Bayesian Decision for Low Flow Evaluation in Non-Stationary Conditions

The investigation of changes in the minimal river runoff was carried out for the Volga river basin, where from the beginning of 80s the changes occurred in the alimentation regime of the rivers. Minimal runoff has grown in all rivers of this region, and the trend of its fluctuations in time allows us to propose a hypothesis about the transition of the river runoff process to the new stationary regime. Thus we can say that there is a non-stationary random sample of minimal runoff for all periods of observation. For the non-stationary case, high uncertainty is inherent in predicting future changes in runoff. This study presents the problem of estimating the statistical characteristics (parameters of distribution) of time series for non-stationary conditions using the Bayesian approach.