Constrained optimization

Techniques developed at MIT and elsewhere will be used in ACES to solve for the values of uncertain physical parameters in planetary models. These optimization methods employ advanced strategies for adjusting non-linear models to optimally fit observations. Studies that will be carried out in ACES include

• Estimating oceanic mixing due to turbulent eddies.
• Estimating vertical mixing in the ocean.
• Estimating viscous properties of the solid earth.

The figure below shows the results for one such experiment. A parallel adjoint of the M.I.T. General Circulation Model and an iterative optimization system are used to estimate the turbulent eddy forcing on the mean flow required to fit a 100 year ocean simulation trajectory to observations. The plots show the eddy mixing coefficient implied by the calculated forcing. The top panel shows the zonal average and the lower panel shows the mean mixing in the upper 1000m.