In recent years, hydrologists have attempted, with encouraging results, to synthesize the accomplishments of time series analysis and deterministic systems methods into the more general framework of stochastic dynamic systems. The purpose of this paper is to attempt an improvement of a stochastic time series model by the use of inputs generated by a distributed, deterministic model.
The analysis is based on a simplified transfer function-noise model, where the system output is modeled as a linear combination of present and past inputs as well as past outputs.
The systems chosen for the analysis are two partially glaciated basins in Iceland.
The results were in general promising and showed that a careful selection of input series could overcome to a great extent the non-linearities as well as the non-stationarities in the relationship between discharge and meteorological variables. The results also showed that transformation of the meteorological series using distributed, deterministic techniques gave superiour results, compared with the standard degree-day approach.
