Oceanviewer Banner

Evaluation, Improvement, and Communication of Short-Term Hazardous-weather Forecasts over Coastal British Columbia

  • Headshot

    Daniel Kirshbaum McGill University

This project focuses on a new short-term weather prediction system called the MEOPAR High-Resolution Ensemble Kalman Filter (MEOPAR/HREnKF), which is currently under development within the MEOPAR project “A Re-locatable Coupled Atmosphere-Ocean Prediction System.”

This project has supported a new collaboration between McGill University and Environment and Climate Change Canada (ECCC) that addresses an important societal problem: precipitation observation and prediction in the mountainous maritime climate of British Columbia and the northwestern United States. The problem is addressed through the creation and observational verification of high-resolution (2.5-km grid-length) 24-hour forecast ensembles from ECCC's Global Environmental Multiscale model, which assimilate observations at high spatial and temporal resolution using an Ensemble Kalman Filter.  

The project achievements thus far include:

  • The creation of an observational precipitation product incorporating all of the available rain-gauge and operational radar in the region
  • Using the geostatistical merging method Kriging with External Drift, the uncertainties associated with this product have been estimated through comparison with other statistical mapping methods and independent precipitation products produced by Canadian and U.S. operational modeling centres 
  • The observational product has been used to statistically evaluate the performance of the experimental ensembles, along with two alternative, and less computationally expensive, GEM ensembles.  

The experimental ensembles are found to outperform one alternative system (the lower-resolution regional ensemble operational at ECCC) but perform similarly to another (identical resolution but no explicit data assimilation).  Thus, improved forecast performance commensurate with that of the experimental system may be achieved at a reduced computational cost by eliminating the high-resolution data assimilation component.  This result is highly beneficial for ECCC, who can devote their available computing resources to other, more pressing, forecasting needs.  It also benefits Canadian industries and the broader public, who rely on ECCC forecasts for business and leisure activities.  This collaboration across societal sectors, lays the groundwork for future collaboration between academia, government, and industry to further refine short-term prediction of natural hazards in BC and other regions.

Partners:

  • Environment Canada - Meteorological Service of Canada

Investigators:

  • Luc Fillion

MEOPeers:

  • Phillippa Cookson-Hills McGill University

The project results will positively impact both model developers and end-users of forecast products.  Our quantification of the benefits and shortcomings of the MEOPAR/HREnKF will steer future improvements to this and similar systems. 

Moreover, to realize the economic benefits of the MEOPAR/HREnKF, its outputs must be effectively communicated to end-users. 

Different end-user groups in BC will be engaged through focus groups, which will provide valuable data on the information they seek in weather forecasts, their preferences in formats, and their perceived value of MEOPAR/HREnKF forecast improvements.  These interactions will help to maximize the impacts of MEOPAR/HREnKF forecasts on end-user decision-making.