policies intertwined to public-private partnership investment in water and sanitation for achieving SDG 6: a systematic literature review. Frontiers in Water8:1703548. doi: 10.3389/FRWA.2026.1703548/TEXT

  • CrossRef
  • Google Scholar

• 54

  MasoudM. H. Z.BasahiJ. M.AlqarawyA.SchneiderM.RajmohanN.NiyaziB. A. M.et al. (2024). Flash flood prediction in Southwest Saudi Arabia using GIS technique and surface water models. Appl Water Sci14:61. doi: 10.1007/s13201-024-02117-2

  • CrossRef
  • Google Scholar

• 55

  MehmoodK.BaoY.SaifullahChengW.KhanM. A.SiddiqueN.et al. (2022). Predicting the quality of air with machine learning approaches: current research priorities and future perspectives. J. Clean. Prod.379:134656. doi: 10.1016/J.JCLEPRO.2022.134656

  • CrossRef
  • Google Scholar

• 56

  MembeleG. M.NaiduM.MutangaO. (2022). Integrating indigenous knowledge and geographical information system in mapping flood vulnerability in informal settlements in a south African context: a critical review. S. Afr. Geogr. J.104, 446–466. doi: 10.1080/03736245.2021.1973907

  • CrossRef
  • Google Scholar

• 57

  MiguezM. G.Pires VerólA.CarnerioP. R. F. (2012). Sustainable Drainage Systems: An Integrated Approach, Combining Hydraulic Engineering Design, Urban Land Control and River Revitalisation Aspects. Available online at: www.intechopen.com

  • Google Scholar

• 58

  Morante-CarballoF.Montalván-BurbanoN.Arias-HidalgoM.Domínguez-GrandaL.Apolo-MasacheB.Carrión-MeroP. (2022). Flood models: an exploratory analysis and research trends. Water14:2488. doi: 10.3390/w14162488

  • CrossRef
  • Google Scholar

• 59

  MunnZ.PetersM. D. J.SternC.TufanaruC.McArthurA.AromatarisE. (2018). Systematic review or scoping review? Guidance for authors when choosing between a systematic or scoping review approach. BMC Med. Res. Methodol.18:143. doi: 10.1186/S12874-018-0611-X,

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 60

  MunyaiR. B.ChikooreH.MusyokiA.ChakwiziraJ.MuofheT. P.XuluN. G.et al. (2021). Vulnerability and adaptation to flood hazards in rural settlements of Limpopo province, South Africa. Water13:3490. doi: 10.3390/w13243490

  • CrossRef
  • Google Scholar

• 61

  NguyenH. D.DangD. K.NguyenQ.-H.BuiQ.-T.PetrisorA.-I. (2022). Evaluating the effects of climate and land use change on the future flood susceptibility in the central region of Vietnam by integrating land change modeler, machine learning methods. Geocarto Int.37, 12810–12845. doi: 10.1080/10106049.2022.2071477

  • CrossRef
  • Google Scholar

• 62

  NooriA. M.PradhanB.AjajQ. M. (2019). Dam site suitability assessment at the greater Zab River in northern Iraq using remote sensing data and GIS. J. Hydrol.574, 964–979. doi: 10.1016/J.JHYDROL.2019.05.001

  • CrossRef
  • Google Scholar

• 63

  PadillaF.HernándezJ. H.JuncosaR.VellandoP. R. (2016). Modelling integrated extreme hydrology. Int. J. Saf. Secur. Eng.6, 685–696. doi: 10.2495/SAFE-V6-N3-685-696

  • CrossRef
  • Google Scholar

• 64

  PetersM. D. J.GodfreyC. M.KhalilH.McInerneyP.ParkerD.SoaresC. B. (2015). Guidance for conducting systematic scoping reviews. Int. J. Evid. Based Healthc.13, 141–146. doi: 10.1097/XEB.0000000000000050,

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 65

  PinosJ.TimbeL. (2019). Performance assessment of two-dimensional hydraulic models for generation of flood inundation maps in mountain river basins. Water Sci. Eng.12, 11–18. doi: 10.1016/j.wse.2019.03.001

  • CrossRef
  • Google Scholar

• 66

  Pugliese ViloriaA. d. J.FoliniA.CarrionD.BrovelliM. A. (2024). Hazard susceptibility mapping with machine and deep learning: a literature review. Remote Sens.16:3374. doi: 10.3390/rs16183374

  • CrossRef
  • Google Scholar

• 67

  Rafiei-SardooiE.AzarehA.ChoubinB.MosaviA. H.ClagueJ. J. (2021). Evaluating urban flood risk using hybrid method of TOPSIS and machine learning. Int. J. Disaster Risk Reduct.66:102614. doi: 10.1016/j.ijdrr.2021.102614

  • CrossRef
  • Google Scholar

• 68

  RashidiyanM.RahimzadeganM. (2024). Investigation and evaluation of land use-land cover change effects on current and future flood susceptibility. Nat. Hazards Rev.25:04023049. doi: 10.1061/NHREFO.NHENG-1854

  • CrossRef
  • Google Scholar

• 69

  SajjadA.LuJ.ChenX.ChisengaC.MazharN.NadeemB. (2022). Riverine flood mapping and impact assessment using remote sensing technique: a case study of Chenab flood-2014 in Multan district, Punjab, Pakistan. Nat. Hazards110, 2207–2226. doi: 10.1007/s11069-021-05033-9

  • CrossRef
  • Google Scholar

• 70

  SandersB. F.SchubertJ. E.GoodrichK. A.HoustonD.FeldmanD. L.BasoloV.et al. (2020). Collaborative modeling with fine-resolution data enhances flood awareness, minimizes differences in flood perception, and produces actionable flood maps. Earth’s Future8:e2019EF001391. doi: 10.1029/2019EF001391

  • CrossRef
  • Google Scholar

• 71

  SchmidtJ.MatchamI.ReeseS.KingA.BellR.HendersonR.et al. (2011). Quantitative multi-risk analysis for natural hazards: a framework for multi-risk modelling. Nat. Hazards58, 1169–1192. doi: 10.1007/s11069-011-9721-z

  • CrossRef
  • Google Scholar

• 72

  SerbanC.MafteiC.DobricaG. (2022). Surface water change detection via water indices and predictive modeling using remote sensing imagery: a case study of Nuntasi-Tuzla Lake, Romania. Water (Switzerland)14:556. doi: 10.3390/w14040556

  • CrossRef
  • Google Scholar

• 73

  SeydiS. T.Kanani-SadatY.HasanlouM.SahraeiR.ChanussotJ.AmaniM. (2022). Comparison of machine learning algorithms for flood susceptibility mapping. Remote Sens.15:192. doi: 10.3390/rs15010192

  • CrossRef
  • Google Scholar

• 74

  TariqM. A. U. R.FarooqR.van de GiesenN. (2020). A critical review of flood risk management and the selection of suitable measures. Appl. Sci. (Switz.)10, 1–18. doi: 10.3390/app10238752

  • CrossRef
  • Google Scholar

• 75

  TriccoA. C.LillieE.ZarinW.O’BrienK. K.ColquhounH.LevacD.et al. (2018). PRISMA extension for scoping reviews (PRISMA-ScR): checklist and explanation. Ann. Intern. Med.169, 467–473. doi: 10.7326/M18-0850

  • CrossRef
  • Google Scholar

• 76

  TripathiA.AttriL.TiwariR. K. (2021). Spaceborne C-band SAR remote sensing–based flood mapping and runoff estimation for 2019 flood scenario in Rupnagar, Punjab, India. Environ. Monit. Assess.193:110. doi: 10.1007/s10661-021-08902-9,

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 77

  TripathyP.MalladiT. (2022). Global flood mapper: a novel Google earth engine application for rapid flood mapping using Sentinel-1 SAR. Nat. Hazards114, 1341–1363. doi: 10.1007/s11069-022-05428-2

  • CrossRef
  • Google Scholar

• 78

  UNDRR (2015). Sendai Framework for Disaster Risk Reduction 2015–2030. Geneva, Switzerland: United Nations Office for Disaster Risk Reduction. Avaliable online at: https://www.preventionweb.net/files/43291_sendaiframeworkfordrren.pdf?startDownload=true

  • Google Scholar

• 79

  UNESCO-IHE. (2024). Flood Vulnerability Index FVI - Institute for Water Education. Available online at: http://unihefvi.free.fr/vulnerability.php

  • Google Scholar

• 80

  WedajoG. K.LemmaT. D.FufaT.GambaP. (2024). Integrating satellite images and machine learning for flood prediction and susceptibility mapping for the case of Amibara, Awash Basin, Ethiopia. Remote Sens.16:2163. doi: 10.3390/rs16122163

  • CrossRef
  • Google Scholar

• 81

  XuH.HouX.LiD.WangX.FanC.DuP.et al. (2022). Spatial assessment of coastal flood risk due to sea level rise in China’s coastal zone through the 21st century. Front. Mar. Sci.9:945901. doi: 10.3389/fmars.2022.945901

  • CrossRef
  • Google Scholar

• 82

  YinJ.HeF.Jiu XiongY.Yu QiuG. (2017). Effects of land use/land cover and climate changes on surface runoff in a semi-humid and semi-arid transition zone in Northwest China. Hydrol. Earth Syst. Sci.21, 183–196. doi: 10.5194/hess-21-183-2017

  • CrossRef
  • Google Scholar

• 83

  ZanchettaA. D. L.CoulibalyP. (2020). Recent advances in real-time pluvial flash flood forecasting. Water12:570. doi: 10.3390/w12020570

  • CrossRef
  • Google Scholar