Amrhein, D.E., Stephenson, D., and Thompson, L. (2024)
Presented at:
Ocean Sciences Meeting 2024This work describes and applies a novel inverse framework for identifying dominant atmospheric drivers of stochastic ocean variability. The approach, which is based on "balanced truncation" approaches to model reduction, combines statistics of atmospheric variability with dynamical constraints from the adjoint of an ocean general circulation model (OGCM) to derive atmospheric patterns optimized to excite variability in a specified ocean quantity of interest. We apply our analysis to the problem of upper ocean heat content (HC) variability in the North Atlantic Subpolar Gyre (SPG) using the adjoint of the MITgcm and atmospheric fluxes of buoyancy and momentum from the ECCOv4-r4 state estimate. Initial principal components analyses reveal a range of spatiotemporal flux patterns over the North Atlantic during 1992-2017, as well as a diversity of potential ocean dynamical pathways by which fluxes might contribute to SPG HC variability. However, by combining flux statistics and adjoint sensitivities in a dynamics-weighted principal components approach, we find a highly reduced subset of atmospheric patterns that is responsible for driving SPG HC variability, each with a distinct mechanism for driving variability. By perturbing the ECCOv4-r4 state estimate, we illustrate these mechanisms and show that they are effective in a nonlinear ocean model. Consistent with previous studies, we find evidence of a leading role for local fluxes associated with the North Atlantic Oscillation in driving SPG HC variability. This technique is applicable across a range of problems in order to discover atmospheric modes responsible for driving the ocean.
