Assessment of the impact of mesoscale eddies on the large-scale physical and biogeochemical responses under climate change

Earth System Models utilized to predict the ocean physical and biogeochemical responses under anthropogenic climate change do not yet routinely resolve mesoscale eddies due to computational resource constraints, and ocean mesoscale eddies are often parameterized. Mesoscale eddy parameterizations are known to affect the large-scale ocean circulation through impacts on the stratification, leading to for example changes of the nutrient stream, which is known to have large-scale impacts for the nutrient supply. Here we examine numerical ocean models with different representations of the mesoscale eddies and their combined physical and biogeochemical response in a hierarchy of models differing in the horizontal resolution, ranging from a non-eddying model with eddy parameterization, eddy permitting models without and with eddy parameterization, and eddy resolving models that serves as the model truth. 

In the case of non-eddying models, we find that existing prescriptions can result in a "better" bulk biogeochemical response but for the wrong physical reasons, relative to the model truth. On the other hand, more recent parameterization schemes can improve the physical response in terms of the sensitivity to changes in forcing, but the biogeochemical response is more subtle. It is confirmed here that the biggest change in the biophysical response stems from a model becoming eddy permitting. However, the eddy permitting model without parameterization is found to overshoot and be "too good" relative to the model truth in the biogeochemical response, attributed to the explicit eddies being too weak. Results are presented from a new approach that supplements the action of explicit eddies in eddy permitting model via eddy parameterization, but without adversely damping the explicit eddies. Combined with more recent parameterizaton schemes, the approach can lead to a biogeochemical response that compares very favorably to the model truth, attributed entirely to the improvement in the physical response in the eddy permitting model, so that we are getting the right answer for the right reasons.