Introduction

To gain an understanding of processes at all space and time scales in the Labrador Sea, a comprehensive observational network is necessary. Theory predicts that eddies operate on scales of 10 km and convective plumes on scales of 1 km in this region (Jones and Marshall, 1993); therefore a field experiment of conventional instruments will only resolve these small scales in a limited domain of the whole Labrador Sea. The Labrador Sea Deep Convection Experiment, which observed wintertime convection primarily through ship hydrographic sections, current meter moorings, and floats, reveals the rapid changes that occur in the the North Atlantic at various locations and times during convection. However, little work has been done to compare these observations with quasi-continuous and spatially-dense sea surface height measurements from the TOPEX/POSEIDON (T/P) satellite altimeter. Satellite altimetry yields a complete spatial picture of the phenomena for any given 10-day period, unlike in-situ instruments which are constrained by the slow velocities of the ocean. We should also consider the fact that general circulation models (GCM's) embody our cumulative, albeit imperfect, knowledge of the physics of the ocean. To form our best estimate of the Labrador Sea's true state, the Deep Convection Experiment should be synthesized with both satellite altimetry and a GCM. To be fair, both TOPEX/POSEIDON altimetry and GCM's have shortcomings in the Labrador Sea. TOPEX/POSEIDON adequately captures the eddy space and time scales (at least in an along-track direction), but aliases such processes as convection to lower frequencies because the 10 day repeat cycle. Also, a computationally realistic GCM can not explicitly resolve convection and must parameterize its effect. In this regard, in-situ observations complement satellite altimetry and GCM's. State estimation attempts to find the best possible estimate of the dynamical structure of the ocean based on a model and a combination of all forms of data. This method will simultaneously eludicate model shortcomings, errors in the forcing fields, and the noise structure of the data, as well as making an observational strategy for future missions more apparent. The first step in determining the feasibility of a state estimation project is a detailed evaluation of a model's consistency with data. In particular, this study will determine whether an eddy-resolving regional version of the MIT GCM in fact does give an adequate representation of deep convection in the Labrador Sea.