Flood Extent Research Articles

Unveiling the role of storm surges as a driver of flooding on the western Mediterranean: a case study of the Ebro Delta

AbstractStorm surges in the Western Mediterranean are generally low in magnitude, making their contribution to coastal flooding less significant compared to wave overtopping. Nonetheless, low-lying, sheltered coasts such as deltas and wetlands, which are frequent along the Mediterranean basin are particularly vulnerable to storm surges. This study, focusing on the Ebro Delta as representative of this type of coastal environment, investigates the flooding caused by storm surge alone and in conjunction with other non-wave related factors like astronomical tides and sea level rise (SLR), using the LISFLOOD-FP model. The findings highlight the significant flooding potential of storm surges on passive, and unprotected coasts, while also indicate that astronomical tides have a minor effect on flood extent under prevalent microtidal conditions. SLR greatly increases the impact of storm surges, amplifying temporary inundation in the short term and becoming the dominant factor over time. The study underscores the importance of accurately representing surge duration and small topographic features in flood models to ensure robust coastal inundation assessments in low-lying areas.

Scaling Terrain-Aware Spatial Machine Learning for Flood Mapping on Large Scale Earth Imagery Data

The accurate and prompt mapping of flood-affected regions is important for effective disaster management, including damage assessment and relief efforts. While high-resolution optical imagery from satellites during disasters presents an opportunity for automated flood inundation mapping, existing segmentation models face challenges due to noises like cloud cover and tree canopies. Thanks to the digital elevation model (DEM) data readily available from sources such as United States Geological Survey (USGS), terrain guidance was utilized by recent graphical models such as hidden Markov trees (HMTs) to improve segmentation quality. Unfortunately, these methods either can only handle a small area where water levels at different locations are assumed to be consistent, or require restricted assumptions such as there is only one river channel. This paper presents an algorithm for flood extent mapping on large-scale Earth imagery, applicable to a large geographic area with multiple river channels. Since water level can vary a lot from upstream to downstream, we propose to detect river pixels in order to partition the remaining pixels into localized zones, each with a unique water level. In each zone, water at all locations flow to the same river entry point. Pixels in each zone are organized by an HMT to capture water flow directions guided by elevations. Moreover, a novel regularization scheme is designed to enforce inter-zone consistency by penalizing pixel-pairs of adjacent zones that violate terrain guidance. Efficient parallelization is made possible by coloring the zone adjacency graph to identify zones and zone-pairs that have no dependency and hence can be processed in parallel, and incremental one-pass terrain-guided scanning is conducted wherever applicable to reuse computations. Experiments demonstrate that our solution is more accurate than existing solutions, and can efficiently and accurately map out flooding pixels in a giant area of size 24,805 × 40,129. Despite the imbalanced workloads caused by a few large zonal HMTs dominating the serial computing time, our parallelization approach is effective and manages to achieve up to 14.3 × speedup on a machine with Intel Xeon Gold 6126 CPU @ 2.60GHz (24 cores, 48 threads) using 32 threads.

Embankment Breach Simulation and Inundation Mapping Leveraging High-Performance Computing for Enhanced Flood Risk Prediction and Assessment

Abstract. Embankment breaches represent significant hazards to communities and infrastructure, precipitating catastrophic flood occurrences. The precise prediction of floods and understanding the scope of inundation stemming from embankment failures are imperative for effective disaster preparedness and response. This research delves into a case study on simulating embankment breaches to evaluate the extent of flooding. Leveraging advanced hydrodynamic models validated through high-performance computing (HPC) systems, and integrating real-time data assimilation, we aim to improve accuracy in flood forecasting. The study endeavours to bolster flood risk management by furnishing detailed inundation maps and insights into embankment breach dynamics, thereby facilitating enhanced preparedness and response strategies. Our findings reveal that simulations conducted on multicore processors offer superior performance compared to single-core setups, yielding enhanced result accuracy and providing administrators with increased lead time. It unlocks high-resolution simulations for intricate basin details, explores a wider range of flood scenarios quickly, and allows for efficient ensemble modelling to assess model uncertainty. Through HPC utilization, we can harness high-resolution digital elevation models (DEMs) for 2D-hydrodynamic modelling, enabling rapid assessment of water spread resulting from embankment breaches within a mere 20-minute timeframe, a significant improvement from the previous 3–4 hours duration.

Flood impact assessment in remote areas using machine learning, SAR, and GIS: a case study of Ngabang District, Indonesia

ABSTRACT Flooding in remote regions presents significant challenges due to data scarcity, complicating impact assessment and mitigation efforts. This research delineates an integrated methodology for quantifying flood impacts in such contexts. By leveraging machine-learning algorithms, Sentinel-1 synthetic aperture radar (SAR) imagery was combined with digital elevation model data and river proximity metrics to predict and accurately demarcate flood extents. Geographic information systems overlay techniques were then employed for spatial analysis of the floods’ impacts on population and infrastructural assets. The methodology was applied in a case study in Ngabang District, Indonesia, demonstrating its utility. Analysis using decision tree, random forest (RF), and gradient boosting machine models provided critical insights into flood prediction factors. The RF model was chosen as the best, successfully identified flood-prone regions, achieving an accuracy of 0.94 and a Kappa of 0.87 on the testing data, demonstrating its robustness. The flood map showed significant impacts, affecting 373.81 hectares, 10,706 people, 1,500 buildings, and 15 km of roads. This study highlights the importance of proximity, elevation, SAR imagery, and iterative model improvements in flood prediction, offering valuable insights for flood management and mitigation efforts in data-scarce regions.

Characterizing interactive compound flood drivers in the Pearl River Estuary: A case study of Typhoon Hato (2017)

Tropical cyclone (TC) induced compound floods involve dynamic interactions among astronomical tides, storm surges, precipitation, and associated river pulses. This study employs a one-way coupled WRF, Delft3D, and HEC-RAS model to investigate the impacts of oceanic, pluvial, and fluvial processes on compound flood dynamics during Typhoon Hato (2017) in the Pearl River Estuary (PRE), South China. Total water levels driven by different combinations of flood drivers are modeled and analyzed. Relative contributions of each type of flood driver are quantified and used to categorize flood zones. This study highlights the persistent impacts of storm surges and their nonlinear interactions with other flood drivers. They can modify both the timing and magnitude of maximum water levels, thereby distorting tidal signals and contributing to post-TC landfall water level peaks. Along coastal regions, water levels exhibit three successive peaks, predominantly driven by storm surge, rainfall, and the combined actions of both factors, respectively. In upstream regions and coastal areas sheltered from islands, a singular water level peak arises exclusively from rainfall-runoff processes. Moreover, nonlinear interactions between surge and rainfall-runoff have non-negligible impacts on the relative contribution of individual flood drivers, which underscores the necessity of considering both rainfall and storm surge in modeling compound flood water levels. During the flooding period, peak storm surge and the following peak rainfall resulted in a time-varying distribution of flood zones. Alternating feedback between compounded and ocean-dominant areas manifests in the midsections of the upper PRE. Maximum flooding depth, extent, and duration are mainly influenced by rainfall. Storm surges play a secondary role, causing intense but short-lived flooding in coastal regions. These findings aid in understanding the generation mechanism of compound floods and provide references for hazard mitigation strategies.

Assessment of socio-economic strategies for managing regional flood risk in an urban coastal catchment

With the cascading flooding events over the past decades, several structural and non-structural measures have been effectively implemented for improved flood management. However, the uneven impacts of the flooding events have highlighted that socio-economic inequality amongst the community needs to be addressed along with hazards to devise long term and sustainable flood management strategies. This study aims to evaluate the impact of socio-economic adaptive scenarios over an urban coastal catchment for reduction in flood risk. Flood risk is estimated utilizing socio-economic aspects (vulnerability) and flood inundation (hazard), for the highly flood-prone coastal city of Mumbai, India. A multivariate approach is used for assessing socio-economic vulnerability, starting with Principal Component Analysis followed by Data Envelopment Analysis, for various socio-economic scenarios, i.e., by reducing percentage of certain feasible socio-economic sensitivity indicators, to identify the key indicators, which contribute to significant change in vulnerability. The flood hazard is assessed through a comprehensive hydrodynamic flood modelling approach to quantify the flood inundation depth and extent. The flood risk is, thereafter, evaluated by combining socio-economic vulnerability and flood hazard using bivariate risk mapping approach. The results indicate that, with the decrease in alteration percentage of certain feasible socio-economic sensitivity indicators, the overall vulnerability and flood risk may be reduced. The areas under very high and high vulnerability classes reduce gradually for all scenarios as the percentage reduction for various indicators is varied from 5 % to 15 %, total illiteracy being the most significant indicator. This study marks the first effort to discern socio-economic vulnerability indicators that have the potential to significantly diminish flood risk, thereby constituting a viable long-term flood risk mitigation strategy. This novel framework is generic and applicable to any flood prone catchments, to devise appropriate long-term flood management measures.

Climate Extremes Walking Together: Evidence from Recent Compounding Climate Hazards after Remal

On May 26, 2024, Tropical Cyclone (TC) Remal made a devastating landfall along the coastal regions of India and Bangladesh. The sequence of devastating events began with prevailing heatwaves, followed by the cyclone’s landfall and subsequent heavy rainfall coupled with early onset of monsoon led to floods, landslides, and the resurgence of hot spells. It illustrates the complexity of managing such cascading effects of compound climate hazards. Multisource Earth Observation (EO) data with Google Earth Engine (GEE) was used to study the dynamics of Remal and its cascading impacts. IbTracks, INSAT-3D, GPM, Sentinel-1, and PlanetScope were involved in analyzing the cyclone's progression and impacts. Remal's slow progression extended the impact of the cyclone as it hovered over Bangladesh for more than 50 hours. As observed from the GPM data, the right side of the cyclone's path experienced more extensive and severe flooding than the left. Flood extent mapping using the Sentinel-1 data and Otsu thresholding depicted Jogipara, Assam in India and Jashore, Narail, and Khulna in Bangladesh as among the worst affected regions. Additionally, the early onset of the monsoon and the triggered cyclonic rainfall in the area led to landslides in Assam and Meghalaya during this period. The present study advocates the usage of multisource EO data and GEE for quick assessment of compound hazards started Remal. The study also highlights the need for a deeper understanding to develop a more robust district disaster management plan integrating EO data for actionable disaster risk reduction strategies for TC.

Simulation of floods under the influence of effective factors in hydraulic and hydrological models using HEC-RAS and MIKE 21

Simulating floods and creating vulnerability maps are essential in urban watershed management systems. The present research aimed to simulate floods and assess floodplain risk in the Khalkai watershed located in Gilan Province, Iran. This river is characterized by extensive agricultural areas and urban areas that become flooded during most rainy seasons. Hence, an accurate assessment of flood extent and behavior is necessary. This study employs HEC-HMS, HEC-RAS, and MIKE 21 for advanced hydrodynamic modeling, focusing on two-dimensional (2D) capabilities essential for urban areas and flood is simulated with probability return periods of 50, 100, and 500 years. The research objectives include comparing HEC-RAS and MIKE 21 to identify the most effective flood simulation model, evaluating the impact of hydrological and hydraulic factors, the influence of GPS topographic mapping compared with DEMs extracted from satellite images and assessing land-use changes on flood behavior. Model accuracy is validated against historical flood data, revealing that MIKE 21 effectively addresses challenges related to grid resolution and flood delineation. Results indicate that while HEC-RAS provides a broader extent of flooding, both models offer critical insights into flood depth and flow characteristics. This research contributes to enhanced flood risk management by illustrating the implications of land-use changes on flood events, offering valuable insights for policymakers in mitigating future flood impacts and developing sustainable urban strategies.

Flood risk assessment of the Narmada Basin, India, under climate change scenarios

ABSTRACT Climate change has led to a significant rise in the occurrence of extreme events such as floods and droughts. It is essential to accurately assess the hazards linked to climate change to effectively tackle flood risks in the future. This study investigates the risk of flooding in the Narmada Basin under climate change scenarios. Future precipitation data of global climate models are used to create intensity–duration–frequency curves and peak flows are estimated using a hydrological model. These peak flows serve as inputs for the hydraulic model to simulate floods across various return periods. The fuzzy inference system (FIS) is employed to optimize releases and reduce peak flows to mitigate flood impacts. The findings indicate that the extent of flooding and maximum water depths increase with higher return periods and future scenarios, and optimizing the reservoir releases, decreases the extent of flooding. This decrease in inundation area ranges from approximately 9 to 10% in the Bargi-to-Indira Sagar section to 20–23% in the Indira Sagar-to-Omkareshwar Sagar segment. In the Omkareshwar Sagar-to-Sardar Sarovar segment, the flooded area diminishes by 17–18% under the Radioactive Concentration Pathway (RCP) 4.5 and by 12–14% under the RCP 8.5 scenarios.

Evaluating the Enhanced Bathtub Model for Coastal Flood Risk Assessment in Table Bay, South Africa

ABSTRACTCoastal zones are susceptible to increasing pressures from urban development and natural hazards, such as storm events, climate change, and rising sea levels. The GIS‐based enhanced bathtub model (eBTM) enables the identification of areas at risk of flooding as a baseline for disaster management and coastal adaptation. This study aims to establish the methodological robustness of the eBTM for coastal flood modeling, by analyzing eight sites flooded during a recent storm event in Table Bay, Cape Town by comparing eBTM outputs with observed flood extent data collected after the storm. The validation showed that for 74% of the 332 validation points the spatial modeling error was < 6 m and for 56% below 3 m. The root mean square error for the model was 4.88 m, indicating an acceptable level of accuracy of the eBTM outputs for coastal risk assessments where more sophisticated models are unavailable.

Automatic Detection of Flood Extent and Volume Estimation from Sentinel-2 Satellite Images using Deep Learning Techniques in India

Floods are unexpected natural disasters that can have a major impact on human life, soil along bank erosion, damage vital infrastructure, road closures, economy standard, and society of various affected regions. An initial step of proper assessment is necessary for flood damage along with accurate measurements to easily restore essential damage of infrastructure, relief, and mitigation as quickly as possible. Nowadays, the rapid development of remote sensing images using deep learning as a most positive tool for accurately estimating the extent of overall flood detection surfaces. The monitoring of flood detections from remote sensing images still extends a few issues due to mostly varying from different weather changes conditions, cloud coverage areas that can have a limit to use of level of visible remote sensing satellite collected data. Moreover, Remote Sensing Satellite based observations may not always be mapped to the distribution's flood point peak, also it is very essential for both the flood extent and flood volume estimation. To overcome this challenge, we have presented a new remote sensing technology that integrates with a high resolution multi-spectral satellite data/information by using an advanced Deep Learning to accurately analyze remote sensing based observations. In our experiment, we use the European Space Agency (ESA) launched Sentinel type-1, Sentinel type-2 data and Digital Elevation Model (DEM) to accurately measure flood monitoring results. In our study, we reviewed a real example of the flood situation that happened in 2019 in Kolhapur. In our results, we evaluated a flood volume estimation at 0.0010 km3 in Kolhapur district. Finally, the proposed methodology provides an effective way to accurately motoring floods using low-cost satellite data and deep learning approaches. This project has the potential to improve the more accurate flood detection and mapping which can prevent an exactly timely response and immediate recovery efforts for flood surrounding affected areas.

Enhancing precision of flood estimation in EOS-04 SAR imagery: a statistical approach

Floods pose a recurrent and devastating natural disaster in India, and accurate and timely assessment of their extent is crucial for effective disaster management and mitigation efforts. In this regard, this study proposes a novel approach for flood inundation estimation through the statistical analysis of two key geospatial datasets: the Normalized Difference Flood Index (NDFI) and temporal Standard Deviation images. To establish a baseline under ordinary conditions, two products were derived. The first product is the mean image during the non-flood season, and the second is the temporal Standard Deviation image for the entire year. These images are derived from Sentinel-1A Synthetic Aperture Radar (SAR) data using Google Earth Engine (GEE). The NDFI is calculated by comparing the mean image to the image obtained during flood event, allowing the identification of flood-affected areas. The process involves constructing Kernel Density Estimation (KDE) plots for NDFI and Standard Deviation, from which a 95% density ellipse is generated using the covariance matrix and eigenvalues. Multiple SAR scenes from diverse regions in India are analyzed individually, yielding density ellipses for each location. The innovative step lies in synthesizing these individual ellipses into the best fit ellipse through the convex hull method. This best fit ellipse encapsulates the general flood characteristics observed across disparate geographic regions, providing a unified representation. The derived best fit ellipse serves as a powerful tool for generating near-real-time flood inundation maps, with an overall accuracy of over 98.2%. Thus, the fusion of statistical insights and historical SAR information enables rapid near-real-time flood mapping, with enhanced accuracy of flood extent predictions.

A review of remote sensing of flood monitoring and assessment in southern Africa

Southern Africa is one of the most vulnerable regions to flooding and this severely impacts its economic development, human livelihoods and ecosystem functioning. In this regard, there is need to identify strategies to monitor flood occurrence, to minimize effects. Remote sensing is one of the key data sources for natural hazards monitoring, over space and time. This paper therefore provides the state-of-the-art review on flood monitoring, using remote sensing in southern Africa, since the emergency of earth observation technologies. Specifically, the review focused on how southern Africa has embraced remote sensing for mapping flood extent, vulnerable areas and impacts, over time. The review also highlights available remote sensing data and products, to monitor floods, including their success, limitations, and prospects for improved flood monitoring in the region. Overall, there has been limited use of remote sensing data in flood monitoring in southern Africa, until 2010. Since then, there was an increase in the use of remote sensing data, for flood monitoring. Most of these studies used the freely available Landsat and MODIS datasets, and these studies focused more on mapping the extent of flooding. However, as much as considerable strides were made, there is still more work to be done. Future research needs to shift towards the use of new generation remote sensing data, including radar, as well as high spatial resolution drones, before, during and after flood occurrence. Advanced cloud-computing, such as Google Earth Engine and machine learning algorithms, also present opportunity for time series analysis of flooding.

Land and Atmospheric Drivers of the 2023 Flood in India

AbstractFloods in India are recurring natural disasters resulting from extreme precipitation during the summer monsoon season (June–September). The recent flood in North India in July 2023 caused substantial damage to lives, agriculture, and infrastructure. However, what led to the 2023 North India flood and the role of atmospheric and land drivers still need to be examined. Using in situ observations, satellite data, and ERA5 reanalysis combined with hydrological and hydrodynamical modeling, we examine the role of land and atmospheric drivers in flood occurrence and its impacts. Extreme precipitation in a large region during 7–10 July 2023 created favorable conditions for the flood in the hilly terrains and plains of north India. More than 300 mm of precipitation fell in just 4 days, which was eight times higher than the long‐term average (2001–2022). Anomalously high moisture transport over northern India was recorded on 7 July 2023, making atmospheric conditions favorable for intense landfall. Increased column water vapor and specific humidity at different pressure levels confirmed the continuous moisture presence before the extreme rainfall that caused floods in northern India from 7 to 12 July 2023. Atmospheric and land (high antecedent soil moisture) conditions contributed to a more than 200% rise in streamflow at several gauge stations. Satellite‐based flood extent shows a considerable flood inundation that caused damage in the Sutlej and Yamuna River basins. Our findings highlight the crucial role of the favorable land and atmospheric conditions that caused floods and flash floods in north India in July 2023.

Hurricane surge and inundation in the Bahamas, part 2: Flood risk assessment

AbstractA hurricane surge and inundation risk assessment has been carried out for Grand Bahama and Eleuthera in The Bahamas. 10,000 years of synthetic hurricane tracks were generated based on a statistical analysis of historical hurricanes from 1979 to 2020 inclusive. The surge and inundation were modelled using a hydrodynamic TELEMAC‐2D model covering sea around The Bahamas and the land of the two islands. Predictions of flood extents and depths were mapped for return periods of up to 1000 years for present day conditions and three climate change scenarios to 2100. The flooding experienced over Grand Bahama during Hurricane Dorian in 2019, was estimated to have a return period of up to approximately 450 years. Under the climate change scenarios the likelihood of flooding similar to Hurricane Dorian was estimated to be approximately 1.7 times more likely in 2050 and up to 3.4 times as likely in 2100 under a high greenhouse gas emissions scenario. The storm surge maps can be used by the Bahamas Department of Meteorology and other government agencies for emergency management, planning of infrastructure and building resilience in response to climate change.

A synergic approach using the model and remote sensing data for flood monitoring in under-observed transboundary rivers

ABSTRACT The southern plain of Nepal recognized as the ‘granary of Nepal’, confronts recurrent monsoon-induced flooding, posing a substantial threat to its pivotal role as a major agricultural contributor to the national economy. As an analysis, this study employs advanced satellite imagery to delineate historical floods in nine flood-prone transboundary basins and compares the rainfall-induced model-based inundation in the West Rapti Basin (WRB) to validate the result. The extent of flooding was mapped between 2015 and 2022 using Sentinel-1 Synthetic Aperture Radar data processed on Google Earth Engine. Hydrodynamic modelling centred on the WRB, incorporated daily measured precipitation data with varying return periods over a 10 m resolution digital elevation model generated through an in situ survey. The model was calibrated for the August 2017 flood event with Nash–Sutcliffe efficiency greater than 70% and validation reasonably with satellite-derived flood maps with Cohen's Kappa value of 0.58 and an overall accuracy metric of 0.84. This synergic approach integrates climatology, remote sensing data, and hydraulics to monitor transboundary river floods in Nepal where precise hydro-meteorological data are limited, thus, offering continuous all-weather monitoring.

Comprehensive Assessment of Flood Risk and Vulnerability for Essential Facilities: Iowa Case Study

In this study, nine different types of essential facilities in the state of Iowa (such as hospitals, fire stations, schools, etc.) were analyzed on a county level in terms of flood depth, functionality and restoration time after flooding, and damage sustained during flooding. These essential facilities were also analyzed on the state level in terms of their location relative to the 100 y and 500 y flood zones. Results show that the number of essential facilities within the flood extent reached up to 39%, and during the 100 y flood scenario all but one of the six chosen counties lost functionality of 100% of their facilities. Most essential facilities were found to have a flood depth of 1 to 4 ft deep and a restoration time of 480 days. The purpose of this study is to bring awareness to decisionmakers regarding the risk that flooding events pose to essential facilities and to highlight the increasing dangers of flooding on a broader scale. This study will be beneficial to improve mitigation strategies, emergency response plans, and ensuring that emergency services and facilities are available in the event of future floods for the affected areas.

Seasonal Controls on Microbial Depolymerization and Oxidation of Organic Matter in Floodplain Soils.

Floodplain soils are vast reservoirs of organic carbon often attributed to anaerobic conditions that impose metabolic constraints on organic matter degradation. What remains elusive is how such metabolic constraints respond to dynamic flooding and drainage cycles characteristic of floodplain soils. Here we show that microbial depolymerization and respiration of organic compounds, two rate-limiting steps in decomposition, vary spatially and temporally with seasonal flooding of mountainous floodplain soils (Gothic, Colorado, USA). Combining metabolomics and -proteomics, we found a lower abundance of oxidative enzymes during flooding coincided with the accumulation of aromatic, high-molecular weight compounds, particularly in surface soils. In subsurface soils, we found that a lower oxidation state of carbon coincided with a greater abundance of chemically reduced, energetically less favorable low-molecular weight metabolites, irrespective of flooding condition. Our results suggest that seasonal flooding temporarily constrains oxidative depolymerization of larger, potentially plant-derived compounds in surface soils; in contrast, energetic constraints on microbial respiration persist in more reducing subsurface soils regardless of flooding. Our work underscores that the potential vulnerability of these distinct anaerobic carbon storage mechanisms to changing flooding dynamics should be considered, particularly as climate change shifts both the frequency and extent of flooding in floodplains globally.

HydroSAR: A Cloud-Based Service for the Monitoring of Inundation Events in the Hindu Kush Himalaya

The Hindu Kush Himalaya (HKH) is one of the most flood-prone regions in the world, yet heavy cloud cover and limited in situ observations have hampered efforts to monitor the impact of heavy rainfall, flooding, and inundation during severe weather events. This paper introduces HydroSAR, a Sentinel-1 SAR-based hazard monitoring service which was co-developed with in-region partners to provide year-round, low-latency weather hazard information across the HKH. This paper describes the end user-focused concept and overall design of the HydroSAR service. It introduces the main processing algorithms behind HydroSAR’s broad product portfolio, which includes qualitative visual layers as well as quantitative products measuring the surface water extent and water depth. We summarize the cloud-based implementation of the developed service, which provides the capability to scale automatically with the event size. A performance assessment of our quantitative algorithms is described, demonstrating the capabilities to map the flood extent and water depth with an accuracy of >90% and <1 m, respectively. An application of the HydroSAR service to the 2023 South Asia monsoon seasons showed that monsoon floods peaked near 6 August 2023 and covered 11.6% of Bangladesh in water. At the peak of the flood season, nearly 13.5% of Bangladesh’s agriculture areas were affected.

Flood inundation monitoring using multi-source satellite imagery: a knowledge transfer strategy for heterogeneous image change detection

Flood emergency mapping is essential for flood management, often requiring near real-time extraction of large-scale flood extents by combining pre- and post-event multi-source remote sensing images. Pre-event optical imagery delineates the normal water extent, providing a benchmark for estimation of post-flood changes. Synthetic aperture radar (SAR) imagery provides rapid and accurate interpretation of flood inundation extent during the rainy period. Post-classification comparison is one of the fundamental methods for flood mapping of multi-source imagery, as the image characteristics of SAR and optical imagery are inherently different. However, the accuracy of flood mapping depends on the accuracy of water body delineation from single temporal imagery, which can lead to error propagation in flood extent determination. Change detection methods can extract flood extent directly from pre- and post-event imagery, but the models need to learn a consistent flood feature description between optical and SAR imagery from a limited flood dataset. Here, we propose an automatic method for flood inundation extent extraction using pre- and post-event multi-source imagery that requires only a small number of pseudo-change labels. The methodology adopts a pseudo-Siamese network framework as a Heterogeneous - Flood Inundation Extraction Network (H-FIENet), to detect flood extent between the pre‐ and post-event heterogeneous images. Spatially inconsistent multi-source pre- and post-event imagery was used to create a pseudo flood dataset, and this dataset was used to transfer water body feature knowledge from the pre-trained model to the flood extraction model using a cross-task knowledge transfer strategy. Pre- and post-event images from different satellite sources and different flood scenarios were used to evaluate the performance of H-FIENet. We found that: (1) The methodology produced an overall accuracy of over 0.94 for flood inundation extraction from multi-source heterogeneous pre- and post-event imagery. (2) H-FIENet can detect both expansion and recession of the flood extent using any dual-temporal imagery. Our work makes it possible to automate time-series flood monitoring from multi-source remote sensing imagery.