Calhoun: The NPS Institutional Archive

DSpace Repository

Theses and Dissertations 1. Thesis and Dissertation Collection, all items

2017-09

Direct and remote effects of topography and

orientation, and the dynamics of mesoscale eddies

Gulliver, Larry T.
Monterey, California. Naval Postgraduate School

http://hdl.handle.net/10945/56130

This publication is a work of the U.S. Government as defined in Title 17, United
States Code, Section 101. Copyright protection is not available for this work in the
United States.

Downloaded from NPS Archive: Calhoun

 NAVAL
POSTGRADUATE
SCHOOL
MONTEREY, CALIFORNIA

THESIS

DIRECT AND REMOTE EFFECTS OF TOPOGRAPHY

AND ORIENTATION, AND THE DYNAMICS OF
MESOSCALE EDDIES

by

Larry T. Gulliver

September 2017

Thesis Advisor: Timour Radko

Second Reader: Justin M. Brown

Approved for public release. Distribution is unlimited.

THIS PAGE INTENTIONALLY LEFT BLANK
 REPORT DOCUMENTATION PAGE Form Approved OMB
No. 0704–0188
Public reporting burden for this collection of information is estimated to average 1 hour per response, including the time for reviewing
instruction, searching existing data sources, gathering and maintaining the data needed, and completing and reviewing the collection
of information. Send comments regarding this burden estimate or any other aspect of this collection of information, including
suggestions for reducing this burden, to Washington headquarters Services, Directorate for Information Operations and Reports, 1215
Jefferson Davis Highway, Suite 1204, Arlington, VA 22202-4302, and to the Office of Management and Budget, Paperwork
Reduction Project (0704-0188) Washington, DC 20503.
1. AGENCY USE ONLY 2. REPORT DATE 3. REPORT TYPE AND DATES COVERED
(Leave blank) September 2017 Master’s thesis
4. TITLE AND SUBTITLE 5. FUNDING NUMBERS
DIRECT AND REMOTE EFFECTS OF TOPOGRAPHY AND
ORIENTATION, AND THE DYNAMICS OF MESOSCALE EDDIES
6. AUTHOR(S) Larry T. Gulliver
7. PERFORMING ORGANIZATION NAME(S) AND ADDRESS(ES) 8. PERFORMING
Naval Postgraduate School ORGANIZATION REPORT
Monterey, CA 93943-5000 NUMBER
9. SPONSORING /MONITORING AGENCY NAME(S) AND 10. SPONSORING /
ADDRESS(ES) MONITORING AGENCY
N/A REPORT NUMBER
11. SUPPLEMENTARY NOTES The views expressed in this thesis are those of the author and do not reflect the
official policy or position of the Department of Defense or the U.S. Government. IRB Protocol number ____N/A____.
12a. DISTRIBUTION / AVAILABILITY STATEMENT 12b. DISTRIBUTION CODE
Approved for public release. Distribution is unlimited.
13. ABSTRACT (maximum 200 words)
Baroclinic instability in the ocean is a primary cause of mesoscale eddies, which are pockets of water
in the scale of 100km that have different density, thermal, and rotational characteristics than their
surroundings. First observed in the early 1900s, eddies are thought to be a predominant reason for the heat
flux between the equator and the poles in both the ocean and the atmosphere. In attempt to understand this
process better, this study uses a series of numerical simulations performed on high performance computing
systems. The calculations are based on the Massachusetts Institute of Technology general circulation
model, which is used to compare lateral heat transport between different simulations. The specific
objectives of this project include
i) Comparison the direct and remote interactions of shear with topographic slope. The direct scenario
is one in which the shear extends throughout the entire ocean depth and is therefore in direct contact with
the sea floor, whereas in the remote scenario there is a spatial separation between the shear in the upper
half of the basin and the bottom topography,
ii) Analysis of the system response to changes in the zonal and meridional seafloor slope and
iii), Investigation of the effect of orientation changes in the mean large-scale current on cross-flow
fluxes.
The lateral heat transport and diffusivity of these simulations are then compared to our analytic model,
known as Growth Rate Balance, which is based on the balance between growth rate (primary) instabilities
deduced from linear theory and numerically generated secondary instabilities.
14. SUBJECT TERMS 15. NUMBER OF
mesoscale variability, ocean eddies, baroclinic instability, topography, Task Force Ocean, PAGES
Massachusetts Institute of Technology general circulation model, ocean modelling, Growth 59
Rate Balance 16. PRICE CODE
17. SECURITY 18. SECURITY 19. SECURITY 20. LIMITATION
CLASSIFICATION OF CLASSIFICATION OF THIS CLASSIFICATION OF OF ABSTRACT
REPORT PAGE ABSTRACT
Unclassified Unclassified Unclassified UU
NSN 7540–01-280-5500 Standard Form 298 (Rev. 2–89)
Prescribed by ANSI Std. 239–18

i
THIS PAGE INTENTIONALLY LEFT BLANK
 Approved for public release. Distribution is unlimited.

DIRECT AND REMOTE EFFECTS OF TOPOGRAPHY AND ORIENTATION,

AND THE DYNAMICS OF MESOSCALE EDDIES

Larry T. Gulliver
Lieutenant Commander, United States Navy
B.S., SUNY Maritime College, 2006

Submitted in partial fulfillment of the

requirements for the degree of

MASTER OF SCIENCE IN METEOROLOGY AND PHYSICAL

OCEANOGRAPHY

from the

NAVAL POSTGRADUATE SCHOOL

September 2017

Approved by: Timour Radko

Thesis Advisor

Justin M. Brown
Second Reader

Peter Chu
Chair, Department of Oceanography

iii
THIS PAGE INTENTIONALLY LEFT BLANK
 ABSTRACT

Baroclinic instability in the ocean is a primary cause of mesoscale eddies, which

are pockets of water in the scale of 100km that have different density, thermal, and
rotational characteristics than their surroundings. First observed in the early 1900s, eddies
are thought to be a predominant reason for the heat flux between the equator and the
poles in both the ocean and the atmosphere. In attempt to understand this process better,
this study uses a series of numerical simulations performed on high performance
computing systems. The calculations are based on the Massachusetts Institute of
Technology general circulation model, which is used to compare lateral heat transport
between different simulations. The specific objectives of this project include

i) Comparison the direct and remote interactions of shear with topographic slope.
The direct scenario is one in which the shear extends throughout the entire ocean depth
and is therefore in direct contact with the sea floor, whereas in the remote scenario there
is a spatial separation between the shear in the upper half of the basin and the bottom
topography,

ii) Analysis of the system response to changes in the zonal and meridional
seafloor slope, and

iii) Investigation of the effect of orientation changes in the mean large-scale

current on cross-flow fluxes.

The lateral heat transport and diffusivity of these simulations are then compared
to our analytic model, known as Growth Rate Balance, which is based on the balance
between growth rate (primary) instabilities deduced from linear theory and numerically
generated secondary instabilities.

v
THIS PAGE INTENTIONALLY LEFT BLANK
 TABLE OF CONTENTS

I. INTRODUCTION..................................................................................................1
A. BACKGROUND ........................................................................................1
B. TOPOGRAPHIC EFFECTS AND SHEAR ............................................3
C. GROWTH RATE BALANCE MODEL ..................................................5
D. ORGANIZATION .....................................................................................6

II. MODEL DESCRIPTION......................................................................................7

A. FORMULATION.......................................................................................7
1. Hydrostatic Equations ...................................................................7
2. Shear................................................................................................9
3. Zonal versus Non-Zonal Currents: Orientation .......................10
B. INITIAL AND BOUNDARY CONDITIONS .......................................10
1. Size and Parameters ....................................................................10
2. Direct and Remote Case ..............................................................11
3. Zonal Case ....................................................................................13
4. Orientation Cases .........................................................................13
C. DIAGNOSTICS OF EDDY HEAT FLUX AND DIFFUSIVITY ........14

III. RESULTS .............................................................................................................17

A. DIRECT VERSUS REMOTE SHEAR OVER MERIDIONAL
SLOPE ......................................................................................................17
B. ZONAL SLOPE .......................................................................................22
C. ORIENTATION.......................................................................................22

IV. COMPARISON WITH GRB ..............................................................................27

A. THEORY ..................................................................................................27
B. RESULTS .................................................................................................29

V. DISCUSSION AND CONCLUSIONS ...............................................................31

VI. RECOMMENDATIONS .....................................................................................33

APPENDIX A. COMPLEMENTARY FIGURES ........................................................35

APPENDIX B. DATA TABLE .......................................................................................37

vii
LIST OF REFERENCES ................................................................................................39

INITIAL DISTRIBUTION LIST ...................................................................................43

 LIST OF FIGURES

Figure 1. Perpetual Ocean Image. Source: NASA Goddard Space Flight

Center (2011). ..............................................................................................2

Figure 2. Diagram of the Direct (left) and Remote (right) Shear Scenarios .............12

Figure 3. Diagram of Zonal Shear Scenarios ............................................................13

Figure 4. Diagram of Orientation Case Scenarios .....................................................14

Figure 5. Direct Versus Remote Case Temperature Flux Comparison at Full

North and Full South Gradients .................................................................17

Figure 6. Direct Case Mean Meridional Temperature Flux and Diffusivity

Versus Slope ..............................................................................................18

Figure 7. Eddy Thermal Flux in Upper Half of Remote 1100-meter Case ...............19

Figure 8. Thermal Diffusivity in Upper Half of Remote 1100-meter Case ..............19

Figure 9. Eddy Thermal Flux in Upper Half of Remote 3300-meter Case ...............21

Figure 10. Thermal Diffusivity with Sopt and Sint, Upper Half of Remote 3300-
meter Case ..................................................................................................21

Figure 11. Zonal Slope Flux Comparison ...................................................................22

Figure 12. Orientation Scenario CFF ..........................................................................23

Figure 13. Orientation Scenario CF-DIFF ..................................................................24

Figure 14. Angle of Mean Flow (right) and Magnitude of Mean Flow (left) .............25

Figure 15. Schematic of GRB Model. Source: Radko et al. 2014. .............................27

Figure 16. Three Timesteps of Model Potential Vorticity ..........................................28

Figure 17. 3-D MITgcm 3300 m Diffusivities compared to GRB ..............................29

Figure 18. 3-D Captures of 1100-meter Direct and Remote Model Runs ..................35

Figure 19. Orientation CFF with Orientation Diagram for Visual Reference.............36

Figure 20. GRB with 3-D 3300-meter and Quasi-Geostrophic Comparison ..............36

ix
THIS PAGE INTENTIONALLY LEFT BLANK
 LIST OF ACRONYMS AND ABBREVIATIONS

ACC Antarctic Circumpolar Current

β Coriolis Beta-effect
BI Baroclinic Instability
CF-DIFF Cross-Flow Diffusivity
CFF Cross-Flow Thermal Flux
KT Thermal Diffusivity
MITgcm Massachusetts Institute of Technology general circulation model
NS Navier-Stokes equations
Sopt Calculated Optimal Slope
Sint Interpolated Optimal Slope
Qf Thermal Heat Flux
QG Quasi-Geostrophic
U Along-flow velocity, representative of orientation
V Cross-flow velocity, respective of orientation

xi
THIS PAGE INTENTIONALLY LEFT BLANK
 ACKNOWLEDGMENTS

First, I want to acknowledge my Naval Postgraduate School cohort; Jeff Scooler,

“Submarine” Andy West, Dave Lorfeld, Alonzo Ingram, Chris Ayala, A-Aron Morone,
and Gary Vines. Additionally, I want to recognize all my fellow METOC and USW
students, including “Timour’s Angels” from the Physical Oceanography Lab. I want to
thank CDR Paula Travis for her words of wisdom, encouragement and support. She gave
me the room to do what I needed to do to make this possible, especially considering my
circumstances during the final portion of my time at NPS.

I also want to thank Cassie Sisti for taking the time to get me started
and answering my daily plethora of questions. I am grateful for Timour Radko for
teaching me numerical modeling through “Bros’ menya k volkam.” I could not ask
for a better mentor. Not only did he share his knowledge of physical oceanography and
mathematics, but his patience with teaching allowed me to achieve and excel.
Furthermore, I am thankful to Justin Brown for helping me understand Linux and
coding, and for guiding me through thermal diffusivity in the Alpha Persei Cluster.

I would also like to thank SPAWAR Systems Pacific for accepting me into
its fellowship program and giving me the opportunity to make much-needed processor
and visualization upgrades to the Physical Oceanography Lab at NPS, which was not
only a great benefit to this research, but will benefit students for years to come.

On a more personal level, there are my boys, Kyle, Logan, and Liam, who give
meaning to my life every day, and my wife, Brigette, who made sure they were taken
care of while I spent long hours behind a computer screen. Last, and most important,
I thank my mother, Theresa, who, is currently in the final stage of her battle against
stage-IV lung cancer. She is, and has always been, an inspiration to my work, my
health and my life. I am here not only because she gave birth to me, but because
her passion for education and self-improvement drove the same values within
me. My mother received her GED in New York state a year before I
graduated high school and continued progressing forward until she earned a
xiii
bachelor’s degree in psychology from Alfred State University. Her hard work
ethic and compassion for people is unmatched. Ma, this is for you.
 I. INTRODUCTION

Modern science has both the distinction of advancing at a very fast rate due to
discoveries in technology and mathematics, and the ignominy of having advanced so
quickly that society has become unaware of the many natural phenomena that still remain
unexplained. This thesis explores one such topic whose effect on mixing has eluded
scientists from multiple disciplines and continues to be an area of debate and research.
The baroclinic instability (BI), which evolves into turbulent flows that transport heat and
momentum, is a predominant source of mesoscale variability in both the ocean and in the
atmosphere. The nonlinear interaction of flow components leading to growing
disturbances creates large-scale flows that can be easily identified in nature through in-
situ and remote observations (Kamenkovich et al. 1986). These disturbances break off
into energetic, relatively long-lasting circular features known as eddies.

A. BACKGROUND

Eddies are everywhere! In the atmosphere, they can be seen in various forms,
such as that of mid-latitude extratropical cyclones. Mesoscale eddies are features that
dominate the ocean basins, especially around strong oceanic current systems like the Gulf
Stream or the Antarctic Circumpolar Current (ACC), and are scientifically important for
many reasons: For one, eddies are believed to be the source of more energy than any
other ocean motion. Accordingly, they also transport tracers, chemicals, and nutrients,
making them a cross-disciplinary topic of study (Robinson 1983). One of the greatest
aspects of eddies is the transfer of heat from the tropics to the poles; thus eddy heat
transport is believed to be a major contributor to the balance of the global heat budget in
both the atmosphere and the ocean. Without large-scale eddy transport in the atmosphere,
for example, the mean meridional circulation would be too balanced to explain the actual
bulk energy transport (Green 1970). Figure 1 illustrates a satellite-derived image of
eddies in the Gulf of Mexico and in the Gulf Stream.

Ocean eddies have spatial scales ranging anywhere from thousands to hundreds of
thousands of meters, and time periods varying from weeks to years (Robinson 1983).

1
Because of this, parameterizations of eddies in climate models remain a major source of
uncertainty which can lead to considerable inaccuracies in model predictions (Gent and
McWilliams 1990). This feature is rather unfortunate, considering the importance of
eddies to global heat budgets and in enhancing model accuracy in the long-range
forecasts. For example, our inability to properly represent eddies serves as a potential
cause of great uncertainty in ice melting in Arctic Ocean models (Maslowski and
Lipscomb 2003). Unphysical eddy closure models can parameterize heat flux to the south
for places like the Canadian Basin where northward (positive) heat flux is observed in
nature (Steiner et al. 2004). The inadequate parameterization of eddies diminishes
accuracy and customer confidence.

Figure 1. Perpetual Ocean Image. Source: NASA Goddard Space

Flight Center (2011).

Traditionally, studies of ocean eddies have been more theoretical than

observational. Pre-satellite eddy observations were few and far between. Eddies were
first documented by Ben Franklin’s grandnephew, Jonathan Williams, who noted a
“warm core ring” while monitoring sea surface temperatures and velocities in the Gulf

2
Stream (Robinson 1983, Williams 1793). Observations for these “rings” was noted again
in the 1930s by P.E. Church who discovered warm core eddies while examining ship
thermograph records from Gulf Stream transits (Church 1932, 1937). Iselin (1936)
followed this up with deep-water temperature and salinity samples taken from The
Atlantis, from which he concluded these rings to be a permanent feature. By the 1970s,
teams of scientists from around the world teamed up on surveys such as MODE1 and
POLYMODE, which was the largest joint U.S.–U.S.S.R. experiment of its time
(Robinson 1983). Now, with the use of satellites, eddies can be observed and tracked
around the globe on a daily basis. Additionally, high performance computer systems and
GCMs (General Circulation Models) can realistically model the ocean in three
dimensions with increasing speed and resolution. Although we may not know as much
about mesoscale variability as we would like, technology has given us the advanced tools
necessary that we may one day be able to solve many of Earth’s mysteries (Knauss
2000).

B. TOPOGRAPHIC EFFECTS AND SHEAR

Of the many topics that require further investigation with regard to ocean eddy
dynamics, the effects of topography on synoptic eddies is rapidly gaining attention of the
oceanographic community. There are many ways that bottom topography can impact the
physical ocean processes such as sea floor roughness, smooth versus rocky geologic
features, and sea floor gradient. Observations indicate that roughness on sloped
topography, even in the deep ocean abyss, will enhance vertical mixing and turbulent
diffusivity (Polzin et al. 1997, Dewar 1998). This creates an issue with models that have a
bathymetric resolution too course to represent roughness, which may cause inaccurate
estimates of diffusivity and other flow properties, especially near the sea floor. This
inaccuracy can be especially prevalent when attempting to model processes like internal
waves or eddies (Goff and Arbic 2010). Thus, there is a need for good synthetic
parameterization of this roughness, particularly with lower-resolution models.

1 Mid-Ocean Dynamics Experiment.

3
 Bottom relief, or slope, is known as a generating mechanism for synoptic
variability. There are various ways slope can affect large- and meso-scale patterns. For
example, topographic Rossby waves are a low-frequency phenomenon that can disappear
when the wave is no longer in contact with the sea floor. Additionally, flow over bottom
irregularities, such as sea mounts and guyots, can create instabilities which, due to the
conservation of potential vorticity, can create cold core eddies above them with warm
core eddies downstream (Kamenkovich et al., 1986). Topographic stress generated on
these eddies can create secondary circulations which may increase upwelling and sea
surface height (Holloway 1987). Sutyrin and Grimshaw (2005, 2010) considered the
effects of sloping topography by applying surface-intensified circular vortices on a β-
plane in a two-layer model with reduced gravity approximation. In their experiment they
specifically looked at frictional effects on deep-ocean flows evaluating the topographic
orientation effects using the β-drift with either along-slope or cross-slope components
(Sutyrin and Grimshaw 2005). Our knowledge of the way various slopes affects eddy
trajectory, for example, and how it changes the structure of the eddy as it forms, is still
developing (Robinson 1983).

A third type of effect caused by bottom relief is the impact it has on the stability
of the current itself. As we will discuss in the next chapter, zonal currents can be
maintained in the ocean without external forces. In a stratified system with a flat (uniform
depth) seafloor and zonal current, shear that remains vertically uniform can be
completely stable with the only varying parameter being the Coriolis force: a fictitious
force that changes with latitude as measured by the aptly-named β-effect, (1),
(Kamenkovich et al. 1986).

f
 (1)
y

Keeping all the same background conditions, the addition of a bottom slope can
create instability (Charney and Flierl 1981). Green (1970) also noted the importance of
shear in meridional heat flux and that constant shear in the troposphere would create an
entropy flux that was independent of height. In this, he suggested that when there is
inconsistent shear in the vertical, the heat flux can change sign. If this holds true in the
4
ocean, we could expect that negative values of heat transfer exist below layers of current-
induced shear. The remote case in our experiment is an example of inconsistent shear.
This invites the question of how the slope of the topography in a motionless lower layer
will impact the heat flux in the upper remote layer, and if there will be a conservation
effect that will shift the sign of heat flux at particular depths, particularly below the level
of no shear.

In this study, we attempt to isolate the effect of slope on eddy dynamics by using
high-resolution numerical simulations in a hydrostatically balanced, eddy-resolving
model. We analyze the equilibrium dynamics of BI in an idealized environment and
compare meridional diffusivity of a flat bottom case with the diffusivity over various
slopes. Meridional slope should have a de-stabilizing effect on shear (Chen and
Kamenkovich 2013) which should affect the eddy-transfer process. It will be shown that
where slope exists, there is an increase in bottom interaction, and a decrease in stability,
both of which enhance eddy formation. The question we look to answer is this: How will
increased positive and negative (north and south) slopes affect large-scale eddy-induced
transfer?

C. GROWTH RATE BALANCE MODEL

One of the great difficulties with our understanding of eddy dynamics is the non-
linearity of BI. In recent decades, many attempts have been made to explain mesoscale
variability in terms of analytical models. Because mesoscale variability is non-linear, full-
fledged models would be computationally prohibitive and mathematically difficult
(McWilliams 2011). Several researchers, such as Thompson (2010) and Visbeck et al.
(1997), have come up with different unique ways to simplify this reasoning using analytical
models. One of these theoretical models, which was developed in 2014, is known as the
growth rate balance model (GRB). This model is based on the assumed balance between
primary (λ1) and secondary (λ2) linear modes of instability, which are linked with an
empirical constant (C). This paper will compare our 3-D model results with those of the GRB
model (Radko et al. 2014).

5
D. ORGANIZATION

The following chapters of this thesis will contain the model description as it is
used in MITgcm. This includes the stability theory that explains how the model creates
eddies. We will then explain the initial conditions and boundary conditions, rationale for
the conditions that were chosen, and differences between the types of model runs that
were conducted in this experiment. Afterward, the diagnostic tools will be discussed
before moving on to Chapter III, numeric results, which will include data, charts and
snapshots that best represent the data and new discoveries. Following that will be our
comparison with GRB theory results in Chapter IV, discussion and conclusions in
Chapter V then recommendations for future research in this topic in Chapter VI.

6
 II. MODEL DESCRIPTION

A. FORMULATION

An important step in modeling ocean eddies is to understand how BI is created

within the model. This chapter gives us a description of the model dynamics, explains the
assumptions, and outlines rationale behind input parameters.

1. Hydrostatic Equations

In this project, we will look at BI in terms of a hydrostatic, continuously stratified

model with a uniform horizontal velocity decreasing with depth. Of the three options,
hydrostatic, quasi-hydrostatic and non-hydrostatic, the former was chosen for the numeric
representation. Although the non-hydrostatic model is the preferred choice because it is
the most general, it is also the most complex and processor intensive for a numeric
model. The hydrostatic model can be as effective as either of its counterparts in large-
scale scenarios (Marshall et al. 1997). By scaling the dimensions of the model, a
determination can be made that hydrostatic approximation is valid. Our length scales are
2000 km, thus L = 2106 m. The height of our basin is 3.3 km, making H ≈ 3103 m, our
average meridional velocity is 0.075 ms-1 in the remote case and 0.15 ms-1 in the direct
which we can approximate U ≈ 0.1 ms-1. Using a standard sea-water potential density of
1103 kgm-3, which means that—with a temperature difference from surface to bottom
of five degrees—we can assume a change in density over depth (based on temperature) of
1 kgm-3. Therefore, we can calculate our buoyancy frequency using Equation (2) below
to be N ≈ 1.7x10-3 s-1.

g 
N2   (2)
 z

Here, g is the standard gravitational constant of 9.8 ms-2 and ρ is water density, which
will be described in Section B. In the direct case, the velocity structure of the basic state
is as follows with v representing meridional velocity and u representing zonal velocity
with the subscripts indicating their vertical location in the water column.

7
 v  0, u surface  umax , u bottom  0,
u (u surface  u bottom ) (3)
 .
z z

In the remote case, however, these expressions change to

z
u surface  umax , ur (h  )  0,
2
u u surface  u r (4)
 .
z 1
z
2

In both cases, remote and direct, the velocity fields are related to the temperature patterns
through the thermal wind balance:

T f u
 . (5)
y g z

Initial boundary conditions were adopted based on Radko et al. (2014) with
changes made to reflect a deeper open-ocean seafloor that would better represent
conditions in the mid-ocean basins. This is particularly pertinent to the Southern Ocean,
which reaches depths over 5 km. Because the model setup had already been tested, this
became the basis for this experiment. This experiment includes β of 110-11 and standard
Coriolis parameter of 110-4 m-1s-1. Density stratification in our model ocean basin was
created by assuming uniform salinity of 35 psu, thereby making density solely a function
of temperature. Velocity is induced via Equations (3) and (4). In order to create both a
horizontal and vertical stratification, the equation for thermal wind (5) is used in
conjunction with shear gradient to generate a temperature gradient in the model
(Kamenkovich et al. 2009). This temperature gradient corresponds to an approximate
10°C overall variation in the vertical and 14°C in the horizontal. A randomly generated
temperature variation of magnitude 0.10°C was induced throughout the model, which
disrupts the initial stratification enough to facilitate instability once it is acted on by
shear.

8
 2. Shear

One of the main drivers of BI is the shear. Vertical shear provides a key
component into destabilizing the stratification and thereby inducing instability. For this
study we chose a magnitude of shear that is observationally reasonable capable of
generating BI. By testing standard surface current velocities, usurface , with a gradient of

u
decreasing velocity with depth, - , a standard gradient was determined that is used in
z
all our shear and depth scenarios. In this research, two shear configurations and two
depths are compared. The first configuration is the direct case where the shear was
induced throughout the water column, reducing linearly in intensity with depth until
ubottom  0 . The second configuration is the remote case, which is where the shear was
applied to the upper half of the basin, but the lower half was given motionless initial
conditions. No meridional flow is initially induced and therefore any flow in the y-
direction is caused by dynamic effects. In doing this, a comparison can be made between
both the direct and remote shear scenarios and between the different meridional slopes.
By measuring the eddy heat flux, we can determine if there is a remote effect on the
topography on the background current, and the role that bottom slope plays on ocean heat
transport.
The remote case scenario was conducted at two different depths: 1100 m and
3300 m. In order to compare the shear effects between scenarios at different depths, we
required that the shear gradient remain the same instead of the mean or maximum
velocities of zonal flow. Therefore, for the 3300m case, the surface velocity measured
0.15 ms-1 and reached zero velocity at a depth (z) of 1550 m whereas the 1100 m case
measured 0.05 ms-1 reaches zero velocity at 550 m. This gives give us a shear gradient
that is equivalent for both depths, despite a greater area covered by shear in the 3300-m
case. Thus, our comparisons are based strictly on a comparable shear gradient rather than
separate shear gradients with the same surface velocity as illustrated by

u u
(h0  1100m)  0.9091e 5 s 1  (h0  3300m) . (6)
z z

9
 3. Zonal versus Non-Zonal Currents: Orientation

Because of the Coriolis force, a current in the zonal direction is the only flow that
can be maintained without the effects of outside forces (Kamenkovich et al. 1986). In this
experiment, the flow direction is turned at 30° intervals over a flat bottom basin, and the
impact of the rotation on the cross-current thermal flux (CFF) is analyzed. The magnitude

of horizontal velocity is U  U 2  V 2 , where U represents along-flow current and V
represents the across-flow current. The term “flow” in this context describes the initial
background current in the direction of orientation. Lower case x and y represent the
longitudinal and latitudinal axes, which is standard in our zonal flow simulations, but in
the orientation cases we use X and Y to indicate the along-flow and cross-flow axes
respectively. Because the dimensions of the basin remain the same, and the basin-relative
flow remains the same, the value of any diagnostic in Y, although no longer meridional, is
our desired quantity. The thermal gradient also changes with each rotation, and therefore,
it is the β-effect that will provide the physical difference between orientation runs.

It should be emphasized that, in the orientation cases the time mean flow is not
strictly oriented in the X-direction. Therefore, the mean temperature flux in Y includes a
cross-flow component associated with the advection of heat by time-mean velocity,
which must be accounted for in order to determine CFF. This is done in the post-
processing analysis as follows:

T 'V '  TV  TV . (7)

B. INITIAL AND BOUNDARY CONDITIONS

1. Size and Parameters

In determining the size of the basin, a few principles of eddy dynamics were
implemented. Since mesoscale eddies are in the scale of tens to hundreds of kilometers
wide, the horizontal dimensions of the box had to be large enough to resolve 100 km
features while the resolution also had to be fine enough to encompass the smaller eddies.
In particular, baroclinic radius of deformation ( Rd ) must be resolved in order to ensure
that baroclinic instability is properly represented. Inherently, the Brunt-Väisälä buoyancy
10
frequency (N) determines radius of deformation. To simplify the equation for determining
density stratification, salinity was kept constant at 35 psu, which means that density(ρ)
becomes a function of temperature: ρ = ρ(T). Because of this, there is a direct relationship
between our temperature gradient and pressure gradient, which can be used to estimate
buoyancy. The expressions below use the equation of state to replace density with
temperature and solve for Rd . Note that α is the thermal expansion coefficient.

Nh
Rd 
f (8)
  0  T
S  0,
0
 
  T  T0  
z z
(9)

 g  T
N   g (10)
0 z z

This equation gives a calculated Rd of approximately 2.4104 m, or 24 km for the

1100-meter runs and 4.20104 m (42 km) for the 3300-meter case. This is consistent with
the size of mesoscale eddies. This estimate led to the optimal choice for basin size for this
experiment of 2000 km by 2000 km, as the basin had to be wide enough to a contain a
large number of eddies. Additionally, the horizontal resolution is also a factor since too
course a resolution will not be able to simulate eddies, and too fine a resolution will cause
model run times to increase dramatically. Two-kilometer resolution was chosen as it is
1/10th the scale of concern of Rd for the 1100-meter case.

2. Direct and Remote Case

As mentioned in the introduction, two major flow regimes were introduced into
the original research plan, which have been aptly named the direct and remote cases. The
direct case is perhaps the more straightforward of the two as the shear is exerted
throughout the water column by introducing a surface current that decreases in magnitude
linearly with depth all the way to the seafloor. Thus, the shear is present throughout the
water column and is only inhibited at the sea floor. In the remote case, the shear is
11
present only in the upper half of the water column, and the velocity becomes zero at half
depth. Thus, the lower half of the water column is completely motionless initially. Figure
2 is a visualization of the remote and direct shear scenarios. A 3-D visualization can be
seen in Appendix A (Figure 18).

Figure 2. Diagram of the Direct (left) and Remote (right) Shear Scenarios

This motionless region is not a “lower layer” and should not be confused with a
two-layer model because the stratification in the upper and lower half is continuous. The
lower half is merely a region of no initial zonal current, and therefore, any flow that
occurs in the lower layer is due to baroclinic instability effects induced in the upper layer.
Also, the magnitude of the shear is the same across the direct and remote cases is the
same. That is the linear decrease in velocity is the same. This also means that the velocity
on the surface is twice the value in the direct case. This allows us to maintain the same
shear magnitude despite having larger surface speeds with the direct flow:

udir u
 h0  1100m   0.055ms 1 , dir  h0  1100m   0.026ms 1
z z1/2
(11)
udir u
 h0  3300m   0.15ms 1 , dir  h0  3300m   0.078ms 1.
z z1/2

12
 3. Zonal Case

The previous model compared several cases of various meridional, or north—

south slopes. In an effort to mimic the somewhat of a mid-Atlantic ridge, a series of runs
were performed to determine the effects of topographic slope in a zonal, or east—west,
direction, as depicted by Figure 3. In these scenarios, the depth of the water remained
constant in y, but changed over x with an apex at length x/2; thus, the flow is
perpendicular to the bottom gradient. This also means that potential vorticity of the mean
flow in the direct case will obviously be affected by the height of the water column as
seen by Equation (12). However, the impact of the bottom topography in the remote case
will be less certain. Additionally, because the zonal slope can deflect the initially zonal
flow, the background meridional flow will have to be taken into account when evaluating
the eddy-induced transport. Equations for the potential vorticity and zonal slope are
f 
 const (12)
h
h h h
 0,  Sm ,   Sm . (13)
y x 0 x /2 x x /2 x 

Figure 3. Diagram of Zonal Shear Scenarios

4. Orientation Cases

For the purposes of simulating initial flows for various orientations, we chose to
retain the flow direction relative to the computational domain and re-orient the basin in
order to simplify our analysis and allow us to induce periodic boundary conditions along
the flow. To simplify our post-processing, we retain the flow direction and shape and re-

13
orient the basin. Future experiments may wish to try alternative methods of measuring
and calculating orientation output across the flow for comparison. Rotating the basin
involved converting from Cartesian grid, which we used on our other runs, to a spherical
grid. The orientation was controlled by using Eulerian coordinates to turn the basin
around the z-axis using the scale 1degree latitude = 111 km. This gave us a dx and dy of
~1.8° vice 2 km. Figure 4 depicts the orientation scenarios and their respective angles.

Figure 4. Diagram of Orientation Case Scenarios

C. DIAGNOSTICS OF EDDY HEAT FLUX AND DIFFUSIVITY

In order to evaluate the impact of topographic effects on mesoscale variability, we

must anticipate what variables that our numeric model outputs that we would expect to
see change. This section outlines some of the rationale behind our diagnostic tools based
on the relationships to each other and to how we would expect them to be impacted under
different conditions. Because of our model setup with a current that is strictly zonal
(except for the “orientation” runs), meridional motion is mostly caused by the baroclinic
instability. Therefore, our measurement of any transient displacement in y-direction will
reflect the intensity of mesoscale variability. This is important because in order to

14
determine the effects of mesoscale variability, it should be clearly isolated from any
direct effects of background flows. To accomplish this, we simplify the eddy equations of
motion and make our initial background currents purely zonal flow. Inherent in this
simplification is the lack of any rotational or non-meridional transport and so divergent
and rotational components are not isolated (Radko et al. 2014). Note that for the
orientation case, this simplification is only possible by applying values of x and y as
“along-flow” and “across-flow,” respectively. Additionally, non-eddy effects are
removed by removing mean flow from our v-velocity.

As previously mentioned, mesoscale eddies are an important component in the

transfer of heat from the equatorial region to the poles. This means eddies are also large
contributors in maintaining the heat balance for our planet. Because the stratification in
our idealized basin is maintained through temperature and not salinity, we are able to
truncate our equation of state to essentially replace density gradient with thermal
gradient. This allows us to approximate density with temperature, and therefore
temperature becomes a key diagnostic variable. MITgcm allows you to output both
temperature and heat flux in both x and y directions (Menemenlis and Fukumori, 2005).
Because we are taking our values over time and we can average gridpoint values over the
entire model area and over all the equilibrium time periods in order to get one value for
the average temperature flux (Qf) over the period from when the model reaches baroclinic
equilibrium to the last integration output.

Q f  V 'T ' (14)

If the heat flux is required, it can be trivially computed from the temperature flux
as follows:
Qhf   C pQ f , (15)

where C p  4000 J kg 1 K 1 is the specific heat and   1000kg m 3 is the density of

seawater. Thermal diffusivity (KT) can be obtained by dividing the temperature flux by
the thermal gradient, as indicated in Equation (16).

15
 T T jj 1  T jj 1
Ty  
y 2y
(16)
Q
KT   f
Ty

Finding the overall thermal gradient in Equation (17) from the model is
accomplished by averaging the temperature gradients for each gridpoint in space and
time. The advantage behind diffusivity is that it can easily be compared with alternative
estimates in literature and analytical models, like GRB. To maintain a statistically steady
state, a relaxation condition is enforced on the surface and bottom in order to prevent
unrealistic build-up of warm and cold temperatures. Additionally, all diagnostic variables
are calculated only after removing the boundaries as far out as 60 km. This was necessary
to be able to focus on what was occurring within the deep water basin and to avoid
contamination from boundary layer dynamics. Think of this as removing the wax layer
from around the cheese in order to prevent a nasty taste in your mouth.

16
 III. RESULTS

A. DIRECT VERSUS REMOTE SHEAR OVER MERIDIONAL SLOPE

We compared the initial meridional slope runs conducted in the 1100-meter basin
to determine any topographic effects on eddy heat flux. We also investigated the remote
scenario, where the flow and shear were not in direct contact with the seafloor, for the
effects of varying the slope. There is a remote effect on eddy heat flux even without
direct interaction. However, the mean total heat flux is over three times greater in the
direct case than in the remote case for an equivalent slope (s = 2.6210-4) after the model
reaches an equilibrium. In both cases, a south slope of the same absolute value as the
north slope yields a greater meridional heat flux as seen in Figure 5.

Figure 5. Direct Versus Remote Case Temperature Flux Comparison at

Full North and Full South Gradients

In our direct case, variations in the meridional bottom slope are expected to affect
variation in total mean heat transport. Chen and Kamenkovich (2013) observed in a two-
layer model that meridional slope changes the PV gradient causing a stabilizing effect.
The results of the 1100-meter model indicate that in the direct shear case, the largest
mean meridional heat flux and diffusivity occur for a flat seafloor (zero slope), which
17
would indicate that a bottom gradient of any kind reduces horizontal diffusivity (Figure
6). However, calculating the meridional heat flux in the upper half of the basin, the
remote case (Figure 7) yields the greatest flux for a slightly negative bottom slope. The
same can be said for diffusivity (Figure 8) which shares the same curvature.

Figure 6. Direct Case Mean Meridional Temperature Flux and Diffusivity

Versus Slope

One of the explanations for the weaker total heat flux in the remote case could be
that the remote case is more strongly affected by conservation of potential vorticity.
Perhaps this is due to the effect of the bottom drag, which may have a stronger impact on
direct flows because of the solid boundary directly below the shear region; the latter
feature is absent in the remote case. In order to explain the pronounced asymmetry of
temperature flux in the remote case (Figure 7), we now attempt to develop a simple
analytical theory predicting the slope that results in maximal heat transport. This slope,
which will be referred to as the optimal slope hereafter, is computed based on the analysis
of the equation for conservation of potential vorticity (Qpv), given by
f   f0   y
Q pv    const . (17)
h h0  sy

18
 Figure 7. Eddy Thermal Flux in Upper Half of Remote 1100-meter Case2

Figure 8. Thermal Diffusivity in Upper Half of Remote 1100-meter Case

Conservation of Qpv implies that the meridional displacement of water columns is

adversely affected by both the β-effect and by the slope individually. However, when
both effects operate concurrently, they can counteract each other and thus have minimal

2 MITgcm thermal flux output was plotted along with the calculated thermal flux for validation.

19
impact on the meridional displacements of water columns. By assuming that this
cancellation occurs when
f 0   y  h0  sy (18)
we can roughly identify an optimal slope (Sopt) based on β, f0 and the slope s, where
β = 110-11 m-1s-1, f0 = 110-4 s-1, and h0 is the maximum depth of the water column. This
assumption can be rationalized as follows. The conservation of Qpv states that if the
height of the water column changes, rotational vorticity (ζ) resists change, and therefore,
Qpv is conserved through a north or south shift, which compensates for changes in f. On
the other hand, if f changes, the water column will want to compensate by stretching or
shrinking. Equation (18) then immediately implies
 h0
Sopt  
f0 . (19)

This formula for optimal slope, Equation (19) shows a calculated Sopt of
approximately -1.110-4 for h0=1100 m, which is comparable to our model results. Using
a splines interpolation, the value of the maximum slope based on the compared model
runs is -0.85810-4.

Using this same equation, the 3300-meter remote case can also be analyzed.
Because the theoretical Sopt is a function of the base depth, it is easily estimated at three
times the 1100-meter case, (Sopt(h0=3300 m) = 3.310-4). However, our results are a bit
more complicated than our simple theory. This meant that additional southern, or
negative slope values had to be run for this case in order to find the slope great enough
for our optimal case. However, even with these extreme slope simulations, the optimal
slope for heat flux has not yet been attained. Figure 9 illustrates that the general shape of
the slope to Qf difference is maintained, the Sopt does not appear to be within the range of
slopes modeled. In fact, it appears that the increased heat flux levels off below -410-4
rather than tapering off sharply like they did in the 1100-meter case.

It initially appears that the optimal slope theory tends to output too gentle a slope
for the h=3300 m case. However, taking a look at the interpolated diffusivity plot in
Figure 10, the diffusivity does begin to drop off at a Sint of -4.5410-4, which is close to
20
our calculated Sopt. A clear difference between the diffusivities and heat flux can be seen
at larger slopes, whereas in the slopes calculated in the 1100-meter case, the KT and Qf
curves were nearly identical.

Figure 9. Eddy Thermal Flux in Upper Half of Remote 3300-meter Case

Figure 10. Thermal Diffusivity with Sopt and Sint, Upper Half of Remote
3300-meter Case
21
B. ZONAL SLOPE

The zonal slope case was analyzed in the same way as the meridional slope case
in order to compare the direct and remote scenarios using the two separate slopes. Results
show that in both the direct and remote cases, and in both slopes, the net heat flux was
equivalent to that of a flat bottom case. Singling out the Qf observed on either the east or
west side of the apex, we discover that they are noticeably large and erratic, which may
be corrected if the turbulent fluxes were subtracted from the total flux. However, it is
apparent to see that the southward mean flow of the western, or “uphill,” side of the basin
is fully countered by the northward flow on the eastern “downhill” side. In Figure 11, the
black and gold lines represent both the flat bottom case equivalent and the net Qf for the
full water column for which the values are nearly identical. This was true both for the
direct and remote cases at both slope angles where the calculated difference between the
net zonal and flat bottom fluxes was negligible.

Figure 11. Zonal Slope Flux Comparison

C. ORIENTATION

As a reminder, diagnostics for the orientation runs were calculated in the frame of
reference associated with the initially imposed basic flow. In particular, we look at the
22
cross flow heat flux, diffusivity, and the velocity and magnitude of the mean current.
Figures 12 and 13 illustrate the different CFF and CF-DIFF patterns for each value of ϴ.
As ϴ increases, the CFF is dampened until ϴ=90 at which point it rises again. At
ϴ=180 it reaches a maximum CFF, three times that of the zonal flow at ϴ=0 and in the
polar opposite direction: southward flux instead of northward. Similar to larger slopes in
the 3300-meter zonal flow model, the orientation models’ CF-DIFF and CFF plots do not
have the same shape. CF-DIFF, for example, reaches its minimum at ϴ=0 and its
maximum at ϴ=120. This is 60 out of phase with the CFF curves of both values. The
results suggest that diffusivity is offset from fluxes when oriented flow is involved.
Additionally, the diffusivities are approximately twice the typical values observed in the
ocean. Appendix A (Figure 19) depicts Figure 12 with orientation angles for better
reference.

Figure 12. Orientation Scenario CFF

23
 Figure 13. Orientation Scenario CF-DIFF

Mean velocities were also computed in order to identify potential changes in the
properties of background flow. In Figure 14 (right), the angle of the mean flow is plotted
in relative to the equator, or x-axis (magenta) and relative to the orientation of the initial
flow (cyan). Not surprisingly, there is a clockwise veer for each of the non-zonal flows,
which are all comparatively similar, roughly in the range of 4–12 degrees. Both of the
zonal flows (0 and 180) maintain their mean flow angle with the direction of
background flow. However, the magnitude differences between the mean flows at
different values of ϴ is noteworthy. Note in Figure 14 (left) that the mean flow velocity
for ϴ=0 is the initial value of the mean current and is therefore approximated. Initially,
the sharp decrease in mean flow velocity begins like the CFF, dropping significantly and
rising again all the way up to ϴ=150˚, it then suddenly drops again around ϴ=180,
unlike the CFF.

24
Figure 14. Angle of Mean Flow (right) and Magnitude of Mean Flow (left)

25
THIS PAGE INTENTIONALLY LEFT BLANK

26
 IV. COMPARISON WITH GRB

This chapter focuses on comparing our 3-D model data to GRB. First by briefly
explaining the theory of GRB, then by comparing our 3-D MITgcm results to the GRB
results for our 3300-meter remote case.

A. THEORY

Oceanographers have often attempted to explain the non-linear effects of BI using

analytical models. GRB attempts to explain BI by linking the growth rates of the primary (λ1)
and of the secondary (λ2) modes of instability. Such analysis can be insightful because, as the
primary mode increases in intensity, the secondary mode, which is a function of the primary,
intensifies and eventually dampens the primary mode. In this theory, the two modes
“compete” and, in doing so, an equilibrium state is created where the instabilities finally
match each other. This “balance” can be calibrated using an empirical constant, (C).

Figure 15. Schematic of GRB Model. Source: Radko et al. 2014.

27
 One of the features of this theoretical model is that the primary modes are a
function of vertical shear, bottom drag, stratification, and β-effect, whereas the secondary
mode is a function of all of those plus the primary mode amplitude. The GRB equation
can be summed up by the following:

2  C1 (20)

In theory, the primary mode would grow continuously if not for the secondary
mode hindering its development. Likewise, if the secondary instabilities (λ2) were to
dominate, the primary instabilities would be overly-dampened which would ultimately
result in no eddy formation. This means the two modes must reach a balance at some
point, at which the output diagnostics can be time averaged to see the mean results.
Examples of this instability and its progression can be seen in a horizontal section of the
numeric models as shown in Figure 16. The Growth Rate Balance theory is discussed in
greater detail by Radko et al. (2014). For this paper, numerical results from 3-D MITgcm
model runs will be used to compare to a 2-D Growth Rate Balance analysis. Figure 16
includes three time steps of potential vorticity from a numeric model that illustrates the
progression of initial growth rate (left), the development of secondary instabilities
generating wave-like patterns (center), and the resulting transient irregularities (right).

Figure 16. Three Timesteps of Model Potential Vorticity. Adapted from

Radko et al. 2014.

28
B. RESULTS

In determining the best fit for the GRB model, Radko et al. (2014) investigated
several values of the non-dimensional empirical constant, C, that would best fit the
Phillips (1951) and Eady (1949) models of linear instability. A range of 3.5-4 was
determined to be the best fit, which was calibrated using PV flux. The 3-D models to
which they compared to had spatial scales similar to our model, lacked sloping
topography and the shear was in the direct contact with the seafloor. For this thesis, the
3300-meter remote case was plotted against calibrated values of PV diffusivity
determined from GRB models for each slope at values of C from 2 to 5. Interestingly, the
GRB models predict a slightly northern optimal slope. Only the diffusivities calculated
from our flat bottom (zero slope) 3-D MITgcm model compared well to the GRB model,
which was just under the C=3.5 plot. As seen by the 3D versus C values in Figure 17,
within a range of slopes from +2.610-4 to -1.410-4 the various diffusivities did stay
within the boundaries of 2 < C < 5, but for any slope outside of that range, the analytic
model was not ideal.

Figure 17. 3-D MITgcm 3300 m Diffusivities compared to GRB

29
 Ultimately, the gradual bell curve of the GRB runs does not match the curve of
the 3-D MITgcm plots (Figure 17). If we look at a value of C = 3.5, which is consistent
for our flat bottom case, the predicted diffusivities from the GRB theory are too low for
negative slopes, and it appears to be too symmetric along the y-axis. It appears that,
although the GRB model compared well to the Eady model in Radko et al. (2014), where
C = 3 fit for each run, for our model, changing the slope of the sea floor creates
variability that the GRB model does not predict. Comparisons with the quasi-geostrophic
model can be seen in Appendix A (Figure 20).

Several factors may be able to explain why GRB does not do well with
topographic slope. One may be the simulation depth. The 3-D model in Radko et al.
(2014) was 1100-meters deep. Here, we are comparing the 3300-meter model. Second,
the assumptions in the GRB model may not accurately account for topography in the
basic equations. Although the GRB model accounts for topographic variation, it may not
be able to represent more complex geophysical domains. Third, there may be a factor in
developing BI that GRB is not able to account for. Ultimately, however, this is a good
indication of why we need a better understanding of the effect bottom topography has on
dynamics.

30
 V. DISCUSSION AND CONCLUSIONS

The first significant conclusion or this research is that the remote case shear
scenario acts differently than the direct case with regard to the effect on BI. First, the
intensity of heat transport in the direct case is greater given the same shear and slope.
This is to be expected since the shear force covers twice the amount of volume as in the
remote case. What is unexpected is that the meridional slope acts on the direct shear case
differently than on the remote case. Any meridional slope dampens the heat flux in the
direct case, whereas in the remote case there is a non-zero optimal slope for maximum Qf.
It should be noted that this is a large-scale experiment, but similar effects may also occur
in smaller scale scenarios. This could be important in straits where there is a meridional
slope with seasonally changing shear depths: a scenario where the difference in heat
transfer may prove more complicated than typical parameterizations account for (Spall
and Chapman 1998).

The distinct optimal slope of the remote case is perplexing since, based on β-
effect alone, the direct case should also have a non-zero optimal slope. One possibility is
that in the direct case, shear acting directly on the motionless seafloor is more influenced
by bottom drag. Since the shear is directly impacted by the bottom drag, the frictional
forces due to any increased slope overcome the variation caused by the β-effect, therefore
making the flat bottom the most efficient for Qf. On the other hand, the shear in the
remote case is shielded from the bottom by an initially quiescent layer, which means the
effect of bottom friction is weaker. This could prove useful in determining
parameterizations of flux and diffusivity based on the depth of the current

With increased slope, the diffusivity curve and heat flux curves no longer match.
This is illustrated quite clearly in the 3300-meter KT and Qf. where the interpolated
optimal slope for KT is of smaller value than that for Qf. Similarly, the orientation cases
show a maximum angle for Qf that is not the same orientation angle as those with
maximum KT. In cases of extreme slope or orientation, heat flux and diffusivities are no
longer proportional. This result is significant in that it justifies making model
comparisons using multiple diagnostics.
31
 In the orientation cases, the thermal gradient is shifted along with the flow to
where it runs east—west and south—north with the same applied Coriolis force,
effectively changing the realm of what we see on earth. With the Coriolis force being
weaker at the equator, one might expect that the reason for this intense Qf in our ϴ=180˚
model is due to our horizontal thermal gradient. If we look at this in terms of what would
happen if warmer water were to the north vice the equator, and neglected any changes in
atmospheric dynamics that would occur from this, we would have a thermal/density
gradient and Coriolis gradient running parallel. Unlike in nature where the warmest water
temperatures occur where there is the least Coriolis force, orientations above 120˚ have a
stronger Coriolis is in the same region as the warmer water. Perhaps this scenario
generates more eddy motion / mixing in the north that what we observe near the equator.
This may act as a “heat engine” that, along with thermal wind, promotes larger amounts
of CFF. Perhaps in the future research is needed to validate this concept.

For the GRB model, the results illustrate that the assumptions built into this quasi-
geostrophic formulation of the GRB model only capture very approximate qualitative
behavior. The 3300 m 3-D model optimal slope falls in line with comparisons done
between the basic two-layer QG model, as seen in Appendix A (Figure 20). However, the
QG slope and magnitude are still half that of the 3-D scenario. This reaffirms that bottom
slope does affect BI and that two-layer and analytic models may lack the ability to
properly account for meridional-sloped topography.

32
 VI. RECOMMENDATIONS

There is a broad range of possible studies that can be done to continue this
research as many other practical comparisons could be made. For example, changing the
strength of shear, or comparing heat flux and diffusivities at different slopes to see when
they match and when they begin to diverge. Below is a short summary of topics that one
may wish to explore, but there are many other possibilities that would undoubtedly be
directly relevant to oceanography and geophysical fluid dynamics.

1. The effects of mesoscale and sub-mesoscale bottom features on meridional

transport: This study could use the same numeric model setup but create one or many
bottom features replicating underwater mountains or atolls with scales from 10 to 100 km
and heights extending through a significant fraction of water depth. A comparison could
then be made between the total meridional heat and PV flux with and without these
features and according to their size and comparing them to the uniform slope data in this
paper.

2. Examining the effects of different shear patterns to determine what shear

characteristic plays the dominant role in the dynamics of BI. This can be accomplished
by varying the speed of the surface current and thus changing the gradient of the flow
over depth, and by introducing various shears in different water depths.

3. Continued work on the orientation cases to improve our understanding of

physical processes controlling the variability of heat transport with the flow direction.

4. Determining the effect of eddies in remote and direct case on sea surface height
and temperature to determine and correct systematic errors in satellite bathymetry
calculations. This experiment can be initiated by attempting to recreate a height-varying
topography on the model seafloor from the sea surface data of the ocean using the model
output data post-equilibrium.

33
THIS PAGE INTENTIONALLY LEFT BLANK

34
 APPENDIX A. COMPLEMENTARY FIGURES

Figure 18. 3-D Captures of 1100-meter Direct and Remote Model Runs

35
Figure 19. Orientation CFF with Orientation Diagram for Visual Reference

Figure 20. GRB with 3-D 3300-meter and Quasi-Geostrophic Comparison

36
 APPENDIX B. DATA TABLE

Below is the data collected from various model runs. Orient_ represents the
orientation runs, 3D_ represents the 1100 m zonal flow remote case, and 8D_ represents
the 3300 m zonal flow remote case.

JOB Depth slope theta Qf Qf MITgcm KT

orient30N 3300 0 30 6.164E‐03 6.155E‐03 1.740E+03
orient60N 3300 0 60 3.847E‐03 1.197E‐02 2.074E+03
orient90N 3300 0 90 2.794E‐03 1.897E‐02 3.549E+03
orient120N 3300 0 120 3.424E‐03 1.960E‐02 9.046E+03
orient150N 3300 0 150 5.827E‐03 1.897E‐02 1.155E+04
orient180 3300 0 180 9.665E‐03 1.611E‐02 1.140E+04
3D_SN_RL_5 1100 ‐2.600E‐04 0 6.637E‐03 6.715E‐03 1.762E+03
3D_SN_RL_25 1100 ‐1.300E‐04 0 7.486E‐03 7.440E‐03 2.135E+03
3D_SN_RL_125 1100 ‐6.550E‐05 0 8.104E‐03 8.030E‐03 2.282E+03
3D_RL_FB 1100 0.000E+00 0 5.422E‐03 5.516E‐03 1.468E+03
3D_NS_RL_125 1100 6.550E‐05 0 4.231E‐03 4.185E‐03 1.065E+03
3D_NS_RL_25 1100 1.300E‐04 0 3.148E‐03 3.110E‐03 7.610E+02
3D_NS_RL_5 1100 2.600E‐04 0 2.059E‐03 1.998E‐03 4.766E+02
8D_SN_RL100 3300 ‐5.244E‐04 0 1.075E‐02 1.078E‐02 3.507E+03
8D_SN_RL_75 3300 ‐3.933E‐04 0 1.061E‐02 1.056E‐02 3.521E+03
8D_SN_RL_5 3300 ‐2.600E‐04 0 1.023E‐02 1.013E‐02 3.328E+03
8D_SN_RL_25 3300 ‐1.300E‐04 0 9.067E‐03 9.148E‐03 2.791E+03
8_SN_RL_125 3300 ‐6.550E‐05 0 7.438E‐03 7.340E‐03 2.194E+03
8D_RL_FB 3300 0.000E+00 0 6.164E‐03 6.155E‐03 1.740E+03
8D_NS_RL_125 3300 6.550E‐05 0 5.483E‐03 5.462E‐03 1.491E+03
8D_NS_RL_25 3300 1.300E‐04 0 5.347E‐03 5.316E‐03 1.435E+03
8D_NS_RL_5 3300 2.600E‐04 0 4.091E‐03 4.059E‐03 1.050E+03

37
THIS PAGE INTENTIONALLY LEFT BLANK

38
 LIST OF REFERENCES

Charney J.G., and Flierl, G.R., 1981: Oceanic analogues of large-scale atmospheric
motion. Evolution of Phys. Oceanogr. B. A. Warren and C. Wunsch, Eds, The
MIT Press, 504–549.
Chen, C., and Kamenkovich, I., 2013: Effects of topography on baroclinic instability. J.
Phys. Oceanogr., 40, 790–804.
Church, P. E., 1932: Progress in the investigation of surface-temperatures of the Western
North Atlantic. Transactions, Am. Geophys. Union, 13, 244–249.
https://doi.org/10.1029/tr013i001p00244.
Church, P. E. 1937: Temperatures of the Eastern North Atlantic from thermograph
records. Union Geod Geophys Int. Association d’Oceanographie Physique, 4,
3–40.
Dewar, W. K., 1998: Topography and barotropic transport control by bottom friction. J.
of Marine Res., 56, 295–328, https://doi.org/10.1357/002224098321822320.
Eady, E. T., 1949: Long waves and cyclone waves. Tellus, 1, 33–52.
https://doi.org/10.1111/j.2153-3490.1949.tb01265.x.
Gent, P. R., and J. C. McWilliams, 1990: Isopycnal mixing in ocean circulation models.
J. Phys. Oceanogr., 20, 150–155.
Goff, J. A., and Arbic, B. K., 2010: Global prediction of abyssal hill roughness statistics
for use in ocean models from digital maps of paleo-spreading rate, paleo-ridge
orientation, and sediment thickness. Ocean Modelling, 32, 36–43.
https://doi.org/10.1016/j.ocemod.2009.10.001
Green, J. S. A., 1970: Transfer properties of the large-scale eddies and the general
circulation of the atmosphere. Quarterly J. of the Roy. Meteor. Soc., 96, 157–185.
https://doi.org/10.1002/qj.49709640802.
Holloway, G., 1987: Systematic forcing of large-scale geophysical flows by eddy-
topography interaction. Journal of Fluid Mechanics, 184, 463.
https://doi.org/10.1017/s0022112087002970.
Iselin, C. O’D., 1936: A study of the circulation of the western North Atlantic. Papers on
Phys. Oceanography and Meteorology, IV(4), 101 pp.
https://doi.org/10.1575/1912/1087.
Kamenkovich, I., P. Berloff, and J. Pedlosky, 2009: Role of eddy forcing in the dynamics
of multiple zonal jets in a model of the North Atlantic. J. Phys. Oceanogr., 39,
1361–1379.

39
Kamenkovich, V. M., Koshlyakov, M. N., and Monin, A. S., 1986: Eddies in the open
ocean. Synoptic Eddies in the Ocean, 265–376. https://doi.org/10.1007/978-94-
009-4502-9_5
Knauss, J. A., 2000: Introduction to Physical Oceanography. 2nd. Prentice-Hall. 303 pp.
Marshall, J., Hill, C., Perelman, L., and Adcroft, A. 1997: Hydrostatic, quasi-hydrostatic,
and nonhydrostatic ocean modeling. Journal of Geophysical Research. Oceans,
102, 5733–5752. https://doi.org/10.1029/96jc02776.
Maslowski, W., and Lipscomb, W. H., 2003: High resolution simulations of Arctic sea
ice, 1979–1993. Polar Res, 22, 67–74.

McWilliams, J. C., 2011: Fundamentals of Geophysical Fluid Dynamics. Cambridge

Univ. Press., 249 pp.

Menemenlis, D., Fukumori, I., and Lee, T., 2005: Using Green’s functions to calibrate an
ocean general circulation model. Monthly Weather Review, 133, 1224–1240.
https://doi.org/10.1175/mwr2912.1.

NASA Goddard Space Flight Center, 2011: Perpetual Ocean. 29 December 2016,
http://svs.gsfc.nasa.gov/3827.

Phillips, N. A., 1951: A simple three-dimensional model for the study of large-scale
extratropical flow patterns. J. Meteor., 8, 381–394.
Radko, T., De Carvalho, D. P., Flanagan, J., 2014: Nonlinear equilibration of baroclinic
instability: The Growth Rate Balance model. J. Phys. Oceanogr. 44, 1919–1940.
https://doi.org/10.1175/jpo-d-13-0248.1.
Robinson, A. R., 1983: Eddies in Marine Science. Springer-Verlag, 609 pp
Polzin, K. L., Toole, J.M., Ledwell, J.R., and Schmitt, R.W., 1997. Spatial variability of
turbulent mixing in the abyssal ocean. Science, 276, 93–96.
https://doi.org/10.1126/science.276.5309.93.
Spall, M. A. and D. C. Chapman, 1998: On the efficiency of baroclinic eddy heat
transport across narrow fronts. J. Phys. Oceanogr., 28, 2275–2287.
Steiner, N., Holloway, G., Gerdes, R., Häkkinen, S., Holland, D., and Karcher, M., 2004:
Comparing modeled streamfunction, heat and freshwater content in the Arctic
Ocean. Ocean Modelling, 6, 265–284. https://doi.org/10.1016/s1463-
5003(03)00013-1.
Sutyrin, G. G., and Grimshaw, R., 2005: Frictional effects on the deep-flow feedback on
the -drift of a baroclinic vortex over sloping topography. Deep Sea Research Part
I: Oceanographic Res. Papers, 52, 2156–2167.
https://doi.org/10.1016/j.dsr.2005.06.017.

40
Sutyrin, G. G., and Grimshaw, R., 2010: The long-time interaction of an eddy with shelf
topography. Ocean Modelling, 32, 25–35.
https://doi.org/10.1016/j.ocemod.2009.08.001.
Thompson, A. F., 2010: Jet formation and evolution in baroclinic turbulence with simple
topography. J. Phys. Oceanogr., 40, 257–278.
Visbeck, M., J. Marshall, and T. Haine, 1997: Specification of eddy transfer coefficients
in coarse-resolution ocean circulation models. J. Phys. Oceanogr., 27, 381–402.
Williams, J. 1793: Memoir of Jonathan Williams on the use of the thermometer in
discovering banks, soundings, etc., Transactions of the American Philosophical
Society, 3, 82–100.

41
THIS PAGE INTENTIONALLY LEFT BLANK

42
 INITIAL DISTRIBUTION LIST

1. Defense Technical Information Center

Ft. Belvoir, Virginia

2. Dudley Knox Library

Naval Postgraduate School
Monterey, California

43