Analysis and Forecasting of Sea Ice Conditions with Three-Dimensional Variational Data Assimilation and a Coupled Ice–Ocean Model

Abstract

A three-dimensional variational data assimilation (3DVAR) system has been developed to provide analyses of the ice-ocean state and to initialize a coupled ice-ocean numerical model for forecasting sea ice conditions. This study focuses on the estimation of the background-error statistics, including the spatial and multivariate covariances, and their impact on the quality of the resulting sea ice analyses and forecasts. The covariances are assumed to be horizontally homogeneous and fixed in time. The horizontal correlations are assumed to have a Gaussian shape and are modeled by integrating a diffusion equation. A relatively simple implementation of the ensemble Kalman filter is used to produce ensembles of the ice-ocean model state that are representative of background error and from which the 3DVAR covariance parameters are estimated.Data assimilation experiments, using various configurations of 3DVAR and simpler assimilation approaches, are conducted over a 7-month period during the winter of 2006/07 for the Canadian east coast region. The only data assimilated are the gridded daily ice charts and RADARSAT image analyses produced by the Canadian Ice Service. All of the data assimilation experiments produce significantly improved shortterm forecasts as compared with persistence. When assimilating the same data, the forecast quality from the experiments employing either the 3DVAR, direct insertion, or nudging is quite similar. However, assimilation of both the daily ice charts and RADARSAT image analyses in 3DVAR results in significant improvements to the sea ice concentration forecasts. This result supports the use of a data assimilation approach, such as 3DVAR, for combining multiple sources of observational data together with a sophisticated forecast model to provide analyses and forecasts of sea ice conditions.

Reconstructing extreme AMOC events through nudging of the ocean surface: a perfect model approach

“…Initialized experiments can additionally or alternatively assimilate salinity (Servonnat et al 2015), sea ice (e.g. Caya et al 2010), land soil moisture (e.g. Han et al 2012), as well as different atmospheric fields, such as wind-stress (Ding et al 2013;Thoma et al 2015) or the 3D winds (Stewart and Haine 2013).…”

Scale-dependent background-error covariance localisation

“…In contrast, areas near the ice edge, where the ice concentration is between 0 and 100 %, have larger values of ensemble spread. A diffusion operator is used for representing the spatial background-error correlations within this sea-ice 3DVar data assimilation system and not a spectral transform (Caya et al, 2010). Shlyaeva et al (2015) also used the diffusion operator to represent the spatial localisation function in preliminary experiments with EnVar.…”

Evaluation of an operational ice–ocean analysis and forecasting system for the Gulf of St Lawrence

“…The largest errors in the ice cover are found in the St Lawrence Estuary and in the northeast gulf and coincide with areas of lowest Radarsat coverage. A 3DVAR sea-ice assimilation system based on Caya et al (2010) is currently in development to permit the use of additional ice datasets.…”