FEMSECT: an Inverse Section Model Based on the Finite Element Method

• Published in: Journal of Geophysical Research
• Main Authors: Losch, Martin, Sidorenko, Dmitry, Beszczynska-Möller, Agnieszka
• Format: Article in Journal/Newspaper
• Language: unknown
• Published: 2005
• Subjects: Fram Strait
• Online Access: https://epic.awi.de/id/eprint/12030/; https://epic.awi.de/id/eprint/12030/1/Los2005a.pdf; https://doi.org/10.1029/2005JC002910; https://hdl.handle.net/10013/epic.22466; https://hdl.handle.net/10013/epic.22466.d001

A new inverse model is presented for the analysis of hydrographic section data in conjunction with velocity measurements. The model offers advantages over commonly applied interpolation techniques because it combines data and physical assumptions such as geostrophic balance in the framework of a finite element discretization. Specifically a quadratic objective function of model-data misfits is minimized to give estimates of transports together with formal error estimates. The finite element method allows the accurate representation of highly irregular bottom topography and ensures consistent interpolation of model variables to measurement points. The model is called FEMSECT for Finite Element Method SECTion model. FEMSECT also gives improved flexibility and performance over standard box models by allowing dynamic adjustment of the model variables temperature and salinity. Idealized test cases illustrate that the finite element methods solve the thermal wind equations far more accurately than standard finite difference methods, especially in the presence of steep topography. For a more realistic test, FEMSECT is applied to hydrographic CTD-section data and moored-instrument current meter measurements from an array in the Fram Strait. Transport estimates by FEMSECT prove to be more robust and less sensitive to the spatial data resolution than estimates by a conventional interpolation method that only uses information from moored instruments. FEMSECT is available as a highly portable Matlab code and can be run on an ordinary desktop computer.