The flow in the upper Niagara River is affected by weed growth in the
summer and by ice formation and accumulation in the winter. Changes in
flow resistance tend to occur abruptly with time and space during ice
-affected flows and make it difficult to forecast flows for power gene
ration. Since 1988, the New York Power Authority and Ontario Hydro hav
e collaborated on the development of a computer-based system for forec
asting flows and optimizing power generation from the Niagara River. A
hydraulic routing model has been developed for forecasting the flow i
n the upper Niagara River under uncertainties due to ice consisting of
: (1) A steady-flow model; (2) an unsteady-flow routing model; (3) a l
inear, time-varying reservoir model for forecasting the flow into the
Grass Island Pool; (4) a time-varying, stochastic model for the stage-
discharge relation at Fort Erie, the upstream boundary of the reach; a
nd (5) Kalman filtering schemes to be coupled with these components mo
dels, to reduce uncertainties in parameters and variables to be estima
ted. This paper presents components (2)-(5) and related concepts and m
odeling techniques.