Janjua, Harveen2015-05-012015-05-012015-05-012015-04-30http://hdl.handle.net/10012/9308The ice cover in the perennial region of the Arctic Circle has reduced signi cantly in recent years. Various models are available to predict the spatial and temporal evolution of the ice cover.Predictions from these models can be improved by incorporating satellite observations by the technique of data assimilation. In this thesis, ice thickness observations from Advanced Microwave Sensing Radiometer Earth (AMSR-E) and Moderate Resolution Imaging Spectro-Radiometer (MODIS) remote sensors are fused with that from an ice-ocean model using an optimal interpolation technique. It is assumed that the background error covariance matrix is static and the spatial correlations are modelled using a di usion operator. The observation error covariance matrix is diagonal and the observation operator is saturated to a threshold of 0.2 m for ice thickness observations from AMSR-E because ice thickness is negatively correlated to the polarization ratio for thin ice up to 0.2 m only. It is observed that when more observations are available the analysis from data fusion is closer to ice charts produced by Canadian Ice Services (CIS). One possible application of the system developed is in areas where ships need to be safely routed through ice infested water. This thesis presents a small example that tries to nd the path for a ship through thick ice. The impact of fusing perturbed observations with ice thickness data inferred from a satellite image on the ice thickness traversed and distance travelled is investigated.enData assimilationsea ice thicknessbackground error covarianceobservation operatorship routingMaster ThesisSystem Design Engineering