The ROMS Tangent Linear and Adjoint Models
Andrew M. Moore
University of Colorado, USA

Many problems in ocean dynamics, data analysis, ocean prediction and predictability can be expressed mathematically as a sensitivity problem. The adjoint of a numerical model is a general sensitivity operator that can be used for a wide range of powerful applications including sensitivity analysis, ensemble prediction and data assimilation. The adjoint of ROMS, and its associated tangent linear and representer models are available, and a suite of tools that utilize these variants of ROMS have been developed. In this talk I will review this new ROMS tool-box which includes tools for computing eigenvectors, singular vectors, and stochastic optimals, and for performing 4D variational data assimilation.