|
Journal of Environmental Hydrology
ISSN 1058-3912 |
||
|
Electronic Journal of the International Association for
Environmental Hydrology
JEH Volume 16 (2008), Paper 26 Posted August 12, 2008 DEVELOPMENT OF A REAL-TIME RIVER FLOOD FORECASTING TRANSFER FUNCTION-NOISE MODEL WITH A KALMAN FILTER FOR SNOWMELT DRIVEN FLOODS
Jan Adamowski1 2 University of Ottawa, Ottawa, Ontario, Canada ABSTRACT A real-time flood forecasting model for the Rideau River in Ottawa, Canada was developed for issuing flood warnings with sufficient lead-time. A Transfer Function-Noise (TFN) stochastic model coupled with recursive parameter estimation via a Kalman prediction algorithm was used to forecast the spring flood at the Ottawa gauging station using an upstream station (at Manotick) and tributary flows (at Jock River) as model inputs. Also, spring snowmelt runoff computed using mean daily temperature, snowfall and areally averaged snowdepth was explicitly represented in the model. The model was calibrated and tested on spring flood data from 2002 and 2004. Comparison of forecast results for a six hour lead-time showed that the new model is better suited to the Rideau River flow than the previously developed Self-Tuning Predictor (STP) model. Reference: Adamowski, J., and K. Adamowski. 2008. Development of a real-time river flood forecasting transfer function-noise model with a Kalman filter for snowmelt driven floods. Journal of Environmental Hydrology, Vol. 16, Paper 26. CONTACT: Jan Adamowski Massachusetts Institute of Technology, E40-476 77 Massachusetts Ave. Cambridge, MA 02139-4307 USA E-mail: adamowsk@mit.edu |
||
|
Return to JEH 2008 Papers
Return to HydroWeb Homepage |
||