Évaluation des modèles portuaires de prévisions océaniques développés dans le cadre du Plan de Protection des Océans : Le Fleuve Fraser

Abstract

Canada's Oceans Protection Plan (OPP) was launched in 2016 to support initiatives aimed at protecting our marine environment from anthropogenic pressures. To this end, the Improving Drift Prediction and Nearshore Modelling sub-initiative of OPP developed six high-resolution port-scale hydrodynamic models, to improve safe navigation, and provide operational emergency response to events such as marine oil spills. The models were downscaled from the Coastal Ice-Ocean Prediction Systems East and West (CIOPS-E, CIOPS-W). Atmospheric forcing was provided by the High-Resolution Deterministic Prediction System (HRDPS). Model performance is assessed against available observational data and contrasted with the parent model using a multi-year hindcast. Evaluations of 48-hour forecasts are performed during a two-month period. This document reports on the performance assessment for the Fraser River model (SF30) and the southern part of the South Salish Sea model (SSS150). SF30 domain covers Fraser River South Arm from Steveston to Port Mann with the horizontal resolution of 30 m and adjacent parts of Sturgeon and Roberts Banks, while SSS150 covers the southern part of Salish Sea and the Fraser River to Mission and to Pitt Lake. All model runs were completed without stability issues including periods of extreme weather events and real-time instrument failure. Results show that the high-resolution models provide either similar or better performance than the lower-resolution models in the areas where they overlap and have adequate performance in the areas not covered by the low-resolution models. The tidal water level error ranges from 5 to 9 cm CRMSE, while the non-tidal water level error is consistently 5-6 cm CRMSE with biases within 6 cm, which brings the total water level error to an order 10 cm. Both tidal and non-tidal current velocity is captured well with a total error of about 0.25 m/s when compared with the observed horizontal ADCP velocities at Woodward’s Landing. The cross-shore structure of the flow in the Fraser River compares well with the data from ship mounted ADCPs with a slight underestimation of the outflow velocities. Surface water temperature in Fraser River has a cold bias of less than 0.5 °C and errors of order 1.1 °C. Vertical distribution of temperature and salinity is represented well and on par with Salish Sea 500 model (SS500). Drift performance is significantly improved in the high-resolution models relative to SS500. The forecast evaluation showed that non-tidal water level error increased gradually with forecast lead time as expected.