Information - NH11.06 Data-driven and computational intelligence methods in flood forecasting
Event Information
Modelling and forecasting such hydrological hazards as floods heavily relies on the so-called process, or physically-based approach. Its basis is the differential equations describing the water motion that are numerically solved.
Along with such process (physically-based) models the so-called data-driven models are gaining popularity. They use the methods mainly developed in the framework of computational intelligence and go considerably further than statistical models. Data-driven models establish a functional relationship between the input and output variables on the basis of historical data and a re able to make predictions. Most popular methods in this category are the artificial neural networks and fuzzy rule-based systems. Recently a number of other methods (like support vector machines, model trees, instance-based techniques and chaos theory) and their combinations (mixtures) are becoming popular.
In the context of flood hazards, it was demonstrated in a number of studies that data-driven models successfully complement more traditional physically-based models, and may even replace them. Their ability to reproduce complex non-linear relationships makes and the associated methods of adequate data preparation for their training (calibration) allow for accurate predictions of water levels and flows. Data-driven models can also be accurate predictors of errors of the physically-based models allowing for the appropriate predictions updates. Successful practical applications include river and coastal floods, urban flash floods, etc.
Concentrating papers on this particular approach to modelling and forecasting will allow researchers and practitioners to form a professional community speaking the same language, and exchange latest research results and details of various types of practical applications.
The session will focus on the following main topics: (1) the latest advances in data-driven modelling; (2) practical applications of data-driven models in forecasting various types of floods and the associated risks
Preliminary List of Solicited Speakers
Co-Sponsorship
HS
General Statement
The information contained hereafter has been compiled and
uploaded by the Session Organizers via the "Organizer Session Form".
The Session Organizers have therefore the sole responsibility
that this information is true and accurate at the date of publication,
and the conference organizer cannot accept any legal responsibility
for any errors or omissions that may be made, and he makes no warranty,
expressed or implied, with regard to the material published.
Back