You are on page 1of 107

NEAR-INERTIAL OSCILLATIONS AND COASTAL TRAPPED WAVES

ALONG THE SOUTHWESTERN ATLANTIC CONTINENTAL SHELF

Pedro Paulo de Freitas

Tese de Doutorado apresentada ao Programa de


Pós-graduação em Engenharia Oceânica,
COPPE, da Universidade Federal do Rio de
Janeiro, como parte dos requisitos necessários à
obtenção do título de Doutor em Engenharia
Oceânica.

Orientadores: Afonso de Moraes Paiva

Mauro Cirano

Rio de Janeiro
Março de 2021
NEAR-INERTIAL OSCILLATIONS AND COASTAL TRAPPED WAVES
ALONG THE SOUTHWESTERN ATLANTIC CONTINENTAL SHELF

Pedro Paulo de Freitas

TESE SUBMETIDA AO CORPO DOCENTE DO INSTITUTO ALBERTO LUIZ


COIMBRA DE PÓS-GRADUAÇÃO E PESQUISA DE ENGENHARIA DA
UNIVERSIDADE FEDERAL DO RIO DE JANEIRO COMO PARTE DOS
REQUISITOS NECESSÁRIOS PARA A OBTENÇÃO DO GRAU DE DOUTOR EM
CIÊNCIAS EM ENGENHARIA OCEÂNICA.

Orientadores: Afonso de Moraes Paiva


Mauro Cirano

Aprovada por: Prof. Afonso de Moraes Paiva


Prof. Mauro Cirano
Prof. Marcos Nicolas Gallo
Prof. Alexandre Macedo Fernandes
Prof Marcelo Dottori

RIO DE JANEIRO, RJ – BRASIL


MARÇO DE 2021
Freitas, Pedro Paulo de
Near-inertial oscillations and Coastal Trapped Waves
along the southwestern Atlantic continental shelf / Pedro
Paulo de Freitas. – Rio de Janeiro: UFRJ/COPPE, 2021.
XVI, 91 p.: il.; 29,7 cm.
Orientadores: Afonso de Moraes Paiva
Mauro Cirano
Tese (doutorado) – UFRJ/ COPPE/ Programa de
Engenharia Oceânica, 2021.
Referências Bibliográficas: p. 79-91.
1. Superinertial/Subinertial variability. 2. Continental
shelf waves. 3. Synoptic winds. I. Paiva, Afonso de
Moraes et al. II. Universidade Federal do Rio de Janeiro,
COPPE, Programa de Engenharia Oceânica. III. Título.

iii
Dedico à minha mãe Célia Freitas.

iv
Acknowledgements

First, I thank my advisors who played a fundamental role in my learning and


professional growth during my doctorate. I thank my advisor Afonso Paiva for guiding
me during these years and for the discussions about thesis research, in which I always
learned something new and I was challenged to improve my analysis/results. I thank my
advisor Mauro Cirano for guiding me during these year, assisting me in the organization
of the research analysis and helping me to structure the thesis in the form of scientific
articles. I extend my thanks to all those from Physical Oceanography Laboratory –
LOF/COPPE who conceived, planned and executed the mooring NAT (part of the INCT
PROOCEANO project); Afonso Paiva, Vladimir Costa, Guilherme Mill, Fernando Dix,
Mariela Gabioux and Natália Dávila. I thank Felipe Amorim who gave much needed
assistance in developing Chapter 2. I thank Bruna Reis who helped me with the
development of Chapter 3. I thank all my friends from Physical Oceanography
Laboratory, especially Vladimir Costa, Guilherme Mill, Fernando Dix, André Grijo,
Luana Bueno, Nilson Moreira, Mariela Gabioux and Bruno Siqueira, for all their shared
knowledge. I thank Cristina Coelho, Lucianita Silva and Marise Santos for helping me
during the doctoral protocols. I thank all my friends from Dynamics Laboratory of
Cohesive Sediments (Luana Gallo, Cristina Coelho, Ernesto Molinas, Bruno Moreira)
for all their support during my doctorate. I thank the Ocean Observation and Modeling
Network (REMO) Project for providing the modeling results. I acknowledge the
Brazilian Navy for providing the oceanographic data from PNBOIA/GLOSS/BRAZIL.
I acknowledge the scholarship provided by the Coordenação de Aperfeiçoamento de
Pessoal de Nível Superior (CAPES) Foundation. Finally, I thank my family, Célia
Freitas, Andreia Alves and Ana Sofia, for their unconditional support.

v
Resumo da Tese apresentada à COPPE/UFRJ como parte dos requisitos necessários para
a obtenção do grau de Doutor em Ciências (D.Sc.)

OSCILAÇÕES QUASE-INERCIAIS E ONDAS CONFINADAS COSTEIRAS AO


LONGO DA PLATAFORMA CONTINENTAL DO ATLÂNTICO SUDOESTE

Pedro Paulo de Freitas

Março/2021

Orientadores: Afonso de Moraes Paiva


Mauro Cirano
Programa: Engenharia Oceânica

Este trabalho investiga o papel que as Oscilações Quase-Inerciais (OQI) e Ondas


Confinadas Costeiras (OCC) desempenham na hidrodinâmica da margem continental
sul-sudeste brasileira. Baseado em seis fundeios com medições de velocidade na quebra
da plataforma continental entre 31.5°S e 16°S, este estudo revela que as OQI
apresentam alta energia e importância relativa para a variância das correntes na camada
superior dos oceanos em aproximadamente 30°S. Os eventos de OQI em Cabo Frio
apresentam propagação vertical de energia que pode envolver os primeiros 175 m da
coluna de água e estrutura vertical ligada à estrutura térmica da coluna de água. Baseado
em dados in situ de velocidade na quebra da plataforma e de altura da superfície do mar
na costa combinado com resultados de modelagem numérica, este trabalho revela uma
ampla área de geração de OCC, desde a plataforma continental da Patagônia até a
porção sul da costa brasileira. A energia das OCC no nível do mar na costa é alta na
porção norte da plataforma da Patagônia e diminue significtivamente ao norte de 22°S,
com grande dissipação de energia no banco de Abrolhos. O parâmetro de estratificação
(S) indica que a plataforma continental brasileira é barotrópica. Entretanto, a
importância dos modos baroclínicos das OCC aumenta na região de desaceleração das
ondas devido forte estratificação.

vi
Abstract of Thesis presented to COPPE/UFRJ as a partial fulfillment of the
requirements for the degree of Doctor of Science (D.Sc.)

NEAR-INERTIAL OSCILLATIONS AND COASTAL TRAPPED WAVES


ALONG THE SOUTHWESTERN ATLANTIC CONTINENTAL SHELF

Pedro Paulo de Freitas

March/2021

Advisors: Afonso de Moraes Paiva


Mauro Cirano
Department: Ocean Engineering

This study investigates the role that Near-Inertial Oscillations (NIO) and Coastal
Trapped Waves (CTW) play in the hydrodynamics of the South-Southeastern Brazilian
continental shelf. Based on six current meter mooring data located at the continental
shelf break between 31.5°S and 16°S, this study reveals that the NIO present the highest
energy and the highest relative contribution to the variance of surface currents in the
upper layer of the ocean at approximately 30°S. The NIO events at the Cabo Frio
mooring present vertical propagation of energy that can involve the first 175 m of the
water column and vertical structure tied to the thermal structure of the water column.
Based on in situ measurements of shelf-break currents and sea surface height (SSH) on
the coast combined with numerical modeling results, this study reveals a wide area of
generation of CTW, ranging from Patagonian continental shelf to the southern portion of
the Brazilian coast. CTW energy is high in SSH in the northern portion of the
Patagonian shelf and decreases significantly north of 22°S, with high energy dissipation
at Abrolhos Bank. The propagation speed is significantly reduced north of 24°S due the
narrowing of the continental shelf. The stratification parameter (S) indicates that the
Brazilian continental shelf is barotropic. However, the importance of the baroclinic
modes of CTW increases in the region of wave deceleration due strong stratification.

vii
Contents

List of Figures x

List of Tables xv

List of Acronyms xvi

1 Introduction 1

2 Near-inertial oscillations along the Brazilian continental shelf break 7

2.1 Introduction……………………………………………………………… 7

2.2 Data and Methods………………………………………………………... 9

2.3 Results and Discussion…………………………………………………... 13

2.3.1 Spectral description…………………………………………… 13

2.3.2 Representative near-inertial events…………………………… 17

2.3.3 NIO at the Cabo Frio upwelling system………………………. 20

2.4 Conclusions……………………………………………………………… 27

3 Coastal Trapped Waves propagation along Southwestern Atlantic


29
Continental Shelf

3.1 Introduction……………………………………………………………… 29

3.2 Data and Methods……………………………………………………….. 32

3.2.1 In situ data…….…………………………………………………. 32

3.2.2 Numerical modeling.…………………………………………….. 35

viii
3.2.3 Analysis………………………………………………………….. 36

3.3 Results and Discussion…………………………………………………... 39

3.3.1 CTW propagation from in situ data……………………………… 39

3.3.1.1 Sea level data at the coast………………………………… 39

3.3.1.2 ADCP data at the continental shelf-break/slope region…... 47

3.3.2 Dynamic characterization of the CTW propagation……………... 52

3.3.3 Cross-shore modal structure of the CTW .………………………. 62

3.4 Conclusions……………………………………………………………… 70

4 Conclusions 73

5 References 79

ix
List of Figures

Figure 2.1: Mooring positions and the bathymetry (shading) of the SSE Brazilian
Continental Shelf. The two dashed isolines represent the 200-m and 1000-m
isobaths. The top inbox shows the temporal coverage of each mooring. PNBOIA
measurements are shown in red and the NAT records are shown in blue. ………... 10
Figure 2.2: Depth-averaged rotary spectra of the raw velocity. The blue line is the
bandpass filter and its response function. The thick (thin) line represents the
anticyclonic (cyclonic) motions. The gray dashed vertical lines represent the tidal
constituents O 1 , K 1 , M 2 , and S 2 (left to right). The black dashed vertical line
is the inertial frequency at: [PS] 0.5516 cpd, [VI] 0.6811 cpd, [CF] 0.7986 cpd,
[SA] 0.8657 cpd, [SC] 0.9544 cpd, and [RG] 1.0456 cpd. See Figure 2.1 for the
geographic location of the stations presented in panels. ………………………….. 14

Figure 2.3: Examples of surface near-inertial events at the six locations presented
in Figure 2.1 of the filtered velocities. ……………………………………………. 16

Figure 2.4: Rotary spectra of the wind velocity extracted at the mooring locations
in Figure 2.1. The thick (thin) line represents anticyclonic (cyclonic) motions,
while the vertical dashed line is the inertial frequency. …………………………... 17

Figure 2.5: [a] Depth-averaged near-inertial energy of the high energy events in
the top 50 m of the water column at the mooring locations presented in Figure
2.1. [b] Relative importance of these near-inertial events in terms of the variance
of the currents. The vertical bars represent one standard deviation. ……………… 18
Figure 2.6: The mean maximum vertical shear of the high energy events in the
top 50 m of the water column at the mooring locations presented in Figure 2.1.
The vertical bars represent one standard deviation. ………………………………. 19

Figure 2.7: Vertical structure of a near-inertial event at Vitória (see Figure 2.1 for
location). [a] A profile of the Brunt-Väisälä frequency obtained from an ARGO
float. [b] The near-inertial velocity ellipses. [c] The vertical shear of the
horizontal velocity. The Brunt-Väisälä frequency increase below the depth of 30
m indicates the change in the stratification profile. This affects the vertical
velocity structure and marks a depth of greater vertical shear. …………………… 20

Figure 2.8: Time series of the near-inertial energy (in cm 2 s−2 ) at Cabo Frio (see
Figure 2.1 for location) during the years of [a] 2014 and [b] 2015. The inverted
triangles represent the near-inertial events, which are described in Table 2.3.……. 21

x
Figure 2.9: [a] Time series of the northward component of the near-inertial
oscillations of an event that started on the 2nd of November 2014 (see Table 2.3).
[b] Wavelength and [c] phase and group velocity obtained based on a least
squares adjustment of data for this near-inertial event. Positive (negative) values
in a are northward (southward)……………………………………………………. 23

Figure 2.10: [a] First EOF mode profiles of the alongshore velocity from to all
events indicated in Table 2.3 (gray lines), with the 22 Oct. 2015 event marked as
a black line. [b] Time evolution of the alongshore velocity of the 22 Oct. 2015
event. Positive (negative) values in b are northward (southward). ……………….. 24

Figure 2.11: Time series of the vertical distribution of the alongshore velocity [a,
d] and temperature [b, d] of two near-inertial events that occurred during the
periods 3 and 19 Jun. 2014 and 5 and the 17 Apr 2015, respectively (see Table
2.3). [c, f] Daily relative vorticity at the surface derived from AVISO for these
events. Positive (negative) values in a, d are northward (southward). ……………. 25

Figure 2.12: Determination of the observed frequency of the near-inertial events


(see Table 2.3) that occurred during [a] 3 and 19 Jun. 2014 and [b] 5 and 17 Apr.
2015 to illustrate the determination of a redshift and blueshift. The vertical
dashed lines represent the inertial frequency at Cabo Frio. ………………………. 26

Figure 3.1: Bathymetry of the Southwestern Atlantic continental shelf (shaded),


indicating the coastal sea level stations (GLOSS/UHSLC - green circles) and
mooring positions (PNBOIA/NAT - red and magenta stars, respectively). The five
isolines represent the 50 m, 200 m (yellow line), 1000 m, 2000 m and 3000 m
isobaths, respectively. The top left panel indicates the two HYCOM model
domains with 1/12° and 1/24° horizontal resolution. The acronyms for the
locations are indicated in Tables 3.1 and 3.2.……………………………………… 33

Figure 3.2: Two-months long band-pass filtered sea level time-series (see text for
details) for GLOSS/UHSLC (blue line) and HYCOM 1/12° (black line) at PD
(from January 1st, 2015 to March 1st, 2015) and HYCOM 1/24° at the remaining
stations: MP (from July 2nd, 2012 to September 1st, 2012), IM (from June 1st,
2009 to August 1st, 2009), CA (from August 1st, 2014 to October 1st, 2014), UB
(from March 1st, 2015 to May 1st, 2015), IF (from August 1st, 2010 to October
1st, 2010), MA (from April 1st, 2012 to June 1st, 2012) and SSA (from
September 1st, 2015 to November 1st, 2015) stations. The scatter plot shows a
strong correlation between GLOSS/UHSLC data and HYCOM, considering all
the available data period for each station (Table 1). The red line represents a
linear fit calculated with a least square adjust. The geographical locations of these
stations and their associated names are illustrated in Figure 3.1 and Table 3.1,
respectively………………………………………………………………………… 37

xi
Figure 3.3: Propagation of CTW along the Brazilian continental shelf based on
coastal sea-level records from SSA, MA, IF, UB, CA, IM, MP, and PD during
May/August 2014, which is the best period in terms of data quality and
synchronicity. The observations were bandpass filtered to eliminate tides and
oscillations with periods exceeding 40 days. The red lines indicate the phase
propagation of 4 CTW, named (from left to right) waves 1, 2, 3, and 4. The width
of continental shelf at each stations is: 325 km (PD), 230 km (MP), 128 km (IM),
175 km (CA), 125 km (UB), 94 km (IF), 62 km (MA) and 10 km (SSA). The
geographical locations of the stations and their names are illustrated in Figure 3.1
and Table 3.1, respectively. ………………………………………………………. 41

Figure 3.4: [a] Correlation coefficient (r) between the filtered sea level at the
coastal stations of PD (black line), MP (blue line) and IM (cyan line) and the
remaining stations located further north. The red line represents the correlation
coefficient between consecutive stations along the propagation of CTW. [b] Time
lagged cross-correlation between the coastal stations of PD (black line), MP (blue
line), and IM (cyan line) and the remaining stations located further north. The y
labels in these figures are not scaled due to the variable spatial distribution of the
data. This analysis was performed during the periods of May 17th, 2014 and
December 31th, 2014 for all stations except MA (22.23°S), which was performed
during May 17th, 2014 and August 07th, 2014, representing the best period in
terms of data quality and synchronicity. The geographical locations of the stations
and their names are illustrated in Figure 3.1 and Table 3.1, respectively. ………… 43

Figure 3.5: Wavelet analysis of the coastal sea level data at the MP, IM, CA, UB,
IF, and MA stations during 2014/2015, which is the best period in terms of data
quality and synchronicity.. The left panels indicate the power spectrum (Morlet),
where the color bar represents the normalized variance and the black contour
indicates the 95% confidence level. The hatched area represents the cone of
influence. The right panels indicate the global wavelet spectrum [cm²/cph] (blue
line) and the 95% confidence level (dashed line). For the stations with gaps in the
series, the global spectrum is the average of the different analysis periods. The
geographical locations of the stations and their names are illustrated in Figure 3.1
and Table 3.2, respectively. ……………………………………………………….. 45

Figure 3.6: Cross-spectrum between the coastal sea-level records in stations PD


(November 20th, 2011 to June 10th, 2013), MP (April 1st, 2010 to December
31th, 2018), IM (January 1st, 2008 to January 15th, 2011), CA (February 1st,
2014 to October 28th, 2016), UB (May 2nd, 2014 to January 26th, 2018), IF
(January 1st, 2008 to June 17th, 2018), MA (December 1st, 2011 to December
31th, 2012), and SSA (December 16th, 2010 to September 04th, 2017) and the
meridional component of wind at 57°W 40°S from MERRA2 reanalysis for the
same period. The gray line represents the 99% confidence interval. The
geographical locations of the stations and their names are illustrated in Figure 3.1
and Table 3.1, respectively.. ………………………………………………………. 48

xii
Figure 3.7: Comparison between the sea level from station IF (black line) and the
top 50 m depth-averaged meridional velocity component at the continental shelf-
break (red line) moorings. The cross-correlation between the time series shows
correlation coefficients (r) of 0.4 (RG), 0.7 (SC), 0.6 (SA), 0.6 (CF), 0.6 (VI) and
0.4 (PS) (from bottom to top). The time series are plotted with maximum
correlation lag (hours) correction of 45 (RG), 27(SC), 10(SA), 8 (CF), -42 (VI) and
-126 (PS). The geographical locations of the stations and their names are
illustrated in Figure 3.1 and Tables 3.1 and 3.2, respectively. ……………………. 49

Figure 3.8: Cross-spectrum between the sea level at the coast [station IF] and the
alongshore velocity at the continental shelf break at stations RG, SC, SA, VI, and
PS for the top 50 m of the water column and the top 200 m of the water column
in CF. The black line represents the 99% confidence interval. The geographical
locations of the stations and their names are illustrated in Figure 3.1 and Tables
3.1 and 3.2, respectively. ………………………………………………………….. 51

Figure 3.9: Hovmöller of the band-pass filtered (Butterworth filter) hourly SSH
along the 50 m isobath during 2015 from the HYCOM 1/12° simulation. The
color scale is limited to highlight the wave signals throughout the year. The black
contours indicate amplitudes larger than 10 cm.…………………………………... 53

Figure 3.10: Cross-spectrum between SSH along the 50 m isobath from 11 years
(2006-2016) of the HYCOM 1/12° simulation and the meridional component of
wind in 57°W, 40°S from MERRA2 reanalysis for the same period. The
confidence interval is 99%. The blank areas represent the coherence coefficient
without statistical confidence. …………………………………………………….. 54
Figure 3.11: Phase speed (in m s-1) of CTW evaluated from SSH at the 50 m
isobath and based on a 11-year (2006-2016) HYCOM 1/24° simulation
considering periods of larger (red line) and smaller (blue line) stratification. The
green dots represent the same property calculated using SSH from the sea level
stations presented in Figure 3.1 and Table 3.1. The yellow dots represent the
theoretical values based on free CSW theory using an exponential depth profile of
the continental shelf with a deviation bar associated with a variation of 10%
the width of the continental shelf. The black dotted line represents the shelf width
(in km). ……………………………………………………………………………. 56

Figure 3.12: [a] Spectral analysis (FFT) of the SSH at the 50 m isobath based on
11-year (2006-2016) HYCOM 1/12° (south of 45.18°S) and 1/24° (north of
45.18°S) simulations. [b] Associated cross-spectrum coherence coefficients and
[c] cross-spectrum lag between station MP and the other stations with a 99%
confidence interval. The red labels indicate values of lag in days. The blank areas
in [b] and [c] represent coherence coefficients and phase without statistical
confidence or phase propagating southward. ……………………………………... 58

xiii
Figure 3.13: Comparison between the frequency–wavenumber relation observed
in the cross-spectrum of the SSH (markers) along the 50 m isobath based on a 11-
year (2006-2016) HYCOM 1/24° simulation and the theoretical dispersion
relation of the first two modes of CSW (lines) derived from the exponential
topographic profiles in 3 different regions [25°S (black), 22°S (blue) and 21.5°S
(green)]. The first mode of each curve is associated with the highest frequencies.
The shadows indicate a variation of 10% of the continental shelf width. The
marks indicate waves with periods of 7 (square), 10 (circle), 18 (star), and 28
days (inverted triangle).……………………………………………………………. 61

Figure 3.14: (left painel) Latitudinal variation in the percentage of importance of


the statistical modes of the EOF analysis of the alongshore velocity along
selected cross-shore sections in the SS, SBB and ACR [mode 1 (blue line), mode
2 (black line) and mode 3 (red line)] based on a 11-year (2006-2016) HYCOM
1/24° simulation. (right panel) Location of the cross-shore sections of velocity…….. 63

Figure 3.15: Spectral analysis (FFT) of the time series of the first (left panel) and
second (right panel) EOF modes of the alongshore velocity at selected cross-
shore sections for the geographic region presented in Figure 3.14 and from the
11-year (2006-2016) HYCOM 1/24° simulation. The spectral power values are
presented as a percentage considering the first four modes. ……………………… 64

Figure 3.16: [a] Mean value of the CLARKE and BRINK (1985) criterion in the
SS, SBB and ACR based on 11 years (2006-2016) of HYCOM 1/24° simulation
and [b] the associated α values. The shadow region in the left panel indicates the
standard deviation. ………………………………………………………………… 66

Figure 3.17: (left panels) The first (a, g), second (b, h), and third (c, i) EOF modes
based on the analysis of the alongshore velocity at two selected cross-shore
sections located at the SBB shelf and based on the outputs of a 11-year (2006-
2016) HYCOM 1/24° simulation. (right panels). The first (e, k) and second (f, l)
baroclinic modes from BRINK and CHAPMAN’s (1987) linear model for the
same two cross-shore sections. The locations of the two cross-shore sections are
indicated in panels d and j. ………………………………………………………… 68

xiv
List of Tables

Table 2.1: Results of the rotary spectral analysis for the near-inertial band
presenting the Relative Frequency Shift (RFS) and the Near-Inertial Variance
(NIV) at the locations presented in Figure 2.1. ……………………………………. 13

Table2.2: Relative vorticity from HYCOM/NCODA Global 1/12° Reanalysis


(http://tds.hycom.org/thredds/catalogs/GLBv0.08/expt_53.X.html) at the locations
presented in Figure 2.1. ……………………………………………………………. 15
Table 2.3: Characterization of the near-inertial events in Cabo Frio (see Figure 2.1
for location. ………………………………………………………………………... 22
Table 3.1: Description of the coastal sea level data………………………………... 34

Table 3.2. Description of the velocity data from PNBOIA and NAT moorings……. 34

xv
List of Acronyms

ACR – Abrolhos-Campos Region


BC – Brazil Current
CSW – Continental Shelf Waves
CTW – Coastal Trapped Waves
ES – Eastern Shelf
NIO – Near-Inertial Oscillations
SBB – South Brazil Bight
SS – Southern Shelf
SSH – Sea Surface Heigh

xvi
Chapter 1

Introduction

The large-scale wind field associated with the South Atlantic Subtropical High

(SASH) and its variability associated with the propagation of synoptic systems are

important forcing mechanisms for the variability of sea surface height (SSH) and upper

ocean currents along the Brazilian continental shelf between 34°S e 15°S. The

anticyclonic winds associated with the SASH blow from east or northeast throughout

the year, whereas the propagation of the synoptic systems changes the wind direction to

southerly (CASTRO et al. 2006). These synoptic atmospheric systems have been

reported in the literature as the main forcing mechanisms of the Near-Inertial

Oscillations (NIO) and Coastal Trapped Waves (CTW) along the South/Southeast

Brazilian continental shelf (STECH and LORENZZETTI, 1992, CASTRO et al. 2006,

ASSIREU et al., 2017, DOTTORI and CASTRO, 2018). The NIO and CTW are

transient waves that play an important role in the dynamics of the continental shelf and

slope waters and may represent a significant fraction of the energy for superinertial

(periods of hours) and subinertial (periods of weeks) variability, respectively.

Direct observations of NIO show the signature of these oscillations in the

horizontal velocity fields in the form of anticyclonic movements, with average

1
amplitudes and spatial scales of the order of 10 cm s −1 and 10 km, respectively

(CHAIGNEAU et al., 2008, ELIPOT and LUMPKIN, 2008). These waves propagate

phase upward and energy downward, and contribute to the turbulent shear near the

surface, and to mixing in the ocean interior ocean (KUNDU 1976; D’ASARO et al.,

1995, ALFORD et al., 2017). The frequency of the NIO varies with latitude as a

function of the inertial frequency ( f ) and of the relative vorticity field of the subinertial

currents (ζ/2) which shifts the lower limit of the internal waveband from f to an effective

ζ
inertial frequency feff = f + (KUNZE 1985, ELIPOT et al. 2010, CHAVANNE et al.
2

2012). For the Southern Hemisphere (f < 0), if the relative vorticity is cyclonic

(anticyclonic), feff results in higher (lower) frequencies when compared with f, defining

the so-called blueshift (redshift). Therefore, boundary currents can play a crucial role in

the NIO characteristics in the coastal ocean.

The CTW are vorticity waves that propagate phase with the coast on the left (right)

in the southern (northern) hemisphere and are widely observed in continental shelves

around the world. These waves present alongshore spatial scales of 10 4 to 106 m, and

sea-level amplitudes of the order of 10-1-100 m, and play an important role in the

subinertial variability of SSH and currents on the continental shelf and slope region

(HUTHNANCE, 2001, HUGHES et al., 2019, WOODWORTH et al., 2019). In general,

CTW are forced by equatorial Kelvin waves (PIETRI et al., 2014, ILLIG et al., 2018) or

by synoptic atmospheric systems that propagate at mid-latitudes (MYSAK, 1980), and

the propagation of these waves transfers energy between different latitudes, and into the

ocean's interior (WISE et al., 2020).

2
The CTW signal in the SSH presents maximum amplitude at the coast and

decreases exponentially offshore, but can be detectable at the continental slope

(HUGHES and MEREDITH, 2006, ROUSSENOV et al. 2008). The cross-shore modal

structure of the CTW presents hybrid characteristics of propagation, from baroclinic

Kelvin waves to topographic Rossby waves, which are governed by the shape of the

continental shelf and slope (width and bottom declivity) and water column stratification

(WANG and MOOERS, 1976, HUTHNANCE, 1978, BRINK, 1991). Abrupt variations

in the width of the continental shelf can lead to energy scattering between different

wave modes, which can also be amplified due to an increase in stratification (WILKIN

and CHAPMAN, 1990).

STECH and LORENZZETTI (1992), using a barotropic hydrodynamic model,

investigated the hydrodynamic response of South Brazil Bight (SBB) region, between

Cabo de Santa Marta (28°40'S) and Cabo Frio (23°S), to the passage of synoptic

systems during winter. This study shows that anticyclonic wind rotation and the passage

of synoptic systems are favorable conditions for the generation of NIO. ASSIREU et al.

(2017), based on satellite-tracked drifters, characterized near-inertial motions in the

Brazil Current (BC) between 24°S e 36°S. This study shows that between 24°S e 36°S

NIO are generated at 4.7 to 15-day bursts, and can account for 45% of the variance of

the currents south of 28°S. In general, these studies have in common the use of

information (model and data) only at the surface, which leaves gaps in the

characterization of NIO along the water column. Besides, these studies do not address

the shelf break region, where part of the BC flows (e.g., SILVEIRA et al. 2000,

ASSIREU et al. 2003), and NIO may play a significant role in the fluxes between the

continental shelf and the adjacent ocean through modulation of the horizontal field of

3
currents. A few questions remain to be addressed regarding the NIO dynamics at the

Brazilian continental shelf between 32°S and 16°S:

1. How does the near-inertial energy in the upper layer of the ocean (mixed layer,

seasonal thermocline, and upper permanent thermocline) vary along the

Brazilian continental shelf-break?

2. What is the relative importance of NIO for the variance of currents at the

interface between the continental shelf and the adjacent ocean?

3. What are the characteristics of near-inertial currents along the water column

during representative events in terms of energy?

Previous studies of CTW propagation along the SBB have shown that remotely

forced Continental Shelf Waves (CSW), with periods between 6 and 12 days, modulate

the SSH and the magnitude and direction of the shelf currents (CASTRO and LEE,

1995, DOTTORI and CASTRO, 2018), and that such waves are primarily barotropic

(DOTTORI and CASTRO, 2009). FILIPPO et al. (2012) identified high energy and

coherence for periods between 3.5 and 28 days in SSH records taken over one year, at

four coastal stations between 43°S and 22°S. FRANÇA (2013), based on numerical

modeling, characterized the propagation of CTW along the South-Southeast continental

shelf of Brazil. The study described that these waves present a marked reduction in

phase speed propagation in Cabo Frio (northern limit of SBB) and Abrolhos bank. Also,

based on the stratification parameter (CLARKE and BRINK, 1985) and EOF analyses

of the filtered alongshore velocity, the study characterized these waves as predominantly

barotropic. However, during summertime in the region between Cabo Frio and Macaé

(upwelling system), the baroclinic component of the CTW increases. These results

presented by FRANÇA (2013) motivated this study to quantify the importance of the

4
topography and stratification of the continental shelf for the phase speed propagation,

and the modal structure of the CTW. Despite the relevant contributions, the spatial and

temporal limitations of the preview studies leave gaps associated with CTW

propagation:

1. What are the geographical limits of the generation area of the CTW that

propagate along the Brazilian continental shelf between 34°S and 11°S?

2. How does CTW energy vary as they propagate towards lower latitudes?

3. How does the phase speed of the CTW vary along the coast?

4. What is the importance of variations in the morphological and stratification

characteristics of the continental shelf for the phase speed propagation, and the

modal structure of the CTW?

Therefore, this present study aims to characterize the hydrodynamic variability

along the Brazilian continental shelf generated by NIO and CTW. The specific

objectives are:

i) To describe the spatial variability of the spectral characteristics at the inertial band.

ii) To quantify the importance of the NIO for the variance of the shelf and slope

currents.

iii) To investigate the vertical structure of the near-inertial events.

iv) To characterize the generation area of CTW that propagate in this region.

v) To describe the propagation of the CTW in terms of variations of SSH and currents.

vi) To investigate the importance of topography and stratification at the continental shelf

for the phase speed propagation of the CTW.

vii) To investigate the modal structure of the CTW.

5
To attain these objectives, the present study uses in situ measurements of currents

covering more than 2000 km of the Brazilian shelf-break region between ~32°S and

16°S, coastal sea level covering approximately 4500 km along the Brazilian coastline

between ~48°S and 13°S, and results from a high-resolution simulation of the HYCOM

hydrodynamic model. The velocity measurements at shelf-break are from 6 current

meter moorings located at the shelf break. Five moorings are part of the Global Ocean

Observing System (GOOS/BRAZIL/PNBOIA), and present measurements from the

surface to 50 m and one mooring was deployed by the Physical Oceanography

Laboratory (LOF/COPPE/UFRJ) at the Cabo Frio upwelling region, with measurements

along the entire water column. The SSH data are from 8 coastal stations, 6 of which are

from the Brazilian component of the Global Ocean Observing System

(GOOS/BRAZIL/GLOSS), and 2 were provided by the University of Hawaii. The

results used from the HYCOM model are from a modeling system developed at the

Physical Oceanography Laboratory (LOF/COPPE/UFRJ) within the context of the

Ocean Observation and Modeling Network (REMO) Project, which is a Brazilian

initiative in operational oceanography for the South Atlantic. A detailed description of

the data and the model is presented in the following chapters.

This thesis is structured into 4 chapters. Chapters 2 and 3 are presented in an article

format, with an introduction and detailed methodology for the study of NIO and CTW,

respectively. Chapter 4 presents the overall conclusions of this study.

6
Chapter 2

Near-inertial oscillations along the


Brazilian continental shelf break

2.1 Introduction
Near-inertial oscillations (NIO) are transient waves with frequencies close to the

inertial frequency and represent a dominant mode of high-frequency variability in the

ocean (ALFORD et al., 2016). These waves exhibit a prominent energy peak in the

internal wave spectrum (GARRETT and MUNK, 1975, GARRETT, 2001) and

combined to the internal tides account for approximately half of the energy required for

vertical mixing in order to maintain the Meridional Overturning Circulation (MOC)

(MUNK and WUNSCH, 1998, SIMMONS and ALFORD, 2012).

The local wind dynamics is crucial for NIO since the wind field oscillations act as

an important mechanism for the NIO generation (POLLARD and MILLARD, 1970,

D’ASARO, 1985). In the coastal region and near mid-latitudes (~ 30°), it is also

observed that diurnal variations in the land-sea breeze become more important for the

NIO, since these movements may resonate with the inertial frequency, resulting in an

increase of the inertial energy in the upper layer of the ocean (GILLE et al., 2005,

7
HYDER et al., 2011, NAM and SEND, 2013). While the wind is the most important

force for the generation of NIO, observational studies indicate that the variability of the

inertial energy is not always directly proportional to the intensity of the winds. In some

cases, weak winds are capable of generating more pronounced NIO when compared

with those generated by stronger winds (SUBBEESH and UNNIKRISHNAN, 2016,

CAO et al., 2018). In general, the transient action of NIO can generate intense currents

on the continental shelf that play an important role in the mixing and distribution of

biogeochemical properties along key areas, such as coastal upwelling regions

(SOBARZO et al., 2007, LUCAS et al., 2014).

Direct observations reveal that NIO present higher velocity amplitudes in the

continental shelf break region, where an inertial energy concentration occurs (CHEN et

al., 1996). Measurements at the external part of the New England continental shelf, for

example, indicate NIO with velocity amplitudes greater than 30 cm s −1, accounting for

up to 20% of the total current variance (SHEARMAN, 2005). RIVAS and PIOLA

(2005), for the continental shelf of Patagonia (43°S), describe NIO reaching velocity

amplitudes greater than 25 cm s−1 and with the first baroclinic mode dominating the

kinetic energy in the outer shelf. SOBARZO et al. (2007) investigated the importance of

the near-inertial band for the variance of currents in the Chile-Peru upwelling system

and computed NIO reaching velocities of 20 cm s−1 over a period of up to 10 days,

which can represent up to 61.5% of the variance of the current at the continental shelf

break.

Despite the fact that NIO have great relevance for high-frequency hydrodynamic

variability, only two studies are reported in the literature for the Brazilian continental

shelf region. STECH and LORENZZETTI (1992), using a hydrodynamic barotropic

8
model, investigated the effect of the passage of cold fronts in the South Brazil Bight.

The authors show that the anticyclonic wind rotation in the region is a favorable factor

for the generation of inertial movements and that strong NIO are generated after the

passage of these cold fronts. ASSIREU et al. (2017), based on drifters trajectories,

describe the near-inertial movements in the BC. The authors argue that the generation of

these movements is related to the passage of cold fronts, low-pressure systems, and the

sea breeze circulation. In addition, they quantify that the inertial band presents greater

energy to the south of 28°S, being responsible for 45% of the variance of the currents.

2.2 Data and Methods

The study of the NIO in the upper layer region along the Brazilian continental

shelf break was performed based on data collected in 6 current meter moorings located

approximately at the 200-m isobath between 16°S and 31.5°S (Figure 2.1). Five of these

locations are part of the Brazilian component of the Global Ocean Observing System

(GOOS/BRAZIL/PNBOIA, http://www.goosbrasil.org/), while the remaining location is

part of the INCT PROOCEANO project that was carried out in the Cabo Frio coastal

upwelling region (e.g., CAMPO et al., 2000, VALENTIN, 2001, CASTELAO and

BARTH, 2006). Each of the PNBOIA moorings was equipped with a 400-kHz acoustic

Doppler current profiler (ADCP), sampling current intensity and direction at hourly

intervals between 5.5 and 53 m with a vertical discretization of 2.5 m. On the other

hand, the NAT mooring, which is the mooring with the greater vertical coverage, has a

150-kHz ADCP and was also equipped with 14 temperature sensors sampling

temperature at every 30 min between 14 m and 185 m with vertical separation cells of 4

m and 10 m, respectively.

9
Figure 2.1: Mooring positions and the bathymetry (shading) of the SSE Brazilian
Continental Shelf. The two dashed isolines represent the 200-m and 1000-m isobaths.
The top inbox shows the temporal coverage of each mooring. PNBOIA measurements
are shown in red and the NAT records are shown in blue.

The raw velocity series of the first 50 m of the water column were analyzed

using the rotary spectral analysis in order to quantify the energy of the anticyclonic

movements of the NIO in the frequency domain (GONELLA, 1972). This analysis has

the purpose of capturing the inertial band signature. Relative frequency shift (RFS) and

near-inertial variance (NIV) were calculated from the depth-averaged spectrum

(ELIPOT et al., 2010). The RFS is defined as the percentage of the difference between

the observed inertial peak frequency (w) and the inertial frequency and characterizes the

10
blueshift/redshift conditions. The observed inertial peak frequency at each location was

quantified based on the maximum energy peak around of the inertial frequency (±0.15

cpd) in the average spectrum. The NIV is the integral of the inertial band spectrum and

represents the importance of NIO for high-frequency dynamics (ELIPOT and

LUMPKIN, 2008, ELIPOT et al., 2010).

The harmonic analysis was also used in the current series in order to remove the

tidal signal (SHEARMAN and LENTZ, 2004, SHEARMAN, 2005). The data was then

bandpass filtered (Butterworth filter) centered on the observed inertial peak frequency

of each station (± 0.15 cpd). Finally, a rotating mask was applied in the filtered near-

inertial series to ensure that only the anticyclonic motions were considered.

Considering that NIO are transients that increase the energy in the mixed layer

due to variations in the intensity and direction of the wind field (WEBSTER, 1968,

POLLARD, 1980), the approach adopted here was to characterize the importance of

NIO based on representative (high energy) near-inertial events only. Hence, the near-

inertial kinetic energy was calculated from the amplitudes of the filtered series of the

velocity components and the identification of their associated positive and negative

peaks. Based on the time series of the energy at the surface, a near-inertial event was

defined as the moment when the energy is greater than the median of the series. The

choice of median as an identification criterion was based on the heterogeneity of the

temporal coverage of each mooring and on the temporal variability of the near-inertial

energy. After the identification of these events, the following quantities were calculated:

(i) the average near-inertial energy, (ii) the relative importance of the near-inertial band

for the variance of the currents, and (iii) the average maximum shear within the first 50

m of the water column.

11
For the characterization of the vertical structure of NIO events, only the NAT

mooring in Cabo Frio (Figure 2.1) was used, since unlike the PNBOIA measurements

that are restricted to the first 50 m of the water column, the NAT mooring has a greater

vertical coverage, allowing better estimates of the phase velocity (C z), the group

velocity (Cg), and the vertical wavelength of the waves (λ). For this characterization of

these events, in addition to the criterion of energy, the condition of existence of

propagation of the phase of the oscillation was also established. For each event, the

wave parameters were quantified using the lagged cross-correlation technique described

in Kundu (1976) and the statistical modes of the EOF. In order to estimate the wave

parameters, a matrix of distances between the depths was constructed and for each pair

of depth, the maximum correlation lag between the velocities was calculated. The C z of

each event was estimated by a least squares adjustment between the depth pair distances

and the lags of maximum correlation. The Cg was calculated from Equation 2.1 below:

[( ) ]
2
f
C g=C z −1 (2.1)
w

f
The ratio was quantified from the least squares fit between the lengths of the
w

minor axis and the major axis of the velocity ellipses for each depth. The vertical

wavelength was calculated from the least square fit between the depth pairs and the

angle resulting from the phase propagation of the alongshore velocity component at

different depths (d) (average angular displacement or turning) according to Equation 2.2

below:

⟨ u1 ⟩ ⟨ u2 ⟩
d=arccos
(√ (⟨ u1 ⟩ 2 ⟨ u2 ⟩ 2) ) (2.2)

12
2.3 Results and discussion

In this section, we present the results and discussion of the spectral

characteristics of the near-inertial band, the importance of the NIO for the variance of

the currents in the contiental shelf break, and the vertical structure of the near-inertial

events in the Cabo Frio upwelling system.

2.3.1 Spectral description

Figure 2.2 shows the mean vertical rotary spectrum of the raw velocity for each

location. The stations south of CF present a prominent peak of energy around the

inertial frequency whereas toward low latitudes the near-inertial energy decreases and

the shape of the near-inertial band becomes flat. This spatial variability of the near-

inertial band occurs because the inertial energy is higher at midlatitudes under storm

tracks (ELIPOT et al., 2010). Table 2.1 summarizes the description of the spectral

parameters RFS and NIV obtained from Figure 2.2.

Table 2.1: Results of the rotary spectral analysis for the near-inertial band presenting the
Relative Frequency Shift (RFS) and the Near-Inertial Variance (NIV) at the locations
presented in Figure 2.1.

Inertial Freq.[cpd] / Observed Freq. [cpd] / RFS NIV


Inertial Period Observed period [cpd] [cm² s-2]
[hours] [hours]
PS (16°S) 0.5516 / 43.5 0.4868 / 49.1 -0.11 1486
VI (19.9°S) 0.6811 / 35.2 0.6379 / 37.6 -0.06 735
CF (23.5°S) 0.7986 / 30.0 0.7746 / 30.9 -0.03 1484
SA (25.2°S) 0.8657 / 27.7 0.7914 / 30.3 -0.08 8096
SC (28.5°S) 0.9544 / 25.1 0.9976 / 24.0 0.04 13038
RG (31.5°S) 1.0456 / 22.9 1.0024 / 23.9 -0.04 16136

13
Figure 2.2: Depth-averaged rotary spectra of the raw velocity. The blue line is the
bandpass filter and its response function. The thick (thin) line represents the
anticyclonic (cyclonic) motions. The gray dashed vertical lines represent the tidal
constituents O1, K1, M2 and S2 (left to right). The black dashed vertical line is the
inertial frequency at: [PS] 0.5516 cpd, [VI] 0.6811 cpd, [CF] 0.7986 cpd, [SA] 0.8657
cpd, [SC] 0.9544 cpd, and [RG] 1.0456 cpd. See Figure 2.1 for the geographic location
of the stations presented in panels.

Due to the geographical location of the moorings, generally, very close to the

continental shelf break, the relative vorticity field of the BC can play a critical role in

the observed frequency shift. Five stations presented a negative RFS, meaning that the

observed frequency is lower than the inertial frequency, which is possibly associated

with the influence of the anticyclonic relative vorticity field (ζ > 0) of the BC that

presents mean position near the 200-m isobath (MIRANDA and CASTRO FILHO,

14
1982, EVANS and SIGNORINI, 1985, PETERSON and STRAMMA, 1991). The SC

station is the only exception of this condition, where the observed frequency is higher

than the inertial frequency (positive RFS). A possible reason for this behavior is the

larger distance between the 200-m and 1000-m isobaths (Figure 2.1), allowing the BC

preferential flow to be placed further offshore, where the dominant effect would be a

cyclonic vorticity field (ζ < 0) associated with the outermost portion of the current at the

continental shelf break. The relative vorticity calculated from HYCOM/NCODA in the

same period of the data (Table 2.2) presents a negative mean value in SC (blue shift)

and positive mean values in the other stations (red shift) that support the results of RFS.

It must be emphasized that the standard deviation of the relative vorticity in the

moorings reveals the high spatial variability of the BC along the shelf break and

indicates that the observed inertial peak frequency may vary over time.

Table 2.2: Relative vorticity from HYCOM/NCODA Global 1/12° Reanalysis


(http://tds.hycom.org/thredds/catalogs/GLBv0.08/expt_53.X.html) at the locations
presented in Figure 2.1.
ζ ζ
Relative vorticity (mean) Relative vorticity (standard
2 2
x 10-6 s-1 deviation) x 10-6 s-1
PS (16°S) 3.2 7.1
VI (19.9°S) 5.2 8.8
CF (23.5°S) 3.2 8.7
SA (25.7°S) 0.6 6.7
SC (28.5°S) -5.3 7.2
RG (31.5°S) 1.1 5.5

The stations located at RG (31.5°S), SC (28.5°S), and SA (25.7°S) are those that

presented the highest values of NIV (Table 2.1), which are represented by higher

velocity amplitudes of NIO (~ 50 cm s−1 ), as shown in Figure 2.3. Toward the locations

at lower latitudes, the average velocity amplitudes decrease from 30 to 5 cm s −1 between

15
CF (23.5°S) and PS (16°S). This spatial distribution reveals that the inertial band has

higher kinetic energy at stations located near 30°S latitude.

Figure 2.3: Examples of surface near-inertial events at the six locations presented in
Figure 2.1 of the filtered velocities.

Wind data of the Climate Forecast System Reanalysis (CFSR) (SAHA et al.,

2010) were also used to evaluate the rotary spectra of the wind at those stations during

the same period where the current data were collected (Figure 2.2 and 2.4). As can be

seen, the inertial frequency at RG and SC is approximately coincident with the

frequency of the anticyclonic movements of the diurnal sea breezes, providing favorable

conditions for resonance (STECH and LORENZZETTI, 1992), hence increasing the

inertial energy in the upper layer of the ocean (SIMPSONS and RIPPETH, 2002,

HYDER et al., 2011). For the stations located north of SA, the inertial frequency

departs from the diurnal breeze frequency, the NIV decreases, and the semidiurnal tidal

band becomes more important for the high-frequency variability (Figure 2.2).

16
Figure 2.4: Rotary spectra of the wind velocity extracted at the mooring locations in
Figure 2.1. The thick (thin) line represents anticyclonic (cyclonic) motions, while the
vertical dashed line is the inertial frequency.

2.3.2 Representative near-inertial events

The importance of the near-inertial band for the kinetic energy and the variance

of the currents in the mixing layer was quantified based on the representative (high

energy) near-inertial events that present average occurrence of about twice per month

(Figure 2.5). In absolute terms, the RG station had the highest near-inertial energy,

followed by SC and SA. This concentration of near-inertial energy around the latitude of

30°S is due to the proximity between the inertial frequency and the frequency of sea

breeze (Figure 2.4) (ELIPOT and LUMPKIN, 2008). The stations north of SA have an

inertial frequency far from the diurnal frequency and lower energy in the near-inertial

band.

The relative importance of the NIO for the variance of the currents (Figure 2.5b)

is, to some extent, correlated with the spatial distribution of the kinetic energy in the

upper layer. For the RG and SC stations, the near-inertial band is responsible for 29%

and 31% of the variance of the currents, whereas for the stations further north the

17
relative importance decreases to around 10%. Although SA presents an average energy

value similar to SC, the relative importance differs due to the average position of the BC

in relation to the location of the measurements. The blueshift condition in SC (Table

2.1) is possibly associated with an offshore displacement of the BC from the continental

shelf break, resulting in an increase of the relative importance of the near-inertial band

in this location. In agreement to what was reported by ASSIREU et al. (2017), in

general, NIO are more important for the variance of the currents in the Brazilian

continental shelf break south of 28°S.

Figure 2.5: [a] Depth-averaged near-inertial energy of the high energy events in the top
50 m of the water column at the mooring locations presented in Figure 2.1. [b] Relative
importance of these near-inertial events in terms of the variance of the currents. The
vertical bars represent one standard deviation.

High energy near-inertial events produce vertical shear in the upper layer, which

increases the vertical mixing (CUYPERS et al., 2013) and may impact the vertical

distribution of nutrients and phytoplankton (FRANKS, 1994, LUCAS et al., 2014). The

vertical shear calculated for each inertial event was calculated and its maximum value

was also computed. The mean and standard deviation of the maximum vertical shear for

each mooring location are shown in Figure 2.6. The near-inertial events between RG

18
and CF generate vertical shear that is statistically similar and higher than those

calculated in VI and PS. This spatial distribution has a similar pattern when compared

with the spatial variation of the depth-average near-inertial kinetic energy (Figure 2.5a).

The analogy is due to the direct relationship with the amplitude of the near-inertial

currents, which for the RG, SC, SA, and CF stations are capable of generating the

vertical shear equivalent to the NIO generated by storms and typhoons in the central

west coast of India and South China Sea (SUBEESH and UNNIKRISHNAN, 2016,

CAO et al., 2018). This level of vertical shear reinforces the importance of these near-

inertial events for vertical mixing in the shelf break.

Figure 2.6: The mean maximum vertical shear of the high energy events in the top 50 m
of the water column at the mooring locations presented in Figure 2.1. The vertical bars
represent one standard deviation.

The vertical stratification of the water column does affect the shape of the

velocity ellipses of an inertial event and consequently its vertical shear. Figure 2.7

illustrates the influence of the Brunt-Väisälä (N) frequency profile on the vertical shear

of an inertial event at the VI station. The region where there is an increase in N values

marks the depth where the velocity ellipses deform and increase the vertical shear.

19
Figure 2.7: Vertical structure of a near-inertial event at Vitória (see Figure 2.1 for
location). [a] A profile of the Brunt-Väisälä frequency obtained from an ARGO float.
[b] The near-inertial velocity ellipses. [c] The vertical shear of the horizontal velocity.
The Brunt-Väisälä frequency increase below the depth of 30 m indicates the change in
the stratification profile. This affects the vertical velocity structure and marks a depth of
greater vertical shear.

2.3.3 NIO at the Cabo Frio upwelling system

The greater vertical coverage of the water column at the CF station (Figure 2.1)

allows the study of the vertical structure of the NIO velocities. The near-inertial energy

series in CF (Figure 2.8a, 2.8b) show periods of high energy and vertical propagation

that exceeds the first 50 m of the water column, but do not show a clear seasonality (at

least between autumn and spring) in relation to the intensity and the vertical penetration

of the energy of the NIO. The vertical structure of a near-inertial event in the upper

portion of the ocean depends on the combination of physical factors that involves the

vertical structure of the water column density and the intensity, direction, and

persistence of the wind field (KUNDU, 1976, POLLARD, 1980, ALFORD et al., 2017).

As a result, the temporal variability of these conditions may increase or decrease the

near-inertial energy in the upper layer for the same given month over two consecutive

20
years. This can be seen, for instance, during the month of October for the years of 2014

(Figure 2.8a) and 2015 (Figure 2.8b).

Figure 2.8: Time series of the near-inertial energy (in cm 2 s−2 ) at Cabo Frio (see Figure
2.1 for location) during the years of [a] 2014 and [b] 2015. The inverted triangles
represent the near-inertial events, which are described in Table 2.3.

Table 2.3 summarizes the description of the near-inertial events in CF based on the

maximum of near-inertial energy, duration (days), phase and group velocities (Cz and Cg,

respectively), vertical wavelength (λ), and percentage importance of the statistical

modes for the variance of currents associated with NIO. The mean value of the

maximum kinetic energy of the near-inertial events is 90.2 cm 2 s-2 (standard deviation of

67 cm2 s-2 ), indicating that the NIO present high temporal variability of the velocity

amplitudes. The kinetic energy maxima of the near-inertial events do not present a

seasonal cycle, as already described for the South Atlantic in other studies (ALFORD

and WHITMONT, 2007, CHAIGNEAU et al., 2008). The mean duration of the events

is 7.6 days and coincides with the period of passage of cold fronts and low-pressure

systems in the Southwest Atlantic, which generally originate the NIO (STECH and

LORENZZETTI, 1992, ASSIREU et al., 2017). The values of wave parameters were

21
obtained from a least squares adjustment for each event (Figure 2.9b, 2.9c) and are

similar to those described in the literature for near-inertial events generated by storms

and hurricanes in other regions (KUNDU, 1976; CHEN et al., 2013, ALFORD et al.,

2017).

Table 2.3: Characterization of the near-inertial events in Cabo Frio (see Figure 2.1 for
location).
Max % EOF
Date Kinetic Duration Cz Cg λ Mode
-1 -1
Energy [days] [cm s ] [cm s ] [m] contribution
[cm² s-²] 1 2
03 Jun. 2014 33 16 0.3 -0.01 127 62% 28%
06 Jul. 2014 60 11 0.2 -0.01 174 66% 26%
27 Sep. 2014 59 10 0.2 -0.02 185 66% 27%
24 Oct. 2014 36 5 0.1 -0.02 183 72% 18%
02 Nov. 2014 60 8,5 0.2 -0.01 189 61% 29%
27 Mar. 2015 98 3,9 0.2 -0.01 185 59% 29%
05 Apr. 2015 269 11 0.3 -0.01 108 73% 11%
05 May 2015 93 4 0.4 -0.01 98 62% 32%
20 Jun. 2015 47 4 0.2 -0.03 185 87% 9%
22 Oct. 2015 147 6 0.2 -0.01 167 65% 25%

The vertical energy propagation of the NIO is marked by a maximum value of −

0.03 cm s−1 (~ 26 m/day). Considering an average duration of events of 7 days, the

maximum Cg represents a vertical injection of energy in the first 175 m of the water

column, thus affecting the upper layer of the oceans (seasonal thermocline), as

evidenced by the vertical velocity profile of the NIO (e.g., Figure 2.9a).

22
Figure 2.9: [a] Time series of the northward component of the near-inertial oscillations
of an event that started on the 2nd of November 2014 (see Table 2.3). [b] Wavelength
and [c] phase and group velocity obtained based on a least squares adjustment of data
for this near-inertial event. Positive (negative) values in a are northward (southward).

The modal structure of the velocity in these near-inertial events was also

investigated by an EOF analysis. On average, the first EOF mode explains 67% of the

total variance of the NIO events (Table 2.3) and resembles a first baroclinic mode

(SHEARMAN, 2005, SUBEESH and UNNIKRISHNAN, 2016), with a two-layer

structure in all events as shown in (Figure 2.10). Nevertheless, the structure of the EOF

profiles varies and the zero-crossing ranges between 40 and 100 m. Figure 2.11, which

presents the water column temperature records, demonstrates that the seasonality of the

thermal structure does impact the vertical structure of the velocity of the near-inertial

events. For instance, the 3 June 2014 (Table 2.3 and Figure 2.11a, 2.11b) event occurs

during late autumn and early winter, when the seasonal thermocline reaches deeper

regions (see the depth of the 20 °C isotherm). This near-inertial event presents a vertical

profile of continuous velocity along the first 150 m of the water column. On the other

hand, the 5 April 2015 event (Figure 2.11d, 2.11e) occurs during a situation where both

23
the mixed layer and the seasonal thermocline are shallower (early fall) and as a

consequence the velocity profile of this near-inertial event shows a more discontinuous

vertical structure, with speeds more restricted in the vertical between 20 and 40 m of the

water column.

Figure 2.10: [a] First EOF mode profiles of the alongshore velocity from to all events
indicated in Table 2.3 (gray lines), with the 22 Oct. 2015 event marked as a black line.
[b] Time evolution of the alongshore velocity of the 22 Oct. 2015 event. Positive
(negative) values in b are northward (southward).

24
Figure 2.11: Time series of the vertical distribution of the alongshore velocity [a, d] and
temperature [b, d] of two near-inertial events that occurred during the periods 3 and 19
Jun. 2014 and 5 and the 17 Apr 2015, respectively (see Table 2.3). [c, f] Daily relative
vorticity at the surface derived from AVISO for these events. Positive (negative) values
in a, d are northward (southward).

In addition to the seasonality presented in the vertical temperature structure, these

two events also reveal different effects of the relative vorticity associated with the BC in

NIO. During the event on 3 June 2014 (Figure 2.11c), the relative vorticity field is

positive, indicating closer proximity of the BC to the continental shelf break, whereas

during the event on 17 Apr. 2015, the relative vorticity is negative (Figure 2.11f), which

means a BC is away from the measurement point.

These two different conditions of relative vorticity affect the vertical mean

spectrum of these two events (Figure 2.12a, 2.12b), since the anticyclonic (positive)

vorticity displaces the inertial peak to frequencies lower than the inertial frequency,

while the (negative) cyclonic vorticity displaces it to higher frequencies (ELIPOT et al.,

2010, ASSIREU et al., 2017), configuring a redshift and a blueshift, respectively.

25
Figure 2.12: Determination of the observed frequency of the near-inertial events (see
Table 2.3) that occurred during [a] 3 and 19 Jun. 2014 and [b] 5 and 17 Apr. 2015 to
illustrate the determination of a redshift and blueshift. The vertical dashed lines
represent the inertial frequency at Cabo Frio.

26
2.4 Conclusions
The NIO play a crucial role in the high-frequency hydrodynamic variability

along the Brazilian continental shelf break region south of 25°S (RG and SC) as a

consequence of the proximity to the latitude of 30°S, where the frequency of diurnal sea

breezes coincides with the inertial frequency, enhancing the near-inertial variability. The

relative importance of the near-inertial band for the variance of currents is also higher in

the stations of SC (31%) and RG (29%) and reinforces the ASSIREU et al. (2017)

results. In addition, NIO in the RG, SC, SA, and CF stations present inertial currents

ranging between 30 and 50 cm s −1 and capable of generating vertical shears equivalent

to those near-inertial events caused by hurricanes and large storms in other regions

(SUBEESH and UNNIKRISHNAN, 2016, CAO et al., 2018), which reinforces the

importance of these oscillations in the vertical mixing along the Brazilian continental

shelf break. The energy of the near-inertial band decreases toward lower latitudes (VI

and PS stations), where the semidiurnal tidal band is more important in terms of energy

for high-frequency variability. The RFS demonstrates the role that the subinertial

circulation has on the frequency of the NIO and can indicate the average spatial

positioning of the BC in relation to the measurement points based on the blueshift and

redshift conditions.

At the continental shelf break of Cabo Frio, a well-known region of upwelling

and where deeper measurements allowed for a vertical characterization of the events, it

was found NIO with average speed amplitudes of 15 cm s −1, duration of 7.6 days and

vertical propagation of energy that can involve the first 175 m of the water column. In

addition, the shape of the vertical velocity profile of these near-inertial events is affected

27
by the thermal structure of the upper part of the water column. When the isotherm of

20°C is deeper (> 100 m), the vertical coherence of the NIO velocities is greater,

whereas when this isotherm is shallower (< 50 m), the vertical structure of the velocities

is more restricted to the first meters of the column of water. However, this cannot be

considered as a general rule, since the shape of the NIO depends on the combination of

physical factors that on top of temporal variability of the stratification also involves the

wind field, the adjacent subinertial circulation, and turbulence in the mixed layer.

Future observational and numerical studies can provide important information

about the temporal variability of the three-dimensional structure of the NIO and their

role for the circulation and vertical mixing in the Southwest Atlantic ocean.

28
Chapter 3

Coastal Trapped Waves propagation


along Southwestern Atlantic Continental
Shelf

3.1 Introduction
The Southwestern Atlantic Continental Shelf, between 55°S and 10°S, has an

extensive continental shelf with different morphological, meteorological and

oceanographic characteristics (CASTRO et al., 2006, PIOLA et al., 2018), where the

propagation of CTW plays an important role in the hydrodynamic variability of the

region. VIVIER and PROVOST (2001) suggest that the propagation of CTW is

responsible for the 70-day fluctuations in the transport of the Malvinas Current, while

SARACENO et al. (2005) suggest that CTW modulate intraseasonal variability in the

sea surface temperature and chlorophyll in the Patagonian shelf break. POLI et al.

(2020) described the propagation of three types of CTWs along the Patagonian shelf

break, where two of them are forced by the large zonal wind stress variations south of

47°S that modulate the intensity of the velocity at the 300 -m isobath from 55°S to

29
47°S, and one propagates in the core of the Malvinas Current and comes from Drake

Passage and Malvinas Escarpment.

Along the Brazilian margin, important contributions have been made to the

understanding of the CTW regarding their forcing mechanisms and some aspects of

their propagation. The remote southerly winds associated with the synoptic atmospheric

systems are the main forcing mechanisms behind the CTW that propagate along the

Brazilian South/Southeast continental shelf (CASTRO et al., 2006, CASTRO, 1990,

STECH and LORENZZETTI, 1992, CASTRO and LEE, 1995, DOTTORI and

CASTRO, 2018). FILIPPO et al. (2012) showed high energy and coherence in periods

between 3.5 and 28 days in SSH records taken over one year at four coastal stations

between 43°S and 22°S.

In the South Brazil Bight (SBB), located between Cabo de Santa Marta

(28°40'S) and Cabo Frio (23°S), the theory of remotely forced Continental Shelf Waves

(CSW) has been proposed to explain the propagation of waves with periods between 6

and 12 days and with average phase speed propagation of 10 m s-1 (CASTRO and LEE,

1995, FILIPPO et al., 2012). The propagation of CSW along the SBB modulates the

magnitude and direction of the shelf currents (CASTRO, 1990, CASTRO and

MIRANDA, 1998). Based on hydrography and current meter data, DOTTORI and

CASTRO (2009) showed that the subinertial response of the central part of the SBB to

wind is mainly barotropic. DOTTORI and CASTRO (2018), based on the

semianalytical model of CLARKE and VAN GORDER (1986) and SSH data collected

along the coast, showed that the use of remote wind improves the forecast of the

subinertial currents associated with CSW and that the phase speed of these waves varies

30
throughout the SBB, which are faster (~11 m s -1) when the shelf is wider and slower

(~6-7 m s-1) when it narrows.

The spatial and temporal limitations of previous studies on the propagation of

CTW along the Brazilian continental shelf between 34°S and 11°S introduce various

knowledge gaps associated with the i) generation area, ii) latitudinal variation of energy,

iii) spatial and temporal variability of phase speed propagation and iv) importance of

alongshore variation in both the morphological characteristics and stratification of the

continental shelf. Therefore, based on in situ measurements of coastal sea level covering

approximately 4500 km along the coastline (~48°S - 13°S) and shelf-break currents

covering over 2000 km of the Brazilian shelf-break region (~32°S - 16°S) combined

with the use of a high-resolution simulation of the HYCOM hydrodynamic model, the

present study aims to characterize the propagation of CTW along the Brazilian

continental shelf between 34°S and 11°S. This involves i) characterizing the formation

area of the CTW that propagate along the region, ii) describing the CTW propagation in

terms of the SSH variations and associated currents, iii) investigating how the

topography and stratification of the continental shelf influence the phase speed

propagation of the CTW and iv) investigating the modal structure of the CTW based on

statistical and dynamical criteria. This study used the geographical partition of the

Brazilian continental shelf adopted by CASTRO et al., (2006), which subdivides the

region into: Southern Shelf (SS) (34°S – 28.5°S), South Brazil Bight (SBB) (28.5°S -

23°S), Abrolhos-Campos Region (ACR) (23°S - 15°S), and Eastern Shelf (ES) (15°S –

8°S).

31
3.2 Data and Methods

3.2.1 In situ data

Hourly sea level data was measured at 8 coastal tidal stations, distributed

between 48°S and 13°S (Figure 3.1 and Table 3.1). The data from the Puerto Deseado

and Mar del Plata stations were provided by University of Hawaii (UHSLC, 2019),

while the data from the other stations were provided by the Brazilian component of the

Global Sea Level System (GOOS/BRAZIL/GLOSS, 2019). Near surface velocity data

was measured at 6 different meter moorings, located at the Brazilian shelf break from

approximately 32°S and 16°S (Figure 3.1 and Table 3.2).

Five moorings were installed as part of the Brazilian component of

GOOS/BRAZIL/PNBOIA (2019) and one mooring (NAT01, located off Cabo Frio, Rio

de Janeiro) was installed by the Physical Oceanography Laboratory –

LOF/COPPE/UFRJ, as part of the INCT PRO-OCEANO Project (Figure 3.1). The

PNBOIA moorings were equipped with 400 kHz acoustic Doppler current profilers -

ADCP (facing down) that measured hourly intensity and current direction between 5.5

m and 53 m, with a vertical resolution of 2.5 m. The NAT01 mooring was equipped

with a 150 kHz ADCP (facing up) that measures current velocities between 14 and 174

m, with a vertical resolution of 4 m, and with a Aquadopp 1MHz (facing down) that

measured current velocities between 184 and 201 m, with vertical resolution of 0.5 m,

both with a 30 minute intervals. The duration of each mooring survey period is listed in

Table 3.2.

32
Figure 3.1: Bathymetry of the Southwestern Atlantic continental shelf (shaded),
indicating the coastal sea level stations (GLOSS/UHSLC - green circles) and mooring
positions (PNBOIA/NAT - red and magenta stars, respectively). The five isolines
represent the 50 m, 200 m (yellow line), 1000 m, 2000 m and 3000 m isobaths,
respectively. The top left panel indicates the two HYCOM model domains with 1/12°
and 1/24° horizontal resolution. The acronyms for the locations are indicated in Tables
3.1 and 3.2.

33
Table 3.1: Description of the coastal sea level data.
Location Lon/Lat Period Length of reliable
data (days)
Salvador [SSA] 38.97°W December 10th, 2010 to 2453
12.97°S September 04th, 2017
Macaé [MA] 41.47°W May 05th, 2008 to 1712
22.23°S May 30th, 2015
Ilha Fiscal [IF] 43.17°W January 1st, 2008 to 3804
22.90°S June 1st, 2018
Ubatuba [UB] 45.12°W May 2nd, 2014 to 1500
23.50°S January 26th, 2018
Cananéia [CA] 47.93°W February 1st, 2014 to 1001
25.02°S October 28th, 2016
Imbituba [IM] 48.40°W January 1st, 2007 to 2473
28.13°S December 31th, 2015
Mar del Plata [MP] 57.55°W April 1st, 2010 to 3196
38.05°S December 31th, 2018
Puerto Deseado 65.17°W January 1st, 2011 to 2413
[PD] 47.75°S September 1st, 2017

Table 3.2. Description of the velocity data from PNBOIA and NAT moorings.

Mooring Lon/Lat Local Period Length


Station depth [m] (days)
Porto Seguro (PS) 37.94°W 133 July 04th, 2012 to 332
16°S May 31th, 2013
Vitória (VI) 39.72°W 83 October 13th, 2015 to 233
19.92°S June 1st, 2016
Cabo Frio (CF) 42.17°W 210 May 03rd, 2014 to 211
23.57°S November 30th, 2014
Santos (SA) 44.93°W 243 April 12th, 2011 to 353
25.27°S March 31th, 2012
Santa Catarina (SC) 47.39°W 203 February 17th, 2011 to 342
28.51°S January 25th, 2012
Rio Grande (RG) 49.88°W 208 March 08th, 2011 to 114
31.58°S July 1st, 2011

34
3.2.2 Numerical modeling

Hydrodynamic modeling was used to accommodate the temporal and spatial

limitations of the in situ data. This study of CTW was carried out by analyzing the

results from a 11-year long (2006-2016) simulation with HYCOM – Hybrid Coordinate

Ocean Model (BLECK, 2002) with data assimilation, developed at the Physical

Oceanography Laboratory of COPPE/UFRJ (GABIOUX et al., 2013, COSTA et al.,

2017) within the context of the REMO Project (LIMA et al., 2013). REMO – The

Ocean Observation and Modeling Network is a Brazilian initiative in operational

oceanography, associated with Godae Ocean View (BELL et al., 2009), with the

objective of producing short term forecasts and long term hindcasts for environmental

studies in the South Atlantic Ocean.

The model was configured with 32 σ2 layers and nominal horizontal resolution

of 1/24° for a domain from 68°W to 18°W and 45.18°S to 10.19°N, nested in a 1/12°

horizontal resolution domain from 98°W to 45°E and 79.54°S to 50.27°N (Figure 3.1).

The model bathymetry was interpolated from ETOPO-1, improved with information

from local data from the Brazilian Navy. This simulation was forced at 3-hour intervals

with synoptic atmospheric fields from the Modern-Era Retrospective analysis for

Research and Applications – MERRA2 (GELARO et al., 2017) from NASA, with bias

correction of the radiative fluxes from the Clouds and the Earth’s Radiant Energy

System (CERES) Energy Balanced and Filled (EBAF) Surface version Ed28 (KATO et

al., 2013). Heat and mass turbulent fluxes (virtual salt flux) were calculated internally

during the simulation with standard bulk formulas, using the model SST and

atmospheric fields from the MERRA2 reanalysis, with additional relaxation of sea

surface salinity to monthly climatology from WOA13 (LOCARNINI et al., 2013,

35
ZWENG et al., 2013), with a 30-day time scale (PAIVA and CHASSIGNET, 2001).

Assimilation of along track altimetry data, sea surface temperature, and ARGO profiles

were carried out with the Tendral Assimilation System, T-SIS (HALLIWELL et al.,

2014). Tidal forcing was implemented in the 1/24o run as lateral boundary conditions for

the main tidal constituents from TPXO7.2. Three-dimensional fields were saved daily

for the 1/12o run and every three hours for the 1/24o run. Additionally, hourly surface

fields were saved for the entire 11-year run with the 1/24 o model, and for the year of

2015 for the 1/12o run. Hourly fields were necessary to improve the computation of the

phase speed propagation.

3.2.3 Analysis

To isolate the frequency band associated with the CTW, the velocity and sea

surface height time series of both the in situ data and the model results were band-pass

filtered (Butterworth filter) with cutoff frequencies of 0.025 and 0.43 cycles per day

(cpd). These limits were established considering the spectral analysis (Fast Fourier

Transform - FFT) of the 11-year long simulated fields, and the observed spectral gaps

at approximately 2.3 and 40 days, fully preserving the signal of waves with periods

between approximately 3 and 30 days. The good correlation (0.9) between model and

data SSH at the coastal stations (see locations in Table 3.1) attests the quality of the

model results, and the realistic representation of CTW propagation in the model (Figure

3.2). The difference between the wave amplitude simulated by HYCOM and that

observed in the in situ data is probably due to the location of the measurements in bays

and coastal inlets, which are not simulated by the model.

36
Figure 3.2: Two-months long band-pass filtered sea level time-series (see text for
details) for GLOSS/UHSLC (blue line) and HYCOM 1/12° (black line) at PD (from
January 1st, 2015 to March 1st, 2015) and HYCOM 1/24° at the remaining stations: MP
(from July 2nd, 2012 to September 1st, 2012), IM (from June 1st, 2009 to August 1st,
2009), CA (from August 1st, 2014 to October 1st, 2014), UB (from March 1st, 2015 to
May 1st, 2015), IF (from August 1st, 2010 to October 1st, 2010), MA (from April 1st,
2012 to June 1st, 2012) and SSA (from September 1st, 2015 to November 1st, 2015)
stations. The scatter plot shows a strong correlation between GLOSS/UHSLC data and
HYCOM, considering all the available data period for each station (Table 1). The red
line represents a linear fit calculated with a least square adjust. The geographical
locations of these stations and their associated names are illustrated in Figure 3.1 and
Table 3.1, respectively

The latitudinal generation range of the CTW that propagated along the Brazilian

continental shelf (SS, SBB, ACR and ES) was characterized according to the SSH

energy level (in situ data and model) and the coefficient of coherence between the SSH

along the Brazilian coast and the associated SSH farther south. Additionally, a cross-

spectral analysis was performed between the SSH along the coast and the remote

southerly winds to quantify the importance of the remote forcing for the SSH variability

in the 3 - 30 day band along the coast. For this purpose, the meridional wind component

of the MERRA2 reanalysis (GELARO et al., 2017) located in the northern portion of

the Argentine continental shelf (57°W, 40°S) was used. This location was chosen as a

37
representative area of the region dominated by mesoscale cold front systems

(GARREAUD, 2000, BARROS et al., 2008) that are identified in the literature as

forcing mechanisms of the CTW that propagate along the SS, SBB and ACR.

The importance of stratification for the dynamics of CTW along the Brazilian

continental shelf between 32° S and 16°S was estimated by the stratification parameter

(S) (CLARKE and BRINK, 1985) defined by Equation 3.1:

−2
S=N ² α ² f (3.1)

in which N2 represents the average Brunt-Väissälä frequency for the continental shelf

region, α is the ratio between the maximum depth (H) and the associated width (L) of

the continental shelf, and f is the Coriolis parameter.

The cross-shore modal structure of the CTW was investigated based on i)

empirical orthogonal function (EOF) analysis of the velocity sections along the

Brazilian continental shelf (between 32°S and 16°S), which were 3 - 30 days band

filtered, and ii) the use of the linear model of Brink and Chapman (1987). The purpose

of the EOF analysis was to quantify the relative importance of the statistical modes for

the variance in currents at the continental shelf between SS and ACR (32°S and 16°S).

The BRINK and CHAPMAN (1987) model, conversely, was used to obtain the cross-

shore structure of the theoretical CTW baroclinic modes using an average 11-year

Brunt-Väissälä profile located at the continental shelf-break region and a bathymetric

profile of the cross-shore sections.

38
3.3 Results and Discussion
This section is divided into 3 subsections, which address i) the CTW

propagation derived from in situ data, ii) the dynamic characterization of the CTW

propagation and iii) the modal structure of the CTW. In the first subsection, the SSH

along the coast and the alongshore velocity at the continental shelf break are used to

characterize the propagation of the CTW in terms of i) the maximum correlation and

associated lag among the SSH coastal stations, ii) the latitudinal and temporal variation

in SSH energy in the band between 3 and 30 days, iii) the latitudinal variation in the

coherence between the SSH and remote winds, and iv) the effect of CTW propagation

on the intensity and direction of the currents along the continental shelf break. In the

second subsection, based on the numerical modeling results, SSH series at the 50 m

isobath are used to characterize the generation region of the CTW and also discusses the

dynamic importance of both topography and stratification of the continental shelf for the

phase speed propagation. In the last subsection, the modal structure of the CTW is

characterized based on the EOF analysis of the alongshore velocity at selected cross-

shore sections from the HYCOM model, as well as the linear theoretical model of

BRINK and CHAPMAN (1987).

3.3.1 CTW propagation from in situ data

3.3.1.1. Sea level data at the coast

The phase propagation of the CTW based on the observed SSH at various

locations along the southwestern Atlantic coast is illustrated in Figure 3.3 and suggests

that the waves that propagate along this region, with periods between 3 - 30 days, come

39
from a wide latitudinal band farther south, which encompasses the PD (47.75° S) and

IM (28.13°S) stations. The red lines in Figure 3.3 illustrate the propagation of four of

these waves. Wave 1 has larger amplitudes near station IM (28.13°), while the other

three waves show a noticeable phase propagation starting either at MP (38.05° S) (wave

2) or PD (47.75° S) stations (waves 3 and 4).

The maximum amplitude of these waves varies along the coastal stations,

increasing from 37 cm at PD (47.75°S) to 93 cm at MP (38.05°S) station (Figure 3.3).

Along the Brazilian continental shelf, the largest amplitudes of the CTW are observed

between the IM (28.13°S) and CA (25.02°S) stations, with values of 51 and 46 cm,

respectively. North of the UB (23.5°S) station, the amplitude of the waves decreases

following the narrowing of the continental shelf (see the 200 m isobath in Figure 3.3),

and between stations MA (22.23°S) and SSA (12.97°S), a drastic reduction from 29 cm

to 10 cm is finally observed.

The relationship between the amplitude of the CTW at the coast and the width of

the continental shelf observed after the UB station (23.5°S) are consistent with the

barotropic nature of CTW described at the SBB by DOTTORI and CASTRO (2009) and

the theory of nondivergent shelf waves presented by GRIMSHAW (1977), which shows

that the wavelength and amplitude of the wave increase with the width of the

continental shelf. The pronounced reduction observed between the MA (22.23°S) and

SSA (12.97°S) stations is possible due to abrupt variations in the width and depth of the

continental shelf (Figure 3.1), especially along ACR, where there is marked narrowing

(Tubarão Bight) followed by a widening of the continental shelf (Abrolhos Bank),

which is also a shallow region (Figure 3.1). These features make this area prone to wave

scattering (WILKIN and CHAPMAN, 1990, WEBSTER, 1987).

40
Figure 3.3: Propagation of CTW along the Brazilian continental shelf based on coastal
sea-level records from SSA, MA, IF, UB, CA, IM, MP, and PD during May/August
2014, which is the best period in terms of data quality and synchronicity. The
observations were bandpass filtered to eliminate tides and oscillations with periods
exceeding 40 days. The red lines indicate the phase propagation of 4 CTW, named (from
left to right) waves 1, 2, 3, and 4. The width of continental shelf at each stations is: 325
km (PD), 230 km (MP), 128 km (IM), 175 km (CA), 125 km (UB), 94 km (IF), 62 km
(MA) and 10 km (SSA). The geographical locations of the stations and their names are
illustrated in Figure 3.1 and Table 3.1, respectively.

The cross-correlation between the filtered SSH series (Figure 3.4a) reveals that

the waves that propagate along the Brazilian continental shelf between 28°S – 12.97°S

are more strongly correlated with the SSH at the MP (38.05°S) station than at the PD

(47.75°S) station. This station has correlation coefficients below 0.4 and a correlation

41
peak that is only distinguishable south of the IF (22.90°S) station (black line). This may

be associated with the fact that the PD (47.75°S) station is located in the portion of the

Patagonian continental shelf (56°S - 45°S) in which the zonal wind stress is the most

important driver of SSH variability. Conversely, the MP station is located in a region

dominated by the synoptic atmospheric systems (GARREAUD, 2000, BARROS, 2008),

which is the main driver of the CTW that propagate along the SBB (STECH and

LORENZZETTI, 1992; CASTRO and LEE, 1995, DOTTORI and CASTRO, 2018).

In addition, the cross-correlation analysis between the SSH series shows that the

CTW signal that propagates along the Brazilian continental shelf (SS, SBB, ACR and

ES) is more strongly correlated with the IM station (28.13°S) than with the MP station

(38.05°S). One explanation for this, which follows the CSW theory, is the fact that the

local meridional wind in the southern portion of the SBB presents higher variance and

can represent an important source of low-frequency energy for the CSW (CASTRO and

LEE, 1995). On the wave propagation (red line) the correlation coefficient increases

after MP station (38.05°S) and the highest values are observed along the Brazilian coast

between IM (28.13°S) and MA (22.23°S) stations. For the sector between the MA

(22.23°S) and SSA (12.97°S) stations, where a decrease in the CTW amplitudes is

observed (Figure 3.3), there is a reduction in the correlation coefficient in all lines

(Figure 3.4a).

42
Figure 3.4: [a] Correlation coefficient (r) between the filtered sea level at the coastal
stations of PD (black line), MP (blue line) and IM (cyan line) and the remaining stations
located further north. The red line represents the correlation coefficient between
consecutive stations along the propagation of CTW. [b] Time lagged cross-correlation
between the coastal stations of PD (black line), MP (blue line), and IM (cyan line) and
the remaining stations located further north. The y labels in these figures are not scaled
due to the variable spatial distribution of the data. This analysis was performed during
the periods of May 17th, 2014 and December 31th, 2014 for all stations except MA
(22.23°S), which was performed during May 17th, 2014 and August 07th, 2014,
representing the best period in terms of data quality and synchronicity. The geographical
locations of the stations and their names are illustrated in Figure 3.1 and Table 3.1,
respectively.

The correlation coefficients between the SSH series along the coast (Figure 3.4a)

allows us to subdivide this wide region from station PD (47.75°S) to station IM

(28.13°S) into 2 parts: i) one south of the MP station (38.05°S), which has a weaker

correlation with the signal of the waves that propagate along the Brazilian continental

shelf (SS, SBB, ACR and ES), and ii) the another between the MP (38.05°S) and IM

(28.13°S) station, which has a well-correlated SSH signal.

43
The latitudinal variation of the time lagged cross-correlation of PD (47.75°S), MP

(38.05°S), and IM (28.13°S) and the remaining stations located further north (Figure

3.4b), shows that the phase speed of the CTW varies along their propagation. For

instance, the increase in lag for the sector between the MA (22.23°S) and SSA (12.97°S)

stations is noteworthy and indicates an area of intense wave deceleration. A more

detailed analysis of these latitudinal variations in phase speed propagation, based on

numerical modeling results, is presented in section 3.2.

The time and space variability of the CTW energy was investigated based on the

wavelet analysis (TORRENCE and COMPO, 1998, LIU et al., 2007) of the SSH series

between stations MP (38.05°S) and MA (22.23°S) (Figure 3.5), which are the locations

presenting significant energy levels in the 3 - 30 day band. The global wavelet spectrum

shows that the highest energy levels at the MP station (38.05°S) are followed by a

decrease towards the MA (22.23°S) station. Furthermore, it reveals a significant

reduction in the occurrence of high-energy events in the 3 - 8 day band with statistical

confidence. This indicates that the attenuation of the CTW energy occurs more

pronouncedly at short wave periods, which explains the recurrent arrival of more

energetically longer period (> 8 days) waves as far north as the MA (22.23°S) station.

The SSH power spectrum wavelets for all coastal station time series show that

CTW energy is present throughout the year and varies over time in the form of pulses.

North of station CA (25.02°S), low-energy phases become more evident, especially

during the southern summer (Dec-Mar), a season during which the propagation of cold

fronts along the Brazilian continental shelf are less frequent and intense (CASTRO et

al., 2006, DOTTORI and CASTRO, 2009).

44
Figure 3.5: Wavelet analysis of the coastal sea level data at the MP, IM, CA, UB, IF, and MA stations during 2014/2015, which is the best
period in terms of data quality and synchronicity.. The left panels indicate the power spectrum (Morlet), where the color bar represents the
normalized variance and the black contour indicates the 95% confidence level. The hatched area represents the cone of influence. The right
panels indicate the global wavelet spectrum [cm²/cph] (blue line) and the 95% confidence level (dashed line). For the stations with gaps in
the series, the global spectrum is the average of the different analysis periods. The geographical locations of the stations and their names
are illustrated in Figure 3.1 and Table 3.2, respectively.

45
The cross spectra between the SSH at the coastal stations and the wind patterns at

40°S (Figure 3.6) reveal the importance of remote forcing for the temporal variability of

the SSH in the 3 - 30 day band. At the PD station (47.75°S), coherence values above the

statistical confidence limit are observed only for periods shorter than 7 days. At station

MP (38.05°S), in turn, the coherence between the SSH and wind increases and is

statistically significant for the whole 3 - 30 day band. Along the Brazilian coast, there is

coherence in the band between 3 - 30 days at stations IM (28.13°S), IF (22.9°S) and MA

(22.23°S) and between 3 - 20 days at stations CA (25.02°S) and UB (23.5°S). At station

SSA (12.97°S), there is no consistency with the remote wind, except for a peak at 16

days that is above the statistical confidence limit.

The coherence increase between stations PD (47.75°S) and MP (38.05°S) indicates

that the SSH variability in the 3 - 30 day band at station PD (47.75°S) is largely

modulated by atmospheric forcing factors different from those at the northern stations.

Along the SS and SBB coast, the latitudinal variation of the coherence between the SSH

and the remote wind is similar to that observed in the variation of the correlation along

the propagation of the CTW (Figure 3.4a), as it presents a high correlation southward of

station MA (22.23°S), which is then followed by a strong northward reduction. Since

the length of the time series is different among the coastal stations, the cross-spectral

analysis between the wind and the SSH along the coast will be explored in more detail

in section 3.2, with 11-year-long numerical simulations.

46
Figure 3.6: Cross-spectrum between the coastal sea-level records in stations PD
(November 20th, 2011 to June 10th, 2013), MP (April 1st, 2010 to December 31th,
2018), IM (January 1st, 2008 to January 15th, 2011), CA (February 1st, 2014 to October
28th, 2016), UB (May 2nd, 2014 to January 26th, 2018), IF (January 1st, 2008 to June
17th, 2018), MA (December 1st, 2011 to December 31th, 2012), and SSA (December
16th, 2010 to September 04th, 2017) and the meridional component of wind at 57°W
40°S from MERRA2 reanalysis for the same period. The gray line represents the 99%
confidence interval. The geographical locations of the stations and their names are
illustrated in Figure 3.1 and Table 3.1, respectively.

3.3.1.2. ADCP data at the continental shelf-break/slope region

The characterization of the CTW propagation at the continental shelf-break region

was based on the analysis of the top 50 m depth-averaged alongshore velocity

47
components at 6 different locations (Figure 3.1 and Table 3.2) and the SSH time series

at the IF (22.9°S) coastal station (Figure 3.1 and Table 3.2), which was chosen because

it presented the best data quality and synchronicity with the current meter data.

Comparisons between the filtered time series of SSH at the IF (22.9°S) station and the

alongshore velocity at the shelf-break along the SS, SBB and ACR (Figure 3.7) show

high coherence and indicate that CTW propagation affects the hydrodynamics of the

entire width of the continental shelf.

These results are in agreement with previous studies carried out in the SBB

showing currents at the inner/medium continental shelf modulated, in terms of intensity

and direction, by the propagation of CSW with periods between 4 and 12 days

(CASTRO, 1990, CASTRO and MIRANDA, 1998, DOTTORI and CASTRO, 2009).

As shown in Figure 3.7, the correlation between the filtered SSH series at the IF

(22.9°S) station and the alongshore velocity series (depth-averaged for the top 50 m) is

moderate at the RG station (31.58°S), strong between stations SC (28.51°S) and VI

(19.92°S), and decreases again at the PS station (16°S), located north of the Abrolhos

Bank.

This decrease in correlation after the VI (19.92°S) station is similar to that

observed in the latitudinal variation of the correlation coefficient of the CTW signal in

the SSH along the coast (Figure 3.4a), which shows a decrease after 22.23°S (station

MA). Nevertheless, the moderate correlation at the PS station (16°S) indicates that the

CTW signal observed in the shelf break velocity is less reduced north of the Abrolhos

Bank when compared to the CTW signal observed in SSH at the coast, which exhibits a

strong attenuation in energy at this location.

48
Figure 3.7: Comparison between the sea level from station IF (black line) and the top 50 m depth-averaged meridional velocity component
at the continental shelf-break (red line) moorings. The cross-correlation between the time series shows correlation coefficients (r) of 0.4
(RG), 0.7 (SC), 0.6 (SA), 0.6 (CF), 0.6 (VI) and 0.4 (PS) (from bottom to top). The time series are plotted with maximum correlation lag
(hours) correction of 45 (RG), 27(SC), 10(SA), 8 (CF), -42 (VI) and -126 (PS). The geographical locations of the stations and their names are
illustrated in Figure 3.1 and Tables 3.1 and 3.2, respectively.

49
It is important to note that the locations of these current meters are under the

influence of the Brazil Current (BC), which on average lies between the 200 m and

1000 m isobaths (PETERSON and STRAMMA, 1991, SILVEIRA et al., 2000, ROCHA

et al., 2014). Therefore, the correlation between this SSH series at the coast and the

current meter velocity is an indication that the propagation of CTW can play an

important role in the variability of the internal portion of the BC. For instance, in the SC

(28.51°S) and SA (25.27°S) station series, the filtered alongshore velocity in the 3 - 30

day band has maximum amplitudes of 0.5 m s -1 and 0.4 m s-1, respectively, with

modulations in the direction consistent with the CTW signal captured by the SSH at the

IF station (Figure 3.7).

A cross-spectral analysis of the raw time series described above was performed

to identify the periods in which coherence between sea level at the coast and the

alongshore velocity component at the shelf break is noticeable (Figure 3.8). The spectra

confirm that there is high coherence, with statistical confidence, in those ranges

associated with the CTW periods that propagate along the SS, SBB and ACR (3, 5, 7,

12, 15 - 20, and 30 days). Between stations RG (31.58°S) and VI (19.92°S), spectral

coherence values larger than 0.9 are observed in various periods associated with the

propagation of CTW, while at the PS (16°S) station, there is a decrease in coherence

(0.4 – 0.5) with statistical confidence, which is restricted to periods of approximately 15

days. The greater vertical coverage of the CF mooring (23.57°S) station shows that this

high coherence persists throughout the water column, indicating that CTW can indeed

play a vital role in modulating currents in the continental shelf break along this

important coastal upwelling region (VALENTIN, 2001, CASTELAO and BARTH,

2006).

50
Tem uma maior energia na faixa de 5 dias, o que
explica a maior energia no evento extremos de 7 dias

Figure 3.8: Cross-spectrum between the sea level at the coast [station IF] and the
alongshore velocity at the continental shelf break at stations RG, SC, SA, VI, and PS for
the top 50 m of the water column and the top 200 m of the water column in CF. The
black line represents the 99% confidence interval. The geographical locations of the
stations and their names are illustrated in Figure 3.1 and Tables 3.1 and 3.2,
respectively.

In order to complement the previous analysis of the in situ data, the next section

focus on the results of two HYCOM models at different horizontal resolutions (1/12°

and 1/24°). The next section presents a characterization of the area of generation of the

CTW as well as a discussion of the dynamic importance of the continental shelf width

51
and its seasonal variations in the stratification in modulating the phase speed

propagation of the CTW.

3.3.2 Dynamic characterization of the CTW propagation

The propagation of CTW along the Southwestern Atlantic continental shelf,

between 54°S and 10.5°S, illustrated in the Hovmöller of Figure 3.9, reveals a wide

generation area for the waves that are seen along the Brazilian shelf (SS, SBB, ACR and

ES). For instance, there are some waves (with amplitudes larger than 10 cm) that show a

clear propagation phase originating at 53°S (at the beginning of October), while others

show a more evident propagation signal starting at 33°S (in mid-November). In general,

the highest amplitudes of these waves are observed between 45°S and 30°S, which is

the portion of the region that includes coastal station MP (38.05°S). This station was

previously characterized (section 3.3.1.1) as the Patagonian shelf location associated

with the highest amplitude and better correlated with the CTW signal that propagates

along the SS and ACR. It is important to note that north of station MA (22.23°S), a

drastic reduction in the amplitude of these waves is also observed.

While Figure 3.9 shows that these waves occur throughout the year, a clear

reduction in the occurrence of waves with amplitudes larger than 10 cm is observed in

late spring and during southern summer of 2015. This is consistent with the less

frequent cold fronts that propagate northward along the southeastern Brazilian

continental shelf (SS, SBB, ACR and ES) during these seasons. This behavior was

observed in all years simulated (2006 – 2016).

52
Se você perceber poucas ondas chegam a 20º, as que chegam normalmente são cristas.
Talvez isso tenha alguma correlação com o aumento de nível no porto de tubarão

Figure 3.9: Hovmöller of the band-pass filtered (Butterworth filter) hourly SSH along the
50 m isobath during 2015 from the HYCOM 1/12° simulation. The color scale is limited
to highlight the wave signals throughout the year. The black contours indicate
amplitudes larger than 10 cm.

The cross-spectral analysis between the 11 years of daily SSH of the HYCOM

1/12° model and the alongshore component of the wind at the northern portion of the

Patagonian continental shelf (57°W, 40°S) (Figure 3.10) shows that there is coherence

present, with statistical confidence, between the SSH along the Brazilian coast (SS,

SBB, ACR and ES) and the remote wind in the 3 - 30 day band. The cross-spectral

analysis considering the SSH in other locations, such as stations MP (38.05°S), CA

(25.02°S) and IF (22.9°S), and the wind patterns between 51°S and 24°S (not shown)

also indicate that the SSH is always well correlated with the remote wind farther south.

This is in agreement with the CSW literature available for the SBB (CASTRO and LEE,

1995; DOTTORI and CASTRO, 2018) and indicates that the SSH signature is

associated with the propagation of synoptic systems along the Brazilian continental

shelf between 34°S and 11°S.

53
Figure 3.10: Cross-spectrum between SSH along the 50 m isobath from 11 years (2006-
2016) of the HYCOM 1/12° simulation and the meridional component of wind in 57°W,
40°S from MERRA2 reanalysis for the same period. The confidence interval is 99%.
The blank areas represent the coherence coefficient without statistical confidence.

The latitudinal variation in the coherence coefficient (Figure 3.10) reveals

maxima at the coast between stations MP (38.05°S) and MA (22.23°S), which is

followed by a decrease towards high and low latitudes away from this region. This

result corroborates that observed in the in situ data presented in Figure 3.7 and is

associated with the atmospheric circulation in the region. While north of 40°S the region

is dominated by the main synoptic systems that force the CTW that propagate along the

Brazilian continental shelf ( SS, SBB, ACR and ES), the region located south of 40°S,

especially between 45°S and 55°S, is more influenced by the westerlies (GARREAUD,

2000, GARREAUD, 2009), that do not force the waves that propagate along the

Brazilian coast between SS and ES.

54
The phase speed propagation of the CTW was characterized based on 11-year

(2006-2016) HYCOM 1/24° simulation of hourly SSH at the 50 m isobath (Fig. 3.1)

between 44°S and 15°S. The speed calculation was performed employing a space-

centered cross-correlation between the filtered series using a window of 2 points

forward and 2 points backward (the approximate distance between two consecutive

points is 127 41 km), which provided the values of the lags associated with the

maximum correlations. Taking into account that the CTW theory demonstrates that the

properties of the wave propagation are affected by the width and stratification of the

continental shelf (HUTHNANCE, 2001, MYSAK, 1980, WANG and MOOERS, 1976),

the phase velocity calculation considered two distinct periods of stratification of the

water column on the continental shelf: summer/autumn (more stratified) and

winter/spring (less stratified).

The latitudinal variation of the CTW phase speed propagation (Figure 3.11) shows

that the waves are faster between 42°S and 41°S, with average speeds larger than 25 m

s-1 that decelerate north of 40°S, presenting an average speed of 11 m s-1 at the MP

(38.05°S) station. Along the SS and SBB, CTW propagate towards station CA (25.02°S)

with speeds between 9 and 12 m s-1, which are values consistent with those presented in

the bibliography for the SBB (CASTRO and LEE, 1995, DOTTORI and CASTRO,

2018; FILIPPO et al., 2012). North of 24°S, the phase propagation speed decreases

rapidly to values of approximately 3 m s-1.

The variability of the CTW phase speed is largest between 42°S and 41°S

(standard deviation bars) and is slightly higher in winter/spring than in summer/autumn

on the coast north of station MP (38.05°S). In addition, it is observed that the latitudinal

variation in the CTW phase speed propagation is directly related to the width of the

55
continental shelf and is slightly modified by seasonal stratification. The relationship

between the CTW phase speed propagation and the width of the continental shelf has

been observed for other continental shelves around the world (SCHULZ et al., 2012,

WOODHAM et al., 2013) and corroborates the CSW barotropic characteristics on the

SBB continental shelf (DOTTORI and CASTRO, 2018, DOTTORI and CASTRO,

2009).

Figure 3.11: Phase speed (in m s-1) of CTW evaluated from SSH at the 50 m isobath and
based on a 11-year (2006-2016) HYCOM 1/24° simulation considering periods of larger
(red line) and smaller (blue line) stratification. The green dots represent the same
property calculated using SSH from the sea level stations presented in Figure 3.1 and
Table 3.1. The yellow dots represent the theoretical values based on free CSW theory
using an exponential depth profile of the continental shelf with a deviation bar
associated with a variation of 10% the width of the continental shelf. The black
dotted line represents the shelf width (in km).

In this study, the theoretical free CSW solution obtained by BUCHWALD and

ADAMS (1968) was applied between 27°S and 21°S to determine how much of the

latitudinal changes in the phase velocity can be explained by variations in the width of

the continental shelf. For this, the shape of the continental shelf was adjusted to

bathymetric profiles with exponential decay at different locations of the SBB, where the

56
phase speed propagation was estimated (yellow dots in Figure 3.11).The error bar is

associated with the calculation of the phase speed with a variation of 10% of the width

of the continental shelf.

Figure 3.11 shows that the free CSW theory provides phase velocity values that

are close to those observed with the in situ data and modeled with the HYCOM 1/24°

and is capable of representing the spatial variation associated with the wave acceleration

and deceleration as a function of changes in the continental shelf width. This result

demonstrates that this feature is the most important factor driving the spatial variability

of the CTW phase velocity and explains the intense deceleration of the waves north of

24°S.

Spectral analysis of the SSH series at the 50 m isobath and between 54°S and

10.5°S was used to characterize the CTW energy in the frequency domain along this

geographical region (Figure 3.12a). Additionally, the cross-spectral analysis between the

SSH at station MP (38.05°S) and the other locations, where values of coherence and lag

were obtained per period, was evaluated (Figure 3.12 b,c).

The power spectrum presented in Figure 3.12a shows that south of 45°S, the

CTW energy is an order of magnitude lower than that observed between 45°S and 22°S.

In this region, it is possible to observe three bands of high energy associated with the

following periods: i) 5 - 12 days, ii) 15 - 22 days and iii) 25 - 40 days. Between 22°S

and the Abrolhos Bank (AB in Figure 3.12a), a strong decrease in energy is observed in

all periods, which allows us to characterize this region as the main area of energy

dispersion for the CTW that propagate along the SS and SBB with periods between 3 -

30 days.

57
The latitudinal variation in the spectral coherence coefficient between the SSH

of station MP (38.05°S) and the other stations presented in Figure 3.12b reveals a

latitudinal band of larger coherence (> 0.8) between 42°S and 32°S, which is coincident

with the region where the maximum energy is observed in the 3 - 30 day periods. South

of 45°S, the coherence coefficient is only moderate (> 0.6 north of 50°S), while along

the SS and SBB, it varies between high and moderate values, from 0.8 (32°S) to 0.5

(22°S), and decreases towards low latitudes.

Figure 3.12: [a] Spectral analysis (FFT) of the SSH at the 50 m isobath based on 11-
year (2006-2016) HYCOM 1/12° (south of 45.18°S) and 1/24° (north of 45.18°S)
simulations. [b] Associated cross-spectrum coherence coefficients and [c] cross-
spectrum lag between station MP and the other stations with a 99% confidence interval.
The red labels indicate values of lag in days. The blank areas in [b] and [c] represent
coherence coefficients and phase without statistical confidence or phase propagating
southward.

The generation area for the CTW (see Figure 3.9) can be divided into two

subregions. One is located south of 45°S, associated with lower energy levels and low

coherence with the signal of the waves that propagate along the SS and SBB. The other

58
is located north of 45°S, distinguished by higher energy levels and consistency with the

CTW that propagate along the SS and SBB.

The choice of the MP station as a reference for the cross-spectrum (Figure 3.12b,

3.13c) is because it is located within the region where the highest energy in the

generation area is found (north of 45°S) and because it presents high/moderate

coherence with the SSH along the Brazilian continental shelf (SS, SBB, ACR and ES).

The cross-spectrum phase (Figure 3.12c) reveals that phase propagation towards lower

latitudes occurs starting at 50°S, which corroborates what was observed in the in situ

SSH data (Figure 3.3) and the HYCOM model results (Figure 3.9). Additionally, it is

observed that the CTW phase propagation speeds vary not only with latitude but also in

terms of the wave period and wavelength. While shorter period waves (between 3 - 10

days) are faster (8 – 10 m s-1) and cease their propagation between 22°S and 15°S,

longer period waves (between 15 - 30 days) are slower (3 – 6 m s-1) and show phase

propagation up to 10.5°S.

With the above results in mind, the theoretical solution of BUCHWALD and

ADAMS (1968) was again applied for the SBB and ACR to dynamically investigate

whether these waves with different phase speeds and propagation ranges represent

different CSW modes. Thus, based on an exponential decay adjustment of the

bathymetric profiles to match the shape of the continental shelf and slope region, the

dispersion relation of the first 2 theoretical modes of the free CSW was obtained.

Following the same method for calculating the phase speed propagation used in Figure

3.11, the cross-spectrum of the HYCOM 1/24° SSH series was used to obtain lag values

and, consequently, wavenumbers in the frequency domain. Thus, based on the highest

energy peaks shown in the power spectrum of Figure 3.12a, wavenumbers associated

59
with periods of 7, 10, 18 and 28 days were chosen for comparison with the theoretical

dispersion relation of free CSW.

Figure 3.13 shows the comparison between the wavenumber (and associated

frequencies) derived from the HYCOM 1/24° simulation (markers) and the dispersion

relation of the first two theoretical modes of the free CSW at three different locations

(25°S, 22°S, and 21.5°S), which are representative of the CTW phase speed propagation

changes along the region (Figure 3.11). The first location (25°S) represents a wave

propagating phase with average speeds above 10 m s-1, while the second location (22°S)

is associated with a region of phase speed reduction, and the third location (21.5°S)

represents the wave after this deceleration region.

At 25°S (black markers), most of the wave periods simulated by the model are

close to the first theoretical CSWs mode, with the exception of the 28-day period wave

(inverted triangle), which is closer to the second mode. At 22°S (blue markers), shorter

period waves lie between the first and second theoretical CSWs modes, while longer

period waves (star and inverted triangle) are closer to the second theoretical CSWs

mode. Finally, at 21.5°S (green markers), while the longer period waves (star and

inverted triangle) are closer to the first theoretical CSWs mode, the shorter period waves

(square and circle) have a dispersion relation far from both theoretical modes.

The behavior observed at 22°S occurs at another location along the continental

shelf where there is a decrease in the phase speed propagation (Figure 3.11) and a

reduction in energy (Figure 3.12a). Here, the CTW exhibit a hybrid dynamic behavior,

since the shorter period waves (7 and 10 days) are in between the theoretical curves of

the first and second free CSW modes, while the longer period waves (18 and 28 days)

are closer to the theoretical curve of the second mode. It is important to emphasize that

60
the comparison between these theoretical curves and the frequency and associated

wavenumber derived from the numerical simulations are not precise, which may be

related to the influence of other forcings that were not taken into account in the

theoretical solution of BUCHWALD and ADAMS (1968).

Figure 3.13: Comparison between the frequency–wavenumber relation observed in the


cross-spectrum of the SSH (markers) along the 50 m isobath based on a 11-year (2006-
2016) HYCOM 1/24° simulation and the theoretical dispersion relation of the first two
modes of CSW (lines) derived from the exponential topographic profiles in 3 different
regions [25°S (black), 22°S (blue) and 21.5°S (green)]. The first mode of each curve is
associated with the highest frequencies. The shadows indicate a variation of 10% of
the continental shelf width. The marks indicate waves with periods of 7 (square), 10
(circle), 18 (star), and 28 days (inverted triangle).

61
3.3.3 Cross-shore modal structure of the CTW

This section is devoted to characterizing the cross-shore modal structure of the

CTW along the SS, SBB and ACR, quantifying the importance of the baroclinic modes

for the variability of the currents in the region. EOF analyses of the filtered alongshore

velocity in the CTW band were applied in selected cross-shore sections between 32°S

and 16°S to investigate the cross-shore structure and the importance of statistical modes

for the variance of currents along the continental shelf and slope region.

The spatial variation of the EOF modes (Figure 3.14) shows that between 32°S

and 23°S, the first mode explains between 71% and 86% of the alongshore current

variance. However, between 23°S and 22°S, there is a decrease in the importance of the

first mode (59% - 63%) accompanied by an increase in the second (23% - 22%) and

third (13% - 10%) statistical modes. This increase in importance of the higher EOF

modes occurs in the same region where i) a decrease in the CTW phase speed

propagation is observed, ii) there is a reduction in the energy in the SSH along the coast

in the 3 - 30 days band and iii) waves with periods of 18 and 28 days have

characteristics closer to the second theoretical mode of the CSW. Thus, the EOF results

in conjunction with the previous results described in section 3.2, indicate that the

changes that the CTW undergo in this region occur from the coast to the continental

slope.

62
Figure 3.14: (left painel) Latitudinal variation in the percentage of importance of the
statistical modes of the EOF analysis of the alongshore velocity along selected cross-
shore sections in the SS, SBB and ACR [mode 1 (blue line), mode 2 (black line) and
mode 3 (red line)] based on a 11-year (2006-2016) HYCOM 1/24° simulation. (right
panel) Location of the cross-shore sections of velocity.

Spectral analysis of the above EOF mode time series was applied to assess the

energy distribution of the statistical modes in the frequency domain. For this purpose,

the percentage of the importance of modes 1 and 2 was calculated taking into account

the first four statistical modes. Figure 3.15 illustrates that the first EOF mode captures

energy across the CTW time band, with maximum percentages concentrated in periods

less than 15 days, while the second EOF mode essentially captures energy in periods

larger than 15 days.

63
Figure 3.15: Spectral analysis (FFT) of the time series of the first (left panel) and
second (right panel) EOF modes of the alongshore velocity at selected cross-shore
sections for the geographic region presented in Figure 3.14 and from the 11-year (2006-
2016) HYCOM 1/24° simulation. The spectral power values are presented as a
percentage considering the first four modes.

The spectra of the time series of the EOF modes suggest that the increase in the

percentual contribution of the second and third modes at 23°S and 22°S (Figure 3.14)

represents a higher influence of the CTW with periods larger than 15 days to alongshore

current variance on the shelf and slope regions. The width variation of the continental

shelf in this region may represent a favorable condition for the energetic spreading of

the CTW, which can also be amplified by stratification and result in energy transfer

between modes (WILKIN and CHAPMAN, 1990). CERDA and CASTRO (2013) show

that South Atlantic Central Water (SACW) intrusions into the SBB continental shelf

(T~15°S) in Ubatuba (23.5°S) and Cabo Frio (23°S) promote the intensification of the

stratification.

64
The importance of stratification for the dynamics of the CTW along the

Brazilian coast between 32°S and 16°S was measured using S (see Equation (1)), which

is based on the baroclinic Rossby radius and the width of the continental shelf (WANG

and MOOERS, 1976, HUTHNANCE et al., 1986). When S << 1, the hydrodynamic

response of the continental shelf to a wind-generated disturbance is barotropic in the

form of a CSW, while when S > 1, the response of the continental shelf is baroclinic. S

(CLARKE and BRINK, 1985) was estimated using a time series of the mean value of

N² for the continental shelf region of the selected cross-shore sections presented in

Figure 3.14 as well as the α values for these same locations.

The average value of S (Figure 3.16) reveals that this portion of the Brazilian

continental shelf, between 32°S and 16°S, exhibits barotropic characteristics. Mean S

values between 32°S and 24°S are less than 0.05, while between 24°S and 22°S, the

region that includes the Cabo Frio coastal upwelling system, the mean value of the S

parameter doubles and reaches 0.19 at 22°S. Between 22°S and 20°S, another

significant increase in the S parameter is also observed, with an average value of 0.55 at

approximately 20°S.

The increase of the S parameter between 24°S and 22°S can be largely explained

by the increase in stratification since the α values of this region are similar to those

farther south, where low values of S (<0.05) are observed. Conversely, the increase of

the S parameter between 22°S and 20°S is directly related to the increase in α and

stratification where the continental shelf is narrow and presents SACW intrusion paths

(PALÓCZY et al., 2016). North of 20°S, the stratification parameter shows a reduction,

especially at approximately 18°S, where α decreases due to the Abrolhos Bank.

Although the S values are less than 1, the standard deviation (shadow region) between

65
22°S and 20°S and between 18°S and 16°S indicates that seasonal stratification may

increase baroclinic conditions in this portion of the Brazilian continental shelf where the

energy reduction of the CTW in the SSH occurs along the coast (Figure 3.12a).

Figure 3.16: [a] Mean value of the CLARKE and BRINK (1985) criterion in the SS,
SBB and ACR based on 11 years (2006-2016) of HYCOM 1/24° simulation and [b] the
associated α values. The shadow region in the left panel indicates the standard
deviation.

The increase of S between 24°S and 22°S coincides with the region where there

is an increase in the importance of the second and third EOF modes (Figure 3.14).

Taking that into account, the cross-shore structure of the CTW theoretical baroclinic

modes was investigated to verify whether there is some correspondence with the

structure of the EOF modes described above. Here, the linear model of BRINK and

CHAPMAN (1987) was used to provide the cross-shore structure of the alongshore

velocity of CTW baroclinic modes derived from an average 11-year Brunt-Väissälä

66
profile located at the continental shelf break region and the cross-shore bathymetric

profile.

Figure 3.17 shows the comparison between the cross-shore structures of the first

three EOF modes and the first two baroclinic modes of the theoretical CTW at two

cross-shore sections of the SBB continental shelf, which were selected based on the S

parameter. The southernmost section is located in a region where the S parameter has a

value on the order of 10-2, while the other section is located between 24°S and 22°S, and

due to the larger stratification, it has a S parameter one order of magnitude higher.

It is important to note that this comparison involves two models with different

levels of complexity in terms of dynamics: the simpler BRINK and CHAPMAN (1987)

and the more sophisticated and complete HYCOM model. Despite this, it appears that at

both sections (Figure 3.17d, 3.17j), there is a similarity between the horizontal cross-

shore structure of the second and third EOF modes (Figure 3.17b, 3.17c, 3.17h, 3.17i)

and the first and second baroclinic CTW modes (Figure 3.17e, 3.17f, 3.17k, 3.17l).

Thus in this region, between 24°S and 22°S, where a decrease in the percentage of

importance of the first EOF mode is observed (Figure 3.14), the first two baroclinic

CTW modes become more important for explaining the variance of the currents at the

continental shelf/slope region.

67
Figure 3.17: (left panels) The first (a, g), second (b, h), and third (c, i) EOF modes
based on the analysis of the alongshore velocity at two selected cross-shore sections
located at the SBB shelf and based on the outputs of a 11-year (2006-2016) HYCOM
1/24° simulation. (right panels). The first (e, k) and second (f, l) baroclinic modes from
BRINK and CHAPMAN’s (1987) linear model for the same two cross-shore sections.
The locations of the two cross-shore sections are indicated in panels d and j.

68
Sensitivity tests were performed with the linear model of BRINK and CHAPMAN

(1987) and with the EOF analysis of the velocity sections along the southeastern

Brazilian coast to quantify the importance of the temporal variability of stratification on

the continental shelf for the CTW modal structure (not shown). Thus, for the BRINK

and CHAPMAN (1987) model, the Brunt-Väissälä profiles representing periods of

higher (summer/autumn) and lower (winter/spring) stratification on the continental shelf

were used, while the analysis of EOF was performed separately for these same two

periods. Thus in both analyses, the cross-shore structures of the CTW modes do not

change significantly, indicating that the topography of the continental shelf is the most

important factor driving the CTW cross-shore modal structure along the SBB shelf.

69
3.4 Conclusions

Analyses based on in situ data and model simulations with HYCOM demonstrate

that CTW with periods between 3 and 30 days play an important role in the

hydrodynamic variability along the Brazilian continental shelf (SS, SBB, ACR and ES),

from the coast to the continental slope. The area of generation of these waves covers a

wide latitudinal band from the Patagonian continental shelf to the southern portion of

the Brazilian continental shelf and can be divided into 2 subregions, based on spectral

and cross-correlation analysis of the SSH along the coast: south 45°S and north of 45°S.

The second subregion shows higher energy and coherence with the signal of the CTW

that propagates along the Brazilian continental shelf (SS, SBB, ACR and ES). CTW

exhibits three high-energy bands in SSH with periods between 5 - 12 days, 15 - 22 days,

and 25 - 40 days. Maximum energy peaks for these three bands are observed between

42°S and 35°S, but are still significant until of ~22oS, showing a drastic reduction north

of this latitude due to abrupt variations in the width and depth of the continental shelf,

narrowing in Tubarão Bight region followed by a widening at Abrolhos Bank, which is

a shallow region and prone to wave scattering.

The signals of CTWs propagating in the SSH along the 50 m isobath of the

continental shelf are faster in the northern portion of the Patagonian continental shelf

(between 42°S and 41°S), with average phase speed larger than 25 m s -1. Northward of

41°S, the phase speed is reduced to approximately 11 m s -1 until 24°S. After this

latitude, in the northern portion of SBB, the propagation speed is significantly reduced,

with values lower than ~3 m s-1. The seasonal stratification of the continental shelf has a

minor effect on this latitudinal variation, and the application of the free CSW theory

70
(BUCHWALD and ADAMS, 1968) confirms that this decrease in the CTW phase speed

after 24°S can be explained by the narrowing of the continental shelf. The cross-spectral

analysis of the simulated SSH fields shows that the higher frequency waves (periods

from 3 to 10 days) are faster and disperse their energy somewhere between 22°S and

15°S, in particular in the Abrolhos Bank, while shorter frequency waves (periods larger

than 15 days) are slower and may be still discernible up to 10.5°S. Since CTW

propagation was investigated in the present study only up to this latitude, one can only

speculate that such low frequency waves may propagate for even longer distances, and

reach lower latitudes.

The comparison between the theoretical dispersion relation of the free CSW and

that derived from the HYCOM model results reveals that at 25°S, the waves with

periods of 7, 10 and 18 days have frequency and wavenumber characteristics similar to

the first CSW theoretical mode, whereas waves with a period of 28 days are closer to

the second CSW theoretical mode. At 22°S, a point located in the region of phase

propagation reduction, the CTW have hybrid characteristics, where waves with periods

of 7 and 10 days have frequencies and wavenumbers between the first and second CSW

theoretical modes, while waves with periods of 18 and 28 days have characteristics

closer to the second CSW theoretical mode. Finally, at 21.5°S, while longer period

waves (18 and 28 days) have frequencies and wavenumbers closer to the first free CSW

mode, the shorter period waves (7 and 10 days) did not fit any of the theoretical modes,

which may be related to the influence of other forcings that were not taken into account

in the theoretical solution of BUCHWALD and ADAMS (1968). The EOF analysis of

cross-shore velocity sections shows that the percentage importance of the first mode is

high throughout the SS, SBB and ACR and decrease between 23°S and 22°S, where the

71
stratification parameter (S) increase due strong stratification. In this region,the first and

second baroclinic modes of the CTW present a similar to structure as the second and

third EOF modes.

72
Chapter 4

Conclusions

This thesis presents an investigation of Near-Inertial Oscillations (NIO) and

Coastal Trapped Waves (CTW) along the Brazilian continental shelf, based on in situ

observations, and the results from a high-resolution numerical simulation of the South

Atlantic Ocean. A general discussion of the main results and analysis was presented in

chapters 2 and 3. In the following pages, the principal conclusions derived from this

investigation are summarized, for each of the general questions posed in the

introductory chapter.

The NIO play an important role in superinertial variability at the shelf break. The

rotary spectral of velocity data shows a prominent energy peak near the inertial

frequency for the anticyclonic component, for all moorings along the region. Red and

blue shifts of the energy peaks were observed and related in part to the vorticity of the

surface currents associated with the relative position of the BC with respect to the

mooring stations. The characterization of the NIO between 32°S and 16°S revealed that

near-inertial currents between 32°S and 23.5°S range between 30 and 50 cm s -1 and

decrease toward lower latitudes locations, where the range is between 5 and 25 cm s −1.

The characterization performed at the Cabo Frio mooring revealed that the near-inertial

events have average amplitudes of 15 cm s-1 and a duration of 7,6 (± 4,1) days.

73
• How does the inertial energy in the upper layer of the ocean (mixed layer,

seasonal thermocline, and upper permanent thermocline) vary along the

Brazilian continental shelf-break?

The NIO present the highest energy (~ 120 cm² s -2) in the upper layer of the ocean

in the Brazilian shelf break at approximately 30°S, where the resonance between the

inertial frequency and the diurnal sea breezes approximately doubles the inertial energy

in relation to the others stations to the north. Towards lower latitudes, the energy in the

inertial band decreases by approximately 80% between the RG (32°S) and PS (16°S)

stations, and the semidiurnal tidal band becomes relatively more important for the high-

frequency variability of the surface currents.

• What is the relative importance of the NIO in relation to tides and the subinertial

band for the variance of currents at the interface between the continental shelf

and the adjacent ocean?

The NIO present the highest relative contribution (~ 30%) to the variance of

surface currents at approximately 30°S, where near-inertial events generate currents

with magnitudes ranging from 30 to 50 cm s-1 and vertical shear relevant for the vertical

mixing. The relative importance of the NIO is correlated with the near-inertial energy

variation along the Brazilian continental shelf break and the average position of the BC

in relation to the moorings.

• What are the characteristics of near-inertial currents along the water column

during representative events in terms of energy?

The NIO events at the Cabo Frio mooring present phase speed of 10 -1 cm s-1 and

vertical propagation of energy that can involve the first 175 m of the water column. The

first EOF mode explains 67% of the variance of the inertial currents in the water

74
column, with a vertical profile that indicates the dominance of the first baroclinic mode.

The vertical structure of NIO is significantly tied to the thermal structure of the water

column. When a seasonal thermocline is formed, velocities associated with NIO are

restricted to the first few meters near the surface. When the mixing layer is deeper, the

NIO deepens as well.

CTW with periods between approximately 3 and 30 days play an important role in

the subinertial variability on the Brazilian continental shelf between SS and ACR. The

in situ data shows that CTW propagation affects not only the SSH at the coast, but also

the intensity and directon of the currents at the shelf break, reaching the domain of the

BC.

• What are the geographical limits of the generation area of the CTW that

propagate along the Brazilian continental shelf between 34°S and 11°S?

The CTW with periods between 3 and 30 days that propagate along the Brazilian

continental shelf between 34°S and 11°S are generated in a wide latitudinal band, from

the Patagonian shelf to the southern portion of the Brazilian continental shelf. However,

the region north of 45°S presents higher energy, and more significant coherence with the

signal of the waves that propagates along the Brazilian coast.

• How does the energy of CTW vary towards lower latitudes?

The spectral analysis of SSH indicates three bands with higher energy, with periods

between approximately 5 and 12 days, 15 and 22 days, and 25 and 40 days. The energy

of the CTW decreases from 104 to 102 cm² cph-1 towards lower latitudes. A drastic

reduction of the CTW signal is observed to occur north of 22°S, due to the abrupt

variations of the width and depth of the continental shelf, which narrows at the Tubarão

Bight and is followed by the wide and shallow Abrolhos Bank.

75
• How does the phase speed of the CTW vary along the coast?

The signals of CTW propagating along the continental shelf are faster in their

generation region, in the northern portion of the Patagonian continental shelf (between

42°S and 41°S), with average phase speed larger than 25 m s -1. Northward of 41°S, the

phase speed is reduced to approximately 11 m s-1 for most of the south-southeastern

Brazilian region, until 24°S. After this latitude, in the northern portion of SBB, the

propagation speed is significantly reduced, with values lower than ~3 m s-1. When

evaluating this signal for different frequency bands, however, it becomes clear that the

propagation speed of the CTW also varies along the frequency domain. CTW with

periods between 3 and 10 days are faster (~10 m s -1), and cease their propagation

between 22°S and 15°S, while waves with periods longer than 15 days are slower (~5 m

s-1) and propagate phase until 10.5°S. This latitude does not indicate the end of the wave

propagation but rather the limit of the analysis domain in this study. It is speculated that

these longer period waves may propagate for longer distances reaching even lower

latitudes.

• What is the importance of variations in the morphological and stratification

characteristics of continental shelf for the phase speed propagation and the

modal structure of the CTW?

The phase speed along the propagation on the coast north of station MP (38.05°S)

is slightly higher (~ 1 m s-1) during winter/spring, although the seasonal stratification

does not change the pattern of phase speed variation along the CTW propagation

described in the previous topic. The application of the free CSW theory (BUCHWALD

and ADAMS, 1968) confirms that the decrease in the phase speed propagation north of

24°S can be explained by the narrowing of the continental shelf.

76
In the SBB, CTW with periods between 5 and 22 days have frequency and

wavenumber characteristics similar to the first CSW theoretical mode, whereas waves

with periods larger than 22 days show proximity to the second CSW theoretical mode.

In the region of phase speed propagation reduction, the CTW have hybrid

characteristics, where the waves with periods between 5 and 12 days have frequencies

and wavenumbers between the first and second CSW theoretical modes, while waves

with periods larger than 15 days have characteristics closer to the second CSW

theoretical mode.

The average value of the stratification parameter S (CLARKE and BRINK, 1985)

indicates that the Brazilian continental shelf between 32°S and 16°S should have a

barotropic response to disturbances generated by the wind, which corroborates the

previous result of phase speed propagation of CTW. However, in the region between

24°S and 22°, where the phase speed is significantly reduced, the parameter S increases

as a result of the larger stratification on the continental shelf. In this region, the EOF

analysis derived from the simulated cross-shore velocity sections show a percentage

reduction of the first mode. The spectral analysis of the time series of the EOF modes

suggests that the increase in importance of the second and third modes in this region is

caused by the increased influence of waves with periods larger than 15 days. The

comparison between the cross-shore structure of the EOF modes and theoretical

baroclinic modes of the CTW (BRINK and CHAPMAN, 1987) reveals that the first and

second baroclinic modes of the CTW present a similar structure to the second and third

EOF modes, indicating that the importance of the baroclinic modes of CTW increases in

this region of wave deceleration.

77
Finally, this thesis shows, based on in situ data and numerical modeling, that NIO

and CTW are important transients for the hydrodynamics of the South-Southeastern

Brazilian continental shelf. The results presented in Chapter 2 and 3 show an interaction

between the hydrodynamics of the continental shelf and the circulation of the BC, which

suggests that the variability of NIO characteristics (frequency of the near-inertial energy

peak of the currents at the continental shelf break) and CTW (variation of the SSH on

the coast band-pass filtered in the period range of 3 - 30 days) have a relationship with

changes in the circulation of the adjacent oceanic region. Therefore, future studies can

provide important information about the relationship between the mesoscale variability

of the BC and the NIO characteristics in the currents at the continental shelf break. In

addition, future research should contribute to the understanding of the importance of

CTW for the hydrodynamics of the South/Southeastern continental shelf of Brazil by

investigating the vertical scope of the correlation between the variability of SSH on the

coast and the intensity and direction of currents along the continental slope.

78
References
ALFORD, M. H., WHITMONT, M., 2007, “Seasonal and spatial variability of near-
inertial kinetic energy from historical moored velocity records.” Journal of
Physical Oceanography, v. 37, n. 8, pp. 2022–2037.

ALFORD, M. H., MACKINNON, J. A., SIMMONS, H. L., NASH, J. D., 2016, “Near-
inertial internal gravity waves in the ocean”. Annual Review of Marine Science,
v. 8, n. 1, pp. 95–123.

ALFORD, M.H., SLOYAN, B.M., SIMMONS, H.L., 2017, “Internal waves in the East
Australian Current.” Geophysical Research Letter, v. 44,n.24, pp. 12,280–12,288.

ASSIREU, A. T., STEVENSON, M. R., STECH, J. L., 2003, “Surface circulation and
kinetic energy in the SW Atlantic obtained by drifters”. Continental Shelf
Research, v. 23, n. 2, pp. 145–157.

ASSIREU, A. T., DAUHUT, T., dos SANTOS, F. A., LORENZZETTI, J. A., 2017,
“Near-inertial motions in the Brazil Current at 24°S-36°S: observations by
satellite tracked drifters.” Continental Shelf Research, v. 145, pp. 1–12.

BARROS, V. R., DOYLE, M. E., CAMILLONI, I. A., 2008, “Precipitation trends in


southeastern South America: relationship with ENSO phases and with low-
level circulation.” Theoretical and Applied Climatology, v. 93, n. 1–2, pp. 19–33.

BELL, M.J., LEFÈBVRE, M., LE TRAON, P.-Y., SMITH, N., WILMER-BECKER, K.,
2009, “GODAE: The Global Ocean Data Assimilation Experiment.”
Oceanography. V. 22, n. 3, pp. 14–21.

BLECK, R., 2002, “An oceanic general circulation model framed in hybrid isopycnic-
Cartesian coordinates.” Ocean Modelling, v. 4, n. 1, pp. 55–88.

79
BRINK, K. H., CHAPMAN, D. C. Programs for computing properties of coastal-
trapped waves and wind-driven motions over the continental shelf and slope. 2 ed.
Woods Hole, MA: Woods Hole Oceanographic Institution, 1987.

BRINK, K. H., 1991, “Coastal-Trapped Waves and Wind-Driven Currents Over the
Continental Shelf.” Annual Review of Fluid Mechanics, v. 23, n. 1, pp. 389–412.

BUCHWALD, V. T., ADAMS, J. K., 1968, “The propagation of continental shelf


waves.” Proceedings of the Royal Society A, v. 305, n. 4, pp. 235–250.

CASTELAO, R. M., BARTH, J. A., 2006, “Upwelling around Cabo Frio, Brazil: The
importance of wind stress curl.” Geophysical Research Letters, v. 33, n. 3, pp.
L03602.

CASTRO, B. M., 1990, “Wind driven currents in the channel of São Sebastião: winter,
1979.” Boletim do Instituto Oceanográfico, v. 38, n. 2, pp. 111–132.

CASTRO, B. M., LEE, T. N., 1995, “Wind-forced sea level variability on the southeast
Brazilian shelf.” Journal of Geophysical Research, v. 100, n. C8, pp. 16045–
16056.

CASTRO FILHO, B. M. DE, MIRANDA, L. B. DE., “Physical oceanography of the


western atlantic continental shelf located between 4°N and 34°S - coastal segment
(4,W). In: The Sea, vol.11. The Global Coastal Ocean: Regional Studies and
Syntheses, John Wiley & Sons, pp. 209-251, 1998.

CASTRO, B. M., BRANDINI, F. P., PIRES-VANIN., “Multidisciplinary oceanographic


processes on the western atlantic continental shelf between 4°N and 34° S.” In:
The Sea, v. 14A, The Global coastal Ocean: Interdisciplinary Regional Studies
and Syntheses, Harvard University Press, pp. 259-293, 2006.

CAMPOS, E. J. D., VELHOTE, D., da SILVEIRA, I. C. A., 2000, “Shelf break


upwelling driven by Brazil Current cyclonic meanders.” Geophysical Research
Letter, v. 27, n. 6, pp. 751–754.

80
CAO, A., GUO, Z., SONG J., LV, X., HE, H., FAN, W., 2018, “Near-inertial waves and
their underlying mechanisms based on the South China Sea Internal Wave
Experiment (2010– 2011).” Journal Geophysical Research: Oceans, v. 123, n. 7,
pp. 5026–5040.

CASTELAO, R. M., BARTH, J. A., 2006, “Upwelling around Cabo Frio, Brazil: the
importance of wind stress curl.” Geophysical Research Letter, v. 33, pp. 2–5.

CERDA, C., CASTRO, B. M., 2013, “Hydrographic climatology of South Brazil Bight
shelf waters between Sao Sebastiao (24°S) and Cabo Sao Tome (22°S).”
Continental Shelf Research, v. 89, pp. 5–14..

CHAIGNEAU, A,. PIZARRO, O., ROJAS, W., 2008, “Global climatology of near-
inertial current characteristics from Lagrangian observations.” Geophysical
Research Letter, v. 35, n. 13, pp. L13603.

CHAVANNE, C. P., FIRING, E., ASCANI, F., 2012, “Inertial oscillations in geostrophic
flow: is the inertial frequency shifted by ζ/2 or by ζ?.” Journal Physical
Oceanography, v. 42, n. 5, pp. 884–888.

CHEN, C., REID, R. O., NOWLIN, W. D., 1996, “Near-inertial oscillations over the
Texas-Louisiana shelf.” Journal Geophysical Research: Oceans, v. 101(C2), pp.
3509–3524.

CHEN, G., XUE, H., WANG, D., XIE, Q., 2013, “Observed near-inertial kinetic energy
in the northwestern South China Sea.” Journal Geophysical Research. v. 118, pp.
4965–4977.

CLARKE, A. J., BRINK, K. H., 1985, “The Response of Stratified, Frictional Flow of
Shelf and Slope Waters to Fluctuating Large-Scale, Low-Frequency Wind
Forcing.” Journal of Physical Oceanography, v. 15, n. 4, pp. 439–453.

CLARKE, A. J., VAN GORDER, S., 1986, “A Method for Estimating Wind-Driven
Frictional, Time-Dependent, Stratified Shelf and Slope Water Flow.” Journal of
Physical Oceanography, v. 16, n. 6, pp. 1013–1028.

81
COSTA, V.S., MILL, G.N., GABIOUX, M., GROSSMANN-MATHESON, G.S.,
PAIVA, A.M., 2017. “The recirculation of the intermediate western boundary
current at the Tubarão Bight – Brazil.” Deep Sea Research. Part I
Oceanographic Research Papers, v. 120, pp. 48–60.

CUYPERS, Y., VAILLANT, X. L., BOURUET-AUBERTOT, P., VIALARD, J.,


MCPHADEN, M. J., 2013, “Tropical storm-induced near-inertial internal waves
during the Cirene experiment: energy fluxes and impact on vertical mixing.”
Journal Geophysical Research:Oceans, v. 118, pp. 358–380.

D’ASARO, E. A., 1985, “Upper ocean temperature structure, inertial currents, and
Richardson numbers observed during strong meteorological forcing.” Journal
Physical Oceanography v. 15, n. 7, pp. 943–962.

D’ASARO, E. A., ERIKSEN, C. C., LEVINE, M. D., NIILER, P., PAULSON, C. A.,
MEURS, P. V., 1995, “Upper-ocean inertial currents forced by a strong storm.
Part I: data and comparisons with linear theory.” Journal Physical
Oceanography, v. 25, n. 11, pp. 2909–2936.

DOTTORI, M., CASTRO, B. M., 2009, “The response of the Sao Paulo Continental
Shelf, Brazil, to synoptic winds.” Ocean Dynamics, v. 59, n. 4, pp. 603–614.

DOTTORI, M., CASTRO, B. M., 2018, “The role of remote wind forcing in the
subinertial current variability in the central and northern parts of the South
Brazil Bight.” Ocean Dynamics, v. 68, n. 6, pp. 677–688.

ELIPOT, S., LUMPKIN, R., 2008, “Spectral description of oceanic near-surface


variability.” Geophysical Research Letter. v. 35, n. 5, pp. L05606.

ELIPOT, S., LUMPKIN, R., PRIETO, G., 2010, “Modification of inertial oscillations
by the mesoscale eddy field.” Journal Geophysical Research: Oceans, v. 115, pp.
C09010.

82
EVANS, D. L., SIGNORINI, S. S., 1985, “Vertical structure of the Brazil Current.”
Nature, v. 315, n. 6014, pp. 48–50.

FILIPPO, A., KJERFVE, B., TORRES JÚNIOR, A.R., FERNANDES, A.M., 2012,
“Low-frequency variability of sea level along the Mid-Atlantic Coast of South
America, in 1983.” Brazilian Journal of Geophysics, v. 30, pp. 5-14.

FRANÇA, B. R. L., 2013, Ondas Confinadas Costeiras na plataforma continental Sul-


Sudeste do Brasil. Dissertação de Mestrado, COPPE / Universidade Federal do
Rio de Janeiro (UFRJ), Rio de Janeiro, Agosto.

FRANKS, P. J. S., 1994, “Thin layers of phytoplankton: a model of formation by near-


inertial wave shear.” Deep-Sea Research, v. 42, n. 1, pp. 75–91.

GABIOUX, M. G., da COSTA, V.S., SOUZA, J.M.A.C., OLIVEIRA, B.F., PAIVA,


A.D.M., 2013, “Modeling the South Atlantic Ocean from medium to high-
resolution.” Revista Brasileira de Geofísica, v. 31, n. 2, pp. 229-242.

GARREAUD, R. D., 2000, “Cold Air Incursions over Subtropical South America:
Mean Structure and Dynamics.” Monthly Weather Review, v. 128, n. 7, p. 2544–
2559.

GARREAUD, R. D., 2009, “The Andes climate and weather.” Advances in


Geosciences, v. 22, pp. 3–11.

GARRETT, C., 2001, “What is the “near-inertial” band and why is it different from the
rest of the internal wave spectrum?” Journal Physical Oceanography, v. 31, n. 4,
pp. 962–971.

GARRETT, C., MUNK, W., 1975, “Space-time scales of internal waves: a progress
report.” Journal Geophysical Research, v. 80, n. 3, pp. 291–297.

GELARO, R., McCARTY, W., SUÁREZ, M.J., TODLING, R., MOLOD, A., TAKACS,
L., RANDLES, C.A., DARMENOV, A., BOSILOVICH, M.G., REICHLE, R.,
WARGAN, K., COY, L., CULLATHER, R., DRAPER, C., AKELLA, S.,

83
BUCHARD, V., CONATY, A., da SILVA, A.M., GU, W., KIM, G.K.,
KOSTER, R., LUCCHESI, R., MERKOVA, D., NIELSEN, J.E., PARTYKA, G.,
PAWSON, S., PUTMAN, W.; RIENECKER, M., SCHUBERT, S.D.,
SIEKIEWICZ, M., ZHAO, B., 2017, “The Modern-Era Retrospective Analysis for
Research and Applications, Version 2 (MERRA-2).” Journal of Climate, v. 30, n.
14, pp. 5419–5454.

GILLE, S. T., SMITH, S. G. L., STATOM, N. M., 2005, “Global observations of the
land breeze.” Geophysical Research Letter, v. 32, n. 5, pp. L05605.

GONELLA, J. A., 1972, “A rotary-component method for analysing meteorological and


oceanographic vector time series.” Deep-Sea Research and Oceanography
Abstracts, v. 19, n. 12, pp. 833–846.

GOOS/BRAZIL/GLOSS Brasil, 2019. Global Ocean Observing System. Disponível


online: <http://www.goosbrasil.org/gloss/>.

GOOS/BRAZIL/PNBOIA Brasil, 2019. Global Ocean Observing System. Disponivel


em: <http://www.goosbrasil.org/pnboia/>.

GRIMSHAW, R., 1977, “The Effects of a Variable Coriolis Parameter, Coastline


Curvature and Variable Bottom Topography on Continental Shelf Waves.” Journal
of Physical Oceanography, v. 7, n. 4, pp. 547–554.

HALLIWELL, G.R., SRINIVASAN, A., KOURAFALOU, V., YANG, H., WILLEY, D.,
LE HÉNAFF, M., ATLAS, R., 2014. “Rigorous Evaluation of a Fraternal Twin
Ocean OSSE System for the Open Gulf of Mexico.” Journal of Atmospheric
and Oceanic Technology, v. 31, pp. 105–130.

HUTHNANCE, J. M., 1978, “On Coastal Trapped Waves: Analysis and Numerical
Calculation by Inverse Iteration.” Journal of Physical Oceanography, v. 8, n. 1,
pp. 74–92.

84
HUTHNANCE, J. M., MYSAK, L. A., WANG, D.-P., “Coastal trapped waves.” In:
Coastal and Estuarine Sciences, v. 3, Baroclinic Processes on Continental Shelves,
Washington, D. C.: American Geophysical Union, pp. 1–18, 1986.

HUTHNANCE, J. M., “Coastal Trapped Waves.” In: Encyclopedia of Ocean Sciences,


v. 1, Elsevier, Academic Press, San Diego, USA, pp. 591–598, 2001.

HUGHES, C. W., MEREDITH, M. P., 2006, “Coherent sea-level fluctuations along the
global continental slope.” Philosophical Transactions of the Royal Society A:
Mathematical, Physical and Engineering Sciences, v. 364, n. 1841, pp. 885–901.

HUGHES, C. W., FUKUMORI, I., GRIFFIES, S.M., HUTHNANCE, J.M., MINOBE,


S., SPENCE, P., THOMPSON, K.R., WISE, A., 2019, “Sea Level and the Role of
Coastal Trapped Waves in Mediating the Influence of the Open Ocean on the
Coast.” Surveys in Geophysics, v. 40, n. 6, pp. 1467–1492.

HYDER, P., SIMPSON, J. H., XING, J., GILLE, S. T., 2011, “Observations over an
annual cycle and simulations of wind-forced oscillations near the critical
latitude for diurnal–inertial resonance.” Continental Shelf Research, v. 31, n. 15,
pp. 1576–1591.

ILLIG, S., BACHÈLERY, M., CADIER, E., 2018, “Subseasonal Coastal‐Trapped Wave
Propagations in the Southeastern Pacific and Atlantic Oceans: 2. Wave
Characteristics and Connection With the Equatorial Variability.” Journal of
Geophysical Research: Oceans, v. 123, n. 6, pp. 3942–3961.

KATO, S., LOEB, N.G., ROSE, F.G., DOELLING, D.R., RUTAN, D.A., CALDWELL,
T.E., YU, L., WELLER, R.A., 2013, “Surface Irradiances Consistent with
CERES-Derived Top-of-Atmosphere Shortwave and Longwave Irradiances.”
Journal of Climate, v. 26, n. 9, pp. 2719–2740.

KUNDU, P. K., 1976, “An analysis of inertial oscillations observed near Oregon coast.”
Journal Physical Oceanography, v. 6, n. 6, pp. 879–893.

85
KUNZE, E., 1985, “Near-inertial wave propagation in geostrophic shear.” Journal
Physical Oceanography, v. 15, n. 5, pp. 544–565.

LIMA, J.A.M., MARTINS, R.P., TANAJURA, C.A.S., PAIVA, A.M., CIRANO, M.,
CAMPOS, E.J.D., SOARES, I.D., FRANÇA, G.B., OBINO, R.S.,
ALVARENGA, J.B.R., 2013. “Design and implementation of the Oceanographic
modeling and observation Network (REMO) for operational oceanography and
ocean forecasting.” Revista Brasileira de Geofísica, v. 31, pp. 209-228.

LIU, Y., SAN LIANG, X., WEISBERG, R. H., 2007, “Rectification of the Bias in the
Wavelet Power Spectrum.” Journal of Atmospheric and Oceanic Technology, v. 24,
n. 12, pp. 2093–2102.

LOCARNINI, R. A., MISHONOV, A.V., ANTONOV, J.I., BOYER, T.P., GARCIA,


H.E., BARANOVA, O.K., ZWENG, M.M., PAVER, C.R., REAGAN, J.R.,
JOHNSON, D.R., HAMILTON, M., SEIDOV, D. World ocean atlas 2013. Volume
1, Temperature. NOAA atlas NESDIS 73, pp. 1-40, 2013.

LUCAS, A. J., PITCHER, G. C., PROBYN, T. A., KUDELA, R. M., 2014, “The
influence of diurnal winds on phytoplankton dynamics in a coastal upwelling
system off southwestern Africa.” Deep-Sea Research II Topical Studies in
Oceanography, v. 101, pp. 50–62.

MIRANDA, L. B., CASTRO, F. B. M., 1982, “Geostrophic flow conditions of the


Brazil Current at 19°S.” Ciencia Interamericana, v. 22, n. 1–2, pp. 44–48.

MUNK, W., WUNSCH, C., 1998, “Abyssal recipes ii: energetics of tidal and wind
mixing.” Deep-Sea Research I Oceanography Research Papers, v. 45, n. 12, pp.
1977–2010.

MYSAK, L. A., 1980, “Topographically Trapped Waves.” Annual Review of Fluid


Mechanics, v. 12, n. 1, pp. 45–76.

86
NAM, S., SEND, U., 2013, “Resonant diurnal oscillations and mean along-shore flows
driven by sea/land breeze forcing in the coastal Southern California Bight.”
Journal Physical Oceanography, v. 43, n. 3, pp. 616–630.

PAIVA, A. M., CHASSIGNET, E. P., 2001, “The Impact of Surface Flux


Parameterizations on the Modeling of the North Atlantic Ocean.” Journal of
Physical Oceanography, v. 31, n. 7, pp. 1860–1879.

PALÓCZY, A., BRINK, K.H., da SILVEIRA, I.C.A., ARRUDA, W.Z., MARTINS, R.P.,
2016, “Pathways and mechanisms of offshore water intrusions on the Espírito
Santo Basin shelf (18°S–22°S, Brazil).” Journal of Geophysical Research:
Oceans, v. 121, n. 7, pp. 5134–5163.

PETERSON, R. G., STRAMMA, L., 1991, “Upper-level circulation in the South


Atlantic Ocean.” Progress in Oceanography, v. 26, n. 1, pp. 1–73.

PIETRI, A., ECHEVIN, V., TESTOR, P., CHAIGNEAU, A., MORTIER, L., GRADOS,
C., ALBERT, A., 2014, “Impact of a coastal-trapped wave on the near-coastal
circulation of the Peru upwelling system from glider data.” Journal of
Geophysical Research: Oceans, v. 119, n. 3, pp. 2109–2120.

PIOLA, A. R., PALMA, E.D., BIANCHI, A.A., CASTRO, B.M., DOTTORI, M.,
GUERRERO, R.A., MARRARI, M., MATANO, R.P., MÖLLER, O.O.,
SARACENO, M., “Physical Oceanography of the SW Atlantic Shelf: A
Review.” In: Plankton Ecology of the Southwestern Atlantic, Cham: Springer
International Publishing, pp. 37–56, 2018.

POLI, A., ARTANA, C., PROVOST, C., SIRVEN, J., SENNÉCHAEL, N., CUYPERS,
Y., LELLOUCHE, J., 2020. “Anatomy of Subinertial Waves Along the Patagonian
Shelf Break in a 1/12° Global Operational Model.” Journal of Geophysical
Research: Oceans, v. 125, pp. 1-21.

POLLARD, R., 1980, “Properties of near-surface inertial oscillations.” Journal


Physical Oceanography, v. 10, pp. 385–398.

87
POLLARD, R., MILLARD, R., 1970, “Comparison between observed and simulated
wind-generated inertial oscillations.”Deep-Sea Research and Oceanography
Abstracts, v. 17, n. 4, pp. 813–821.

ROCHA, C. B., da SILVEIRA, I. C. A., CASTRO, B. M., LIMA, J. A. M. L., 2014,


“Vertical structure, energetics, and dynamics of the Brazil Current System at
22°S-28°S.” Journal of Geophysical Research: Oceans, v. 119, n. 1, pp. 52–69.

ROUSSENOV, V. M., WILLIAMS, R. G., HUGHES, C. W., BINGHAM, R. J., 2008,


“Boundary wave communication of bottom pressure and overturning changes for
the North Atlantic.” Journal of Geophysical Research, v. 113, n. C08042, pp. 1-12.

RIVAS, A., PIOLA, A., 2005, “Near inertial oscillations at the shelf off northern
Patagonia.” Atlântica (Rio Grande), v. 27, n. 2, pp. 75–86.

SAHA, S., MOORTHI, S., PAN, H., WU, X., WANG, J., NADIGA, S., TRIPP, P.,
KISTLER, R., WOOLLEN, J., BEHRINGER, D., LIU, H., STOKES, D.,
GRUMBINE, R., GAYNO, G., WANG, J, HOU, Y., CHUANG, H., JUANG, H.
H., SELA, J., IREDELL, M., TREADON, K. R., DELST, D. P. V., KEYSER,
D., DERBER, J., EK, M., MENG, J., WEI, R. Y. H., LORD, S., VAND DEN
DOOL H., KUMAR, A., WANG, W., LONG, C., CHELLIAH, M., XUE, Y.,
HUANG, B., SCHEMM, J., EBISUZAKI, W., LIN, R., XIE, P., CHEN, M.,
ZHOU, S., HIGGINS, W., ZOU, C., LIU, Q., CHEN, Y., HAN, Y.,
CUCURULL, L., REYNOLDS, R. W., RUTLEDGE, G., GOLDBERG, M.,
2010, “The NCEP climate forecast system reanalysis.” Bulletin of the American
Meteorological Society, v. 91, pp. 1015–1057.

SARACENO, M., PROVOST, C., PIOLA, A. R., 2005, “On the relationship between
satellite-retrieved surface temperature fronts and chlorophyll a in the western
South Atlantic.” Journal of Geophysical Research, v. 110, n. C11, pp. C11016.

SCHULZ, W. J. Jr., MIED, R. P., SNOW, C. M., 2012, “Continental Shelf Wave
Propagation in the Mid-Atlantic Bight: A General Dispersion Relation.” Journal
of Physical Oceanography, v. 42, n. 4, p. 558–568.

88
SHEARMAN, R. K., 2005, “Observations of near-inertial current variability on the new
England shelf.” Journal Geophysical Research: Oceans. v. 110(C2), C02012.
SHEARMAN, R. K., LENTZ, S. J., 2004, “Observations of tidal variability on the New
England shelf.” Journal Geophysical Research: Oceans, v. 109(C6):C06010.

SILVEIRA, I. C. A., SCHMIDT, A. C. K., CAMPO, E. J. D., GODOI, S. S., IKEDA, Y.,
2000, “The Brazil Current off the Eastern Brazilian Coast.” Revista Brasileira de
Oceanografia, v. 48, n. 2, pp. 171–183.

SIMMONS, H. L., ALFORD, M. H., 2012, “Simulating the long-range swell of internal
waves generated by ocean storms.” Oceanography, v. 25, n. 2, pp. 30–41.

SIMPSONS, J. H., RIPPETH, T., 2002, “Forced oscillations near the critical latitude for
diurnal-inertial resonance.” Journal Physical Oceanography, v. 32, pp. 177–187.

SOBARZO, M., SHEARMAN, R. K., LENTZ, S., 2007, “Near-inertial motions over
the continental shelf off concepción, central Chile.” Progress in Oceanography,
v. 75, n. 3 pp. 348–362.

STECH, J. L., LORENZZETTI, J. A., 1992, “The response of the South Brazil Bight to
the passage of wintertime cold fronts.” Journal Geophysical Research: Oceans,
v. 97, C6, pp. 9507–9520.

SUBEESH, M., UNNIKRISHNAN, A., 2016, “Observed internal tides and near-inertial
waves on the continental shelf and slope off Jaigarh, central west coast of India.”
Journal of Marine System, v. 157, pp. 1–19.

TORRENCE, C., COMPO, G. P. A., 1998, “Practical Guide to Wavelet Analysis.”


Bulletin of the American Meteorological Society, v. 79, n. 1, pp. 61–78.

UHSLC Legacy Data Portal, 2019. University of Hawaii Sea Level Center. Disponível
em:<https://uhslc.soest.hawaii.edu/data/>.

89
VALENTIN, J. L., “The Cabo Frio Upwelling System, Brazil.” In: Coastal Marine
Ecosystems of Latin America. v. 144, Ecological Studies, Berlin, Heidelberg:
Springer Berlin Heidelberg, pp. 97–105, 2001.

VIVIER, F., PROVOST, C., MEREDITH, M., 2001, “Remote and Local Forcing in the
Brazil-Malvinas Region.”Journal of Physical Oceanography,v. 31, pp. 892–
913.

WANG, D.-P., MOOERS, C. N. K., 1976, “Coastal-Trapped Waves in a Continuously


Stratified Ocean.” Journal of Physical Oceanography, v. 6, n. 6, pp. 853–863.

WEBSTER, F., 1968, “Observations of inertial-period motions in the deep sea.”


Reviews of Geophysics, v. 6, n. 4, pp. 473–490.

WEBSTER, I., 1987, “Scattering of Coastally Trapped Waves by Changes in


Continental Shelf Width.” Journal of Physical Oceanography, v. 17, n. 7, pp.
928–937.

WILKIN, J. L., CHAPMAN, D. C., 1990, “Scattering of Coastal-Trapped Waves by


Irregularities in Coastline and Topography.” Journal of Physical Oceanography,
v. 20, n. 3, pp. 396–421.

WOODHAM, R., BRASSINGTON, G.B., ROBERTSON, R.; ALVES, O., 2013,


“Propagation characteristics of coastally trapped waves on the Australian
Continental Shelf.” Journal of Geophysical Research: Oceans, v. 118, n. 9, p.
4461–4473.

WOODWORTH, P. L., MELET, A., MARCOS, M., RAY, R.D., WÖPPELMANN, G.,
SASAKI, Y.N., CIRANO, M., HIBBERT, A., HUTHNANCE, J.M.,
MONSERRAT, S., MERRIFIELD, M.A., 2019, “Forcing Factors Affecting Sea
Level Changes at the Coast.” Surveys in Geophysics, v. 40, n. 6, pp. 1351–1397.

WISE, A., HUGHES, C.W., POLTON, J.A., HUTHNANCE, J.M., 2020, “Leaky Slope
Waves and Sea Level: Unusual Consequences of the Beta Effect along Western

90
Boundaries with Bottom Topography and Dissipation.” Journal of Physical
Oceanography, v. 50, n. 1, pp. 217–237.

ZWENG, M. M., REAGAN, J.R., ANTONOV, J.I., LOCARNINI, R.A., MISHONOV,


A.V., BOYER, T.P., GARCIA, H.E., BARANOVA, O.K., JOHNSON, D.R.,
SEIDOV, D., BIDDLE, M.M. World ocean atlas 2013. Volume 2, Salinity. NOAA
atlas NESDIS 74, pp. 1-39, 2013.

91

You might also like