長期間流出機構の情報理論的研究

Abstract

Because the long range runoff system is essentially the stochastic one, f or theanalysis and synthesis of the system, it is necessary to disclose not only the deterministiccharacteristics but the statistical laws contained in the transition system from precipi-tation to river discharge. From this point of view, this paper proposes the methodsof the analysis and synshesis of the long range runoff system based on the conceptsand the techniques of Information theory.First, "Statistical UnitHydrograph Method" is presented using Wiener's Filtering theoryafter verifining the stationality of the daily precipitation series and making time-invariantand linearization the system. Second, in order to clarify the interior statistical mecha-nism of the system, a method of the transition probabilities among the runoff states whichconsist of surface, sub-surface and ground water runoff states, is proposed based on theassumption which the Entropy defined by Shannon becomes maximum. The applyingresults of these stochastic procedures to Yura River Basin were good agreement withthe natural stream records.