Pillar, H., Nguyen, A.T., Heimbach, P., Bigdeli, A., Loose, N., Ocaña, V., Lozier, S., and Li, F. (2020)
Presented at: Ocean Sciences Meeting 2020
The disproportionate importance of the subpolar North Atlantic in the global climate system arises from the strong heat loss and associated water mass transformation in the region, which shape the strength and structure of the AMOC and its associated heat redistribution. Due to the complex nature of the circulation here, and a prior lack of in-situ constraints, state-of-the-art GCMs exhibit a wide range of ability in simulating the local stratification and deep convection. The recently acquired in-situ OSNAP observations are beginning to provide crucial ground-truth for assessing model performance in the subpolar North Atlantic. However, these data hold even greater potential for constraining the hydrography throughout the model domain via adjoint-based assimilation. This model-data synthesis allows the array-based constraints to propagate through space and time, as governed by the underlying dynamics, and impact the evolution of the entire modeled state. Importantly, by only adjusting the independent variables, we obtain an estimate of the ocean state that is simultaneously consistent with both the assimilated oceanic data and the governing equations, to within prescribed uncertainty.
Here we present results from assimilation of the first 21-month record from OSNAP into the Arctic and Subpolar gyre sTate Estimate (ASTE). We discuss the dynamical mechanisms underlying the OSNAP-driven model adjustments and the associated reduction in model bias. We highlight differences in the constraint provided by (1) East vs. West sections of the array and (2) temperature vs. salinity observations. The total OSNAP-driven bias reduction is significant throughout the water column, both at the location of the moorings and throughout the larger subpolar North Atlantic region, strengthening the modeled MOC and associated tracer transports throughout the Atlantic basin. Since ASTE has already been optimized to several million diverse satellite and in-situ oceanographic observations, these results confirm the high independent information potency of OSNAP. We conclude with a brief assessment of the differences in constraint imposed by RAPID vs. OSNAP mooring data, set by differences in adjustment pathways/timescales and differences in dominant forcings in the subtropical vs. subpolar North Atlantic.