Efficient uncertainty quantification methods for flood modelling with multiple uncertain inputs (C++, Matlab)
In reality, the inputs of a flood model might be uncertain. These multiple probabilistic uncertain inputs propagate from a model (it is LISFLOOD here) into outputs and make them probabilistic as well.
Different uncertainty quantification methods are available to study this propagation and the classical one is the standard Monte-Carlo (SMC) method. Unfortunately, the standard Monte-Carlo is too costly as we will need a large sample size to perform the quantification. In this project, alternative uncertainty quantification methods are employed and compared for flood modelling aplication. These alternatives to the standard Monte-Carlo methods are: Latin Hypercube Sampling (LHS), Adaptive Stratified Sampling (ASS), Quasi Monte-Carlo (QMC), Haar-Wavelet Expansion (HWE)
Findings (based on more than 4,800,000 simulation runs) indicate these alternative methods can be much more accurate and efficient than the standard Monte-Carlo method.
Presentation in annual steering commitee (July 2022).
Oral presentation in Water Group at University of Sheffield (February 2022).
A paper is already ready. The publication to be updated soon.