Modeling and parameter identification of ice-affected streamflow

An extended Kalman filter was developed to automate the real-time estimation of ice-affected streamflow. The filter uses routine measurements of stream stage, air temperature, and the open-water relation between stage and streamflow to compute ice-affected streamflow. The filter accommodates three dynamic modes of ice effects: sudden formation/ablation, stable-ice conditions, and eventual elimination. The utility of the filter was evaluated by applications to historical data from two long-term streamflow-gaging stations, St. John River at Dickey, Maine and Platte River at North Bend, Nebr. Results indicate that the filter is stable and that site-specific parameters converged for both stations, producing streamflow estimates that closely match published values. Implementation of the filtering technique would provide a basis for quality-assuring real-time data on ice-affected streamflow.