Stochastic modelling of the spatial structure of rainfall

When estimating the risk of a flood event, the need to study the spatial structure of a precipitation event arises. As far as this study can be done based on stochastic simulations, the need to construct stochastic models consistent with observed data arises. As observed from precipitation data, it seems that the spatial distribution of the rain has a scaling behaviour or long scale dependence, a behaviour also known as the Hurst phenomenon. This thesis aims to construct a stochastic model that can reproduce satisfactorily the behaviour as observed in Nature from field measurements. Specifically, in this thesis an effort is done to expand the stochastic model SMA (Symmetric Moving Average) from one dimension to two dimensions and use it for generation of synthetic rainfall fields.