AI-based warning system for heavy rain and urban flash floods
08. May 2021
JOINT PROJECT DEVELOPS SYSTEM FOR CIVIL SECURITY
Gelsenkirchen / Mülheim an der Ruhr, 04 May 2021: Heavy rain and flash floods are not a new phenomenon. But extreme weather events, especially in the summer months, are increasing as a result of climate change. Especially in urban areas, i.e. Ruhr cities such as Gelsenkirchen, sewer networks quickly become overloaded and streets and subways flood. This blocks rescue routes, e.g. for the fire brigade. In terms of warning time, geographically precise location and expected precipitation amounts, heavy rainfall events can hardly be adequately predicted. This makes it all the more important to further develop the forecasting models in terms of technology and content. This is where the BMBF joint research project "KIWaSuS" comes in.
KIWaSuS (KI-basiertes Warnsystem vor Starkregen und urbanen Sturzfluten) stands for „AI-based warning system for heavy rain and urban flash floods“. The aim of the project is to significantly increase the warning times for flash floods in cities, to localise them better and at the same time to provide important information for municipal crisis management in order to better protect citizens. To this end, an intuitive, digital map is to be created that reliably depicts the location and extent of the resulting flooding at an early stage, depending on the impending heavy rainfall event. This will provide meaningful support for local actors: Deployment plans for the fire brigade, disaster control and sewer network operators can be individually adapted to the event. Citizens can be warned in time and initiate their own protective measures.
In the cities of the Ruhr region, the situation is aggravated by the separation of entire districts by subways due to numerous motorways and railway lines. Past heavy rainfall events have shown that not only low points in the terrain such as subways can become obstacles, but also entire streets can turn into torrents within a very short time. The need for an efficient real-time warning system is not limited to Gelsenkirchen, but exists throughout Germany.
In the KIWaSuS project, 'artificial intelligence' (AI) is to be used as a central element. This is a technology that is normally used more in modern smartphones, cars or voice assistants for pattern recognition in images and speech. In KIWaSuS, AI will be used to learn correlations and patterns in the formation of heavy rain cells in order to better predict the temporal and spatial development of heavy rain cells in the future. On the other hand, AI will be used to learn the relationship between precipitation and the resulting runoff. This should make it easier to describe overloads of the sewer network and floods.
A prerequisite for the efficient use of AI is an intensive training process that requires a large database. The data is collected from various sources: For precipitation, measurement data has already been collected by municipalities and water associations for several decades. For precipitation-related runoff, on the other hand, hardly any data is currently available. Here, physically based runoff models are used to generate artificial training data. In addition, an innovative sensor system is to be set up to consolidate and supplement the database. All data streams will be combined in a central data platform and transformed into a format suitable for ML and made available for the forecast models.
The companies involved in this joint project are neusta software development west GmbH, Gelsenwasser AG, Abwassergesellschaft Gelsenkirchen, the Institute of Hydraulic Engineering and Water Management at the University of Duisburg-Essen and the Institutes of Civil Engineering and Measurement and Sensor Technology at the Ruhr West University of Applied Sciences. The consortium leader is Prof. Dr. Markus Quirmbach from the Institute of Civil Engineering. Requirements and data are provided by the Gelsenkirchen Fire Department, the State Office for Nature, Environment and Consumer Protection and the Emschergenossenschaft. The project started in April 2021 and will run until March 2024. The project is funded with approximately 1.5 million euros from the Federal Ministry of Education and Research funding guideline: „Artificial Intelligence in Civil Security Research“ in the programme „Research for Civil Security 2018 to 2023“.
Die Hochschule Ruhr West verwendet Cookies, um Zugriffe und Marketingmaßnahmen zu analysieren. Diese Informationen geben wir an unsere Partner für Medienwerbung und Analysen weiter. Wir setzen Cookies von Analyse- und Retargeting-Software ein, um unsere Website bestmöglich an die Bedürfnisse unserer Besucher anpassen zu können. Sie können der Nutzung von Cookies zustimmen oder widersprechen. Weitere Informationen finden Sie in unserer