| KEYWORD |
Machine learning analysis of the shape characteristics of flood hydrograms to support the design and verification of flood attenuation reservoirs
Reference persons PIERLUIGI CLAPS, DANIELE GANORA
External reference persons GIULIA EVANGELISTA
Research Groups IDROLOGIA
Thesis type NUMERICAL-EXPERIMENTAL
Description The work is based on the application of Artificial Intelligence techniques (more precisely Machine Learning) for the maximisation of a scarcely available piece of information, i.e. the tendential shape of the flood waves coming from a given river basin. In fact, the recent flooding in Emilia Romagna (2023) showed how critical it is to study flood waves well before designing expansion basins, which may prove insufficient for very significant rainfall durations.
In a recent work: https://doi.org/10.1080/02626667.2022.2153051 we have offered, for the first time in Italy, a methodology to estimate this shape from geomorphological characteristics of basins. To broaden the scope of this research and extend it to other Italian and international contexts, we would like to investigate the shapes of flood hydrographs with non-parametric systems, such as those used in this work: https://www.researchgate.net/publication/281455079_Monthly_Runoff_Regime_Regionalization_Through_Dissimilarity-Based_Methods
but experimenting with tools typical of Machine Learning, such as the Random Forest method.
The final aim of the work is to indicate the amount of local information needed to predict the most probable shapes of flood waves exiting a river basin in order to effectively support the design of mitigation measures.
Required skills Basic programming elements. Ability to organise data using data analysis systems such as Matlab, R or similar tools
Deadline 15/08/2024
PROPONI LA TUA CANDIDATURA