Eddy-permitting simulations of the sub-polar North Atlantic: impact of the model bias on water mass properties and circulation

Some previous studies demonstrated that model bias has a strong impact on the quality of long-term prognostic model simulations of the sub-polar North Atlantic Ocean. Relatively strong bias of water mass characteristics is observed in both eddy-permitting and eddy-resolving simulations, suggesting that an increase of model resolution does not reduce significantly the model bias. This study is an attempt to quantify the impact of model bias on the simulated water mass and circulation characteristics in an eddy-permitting model of the sub-polar ocean. This is done through comparison of eddy-permitting prognostic model simulations with the results from two other runs in which the bias is constrained by using spectral nudging. In the first run, the temperature and salinity are nudged towards climatology in the whole column. In the second run, the spectral nudging is applied in the surface 30 m layer and at depths below 560 m only. The biases of the model characteristics of the unconstrained run are similar to those reported in previous eddy-permitting and eddy-resolving studies. The salinity in the surface and intermediate waters of the Labrador Sea waters increases with respect to the climatology, which reduces the stability of the water column. The deep convection in the unconstrained run is artificially intensified and the transport in the sub-polar gyre stronger than in the observations. In particular, the transport of relatively salty and warm Irminger waters into the Labrador Sea is unrealistically high. While the water mass temperature and salinity in the run with spectral nudging in the whole column are closest to the observations, the depth of the winter convection is underestimated in the model. The water mass characteristics and water transport in the run with spectral nudging in the surface and deep layers only are close to observations and at the same time represent well the deep convection in terms of its intensity and position. The source of the bias in the prognostic model run is discussed.     

Development of a stochastic weather generator for the sub-polar North Atlantic

The article presents an approach for creating a computationally efficient stochastic weather generator. In this work the method is tested by the stochastic simulation of sea level pressure over the sub-polar North Atlantic. The weather generator includes a hidden Markov model, which propagates regional circulation patterns identified by a self organising map analysis, conditioned on the state of large-scale interannual weather regimes. The remaining residual effects are propagated by a regression model with added noise components. The regression step is performed by one of two methods, a linear model or artificial neural networks and the performance of these two methods is assessed and compared. The resulting simulations express the range of the major regional patterns of atmospheric variability and typical time scales. The long term aims of this work are to provide ensembles of atmospheric data for applied regional studies and to develop tools applicable in down-scaling large-scale ocean and atmospheric simulations.     

Model simulations of mesoscale eddies and deep convection in the Labrador Sea

Deep convection in the Labrador Sea is confined within a small region in the southwest part of the basin.The strength of deep convection in this region is related to the local atmospheric and ocean characteristics,which favor processes of deep convection preconditioning and intense air-sea exchange during the winter season.In this study,we explored the effect of eddy-induced flux transport on the stratification of the Labrador Sea and the properties of deep convection.Simulations from an eddy-resolving ocean model are presented for the Labrador Sea.The general circulation was well simulated by the model,including the seasonal cycle of the deep Labrador Current.The simulated distribution of the surface eddy kinetic energy was also close to that derived from Topex-Poseidon satellite altimeter data,but with smaller magnitude.The energy transfer diagnostics indicated that Irminger rings are generated by both baroclinic and barotropic processes; however,when they propagate into the interior basin,the barotropic process also disperses them by converting the eddy energy to the mean flow.In contrast to eddy-permitting simulations,deep convection in the Labrador Sea was better represented in the eddyresolving model regarding their lateral position.Further analysis indicated that the improvement might be due to the lateral eddy flux associated with the resolved Irminger rings in the eddy-resolving model,which contributes to a realistic position of the isopycnal dome in the Labrador Sea and correspondingly a realistic site of deep convection.