Dry period statistics of monthly flow models.

Bayazit M., Şen Z.

Modeling hydrologic processes. Proc. Fort Collins 3rd international hydrology symposium, July 1978, pp.18-31, 1979 (Scopus) identifier


Modelling of periodic-stochastic processes of monthly flows is studied from the point of view of economy in parameters and preservation of dry period statistics. It is found that ARIMA models with very few parameters provide satisfactory fit with the observed series. Autocorrelation structure of seasonal ARIMA models is investigated. Analytical expressions are derived for the probability distribution of negative run-lenths of stationary first order Markov models. These are generalised to monthly flow models. Run-lenth statistics of Thomas-Fiering and ARIMA models are obtained by simulation. Analysis of observed monthly flows of a number of streams has shown that their run-lenth statistics are satisfactorily preserved by Thomas-Fiering and ARIMA models. Models with constant autocorrelation coefficient cannot preserve the variance of run-lenths. - Authors