Um modelo markoviano para simulação de vazões diárias

Abstract

Research of a model to simulate dayly flows under the base of Markov's chains theory. To reach this goal, a historic sucession of mentioned flows is analysed as a first and second order process prevailing that, for both cases, the chain ergodicity is determined and its stationary distribution is observed. A theoretic adjustment for markovian matrix lines is analysed through Pearson's indexes, drying out gama, normal and lognormal distributions. Dry and wet periods of the hydrologic year are observed separately, while simulated series are compared with the corresponding historics in order to point out the best adjustable process for the period under analysis.