A multiple distribution Markovian model for annual hydrologic time series (Q1115020)

It is assumed that the geophysical mechanisms responsible for the amount of annual rainfall can be in each year in one of the states \(O_ 1,...,O_ M\). The occurrence of such states is described by a Markov chain. The statistical distribution of the annual value of the studied time series depends on the state \(O_ j(t)\) occurring at time t. The case \(M=2\) is analyzed in detail and then it is applied to the annual rainfall at Fortaleza in Northeast Brazil. The model can preserve non- linear structures of serial dependence and it produces a skewed marginal distribution without any transformation.