Elsevier

Ocean Modelling

Volume 167, November 2021, 101889
Ocean Modelling

The impact of ocean data assimilation on the simulation of mesoscale eddies at São Paulo plateau (Brazil) using the regional ocean modeling system

https://doi.org/10.1016/j.ocemod.2021.101889Get rights and content

Highlights

  • Ocean currents in the São Paulo plateau are driven by mesoscale eddies and dipoles.

  • Local and remote forcing contribute to generate eddy dipoles at São Paulo Plateau.

  • Ocean model results were evaluated with measured current data at São Paulo plateau.

  • Non-assimilative models showed limitations to simulate ocean eddies in the plateau.

  • Data assimilation is critical to simulate mesoscale eddies at the São Paulo plateau.

Abstract

This work presents simulations of ocean circulation in the southwestern South Atlantic Ocean with emphasis on the São Paulo Plateau region, a unique physiographic feature at the foot of the continental slope offshore southeast Brazil, where the flow is dominated by ocean eddies and dipoles. Surface dynamics south of 23°S presents a heterogeneous circulation pattern intimately related to the presence of mesoscale features, such as meanders of the Brazil Current along the continental slope, and eddies that interacts with deep ocean features. The reproduction of these eddies at the correct time and spatial structure are quite challenging in the context of a non-assimilative model, considering the limitations imposed by the model’s numerical schemes, physical parameterizations, boundary conditions and surface forcing. In this sense, the Regional Ocean Modeling System (ROMS) 4-dimensional Variational Data Assimilation (4D-Var) module was used to increase model skill and minimize deviations from oceanic observations. Weekly observations of Sea Surface Height (SSH), daily Sea Surface Temperature (SST) fields and in situ vertical profiles of temperature and salinity from multiple sources were used in consecutive assimilation cycles to produce an accurate representation of ocean features and their variability. Both the assimilative and free run were evaluated against observations, including in situ current measurements. Results showed significant improvements of the assimilation procedure on both assimilated and non-assimilated variables, with RMSE reductions of 27% and 47% for SST and SSH, respectively, and 51% for current speed measurements during a dipole event in the São Paulo Plateau. The assimilative run was able to reproduce a remarkable event of mesoscale eddies interaction (dipole) and consequent current intensification, leading to important improvements on model skill. The correct use of observational information of open ocean surface features through data assimilation is a key factor for a proper simulation of such eddy dipole events.

Introduction

The São Paulo plateau is a unique physiographic feature of the southeast Brazilian continental margin. It is a low declivity region located between the 2000 m and 3000 m isobaths (approximately 2 degrees apart) and zonally from 21°S, just south of the Vitoria-Trindade Ridge, to 28°S (de Almeida and Kowsmann, 2016, Kumar et al., 1977, Kumar and Gambôa, 1979). This study is concerned with the ocean circulation in the São Paulo plateau region south of 23°S, approximate latitude of Cape Frio, where the coastline orientation has an abrupt change from NE-SW to E-W (Fig. 1). In this southern stretch, the plateau has a large width, greater than 300 km on average. It is a region of intense mesoscale eddies activity (Andrioni et al., 2012).

The surface ocean circulation off southeast Brazil is marked by the presence of the Brazil Current (BC), the western boundary current associated with the Sverdrup balance of the wind-driven South Atlantic Subtropical Gyre. Originated from the bifurcation of the South Equatorial Current (SEC), the BC flows southward approximately following the continental slope (Stramma and England, 1999). Along its path, the BC develops intense mesoscale activity and large meanders that sometimes encloses eddies. The mechanism of baroclinic instability is one of the main drivers of such mesoscale activity (da Silveira et al., 2008, Rocha et al., 2014). The most recurrent eddies of this kind are found off Cape São Tomé (22°S) and off Cape Frio (23°S). Feature oriented numerical modeling has been applied to study these eddies (Calado et al., 2008, Calado et al., 2006) reinforcing the point that the use of observational data (hydrographic, current meter and satellite data) and numerical models are powerful tools to understand and study these mesoscale features along the southeast Brazilian continental slope.

South of 23°S, on the so-called Southeast Brazilian Bight (SBB), which extends from Cape Frio to Cape Santa Marta (28°S), the BC meanders and its eddies can induce shelf break upwelling events (Campos et al., 2000). Distinct methods were used to evaluate the Brazil Current structure and velocities on this region (Biló et al., 2014). Observational data, both direct measured ADCP current velocities as well as derived geostrophic velocities from CTD and XBT data along transects, have shown that the BC meanders along its path on the continental slope from 23°S to 28°S (Belo, 2011, Biló et al., 2014), but it does not dominate the deep ocean circulation on the São Paulo plateau further offshore. The ocean circulation on this plateau was evaluated by Belo (2011) with a series of measured current and hydrographic transects, revealing a pattern of successive eddy structures. These vortical features can sometimes be enhanced by the Cape Frio cyclonic eddy coupled with westward propagating anticyclonic features, leading to the formation of an eddy dipole structure and generating a strong intensification of surface currents in its middle section. The dipole structure commonly moves in the south-southwest direction and significantly changes the circulation pattern of the São Paulo plateau. This seems to be a recurrent feature as its surface signature can be frequently observed in satellite derived sea surface height (Andrioni et al., 2012). The plateau acts as a natural eddy corridor where dipoles gradually move westward, as shown in Fig. 2, where a sequence of snapshots of sea surface height and geostrophic velocity fields during August and September, 2010, is presented. An intense cyclone centered at 25°S, 43°W in the first snapshot is joined by an anticyclone giving rise to a strong dipole. Its bounds are roughly indicated in Fig. 2 by the black dashed line circle. Its influence on currents behavior, as will be shown later, could be captured by a current meter mooring, the P1 mooring, indicated by the red square.

The purpose of this work is to assess the impact of data assimilation on the simulation of eddy dipoles at the São Paulo Plateau, and not to identify their possible local or remote driving mechanisms. However, besides the local baroclinic instability mechanism discussed earlier, it has been observed that the arrival of westward propagating Agulhas Current eddies can contribute to generate enhanced dipole structures (Guerra et al., 2018). Also, the vorticity contribution of Rossby waves reaching the western boundary of the South Atlantic can contribute to the modulation of the BC transport as well as its meandering and eddy activity (Majumder et al., 2019).

Considering that the generation of dipole structures may depend on the conjugation of many distinct mechanisms, a precise reproduction of these features in time and space is a challenging task for a non-assimilative model. In fact, inaccuracies in the bathymetry representation, initial conditions, lateral boundary conditions and surface forcing, as well as limitations of sub-grid processes parametrization may introduce large uncertainties (Lima et al., 2019) and contribute to deviations of the numerical simulations from the ocean real state. Despite the limitations imposed by the numerical nature of the simulation, there must exist an optimum state or initialization field that minimizes the differences between model and observations. Finding these states and/or initial conditions is the core of the data assimilation methods and it is a requirement, if one is interested in simulating and forecasting high variability mesoscale features. Other studies have shown a positive impact of assimilation on the simulation of the ocean circulation at the Southeast Brazilian oceanic region (da Rocha Fragoso et al., 2016, Lima and Tanajura, 2013, Tanajura et al., 2016, Tanajura et al., 2015).

Due to the importance of data assimilation to operational oceanography and physical oceanography, the Oceanographic Modeling and Observation Network (REMO, Portuguese acronym for Rede de Modelagem e Observação Oceanográfica) was established in 2007 (Lima et al., 2013). The main goal was to implement an operational forecasting system in the Brazilian Navy Hydrography Center (CHM) with focus on the South Atlantic. Efforts were first dedicated to build a data assimilation system into the Hybrid Coordinate Ocean Model (HYCOM) (Santana et al., 2020, Tanajura et al., 2020). Another effort is also being conducted by the present work under REMO to explore the Regional Ocean Modeling System (ROMS) with the 4-dimensional variational data assimilation (4D-Var) method (Moore et al., 2011b) in the southwestern Atlantic. Several works have demonstrated this system’s capacity to produce accurate representation and forecasts in regions of high oceanic variability (da Rocha Fragoso et al., 2016, Lee et al., 2018, Moore et al., 2011a, Powell et al., 2008).

The present work aims to describe the implementation of ROMS primal formulation of incremental strong constraint 4D-Var over the southwestern Atlantic to investigate the ocean circulation at the São Paulo plateau region, particularly to simulate its characteristic mesoscale eddies. A particular event of an intense observed dipole structure that occurred in September 2010 is used to assess the impact of the data assimilation in model skill, especially its ability to reproduce the dipole with the correct spatial and time scales as well as their associated intense currents.

The paper is organized as follows. Section 2 describes the model setup, the data assimilation system and the assimilated observations. Section 3 presents simulation results with particular emphasis in model skills and in improvements achieved with assimilation, including comparisons with non-assimilated variables and observations, as well as global reanalysis products. Section 4 summarizes the main aspects and conclusions of this study.

Section snippets

Model description and free run setup

The model employed in this study is the Regional Ocean Modeling System (ROMS) (revision 904) along with the incremental strong constrain 4D-Var, better described in the next section. ROMS is a split-explicit, free-surface, 3D primitive equation and terrain following vertical coordinate oceanic model and has been widely use by the oceanographic community. A comprehensive description of ROMS and its 4D-VAR formulations can be found in Shchepetkin and McWilliams, 2005, Shchepetkin and McWilliams,

Assimilation performance

To monitor the operation of the I4D-Var assimilation procedure, it is useful to analyze the cost function over consecutive cycles. Fig. 7a shows the behavior of the TLM cost function during inner loop iterations of all assimilative cycles and Fig. 7b the final TLM (J) and NLM (JNL) cost functions at the end of the minimization algorithm. The variables are normalized by their values at the beginning of each cycle.

A decrease in J is achieved in all cycles indicating that the system is behaving

Conclusions

This paper presents an application of the Regional Ocean Modeling System (ROMS) to simulate the ocean circulation in the southwestern Atlantic Ocean, within a domain covering the entire south-southeast Brazilian shelf. A hindcast for the year of 2010 using ROMS I4D-Var assimilation module was produced assimilating satellite SSH, SST and in situ TS vertical profiles. Results were compared against both assimilated and non-assimilated variables to assess the assimilation convergence and evaluate

CRediT authorship contribution statement

Thiago Pires de Paula: Methodology, Software, Data curation, Visualization, Writing – original draft. Jose Antonio Moreira Lima: Conceptualization, Supervision, Data curation, Writing – original draft. Clemente Augusto Souza Tanajura: Conceptualization, Writing – review & editing. Marcelo Andrioni: Software, Data curation, Writing – review & editing. Renato Parkinson Martins: Project Administration, Resources, Writing – review & editing. Wilton Zumpichiatti Arruda: Supervision, Resources,

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgments

The authors acknowledge Petróleo Brasileiro S.A. (PETROBRAS) for providing funding, computational resources, observed currents and TS profiles measurements in Santos Basin to execute and validate our simulations, and the Centro de Hidrografia da Marinha (CHM) and the Instituto de Física of Universidade Federal da Bahia (UFBA) for kindly supplying qualified data from Argo floats assimilated into ROMS. We also thank the anonymous reviewers that made valuable comments and suggestions to improve

References (67)

  • BroquetG. et al.

    Application of 4D-variational data assimilation to the california current system

    Dyn. Atmos. Ocean

    (2009)
  • BroquetG. et al.

    Ocean state and surface forcing correction using the ROMS-IS4DVAR data assimilation system

    Mercat. Ocean Q. Newsl.

    (2009)
  • CaladoL. et al.

    A parametric model for the Brazil current meanders and eddies off southeastern Brazil

    Geophys. Res. Lett.

    (2006)
  • CaladoL. et al.

    Feature-oriented regional modeling and simulations (FORMS) for the western South Atlantic: Southeastern Brazil region

    Ocean Model.

    (2008)
  • CamposE.J.D. et al.

    Shelf break upwelling driven by Brazil current cyclonic meanders

    Geophys. Res. Lett.

    (2000)
  • ChapmanD.C.

    Numerical treatment of cross-shelf open boundaries in a barotropic coastal ocean model

    J. Phys. Ocean

    (1985)
  • ChinT.M. et al.

    A multi-scale high-resolution analysis of global sea surface temperature

    Remote Sens. Environ.

    (2017)
  • DaherV.B. et al.

    Regional altimetry product to the area of interests of the oceanographic modeling and observation network (REMO)

    Rev. Pesqui. Nav.

    (2014)
  • DonlonC.J. et al.

    The operational sea surface temperature and sea ice analysis (OSTIA) system

    Remote Sens. Environ.

    (2012)
  • FairallC.W. et al.

    Cool-skin and warm-layer effects on sea surface temperature

    J. Geophys. Res. C Ocean

    (1996)
  • FairallC.W. et al.

    Bulk parameterization of air-sea fluxes for tropical ocean global atmosphere coupled-ocean atmosphere response experiment

    J. Geophys. Res. C Ocean

    (1996)
  • FlatherR.A.

    A tidal model of the north-west European continental shelf

    Mem. Soc. R. Sci. Liege, Ser.

    (1975)
  • Fox-KemperB. et al.

    Challenges and prospects in ocean circulation models

    Front. Mar. Sci.

    (2019)
  • GoesM. et al.

    Long-term monitoring of the Brazil Current Transport at 22°S from XBT and altimetry data: Seasonal, interannual, and extreme variability

    J. Geophys. Res. Ocean

    (2019)
  • GuerraL.A.A. et al.

    On the translation of agulhas rings to the western south atlantic ocean

    Deep. Res. Part I Oceanogr. Res. Pap.

    (2018)
  • HersbachH. et al.

    The ERA5 global reanalysis

    Q. J. R. Meteorol. Soc.

    (2020)
  • KumarN. et al.

    Evolution of the São Paulo Plateau (southeastern Brazilian margin) and implications for the early history of the South Atlantic

    Bull. Geol. Soc. Am.

    (1979)
  • KumarN. et al.

    Geologic history and origin of sao paulo plateau (southeastern Brazilian margin)

  • LeeJ.H. et al.

    4DVAR data assimilation with the regional ocean modeling system (ROMS): Impact on the Water Mass Distributions in the Yellow Sea

    Ocean Sci. J.

    (2018)
  • LelloucheJ.M. et al.

    Recent updates to the copernicus marine service global ocean monitoring and forecasting real-time 1g 12° high-resolution system

    Ocean Sci.

    (2018)
  • LimaJ.A.M. et al.

    Design and implementation of the oceanographic modeling and observation network (REMO) for operational oceanography and ocean forecasting

    Rev. Bras. Geofis.

    (2013)
  • LimaL.N. et al.

    An investigation of ocean model uncertainties through ensemble forecast experiments in the southwest atlantic ocean

    J. Geophys. Res. Ocean

    (2019)
  • LimaL.N. et al.

    A study of the impact of altimetry data assimilation on short-term predictability of the hycom ocean model in regions of the tropical and south atlantic ocean

    Rev. Bras. Geofis.

    (2013)
  • Cited by (6)

    • The Brazil current cyclonic meandering and shelf-slope water exchanges at 27°S–31°S

      2024, Deep-Sea Research Part I: Oceanographic Research Papers
    View full text