A First Attempt of Data Assimilation for Operational Sea Ice Monitoring in Canada

Abstract

A three-dimensional variational data assimilation (3D-Var) system is developed as a first attempt to explore the potential use of data assimilation to improve a coupled ice-ocean model (CIOM) forecast of sea ice near the east coast of Canada. The accuracy of the resulting analysis is largely dependent upon the forecast-error covariance matrix. This study focuses on the estimation of forecast-error statistics required in a 3D-Var system and their effect on the ocean part of the CIOM. This is accomplished by comparing CIOM output according to different specifications of forecast-error statistics used during the data assimilation. The results show no improvement in the ice forecast with respect to persistence. It has been concluded that the assimilation system still needs significant improvement including assimilation of many different types of observations such as sea surface temperature, ice drift, ocean current, etc. An improved forecast-error covariance matrix is needed for a more accurate description of the system's behaviour.