Model sensitivity experiments on data assimilation, downscaling and tides for the representation of the Cape São Tomé Eddies

• Published in: Ocean dynamics Vol. 70; no. 1; pp. 77 - 94
• Main Authors: Santana, Rafael, Costa, Filipe B., Mignac, Davi, Santana, Alex N., Vidal, Vitor F. da S., Zhu, Jiang, Tanajura, Clemente A. S.
• Format: Journal Article
• Language: English
• Published: Berlin/Heidelberg Springer Berlin Heidelberg 2020; Springer Nature B.V
• Subjects: 25-28 June 2018; Algorithms; Atmospheric Sciences; Brazil; Brazil Current; Computational fluid dynamics; Computer simulation; Data assimilation; Data collection; Earth and Environmental Science; Earth Sciences; Eddies; Energy levels; Errors; Fluid- and Aerodynamics; Geophysics/Geodesy; Kinetic energy; Monitoring/Environmental Analysis; Oceanography; Profiles; Remote sensing; Representations; Resolution; Salinity; Santos; Sea level; Sea level anomalies; Sea surface; Sea surface temperature; Sensitivity; Surface temperature; Tides; Topical Collection on the 10th International Workshop on Modeling the Ocean (IWMO); Turbulent kinetic energy; Vortices; Downscaling; Tides; Cape São Tomé Eddy; Data assimilation
• ISSN: 1616-7341; 1616-7228
• DOI: 10.1007/s10236-019-01307-w

The impacts of data assimilation, downscaling, and tidal forcing were investigated with focus on the representation of the Cape São Tomé Eddy (CSTE). Sea level anomaly (SLA), sea surface temperature (SST), and temperature and salinity (TS) profiles were assimilated into a HYCOM nested grid system composed by three grids with horizontal resolutions from 1/4° to 1/24°. Tides were included in the highest resolution domain, which covers the Brazil Current (BC) northern region. Sensitivity experiments were conducted using the high-resolution domain when each observation type was assimilated alone and all together. In addition, the non-tidal and mid-resolution (1/12°) assimilative models were used in the comparison to evaluate the impacts of tides and downscaling. All assimilated observations positively contributed to reducing the temperature errors in the upper 500 m. However, the salinity was degraded in these depths by the assimilation of SST and SLA alone. Assimilation of all observations together decreased T and S errors by 34% and 17%, respectively, as well as raised (and reduced) in 77% (60%) the correlation (deviation) of SLA (SST) between remotely sensed observations when compared to the control run without assimilation. Six observed eddies were objectively compared against the simulated eddies using an eddy tracking algorithm. Downscaling to the higher resolution model was crucial to better represent the observed eddies, as well to increase the simulated energetic levels in the studied region. The inclusion of tides improved the turbulent kinetic energy levels and the CSTE SLA amplitude simulation. SLA data assimilation is of utmost importance for the representation of the CSTE. However, the assimilation of SST and mostly TS was also relevant for correcting the thermohaline field and enabling the correct eddy translation. Finally, possible mechanisms were discussed to explain the CSTE northward migration, against the BC main flow.