Recent Advances in Flood Monitoring & Prediction Methods: A Systematic Review

Recent Advances in Flood Monitoring & Prediction Methods: A Systematic Review | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Recent Advances in Flood Monitoring & Prediction Methods: A Systematic Review Adrita Talapatra, Narendra Kumar Rana This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-6948732/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 07 Jan, 2026 Read the published version in Environmental Science and Pollution Research → Version 1 posted 4 You are reading this latest preprint version Abstract Floods are one of the most dreaded hazards adversely affecting life and property which demands and necessitates singular attention. This paper aims at documenting the trajectory of the recent trends and advancements in monitoring and prediction, including probabilities of associated uncertainties from a “flood” point of view. The primary focus is to convey a comparative vision of vulnerability and risk assessment bringing a viable grasp in approaching problems associated with the sudden extremities of floods. Over a timeline spanning 17 years (2007–2024), a systematic review is conducted utilising Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines, dividing the findings into two sections: monitoring and forecasting techniques. The monitoring section encompasses a range of technologies, from ground-level observations to advanced tools, involving global scales to community levels, in perceiving impacts of flood events. Monitoring is enhanced by drone technology, sensor cameras, remote sensing and digital image analysis allowing spatial risk visualisation. The forecasting section pays attention to the utilisation of geospatial guides, GIS tools and machine learning algorithms in predicting the temporal probabilities of flood. The prediction often leverages computer training ensemble models, Artificial Neural Networks (ANN) and Convolutional Neural Networks (CNN). This review article serves as an efficacious resource for researchers and decision-makers, streamlined to enhance understanding of the entire flood hazard management continuum—from initiation and response to mitigation and recovery. By scientifically compiling relevant techniques discussed here, the present work will steer future studies on flood hazards and actively promote adaptive strategies to mitigate their disastrous effects. disaster risk management (DRM) flood forecasting flood hazard mapping (FHM) flood monitoring machine learning algorithms Figures Figure 1 Figure 2 Figure 3 Figure 4 1. Introduction Floods in the present century stand as one of the prime adversities affecting both developed and developing countries despite their physiographic or morphological stability. In a recent evaluation made by C. Van Westen, under the influence of changing hydro-meteorological conditions, flood globally accounts for 30% of total losses caused by environmental hazards ranking as the second most disastrous event after losses from earthquakes(Van Westen, 2013 ). The global vulnerability to floods has increased by more than 40% over the past two decades accounting for heavy urbanisation and global climate change(Wahba et al., 2024 ). This is only projected to grow in the future with multitudes rise in its frequency and severity(Diaconu et al., 2024 ; Wahba et al., 2024 ). From a scientific standpoint, the ability to predict an uncertain phenomenon effectively contributes to assessment as well as adaptability to the calamities related to the hazard. Flood Hazard Mapping (FHM), automatically becomes an essential tool in this context. This along with historical data when utilised with understanding and specification becomes a powerful assistance in the identification of hazard potential, areas, extent, depth and spatial damage potential(Mudashiru et al., 2021 ). The collective objectives of this review article are highlighted in the following points. Synthesis of the existing knowledge base is the primary aim to compile and analyse the trends in the techniques and methods used in existing literatures and discuss their utility while drawing attention to existing gaps. An array of methodologies used in different flood monitoring and prediction studies to be discussed through evaluation to understand the present trend as well as the technological standpoint. The findings from the literatures would be effectively used in an attempt to point out data types used in visualising flood hazards and projecting their occurrences. The shortcomings in data availability to be highlighted alongside as a part of the findings. Examination of contemporary technological advancements, highlighting the abilities of GIS and machine learning involved in actively assisting the enhancement of flood monitoring and prediction capabilities. Based on the case studies and evaluations from existing literature, best practices and methods to be underlined for assistance in flood hazard risk assessment, its monitoring, evaluation of factors-at-risk, damage assessment and more elaborately the prediction that can be applied in all the different contexts. Finally, the present study ventures into encouraging specialists from various disciplines such as hydrologists, climatologists, engineers, urban planners, and policymakers to come together in collaboration to develop holistic flood risk management strategies. The Centre for Research on the Epidemiology of Disasters (CRED) reports a total of more than 2156 flood occurrences over the last three decades making this study the prime of the hour to bring effective ways to look at a natural calamity that affects more than 2.6 billion of lives and huge sum of economic losses worldwide(Bandyopadhyay et al., 2016 ). The review article assists in streamlining the knowledge of the recent trends relating to means of monitoring and predicting flood hazards. It is an effort to bring out elaborate documentation of compiled techniques and measures to address the improvisations in preparing FHM and FSM maps. The paper tries to pinpoint the root causes while accounting implementation of evolved solutions to one of the most serious natural hazards of the twenty-first-century world. 2. Methodology Literatures were selected using the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines. A systematic review is targeted keeping in mind the specific inclusion and exclusion criteria to assess the relevance of the records being studied(Arshad et al., 2019 ). The stepwise process of evaluation primarily included quantitative investigations based on keyword searches across three different databases namely- Scopus, Web of Science and Science Direct (Table 1 ). This returned a total of n = 12,889 manuscript records from all three databases. Table 1 Quantitative results based on the keywords search across different scholarly databases Search Keywords Scopus Web of Science Science Direct “recent AND techniques AND in AND flood AND monitoring AND forecasting” 26 18 918 “flood AND monitoring AND techniques” 1129 54 3753 “flood AND forecasting AND techniques” 1184 34 1384 “hydrological AND monitoring AND forecasting” 794 20 878 “flood AND prediction AND recent AND trends” 76 11 1897 “advanced AND flood AND monitoring AND forecasting” 53 12 648 Total 3262 149 9478 This was followed by the removal of duplicates with assistance from the referencing application Zotero called the ‘Duplicate Items’ feature. From here, a tedious manual assessment was conducted where filtering was carried out based on titles. This resulted in screening a total of 2694 records for inclusion. Afterward, the inclusion and exclusion criteria (discussed later in this section) were included in order to validate the manuscripts taken into consideration for this review report (Fig. 1 ). The final report considers a total of 56 extensively reviewed articles to bring out the desired findings. The present work is achieved through record analysis assisted by systematic literature studies and cautious selections. This is done based on layers of inclusion and exclusion criteria as stated below- 1. General criteria: The timeline for the study is set from 2007 to 2024. The extensive period of 17 years elaborately depicts the changes in the methodological trends over time. 2. Preference in first order: Here, the precise subject domain is selected for narrowing the search criteria. The following subject areas are considered to be included: i. Earth and Planetary Sciences, ii. Social Sciences, and iii. Environmental Sciences. 3. Preference in second order- The types of registers considered need to be regularised and limited to certain preset to maintain similarities and lesser variations in the findings collected reviewing the existing literature. The following registers are considered to be included: i. Conference Papers, ii. Articles, here the inclusion was considered for case-study-based and application-based studies, and iii. Review Reports. Visualisation of the studied literature further proceeds using the Biblioshiny web interface (Aria & Cuccurullo, 2017 ) and Microsoft Excel. The annual scientific production (Fig. 2 ), shows the overall pieces of literature produced over the three different databases during the suggested review period. The graphical representation shows an increasing trend in flood monitoring and prediction studies in the recent decade. It further shows that articles, papers and journals dominate the later part of the study period (2016–2024) showing a steady rise in annual production levels over all the databases. Moreover, the global scientific production map (Fig. 3 ) generated in Biblioshiny using R-Studio, illustrates the country-wise production of literature in the field of study. The largest production is contributed by USA, followed closely by the China. India ranks as the fourth largest contributor in the field right after Italy. Among these the most cited country is USA with almost 3,404 citations within the 17 years study span (data based on Biblioshiny, India results). The three-factor analysis (Fig. 4 ) generated in Biblioshiny using R-Studio, shows the relationship between the sources (SO), countries (AU_CO) and the focused keywords (DE) in the literature to show the nature of interdependencies. It also tries to show the interdisciplinary nature of the present study. The review tried to analyse the various techniques and methods from different fields to bring out a collaborative approach to show recent trends in flood monitoring and forecasting. 3. Overview of Flood Hazards Flood hazards can be classified into various forms and divisions based on both physiological as well as meteorological drivers. Under broad divisions, they can be classified as floods resulting from catastrophic events ensuing in dam failures or glacial lake outbursts(W. Wu et al., 2020 ). These are often associated with improper standards of hard engineering imposed on riverbeds(Bera et al., 2024 ), or causative events such as groundwater overflow and extensive rainfall(Zanchetta & Coulibaly, 2020 ). The other major division would include conditions of flash floods resulting from either pluvial or fluvial indices. Pluvial flash floods (PFFs) are caused by overwhelming localised convective precipitation, orogenic precipitation, sudden cyclonic events, or storm surges in an area(Zanchetta & Coulibaly, 2020 ). Flash floods are typically characterised by their rapid onset (within six hours of rainfall onset) often paired with largescale destructions caused by landslides, extreme sediment depositions, bridge and dam failures, increased urbanisation and demographic imbalances in a region(Hapuarachchi et al., 2011 ). Apart from this, coastal flooding often results from aggrieved coastlines resulting from erosion or problems related to siltation(Appeaning Addo et al., 2018 ). Sometimes, floods result from factors such as sediment transport phenomenon as seen in the case of young river channels(Gharbi et al., 2016 ). Floods are often associated with soil erosion, mass-wasting, landslides and geo-climatic distress. In order to have timely assistance in flood management it is rather very important to understand the triggering factors such as meteorological and hydrological drivers in the region. This along with heavy rainfall often causes flash floods, especially in regions of shallow stream beds(Das, 2012 ). Channel bed aggradation resulting in changes in river bed terrain is also a direct impact of flooding and has a clear relationship between each other(D. Ghosh & Saha, 2014 ; Paul & Biswas, 2019 ). The impact of flood is also evident in regions where sedimentation as an outcome of extensive downwasting from heavy rainfall has resulted in changing the channel morphology affecting the banks on either side of the river(Ma et al., 2022 ; Paul & Biswas, 2019 ). These phenomena are evident in the sub-Himalayan foothills along the northern parts of West Bengal among others. In multiple studies throughout the selected timeline, a direct relationship exists with rainfall as well as the slope of a region. Greater slope associated with heavier rainfall has resulted in extensive flood conditions from flow accumulation in river channels in a small period often resulting in flash floods or extreme cases of pluvial floods. This is often related to analysis and prediction of flood hazard where the “lag time” or the time delay between the peak intensity of rainfall and peak discharge event is considered to enhance prediction of the chances of flooding as well as assessing the associated risks. Understanding lag time differences becomes crucial in selection of methods as it determines the pace of flooding within a watershed. Elevation, soil and the land cover of the areas are also seen to have an effective relation with the flood hazard and pose to be necessary factors in FHM. 4. Recent Methods for Monitoring Flood Hazards Monitoring of flood events is often done through satellite-driven, real-time observation of the flood incidents aiming to build awareness and promptness in understanding and combating the hazard and its atrocities(Revilla-Romero et al., 2015 ). The type of method to be selected for flood monitoring depends on the region of study, the type of desired output, the accuracy requirements and use of results as well as the funding available(Appeaning Addo et al., 2018 ). Often traditional techniques intertwined with advanced models and technologies bring out efficient flood monitoring where input data clustering and filtering, modelling of river terrain and showcasing morphometric variations through digital maps of relief and elevation, heterogenous data processing and hydrological modelling are used to observe flood incidents to bring preparedness to the hazard and its destructions(Merkuryeva et al., 2015 ). Techniques involve a large array of methods utilising drone and sensor-based outputs. Image processing and geoinformation technology play a significant role in the visualisation of the data. 4.1. Real-time Monitoring Systems: Real-time flood monitoring often gets hampered due to the lack of or unavailability of in situ streamflow data in many parts of the world(Revilla-Romero et al., 2015 ). To fill this gap, satellite-based monitoring is used to calibrate and validate the data as well as assimilate it. Real-time forecasting drives as a key resource while used in flash flood early warning systems (FFEWS). In the case of PFFs, real-time forecasting helps in assessing multiple criteria such as the flash flood susceptibility assessment (FFSA), rainfall comparison with surface conditions neglected (RC-SN), rainfall comparison with surface conditions included (RC-SC), flow comparisons in real-time (FC), quantitative and qualitative precipitation forecast and estimation respectively (QPF, QPE)(Zanchetta & Coulibaly, 2020 ). While dealing with forecasting to assist in lead time support along with early warning to flood events, weather radar precipitation mosaic products are accessible, run-off threshold exceedance is best utilised to project the available data and applied in preparing flood forecast maps. In situations where radar data is insufficient the satellite-based precipitation data plays a huge role in assistance to real-time flood observation. Andre Zanchetta and Paulin Coulibaly conducted extensive research in the year 2020 where they highlighted the importance of global navigation systems (GNSSs) and global positioning systems (GPSs) in developing multi-constellation approaches to retrieve and enhance accurate real-time estimation of precipitation in an area. In order to properly orient data fetched in real-time, it is important to have them properly scaled as per requirements. Various extrapolation techniques such as kriging interpolation or inverse distance weighted (IDW) interpolations are involved in processing rainfall data to have a refined computer vision while observing flood conditions in an area. Dynamic downscaling of numerical weather predictions based on convection-permitting (CP-NWP) uses hydrodynamic processes as a baseline with a spatial resolution of 4 km or higher to present local convections and mesoscale event visualisation. Temporal extrapolation becomes vital in order to not only monitor but also project the data to understand future scenarios. In such cases, uncertainties in grid scale are common, where pixel-to-pixel analysis is used involving ensemble models to reduce the existence of spatial noise in the data being monitored. 4.2. Sensor Networks- Internet of Things (IoT) based sensors and warning systems: In the systematic review conducted by Arshad et al., various techniques using sensors to monitor floods are highlighted. The IoT-based sensor called the automatic pressure transducer assists in measuring up to 0.001 m accuracy in water levels and is compatible with multiple controllers for logging the data and visualise them in real-time(Arshad et al., 2019 ). Here, using computer-based technology and IoT sensors helps in warning in case there is a water level rise resulting from flood hazard via, short message service (SMS), email, or web servers. Resource from Arshad et al. also state, the comparison between three techniques: (1) differencing image technique, (2) dictionary method and (3) convolution neural networks (CNN) in early warning of rising water levels and flood situation. The differencing image technique uses images every 10-minute intervals to check the previous image and current image state of water level pixels. It uses Gaussian smoothening to remove any noise. The dictionary method classifies the region of interest (ROI) into ruler and water classes to detect the changes in water level. The CNN technique however involves the best results of all these, where a filter is used as the algorithm trains from features extracted from the input images, thus reducing errors to a large extent(Arshad et al., 2019 ). While studying images the bandpass technique is useful to reduce noise in images captured using a charged couple device (CCD) camera, which is converted from a time series to a frequency domain using discrete cosine transform (DCT) for calculating change in water level(Park et al., 2009 ). The histogram variations studied, on the other hand, also help in image analysis and condition judgments. An accumulated histogram can be prepared using images gathered from water level surveillance cameras to detect changes in water level by comparison of images collected at different points in time. Such detections are attempted in a conference proceeding by Park et al. through their elaborate studies in Seoul, Korea. Wireless sensors used in IoT coupled with ANNs have built substantial real-time monitoring systems of change detection in the inundation level of urban drainage networks. 4.3. Drones and UAVs acting as a high-resolution monitoring tool: Datasets produced by drones have a high spatial resolution (< 10cm) and produce very high-resolution digital elevation models (DEMs). The use of drone technology in coastal monitoring is a common approach as far as coastal surveillance and management is concerned. It also provides timely, event-based rapid availability and superior quality outputs making it highly advantageous in immediate monitoring and preparedness. Specifically, a little after 2014 there has been a significant boom in innovations and advancements in drone technology, hardware and their availability(Iqbal et al., 2023 ). A drone-assisted study across the Volta Delta in Ghana efficiently detected accretion, erosion and change in shoreline over 12 years. Drone image comparisons show vital changes in the shoreline and vulnerability of settlements located along the shoreline(Appeaning Addo et al., 2018 ). The very high-resolution data provided by drone cameras makes it powerful in monitoring flood incidents as well as account the rate of population displacements in the affected area as a post-flood situation. This in turn helps the region to cope and convalesce from the extreme effects of the hazard immediately after its occurrence. However, drones are limited in strong wind conditions as they are vulnerable to distortions from environmental indices. Thus, they are subjected to limited use in times of gentle wind conditions. Various other associated challenges include platform instability, distortions in view angle, or erroneous data processing tools. One of the other major constraints of drone usage is their low battery lifespan restricting them to only short flight times. The utilisation of Unmanned Aerial Vehicles (UAVs) also has numerous perks of its edge as it has no requirement of installation points as in the case of remote sensing images, has a large spatial coverage and is comparatively less costly allowing frequent missions(Vousdoukas et al., 2011 ). Studies conducted using Small Unmanned Aerial Vehicles (SUAVs) use a series of images collected after flying a camera-mounted device across the shoreline to detect the changes. The different time frames are processed in a semi-automated fashion where each frame is synthesised and masked to detect the change in shoreline elevation, water variations and displacements in a short time over a wide, large-scale area. Each frame of images is geo-rectified and mosaics are obtained to view the results with better precision. The shoreline elevation can also be collected from tidal and wave data using the Argus video monitoring systems(Holman & Stanley, 2007 ). However, while conducting geo-rectification on each frame of images it is evident that errors may accumulate over several iterations, demanding a robust frame-to-frame geo-referencing(Vousdoukas et al., 2011 ). Transforming the image pixel to world coordinates often poses one of the greatest challenges in generating accurate velocity estimation during flash flood monitoring even under controlled settings and parameters(Perks et al., 2016 ). Various algorithms are drawn to address such errors and minimise them to view the free surface velocity of real-time flash flood monitoring. 4.4. Community-based Monitoring and Crowdsourcing: Community-based disaster risk management in monitoring any disaster often helps in building early warning systems. In the case of community-based monitoring, this is often done by locals and villagers who know the region affected by the hazard very well. This in turn helps to empower communities through participation and build awareness to prepare for and confront hazards(Macherera & Chimbari, 2016 ). Research accounts conduction of various workshops in the Nordic countries by several organisations, namely HydroLogic and WeSenseIt to provide insight into ‘state-of-art tools’ based on citizen observations(Henriksen et al., 2018 ). Under the same research, directives such as the Citizen Observatories for Water Management (COWM2016, Venice), European Water Framework Directive (2000/60/EC) and Flood Directive (2007/60/EC) exist to strengthen data based on community observations relating to flood hazards and its management. Systematic reviews on urban pluvial flooding (UPF) show extensive use of participatory science and crowdsourcing to understand floods in areas of vulnerability(Azizi et al., 2022 ). Crowdsourced data proposes to address uncertainties driven by localised topographic and land cover variations, data unavailability, validation and calibration errors by providing context-specific observations(Assumpção et al., 2018 ; R. Q. Wang et al., 2018 ). Major crowdsourcing tools that are utilised in recent studies include public webcams, social media and citizen science(Helmrich et al., 2021 ) as they can provide large volumes of spatially relevant observational data points. In the case of urban flood monitoring, which mostly occurs from pluvial flooding due to various factors relating to physiographic as well as anthropogenic drivers, crowdsourcing measures become an effective tool in the observation of flood situations. Application-based monitoring is carried out where individuals located within the affected regions can mark themselves on a map generating direct real-time data. These data can be treated as secondary data and are most of the time available over large spaces and generative data can be narrowed or adjusted as per the needs of the study. Two such applications are STORM and HAZE used by Helmrich et.al. in Norfolk, Virginia flood hazard monitoring. Public webcams, even though limited to only daylight visualisation of floods, are most effective when located in high-risk flood-prone areas. Webcams using IR sensors can be used to carry out night-time observations with improved, effective results. Citizen science data available in various secondary resources often provides detailed information for each data provided. Social media accounts to quick, effective reporting via various media platforms’ posts, Twitter posts giving out information and geo-tagged photographs in real-time which can be geo-coded to map flooding in an area promptly in order to bring measures and relief as per necessities. 5. Recent Methods for Predicting Flood Hazards The trends in flood prediction have moved from a deterministic to a probabilistic approach over the two decades of the literature study period. In the first half, for most parts, studies were more data-oriented giving one proper idea or nothing as we see in deterministic prediction models. An increasing number of physically based hydrological models are developed and used in flash flood prediction which are found to be giving much better, plausible results as compared to traditional approaches in forecasting the hazard(Hapuarachchi et al., 2011 ). With a probabilistic approach, flexibility is introduced which helps in definitively including the variability and changes associated with physiographical as well as meteorological factors of flood hazards. Prediction of flood hazards involves the exercise of various algorithms, mathematical computations and machine learning assistance involving multi-criteria analysis (MCA) to develop hydro-dynamic models and forecasting tools that give high-level accuracy of data. This in turn enhances visualisation and probabilistic assumptions of the occurrences of the hazard worldwide under similar parameters. Forecasting can be carried out as extended-range forecasting over 7–14 days making predictions using advanced numerical weather prediction models, medium-range forecasting, long-range forecasting extending from a few weeks to a few months combining statistical analysis, historical weather data, climate indices, and global climate models to identify patterns and signals indicative of future weather conditions on longer timescales and nowcasting that utilises real-time observational data from weather radar, satellite imagery, ground stations, and other sources to track the movement and intensity of weather phenomena such as thunderstorms, rainfall, snowfall, and severe weather events. 5.1. Statistical Methods: Effective discussion on flood forecasting methods (FFM) primarily depends on the basic understanding that often speaks about the utilisation of probabilistic methodologies which is closely intertwined with deterministic approaches of model building. Empirical models prepared based on statistical data often utilise historical data based on observed relationships between input variables and river flow changes to identify patterns within the variables and their effects on flooding and influx in streamflow(Granata & Di Nunno, 2024 ). In this context, the Bayesian forecasting system (BFS) is a theoretical model to forecast an extensively occurring hazard. It involves probabilistic forecasting from a deterministic hydrological model. BFS can be used for probabilistic river stage forecast (PRSF), probabilistic river discharge forecast (PRDF), or probabilistic runoff volume forecast (PRVF)(Han & Coulibaly, 2017 ). Here generally two sources are used: (1) Precipitation uncertainty in the future and (2) other hydrological parameters uncertainties. These two are individually calculated and then combined for probabilistic forecasting. Sasha Han along with Paulin Coulibaly, in their elaborate research on BFS utilisation, noted that the uncertainties observed in the overall data are considered in their research as occurrences and non-occurrences. For precipitation forecasting, they are employed. A modified Bayesian processor of output (BPO) is implemented in one study to process the quantitative precipitation forecast output from the numerical weather prediction (NWP) model. It offers an ideal theoretical structure and quantifies all sources of uncertainties, thus reducing predictive uncertainty to some extent and leading to a more reliable and accurate forecast. However, it is not implemented in large watersheds and under climatic variabilities to understand its efficiency beyond river basins. Refinements based on conditional random fields (CRFs) in a Bayesian statistical modelling network are used in pattern recognition and mapping of flood areas combining intensity and coherences(Schumann et al., 2023 ). In a case study of the Zhejiang province of China, researchers X. Wu. et. al. conducted a multi-criteria decision analysis (MCDA) along with the CA-Markov model to predict the future flood risk changes in the region. Analytic Hierarchy Process (AHP) was implemented to obtain a ‘pairwise comparison matrix’ in the MCDA approach. These criteria were divided into ‘the hazard-causing factor’ containing precipitation and flood hazard parameters for 24 hours and 72 hours respectively; ‘the hazard-gestating factor’ containing altitude, slope, standard deviation of elevation, drainage density, and buffer zone of water parameters; and ‘the hazard-bearing factor’ containing the population, GDP, and land cover parameters(X. Wu et al., 2024 ). Different interpolation methods to arrange and understand the various standards of data can be used based on the variogram results showing the spatial geometry of regionalised variables. Utilisation of the CA-Markov model in the study of Zhejiang province elaborately shows parameters such as the population, GDP and land cover of an area as indirect parameters of flood hazard studies. In another study conducted in north-central Italy, a probability error study is attempted to estimate the uncertainty of hydrological forecast in two watersheds. In this literature, the meta-Gaussian model in statistics takes into account the errors and uncertainties in the forecast within the same time frame, where the transformation of preliminary errors is done by Normal Quantile Transform (NQT). In this domain, it is assumed that “forecast errors can be explained as the linear combination of the selected explanatory variables.”(Montanari & Grossi, 2008 ). Here it is shown how utilisation of probabilistic forecasting can be useful in prediction while computing the error to identify and account for the uncertainties. 5.2. Hydrological Models: Detection of pre-flood and post-flood measures and identification of flood hazard zones are attempted to understand the vulnerable sites. GIS and data-driven statistical assessment helped in drafting as well as addressing the issues related to flood mapping systems as well as the warning systems associated with them (Kourgialas & Karatzas, 2011 ). The physically based models comprise mathematical equations showing hydrodynamics principles and assist hydraulic modelling making them extremely efficient in comprehending hydrological processes and preparing models(Granata & Di Nunno, 2024 ). However, these are very data-intensive processes requiring levels of validation and calibration in steps along the entire course of its execution. The Hazards US Multi-Hazard (HAZUS-MH) flood model was prepared to quantify the human, property, financial and social impacts of flood hazards and draw mitigation methods based on the results from the software. Kourgialas et. al., described hydrological and hydraulic modelling as the classical approach to delineating flood-prone regions with different levels of hazard effects. These models are also adapted while simulating run-off in small catchment areas(Wahba et al., 2024 ). River flood stage analysis is a complex dynamic process characterised by successive spatial and temporal variations(Le et al., 2019 ). Different hydrological models are prepared based on their dimension of features considered. In case of one-dimensional models, only one feature such as the channel flow is considered to understand the flood extent setting up cross-sections, two-dimensional models use grid-like meshes to solve the fluid equations of more than one feature whereas the three-dimensional models uses complex equation to solve multi-dimensional fluid equations(DigitalCommons et al., 2023 ). An extensive study on such models is prepared where five different models are speculated while observing the Waterford River watershed in Canada. These include the GR4J model, a simple conceptual model using a minimum of four to up to nine parameters to model flooding conditions in a region; the MAC-HBV model which is a lumped conceptual model using combined modelling methods where enhanced modelling is carried out considering fifteen parameters such as flow accumulation, soil moisture and non-linear responses; SAC-SMA model is a conceptual watershed model used commonly by the National Weather Services, USA, in operational flood forecasting that uses nineteen input parameters; HEC-HMS model is prepared by the Hydrologic Engineering Centre of the US Army Corps of Engineers to devise simulation hydrologic processes in dendritic watersheds over both space and time; and lastly, WATFLOOD model which is a physically-based distributed hydrological model that subdivides watersheds into Group Response Units (GRU) based on the assumption of surface areas with similar land use and hydrologic processes(Wijayarathne & Coulibaly, 2020 ). Each of these models is utilised to produce different scales of simulations based on the parameters used. Often these parameters help in showing the variations and changes in water levels within a watershed and devise in forecasting the probability of flood hazards. Operational flow forecasting is used to understand the model that is the best fit for predicting the streamflow within a watershed both spatially and temporally. 5.3. Remote Sensing Techniques alongside GIS tools and environments: Remote Sensing techniques are especially beneficial since in situ measurements are often cumbersome to attain due to the inability to obtain point-based data as a region might become inaccessible during flood incidence, the period of high water flow during floods typically does not allow data collection, moreover, problems related to high operational and maintenance costs in rendering data is associated. Flood prediction utilising earth observation data is effective in this sense using satellite data and remote sensing observation to check the flood damage footprints by overlaying elements-at-risk and the hazard affected points then proceed to carry out statistical analysis and mapping(Schumann et al., 2023 ; Van Westen, 2013 ). Employing earth observatory (EO) instruments, there are multiple befitting options such as optical imageries- using MODIS or NOAA VIIRS satellite data to extract flood pixels for making flood binary maps which can be used as an inventory in flood analysis; SAR data can provide radar imageries operable under all-weather conditions extending day-night imaging capabilities involving statistical model, CNNs, thresholding and other techniques using Sentinel-1, Envisat or InSAR coherence; CNNs gives edge in interpretation as well as detection; InSAR helps in enhancement of ground observation utilising the radar data; Lidar DEMs are available at 50cm resolution in altimetric studies with very limited availability; radar altimeters are used at different levels with varying resolutions and gestation periods(Schumann et al., 2023 ). A list of commonly used satellites in remote sensing are shown in Table 2 which assists in flood mapping at present. However, the selection of the satellite data becomes vital to carry out the study across an area that best fits the desired results. This hence requires expert opinions and understanding as well as the availability of data under the time frame of the study. Table 2 A list of satellites along with their sensors that are commonly used in the practice for flood detection and forecast Satellites Sensors Spatial Resolutions Output/ Uses under Desired Parameters NASA’s Terra or Aqua MODIS 250 meters (bands 1 and 2) Provides real-time flooded and non-flooded pixel values to prepare FHM. NOAA VIIRS 375–750 meters at nadir Collects visible and infrared images flood pixels from Earth’s surface. Copernicus Sentinel-1 and Sentinel-2 SAR, InSAR, Multispectral 30 meters (common); up to 50 cm (high resolution); 10–60 meters (MSS in Sentinel-2) Captures radar and multispectral data used in flood detection under all weather conditions, allowing day and night imaging. Envisat ASAR, InSAR 30 meters Detects hazards and floods in real-time, assisting with rapid mapping under any weather conditions. INSAT-3D Multi-spectral Imager 1 km x 1 km (visible/SWIR); 4 km x 4 km (MIR/TIR); 8 km x 8 km (WV) Uses optical radiometry to generate images across six wavelength bands to detect flooded pixels. Satellite altimeters Lidar DEMs 50 cm resolution on the ground Utilises altimetric data for flood detection, often paired with radar altimeters for more accurate results. Radarsat-2 C-band SAR 10 meters (high resolution), up to 100 meters (low resolution) Provides all-weather radar data for flood mapping, particularly in coastal and inland flood areas. TerraSAR-X X-band SAR 1 meter (high resolution), 3 meters (low resolution) Provides high-resolution radar images, effective for detecting flood boundaries, even in adverse weather. Copernicus Sentinel-3 OLCI, SLSTR 300 meters (OLCI); 1 km (SLSTR) Offers multispectral imagery for flood monitoring, especially surface water mapping with optical data. Cartosat-2 High-Resolution Panchromatic 0.8 meters (panchromatic); 2 meters (multispectral) Provides very high-resolution images for flood mapping and monitoring urban and rural flood-prone areas. The global flood detection system (GFDS) estimates changing water levels based on signal information received from satellite imageries, specifically passive microwave sensors. This is assisted by MODIS flood maps prepared by MODIS Near Real-Time Global Flood Mapping Project. Beyond detection, the GloFAS flood forecasting is carried out where land use changes, hydrological changes and meteorological data are obtained from the Ensemble Prediction System (ENS), European Centre for Medium-range Weather Forecasting (ECMWF). Once data is gathered, the magnitude of flood is compared for all the days of reported floods showing that in places where the in situ data was unavailable, and low magnitude floods, remotely sensed GFDS maps gave better results than the GloFAS forecasts(Revilla-Romero et al., 2015 ). The Earth Observations (EO) and Geographical Information Systems (GIS) are integrated well-developed tools in disaster risk management where spatial methodologies can be fully explored throughout the risk assessment processes(Van Westen, 2013 ). C. Van Westen, in his extensive works from 2009 followed by 2013 on “Natural Hazards Assessment and Disaster Risk Management”, clearly stated that in studies with large ROI, data is often unavailable or unreliable, remotely sensed data along with GIS tools based on numerous other factors such as elevation, precipitation, landcover and illumination while sensors are working helps in assistance and guided studies. 5.4. Machine Learning: Machine learning involves the identification of factors triggering flood incidents and training a computer-driven algorithm through training and testing sets of data to become self-reliant in predicting the probability of the hazard reoccurring in the future. These models are data-driven and can handle complex datasets by capturing non-linear relationships adapting to changing conditions(Granata & Di Nunno, 2024 ). This helps in adaptability as a precaution to the disastrous effects of the hazard. The algorithm goes through iterations on the distributed training and testing sets to understand patterns, identify and recognise the same by correlating inputs to achieve desired outputs. Advanced machine learning helps in data viewing and enhances the automation of data visualisation as well as analysis. It assists in data augmentation to interpret predictions(Diaconu et al., 2024 ). While discussing the inclusion of machine learning (ML) algorithms in FSM, data becomes a very vital tool. Thus, extraction, interpolation, utilisation and validation become essential. In almost every study conducted with ML assistance environmental and morphometric factors are utilised. This is mainly to consider multiple criteria including all the flood triggering factors. These include digital elevation model (DEM) data involving slope, elevation, curvature and slope aspect of the basin or catchment being studied. Other environmental data involve precipitation, temperature, topographic wetness index (TWI), flow accumulation, stream-transport index (STI), land use and land cover, and soil porosity data. The factors in inclusion in any study can be as many and as varied as per the requirements and motive of flood maps being prepared. Physiographical data such as the change in population, GDP and economic activities in a region also act as indirect factors inducing floods in any region. The inter-correlation between all these multiple factors is identified as a multi-collinearity problem. The multi-collinearity test is carried out in this context to reduce the bias and collinearity problem(S. Ghosh et al., 2022 ). This is done by calculating tolerance and variance inflation factor (VIF). VIF values better than 5 (significant multicollinearity) and tolerance less than 0.1 directly indicate multi-collinearity problems within the dataset. Based on the values (above the set threshold of acceptance) obtained changes are made in criteria specification and adjustments can be made to achieve the least distortions in outputs, while working with training and test sets. Flood inventory maps are made describing flooded and non-flooded points to be used in the stages of map-making. This is often done using Google Earth Engine (GEE) to find and demarcate ‘presence’ and ‘absence’ points of water in a flooded region (Bashirgonbad et al., 2024 ). The comprehensive sampling based on validation and region-specific segregation of flooded from non-flooded regions is essential in the enhancement of flood maps prepared to delineate the regions prone to risks of flood hazard(Wahba et al., 2024 ). Bashirgonbad et al. in their work also mentioned that based on the VIF values from the multi-collinearity test, the factors most to least responsible in influencing flood in the region can also be assessed. The delineation of basin or catchment boundaries also becomes crucial before considering factors affecting the flood hazard as it will determine the input data which are ultimately responsible for output generation. Researchers have worked on focused studies relating to different models used in flood forecasting. The preparatory data in these studies comprise of the data maps prepared from the DEM data, environmental factors along with the flooded and non-flooded datasets to train the machine learning models such as the support vector regression (SVR) model to not only predict but also build pattern recognition of the hazard; least absolute shrinkage and selection operator (LASSO) linear regression model that serves as variable selection method assisting in reduction of the number of variables used in each tier of the model(Wahba et al., 2024 ). The support vector machine (SVM), a kernel-based technique that falls under supervised machine learning algorithms calculates both associations among the input variables and projects a linear structure by combining the output from training samples(Y. Wang et al., 2020 ); random forest (RF) on the other hand, is based on ensemble ML algorithm where it creates multiple classification trees and then average the output to predict the data(Bashirgonbad et al., 2024 ; S. Ghosh et al., 2022 ). Forecasts based on daily data are carried out using daily precipitation and flow rate data as inputs in artificial neural network (ANN) and recurrent neural network (RNN) models for a flood-prone area. The long short-term memory (LSTM) model is used for one to three-day flow rate forecasting exhibiting outstanding advantages in its ability to learn short dependencies effectively, and this model can completely be applied to forecast the flow two days or even three days ahead with accuracy of over 86%(Le et al., 2019 ). For terrain or hydrological variations consideration in the forecasting model, LSTM-NN turns out to be an extremely effective data-driven method to be utilised. However, limitations exist as physical differences are not extensively considered and thus, limits the model's effective parameterisation. It successfully gives highly accurate predictions at specific stations. Assessment of the accuracy of the models is carried out by various statistical validation methods. Area under receiver operating characteristic curve (AUC-ROC), residual analysis using R-squared error calculation, mean squared error (MSE) and mean absolute error (MAE) calculations to validate the FSM map prepared(Wahba et al., 2024 ). Other techniques involve the Kling Gupta efficiency (KGE), mean absolute percentage error (MAPE)(Granata & Di Nunno, 2024 ); Friedman test and Wilcoxon Signed Rank test(S. Ghosh et al., 2022 ) to validate the outputs and identify the model most fitting in forecasting of flood in the region of study. On an ensemble model based on multiple meta-learners, KGE renders more precise error detection than r-squared error. The accuracy also considers various statistical indicators such as sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV). 5.5. Integrated Approaches: Hybrid or Ensemble Models In the recent decade of the articles studied it is noted that there is an accent of ML algorithms being widely used in the forecast of a hazard. A few of the more recent studies also show the coupled implementation of efforts put toward the development of hybrid or ensemble models. They provide advantages in terms of levels of accuracy, model stability and reduction of over or under-fitting of models. Stacked models made in two or more layers can be effective ensemble models to study and forecast short to medium-period data(Granata & Di Nunno, 2024 ). Different basins and their varying stream or water flow rates can be extensively studied. Models, integrated ensembles, or simple regressions, are a simplified depiction of the reality where choosing the input factors becomes extremely crucial and sensitive to the output. Francesco Granata and Fabio Di Nunno(Granata & Di Nunno, 2024 ) in their ensemble study of short-term and long-term streamflow used a multilayer perceptron algorithm combined with the random forest algorithm to build the MLP-RF, a stacked forecasting deep learning model, utilised to forecast flood data. Comparison is drawn with different meta-learners such as isotonic regression (IR), pace regression (PR), and radial basis function (RBF-NN) neural networks. Each of these regressors works and gives variable results over different frames of time ranging from one day to up to 15 days forecast or short to medium range forecasting. Out of these, RBF-NN is seen to be returning effective results even with less data in the training set. Comparison analysis depicts that the predictions became more delicate and precise toward the longer period of a 15-day forecast of streamflow in a basin. This comparison is also drawn against Elastic Net (EN), which uses meta-learners between two different stacked models. Some other literature showed using artificial neural network multi-layer perceptron (ANN-MLP) regressor model is used to predict the chances of reoccurrence of flood in an already flooded area; gradient boosting regressor (GBR) which simply helps in boosting other weak models specifically decision tree model by combining two models to increase its efficiency(Wahba et al., 2024 ); the species distribution modelling (SDM) is used to combine strengths of SVM and RF models in a weighted manner to have the best FSM mapping outcome with enriched prediction abilities.(Bashirgonbad et al., 2024 ). The stacking of models is generally done on multiple levels. Level 0 models consist of multiple models that are trained on the same data utilising different algorithms and hyperparameters. Level 1 also called ‘meta-level’ uses outputs from level 0 as inputs in this level to train the model to predict the occurrences of the phenomenon in question. Studies like that conducted by Diaconu et al. recognise dependent factors to understand the condition of flood proneness in the study area. It utilises four ML techniques- Deep Learning Neural Network- Statistical Index (DLNN-SI), Particle Swarm Optimization-Deep Learning Neural Network- Statistical Index (PSO-DLNN-SI), Support Vector Machine-Statistical Index (SVM-SI), and Particle Swarm Optimization-Support Vector Machine-Statistical Index (PSO-SVM-SI)(Diaconu et al., 2024 ). The Flood Susceptibility Index (FSI) calculated for each of the models helps in understanding the percentage of whether the areas in the region are vulnerable or not to flooding. The Gain Ratio method is utilised where slope has the highest predictive ability whereas rainfall, hydrological soil index and convergence index have low predictive ability. Overall, the PSO ensembled with DLNN and SVM improved the efficiency and PSO-DLNN-SI outperformed all the other models. Another paper by W. Wu et al., prospects into understanding the improvement of technology and inclusion of ML in predicting floods and understanding better data assimilation and the extent of flood inundation. The study attempts to extend the forecast variable from simple river flow dynamics to inundation variability and water flow changes as an important consideration aspect in ensemble forecasting, targeting longer prediction periods, seasonal variations and climatic constraints(W. Wu et al., 2020 ). Seamless prediction is preferred combining short-term and long-term forecasts. Thus, the ability of ensemble studies elaborately shows the enormous capability of hybrid ML algorithms in prediction and forecast technology. 6. Discussion Monitoring and forecasting flood data is an extensive and prolonged study of all the factors that influence the hazard. This aims ultimately at understanding the ‘factors-at-risk’ and routing towards their mitigation. Thus, a careful evaluation of each technique is attempted in this review article. In order to determine flood vulnerabilities identification of sources of danger is highly necessary. Knowledge of these sources is gathered from different techniques of monitoring and observations. On comparing the various techniques used in flood monitoring it is seen that the role of sensors and communities is primary during the initiation period of flood. GIS tools and software become incredibly essential to map and build simulations in real-time to look at the affected region. In places of inaccessibility, remote sensing and satellite imageries give high spatial resolution images to carry out necessary investigations. ML algorithms and neural networking are of the highest efficiency at present in building elaborate and active models as per requirements. Utilisation of multiple criteria to include all the regional and local factors triggering floods is essential in this regard for both enhancements as well as the accuracy of the model. It can foresee the occurrence of the hazard reducing uncertainties while combating the effects of flood in an area. In these terms, integrated machine learning algorithms are noted to be most effective in building powerful models supporting the prediction of streamflow, river flow and other factors affecting flood hazards. The use of ensemble models needs careful selection and calibration as there are chances of increased complexity in computations and prone to multi-layered interpretations. However, when taken well care of, it poses out to be one of the best methods in the prediction of flood situations giving a robust, efficient and highly accurate level of prediction in both short-term and long-term forecasting. The development and implementation of any model effectively also depend on its ability to reproduce results over similar scenarios using variables and factors as per requirements of the basin or channel being studied for the hazard assessment. The assessment of multi-hazards and their risk monitoring is an extremely data-intensive procedure where the availability of certain types of spatial data is unavailable(Van Westen, 2013 ). This is more prevalent in developing nations, where hazard inventories are scarce and underdeveloped, often unfit for use in analysis. The uncertainties related to flood forecasting can be tackled using either statistical or non-statistical approaches. Due to the inherent heterogeneous nature of physical processes related to hydrological or river basins, non-statistical approaches are preferred over statistical approaches(Montanari & Grossi, 2008 ). The problems related to missing data are also correlated effect of insufficient maintenance and the inability to acquire consistent data collection in regular intervals. In the context of developing nations, the availability of stations in the basin being studied is so scarce that dependency on data in carrying out a study is cumbersome and less encouraged. Large-scale research and development are being carried out to enhance this very problem and assistance from the government through directories and district-wise data handbooks is under progress. This paper also showed a collective work on satellite-driven data extraction for data-sparse regions(Revilla-Romero et al., 2015 ), showing how each has its advantages and disadvantages. In the case of data derived from news reports, it is often seen to be insufficient or biased. Remote sensing data is rather insufficient for narrow mountain valleys facing flash floods whereas cloud cover is a major problem in viewing flood-affected regions using satellite imageries. Satellite missions and remote sensing techniques are being developed for directly measuring water depths and flow monitoring in basins and channels. Currently, significant attention is being directed towards flood monitoring as it affects and has a widespread impact on the global population. Enhanced understanding and rigorous research in this area has become imperative. Advances in satellite missions and the increasing acceptance of technology in managing flood hazards offer hope to communities around the world. In particular, the integration of artificial intelligence (AI) and machine learning (ML) algorithms is opening new avenues for forecasting flood risks, helping to reduce the uncertainties surrounding this natural hazard. The development of ensemble models and the refinement of algorithms to train predictive software are key areas of focus, shaping a future that is better equipped to both address and mitigate the impacts of floods. 7. Conclusion One of the major challenges related to flood hazards is the inherent uncertainty associated with them. Monitoring of the hazard plays a crucial role in addressing this issue. In addition to monitoring, predictive analysis and forecasting are essential tools to assist in future preparedness and adaptability. This review aims to provide a comprehensive understanding of flood hazards, focusing on their occurrence and the application of technologies and advancements to address the associated challenges. Among the various methodologies available, the choice of an appropriate model depends on factors such as the size of a region, its physiographic conditions, and, to some extent, the anthropogenic activities contributing to changes associated with the region. The key findings of this paper are summarised as follows: The paper addresses the primary objective of synthesising existing knowledge and collecting techniques and methods to assess flood extents and their impacts. In discussing various methodologies, the paper acknowledges the challenges associated with each method individually and in comparison. It also evaluates data acquisition, assimilation, and availability through an extensive review process. Contemporary methods are thoroughly examined to facilitate their multi-faceted application, allowing for a broader understanding of floods not just as a temporal event, but also as a spatial phenomenon. The use of GIS and advanced machine learning techniques for flood prediction and forecasting is emphasised. The review highlights that ensemble and integrated machine learning algorithms are most effective in this context. Best practices and optimal methods are identified, offering practical insights for implementation through multidisciplinary approaches. The paper also stresses the importance of involving experts from diverse disciplines, making it accessible for them to grasp the requirements related to flood hazard monitoring and prediction, encouraging their contributions. Overall, this review article provides a composite analysis of methods that can be directly applied in real-world scenarios, benefiting communities and fostering advancements in the field. It stresses the application of the techniques reviewed to ultimately contribute to their implementation to bring changes. It encourages researchers, government agencies, and organizations to leverage the findings to explore new directions and approaches in flood hazard studies. Declarations Acknowledgments The authors acknowledge the support and motivation from the Department of Geography, Banaras Hindu University, Varanasi towards the completion of this paper. Authors contributions All authors contributed to the conceptualisation and design of the study. Literature studying and manual selection of articles for review purposes were carried out by the authors, Adrita Talapatra and Narendra Kumar Rana. The first draft of the manuscript was written by Adrita Talapatra and was redrafted with comments and considerations from Narendra Kumar Rana. The final manuscript was read and approved by the authors. Fundings No funding was received for this work. Data availability All data and material are published and available on request from author ( [email protected] ) Ethics approval and consent to participate Not applicable. Consent to publish Both author and co-author consented to the publication of the manuscript. Competing interests There is no conflict of interest among the authors. Clinical trial number Not applicable. References Appeaning Addo, K., Jayson-Quashigah, P. N., Codjoe, S. N. A., & Martey, F. (2018). Drone as a tool for coastal flood monitoring in the Volta Delta, Ghana. Geoenvironmental Disasters , 5 (1). https://doi.org/10.1186/s40677-018-0108-2 Aria, M., & Cuccurullo, C. (2017). bibliometrix : An R-tool for comprehensive science mapping analysis. Journal of Informetrics , 11 (4), 959–975. https://doi.org/10.1016/j.joi.2017.08.007 Arshad, B., Ogie, R., Barthelemy, J., Pradhan, B., Verstaevel, N., & Perez, P. (2019). Computer vision and iot-based sensors in flood monitoring and mapping: A systematic review. Sensors (Switzerland) , 19 (22), 1–28. https://doi.org/10.3390/s19225012 Assumpção, T. H., Popescu, I., Jonoski, A., & Solomatine, D. P. (2018). Just a me-too: No advantage in anaemia of chronic renal failure. Prescrire International , 12 (63), 14–16. Azizi, K., Diko, S. K., Saija, L., Zamani, M. G., & Meier, C. I. (2022). Integrated community-based approaches to urban pluvial flooding research, trends and future directions: A review. Urban Climate , 44 (July), 101237. https://doi.org/10.1016/j.uclim.2022.101237 Bandyopadhyay, S., Ghosh, P. K., Jana, N. C., & Sinha, S. (2016). Probability of flooding and vulnerability assessment in the Ajay River, Eastern India: implications for mitigation. Environmental Earth Sciences , 75 (7). https://doi.org/10.1007/s12665-016-5297-y Bashirgonbad, M., Farokhzadeh, B., & Gholami, V. (2024). Enhancing flood mapping through ensemble machine learning in the Gamasyab watershed, Western Iran. Environmental Science and Pollution Research , March 2019 . https://doi.org/10.1007/s11356-024-34501-5 Bera, B., Bhattacharjee, S., Sengupta, N., Chamling, M., Ghosh, S., & Ghosh, A. (2024). Impacts of channelization of River Bala, eastern Himalayan foothills, India. Geology, Ecology, and Landscapes , 8 (3), 241–253. https://doi.org/10.1080/24749508.2022.2130857 Das, B. K. (2012). Losing Biodiversity , Impoverishing Forest Villagers : Analysing forest policies in the context of Flood Disaster in a National Park of Sub Himalayan Bengal , India Losing Biodiversity , Impoverishing Forest Villagers : Analysing forest policies in the co . 91 (July). Diaconu, D. C., Costache, R., Towfiqul Islam, A. R. M., Pandey, M., Pal, S. C., Mishra, A. P., & Pande, C. B. (2024). Developing flood mapping procedure through optimized machine learning techniques. Case study: Prahova river basin, Romania. Journal of Hydrology: Regional Studies , 54 (June), 101892. https://doi.org/10.1016/j.ejrh.2024.101892 DigitalCommons, U., All Graduate Theses, U., & Kesler, T. (2023). Comparing 1D, 2D, and 3D Hydraulic Models in Urban Flooding Comparing 1D, 2D, and 3D Hydraulic Models in Urban Flooding Applications Applications . https://digitalcommons.usu.edu/etd/8691 Gharbi, M., Soualmia, A., Dartus, D., & Masbernat, L. (2016). Floods effects on rivers morphological changes application to the Medjerda River in Tunisia. Journal of Hydrology and Hydromechanics , 64 (1), 56–66. https://doi.org/10.1515/johh-2016-0004 Ghosh, D., & Saha, S. (2014). Channel bed aggradation in relation to channel Morphometry : A case study of river Jainti , Jalpaiguri , West Bengal. INTERNATIONAL JOURNAL OF GEOMATICS AND GEOSCIENCES Volume 5, No 1, 2014 Research , 5 (1), 192–208. Ghosh, S., Saha, S., & Bera, B. (2022). Flood susceptibility zonation using advanced ensemble machine learning models within Himalayan foreland basin. Natural Hazards Research , September . https://doi.org/10.1016/j.nhres.2022.06.003 Granata, F., & Di Nunno, F. (2024). Forecasting short- and medium-term streamflow using stacked ensemble models and different meta-learners. Stochastic Environmental Research and Risk Assessment , 20 (12), 2175–2177. https://doi.org/10.1007/s00477-024-02760-w Han, S., & Coulibaly, P. (2017). Bayesian flood forecasting methods: A review. Journal of Hydrology , 551 , 340–351. https://doi.org/10.1016/j.jhydrol.2017.06.004 Hapuarachchi, H. A. P., Wang, Q. J., & Pagano, T. C. (2011). A review of advances in flash flood forecasting. Hydrological Processes , 25 (18), 2771–2784. https://doi.org/10.1002/hyp.8040 Helmrich, A. M., Ruddell, B. L., Bessem, K., Chester, M. V., Chohan, N., Doerry, E., Eppinger, J., Garcia, M., Goodall, J. L., Lowry, C., & Zahura, F. T. (2021). Opportunities for crowdsourcing in urban flood monitoring. Environmental Modelling and Software , 143 , 105124. https://doi.org/10.1016/j.envsoft.2021.105124 Henriksen, H. J., Roberts, M. J., van der Keur, P., Harjanne, A., Egilson, D., & Alfonso, L. (2018). Participatory early warning and monitoring systems: A Nordic framework for web-based flood risk management. International Journal of Disaster Risk Reduction , 31 , 1295–1306. https://doi.org/10.1016/j.ijdrr.2018.01.038 Holman, R. A., & Stanley, J. (2007). The history and technical capabilities of Argus. Coastal Engineering , 54 (6–7), 477–491. https://doi.org/10.1016/j.coastaleng.2007.01.003 Iqbal, U., Riaz, M. Z. Bin, Zhao, J., Barthelemy, J., & Perez, P. (2023). Drones for Flood Monitoring, Mapping and Detection: A Bibliometric Review. Drones , 7 (1), 1–29. https://doi.org/10.3390/drones7010032 Kourgialas, N. N., & Karatzas, G. P. (2011). Gestion des inondations et méthode de modélisation sous SIG pour évaluer les zones d’aléa inondation-une étude de cas. Hydrological Sciences Journal , 56 (2), 212–225. https://doi.org/10.1080/02626667.2011.555836 Le, X. H., Ho, H. V., Lee, G., & Jung, S. (2019). Application of Long Short-Term Memory (LSTM) neural network for flood forecasting. Water (Switzerland) , 11 (7). https://doi.org/10.3390/w11071387 Ma, C., Qiu, D., Mu, X., & Gao, P. (2022). Morphological Evolution Characteristics of River Cross-Sections in the Lower Weihe River and Their Response to Streamflow and Sediment Changes. Water (Switzerland) , 14 (21). https://doi.org/10.3390/w14213419 Macherera, M., & Chimbari, M. (2016). A review of studies on community based early warning systems. Journal of Disaster Risk Studies , 1–11. Merkuryeva, G., Merkuryev, Y., Sokolov, B. V., Potryasaev, S., Zelentsov, V. A., & Lektauers, A. (2015). Advanced river flood monitoring, modelling and forecasting. Journal of Computational Science , 10 , 77–85. https://doi.org/10.1016/j.jocs.2014.10.004 Montanari, A., & Grossi, G. (2008). Estimating the uncertainty of hydrological forecasts: A statistical approach. Water Resources Research , 44 (12), 1–9. https://doi.org/10.1029/2008WR006897 Mudashiru, R. B., Sabtu, N., Abustan, I., & Balogun, W. (2021). Flood hazard mapping methods: A review. Journal of Hydrology , 603 (March). https://doi.org/10.1016/j.jhydrol.2021.126846 Park, S., Lee, N., Han, Y., & Hahn, H. (2009). The Water Level Detection Algorithm using Accumulated Histogram with Band Pass Filter. International Conference on Signal Processing, Communications and Networking 2009 . Paul, A., & Biswas, M. (2019). Changes in river bed terrain and its impact on flood propagation–a case study of River Jayanti, West Bengal, India. Geomatics, Natural Hazards and Risk , 10 (1), 1928–1947. https://doi.org/10.1080/19475705.2019.1650124 Perks, M. T., Russell, A. J., & Large, A. R. G. (2016). Technical note: Advances in flash flood monitoring using unmanned aerial vehicles (UAVs). Hydrology and Earth System Sciences , 20 (10), 4005–4015. https://doi.org/10.5194/hess-20-4005-2016 Revilla-Romero, B., Hirpa, F. A., Pozo, J. T. del, Salamon, P., Brakenridge, R., Pappenberger, F., & de Groeve, T. (2015). On the use of global flood forecasts and satellite-derived inundation maps for flood monitoring in data-sparse regions. Remote Sensing , 7 (11), 15702–15728. https://doi.org/10.3390/rs71115702 Schumann, G., Giustarini, L., Tarpanelli, A., Jarihani, B., & Martinis, S. (2023). Flood Modeling and Prediction Using Earth Observation Data. Surveys in Geophysics , 44 (5), 1553–1578. https://doi.org/10.1007/s10712-022-09751-y Van Westen, C. J. (2013). Remote Sensing and GIS for Natural Hazards Assessment and Disaster Risk Management. Treatise on Geomorphology: Volume 1-14 , 1 – 14 , 259–298. https://doi.org/10.1016/B978-0-12-374739-6.00051-8 Vousdoukas, M. I., Pennucci, G., Holman, R. A., & Conley, D. C. (2011). A semi automatic technique for Rapid Environmental Assessment in the coastal zone using Small Unmanned Aerial Vehicles (SUAV). Journal of Coastal Research , SPEC. ISSUE 64 , 1755–1759. Wahba, M., Sharaan, M., Elsadek, W. M., Kanae, S., & Hassan, H. S. (2024). Examination of the efficacy of machine learning approaches in the generation of flood susceptibility maps. Environmental Earth Sciences , 83 (14), 429. https://doi.org/10.1007/s12665-024-11696-x Wang, R. Q., Mao, H., Wang, Y., Rae, C., & Shaw, W. (2018). Hyper-resolution monitoring of urban flooding with social media and crowdsourcing data. Computers and Geosciences , 111 , 139–147. https://doi.org/10.1016/j.cageo.2017.11.008 Wang, Y., Sun, D., Wen, H., Zhang, H., & Zhang, F. (2020). Comparison of Random Forest Model and Frequency Ratio Model for Landslide Susceptibility Mapping (LSM) in Yunyang County (Chongqing, China). International Journal of Environmental Research and Public Health , 17 (12), 4206. https://doi.org/10.3390/ijerph17124206 Wijayarathne, D. B., & Coulibaly, P. (2020). Identification of hydrological models for operational flood forecasting in St. John’s, Newfoundland, Canada. Journal of Hydrology: Regional Studies , 27 (March 2019), 100646. https://doi.org/10.1016/j.ejrh.2019.100646 Wu, W., Emerton, R., Duan, Q., Wood, A. W., Wetterhall, F., & Robertson, D. E. (2020). Ensemble flood forecasting: Current status and future opportunities. Wiley Interdisciplinary Reviews: Water , 7 (3), 1–32. https://doi.org/10.1002/WAT2.1432 Wu, X., Shen, X., Li, J., & Xie, X. (2024). Determination and projection of flood risk based on multi-criteria decision analysis (MCDA) combining with CA-Markov model in Zhejiang Province, China. Urban Climate , 53 (October 2023). https://doi.org/10.1016/j.uclim.2023.101769 Zanchetta, A. D. L., & Coulibaly, P. (2020). Recent advances in real-time pluvial flash flood forecasting. Water (Switzerland) , 12 (2). https://doi.org/10.3390/w12020570 Cite Share Download PDF Status: Published Journal Publication published 07 Jan, 2026 Read the published version in Environmental Science and Pollution Research → Version 1 posted Reviewers agreed at journal 08 Jul, 2025 Reviewers invited by journal 08 Jul, 2025 Editor assigned by journal 25 Jun, 2025 First submitted to journal 23 Jun, 2025 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-6948732","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":482384650,"identity":"173cc15a-0c50-40d6-b97f-605c252b1448","order_by":0,"name":"Adrita Talapatra","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAABEElEQVRIiWNgGAWjYFCCBIYDIIoPypUDEQceENJygMGAgQ2slIHBGEwnENDCgKwlsQEqiBPwt+c+PPyh5o88G/vZg58/7rFLnx92+CHQFjs53QbsWiTOPDc4cOCYgWEbT16yxIFnybkbb6cZALUkG5sdwGHNjTSgX9gMGNsYcgwkDhxgzt04OwGk5UDiNhxa5MFa/hnYt/G/Mf5x4EB9uuHs9A94tRiAtBxsM0hsk8gxA9pyOEFeOge/LYZnnjEcONtnnNwm8S7N4syB44YbpHMKDiQY4PaL3PE05g8V3+Rs+/lzD9+oOFAtLz87ffOHDxV2cji9jwA8UKeCVRoQVI6kRb6BKNWjYBSMglEwggAATqprrcSa17QAAAAASUVORK5CYII=","orcid":"https://orcid.org/0009-0008-2088-1657","institution":"Banaras Hindu University","correspondingAuthor":true,"prefix":"","firstName":"Adrita","middleName":"","lastName":"Talapatra","suffix":""},{"id":482384651,"identity":"e5bcf7b6-9f83-47eb-b012-317980940a8b","order_by":1,"name":"Narendra Kumar Rana","email":"","orcid":"","institution":"Banaras Hindu University","correspondingAuthor":false,"prefix":"","firstName":"Narendra","middleName":"Kumar","lastName":"Rana","suffix":""}],"badges":[],"createdAt":"2025-06-22 09:30:47","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-6948732/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-6948732/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1007/s11356-025-37366-4","type":"published","date":"2026-01-07T15:58:16+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":86655699,"identity":"7603ed74-b8e0-42dc-8cbc-2d4eb06a2a28","added_by":"auto","created_at":"2025-07-14 10:14:54","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":398336,"visible":true,"origin":"","legend":"\u003cp\u003ePRISMA 2020 flow diagram showing records from databases and registers\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-6948732/v1/cf71113f3c388110dd581e00.png"},{"id":86657290,"identity":"618977a0-e0dd-478a-911b-d6aae78a41d9","added_by":"auto","created_at":"2025-07-14 10:22:54","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":32228,"visible":true,"origin":"","legend":"\u003cp\u003eAnnual scientific production across databases (Generated by authors)\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-6948732/v1/0444cec9a6f0804dee75d5fa.png"},{"id":86655698,"identity":"d6f24e0e-f0c5-4b42-b462-2cc97c3790ca","added_by":"auto","created_at":"2025-07-14 10:14:54","extension":"jpeg","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":173147,"visible":true,"origin":"","legend":"\u003cp\u003eGlobal scientific production map (Generated by authors through Biblioshiny, India, 2022)\u003c/p\u003e","description":"","filename":"3.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-6948732/v1/b0d707a4a501894d74d5ee9e.jpeg"},{"id":86657294,"identity":"82bfeaba-87c3-4bc2-986d-b850f702eb73","added_by":"auto","created_at":"2025-07-14 10:22:54","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":175450,"visible":true,"origin":"","legend":"\u003cp\u003eThree factor analysis showing interdisciplinary nature of the study\u003c/p\u003e\n\u003cp\u003e(Generated by authors through Biblioshiny, India, 2022)\u003c/p\u003e","description":"","filename":"4.png","url":"https://assets-eu.researchsquare.com/files/rs-6948732/v1/2ec3c3913207408c8d601686.png"},{"id":100070733,"identity":"98a14dc8-b0ea-43de-8f79-42f8aac63582","added_by":"auto","created_at":"2026-01-12 16:18:20","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1546745,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6948732/v1/9ed7959d-d292-4e7e-95e6-0ca788a02ce2.pdf"}],"financialInterests":"","formattedTitle":"Recent Advances in Flood Monitoring \u0026amp; Prediction Methods: A Systematic Review","fulltext":[{"header":"1. Introduction","content":"\u003cp\u003eFloods in the present century stand as one of the prime adversities affecting both developed and developing countries despite their physiographic or morphological stability. In a recent evaluation made by C. Van Westen, under the influence of changing hydro-meteorological conditions, flood globally accounts for 30% of total losses caused by environmental hazards ranking as the second most disastrous event after losses from earthquakes(Van Westen, \u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e2013\u003c/span\u003e). The global vulnerability to floods has increased by more than 40% over the past two decades accounting for heavy urbanisation and global climate change(Wahba et al., \u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). This is only projected to grow in the future with multitudes rise in its frequency and severity(Diaconu et al., \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e2024\u003c/span\u003e; Wahba et al., \u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). From a scientific standpoint, the ability to predict an uncertain phenomenon effectively contributes to assessment as well as adaptability to the calamities related to the hazard. Flood Hazard Mapping (FHM), automatically becomes an essential tool in this context. This along with historical data when utilised with understanding and specification becomes a powerful assistance in the identification of hazard potential, areas, extent, depth and spatial damage potential(Mudashiru et al., \u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e2021\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eThe collective objectives of this review article are highlighted in the following points.\u003c/p\u003e\u003cp\u003e\u003cul\u003e\u003cli\u003e\u003cp\u003eSynthesis of the existing knowledge base is the primary aim to compile and analyse the trends in the techniques and methods used in existing literatures and discuss their utility while drawing attention to existing gaps.\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003eAn array of methodologies used in different flood monitoring and prediction studies to be discussed through evaluation to understand the present trend as well as the technological standpoint.\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003eThe findings from the literatures would be effectively used in an attempt to point out data types used in visualising flood hazards and projecting their occurrences. The shortcomings in data availability to be highlighted alongside as a part of the findings.\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003eExamination of contemporary technological advancements, highlighting the abilities of GIS and machine learning involved in actively assisting the enhancement of flood monitoring and prediction capabilities.\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003eBased on the case studies and evaluations from existing literature, best practices and methods to be underlined for assistance in flood hazard risk assessment, its monitoring, evaluation of factors-at-risk, damage assessment and more elaborately the prediction that can be applied in all the different contexts.\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003eFinally, the present study ventures into encouraging specialists from various disciplines such as hydrologists, climatologists, engineers, urban planners, and policymakers to come together in collaboration to develop holistic flood risk management strategies.\u003c/p\u003e\u003c/li\u003e\u003c/ul\u003e\u003c/p\u003e\u003cp\u003eThe Centre for Research on the Epidemiology of Disasters (CRED) reports a total of more than 2156 flood occurrences over the last three decades making this study the prime of the hour to bring effective ways to look at a natural calamity that affects more than 2.6\u0026nbsp;billion of lives and huge sum of economic losses worldwide(Bandyopadhyay et al., \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e2016\u003c/span\u003e). The review article assists in streamlining the knowledge of the recent trends relating to means of monitoring and predicting flood hazards. It is an effort to bring out elaborate documentation of compiled techniques and measures to address the improvisations in preparing FHM and FSM maps. The paper tries to pinpoint the root causes while accounting implementation of evolved solutions to one of the most serious natural hazards of the twenty-first-century world.\u003c/p\u003e"},{"header":"2. Methodology","content":"\u003cp\u003eLiteratures were selected using the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines. A systematic review is targeted keeping in mind the specific inclusion and exclusion criteria to assess the relevance of the records being studied(Arshad et al., \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). The stepwise process of evaluation primarily included quantitative investigations based on keyword searches across three different databases namely- Scopus, Web of Science and Science Direct (Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e). This returned a total of n\u0026thinsp;=\u0026thinsp;12,889 manuscript records from all three databases.\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eQuantitative results based on the keywords search across different scholarly databases\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"4\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eSearch Keywords\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eScopus\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003eWeb of Science\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003eScience Direct\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u0026ldquo;recent AND techniques AND in AND flood AND monitoring AND forecasting\u0026rdquo;\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e26\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e18\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e918\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u0026ldquo;flood AND monitoring AND techniques\u0026rdquo;\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e1129\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e54\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e3753\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u0026ldquo;flood AND forecasting AND techniques\u0026rdquo;\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e1184\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e34\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e1384\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u0026ldquo;hydrological AND monitoring AND forecasting\u0026rdquo;\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e794\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e20\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e878\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u0026ldquo;flood AND prediction AND recent AND trends\u0026rdquo;\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e76\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e11\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e1897\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u0026ldquo;advanced AND flood AND monitoring AND forecasting\u0026rdquo;\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e53\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e12\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e648\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eTotal\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e3262\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e149\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e9478\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003eThis was followed by the removal of duplicates with assistance from the referencing application Zotero called the \u0026lsquo;Duplicate Items\u0026rsquo; feature. From here, a tedious manual assessment was conducted where filtering was carried out based on titles. This resulted in screening a total of 2694 records for inclusion. Afterward, the inclusion and exclusion criteria (discussed later in this section) were included in order to validate the manuscripts taken into consideration for this review report (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e). The final report considers a total of 56 extensively reviewed articles to bring out the desired findings.\u003c/p\u003e\u003cp\u003eThe present work is achieved through record analysis assisted by systematic literature studies and cautious selections. This is done based on layers of inclusion and exclusion criteria as stated below-\u003c/p\u003e\u003cp\u003e1. General criteria: The timeline for the study is set from 2007 to 2024. The extensive period of 17 years elaborately depicts the changes in the methodological trends over time.\u003c/p\u003e\n\u003cp\u003e2. Preference in first order: Here, the precise subject domain is selected for narrowing the search criteria. The following subject areas are considered to be included:\u0026nbsp;\u003c/p\u003e\n\u003cp\u003ei.\u0026nbsp;Earth and Planetary Sciences,\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eii.\u0026nbsp; \u0026nbsp; \u0026nbsp;Social Sciences, and\u003c/p\u003e\n\u003cp\u003eiii.\u0026nbsp; \u0026nbsp;\u0026nbsp;Environmental Sciences.\u003c/p\u003e\n\u003cp\u003e3. Preference in second order- The types of registers considered need to be regularised and limited to certain preset to maintain similarities and lesser variations in the findings collected reviewing the existing literature. The following registers are considered to be included:\u003c/p\u003e\n\u003cp\u003ei.\u0026nbsp;Conference Papers,\u003c/p\u003e\n\u003cp\u003eii.\u0026nbsp; \u0026nbsp;Articles, here the inclusion was considered for case-study-based and application-based studies, and\u003c/p\u003e\n\u003cp\u003eiii. \u0026nbsp;Review Reports.\u003c/p\u003e\u003cp\u003eVisualisation of the studied literature further proceeds using the Biblioshiny web interface (Aria \u0026amp; Cuccurullo, \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2017\u003c/span\u003e) and Microsoft Excel. The annual scientific production (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e), shows the overall pieces of literature produced over the three different databases during the suggested review period. The graphical representation shows an increasing trend in flood monitoring and prediction studies in the recent decade. It further shows that articles, papers and journals dominate the later part of the study period (2016\u0026ndash;2024) showing a steady rise in annual production levels over all the databases.\u003c/p\u003e\u003cp\u003eMoreover, the global scientific production map (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e3\u003c/span\u003e) generated in Biblioshiny using R-Studio, illustrates the country-wise production of literature in the field of study. The largest production is contributed by USA, followed closely by the China. India ranks as the fourth largest contributor in the field right after Italy. Among these the most cited country is USA with almost 3,404 citations within the 17 years study span (data based on Biblioshiny, India results).\u003c/p\u003e\u003cp\u003eThe three-factor analysis (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e4\u003c/span\u003e) generated in Biblioshiny using R-Studio, shows the relationship between the sources (SO), countries (AU_CO) and the focused keywords (DE) in the literature to show the nature of interdependencies. It also tries to show the interdisciplinary nature of the present study. The review tried to analyse the various techniques and methods from different fields to bring out a collaborative approach to show recent trends in flood monitoring and forecasting.\u003c/p\u003e"},{"header":"3. Overview of Flood Hazards","content":"\u003cp\u003eFlood hazards can be classified into various forms and divisions based on both physiological as well as meteorological drivers. Under broad divisions, they can be classified as floods resulting from catastrophic events ensuing in dam failures or glacial lake outbursts(W. Wu et al., \u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e2020\u003c/span\u003e). These are often associated with improper standards of hard engineering imposed on riverbeds(Bera et al., \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e2024\u003c/span\u003e), or causative events such as groundwater overflow and extensive rainfall(Zanchetta \u0026amp; Coulibaly, \u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e2020\u003c/span\u003e). The other major division would include conditions of flash floods resulting from either pluvial or fluvial indices. Pluvial flash floods (PFFs) are caused by overwhelming localised convective precipitation, orogenic precipitation, sudden cyclonic events, or storm surges in an area(Zanchetta \u0026amp; Coulibaly, \u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e2020\u003c/span\u003e). Flash floods are typically characterised by their rapid onset (within six hours of rainfall onset) often paired with largescale destructions caused by landslides, extreme sediment depositions, bridge and dam failures, increased urbanisation and demographic imbalances in a region(Hapuarachchi et al., \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e2011\u003c/span\u003e). Apart from this, coastal flooding often results from aggrieved coastlines resulting from erosion or problems related to siltation(Appeaning Addo et al., \u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e2018\u003c/span\u003e). Sometimes, floods result from factors such as sediment transport phenomenon as seen in the case of young river channels(Gharbi et al., \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e2016\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eFloods are often associated with soil erosion, mass-wasting, landslides and geo-climatic distress. In order to have timely assistance in flood management it is rather very important to understand the triggering factors such as meteorological and hydrological drivers in the region. This along with heavy rainfall often causes flash floods, especially in regions of shallow stream beds(Das, \u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e2012\u003c/span\u003e). Channel bed aggradation resulting in changes in river bed terrain is also a direct impact of flooding and has a clear relationship between each other(D. Ghosh \u0026amp; Saha, \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e2014\u003c/span\u003e; Paul \u0026amp; Biswas, \u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). The impact of flood is also evident in regions where sedimentation as an outcome of extensive downwasting from heavy rainfall has resulted in changing the channel morphology affecting the banks on either side of the river(Ma et al., \u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e2022\u003c/span\u003e; Paul \u0026amp; Biswas, \u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). These phenomena are evident in the sub-Himalayan foothills along the northern parts of West Bengal among others. In multiple studies throughout the selected timeline, a direct relationship exists with rainfall as well as the slope of a region. Greater slope associated with heavier rainfall has resulted in extensive flood conditions from flow accumulation in river channels in a small period often resulting in flash floods or extreme cases of pluvial floods. This is often related to analysis and prediction of flood hazard where the \u0026ldquo;lag time\u0026rdquo; or the time delay between the peak intensity of rainfall and peak discharge event is considered to enhance prediction of the chances of flooding as well as assessing the associated risks. Understanding lag time differences becomes crucial in selection of methods as it determines the pace of flooding within a watershed. Elevation, soil and the land cover of the areas are also seen to have an effective relation with the flood hazard and pose to be necessary factors in FHM.\u003c/p\u003e"},{"header":"4. Recent Methods for Monitoring Flood Hazards","content":"\u003cp\u003eMonitoring of flood events is often done through satellite-driven, real-time observation of the flood incidents aiming to build awareness and promptness in understanding and combating the hazard and its atrocities(Revilla-Romero et al., \u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e2015\u003c/span\u003e). The type of method to be selected for flood monitoring depends on the region of study, the type of desired output, the accuracy requirements and use of results as well as the funding available(Appeaning Addo et al., \u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e2018\u003c/span\u003e). Often traditional techniques intertwined with advanced models and technologies bring out efficient flood monitoring where input data clustering and filtering, modelling of river terrain and showcasing morphometric variations through digital maps of relief and elevation, heterogenous data processing and hydrological modelling are used to observe flood incidents to bring preparedness to the hazard and its destructions(Merkuryeva et al., \u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e2015\u003c/span\u003e). Techniques involve a large array of methods utilising drone and sensor-based outputs. Image processing and geoinformation technology play a significant role in the visualisation of the data.\u003c/p\u003e\u003cdiv id=\"Sec5\" class=\"Section2\"\u003e\u003ch2\u003e4.1. Real-time Monitoring Systems:\u003c/h2\u003e\u003cp\u003eReal-time flood monitoring often gets hampered due to the lack of or unavailability of in situ streamflow data in many parts of the world(Revilla-Romero et al., \u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e2015\u003c/span\u003e). To fill this gap, satellite-based monitoring is used to calibrate and validate the data as well as assimilate it.\u003c/p\u003e\u003cp\u003eReal-time forecasting drives as a key resource while used in flash flood early warning systems (FFEWS). In the case of PFFs, real-time forecasting helps in assessing multiple criteria such as the flash flood susceptibility assessment (FFSA), rainfall comparison with surface conditions neglected (RC-SN), rainfall comparison with surface conditions included (RC-SC), flow comparisons in real-time (FC), quantitative and qualitative precipitation forecast and estimation respectively (QPF, QPE)(Zanchetta \u0026amp; Coulibaly, \u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e2020\u003c/span\u003e). While dealing with forecasting to assist in lead time support along with early warning to flood events, weather radar precipitation mosaic products are accessible, run-off threshold exceedance is best utilised to project the available data and applied in preparing flood forecast maps. In situations where radar data is insufficient the satellite-based precipitation data plays a huge role in assistance to real-time flood observation. Andre Zanchetta and Paulin Coulibaly conducted extensive research in the year 2020 where they highlighted the importance of global navigation systems (GNSSs) and global positioning systems (GPSs) in developing multi-constellation approaches to retrieve and enhance accurate real-time estimation of precipitation in an area. In order to properly orient data fetched in real-time, it is important to have them properly scaled as per requirements. Various extrapolation techniques such as kriging interpolation or inverse distance weighted (IDW) interpolations are involved in processing rainfall data to have a refined computer vision while observing flood conditions in an area. Dynamic downscaling of numerical weather predictions based on convection-permitting (CP-NWP) uses hydrodynamic processes as a baseline with a spatial resolution of 4 km or higher to present local convections and mesoscale event visualisation. Temporal extrapolation becomes vital in order to not only monitor but also project the data to understand future scenarios. In such cases, uncertainties in grid scale are common, where pixel-to-pixel analysis is used involving ensemble models to reduce the existence of spatial noise in the data being monitored.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec6\" class=\"Section2\"\u003e\u003ch2\u003e4.2. Sensor Networks- Internet of Things (IoT) based sensors and warning systems:\u003c/h2\u003e\u003cp\u003eIn the systematic review conducted by Arshad et al., various techniques using sensors to monitor floods are highlighted. The IoT-based sensor called the automatic pressure transducer assists in measuring up to 0.001 m accuracy in water levels and is compatible with multiple controllers for logging the data and visualise them in real-time(Arshad et al., \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). Here, using computer-based technology and IoT sensors helps in warning in case there is a water level rise resulting from flood hazard via, short message service (SMS), email, or web servers. Resource from Arshad et al. also state, the comparison between three techniques: (1) differencing image technique, (2) dictionary method and (3) convolution neural networks (CNN) in early warning of rising water levels and flood situation. The differencing image technique uses images every 10-minute intervals to check the previous image and current image state of water level pixels. It uses Gaussian smoothening to remove any noise. The dictionary method classifies the region of interest (ROI) into ruler and water classes to detect the changes in water level. The CNN technique however involves the best results of all these, where a filter is used as the algorithm trains from features extracted from the input images, thus reducing errors to a large extent(Arshad et al., \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e2019\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eWhile studying images the bandpass technique is useful to reduce noise in images captured using a charged couple device (CCD) camera, which is converted from a time series to a frequency domain using discrete cosine transform (DCT) for calculating change in water level(Park et al., \u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e2009\u003c/span\u003e). The histogram variations studied, on the other hand, also help in image analysis and condition judgments. An accumulated histogram can be prepared using images gathered from water level surveillance cameras to detect changes in water level by comparison of images collected at different points in time. Such detections are attempted in a conference proceeding by Park et al. through their elaborate studies in Seoul, Korea. Wireless sensors used in IoT coupled with ANNs have built substantial real-time monitoring systems of change detection in the inundation level of urban drainage networks.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec7\" class=\"Section2\"\u003e\u003ch2\u003e4.3. Drones and UAVs acting as a high-resolution monitoring tool:\u003c/h2\u003e\u003cp\u003eDatasets produced by drones have a high spatial resolution (\u0026lt;\u0026thinsp;10cm) and produce very high-resolution digital elevation models (DEMs). The use of drone technology in coastal monitoring is a common approach as far as coastal surveillance and management is concerned. It also provides timely, event-based rapid availability and superior quality outputs making it highly advantageous in immediate monitoring and preparedness. Specifically, a little after 2014 there has been a significant boom in innovations and advancements in drone technology, hardware and their availability(Iqbal et al., \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). A drone-assisted study across the Volta Delta in Ghana efficiently detected accretion, erosion and change in shoreline over 12 years. Drone image comparisons show vital changes in the shoreline and vulnerability of settlements located along the shoreline(Appeaning Addo et al., \u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e2018\u003c/span\u003e). The very high-resolution data provided by drone cameras makes it powerful in monitoring flood incidents as well as account the rate of population displacements in the affected area as a post-flood situation. This in turn helps the region to cope and convalesce from the extreme effects of the hazard immediately after its occurrence. However, drones are limited in strong wind conditions as they are vulnerable to distortions from environmental indices. Thus, they are subjected to limited use in times of gentle wind conditions. Various other associated challenges include platform instability, distortions in view angle, or erroneous data processing tools. One of the other major constraints of drone usage is their low battery lifespan restricting them to only short flight times.\u003c/p\u003e\u003cp\u003eThe utilisation of Unmanned Aerial Vehicles (UAVs) also has numerous perks of its edge as it has no requirement of installation points as in the case of remote sensing images, has a large spatial coverage and is comparatively less costly allowing frequent missions(Vousdoukas et al., \u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e2011\u003c/span\u003e). Studies conducted using Small Unmanned Aerial Vehicles (SUAVs) use a series of images collected after flying a camera-mounted device across the shoreline to detect the changes. The different time frames are processed in a semi-automated fashion where each frame is synthesised and masked to detect the change in shoreline elevation, water variations and displacements in a short time over a wide, large-scale area. Each frame of images is geo-rectified and mosaics are obtained to view the results with better precision. The shoreline elevation can also be collected from tidal and wave data using the Argus video monitoring systems(Holman \u0026amp; Stanley, \u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e2007\u003c/span\u003e). However, while conducting geo-rectification on each frame of images it is evident that errors may accumulate over several iterations, demanding a robust frame-to-frame geo-referencing(Vousdoukas et al., \u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e2011\u003c/span\u003e). Transforming the image pixel to world coordinates often poses one of the greatest challenges in generating accurate velocity estimation during flash flood monitoring even under controlled settings and parameters(Perks et al., \u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e2016\u003c/span\u003e). Various algorithms are drawn to address such errors and minimise them to view the free surface velocity of real-time flash flood monitoring.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec8\" class=\"Section2\"\u003e\u003ch2\u003e4.4. Community-based Monitoring and Crowdsourcing:\u003c/h2\u003e\u003cp\u003eCommunity-based disaster risk management in monitoring any disaster often helps in building early warning systems. In the case of community-based monitoring, this is often done by locals and villagers who know the region affected by the hazard very well. This in turn helps to empower communities through participation and build awareness to prepare for and confront hazards(Macherera \u0026amp; Chimbari, \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e2016\u003c/span\u003e). Research accounts conduction of various workshops in the Nordic countries by several organisations, namely HydroLogic and WeSenseIt to provide insight into \u0026lsquo;state-of-art tools\u0026rsquo; based on citizen observations(Henriksen et al., \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e2018\u003c/span\u003e). Under the same research, directives such as the Citizen Observatories for Water Management (COWM2016, Venice), European Water Framework Directive (2000/60/EC) and Flood Directive (2007/60/EC) exist to strengthen data based on community observations relating to flood hazards and its management. Systematic reviews on urban pluvial flooding (UPF) show extensive use of participatory science and crowdsourcing to understand floods in areas of vulnerability(Azizi et al., \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e2022\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eCrowdsourced data proposes to address uncertainties driven by localised topographic and land cover variations, data unavailability, validation and calibration errors by providing context-specific observations(Assump\u0026ccedil;\u0026atilde;o et al., \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e2018\u003c/span\u003e; R. Q. Wang et al., \u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e2018\u003c/span\u003e). Major crowdsourcing tools that are utilised in recent studies include public webcams, social media and citizen science(Helmrich et al., \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e2021\u003c/span\u003e) as they can provide large volumes of spatially relevant observational data points. In the case of urban flood monitoring, which mostly occurs from pluvial flooding due to various factors relating to physiographic as well as anthropogenic drivers, crowdsourcing measures become an effective tool in the observation of flood situations. Application-based monitoring is carried out where individuals located within the affected regions can mark themselves on a map generating direct real-time data. These data can be treated as secondary data and are most of the time available over large spaces and generative data can be narrowed or adjusted as per the needs of the study. Two such applications are STORM and HAZE used by Helmrich et.al. in Norfolk, Virginia flood hazard monitoring. Public webcams, even though limited to only daylight visualisation of floods, are most effective when located in high-risk flood-prone areas. Webcams using IR sensors can be used to carry out night-time observations with improved, effective results. Citizen science data available in various secondary resources often provides detailed information for each data provided. Social media accounts to quick, effective reporting via various media platforms\u0026rsquo; posts, Twitter posts giving out information and geo-tagged photographs in real-time which can be geo-coded to map flooding in an area promptly in order to bring measures and relief as per necessities.\u003c/p\u003e\u003c/div\u003e"},{"header":"5. Recent Methods for Predicting Flood Hazards","content":"\u003cp\u003eThe trends in flood prediction have moved from a deterministic to a probabilistic approach over the two decades of the literature study period. In the first half, for most parts, studies were more data-oriented giving one proper idea or nothing as we see in deterministic prediction models. An increasing number of physically based hydrological models are developed and used in flash flood prediction which are found to be giving much better, plausible results as compared to traditional approaches in forecasting the hazard(Hapuarachchi et al., \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e2011\u003c/span\u003e). With a probabilistic approach, flexibility is introduced which helps in definitively including the variability and changes associated with physiographical as well as meteorological factors of flood hazards. Prediction of flood hazards involves the exercise of various algorithms, mathematical computations and machine learning assistance involving multi-criteria analysis (MCA) to develop hydro-dynamic models and forecasting tools that give high-level accuracy of data. This in turn enhances visualisation and probabilistic assumptions of the occurrences of the hazard worldwide under similar parameters. Forecasting can be carried out as extended-range forecasting over 7\u0026ndash;14 days making predictions using advanced numerical weather prediction models, medium-range forecasting, long-range forecasting extending from a few weeks to a few months combining statistical analysis, historical weather data, climate indices, and global climate models to identify patterns and signals indicative of future weather conditions on longer timescales and nowcasting that utilises real-time observational data from weather radar, satellite imagery, ground stations, and other sources to track the movement and intensity of weather phenomena such as thunderstorms, rainfall, snowfall, and severe weather events.\u003c/p\u003e\u003cdiv id=\"Sec10\" class=\"Section2\"\u003e\u003ch2\u003e5.1. Statistical Methods:\u003c/h2\u003e\u003cp\u003eEffective discussion on flood forecasting methods (FFM) primarily depends on the basic understanding that often speaks about the utilisation of probabilistic methodologies which is closely intertwined with deterministic approaches of model building. Empirical models prepared based on statistical data often utilise historical data based on observed relationships between input variables and river flow changes to identify patterns within the variables and their effects on flooding and influx in streamflow(Granata \u0026amp; Di Nunno, \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2024\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eIn this context, the Bayesian forecasting system (BFS) is a theoretical model to forecast an extensively occurring hazard. It involves probabilistic forecasting from a deterministic hydrological model. BFS can be used for probabilistic river stage forecast (PRSF), probabilistic river discharge forecast (PRDF), or probabilistic runoff volume forecast (PRVF)(Han \u0026amp; Coulibaly, \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e2017\u003c/span\u003e). Here generally two sources are used: (1) Precipitation uncertainty in the future and (2) other hydrological parameters uncertainties. These two are individually calculated and then combined for probabilistic forecasting. Sasha Han along with Paulin Coulibaly, in their elaborate research on BFS utilisation, noted that the uncertainties observed in the overall data are considered in their research as occurrences and non-occurrences. For precipitation forecasting, they are employed. A modified Bayesian processor of output (BPO) is implemented in one study to process the quantitative precipitation forecast output from the numerical weather prediction (NWP) model. It offers an ideal theoretical structure and quantifies all sources of uncertainties, thus reducing predictive uncertainty to some extent and leading to a more reliable and accurate forecast. However, it is not implemented in large watersheds and under climatic variabilities to understand its efficiency beyond river basins. Refinements based on conditional random fields (CRFs) in a Bayesian statistical modelling network are used in pattern recognition and mapping of flood areas combining intensity and coherences(Schumann et al., \u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e2023\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eIn a case study of the Zhejiang province of China, researchers X. Wu. et. al. conducted a multi-criteria decision analysis (MCDA) along with the CA-Markov model to predict the future flood risk changes in the region. Analytic Hierarchy Process (AHP) was implemented to obtain a \u0026lsquo;pairwise comparison matrix\u0026rsquo; in the MCDA approach. These criteria were divided into \u0026lsquo;the hazard-causing factor\u0026rsquo; containing precipitation and flood hazard parameters for 24 hours and 72 hours respectively; \u0026lsquo;the hazard-gestating factor\u0026rsquo; containing altitude, slope, standard deviation of elevation, drainage density, and buffer zone of water parameters; and \u0026lsquo;the hazard-bearing factor\u0026rsquo; containing the population, GDP, and land cover parameters(X. Wu et al., \u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). Different interpolation methods to arrange and understand the various standards of data can be used based on the variogram results showing the spatial geometry of regionalised variables. Utilisation of the CA-Markov model in the study of Zhejiang province elaborately shows parameters such as the population, GDP and land cover of an area as indirect parameters of flood hazard studies.\u003c/p\u003e\u003cp\u003eIn another study conducted in north-central Italy, a probability error study is attempted to estimate the uncertainty of hydrological forecast in two watersheds. In this literature, the meta-Gaussian model in statistics takes into account the errors and uncertainties in the forecast within the same time frame, where the transformation of preliminary errors is done by Normal Quantile Transform (NQT). In this domain, it is assumed that \u0026ldquo;forecast errors can be explained as the linear combination of the selected explanatory variables.\u0026rdquo;(Montanari \u0026amp; Grossi, \u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e2008\u003c/span\u003e). Here it is shown how utilisation of probabilistic forecasting can be useful in prediction while computing the error to identify and account for the uncertainties.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec11\" class=\"Section2\"\u003e\u003ch2\u003e5.2. Hydrological Models:\u003c/h2\u003e\u003cp\u003eDetection of pre-flood and post-flood measures and identification of flood hazard zones are attempted to understand the vulnerable sites. GIS and data-driven statistical assessment helped in drafting as well as addressing the issues related to flood mapping systems as well as the warning systems associated with them (Kourgialas \u0026amp; Karatzas, \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2011\u003c/span\u003e). The physically based models comprise mathematical equations showing hydrodynamics principles and assist hydraulic modelling making them extremely efficient in comprehending hydrological processes and preparing models(Granata \u0026amp; Di Nunno, \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). However, these are very data-intensive processes requiring levels of validation and calibration in steps along the entire course of its execution.\u003c/p\u003e\u003cp\u003eThe Hazards US Multi-Hazard (HAZUS-MH) flood model was prepared to quantify the human, property, financial and social impacts of flood hazards and draw mitigation methods based on the results from the software. Kourgialas et. al., described hydrological and hydraulic modelling as the classical approach to delineating flood-prone regions with different levels of hazard effects. These models are also adapted while simulating run-off in small catchment areas(Wahba et al., \u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). River flood stage analysis is a complex dynamic process characterised by successive spatial and temporal variations(Le et al., \u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e2019\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eDifferent hydrological models are prepared based on their dimension of features considered. In case of one-dimensional models, only one feature such as the channel flow is considered to understand the flood extent setting up cross-sections, two-dimensional models use grid-like meshes to solve the fluid equations of more than one feature whereas the three-dimensional models uses complex equation to solve multi-dimensional fluid equations(DigitalCommons et al., \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). An extensive study on such models is prepared where five different models are speculated while observing the Waterford River watershed in Canada. These include the GR4J model, a simple conceptual model using a minimum of four to up to nine parameters to model flooding conditions in a region; the MAC-HBV model which is a lumped conceptual model using combined modelling methods where enhanced modelling is carried out considering fifteen parameters such as flow accumulation, soil moisture and non-linear responses; SAC-SMA model is a conceptual watershed model used commonly by the National Weather Services, USA, in operational flood forecasting that uses nineteen input parameters; HEC-HMS model is prepared by the Hydrologic Engineering Centre of the US Army Corps of Engineers to devise simulation hydrologic processes in dendritic watersheds over both space and time; and lastly, WATFLOOD model which is a physically-based distributed hydrological model that subdivides watersheds into Group Response Units (GRU) based on the assumption of surface areas with similar land use and hydrologic processes(Wijayarathne \u0026amp; Coulibaly, \u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e2020\u003c/span\u003e). Each of these models is utilised to produce different scales of simulations based on the parameters used. Often these parameters help in showing the variations and changes in water levels within a watershed and devise in forecasting the probability of flood hazards. Operational flow forecasting is used to understand the model that is the best fit for predicting the streamflow within a watershed both spatially and temporally.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec12\" class=\"Section2\"\u003e\u003ch2\u003e5.3. Remote Sensing Techniques alongside GIS tools and environments:\u003c/h2\u003e\u003cp\u003eRemote Sensing techniques are especially beneficial since in situ measurements are often cumbersome to attain due to the inability to obtain point-based data as a region might become inaccessible during flood incidence, the period of high water flow during floods typically does not allow data collection, moreover, problems related to high operational and maintenance costs in rendering data is associated. Flood prediction utilising earth observation data is effective in this sense using satellite data and remote sensing observation to check the flood damage footprints by overlaying elements-at-risk and the hazard affected points then proceed to carry out statistical analysis and mapping(Schumann et al., \u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e2023\u003c/span\u003e; Van Westen, \u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e2013\u003c/span\u003e). Employing earth observatory (EO) instruments, there are multiple befitting options such as optical imageries- using MODIS or NOAA VIIRS satellite data to extract flood pixels for making flood binary maps which can be used as an inventory in flood analysis; SAR data can provide radar imageries operable under all-weather conditions extending day-night imaging capabilities involving statistical model, CNNs, thresholding and other techniques using Sentinel-1, Envisat or InSAR coherence; CNNs gives edge in interpretation as well as detection; InSAR helps in enhancement of ground observation utilising the radar data; Lidar DEMs are available at 50cm resolution in altimetric studies with very limited availability; radar altimeters are used at different levels with varying resolutions and gestation periods(Schumann et al., \u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). A list of commonly used satellites in remote sensing are shown in Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e which assists in flood mapping at present. However, the selection of the satellite data becomes vital to carry out the study across an area that best fits the desired results. This hence requires expert opinions and understanding as well as the availability of data under the time frame of the study.\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eA list of satellites along with their sensors that are commonly used in the practice for flood detection and forecast\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"4\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eSatellites\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eSensors\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003eSpatial Resolutions\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003eOutput/ Uses under Desired Parameters\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eNASA\u0026rsquo;s Terra or Aqua\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eMODIS\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e250 meters (bands 1 and 2)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003eProvides real-time flooded and non-flooded pixel values to prepare FHM.\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eNOAA\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eVIIRS\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e375\u0026ndash;750 meters at nadir\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003eCollects visible and infrared images flood pixels from Earth\u0026rsquo;s surface.\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCopernicus\u003c/p\u003e\u003cp\u003eSentinel-1 and Sentinel-2\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eSAR, InSAR, Multispectral\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e30 meters (common); up to 50 cm (high resolution); 10\u0026ndash;60 meters (MSS in Sentinel-2)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003eCaptures radar and multispectral data used in flood detection under all weather conditions, allowing day and night imaging.\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eEnvisat\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eASAR, InSAR\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e30 meters\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003eDetects hazards and floods in real-time, assisting with rapid mapping under any weather conditions.\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eINSAT-3D\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eMulti-spectral Imager\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e1 km x 1 km (visible/SWIR); 4 km x 4 km (MIR/TIR); 8 km x 8 km (WV)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003eUses optical radiometry to generate images across six wavelength bands to detect flooded pixels.\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eSatellite altimeters\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eLidar DEMs\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e50 cm resolution on the ground\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003eUtilises altimetric data for flood detection, often paired with radar altimeters for more accurate results.\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eRadarsat-2\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eC-band SAR\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e10 meters (high resolution), up to 100 meters (low resolution)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003eProvides all-weather radar data for flood mapping, particularly in coastal and inland flood areas.\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eTerraSAR-X\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eX-band SAR\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e1 meter (high resolution), 3 meters (low resolution)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003eProvides high-resolution radar images, effective for detecting flood boundaries, even in adverse weather.\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCopernicus Sentinel-3\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eOLCI, SLSTR\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e300 meters (OLCI); 1 km (SLSTR)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003eOffers multispectral imagery for flood monitoring, especially surface water mapping with optical data.\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCartosat-2\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eHigh-Resolution Panchromatic\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0.8 meters (panchromatic); 2 meters (multispectral)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003eProvides very high-resolution images for flood mapping and monitoring urban and rural flood-prone areas.\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003eThe global flood detection system (GFDS) estimates changing water levels based on signal information received from satellite imageries, specifically passive microwave sensors. This is assisted by MODIS flood maps prepared by MODIS Near Real-Time Global Flood Mapping Project. Beyond detection, the GloFAS flood forecasting is carried out where land use changes, hydrological changes and meteorological data are obtained from the Ensemble Prediction System (ENS), European Centre for Medium-range Weather Forecasting (ECMWF). Once data is gathered, the magnitude of flood is compared for all the days of reported floods showing that in places where the in situ data was unavailable, and low magnitude floods, remotely sensed GFDS maps gave better results than the GloFAS forecasts(Revilla-Romero et al., \u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e2015\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eThe Earth Observations (EO) and Geographical Information Systems (GIS) are integrated well-developed tools in disaster risk management where spatial methodologies can be fully explored throughout the risk assessment processes(Van Westen, \u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e2013\u003c/span\u003e). C. Van Westen, in his extensive works from 2009 followed by 2013 on \u0026ldquo;Natural Hazards Assessment and Disaster Risk Management\u0026rdquo;, clearly stated that in studies with large ROI, data is often unavailable or unreliable, remotely sensed data along with GIS tools based on numerous other factors such as elevation, precipitation, landcover and illumination while sensors are working helps in assistance and guided studies.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec13\" class=\"Section2\"\u003e\u003ch2\u003e5.4. Machine Learning:\u003c/h2\u003e\u003cp\u003eMachine learning involves the identification of factors triggering flood incidents and training a computer-driven algorithm through training and testing sets of data to become self-reliant in predicting the probability of the hazard reoccurring in the future. These models are data-driven and can handle complex datasets by capturing non-linear relationships adapting to changing conditions(Granata \u0026amp; Di Nunno, \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). This helps in adaptability as a precaution to the disastrous effects of the hazard. The algorithm goes through iterations on the distributed training and testing sets to understand patterns, identify and recognise the same by correlating inputs to achieve desired outputs. Advanced machine learning helps in data viewing and enhances the automation of data visualisation as well as analysis. It assists in data augmentation to interpret predictions(Diaconu et al., \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e2024\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eWhile discussing the inclusion of machine learning (ML) algorithms in FSM, data becomes a very vital tool. Thus, extraction, interpolation, utilisation and validation become essential. In almost every study conducted with ML assistance environmental and morphometric factors are utilised. This is mainly to consider multiple criteria including all the flood triggering factors. These include digital elevation model (DEM) data involving slope, elevation, curvature and slope aspect of the basin or catchment being studied. Other environmental data involve precipitation, temperature, topographic wetness index (TWI), flow accumulation, stream-transport index (STI), land use and land cover, and soil porosity data. The factors in inclusion in any study can be as many and as varied as per the requirements and motive of flood maps being prepared. Physiographical data such as the change in population, GDP and economic activities in a region also act as indirect factors inducing floods in any region. The inter-correlation between all these multiple factors is identified as a multi-collinearity problem. The multi-collinearity test is carried out in this context to reduce the bias and collinearity problem(S. Ghosh et al., \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). This is done by calculating tolerance and variance inflation factor (VIF). VIF values better than 5 (significant multicollinearity) and tolerance less than 0.1 directly indicate multi-collinearity problems within the dataset. Based on the values (above the set threshold of acceptance) obtained changes are made in criteria specification and adjustments can be made to achieve the least distortions in outputs, while working with training and test sets.\u003c/p\u003e\u003cp\u003eFlood inventory maps are made describing flooded and non-flooded points to be used in the stages of map-making. This is often done using Google Earth Engine (GEE) to find and demarcate \u0026lsquo;presence\u0026rsquo; and \u0026lsquo;absence\u0026rsquo; points of water in a flooded region (Bashirgonbad et al., \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). The comprehensive sampling based on validation and region-specific segregation of flooded from non-flooded regions is essential in the enhancement of flood maps prepared to delineate the regions prone to risks of flood hazard(Wahba et al., \u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). Bashirgonbad et al. in their work also mentioned that based on the VIF values from the multi-collinearity test, the factors most to least responsible in influencing flood in the region can also be assessed. The delineation of basin or catchment boundaries also becomes crucial before considering factors affecting the flood hazard as it will determine the input data which are ultimately responsible for output generation.\u003c/p\u003e\u003cp\u003eResearchers have worked on focused studies relating to different models used in flood forecasting. The preparatory data in these studies comprise of the data maps prepared from the DEM data, environmental factors along with the flooded and non-flooded datasets to train the machine learning models such as the support vector regression (SVR) model to not only predict but also build pattern recognition of the hazard; least absolute shrinkage and selection operator (LASSO) linear regression model that serves as variable selection method assisting in reduction of the number of variables used in each tier of the model(Wahba et al., \u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). The support vector machine (SVM), a kernel-based technique that falls under supervised machine learning algorithms calculates both associations among the input variables and projects a linear structure by combining the output from training samples(Y. Wang et al., \u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e2020\u003c/span\u003e); random forest (RF) on the other hand, is based on ensemble ML algorithm where it creates multiple classification trees and then average the output to predict the data(Bashirgonbad et al., \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e2024\u003c/span\u003e; S. Ghosh et al., \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). Forecasts based on daily data are carried out using daily precipitation and flow rate data as inputs in artificial neural network (ANN) and recurrent neural network (RNN) models for a flood-prone area. The long short-term memory (LSTM) model is used for one to three-day flow rate forecasting exhibiting outstanding advantages in its ability to learn short dependencies effectively, and this model can completely be applied to forecast the flow two days or even three days ahead with accuracy of over 86%(Le et al., \u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). For terrain or hydrological variations consideration in the forecasting model, LSTM-NN turns out to be an extremely effective data-driven method to be utilised. However, limitations exist as physical differences are not extensively considered and thus, limits the model's effective parameterisation. It successfully gives highly accurate predictions at specific stations.\u003c/p\u003e\u003cp\u003eAssessment of the accuracy of the models is carried out by various statistical validation methods. Area under receiver operating characteristic curve (AUC-ROC), residual analysis using R-squared error calculation, mean squared error (MSE) and mean absolute error (MAE) calculations to validate the FSM map prepared(Wahba et al., \u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). Other techniques involve the Kling Gupta efficiency (KGE), mean absolute percentage error (MAPE)(Granata \u0026amp; Di Nunno, \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2024\u003c/span\u003e); Friedman test and Wilcoxon Signed Rank test(S. Ghosh et al., \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e2022\u003c/span\u003e) to validate the outputs and identify the model most fitting in forecasting of flood in the region of study. On an ensemble model based on multiple meta-learners, KGE renders more precise error detection than r-squared error. The accuracy also considers various statistical indicators such as sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV).\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec14\" class=\"Section2\"\u003e\u003ch2\u003e5.5. \u003cb\u003eIntegrated Approaches: Hybrid or Ensemble Models\u003c/b\u003e\u003c/h2\u003e\u003cp\u003eIn the recent decade of the articles studied it is noted that there is an accent of ML algorithms being widely used in the forecast of a hazard. A few of the more recent studies also show the coupled implementation of efforts put toward the development of hybrid or ensemble models. They provide advantages in terms of levels of accuracy, model stability and reduction of over or under-fitting of models. Stacked models made in two or more layers can be effective ensemble models to study and forecast short to medium-period data(Granata \u0026amp; Di Nunno, \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). Different basins and their varying stream or water flow rates can be extensively studied. Models, integrated ensembles, or simple regressions, are a simplified depiction of the reality where choosing the input factors becomes extremely crucial and sensitive to the output.\u003c/p\u003e\u003cp\u003eFrancesco Granata and Fabio Di Nunno(Granata \u0026amp; Di Nunno, \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2024\u003c/span\u003e) in their ensemble study of short-term and long-term streamflow used a multilayer perceptron algorithm combined with the random forest algorithm to build the MLP-RF, a stacked forecasting deep learning model, utilised to forecast flood data. Comparison is drawn with different meta-learners such as isotonic regression (IR), pace regression (PR), and radial basis function (RBF-NN) neural networks. Each of these regressors works and gives variable results over different frames of time ranging from one day to up to 15 days forecast or short to medium range forecasting. Out of these, RBF-NN is seen to be returning effective results even with less data in the training set. Comparison analysis depicts that the predictions became more delicate and precise toward the longer period of a 15-day forecast of streamflow in a basin. This comparison is also drawn against Elastic Net (EN), which uses meta-learners between two different stacked models.\u003c/p\u003e\u003cp\u003eSome other literature showed using artificial neural network multi-layer perceptron (ANN-MLP) regressor model is used to predict the chances of reoccurrence of flood in an already flooded area; gradient boosting regressor (GBR) which simply helps in boosting other weak models specifically decision tree model by combining two models to increase its efficiency(Wahba et al., \u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e2024\u003c/span\u003e); the species distribution modelling (SDM) is used to combine strengths of SVM and RF models in a weighted manner to have the best FSM mapping outcome with enriched prediction abilities.(Bashirgonbad et al., \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e2024\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eThe stacking of models is generally done on multiple levels. Level 0 models consist of multiple models that are trained on the same data utilising different algorithms and hyperparameters. Level 1 also called \u0026lsquo;meta-level\u0026rsquo; uses outputs from level 0 as inputs in this level to train the model to predict the occurrences of the phenomenon in question. Studies like that conducted by Diaconu et al. recognise dependent factors to understand the condition of flood proneness in the study area. It utilises four ML techniques- Deep Learning Neural Network- Statistical Index (DLNN-SI), Particle Swarm Optimization-Deep Learning Neural Network- Statistical Index (PSO-DLNN-SI), Support Vector Machine-Statistical Index (SVM-SI), and Particle Swarm Optimization-Support Vector Machine-Statistical Index (PSO-SVM-SI)(Diaconu et al., \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). The Flood Susceptibility Index (FSI) calculated for each of the models helps in understanding the percentage of whether the areas in the region are vulnerable or not to flooding. The Gain Ratio method is utilised where slope has the highest predictive ability whereas rainfall, hydrological soil index and convergence index have low predictive ability. Overall, the PSO ensembled with DLNN and SVM improved the efficiency and PSO-DLNN-SI outperformed all the other models.\u003c/p\u003e\u003cp\u003eAnother paper by W. Wu et al., prospects into understanding the improvement of technology and inclusion of ML in predicting floods and understanding better data assimilation and the extent of flood inundation. The study attempts to extend the forecast variable from simple river flow dynamics to inundation variability and water flow changes as an important consideration aspect in ensemble forecasting, targeting longer prediction periods, seasonal variations and climatic constraints(W. Wu et al., \u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e2020\u003c/span\u003e). Seamless prediction is preferred combining short-term and long-term forecasts. Thus, the ability of ensemble studies elaborately shows the enormous capability of hybrid ML algorithms in prediction and forecast technology.\u003c/p\u003e\u003c/div\u003e"},{"header":"6. Discussion","content":"\u003cp\u003eMonitoring and forecasting flood data is an extensive and prolonged study of all the factors that influence the hazard. This aims ultimately at understanding the \u0026lsquo;factors-at-risk\u0026rsquo; and routing towards their mitigation. Thus, a careful evaluation of each technique is attempted in this review article. In order to determine flood vulnerabilities identification of sources of danger is highly necessary. Knowledge of these sources is gathered from different techniques of monitoring and observations. On comparing the various techniques used in flood monitoring it is seen that the role of sensors and communities is primary during the initiation period of flood. GIS tools and software become incredibly essential to map and build simulations in real-time to look at the affected region. In places of inaccessibility, remote sensing and satellite imageries give high spatial resolution images to carry out necessary investigations. ML algorithms and neural networking are of the highest efficiency at present in building elaborate and active models as per requirements. Utilisation of multiple criteria to include all the regional and local factors triggering floods is essential in this regard for both enhancements as well as the accuracy of the model. It can foresee the occurrence of the hazard reducing uncertainties while combating the effects of flood in an area. In these terms, integrated machine learning algorithms are noted to be most effective in building powerful models supporting the prediction of streamflow, river flow and other factors affecting flood hazards. The use of ensemble models needs careful selection and calibration as there are chances of increased complexity in computations and prone to multi-layered interpretations. However, when taken well care of, it poses out to be one of the best methods in the prediction of flood situations giving a robust, efficient and highly accurate level of prediction in both short-term and long-term forecasting. The development and implementation of any model effectively also depend on its ability to reproduce results over similar scenarios using variables and factors as per requirements of the basin or channel being studied for the hazard assessment.\u003c/p\u003e\u003cp\u003eThe assessment of multi-hazards and their risk monitoring is an extremely data-intensive procedure where the availability of certain types of spatial data is unavailable(Van Westen, \u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e2013\u003c/span\u003e). This is more prevalent in developing nations, where hazard inventories are scarce and underdeveloped, often unfit for use in analysis. The uncertainties related to flood forecasting can be tackled using either statistical or non-statistical approaches. Due to the inherent heterogeneous nature of physical processes related to hydrological or river basins, non-statistical approaches are preferred over statistical approaches(Montanari \u0026amp; Grossi, \u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e2008\u003c/span\u003e). The problems related to missing data are also correlated effect of insufficient maintenance and the inability to acquire consistent data collection in regular intervals. In the context of developing nations, the availability of stations in the basin being studied is so scarce that dependency on data in carrying out a study is cumbersome and less encouraged. Large-scale research and development are being carried out to enhance this very problem and assistance from the government through directories and district-wise data handbooks is under progress. This paper also showed a collective work on satellite-driven data extraction for data-sparse regions(Revilla-Romero et al., \u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e2015\u003c/span\u003e), showing how each has its advantages and disadvantages. In the case of data derived from news reports, it is often seen to be insufficient or biased. Remote sensing data is rather insufficient for narrow mountain valleys facing flash floods whereas cloud cover is a major problem in viewing flood-affected regions using satellite imageries. Satellite missions and remote sensing techniques are being developed for directly measuring water depths and flow monitoring in basins and channels.\u003c/p\u003e\u003cp\u003eCurrently, significant attention is being directed towards flood monitoring as it affects and has a widespread impact on the global population. Enhanced understanding and rigorous research in this area has become imperative. Advances in satellite missions and the increasing acceptance of technology in managing flood hazards offer hope to communities around the world. In particular, the integration of artificial intelligence (AI) and machine learning (ML) algorithms is opening new avenues for forecasting flood risks, helping to reduce the uncertainties surrounding this natural hazard. The development of ensemble models and the refinement of algorithms to train predictive software are key areas of focus, shaping a future that is better equipped to both address and mitigate the impacts of floods.\u003c/p\u003e"},{"header":"7. Conclusion","content":"\u003cp\u003eOne of the major challenges related to flood hazards is the inherent uncertainty associated with them. Monitoring of the hazard plays a crucial role in addressing this issue. In addition to monitoring, predictive analysis and forecasting are essential tools to assist in future preparedness and adaptability. This review aims to provide a comprehensive understanding of flood hazards, focusing on their occurrence and the application of technologies and advancements to address the associated challenges. Among the various methodologies available, the choice of an appropriate model depends on factors such as the size of a region, its physiographic conditions, and, to some extent, the anthropogenic activities contributing to changes associated with the region. The key findings of this paper are summarised as follows:\u003c/p\u003e\u003cp\u003e\u003cul\u003e\u003cli\u003e\u003cp\u003eThe paper addresses the primary objective of synthesising existing knowledge and collecting techniques and methods to assess flood extents and their impacts.\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003eIn discussing various methodologies, the paper acknowledges the challenges associated with each method individually and in comparison. It also evaluates data acquisition, assimilation, and availability through an extensive review process.\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003eContemporary methods are thoroughly examined to facilitate their multi-faceted application, allowing for a broader understanding of floods not just as a temporal event, but also as a spatial phenomenon.\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003eThe use of GIS and advanced machine learning techniques for flood prediction and forecasting is emphasised. The review highlights that ensemble and integrated machine learning algorithms are most effective in this context.\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003eBest practices and optimal methods are identified, offering practical insights for implementation through multidisciplinary approaches.\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003eThe paper also stresses the importance of involving experts from diverse disciplines, making it accessible for them to grasp the requirements related to flood hazard monitoring and prediction, encouraging their contributions.\u003c/p\u003e\u003c/li\u003e\u003c/ul\u003e\u003c/p\u003e\u003cp\u003eOverall, this review article provides a composite analysis of methods that can be directly applied in real-world scenarios, benefiting communities and fostering advancements in the field. It stresses the application of the techniques reviewed to ultimately contribute to their implementation to bring changes. It encourages researchers, government agencies, and organizations to leverage the findings to explore new directions and approaches in flood hazard studies.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eAcknowledgments\u0026nbsp;\u003c/strong\u003eThe authors acknowledge the support and motivation from the Department of Geography, Banaras Hindu University, Varanasi towards the completion of this paper.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthors contributions\u003c/strong\u003e All authors contributed to the conceptualisation and design of the study. Literature studying and manual selection of articles for review purposes were carried out by the authors, Adrita Talapatra and Narendra Kumar Rana. The first draft of the manuscript was written by Adrita Talapatra and was redrafted with comments and considerations from Narendra Kumar Rana. The final manuscript was read and approved by the authors.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFundings\u003c/strong\u003e No funding was received for this work.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData availability\u0026nbsp;\u003c/strong\u003eAll data and material are published and available on request from author ([email protected])\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthics approval and consent to participate\u0026nbsp;\u003c/strong\u003eNot applicable.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent to publish\u0026nbsp;\u003c/strong\u003eBoth author and co-author consented to the publication of the manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interests\u003c/strong\u003e There is no conflict of interest among the authors.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eClinical trial number\u0026nbsp;\u003c/strong\u003eNot applicable.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eAppeaning Addo, K., Jayson-Quashigah, P. N., Codjoe, S. N. A., \u0026amp; Martey, F. (2018). Drone as a tool for coastal flood monitoring in the Volta Delta, Ghana. \u003cem\u003eGeoenvironmental Disasters\u003c/em\u003e, \u003cem\u003e5\u003c/em\u003e(1). https://doi.org/10.1186/s40677-018-0108-2\u003c/li\u003e\n\u003cli\u003eAria, M., \u0026amp; Cuccurullo, C. (2017). bibliometrix : An R-tool for comprehensive science mapping analysis. \u003cem\u003eJournal of Informetrics\u003c/em\u003e, \u003cem\u003e11\u003c/em\u003e(4), 959\u0026ndash;975. https://doi.org/10.1016/j.joi.2017.08.007\u003c/li\u003e\n\u003cli\u003eArshad, B., Ogie, R., Barthelemy, J., Pradhan, B., Verstaevel, N., \u0026amp; Perez, P. (2019). Computer vision and iot-based sensors in flood monitoring and mapping: A systematic review. \u003cem\u003eSensors (Switzerland)\u003c/em\u003e, \u003cem\u003e19\u003c/em\u003e(22), 1\u0026ndash;28. https://doi.org/10.3390/s19225012\u003c/li\u003e\n\u003cli\u003eAssump\u0026ccedil;\u0026atilde;o, T. H., Popescu, I., Jonoski, A., \u0026amp; Solomatine, D. P. (2018). Just a me-too: No advantage in anaemia of chronic renal failure. \u003cem\u003ePrescrire International\u003c/em\u003e, \u003cem\u003e12\u003c/em\u003e(63), 14\u0026ndash;16.\u003c/li\u003e\n\u003cli\u003eAzizi, K., Diko, S. K., Saija, L., Zamani, M. G., \u0026amp; Meier, C. I. (2022). Integrated community-based approaches to urban pluvial flooding research, trends and future directions: A review. \u003cem\u003eUrban Climate\u003c/em\u003e, \u003cem\u003e44\u003c/em\u003e(July), 101237. https://doi.org/10.1016/j.uclim.2022.101237\u003c/li\u003e\n\u003cli\u003eBandyopadhyay, S., Ghosh, P. K., Jana, N. C., \u0026amp; Sinha, S. (2016). Probability of flooding and vulnerability assessment in the Ajay River, Eastern India: implications for mitigation. \u003cem\u003eEnvironmental Earth Sciences\u003c/em\u003e, \u003cem\u003e75\u003c/em\u003e(7). https://doi.org/10.1007/s12665-016-5297-y\u003c/li\u003e\n\u003cli\u003eBashirgonbad, M., Farokhzadeh, B., \u0026amp; Gholami, V. (2024). Enhancing flood mapping through ensemble machine learning in the Gamasyab watershed, Western Iran. \u003cem\u003eEnvironmental Science and Pollution Research\u003c/em\u003e, \u003cem\u003eMarch 2019\u003c/em\u003e. https://doi.org/10.1007/s11356-024-34501-5\u003c/li\u003e\n\u003cli\u003eBera, B., Bhattacharjee, S., Sengupta, N., Chamling, M., Ghosh, S., \u0026amp; Ghosh, A. (2024). Impacts of channelization of River Bala, eastern Himalayan foothills, India. \u003cem\u003eGeology, Ecology, and Landscapes\u003c/em\u003e, \u003cem\u003e8\u003c/em\u003e(3), 241\u0026ndash;253. https://doi.org/10.1080/24749508.2022.2130857\u003c/li\u003e\n\u003cli\u003eDas, B. K. (2012). \u003cem\u003eLosing Biodiversity , Impoverishing Forest Villagers : Analysing forest policies in the context of Flood Disaster in a National Park of Sub Himalayan Bengal , India Losing Biodiversity , Impoverishing Forest Villagers : Analysing forest policies in the co\u003c/em\u003e. \u003cem\u003e91\u003c/em\u003e(July).\u003c/li\u003e\n\u003cli\u003eDiaconu, D. C., Costache, R., Towfiqul Islam, A. R. M., Pandey, M., Pal, S. C., Mishra, A. P., \u0026amp; Pande, C. B. (2024). Developing flood mapping procedure through optimized machine learning techniques. Case study: Prahova river basin, Romania. \u003cem\u003eJournal of Hydrology: Regional Studies\u003c/em\u003e, \u003cem\u003e54\u003c/em\u003e(June), 101892. https://doi.org/10.1016/j.ejrh.2024.101892\u003c/li\u003e\n\u003cli\u003eDigitalCommons, U., All Graduate Theses, U., \u0026amp; Kesler, T. (2023). \u003cem\u003eComparing 1D, 2D, and 3D Hydraulic Models in Urban Flooding Comparing 1D, 2D, and 3D Hydraulic Models in Urban Flooding Applications Applications\u003c/em\u003e. https://digitalcommons.usu.edu/etd/8691\u003c/li\u003e\n\u003cli\u003eGharbi, M., Soualmia, A., Dartus, D., \u0026amp; Masbernat, L. (2016). Floods effects on rivers morphological changes application to the Medjerda River in Tunisia. \u003cem\u003eJournal of Hydrology and Hydromechanics\u003c/em\u003e, \u003cem\u003e64\u003c/em\u003e(1), 56\u0026ndash;66. https://doi.org/10.1515/johh-2016-0004\u003c/li\u003e\n\u003cli\u003eGhosh, D., \u0026amp; Saha, S. (2014). Channel bed aggradation in relation to channel Morphometry : A case study of river Jainti , Jalpaiguri , West Bengal. \u003cem\u003eINTERNATIONAL JOURNAL OF GEOMATICS AND GEOSCIENCES Volume 5, No 1, 2014 Research\u003c/em\u003e, \u003cem\u003e5\u003c/em\u003e(1), 192\u0026ndash;208.\u003c/li\u003e\n\u003cli\u003eGhosh, S., Saha, S., \u0026amp; Bera, B. (2022). Flood susceptibility zonation using advanced ensemble machine learning models within Himalayan foreland basin. \u003cem\u003eNatural Hazards Research\u003c/em\u003e, \u003cem\u003eSeptember\u003c/em\u003e. https://doi.org/10.1016/j.nhres.2022.06.003\u003c/li\u003e\n\u003cli\u003eGranata, F., \u0026amp; Di Nunno, F. (2024). Forecasting short- and medium-term streamflow using stacked ensemble models and different meta-learners. \u003cem\u003eStochastic Environmental Research and Risk Assessment\u003c/em\u003e, \u003cem\u003e20\u003c/em\u003e(12), 2175\u0026ndash;2177. https://doi.org/10.1007/s00477-024-02760-w\u003c/li\u003e\n\u003cli\u003eHan, S., \u0026amp; Coulibaly, P. (2017). Bayesian flood forecasting methods: A review. \u003cem\u003eJournal of Hydrology\u003c/em\u003e, \u003cem\u003e551\u003c/em\u003e, 340\u0026ndash;351. https://doi.org/10.1016/j.jhydrol.2017.06.004\u003c/li\u003e\n\u003cli\u003eHapuarachchi, H. A. P., Wang, Q. J., \u0026amp; Pagano, T. C. (2011). A review of advances in flash flood forecasting. \u003cem\u003eHydrological Processes\u003c/em\u003e, \u003cem\u003e25\u003c/em\u003e(18), 2771\u0026ndash;2784. https://doi.org/10.1002/hyp.8040\u003c/li\u003e\n\u003cli\u003eHelmrich, A. M., Ruddell, B. L., Bessem, K., Chester, M. V., Chohan, N., Doerry, E., Eppinger, J., Garcia, M., Goodall, J. L., Lowry, C., \u0026amp; Zahura, F. T. (2021). Opportunities for crowdsourcing in urban flood monitoring. \u003cem\u003eEnvironmental Modelling and Software\u003c/em\u003e, \u003cem\u003e143\u003c/em\u003e, 105124. https://doi.org/10.1016/j.envsoft.2021.105124\u003c/li\u003e\n\u003cli\u003eHenriksen, H. J., Roberts, M. J., van der Keur, P., Harjanne, A., Egilson, D., \u0026amp; Alfonso, L. (2018). Participatory early warning and monitoring systems: A Nordic framework for web-based flood risk management. \u003cem\u003eInternational Journal of Disaster Risk Reduction\u003c/em\u003e, \u003cem\u003e31\u003c/em\u003e, 1295\u0026ndash;1306. https://doi.org/10.1016/j.ijdrr.2018.01.038\u003c/li\u003e\n\u003cli\u003eHolman, R. A., \u0026amp; Stanley, J. (2007). The history and technical capabilities of Argus. \u003cem\u003eCoastal Engineering\u003c/em\u003e, \u003cem\u003e54\u003c/em\u003e(6\u0026ndash;7), 477\u0026ndash;491. https://doi.org/10.1016/j.coastaleng.2007.01.003\u003c/li\u003e\n\u003cli\u003eIqbal, U., Riaz, M. Z. Bin, Zhao, J., Barthelemy, J., \u0026amp; Perez, P. (2023). Drones for Flood Monitoring, Mapping and Detection: A Bibliometric Review. \u003cem\u003eDrones\u003c/em\u003e, \u003cem\u003e7\u003c/em\u003e(1), 1\u0026ndash;29. https://doi.org/10.3390/drones7010032\u003c/li\u003e\n\u003cli\u003eKourgialas, N. N., \u0026amp; Karatzas, G. P. (2011). Gestion des inondations et m\u0026eacute;thode de mod\u0026eacute;lisation sous SIG pour \u0026eacute;valuer les zones d\u0026rsquo;al\u0026eacute;a inondation-une \u0026eacute;tude de cas. \u003cem\u003eHydrological Sciences Journal\u003c/em\u003e, \u003cem\u003e56\u003c/em\u003e(2), 212\u0026ndash;225. https://doi.org/10.1080/02626667.2011.555836\u003c/li\u003e\n\u003cli\u003eLe, X. H., Ho, H. V., Lee, G., \u0026amp; Jung, S. (2019). Application of Long Short-Term Memory (LSTM) neural network for flood forecasting. \u003cem\u003eWater (Switzerland)\u003c/em\u003e, \u003cem\u003e11\u003c/em\u003e(7). https://doi.org/10.3390/w11071387\u003c/li\u003e\n\u003cli\u003eMa, C., Qiu, D., Mu, X., \u0026amp; Gao, P. (2022). Morphological Evolution Characteristics of River Cross-Sections in the Lower Weihe River and Their Response to Streamflow and Sediment Changes. \u003cem\u003eWater (Switzerland)\u003c/em\u003e, \u003cem\u003e14\u003c/em\u003e(21). https://doi.org/10.3390/w14213419\u003c/li\u003e\n\u003cli\u003eMacherera, M., \u0026amp; Chimbari, M. (2016). A review of studies on community based early warning systems. \u003cem\u003eJournal of Disaster Risk Studies\u003c/em\u003e, 1\u0026ndash;11.\u003c/li\u003e\n\u003cli\u003eMerkuryeva, G., Merkuryev, Y., Sokolov, B. V., Potryasaev, S., Zelentsov, V. A., \u0026amp; Lektauers, A. (2015). Advanced river flood monitoring, modelling and forecasting. \u003cem\u003eJournal of Computational Science\u003c/em\u003e, \u003cem\u003e10\u003c/em\u003e, 77\u0026ndash;85. https://doi.org/10.1016/j.jocs.2014.10.004\u003c/li\u003e\n\u003cli\u003eMontanari, A., \u0026amp; Grossi, G. (2008). Estimating the uncertainty of hydrological forecasts: A statistical approach. \u003cem\u003eWater Resources Research\u003c/em\u003e, \u003cem\u003e44\u003c/em\u003e(12), 1\u0026ndash;9. https://doi.org/10.1029/2008WR006897\u003c/li\u003e\n\u003cli\u003eMudashiru, R. B., Sabtu, N., Abustan, I., \u0026amp; Balogun, W. (2021). Flood hazard mapping methods: A review. \u003cem\u003eJournal of Hydrology\u003c/em\u003e, \u003cem\u003e603\u003c/em\u003e(March). https://doi.org/10.1016/j.jhydrol.2021.126846\u003c/li\u003e\n\u003cli\u003ePark, S., Lee, N., Han, Y., \u0026amp; Hahn, H. (2009). The Water Level Detection Algorithm using Accumulated Histogram with Band Pass Filter. \u003cem\u003eInternational Conference on Signal Processing, Communications and Networking 2009\u003c/em\u003e.\u003c/li\u003e\n\u003cli\u003ePaul, A., \u0026amp; Biswas, M. (2019). Changes in river bed terrain and its impact on flood propagation\u0026ndash;a case study of River Jayanti, West Bengal, India. \u003cem\u003eGeomatics, Natural Hazards and Risk\u003c/em\u003e, \u003cem\u003e10\u003c/em\u003e(1), 1928\u0026ndash;1947. https://doi.org/10.1080/19475705.2019.1650124\u003c/li\u003e\n\u003cli\u003ePerks, M. T., Russell, A. J., \u0026amp; Large, A. R. G. (2016). Technical note: Advances in flash flood monitoring using unmanned aerial vehicles (UAVs). \u003cem\u003eHydrology and Earth System Sciences\u003c/em\u003e, \u003cem\u003e20\u003c/em\u003e(10), 4005\u0026ndash;4015. https://doi.org/10.5194/hess-20-4005-2016\u003c/li\u003e\n\u003cli\u003eRevilla-Romero, B., Hirpa, F. A., Pozo, J. T. del, Salamon, P., Brakenridge, R., Pappenberger, F., \u0026amp; de Groeve, T. (2015). On the use of global flood forecasts and satellite-derived inundation maps for flood monitoring in data-sparse regions. \u003cem\u003eRemote Sensing\u003c/em\u003e, \u003cem\u003e7\u003c/em\u003e(11), 15702\u0026ndash;15728. https://doi.org/10.3390/rs71115702\u003c/li\u003e\n\u003cli\u003eSchumann, G., Giustarini, L., Tarpanelli, A., Jarihani, B., \u0026amp; Martinis, S. (2023). Flood Modeling and Prediction Using Earth Observation Data. \u003cem\u003eSurveys in Geophysics\u003c/em\u003e, \u003cem\u003e44\u003c/em\u003e(5), 1553\u0026ndash;1578. https://doi.org/10.1007/s10712-022-09751-y\u003c/li\u003e\n\u003cli\u003eVan Westen, C. J. (2013). Remote Sensing and GIS for Natural Hazards Assessment and Disaster Risk Management. \u003cem\u003eTreatise on Geomorphology: Volume 1-14\u003c/em\u003e, \u003cem\u003e1\u003c/em\u003e\u0026ndash;\u003cem\u003e14\u003c/em\u003e, 259\u0026ndash;298. https://doi.org/10.1016/B978-0-12-374739-6.00051-8\u003c/li\u003e\n\u003cli\u003eVousdoukas, M. I., Pennucci, G., Holman, R. A., \u0026amp; Conley, D. C. (2011). A semi automatic technique for Rapid Environmental Assessment in the coastal zone using Small Unmanned Aerial Vehicles (SUAV). \u003cem\u003eJournal of Coastal Research\u003c/em\u003e, \u003cem\u003eSPEC. ISSUE 64\u003c/em\u003e, 1755\u0026ndash;1759.\u003c/li\u003e\n\u003cli\u003eWahba, M., Sharaan, M., Elsadek, W. M., Kanae, S., \u0026amp; Hassan, H. S. (2024). Examination of the efficacy of machine learning approaches in the generation of flood susceptibility maps. \u003cem\u003eEnvironmental Earth Sciences\u003c/em\u003e, \u003cem\u003e83\u003c/em\u003e(14), 429. https://doi.org/10.1007/s12665-024-11696-x\u003c/li\u003e\n\u003cli\u003eWang, R. Q., Mao, H., Wang, Y., Rae, C., \u0026amp; Shaw, W. (2018). Hyper-resolution monitoring of urban flooding with social media and crowdsourcing data. \u003cem\u003eComputers and Geosciences\u003c/em\u003e, \u003cem\u003e111\u003c/em\u003e, 139\u0026ndash;147. https://doi.org/10.1016/j.cageo.2017.11.008\u003c/li\u003e\n\u003cli\u003eWang, Y., Sun, D., Wen, H., Zhang, H., \u0026amp; Zhang, F. (2020). Comparison of Random Forest Model and Frequency Ratio Model for Landslide Susceptibility Mapping (LSM) in Yunyang County (Chongqing, China). \u003cem\u003eInternational Journal of Environmental Research and Public Health\u003c/em\u003e, \u003cem\u003e17\u003c/em\u003e(12), 4206. https://doi.org/10.3390/ijerph17124206\u003c/li\u003e\n\u003cli\u003eWijayarathne, D. B., \u0026amp; Coulibaly, P. (2020). Identification of hydrological models for operational flood forecasting in St. John\u0026rsquo;s, Newfoundland, Canada. \u003cem\u003eJournal of Hydrology: Regional Studies\u003c/em\u003e, \u003cem\u003e27\u003c/em\u003e(March 2019), 100646. https://doi.org/10.1016/j.ejrh.2019.100646\u003c/li\u003e\n\u003cli\u003eWu, W., Emerton, R., Duan, Q., Wood, A. W., Wetterhall, F., \u0026amp; Robertson, D. E. (2020). Ensemble flood forecasting: Current status and future opportunities. \u003cem\u003eWiley Interdisciplinary Reviews: Water\u003c/em\u003e, \u003cem\u003e7\u003c/em\u003e(3), 1\u0026ndash;32. https://doi.org/10.1002/WAT2.1432\u003c/li\u003e\n\u003cli\u003eWu, X., Shen, X., Li, J., \u0026amp; Xie, X. (2024). Determination and projection of flood risk based on multi-criteria decision analysis (MCDA) combining with CA-Markov model in Zhejiang Province, China. \u003cem\u003eUrban Climate\u003c/em\u003e, \u003cem\u003e53\u003c/em\u003e(October 2023). https://doi.org/10.1016/j.uclim.2023.101769\u003c/li\u003e\n\u003cli\u003eZanchetta, A. D. L., \u0026amp; Coulibaly, P. (2020). Recent advances in real-time pluvial flash flood forecasting. \u003cem\u003eWater (Switzerland)\u003c/em\u003e, \u003cem\u003e12\u003c/em\u003e(2). https://doi.org/10.3390/w12020570\u003cstrong\u003e\u003c/strong\u003e\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":true,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"environmental-science-and-pollution-research","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"espr","sideBox":"Learn more about [Environmental Science and Pollution Research](https://www.springer.com/journal/11356)","snPcode":"11356","submissionUrl":"https://submission.nature.com/new-submission/11356/3","title":"Environmental Science and Pollution Research","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"disaster risk management (DRM), flood forecasting, flood hazard mapping (FHM), flood monitoring, machine learning algorithms","lastPublishedDoi":"10.21203/rs.3.rs-6948732/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-6948732/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eFloods are one of the most dreaded hazards adversely affecting life and property which demands and necessitates singular attention. This paper aims at documenting the trajectory of the recent trends and advancements in monitoring and prediction, including probabilities of associated uncertainties from a \u0026ldquo;flood\u0026rdquo; point of view. The primary focus is to convey a comparative vision of vulnerability and risk assessment bringing a viable grasp in approaching problems associated with the sudden extremities of floods. Over a timeline spanning 17 years (2007\u0026ndash;2024), a systematic review is conducted utilising Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines, dividing the findings into two sections: monitoring and forecasting techniques. The monitoring section encompasses a range of technologies, from ground-level observations to advanced tools, involving global scales to community levels, in perceiving impacts of flood events. Monitoring is enhanced by drone technology, sensor cameras, remote sensing and digital image analysis allowing spatial risk visualisation. The forecasting section pays attention to the utilisation of geospatial guides, GIS tools and machine learning algorithms in predicting the temporal probabilities of flood. The prediction often leverages computer training ensemble models, Artificial Neural Networks (ANN) and Convolutional Neural Networks (CNN). This review article serves as an efficacious resource for researchers and decision-makers, streamlined to enhance understanding of the entire flood hazard management continuum\u0026mdash;from initiation and response to mitigation and recovery. By scientifically compiling relevant techniques discussed here, the present work will steer future studies on flood hazards and actively promote adaptive strategies to mitigate their disastrous effects.\u003c/p\u003e","manuscriptTitle":"Recent Advances in Flood Monitoring \u0026amp; Prediction Methods: A Systematic Review","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-07-14 10:14:49","doi":"10.21203/rs.3.rs-6948732/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"reviewerAgreed","content":"","date":"2025-07-08T13:37:18+00:00","index":0,"fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-07-08T12:48:19+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-06-25T04:51:31+00:00","index":"","fulltext":""},{"type":"submitted","content":"Environmental Science and Pollution Research","date":"2025-06-23T09:30:22+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"environmental-science-and-pollution-research","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"espr","sideBox":"Learn more about [Environmental Science and Pollution Research](https://www.springer.com/journal/11356)","snPcode":"11356","submissionUrl":"https://submission.nature.com/new-submission/11356/3","title":"Environmental Science and Pollution Research","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"7d16f695-5b33-461b-bcd0-110c80387be6","owner":[],"postedDate":"July 14th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[],"tags":[],"updatedAt":"2026-01-12T16:15:40+00:00","versionOfRecord":{"articleIdentity":"rs-6948732","link":"https://doi.org/10.1007/s11356-025-37366-4","journal":{"identity":"environmental-science-and-pollution-research","isVorOnly":false,"title":"Environmental Science and Pollution Research"},"publishedOn":"2026-01-07 15:58:16","publishedOnDateReadable":"January 7th, 2026"},"versionCreatedAt":"2025-07-14 10:14:49","video":"","vorDoi":"10.1007/s11356-025-37366-4","vorDoiUrl":"https://doi.org/10.1007/s11356-025-37366-4","workflowStages":[]},"version":"v1","identity":"rs-6948732","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-6948732","identity":"rs-6948732","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}