Transfer function-noise model applied to flow forecasting / Modèle Box-Jenkins pour la prévision des débits des cours d'eau

Abstract

The combined transfer function-noise model investigated by Box and Jenkins has been used to identify a model and estimate parameters for the simulation and forecasting of streamflow. The model reliability was tested with hourly storm data recorded in three basins of different size (7.6, 5.3 and 841 km²). The model proved to be satisfactory in short term (few hours) forecasting. As real time streamflow forecasting may be of practical importance, details are given about program dimensions, data requirements and job execution costs.