• Rainfall

The aim of this thesis is studying rainfall models, capable of simulating rainfall data to serve as input for a general model aimed at the evaluation of the hydrogeological risk. A space-time model for rainfall is first analysed in detail and it is subsequently generalized by mean of Hidden Markov Models. Different techinques for fitting the new model to data are presented in theory: maximum likelihood estimation through the Expectation Mazimization algorithm and Approximate Bayesian Computation. They are then tested and applied in numerical simulations.