Adequacy of Markovian Models with Cyclic Components for Stochastic Streamflow Simulation

Publisher Website
Google Scholar
Abstract
The concept of deterministic hydrologic cycles is questioned from physical and statistical points of view. The spectra of streamflow series are shown to vary with time, and this variation suggests the nonstationary character of the data. Demodulation techniques are used to study the statistical character of the amplitude and phase of the annual streamflow cycle, and the use of the results in models for simulation is suggested. A case study performed with several Markovian type models shows that the seasonal variation in the standard deviation should be incorporated in any model used for design. It is also shown that the type of distribution used for the random component does not have a great influence on streamflows generated for storage design. Great care should be taken when using logarithmic models because they may produce very distorted synthetic traces when improperly used.

This publication has 6 references indexed in Scilit: