A new method for modelling the dynamics of rain sampled by a tipping bucket rain gauge is proposed. The considered models belong to the class of integer valued autoregressive processes. The models take the autocorelation and discrete nature of the data into account. A first order, a second order and a threshold model are presented together with methods to estimate the parameters of each model. The models are demonstrated to provide a good description of dt from actual rain events requiring only two to four parameters.
|Publication status||Published - 1999|