Floods are one of the most common  natural disasters, with a disproportionate impact in developing countries that often lack dense streamflow gauge networks. Accurate and timely warnings are critical for mitigating flood risks, but hydrological simulation models typically must be calibrated to long data records in each watershed. Here we show that AI-based forecasting achieves reliability in predicting extreme riverine events in ungauged watersheds at up to a 5-day lead time that is similar to or better than the reliability of nowcasts (0-day lead time) from a current state of the art global modeling system (the Copernicus Emergency Management Service Global Flood Awareness System). Additionally, we achieve accuracies over 5-year return period events that are similar to or better than current accuracies over 1-year return period events. This means that AI can provide flood warnings earlier and over larger and more impactful events in ungauged basins. The model developed in this paper was incorporated into an operational early warning system that produces publicly available (free and open) forecasts in real time in over 80 countries. This work highlights a need for increasing the availability of hydrological data to continue to improve global access to reliable flood warnings.

Nearing, Grey, et al. "AI Increases Global Access to Reliable Flood Forecasts." arXiv preprint arXiv:2307.16104 (2023).

How to cite: Nearing, G., Cohen, D., Dube, V., Gauch, M., Gilon, O., Harrigan, S., Hassidim, A., Klotz, D., Kratzert, F., Metzger, A., Nevo, S., Pappenberger, F., Prudhomme, C., Shalev, G., Shenzis, S., Tekalign, T., Weitzner, D., and Matias, Y.: AI Increases Global Access to Reliable Flood Forecasts, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-4435, https://doi.org/10.5194/egusphere-egu24-4435, 2024.

It is essential that we work towards better preparation for flooding, as the impacts and risks associated increase with a changing climate. Standard methods for flood risk assessment are typically static, based on flood depths corresponding to return levels. In contrast flood risk changes over time, with the time of day and weather conditions, driving the location and extent of potential debris (e.g. vehicles or trees may cause blockages in culverts) affecting the associated risks. To this end, we aim to provide a platform for dynamic flood risk assessment, to better inform decision making, allowing for improved flood preparation at a local level. With stakeholder collaboration at a local level, a web-platform demonstrator is presented, for the city of Newcastle upon Tyne (U.K.) and the wider catchment, providing interactive visualisations and dynamic flood risk maps.

To achieve this, near real-time updates are incorporated as part of a fully integrated workflow of models, with traditional datasets combined with novel, hidden data. More realistic high-resolution data, citizen science data and novel data sources are combined, making use of data scraping and APIs to obtain additional sensor data. Using machine learning methods, more complex datasets are generated, using artificial intelligence algorithms and object detection to identify potential debris information from satellites, LIDAR point clouds and trash screen images. The model framework involves hyper-resolution hydrodynamic modelling (HIPIMS), with a hydrological catchment model (SHETRAN), working towards a digital twin.

How to cite: Green, A., Lewis, E., Tong, X., Wang, S., Smith, B., and Fowler, H.: PYRAMID: A Platform for dynamic, hyper-resolution, near-real time flood risk assessment integrating repurposed and novel data sources, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-14944, https://doi.org/10.5194/egusphere-egu24-14944, 2024.

The FURBAS project (Forecasting urban floods and strong rainfall events, 2022-2025) aims at establishing a real-time pluvial flood forecast system for the city of Hanover in Germany with a total catchment area of 260 km². The forecast system consists of radar rainfall with nowcasting  and an Artificial Neuronal Network (ANN) to forecast the dynamical evolution of surface water levels in the city with a spatial resolution of 5x5 meter and a temporal resolution of 5 minutes (Berkhahn & Neuweiler, 2023). The ANN is trained based on results of a physically based hydrodynamic model (HYSTEM EXTRAN 2D) with bi-directional coupled storm-sewer (1d) to surface (2d) domain.

The terrain has a slight gradient which causes a long travel time in the sewer system. The pipe network is partially built as combined sewer and separate sewer system. In the latter, rainfall is partially routed via trenches that are modelled as part of the 2d surface. These trenches are required to get an impression of the overall extent of the flooding and the interaction with the sewer system. In classical modelling approaches the drainage volume vanishes when reaching the outlets of the pipe network which leads to an underestimation of the flood level. We show the effects of modelling trenches as elements of the surface and provide tips for the correct arrangement of trenches.

The underground transport tunnels of a city are a rarely modelled factor in the flooding of a city. The underground stations are connected to the public drainage system where it can lead to a drainage delay, due to underground storage tanks. Flooding or overload of the drainage pipes along the underground tunnels have an important influence on the operability of the trains because the inflowing water is pumped into the regular drainage network. The fill level of the pipes is therefore the decisive limit value. We show the effects of station entrances and tunnel ramps to the water level at the surface and how the precipitation intensity is decisive for the operation of the trains.

Funding:
The FURBAS research project is a cooperation of the Institute for Technical and Scientific Hydrology (itwh) GmbH, the Institute of Fluid Mechanics and Environmental Physics in Civil Engineering, Leibniz University of Hanover, and the municipal operation for Hanover city drainage. The project is funded by the German Federal Ministry for the Environment, Nature Conservation, Nuclear Safety and Consumer Protection under grand number 67DAS224.

Literature:
Berkhahn, S., & Neuweiler, I. (2023). Data driven real-time prediction of urban floods with spatial and temporal distribution. Journal of Hydrology X, 100167.

How to cite: Sämann, R., Treindl, J., Fuchs, L., Beeneken, T., and Berkhahn, S.: Influence of trenches and subway system on pluvial flood forecast , EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-18526, https://doi.org/10.5194/egusphere-egu24-18526, 2024.

The increasing frequency and intensity of floods pose a significant threat to lives, property, and infrastructure. Real-time flood forecasting is crucial for early warning systems and disaster risk reduction. However, traditional forecasting methods often have limitations in terms of accuracy and timeliness.

This paper, developed under the framework of AI4Copernicus 5th Calls project, presents a data-driven approach for real-time flood and water level forecasting using AI and machine learning algorithms. The proposed system is based on a hybrid model that combines multiple machine learning algorithms, including DLinear/NLinear, LSTM Hindsight Modelling, and FLEX. The system is trained on historical data on hydrological and meteorological features, and is able to predict water levels at river gauging stations up to the next 9 hours.

The system has been tested on data from the Lamone River in Italy, and has been shown to achieve a mean-absolute-error of only a few (<5) centimetres. This is a very low error margin for this kind of river, and is comparable to or better than the performance of other alternative forecast approaches.

The system has been integrated into GECOSistema's Flood Risk Intelligence platform, named SaferPlaces (www.saferplaces.co). This platform provides a user-friendly interface for accessing flood risk information, and includes features such as real-time flood maps, early warning alerts, and detailed flood risk assessments.

The proposed system has the potential to be a valuable tool for flood forecasting and disaster risk reduction. It can be used to support decision-making at both the local and regional levels, and can help to save lives and property.

How to cite: Bagli, S., Mazzoli, P., van der Brink, K., luzzi, V., and papa, M.: Real-time flood and water level forecasting using AI-based models for early warning and disaster risk reduction, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-3187, https://doi.org/10.5194/egusphere-egu24-3187, 2024.

This presentation aims to offer valuable insights into the state-of-the-art technologies and methodologies employed to enhance Real-time Flood Forecasting (RTFF) systems by highlighting recent progress in the development of high-resolution RTFF systems. Given the existing uncertainties associated with RTFF and the documented increasing trends in flood peaks in the western and central regions of Europe [1], continuous improvement of RTFF systems is essential for better disaster risk preparedness.

Recurrent and severe flooding events have impacted Germany in recent years. The predictability of two recent flooding events, including the extensive flooding from December 2023 to January 2024 across the entire country, and the 2021 summer flood in Ahrtal [5] will be explored by introducing an experimental RTFF chain. This chain utilizes high-resolution weather forecasts from Germany's National Meteorological Service, Deutscher Wetterdienst (DWD). The chain incorporates high-resolution streamflow and water level forecasts at 1 km using the mesoscale Hydrologic Model (mHM) [2,3]. Additionally, it features a fast hydrodynamic model (RIM2D) at 10 m resolution [4] with an extended component for impact forecasting tailored to the scale of individual buildings. We showcase how the newly developed RTFF system enables tailored decision-making compared to the common practices currently used by local authorities.

References

[1] Blöschl, G., Hall, J., Viglione, A., Perdigão, R. A., Parajka, J., Merz, B., ... & Živković, N. (2019). Changing climate both increases and decreases European river floods. Nature, 573(7772), 108-111. DOI: 10.1038/s41586-019-1495-6

[2] Samaniego, L., Kumar, R., & Attinger, S. (2010). Multiscale parameter regionalization of a grid‐based hydrologic model at the mesoscale. Water Resources Research, 46(5).

[3] Samaniego, L., Kumar, R., Thober, S., Rakovec, O., Zink, M., Wanders, N., ... & Attinger, S. (2017). Toward seamless hydrologic predictions across spatial scales. Hydrology and Earth System Sciences, 21(9), 4323-4346.

[4] Apel, H., Vorogushyn, S., & Merz, B. (2022). Brief communication: Impact forecasting could substantially improve the emergency management of deadly floods: case study July 2021 floods in Germany. Nat. Hazards Earth Syst. Sci., 22(9), 3005-3014. doi:10.5194/nhess-22-3005-2022.

[5] Najafi, H., Shrestha, PK., Rakove, O.,  Apel, H., Thober, S., Kumar, R., Vorogushyn, S., & Merz, B., Samaniego, L. (in review). Advancing a High-Resolution Impact-based Early Warning System for Riverine Flooding. Nature communications.

How to cite: Najafi, H., Rakovec, O., Kumar Shrestha, P., Kumar, R., Vorogushyn, S., Apel, H., Thober, S., Merz, B., and Samaniego, L.: Recent advances in developing high-resolution flood forecasting systems in Germany, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-13854, https://doi.org/10.5194/egusphere-egu24-13854, 2024.

Natural disasters cause extensive losses worldwide annually. Flood events are responsible for economic and life-threatening damages[1]. To mitigate flood risks and resulting damages, particularly in the construction of residential buildings, two approaches exist. First: constructing in areas with lower flood susceptibility, and second: implementing architectural solutions to fortify structures against floods and associated hazards. Due to the presence of water resources, rivers, etc., prompting urban expansion due to reasons like transportation, trade, agricultural use, household consumption, etc., construction near rivers and flood-prone areas becomes inevitable[2]. This underscores the importance of the second approach—architectural fortification.

In this study, areas highly susceptible to flooding were identified from flood zoning maps using artificial intelligence to adapt these maps and estimate the most hazardous regions[3]. Subsequently, by examining the specific elements of traditional architecture in each of these areas and exploring the cause and function of each element in facing floods over time, attention is given to the particular and regional (indigenous) architectural features that have responded to floods. Finally, appropriate architectural measures and responses to reduce flood risks, such as constructing at elevation or suitable gradients, is combined with early warning systems to provide a proper route for the future construction projects.

Keywords: Flood forecasting; Flood prone areas; Architectural fortification

[1] Naghedi, S.R., Huang, X. and Gheibi, M., 2023. A smart dashboard for forecasting disaster casualties: An investigation from sustainable development dimensions (No. EGU23-17237). Copernicus Meetings.

[2] Yan, J., Naghedi, R., Huang, X., Wang, S., Lu, J. and Xu, Y., 2023. Evaluating simulated visible greenness in urban landscapes: An examination of a midsize US city. Urban Forestry & Urban Greening, 87, p.128060.

[3] Vahid, R., Farnood Ahmadi, F., & Mohammadi, N. (2021). Earthquake damage modeling using cellular automata and fuzzy rule-based models. Arabian Journal of Geosciences, 14, 1-14.

How to cite: Naghedi, S. N., Maleki, A., Vahid, R., Piadeh, F., and Behzadian, K.: Architectural Strategies for Flood Mitigation in Urban Environments: A Study of Traditional Elements and Contemporary Resilience, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-19111, https://doi.org/10.5194/egusphere-egu24-19111, 2024.

This study introduces a methodology for enhancing early-warning systems (EWS) for climate variables such as temperature, humidity, and precipitation. These systems are crucial for predicting hydrological extremes, including heat waves and floods. Traditional forecasting methods face challenges due to the complex nature of climate systems, limitations of global circulation models, and computational demands, often resulting in predictions with coarse spatial and temporal resolutions.

Our approach integrates advanced Machine Learning (ML) models with comprehensive data collection for global climate forecasting and regional downscaling. The methodology centers on the use of the ERA5 reanalysis dataset from the European Center for Medium-Range Weather Forecasts (ECMWF) and the CMCC dataset, which provides high-resolution climate data.

The core of our global forecasting relies on FourCastNet, a cutting-edge deep learning model developed by NVIDIA. Utilizing Fourier Neural Networks, FourCastNet excels in generating high-resolution global climate forecasts quickly and accurately. It offers a lead time of up to 96 hours for various atmospheric variables, with a specific focus on precipitation forecasts up to 36 hours ahead. This model’s ability to handle complex climate patterns makes it ideal for initial global forecasting.

For regional downscaling, we employ Stacked Super-Resolution Convolutional Neural Network (SRCNN) and Super-Resolution Generative Adversarial Network (SRGAN) models, which are trained on the CMCC dataset. This dataset contains dynamically downscaled ERA5 reanalysis and has a 2.2 km spatial resolution and a 6-hourly temporal resolution, matching the temporal resolution of FourCastNet outputs. This compatibility enables seamless linking of global and regional forecasts. The downscaling aims to increase spatial resolution by eight times, providing detailed local climatic insights.

All computational models and simulations are conducted on the Google Cloud platform. This platform provides the necessary computational resources, including GPUs, to handle the large-scale processing of climate datasets and the execution of complex ML models efficiently.

In summary, this methodology combines advanced ML models and detailed data collection from both ERA5 and CMCC datasets for both global forecasting and regional downscaling. This integrated approach aims to deliver accurate, high-resolution predictions of climate variables, significantly enhancing the capabilities of early-warning systems. The selection of suitable high-resolution datasets for training downscaling models is a key step, ensuring the generation of detailed regional forecasts.

How to cite: Shafei, A. and Cioffi, F.: Designing an Early-Warning System to Forecast Extreme Climate Conditions Using Data-Driven Approaches with Machine-Learning and Deep-Learning Methods, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-5526, https://doi.org/10.5194/egusphere-egu24-5526, 2024.

Urban flooding poses a significant challenge to cities, requiring the development of advanced predictive models to mitigate potential risks and enhance urban resilience. In the present study, we test an artificial neural network (ANN)-based model for predicting urban flooding in the city of Hanover, Germany. The model provides high-resolution spatial analysis on a 5 x 5 meter grid, providing detailed insights into potential flood-prone areas. With a temporal resolution of 5 minutes, the ANN model uses radar-based precipitation data to predict water levels during extreme weather events. The study is part of the FURBAS project (Forecasting urban floods and strong rainfall events, 2022-2025). This research project is a cooperation of the Institute for Technical and Scientific Hydrology (itwh) GmbH, the Institute of Fluid Mechanics and Environmental Physics in Civil Engineering, Leibniz University of Hanover and the municipal operation for Hanover city drainage. The project is funded by the Federal Ministry for the Environment, Nature Conservation, Nuclear Safety and Consumer Protection under grand number 67DAS224.

The data driven flood prediction model uses pre-simulated flood scenarios from a physically based model as training data. The model approach of Berkhahn and Neuweiler (2023) was adapted for the present study to cope with the large catchment area of about 260 km².

The proposed model could improve timely decision making for urban planning and emergency response in the future. Despite the focus on the specific challenges of the city of Hanover, the chosen modeling approach could also be applied to flood forecasting and management in other cities. With this conference contribution we want to highlight the challenges of real-time forecasting of pluvial urban floods in large catchments and present first preliminary results.

Berkhahn, S., & Neuweiler, I. (2023). Data driven real-time prediction of urban floods with spatial and temporal distribution. Journal of Hydrology X, 100167.

How to cite: Berkhahn, S., Sämann, R., Fuchs, L., and Neuweiler, I.: A High-Resolution Artificial Neural Network-Based Model for Predicting Urban Flooding in Hanover, Germany, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-17244, https://doi.org/10.5194/egusphere-egu24-17244, 2024.

Abstract

While the accuracy of flood predictions is likely to improve with increasing gauging station networks and robust radar coverage, challenges arise when such sources are spatially limited [1]. For instance, severe rainfall events in the UK come mostly from the North Atlantic area where gauges are ineffective and radar instruments are limited to it 250km range.  In these cases, NASA’s IMERG is an alternative source of precipitation estimates offering global coverage with 0.1-degree spatial resolution at 30-minute intervals. The IMERG estimates for the UK’s case can offer an opportunity to extend the zone of rainfall detection beyond the radar range and increase lead time on flood risk predictions [2].

This study investigates the ability of machine learning (ML) models to capture the patterns between rainfall and stream level, observed during 20 years in the River Crane in the UK. To compare performances, the models use two sources of rainfall data as input for stream level prediction, the IMERG final run estimates and rain gauge readings. Among the three IMERG products (early, late, and final), the final run was selected for this study due to its higher accuracy in rainfall estimates. The rainfall data was retrieved from rain gauges and the pixel in the IMERG dataset grid closest to the point where stream level readings were taken.

These datasets were assessed regarding their correlation with stream level using cross-correlation analysis. The assessment revealed a small variance in the lags and correlation coefficients between the stream-level and the IMERG dataset compared to the lags and coefficients found between stream-level and the gauge’s datasets. To evaluate and compare the performance of each dataset as input in ML models for stream-level predictions, three models were selected: NARX, LSTM, and GRU. Both inputs performed well in the NARX model and produced stream-level predictions of high precision with MSE equal to 1.5×10^-5 while using gauge data and 1.9×10^-5 for the IMERG data. The LSTM model also produced good predictions, however, the MSE was considerably higher,  MSE of 1.8×10^-3 for gauging data and 4.9×10^-3 for IMERG data. Similar performance was observed in the GRU predictions with MSE of 1.9×10^-3 for gauging data and 5.6×10^-3 for IMERG.  Nonetheless, the results of all models are within acceptable ranges of efficacy confirming the applicability of ML models on stream-level prediction based just on rainfall and stream-level information. More importantly, the small difference between the results obtained from IMERG estimates and gauging data seems promising for future tests of IMERG rainfall data sourced from other pixels of the dataset’s grid and to explore the potential for increased lead time of predictions. 

[1] Piadeh, F., Behzadian, K. and Alani, A. (2022). A critical review of real-time modelling of flood forecasting in urban drainage systems. Journal of Hydrology, 607, p.127476.

[2] Foelsche, U., Kirchengast, G., Fuchsberger, J., Tan, J., Petersen, W. (2017). Evaluation of GPM IMERG Early, Late, and Final rainfall estimates using WegenerNet gauge data in southeastern Austria. Hydrology and Earth System Sciences, 21(12), pp. 6559-6572.

How to cite: Girotto, C., Piadeh, F., Behzadian, K., Zolgharni, M., Campos, L., and Chen, A.: Machine learning models for stream-level predictions using readings from satellite and ground gauging stations , EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-13359, https://doi.org/10.5194/egusphere-egu24-13359, 2024.

The occurrences of floods in the recent past have significantly increased due to climate change and anthropogenic activities. Hence, reliable streamflow forecasts are crucial for minimizing the detrimental effects of flooding. However, forecast accuracy deteriorates besides elevated uncertainty when the lead time increases. Therefore, streamflow forecast should have improved accuracy with simultaneous uncertainty quantification to increase the model confidence for effective decision-making. The study proposes a novel two-stage multi-step dynamic error correction model to forecast up to 7 days ahead of streamflow, with the objective of no significant deterioration in accuracy. The framework is developed by integrating the process-based hydrological HBV model with the Bayesian-based Particle filter (PF) and machine learning-based Random Forest algorithm (RF). This facilitates combining the advantages of each model, i.e., process understanding ability of the HBV model, robust uncertainty quantifying ability of the PF technique, and relatively superior predictive ability of the RF algorithm. The model performance is quantified through several statistical performance error measures and uncertainty indices, with graphical performance indicators. The framework tested on the Beas and Sunkoshi river basins of India and Nepal exemplified the NSE of 0.94 and 0.98 in calibration and 0.95 and 0.99 in validation respectively for the 7-day ahead streamflow forecast. Hence, the proposed dynamic modeling framework can be considered as a potential tool to forecast streamflow without significant deterioration in the model accuracy even at increased lead times.  

How to cite: Roy, A., Kasiviswanathan, K. S., and Patidar, S.: A novel two-stage multi-step dynamic error correction model for improving streamflow forecast accuracy, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-315, https://doi.org/10.5194/egusphere-egu24-315, 2024.

Antecedent conditions play a crucial role in flooding, and hence it is essential to simulate them when floods are forecasted. Wet antecedent conditions can lead to significant flooding even if the rainfall during an event is not very intense, prolonged or widespread. An example of this type of flooding is the European floods in the summer of 2013, which affected Germany, Czech Republic and Austria; leading to 25 fatalities and financial losses amounting to 16 billion USD (Munich Re, 2013).

This study presents a modelling framework aimed focusing on the interplay between antecedent conditions and flood events. Our approach integrates a new component related to antecedent conditions to Previsico’s proprietary FloodMap Live by leveraging geospatial datasets as well as past precipitation data. The ground parameters are modified automatically without the need of manual intervention.

In this study we discuss the data processing spatially and temporally, and the impact of antecedent conditions for various events by showcasing different scenarios in the United Kingdom. Moreover, we investigate the model sensitivity and performance when compared with observation points which were flooded.

This research investigates the importance of antecedent conditions on flood modelling and  contributes to our understanding of how scenario-based events should be modelled in order to improve forecast performance. These improvements improve the accuracy of Previsico’s flood forecasts as they add a new component related to how the ground conditions changed a few days before a flood event.

Reference:

Munich Re, 2013. Floods dominate natural catastrophe statistics in first half of 2013. Available at: https://web.archive.org/web/20130714234357/http://www.munichre.com/en/media_relations/press_releases/2013/2013_07_09_press_release.aspx

How to cite: Tsaknias, D. and Yu, D.: Modelling the impact of antecedent conditions on flood forecasting, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-21875, https://doi.org/10.5194/egusphere-egu24-21875, 2024.

Accurate rainfall estimation is vital for real-time forecasting of water resources used for water demand management. This needs an optimum density of rain gauges. While numerous geostatistical methods exist for optimizing rain gauge networks, many may suffer from limitations. This study aims to develop a novel geostatistical method to redesign rain gauge networks that was demonstrated in a real-world case study of Khorasan Razavi province, specifically in the Qarahqoom basin, Iran, aiming to minimize errors.

The methodology involves analyzing the number and locations of rain gauges and assessing each rain station’s contribution to the region’s overall rainfall estimation accuracy. Initially, station homogeneity in the study area is verified using the linear moment method. Subsequently, a suitable semi-variogram is selected to calculate the acceptance probability for different areas within the province. This approach determines acceptance accuracy (AP) values at various probability levels.

Considering the basin’s characteristics, including its homogeneity, the acceptance probability method was implemented at an 80% probability level. The findings reveal that current networks of 66 rain gauges achieves a 61% acceptance accuracy. Of these, only 42 rain gauges significantly influence the estimated basin rainfall (i.e. forming the base network) while the remaining 24 rain gauges have a minor impact (i.e. non-base network). It is proposed that adding 24 strategically placed stations could evaluate the rainfall estimation accuracy in the Qaraqoom basin to 95%.

Keywords: Rain gauge network, variogram, acceptance probability, acceptance accuracy, Qarahqoom basin

How to cite: Nouriabouzari, H., Abbasi, A., Shafiei, M., and Behzadian, K.: Cumulative Impact Analysis of Rain Gauge Networks on Rainfall Forecasting Accuracy: A Case Study , EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-15264, https://doi.org/10.5194/egusphere-egu24-15264, 2024.

Accurate forecasting of water quality variables in river systems is crucial for relevant administrators to identify potential water quality degradation issues and take countermeasures promptly. However, pure data-driven forecasting models are often insufficient to deal with the highly varying periodicity of water quality in today’s more complex environment. This study presents a new holistic framework for time-series forecasting of water quality parameters by combining advanced deep learning algorithms (i.e., Long Short-Term Memory (LSTM) and Informer) with causal inference, time-frequency analysis, and uncertainty quantification. The framework was demonstrated for total nitrogen (TN) forecasting in the largest artificial lakes in Asia (i.e., the Danjiangkou Reservoir, China) with six-year monitoring data from January 2017 to June 2022. The results showed that the pre-processing techniques based on causal inference and wavelet decomposition can significantly improve the performance of deep learning algorithms. Compared to the individual LSTM and Informer models, wavelet-coupled approaches diminished well the apparent forecasting errors of TN concentrations, with 24.39%, 32.68%, and 41.26% reduction at most in the average, standard deviation, and maximum values of the errors, respectively. In addition, a post-processing algorithm based on the Copula function and Bayesian theory was designed to quantify the uncertainty of predictions. With the help of this algorithm, each deterministic prediction of our model can correspond to a range of possible outputs. The 95% forecast confidence interval covered almost all the observations, which proves a measure of the reliability and robustness of the predictions. This study provides rich scientific references for applying advanced data-driven methods in time-series forecasting tasks and a practical methodological framework for water resources management and similar projects.

How to cite: Zhang, C., Nong, X., Behzadian, K., Campos, L., and Shao, D.: A new framework for water quality forecasting coupling causal inference, time-frequency analysis and uncertainty quantification, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-7047, https://doi.org/10.5194/egusphere-egu24-7047, 2024.

Rainfall data sources constitute a vital component of flood early warning systems (EWS), and their inseparability from these systems is evident [1]. However, the information derived from these sources is typically confined to the duration, intensity and peak time for ground-based stations and cloud density and temperature for satellite productions [2]. Therefore, more details into the current rainfall occurrence and predictions regarding its future characteristics can significantly assist real-time flood forecasting systems to perform more accurate and reliable measures [3]. One of the rainfall characteristics that can bring valuable insight into the EWS are return period (RP) or position of rainfall into the intensity-duration-frequency (IDF) curves. This new parameter can offer a more nuanced understanding of rainfall events and significantly enhance the capabilities of early warning systems [4].

In this study, a novel Back Propagation Neural Network model is designed to enhance the accuracy of rainfall predictions in EWS. The model incorporates five rainfall inputs of (1) current Intensity, (2) intensity gradient determined from an intensity library, (3) current duration, (4) current RP determined using rules from the IDF curve library, (5) RP gradient, (6) absolute energy, and (7) anthropic class. The model employs two 5-neuron hidden layers to predict the RP class of current rainfall, i.e. a 5-year or 3-month RP for instance, depending on the desired lead time. To evaluate its accuracy, the model is tested for various time predictions with 15-minute intervals. Subsequently, a real case study of an urban drainage system in the UK is chosen to assess how this additional input enhances previously developed models [3-4].

The results demonstrate that the model excels in predicting the RP for a 2-hour lead time, achieving a performance accuracy exceeding 90%. Moreover, an acceptable accuracy rate of over 75% is achieved for a 4-hour lead time. Additionally, the incorporation of an added parameter into a benchmark EWS results in a 10.8% increase in accuracy for 15-min, escalating to 37.8% for 4-hour lead time. Although the influence of the added parameter may be minimal for near timesteps, its impact becomes significantly more pronounced when dealing with longer lead time predictions, exactly when conventional EWS performance tends to be reduced.

References

[1] Piadeh, F., Behzadian, K., Chen, A.S., Kapelan, Z., Rizzuto, J., Campos, L.C. (2023). Enhancing urban flood forecasting in drainage systems using dynamic ensemble-based data mining. Water Research, 247, p.120791.

[2] Piadeh, F., Behzadian, K., Chen, A.S., Campos, L.C., Rizzuto, J., Kapelan, Z. (2023). Event-based decision support algorithm for real-time flood forecasting in urban drainage systems using machine learning modelling. Environmental Modelling & Software, 167, p.105772.

[3] Piadeh, F., Behzadian, K., Chen, A.S., Campos, L.C., Rizzuto, J.P. (2023). Real-time flood overflow forecasting in Urban Drainage Systems by using time-series multi-stacking of data mining techniques, EGU General Assembly 2023, Vienna, Austria, EGU23-8574, https://doi.org/10.5194/egusphere-egu23-8574, 2023.

[4] Piadeh, F., Piadeh, F., Behzadian, K. (2023). Time-series Boosting in Ensemble Modelling of Real-Time Flood Forecasting Application, EGU General Assembly 2023, Vienna, Austria, EGU23-4183, https://doi.org/10.5194/egusphere-egu23-4183, 2023.

How to cite: Piadeh, F. and Piadeh, F.: Rule-based BPNN model for real-time IDF prediction of rainfall: Valuable Input for Early Warning Systems, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-10462, https://doi.org/10.5194/egusphere-egu24-10462, 2024.

Hydrological ensemble forecasting is crucial for regional or urban flood forecasting. The development of real-time hydrological ensemble forecasting methods and early warning systems that produce accurate and timely forecasts and flood warnings is of paramount importance for effective disaster risk reduction and the mitigation of loss of life.

However, the majority of current hydrological ensemble forecasting systems are centralized, requiring researchers to collect data, download executable programs for models and related methods, and configure the runtime environment on local computers based on specific scenarios (e.g., simulation and forecasting of a specific city or watershed). This method is extremely time-consuming and labour-intensive, and there is a high level of coupling between modelling resources such as data, models (or methods), and parameters. When researchers simulate other scenarios, the models used in certain hydrological processes may not be applicable to the new environment due to changes in the natural environment, and new models may need to be implemented (for instance, the models for runoff yield under saturated storage and runoff yield under excess infiltration conditions are distinctly different). Substantial amounts of time and effort must be invested in recollecting and deploying forecasting resources in local computer, which leads to repetitive labour. This involves downloading models, configuring the operating environment for each ensemble forecasting process, collecting pertinent data, compiling data and model adaptation methods, designing optimization schemes and evaluating the model based on results.

Therefore, to change the current complicated download and installation usage patterns associated with hydrological ensemble forecasting and to facilitate the seamless replacement and integration of various hydrological process model components, we propose an open online simulation strategy. This strategy utilizes a service-oriented web architecture to support the online sharing, invocation, integration, and optimization of simulation resources at the three perspectives: model, input data, and model parameters. Specifically, we explore (1) a service-oriented hydrological ensemble forecasting model sharing method and a document-based model service integration and management method, (2) a hydrological ensemble forecasting data sharing and Python-based data adaptation method, and (3) an online optimization and recommendation method for model parameters. By applying the strategy proposed in this paper to hydrological ensemble forecasting, it is possible to reduce the cost of using models, encourage the sharing of hydrological resources and the exchange of knowledge, and ultimately improve the accuracy of flood forecasting.

How to cite: He, Y. and Chen, M.: An open online simulation strategy for hydrological ensemble forecasting, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-3280, https://doi.org/10.5194/egusphere-egu24-3280, 2024.

Pluvial (flash) floods regularly cause significant damage in both rural and urban catchments. Such pluvial floods are usually caused by short-term local precipitation events of extreme intensity, resulting in infiltration excess and overland flow. In contrast to fluvial floods, the hazard of pluvial floods is mainly due to overland flow and flow in small ditches and creeks. Therefore, pluvial floods cannot be evaluated with common extreme value statistics, which are based on fluvial discharge records at river gages. On the other hand, pluvial floods are not only influenced by precipitation alone, but also by hydrological and hydrodynamic processes. Thus, precipitation statistics are not sufficient to evaluate and predict pluvial floods, either. Therefore, we suggest a new pluvial flood index (PFI), which evaluates the danger resulting from overland flow and surface flooding during pluvial floods and takes into account precipitation along with hydrological and hydrodynamic processes.

The new PFI is based on the proportion of pluvial flood hazard areas (PFHA). We define PFHA as areas where pedestrians or vehicles are at risk because water depth, flow velocity or the combination of both (flow rate) exceed defined thresholds. Based on historical events and design events (combining different probabilities of precipitation and initial soil moisture), we defined thresholds of PFHA to generate four classes of PFI ranging from no flood danger to very large flood danger. Hence, PFI is a simple, dimensionless measure, which can convey valuable information about the occurrence and severity of a pluvial flood to the general public and authorities.

PFHA and PFI for different events are determined from precipitation input, dynamic simulation of infiltration and saturation excess and hydrodynamic simulation of surface runoff concentration. Thus, simulation and forecasting PFI does not only require quantitative precipitation input and appropriate hydrodynamic overland flow models, but also adequate distributed, process-based hydrological models that consider infiltration excess and saturation runoff resulting from different initial soil moisture and land surface conditions. We demonstrate the application and usefulness of the new PFI in case studies for historical events and for a large-scale test area and show the potential using ML methods to allow real-time forecasting. We will also demonstrate that this information is much more valuable than rainfall warning alone. Moreover, the PFI can be linked to detailed local data to improve decision making of local municipalities. Therefore, the PFI is a valuable core piece for operational, real-time pluvial flood forecasting and early warning systems. The proposed system provides information on whether pluvial flooding will occur in a certain area on the scale of several hectares to square kilometers and how extreme this flood will be.

How to cite: Weiler, M., Haag, I., Hänsler, A., Krumm, J., Leistert, H., Schmit, M., and Steinbrich, A.: A new Pluvial Flood Index (PFI) considering meteorological, hydrological and hydrodynamic processes for real-time flash flood forecasting, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-11503, https://doi.org/10.5194/egusphere-egu24-11503, 2024.

Urban systems are highly exposed and vulnerable to extreme rainfall and flooding. Flood impacts span across various sectors, causing disruptions in transportation network, power supply, and access to emergency services. These impacts are expected to increase with expanding urban development, aging flood control infrastructure, and intensifying rainfall events. Reliable prediction of flood hazards is crucial to inform the design of sustainable risk management strategies. This study aims to advance predictive understanding of flood hazards by leveraging recent advances in numerical weather prediction, machine learning, satellite observations and high-performance computing. We compare the predictive skill of standalone machine learning with the hybrid models built by integrating process-based hydrodynamic model outputs with machine learning algorithms. We demonstrate the ability of machine learning surrogate models to capture spatio-temporal flood dynamics with reduced computational expense. This work contributes to strengthening the scientific foundation for flood-risk prediction that is of utmost importance to enhance community resilience in the face of evolving weather and climate extremes.

How to cite: Sharma, S., Bhatttarai, Y., Bista, S., and Talchabhadel, R.: Leveraging machine learning and satellite observation for skillful flood forecasts, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-14019, https://doi.org/10.5194/egusphere-egu24-14019, 2024.

Floods are one of the most frequent and severe natural disasters, and it is important to be prepared to predict them. Accurate prediction of floods requires the provision of accurate estimates of river discharge. Data assimilation (DA) as a technique for integrating background fields and observations can be a helpful solution to improve the accuracy of the river discharge prediction. DA can be a highly effective technology, however, when DA is performed on a large amount of data or high dimensional data, it results to be computationally very expensive, which is inappropriate for flood prediction, where timely results are required. Also, DA is used to merge data from diverse sources of information and, when the background fields and the observations are not from the same place, e.g. the observations are sparse, data must be interpolated on different grinds which increase the errors’ accumulation. In this work, latent neural mapping is designed to mitigate problems related to errors propagation and computational cost. We integrated DA with neural network (NN) and the resulting model helps on saving computational cost and solve the problem of sparse observation. Convolutional NN are employed to build a mapping function which converts data from the background space to the observation space (and vice versa). We tested the model with real data and flooding events in the UK. Data provided by the National River Flow Archive (NRFA) served as observations and the data provided by the European Flood Awareness System (EFAS) served as background fields. The Result shows that the accuracy is improved by 54.4% in MSE and the runtime of the model in 50s for 300 iterations. 

How to cite: Wang, K., Cheng, S., Piggott, M., Dance, S. L., and Arcucci, R.: Latent Neural Mapping for Hydrological Data Assimilation in Flood Prediction, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-20164, https://doi.org/10.5194/egusphere-egu24-20164, 2024.

Dissolved oxygen (DO) is an essential indicator for assessing water quality and managing aquatic environments, but it is still a challenging topic to accurately understand and predict the spatiotemporal variation of DO concentrations under the complex effects of different environmental factors. In this study, a practical prediction framework was proposed for DO concentrations based on the support vector regression (SVR) model coupling multiple intelligence techniques (i.e., four data denoising techniques, three feature selection rules, and four hyperparameter optimization methods). The holistic framework was tested using a data matrix (17532 observation data in total) of 12 indicators from three vital water quality monitoring stations of the longest inter-basin water diversion project in the world (i.e., the Middle-Route of the South-to-North Water Diversion Project of China), during the year 2017 to 2020 period. The results showed that the framework we advocated for could successfully and accurately predict DO concentration variations in different geographical locations. The model used the “wavelet analysis–LASSO regression–random search–SVR” combination of the Waihuanhe station has the best prediction performance, with the Root Mean Square Error (RMSE), Mean Square Error (MSE), Mean Absolute Error (MAE), and coefficient of determination (R2) values of 0.251, 0.063, 0.190, and 0.911, respectively. The combined methods using feature selection and hyperparameter optimization techniques can significantly promote the robustness and accuracy of the prediction model and can provide a new universal and practical way for investigating and understanding the environmental drivers of DO concentration variations. For the water quality management department, this proposed comprehensive framework can also identify and reveal the key parameters that should be concerned and monitored under different environmental factors change. More studies in terms of assessing potential integrated water quality risk using multi-indicators in mega water diversion projects and/or similar water bodies are required in the future.

How to cite: Nong, X., Lai, C., Chen, L., and Wei, J.: Prediction modelling framework comparative analysis of dissolved oxygen concentration variations using support vector regression coupled with multiple feature engineering and optimization methods: A case study in China, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-7941, https://doi.org/10.5194/egusphere-egu24-7941, 2024.

This research employs a systems dynamics approach to simulate the intricate dynamics of the agricultural sector, a predominant consumer of resources, with a focus on the Khuzestan region. By meticulously reviewing reports and conducting field visits, we extracted essential inputs for both hydrological and agricultural models. Our objective was to formulate the behaviors of hydrology and agriculture, providing a comprehensive understanding of the region's phenomena. Employing Vensim software, we modeled and integrated the agriculture and hydrology of the region, subsequently deriving a mathematical model for analysis[1].

The model's performance was assessed over a 96-month period from 2011 to 2019, utilizing evaluation metrics such as Lux indices, protein production index, and yield index. Notably, the study reveals that, with the exception of 2018 when Khuzestan experienced flooding, the region consistently faces high water stress[2]. Remarkably, the environmental sector claims the largest share of resource consumption in the region, shaping the allocation dynamics[3]. Analyzing the prevailing agricultural patterns, our findings indicate that sugarcane, wheat, and rice exhibit the highest financial income per cubic meter of water consumption.

This research contributes valuable insights into the sustainability challenges of resource allocation in the Khuzestan agricultural sector. The integrated modeling approach provides a nuanced understanding of the complex interplay between hydrological and agricultural components, shedding light on potential strategies for optimizing resource management. The findings hold significance for policymakers, researchers, and practitioners seeking sustainable solutions to address water stress and enhance agricultural productivity in comparable regions.

Keywords: Systems Dynamics; Agricultural Modeling; Hydrological Modeling; Resource Allocation, Khuzestan Region

[1] Mehranfar, N., Kolahdoozan, M., & Faghihirad, S. (2023). Development of multiphase solver for the modeling of turbidity currents (the case study of Dez Dam). International Journal of Multiphase Flow, 168, 104586.

[2] Vahid, R., Farnood Ahmadi, F., & Mohammadi, N. (2021). Earthquake damage modeling using cellular automata and fuzzy rule-based models. Arabian Journal of Geosciences, 14, 1-14.

[3] Naghedi, S.R., Huang, X. and Gheibi, M., 2023. A smart dashboard for forecasting disaster casualties: An investigation from sustainable development dimensions (No. EGU23-17237). Copernicus Meetings.

How to cite: Maleki, A., Darabi Kerchi, H., and Piadeh, F.: Integrated Modeling of Agricultural and Hydrological Systems for Sustainable Resource Allocation: A Case Study of the Khuzestan Region, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-18890, https://doi.org/10.5194/egusphere-egu24-18890, 2024.

Flooding's impact on transportation infrastructure is crucial, influencing urban mobility, economic activities, and societal resilience [1]. Disruptions in transportation networks during flood events significantly impede access to essential services, intensifying the vulnerability of communities and hindering recovery efforts. Understanding the multifaceted consequences of flooding on transportation is fundamental for fortifying these critical systems against the escalating risks posed by changing climate patterns and extreme weather events [2].

Floods, stemming from various sources like heavy rainfall, storm surges, or river overflow, profoundly affect transportation infrastructure. Bridges, roads, and rail networks face damage or complete destruction, impeding travel and access to crucial services. Moreover, inundated areas and compromised roadways exacerbate accessibility challenges for specific demographic groups [3]. Vulnerable communities, including low-income populations or geographically isolated areas, bear a disproportionate burden, experiencing limited access to jobs, healthcare, and emergency services during and after flood events.

Research exploring the nexus between early warning systems and transportation resilience remains sparse but holds significant promise. Early warnings tailored to transportation vulnerabilities could mitigate disruptions, enhancing evacuation plans and rerouting strategies. Enabling timely and targeted information dissemination to affected areas or populations, especially those with limited mobility or access, can substantially reduce the adverse impacts on their daily lives and crucial infrastructure. Understanding the gaps in the interconnection of early warning systems and transportation resilience is crucial for bolstering the adaptive capacity of transportation networks, ensuring equitable access, and minimizing the disproportionate impacts of floods on vulnerable communities.

[1] Naghedi, S., Huang, X., Gheibi, M., (2023). A smart dashboard for forecasting disaster casualties: An investigation from sustainable development dimensions EGU General Assembly 2023, Vienna, Austria. https://doi.org/10.5194/egusphere-egu23-17237

[2] Piadeh, F., Behzadian, K., Chen, A.S., Campos, L.C., Rizzuto, J., Kapelan, Z. (2023). Event-based decision support algorithm for real-time flood forecasting in urban drainage systems using machine learning modelling. Environmental Modelling & Software, 167, p.105772.

[3] Yan, J., Naghedi, R., Huang, X., Wang, S., Lu, J. and Xu, Y., 2023. Evaluating simulated visible greenness in urban landscapes: An examination of a midsize US city. Urban Forestry & Urban Greening, 87, p.128060.

How to cite: Naghedi, S. N., Piadeh, F., Behzadian, K., and Hemmati, M.: Unveiling the Interplay: Flood Impacts on Transportation, Vulnerable Communities, and Early Warning Systems, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-13189, https://doi.org/10.5194/egusphere-egu24-13189, 2024.