Modelling the runs' properties of annual flows

Abstract

Abstract A model is developed that preserves the statistical properties of run-lengths and run-sums of a stationary flow series at a given truncation level. The model first generates the positive and negative run-lengths, and then the corresponding run-sums by a regression equation. It is applied to independent and first-order autoregressive processes. It is shown that the deficit of the process can be expressed in terms only of the runs' properties. Thus the model reproduces the statistical properties of the deficit for a certain yield by choosing the truncation level accordingly. The simulation results show that the model works more satisfactorily for independent processes.