Can a Bivariate AR(1) Process Model the Variability of the Inflow into Stochastic Reservoirs?

A. Eriksson and D. Preve (Sweden)


Hydrological models; autoregression; non-Gaussian time series; range modeling.


Here we investigate the statistical evidence for a novel class of constrained bivariate AR(1) processes to capture the temporal dynamics of the maximum and minimum for the hourly/daily inflow into a stochastic reservoir. The estima tion is done using a perturbed maximum likelihood tech nique and is illustrated by an empirical example.

Important Links:

Go Back