While floods are considered as costliest natural disasters globally, their impact on developing countries is disproportionately high in terms of lost lives and irreversible property damage. In India, 27.75 percentage of the country’s population is exposed to high flood risk areas out of which majority of the population are concentrated in the Ganges- Brahmaputra plain. The Brahmaputra River has a history of yearly flooding, and over 674 deaths and more than $1 million worth of property loss were reported during the years 2012–2023 due to the river’s annual flooding.

To mitigate the forthcoming effects of flooding, we introduce a spatial flood forecasting system in the middle Brahmaputra basin that predicts the spatial locations as well as magnitude of the impending flood with the help of multitemporal Synthetic Aperture Radar images. By taking the advantage of the abundant Earth Observation data available, we demonstrated the forecasting system that predict locations experiencing maximum flooding in a single flood plain of the middle Brahmaputra basin.

In the initial phase, 169 SAR images from 2006-2022 were processed to create an inventory of flood extents and levels in a 1Sq.Km spatial zones. The forecasting model has been established using LSTM network and is calibrated to predict the upcoming flood hit locations as well as the flood depths in predicted spatial zones. This paper presents the procedures and findings of the proposed forecasting model along with potential future directions of this work.

How to cite: Surampudi, S. and Kumar, V.: Spatial Flood Forecasting in Middle Brahmaputra Basin with the help of Time Series Spatio temporal SAR Images using LSTM Network, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-333, https://doi.org/10.5194/egusphere-egu24-333, 2024.

Rivers are corridors of freshwater that provide vital services for sustainable development and ecosystem functioning. Moreover, increase in frequency and severity of droughts and floods due to climatic change necessitates innovative and reliable techniques enabling continuous monitoring of river discharge to effectively manage risk. Since ground-based flow gauging stations are difficult to install and maintain, especially in remote regions, remote sensing methodologies have gained attention over the last decades.

In this study, we integrate Sentinel-1 Synthetic Aperture Radar (SAR) data and Sentinel-2 optical data to make best use of their advantages, namely, observation capability on cloudy-days and higher spatio-temporal resolutions, respectively. In our methodology, we first identify the water surface area at selected river reaches where flow observations are also available. The conceptual framework for computing water surface areas within the specified study boundaries entails the utilization of water indices, specifically the Normalized Difference Water Index (NDWI) and Modified Normalized Water Index (MNDWI), for Sentinel-2 and histogram-based backscattering intensity thresholding for the Sentinel-1 platform. Later, we establish relationships between the computed surface water areas and corresponding flow measurements. The Google Earth Engine (GEE) platform serves as the operational foundation for executing the methodology. We validate the satellite-based discharge estimations using observed in-situ discharge data obtained from three selected USGS gauging stations along the Mississippi River, USA. According to our preliminary results, the coefficient of determination values between estimated and observed discharge datasets range between 0.49-0.79, 0.44-0.77 and 0.49-0.74 for the studied river reaches. The methodology is being tested for other river reaches along the globe to test and improve its river discharge estimation accuracy.

How to cite: Tural, C. Y., Yilmaz, K. K., and Tarpanelli, A.: Integration of Sentinel-1 and Sentinel-2 Datasets for River Discharge Estimation, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-969, https://doi.org/10.5194/egusphere-egu24-969, 2024.

Floods stand out as the most frequent and costly natural disaster in Europe. In the EU alone, there have been documented more than 1500 flood events since 1980, causing over 4300 deaths and more than €170 billion in economic damages.

Due to the compounded developments of urbanization and climate change, the frequency of floods is expected to increase with severe impacts, possibly endangering lives and leading to economic losses. Moreover, floods mobilize pollutants stored in the subsurface and urban areas. Thus, current efforts, such as coastal barriers, restoration of river courses, or resilient city and landscape planning, focus on reducing vulnerability and risks from flooding. But to implement such measures, detailed information on where and when flooding occurs is necessary. This study aims to improve and implement satellite-based mapping of flood extent under Danish conditions by presenting different methods and algorithms utilizing Sentinel-1 (S1) Synthetic Aperture Radar (SAR) imagery, digital elevation models (DEM) and river geometry. In the broader literature, various methods have been proven to successfully map flood extent, such as deep learning (DL) and change detection (CD) as employed in the Global Flood Awareness System. However, DL require extensive training and labeled data that are often not available, and CD is reliant on a comprehensive pre-processing procedure of antecedent satellite imagery or accompanied with a datacube-based algorithm that exploits the satellite orbit repetition. While these methods can provide excellent results, the steep data requirements and pre-processing procedures hinder practical usage. On the other hand, single-temporal image flood extent mapping algorithms relying on histogram analysis offering a straightforward approach potentially yielding satisfying results, especially when accompanied by techniques such as image decomposition, region-growing, active contour models or image texture algorithms. But for single-temporal image histogram analysis to work in an automated setup, the two main problems, namely class imbalance and class overlap must be addressed properly. This study proposes a novel approach for single-temporal image histogram analysis by combining automatic local histogram thresholding with two image decomposition techniques for image tiling using a quadtree and a novel combination of k-means clustering and box tiling. This study implements a bimodality test and a subsequent local-threshold selection using gaussian mixture modelling and kernel-density smoothening, followed by contextual segmentation using region-growing. Furthermore, a novel approach for improving flood extent segmentation using a combination of DEM information, geographical stream location and region-growing is presented. The proposed method is showcased for two different flood events in Denmark from 2015 to 2022 using 10 x 10 m interferometric wide swath S1 SAR imagery. Results are evaluated using Sentinel-2 optical imagery where available, and otherwise evaluated against high-precision permanent water maps. Moreover, we utilize gauged timeseries of stream water level to evaluate the temporal evolution of flood extent over the period of a flood event.  

How to cite: Hansen, M., Vejby, J., and Koch, J.: Satellite-based Mapping of Flood Extent in Denmark, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-5543, https://doi.org/10.5194/egusphere-egu24-5543, 2024.

The necessity of flood risk mapping is critical for effective planning and disaster response, particularly in flood-prone regions like the Qaa'Jahran watersheds in Dhamar, Yemen. This research implements various machine learning methods, including Support Vector Machines (SVM), K-Nearest Neighbors (kNN), Random Forest (RF), Artificial Neural Networks (ANN), and Logistic Regression (LR), with the latter also functioning as the meta-model in our stacking ensemble approach for mapping flood susceptibility. The process began with creating a flood inventory map using SAR images and historical flood records. Our model integrates the individual strengths of each technique and employs a meta-model to synthesize their forecasts. This stacked ensemble approach demonstrated superior performance over each model alone, achieving a remarkable AUC score of 0.9848 compared to the individual scores of SVM, LR, kNN, ANN, and RF. It also surpassed two innovative models, ABRBF and TPOT, in accurately pinpointing all high-risk zones identified in historical flood data. This advancement in flood risk mapping for the Qaa'Jahran watersheds exemplifies the potential of our model in enhancing disaster management and prevention efforts. It offers a significant tool for identifying at-risk areas and guiding mitigation strategies to safeguard communities in Dhamar, Yemen, against the catastrophic impacts of flooding.

How to cite: Ghaleb, M., Al-Areeq, A., Al-Areeq, N., Saleh, R., AbuDaqa, A., and kawara, A.: A Stacking Ensemble Method for Comprehensive Flood Susceptibility Mapping in Yemen, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-6430, https://doi.org/10.5194/egusphere-egu24-6430, 2024.

Estimating water flows and stocks in surface hydrology is crucial for addressing important socio-economic issues, such as managing water resources and predicting extreme events like floods and droughts. These challenges become more significant with the ongoing global climate change, which may intensify the hydrological cycle. Advanced modeling tools are necessary for making precise and reliable local forecasts. However, hydrological models, regardless of their complexity and status, encounter difficulties in accurately and reliably predicting quantities of interest such as river flows or soil moisture states, and in accounting for meteorological-climatic effects on hydrology. Given the complexity of the physical processes involved and their heterogeneous and limited observability, hydrological modeling is a challenging task, and internal flows often have significant uncertainties. These uncertainties could be reduced by integrating new observations from remote sensing applied to continental surfaces, which is rapidly evolving. A variety of satellites and sensors now allow the observation of watershed surface characteristics and hydrological responses with increasing spatial-temporal resolutions. In particular, products of soil moisture, evaporation, and land use are now available at relatively high spatial-temporal resolution. This work focuses on improving the integration of satellite and in-situ land surface data into spatially distributed hydrological models. The Hybrid Data Assimilation and Parameter Regionalization (HDA-PR) approach incorporating learnable regionalization mappings, based on neural networks into the differentiable hydrological model SMASH, is modified to account for satellite moisture maps in addition to discharge at gauging stations and basins physical descriptors maps. Regional optimizations are performed on flash-flood-prone areas located in the South of France and their accuracy and robustness is evaluated in terms of simulated discharge and moisture against observations. 

How to cite: Ettalbi, M., Garambois, P.-A., Baghdadi, N., Ferreira, E., and Huynh, N.-N.-T.: Multi-source in situ and satellite variational data assimilation into a fully distributed hydrological model for floods and droughts modeling over poorly gauged and ungauged areas, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-8214, https://doi.org/10.5194/egusphere-egu24-8214, 2024.

Flood mapping, particularly in data-scarce regions, poses challenges including inadequate observational data to understand the hydrological characteristics of the floods. This study addresses this research gap by utilizing remotely sensed data, specifically Sentinel-1 Synthetic Aperture Radar (SAR) images, to delineate flood extent related to the September 11, 2023 Derna flood event in Libya. The objective is to extract flood extent from both SAR intensity and coherence and integrate these characteristics to generate a confidence flood map.

Our approach involves radiometric terrain correction of SAR data, flood pixel identification using anomaly detection techniques based on SAR intensity, and coherence analysis of pre-and post-flood SAR images. Flooded areas are categorized into 3 main classes. These include (1) High Confidence Flood (HCF), which is the intersection of SAR intensity and coherence in VV and VH bands in both Ascending and Descending directions; (2) Medium Confidence Flood (MCF), extracted from intensity and coherence in either the Ascending or Descending direction in both VV and VH bands; and (3) Low Confidence Flood (LCF), extracted from a single direction in either VV or VH band. LCF includes all pixels not confidently identified as part of either HCF or MCF.  The effectiveness of flood segmentation utilizing the integration of anomaly detection of SAR intensity and coherence analysis method is evaluated through a comparison between Sentinel-1 SAR data and optical Planet imagery.

Our findings indicate HCF covering approximately 8 hectares, MCF covering around 24 hectares, and LCF covering more than 227 hectares. These findings offer valuable insights into the observed flood extent at varying confidence levels. However, the moderate temporal resolution of Sentinel-1 data, with a revisit time of 12 days, introduces challenges in promptly detecting the entire extent of the flood. Overall, this study underscores the significance of remote sensing technology in near-real-time flood monitoring, emphasizing its role in identifying vulnerable areas, prioritizing resources, planning for potential risks, and supporting decision-making in relief efforts.

How to cite: Hotaki, M. S. F., Motagh, M., and Haghshenas Haghighi, M.: Flood inundation Mapping for the Sept. 2023 Derna, Libya flood event using Sentinel-1 SAR data: An Integration of SAR intensity and interferometric coherence, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-11215, https://doi.org/10.5194/egusphere-egu24-11215, 2024.

Climate change intensifies the occurrence of severe flood events, increasing the demand for flood modeling studies. Hydrodynamic models can effectively represent flood events, but they are limited by the quality of available observations. Accurate topographic elevation is essential to replicate channel-floodplain interaction. Elevation is normally retrieved using satellite-based DEMs. However, freely available DEMs have a low spatial resolution, which is a limitation in identifying small-scale channel features in complex floodplain topography. In addition, these products can present issues such as vertical offset, random noise, or vegetation biases. These issues can lead to large errors when used in hydraulic modeling to simulate water levels and inundation extent. FABDEM is a 1 arcsec DEM, that removes forest and building artifacts from Copernicus DEM, but to map complex floodplain topography, finer resolution is needed. ICESat-2 mission offers a large spatial coverage with an along-track resolution down to 70 cm in the ATL03 product. This data product has shown great potential when mapping river topography and identifying small-scale channel features. ATL03 can be used as a control point dataset, to correct biases and refine DEMs

To improve the accuracy of 2D hydraulic models, FABDEM was corrected on selected floodplain areas using supplementary data and machine learning methods. Artificial Neural Network (ANN) was used in the correction of FABDEM. This regression algorithm can predict differences between FABDEM floodplain elevation and ATL03 reference elevation, inputting data from Sentinel-2 and water occurrence maps produced from spectral and SAR imagery. The output floodplain elevation has a reduced vertical offset and a spatial resolution of 10 m, which can detect small-scale channel features. Flood inundation was simulated using the updated DEM. The high computational cost of 2D hydraulic models is a limitation when using discharge time series. To deal with computational cost, discharge classes were defined to represent different inundation scenarios that provide a good indicator for flood risk management, and steady-state inundation patterns were simulated for each discharge class.

The method is demonstrated in a section of the Upstream Yellow River characterized by large floodplains with complex topography, and small-scale channels. Discharge observations from the Jimai in-situ station are used to define discharge classes. The discharge classes are defined by calculating the exceedance probability of a discharge value. The inundation scenarios are simulated for high flow discharge values for an exceedance probability of 25% (Q25) and 10% (Q10), and medium flow discharge values with 50% (Q50), and are compared with the corresponding water occurrence map produced from spectral and SAR imagery for the given discharge class. The critical success index (CSI) of the inundation map improves by about 5% using FABDEM corrected version for Q10 and Q25, and about 4% for Q50. In addition, we observe a consistent Bias reduction of about 20% for Q10 and Q25.

How to cite: Coppo Frias, M., Liu, S., Mo, X., Druce, D., Yamazaki, D., Folkmann Musaeus, A., Nielsen, K., and Bauer-Gottwein, P.: Improving 2D hydraulic modeling in floodplain areas with ICESat-2 data: A case study in Upstream Yellow River, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-14669, https://doi.org/10.5194/egusphere-egu24-14669, 2024.

Changing climate, intense rainfall, and geomorphological conditions within the Indo-Gangetic Basin (IGB) have led to recurring flooding within the area in the recent past. The devastating August 2022 floods in Pakistan affected 33 million people causing severe loss of life and property. The occurrence of such flooding events has increased the need to understand the complexities of the interplay between flood hazards, exposure, vulnerability, and risk. This study delves into flood risk within India's Ganga Basin, examining the flood-inducing factors, vulnerability, and exposure through an innovative approach using NASA's Black Marble Nighttime Lights Product Suite (VNP46).  The product suite, available globally on daily, monthly, and annual composite scales, corrects extraneous sources of noise in nighttime light (NTL) radiance signals and has proven effective in disaster monitoring, risk assessment and reduction, humanitarian response, preparedness, resilience, and sustainable development.

Our work to date has successfully utilized these NTL data to quantify flood exposure and the impact of flooding in both urban and rural areas by analyzing changes in radiance across time and space. However, to improve our understanding of human response to floods, we now focus on a more intricate analysis: incorporating geomorphological and socio-hydrological factors into a risk assessment approach.

Our study evaluates flood hazard, exposure, and vulnerability as three separate entities and combines them using a multi-criterion decision tool to assess flood risk within the basin. Flood hazards are studied as a relationship between geomorphological and hydrological parameters, whereas flood vulnerability is studied using land use and land cover data. The novelty of this research is using NASA’s Black Marble nightlights as a proxy to study flood exposure. We argue that the NTL data can more effectively capture the human presence and economic activities compared to some conventional parameters for flood exposure such as population count, household density, and literacy amongst others. By integrating these diverse data layers using the robust Analytical Hierarchical Process (AHP), we generate comprehensive flood risk maps across the Ganga Basin spanning a decade. The accuracy of these maps is validated against historical flood event data from the EM-DAT database. Ultimately, our research culminates in a spatially explicit and data-driven approach to flood risk assessment, which can empower targeted mitigation strategies and proactive planning within the basin.

How to cite: Aggarwal, E., de Ruiter, M. C., Buijs, S., Whittaker, A. C., Gupta, S., Sangwan, K. S., and Shrestha, R.: Flood risk assessment in the Ganga Basin, India: A multi-criteria geospatial analysis with NASA’s Black Marble Nighttime light Data, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-15280, https://doi.org/10.5194/egusphere-egu24-15280, 2024.

Floods are widespread natural disasters on Earth affecting the planet with increasing frequency and intensity. Climate changes are responsible of the increasing number of heavy and persistent rains generating these destructive events often resulting in fatalities, injuries, and extensive infrastructural damages. A near real time monitoring system able to provide timely and accurate information about location and extent of the flooded areas is crucial for the authorities to implement the right mitigation actions. Currently, the Copernicus Emergency Management Service (CEMS) supports at European level the crisis management activities in the immediate aftermath of a flood, exploiting multi-source satellite data to provide flood delineation with a release time ranging from 7 to 48 hours (from the satellite acquisition). Map characterization and relative information are retrieved through semi-automatic or manual methodologies which do not allow for a complete automation of the analysis crucial to speed up the procedure and shorten the release time.

In this study, carried out in the framework of the MITIGO project (funded by MIUR PON R&I 2014-2020 Program), results coming from a multi-temporal optical satellite technique able to quick detect and accurately map flooded areas will be presented. This technique, namely RST-Flood, exploits the statistical characterization of the satellite observed signal to retrieve accurate background information useful to promptly and automatically identify ground changes directly linked to events occurrence. RST-Flood has already been successfully implemented with mid-low spatial resolution (from 1000 to 375m) optical satellite data sensors (i.e., Advanced Very High Resolution Radiometer, Moderate Resolution Imaging Spectroradiometer, Visible Infrared Imaging Radiometer Suite), and here is for the first time exported to Sentinel-2 Multi Spectral Instrument (MSI) data at mid-high spatial resolution (20m) to study recent floods events. The achieved results demonstrated the easy implementation of RST-Flood to different sensors and geographic areas and its capability in providing fast (processing time less than 15 min from data availability) and robust mapping of flooded areas. Furthermore, its design developed to work in the Google Earth Engine (GEE) environment makes it suitable for global scale implementation without altering its performance.

How to cite: Satriano, V., Ciancia, E., Pergola, N., and Tramutoli, V.: Floods automatic rapid mapping through Sentinel-2 MSI multitemporal data, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-15446, https://doi.org/10.5194/egusphere-egu24-15446, 2024.

Timely and accurate flood map production plays a key role in enhancing effective flood risk assessment and management. Satellite imagery is frequently employed in flood mapping as it can capture flooding across vast spatial and temporal scales. Floods are usually caused by prolonged or heavy precipitation correlated with dense cloudy conditions, posing challenges for accurate mapping.

The Synthetic Aperture Radar (SAR) active sensor is a popular option due to its feature of being weather agnostic, penetrating through clouds, fog, and darkness, providing images for the detection of flooded areas regardless of the weather conditions. However, this advantage is at the expense of low temporal resolution and double bounces in urban and heavily vegetated areas, which increase signal processing difficulty and misinterpretation. Passive microwave radiometry has also been explored for flood mapping, but its coarse spatial resolution limits the utility of the resulting flood maps. Multispectral optical imagery offers a balanced trade-off between temporal and spatial resolutions, with the only limitation that the acquired images might be hindered by the presence of clouds. Capitalizing on the utility of optical imagery, FloodSENS, a machine-learning (ML) algorithm consisting of a SENet and UNet, precisely delineates flooded areas from non-flooded areas in clear and partially clouded optical imagery. Although the current algorithm version enforces flood delineation involving topography-derived information in the ML processing, it is not capable of detecting floods under clouds; thus, we propose a new iteration of FloodSENS that utilizes auxiliary data in post-processing to improve the inferred flood maps.

The post-processing pipeline utilizes the inferred flood map generated by FloodSENS and the Digital Elevation Model (DEM) of the target area to accurately delineate the flood extent beneath clouds, adhering to the physical constraints in the topography. First, Pixels at elevations equal to or lower than the water level are designated as flooded pixels. These pixels are further refined with geoprocessing to establish hydrological connectivity and topographic consistency. Pixels that are both marked as flooded and hydrologically connected are confirmed as flooded pixels for the final flood map.

The post-processing proves essential in tropical and subtropical regions that frequently have high cloud cover during the monsoon seasons, making it imperative to map the affected areas during flooding events. The FloodSENS detection with the post-processing pipeline has been tested on partly clouded optical imagery obtained from the 2023 autumn flooding in southern Somalia.

How to cite: Campo, C., Tamagnone, P., Gallion, G., and Schumann, G.: Flood Segmentation with Optical Satellite Images Under Clouds Using Physically Constrained Machine Learning, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-16923, https://doi.org/10.5194/egusphere-egu24-16923, 2024.

This research aims to simplify and enhance the analysis and visualization of flood-prone areas in Norwegian rivers, with a primary emphasis on Sokna river. Utilizing remote sensing and GIS analysis, our objective is to advance flood risk assessment and management by integrating hydraulic data from numerical models, remotely sensed geomorphic features, and publicly available natural hazard maps. In this study, we develop GIS models, analyze geomorphic features related to erosion and deposition processes, and optimize flood risk analysis using hydro-morphodynamic indicators such as shear stress, stream power, Froude number, Shields formula, and the Hjulström diagram.

To locate flood-prone areas and estimate their severity, different influencing factors to flood risk were identified, among them fluvial dynamics, terrain characteristics, land use, and anthropic activities. Within the 12.65 km lowermost reach of Sokna river, near its confluence with Gaula river and Lundamo urban area, we conducted a comprehensive analysis of the geomorphological features (e.g. river width, soil type), natural hazards maps, and anthropic footprint (i.e. land use, infrastructure, safety measures), supported by hydrodynamics information from HEC-RAS models. Special attention was given to the analysis of sediments, erodible materials, and land use along the riverbanks while integrating flood areas with return periods ranging from 10 to 500 years, as well as other natural hazards such as rockfall- and snow erosion and deposition areas, avalanche records, landslides, debris, and quick clay landslide areas.

A temporal analysis was conducted using orthophotos from 1956, 2011, and 2021. The river channels in these orthophotos, captured in the same month to ensure similar discharges, were digitized to assess changes in river width and deposition processes. Additionally, DEM of Differences (DoD) supported refining documented river changes. The erodible sediment particle size was estimated using the Shields formula based on HEC-RAS model outputs, including Froude number, shear stress, and stream power. The erodible fraction was plotted into Shields and Hjulström diagrams and compared with the soil map. Identified locations with erodible material were complemented with land use data and other anthropic activities. Vulnerable infrastructure to erosion and deposition processes, such as culverts and bridges, were considered in flood risk assessments, with areas having safety measures (such as channel embankments) marked as having lower flood risk.

The step-by-step workflow, integrated into a GIS model using the Model Builder feature in ArcGIS Pro, is replicable for other rivers. These findings provide insights into the factors influencing flood risk, including potential erosion areas, the impact of natural hazards, and the temporal evolution of river channels. This methodology serves as a versatile tool for flood risk assessment and management in other river systems, contributing to the broader field of fluvial geomorphology and hydraulic engineering.

How to cite: Moraru, A., Ahmed, R., Tekle, M. G., Alfredsen, K., and Bruland, O.: Integrated GIS analysis for flood risk assessment in Norwegian Rivers: a case study of Sokna river, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-19378, https://doi.org/10.5194/egusphere-egu24-19378, 2024.

Remote sensing (RS)-derived Flood Inundation Maps (RS-FIM) are, in principle, a desirable source of observed data for the development, calibration, and validation of (model) predicted FIM. Advantages of using RS-FIM for evaluating predicted FIM include its spatial continuity (compared to point observations), and the representation of real flooding events (compared to synthetic events (e.g. 100-yr) or other models). Disadvantages may include low/mismatched spatial resolution, insufficient classification accuracy, lack of water depth information, and gaps in coverage (due to dense vegetation, buildings, clouds, etc.). Gaps in inundation coverage are very common in RS-FIM. While these may not be a major issue for some RS-FIM applications, they are a major, yet unacknowledged, issue for fair and robust evaluation of predicted FIM. This is because the evaluated model may correctly predict flooding in these gaps while the (RS-FIM) benchmark data is classified as non-flooded (leading to inaccurate identification of 'False-positives'). Techniques for 'closing the gaps' in RS-FIM using hydraulic models or terrain analysis can yield improved FIM but, depending on the scale of the 'gap-filling', can result in an RS-model hybrid which undermines the observational nature of RS-FIM. Here we will demonstrate and discuss the challenges in using RS-FIM for the evaluation of predicted FIM and present tools and analysis demonstrating a new framework for fair and robust evaluation of FIM predictions using RS-FIM.

How to cite: Cohen, S., Tian, D., Baruah, A., Liu, H., and Nikrou, P.: Use of Remote Sensing Flood Inundation Maps (FIM) for Evaluating Model-predicted FIM: Challenges and Strategies, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-2941, https://doi.org/10.5194/egusphere-egu24-2941, 2024.

Despite the increase in the number of Earth Observation satellites with active microwave sensors suitable for flood mapping, the frequency of observations still limits adequate characterization of inundation dynamics. Particularly, capturing the flood peak or maximum inundation extent, still remains elusive and a major research gap in the remote sensing of floods. Rapidly growing archives of multimodal satellite hydrology datasets combined with the recent deep learning revolution provide an opportunity to solve this problem adequate observation frequency. DeepFuse is a scalable data fusion methodology, leveraging deep learning (DL) and Earth Observation data, to estimate daily flood inundation at scale with a high spatial resolution. In this proof-of-concept study, the potential of Convolutional Neural Networks (CNN) to simulate flood inundation at the Sentinel-1 (S1) spatial resolution is demonstrated. Leveraging coarse resolution but temporally frequent datasets such as soil moisture/accumulated precipitation data from NASA’s SMAP/GPM missions and static topographical/land-use predictors, a CNN was trained on flood maps derived from S1 to predict high-resolution flood inundation. The proposed methodology was tested in southwest France at the confluence of its two main rivers, Adour and Luy, for the December 2019 flood event. The predicted high-resolution maps were independently evaluated against flood masks derived from Sentinel-2 using the Random Forest Classifier. First results confirm that the CNN can generalize some hydrological/hydraulic relationships leading to inundation based on the provided inputs, even for some rather complex topographies. However, further tests in catchments with strongly divergent land-use, hydrological, and elevation profiles is necessary to evaluate model sensitivity towards different land surface conditions. Achieving daily cadence for flood monitoring will enable an improved understanding of spatial inundation dynamics, as well as help develop better parametric hazard re/insurance products to effectively bridge the flood protection gap.

How to cite: Dasgupta, A., Sahu, R., Hybbeneth, L., and Waske, B.:  DeepFuse: Towards Frequent Flood Inundation Monitoring using AI and EO, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-20702, https://doi.org/10.5194/egusphere-egu24-20702, 2024.

During the winter of 2023 in the Southern Hemisphere, Central Chile experienced two extreme rain events, impacting an area where over 70% of the country's population resides. The initial rain event transpired in late June, accumulating over 700 mm within five days. This led to significant damage to critical infrastructure, isolating towns and villages, particularly in the Mataquito basin. Subsequently, a second rain event in August affected the southern part of central Chile, specifically the Maule basin, mirroring the impacts observed in Mataquito. Given the infrastructure affectation and the need to mitigate risks it is imperative to identify flood-affected areas. This task is complicated considering the lack of in-situ measurements, but of increasing relevance considering the heightened likelihood of more intense rain events attributable to climate warming. Here, we employed a suite of satellite remote sensors, utilizing Google Earth Engine for analysis. Our approach integrated active and passive sensors, employing change detection algorithms such as Spectral Euclidean Distance (SED) and Spectral Angle Measurement (SAM) on Sentinel-2 images. Additionally, we employed supervised classification, utilizing the random forest algorithm to train the model for flood detection. To enhance accuracy, we calibrated our Sentinel-1 GRD results to the flood-identified areas in Sentinel-2 imagery. Our findings revealed a notable correlation between the identified flooded areas and in-situ measurements. Surprisingly, SAM demonstrated high accuracy in identifying flooding areas with elevated values, whereas SED presented a broader range of results that were often inconclusive, however, their use is limited by cloud presence. Despite these limitations, the calibration of active remote sensors using detected floods as ground truth proved valuable in establishing change thresholds for future rain events in similar regions.

How to cite: Olivares, L. and González, G.: Flooding in Central Chile due to two extreme rain events: Insights from satellite remote sensors, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-21882, https://doi.org/10.5194/egusphere-egu24-21882, 2024.

Continuous megacities' development, aging infrastructure, and increasing frequency and magnitude of extreme events, the lack of flood resilience becomes a pressing issue due to inadequate planning of existing hydraulic structures to handle future threats. A more resilient urban flood risk management strategy is required to efficiently mitigate the impacts of climate change, particularly floods resulting from river and urban channel overflows. This is evident within the Aricanduva River watershed area in the east zone of São Paulo City, Brazil, a region with flood challenges arise because existent hydraulic infrastructures are ineffective in inundation control, due to extensive urbanization in the lower and middle parts of the basin. To achieve resilience in urbanized areas and reduce the risk of flash floods, the development of Early Warning Systems (EWS) is crucial. An EWS serves as a predictive tool for accurately forecasting water levels in rivers or channels in real-time, providing enough time to take action in order to reduce potential risk. Hydrologic-hydrodynamic models are increasingly employed in EWS to enhance their effectiveness. However, many urban basins lack monitoring systems, whereas products such as meteorological radar represent a feasible option since they effectively capture the spatial and temporal distribution of rainfall. In urban basins like the Aricanduva River, where the quantity and distribution of pluviometers are insufficient to spatially represent an event, the use of Quantitative Precipitation Estimation (QPE) from meteorology radar becomes essential to improve hydrological-hydrodynamic analyses. The objective of this work is to propose the presentation of a distributed hydrological-hydrodynamic model (HydroPol2D) for the Aricanduva basin, calibrated with QPEs from meteorological radar. Additionally, rainfall data from 15 gauges within and around the basin were utilized, covering a 5-year period, to generate spatial rainfall using Inverse Distance Weighted (IDW) interpolation. The results of the two rainfall databases were compared using metrics such as the Nash-Sutcliffe efficiency index, Efficiency of Kling-Gupta (KGE) index, and the percentage of bias to assess model accuracy. The findings indicate that (i) the distributed model coupled with QPEs produces favorable results and better represents the basin's dynamics, (ii) the model accurately reflects the hydraulics of existing flood control infrastructure within the basin, and (iii) the generation of an accurate and rapid rainfall-runoff model forms the initial steps in identifying risk areas, establish critical points for the early warning systems and analyzing the factors contributing to or generating the risk. The next step of this work is to assess the model with more events and to include in the model strategies to automate flow control in existing flood control infrastructures.  

Keywords: Urban flooding risk management, Early Warning Systems (EWS), Hydrological-hydrodynamic models, Radar Quantitative Precipitation Estimation (QPE), Climate Change.

How to cite: Hernandez Sanchez, M., Castillo Rapalo, L. M., Silva, P. G., and Mediondo, E. M.: Initial steps towards implementation of an early warning system with distributed hydrologic-hydrodynamic modeling for an urban basin with quantitative precipitation estimation (QPE) from meteorology radar., EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-13550, https://doi.org/10.5194/egusphere-egu24-13550, 2024.

In July 2021, the northwestern European continent experienced a devastating flood caused by unusually high rainfall, resulting in significant socio-economic destruction. One of the areas that was highly affected by this flooding was Southern Limburg (Geul River) in the Netherlands. The Geul River, located between Belgium, Germany, and the Netherlands, posed a challenging situation for modeling the catchment due to its cross-boundary nature. The need to harmonize input datasets from different countries with varying characteristics arose despite the abundance of available data in the study area. This study assesses the feasibility of combining multi-national Digital Elevation Models (DEM) for cross-boundary flash flood modeling purposes.

The quality of the DEM significantly impacts the accuracy of flood dynamics. However, it should be noted that elevation data from various sources creates elevation mismatches, particularly in the overlapping areas between different DEMs. A comprehensive quality assessment is indispensable to ensure the compatibility and reliability of these datasets for hydrology and flood modeling. Thus, To evaluate the accuracy of the DEMs, various statistical measures such as Root Mean Square Error (RMSE), Mean and Standard Deviation (STD) have been calculated. Initially, a pixel-by-pixel-based elevation difference map was generated. Upon analysis, it was observed that the overall elevation differences ranged from -7.0 to 7.0 meters. Despite certain pixels exhibiting pronounced differences in elevation, the overall statistical analysis indicated minimal variation. The calculated RMSE and STD values were both ≤ 0.3 meters for the overlapping parts of the DEM. These errors were considered negligible in relation to the actual slope values and had no significant impact on the flow direction within the catchment. This merged dataset provides a comprehensive representation of the terrain, enabling more accurate and reliable flood modeling simulations compared with calibrated result.

How to cite: Mohammed, M. J.: Applicability Of Multi-National Digital Elevation Model (DEM) For Cross-Boundary Flash Flood Modeling, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-1789, https://doi.org/10.5194/egusphere-egu24-1789, 2024.

While access to discharge data is key to hydrologic studies, it is a serious obstacle in ungauged basins. Currently, Sentinel-2 imagery at high spatiotemporal resolution offers a unique opportunity to infer the relation between pixel-based discharge rate and surface reflectance. One promising approach in this respect has been to find the complex relationship between river discharge and the spectral ratio between two benchmark pixels, namely the wet and dry pixels, whose dynamics resembles river discharge variation. However, this has been challenging due to the adverse impact of soil moisture and mixed land cover on the spectral behavior of the dry pixel. The selection of the wet pixel must also guarantee sufficient sensitivity of its spectral response to water depth fluctuations. To tackle the above issues, in this study, we developed a novel framework that automatizes the selection of the wet and dry pixels by using Sentinel-2 imagery. We also introduced the Normalized Difference Discharge Index (NDDI), as the best band combination, to predict river discharge. We used linear regression with leave-one-out cross-validation as the prediction model, which leverages limited satellite data due to the cloud cover. By implementing the developed framework at multiple gauged points across the continental United States, the best location of the dry pixel was consistently found in urban pixels whose longwave reflectance fall within a certain range. By analyzing the pixel-wised correlation coefficient between surface reflectance at NIR band and river discharge across the studied river widths, we found that the best wet pixels are located along river banks with shallow water depth. These pixels were characterized by the average reflectance higher than the 98th percentile in the green band. Finally, by testing over 4000 band combinations as input to the river discharge prediction model, we found that the normalized difference between B11 and B4 for the wet pixel, as well as the B11 ratioing between the dry and wet pixels yielded the most accurate predictions with R2 = 0.88 and R2 = 0.73, respectively.

How to cite: Tayebi-Alashti, A. and Danesh-Yazdi, M.: A novel framework for the selection of spectral input to pixel-based river discharge estimation model using Sentinel-2 imagery, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-5438, https://doi.org/10.5194/egusphere-egu24-5438, 2024.

Digital Elevation Models (DEMs)are used for extracting the geomorphological attributes of catchments. These attributes play crucial role in determining the hydrological responses of the catchments. However past research has highlighted the sensitivity of the geomorphological attributes to various DEM sources as well as special resolutions. This study is envisaged to assess the impact of different DEM sources and DEM resolutions on geomorphological attributes proposed by Moussa (2008) on Upper Yamuna River Basin which flows in 5 states (Uttarakhand, Himachal Pradesh, Uttar Pradesh, and Haryana) of India. For this purpose, different DEM sources, Shuttle Radar Topography Mission (SRTM), Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER), and Advanced Land Observing Satellite (ALOS) Phased Array type L-band Synthetic Aperture Radar (PALSAR) DEMs are used. To investigate the impact of DEM spatial resolution, 5 different resampled scenarios (grid size 90m, 120m, 150m, 180m, 210m) for each source of DEM was considered using the SRTM 30m DEM as a base DEM. The comparative assessment revealed notable discrepancies in the derived attributes among the DEMs of different resolutions and sources. The evaluation of variation in geomorphological attributes derived from various DEM sources and resolutions, yielded insightful observations. Furthermore, variations were observed between the different satellite sources, highlighting inherent differences in elevation data acquisition and processing methodologies. These findings underscore the critical influence of spatial resolution and data source on the accuracy and reliability of geomorphological attributes derived from DEMs, emphasizing the significance of careful consideration in selecting DEMs for terrain analysis and related applications.

How to cite: Rana, S. and Chavan, S. R.: Assessing the sensitivity of geomorphological attributes to DEM source and Spatial Resolutions , EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-16998, https://doi.org/10.5194/egusphere-egu24-16998, 2024.