Sensitivity of Inverse Box Model Estimates to Mesoscale and Interannual Variability

Inverse models applied to hydrographic sections provide one of the few observation-based estimates of the global overturning circulation and global-scale mixing, against which ocean models are frequently compared. These inverse models work under the assumptions of a steady-state ocean circulation, that the mesoscale variability that occurs on the time scale it takes to gather hydrographic data is negligible, that the hydrographic data adequately samples mean properties, and that the inverse model’s dynamical constraints are good time-mean approximations. If the presence of eddies or interannual variability in the observations utilized violate these assumptions, the inferred overturning streamfunction and diapycnal diffusivities inferred from the inverse box model may have substantial uncertainty that is not rigorously accounted for in traditional error budgets. Here we assess these sources of error by sampling a strongly-eddying (1/10th degree) climate model as if by a ship and applying inverse methods to the virtual hydrographic transects.