Statistical Analysis of the 30-Day Water Level Forecasts in the St. Lawrence River

Abstract

Forecasts of water levels in the St. Lawrence River on the east coast of Canada have been issued every working day since 1997 using a one-dimensional hydrodynamic model (One-D) for periods extending up to 30 days. In order to assess the performance of these forecasts, a comparison between the model forecasts and the observations of the water levels during 2005 was done at 12 stations of the SINECO network located between Montreal and Quebec City. The statistical analysis shows that mean errors are small compared to the water level fluctuations. Confidence intervals of the forecasted values for all stations are evaluated.

Keywords:

• Water level
• One-D
• one dimensional
• numerical hydrodynamic model
• tides
• St. Lawrence
• river
• forecast
• statistical analysis
• fresh water
• observation

Acknowledgements

The authors wish to acknowledge the work of François J. Saucier, then at the Ocean Science Branch of the DFO Maurice Lamontagne Institute, who first implemented the system for operations and directed the work of Daniel Bourgault and Diane Lavoie to improve it. Afterwards, Nathalie Roy, Gilles Mercier, and Alain D'Astous continued the operations and improvements under the supervision of D. Lefaivre. Bernard Labrecque (CHS) provided the observations and Bernard Tessier (CHS) supervised the forecasted data products. Agathe Roy (CHS) helped in shaping the figures. The wind forecast data for the area is made available by the Meteorological Service of Canada of Environment Canada through a joint program of marine forecast.