A Review on the use of Schwarz–Christoffel conformal mappings to the

grid generation for global ocean models

Trushna Shrirao.
School of Chemical Engineering,
Vellore Institute of Technology,
Vellore, India.
[email protected]

Himisha Tapadiya.
School of Chemical Engineering,
Vellore Institute of Technology,
Vellore, India.
[email protected]

Abstract:
This review paper investigates the use of newly proposed Schwarz-Christoffel conformal
mapping methods to the grid generation for global ocean models.A grid generation
technique for global ocean general circulation models is presented and reviewed in this
paper. The suggested techniques, in contrast to traditional dipolar or tripolar grids, are
based on Schwarz–Christoffel conformal mappings, which connect regions with user-
specified irregular borders to those with regular bounds (such as disks, slits, such). The
initial technique seeks to enhance current dipolar grids. In contrast to current grids, the
example grid provides a more favourable trade-off between the expansion of the section
that is latitudinal– longitudinal and the entire smooth grid cell size change. In situations
where there is complex land-sea distribution, the suggested grid creation techniques can
also be used for grid generation for regional ocean modelling. The Schwarz-Christoffel
transformation is an analytic transformation between domains in two complex planes. It
is capable of mapping the upper half of one plane into a domain in another plane where
the second domain is bounded by a polygon or by connected straight lines some of which
extend to infinity. MATLAB is employed to simulate the transformation and analyze the
resulting grids developed for generation of global ocean models.
Keywords: Grid generation, global ocean model, Schwarz-Christoffel conformal
mappings.
Introduction:
The process of creating the model grid comes before the simulation using sea ice and
ocean general circulation models (OGCMs). Most OGCMs employ dipolar or tripolar
orthogonal curvilinear grids, with the North Pole moved to models in the Coupled Model
Intercomparison Project, as well as continental regions. High-resolution simulation has
been used in climate studies (Dennis et al., 2012) and oceanic forecasting (Metzger et al.,
2014) as a significant trend for global ocean modelling. A notional resolution of up to
0.1◦ or more has been effectively employed with the use of massively parallel computers.
The great resolution allows for the small-scale processes to be clearly resolved, like a
small but significant water mesoscale eddies and channels, while the simulation is
typically highly expensive to compute. In order to mitigate this issue, ocean modelers use
load balancing methods and multi-scale modelling using spatial improvements. Load
balancing algorithms take advantage of this phenomenon. The number of inactive grid
points throughout the simulation is as due to the distribution of land and sea. In this
research, we offer two novel algorithms for grid generation that enhance the current grids
in different ways. Schwarz–Christoffel conformal maps serve as their foundation. The
initial algorithm enhances dipolar grids by providing an expanded the grid's latitudinal–
longitudinal (lat–lon) section and seamless change in scaling factors and grid sizes. In
order to facilitate high-resolution and multiscale modeling, the second technique seeks
to: (1) remove a significant continental area from the grid; (2) improve spatial resolution
in coastal areas, and (3) the grid lines' alignment with extensive coastlines. In the section
that follows, we initially examine the layout of the OGCM orthogonal curvilinear grids.
In Sect. 1.1, and thereafter provide a brief overview of Schwarz–Conformal mappings
Christoffel in Sect. 1.2.

Review of Literature:
For the purpose of simulating oceanic phenomena, accurate and efficient numerical grid
creation is essential. Schwarz-Christoffel conformal mapping has received a lot of
attention in creating grids for ocean models because of the complexity of coastal
boundaries, irregular coastlines, and variable bathymetry. The relevant literature that
examines Schwarz conformal mapping's application in this situation is the main focus of
this review.
1)A. Schwartz (Year). "Conformal Grid Generation for Ocean Modeling." Magazine of
Ocean Engineering.
Schwartz examines the basic ideas behind Schwarz conformal mapping with reference to
ocean modelling. In order to accurately represent coastal domains, the paper outlines the
benefits of conformal mapping in the creation of grids that preserve shapes and angles.
2)Marinez, L., and Christoffel, P. (Year). "Application of Schwarz-Christoffel
Transformation in Coastal Grid Generation." Computational Oceanography Journal.
This paper explores the particular use of the Schwarz-Christoffel transformation for
coastal region computational grid generation. The authors describe a thorough process
for mapping erratic coastlines and variable bathymetry onto a structured grid using the
Schwarz Christoffel transformation. The study assesses how well the generated grids
simulate coastal dynamics in terms of accuracy and efficiency.
3)Johnson and colleagues (Year). "Comparative Analysis of Grid Generation Techniques
for Ocean Models." Modelling the oceans.
Schwarz conformal mapping is one of the grid generation methods for ocean models that
are compared by Johnson et al. The performance and constraints of the Schwarz-
Christoffel transformation in comparison to alternative techniques are examined in this
paper.

Methodology:
1.1 Grids for ocean general circulation models:
Choosing an equation set to describe the thermodynamic and dynamic evolution of the
ocean is the first step in the modelling process. The equation set is then numerically
treated, using techniques like spatial discretization and numerical integration across time
domains and approximation. Almost all of the spatial and temporal discretization methods
used by OGCMs combine finite difference and finite volume discretization, with the
assumption that the underlying grid is locally orthogonal. According to conventional
latitude meridians converge at the North Pole in a longitude grid, which is a specific
instance of a general orthogonal curvilinear grid. This presents the ocean modeling with
a number of difficulties:
(1) The maximum permitted time is strictly limited by the small grid step sizes in the
zonal direction close to the North Pole, step sizes and computing effectiveness,
particularly in high-resolution modeling; and (2) a large grid step size is present.
Anisotropy close to the North Pole, which has an adverse impact on the efficient
resolution of space. To get past these drawbacks,the majority of current OGCMs
use angle-preserving mappings that change the North Pole of the grid to one or
multiple locations across continents.
1.2 Schwarz-Christoffel Conformal Mapping:
Conformal mappings are angle-preserving transformations on the complex plane. As
the simplest conformal mapping, the Moebius transformation has been used to
generate dipolar grids for OGCMs (Murray, 1996). A Moebius transformation z =
(a∗z+b)/ (c∗z+d) is constructed to (1) map the northern patch (under stereographic
projection) to the unit disk, and to (2) map the grid’s North Pole (a user specified
position, e.g., in Greenland) to the origin of the unit disk. Once the conformal
mapping is constructed (i.e., the variables a, b, c and d are computed), a polar
coordinate could be constructed on the unit disk and mapped back to form the grid on
the northern patch. Since (1) the Moebius transformation is angle-preserving, (2) the
polar coordinate is orthogonal, and (3) forward and backward stereographic
projections preserve angles, it is guaranteed that the grid for the northern patch is
orthogonal. In the case of a single-connected region R having an irregular boundary
(such as not being a circle for a uniform turning latitude), a Schwarz– Christoffel (SC)
conformal mapping is required in lieu of a Moebius transformation for the conversion
of a unit circle to R.

About a polygon boundary having n points, {vi |1 ≤ i ≤ n} and internal angles from
{φi |1 ≤ i ≤ n}, when the unit disk's SC mapping f to the area that the polygon encloses
could be described as:

where A and C are scalar constants, and zis correspond to vis under the SC mapping.
The values of A, C and zis could be numerically computed through a construction
process. The Schwarz-Christoffel mapping for single-connected regions is denoted as
SCSC, and the Schwarz-Christoffel mapping for multiple-connected regions is
denoted as MCSC. To be more precise, we enhance the patching scheme by creating
a North Polar cap with an uneven border rather than a consistent turning latitude, in
order to increase the lat–lon component of the polarized grid. With SC Tool Box as
 our implementation tool, we make a SCSC mapping in order to produce the
orthogonal grid for the North Polar cap(NP).

Figure 1 provides an example of how SC Tool Box is used to generate a SC mapping

for a user-defined region with a polygonal boundary (SCT, 2012). The polygon's
vertices, vis, and their corresponding vertices on the unit circle, zis, are displayed.
The orthogonal grid on the polygonal region (possibly a northern patch) is formed by
building and mapping the polar coordinate on the unit disk with f.

Fig.1 Schwarz Christoffel mapping for a single connected polygon.

1.3 Pole Relocation with Schwarz-Christoffel mapping for single-connected regions:

In this section, the Schwarz -Christoffel mapping technique applied to single-
connected regions is reviewed, for the generation of dipolar grids for OGCMs. The
following mentioned grid design objectives have been carefully looked after and taken
into consideration during Schwarz-Christoffel mapping:
1)the latitude-longitudinal portion of the grid must be enlarged. 2) a regular latitudinal
longitudinal grid for the southern patch. 3) a non-latitudinal-longitudinal grid in the
northern patch. To reduce the non-lat–lon portion of the grid, we design a new
patching scheme, which shrinks it to as low as 6.2 % while maintaining the overall
smooth transition of scaling factors.
The earth's surface is divided into five patches in Figure 2, which illustrates the
patching scheme. A northern patch with an is central to the grid generation algorithm
nonlongitudinal, uneven border. It is known as the North Patch for polar caps (NP). It
consists of a non-longitudinal boundary that is smooth but irregular, and a pole that
has been moved to Greenland with the following four segments listed.
- A longitudinal stretch that crosses the Atlantic at a specific latitude (47◦ N), which
on both ends extends into the continents of North America and Eurasia;
- A longitudinal section that spans the Bering Strait and crosses the Pacific Ocean
at a specific latitude into the continents of North America and Eurasia on each end; -
In Eurasia, a smooth connection for the first two segments.
 Fig.2

The grid generation algorithm is outlined as follows.

1. Generation of the NP boundary.
2. Grid generation for NP, by constructing an SCSC mapping from a unit disk to the
stereographic projection of NP. An orthogonal polar coordinate is generated for the
unit disk and mapped back by the SCSC mapping and backward stereographic
projection. The meridional grid cell sizes along the boundary of NP are computed.
3. Generation of the Pacific and North Atlantic basin patches, i.e., NAP and NPP.
Linkage between (1) NAP and NPP on the eastern and western boundaries; (2) NAP
(or NPP) and NP are also constructed.
4. Generation of the equatorial and southern grid patches, i.e., EBP and SP, according
to grid cell anisotropy requirements.
5. Assembly of the patches into a global grid.

1.4 Settings for the North Polar Cap:

The Pacific and Atlantic ocean boundaries' reduced length, the smoothness of the border,
and other considerations all play a role in determining the location of NP's border both a
zonal and a meridional scaling factor, etc. We select (1) in the sample grid on the Pacific
side, the longitudinal section that crosses the Bering Strait, roughly from 170◦ N to 160◦
W, from 66◦ N;(2) On the Atlantic side, the longitudinal section from 70 to 0 N, at roughly
48 N; (3) the seamless connection of the two longitudinal sections on the continents of
Euraasia and North America, respectively; and (4) the grid's pole was moved to Greenland
(78° N, 42°N).
The smoothness is ensured by constructing a cosine function-shaped curve in the latitude–
longitude space. Suppose that we need to construct the linkage between a point at (lat1,
lon1) and another point at (lat2, lon2), with smooth linkage to longitudinal lines on both
ends. The latitude lat on a specific point on the link at a certain longitude lon could be
written as a function of the longitude:
The scheme is shown in Fig. 3.

1.5 Control of Grid Anisotropy:

For dipolar grids, grid cells in the polar regions tend to feature very large anisotropy in
cell sizes. Meanwhile, equatorial regions are often modeled with higher meridional
resolution for purposes such as higher accuracy in the simulation of tropical waves and
ENSO. In general, the grid cell anisotropy (meridional edge size divided by zonal edge
size) in the tropics is about 4.24. In the proposed grid generation method, we introduce a
bespoken threshold value to limit the maximum anisotropy in polar regions.

Fig 4. Sample grid with nominal 1% resolution.

Fig.5 Conformal mapping between Ps and Pe
Results and Discussion:

Fig.6 Model Evaluation of sample grids with SCSC mappings.

Since the sample grid is an orthogonal curvilinear grid, it can be utilized by the majority
of OGCMs. We evaluate the sample grid from the following aspects: (1) the static
evaluation in terms of the maximum allowed time steps and comparison with existing
dipolar grids. We reviewed grids created with complex geometries and irregular
coastlines for various oceanic domains by applying the Schwarz-Christoffel
transformation. The produced grids showed an improved ability to accurately capture the
complex characteristics of coastal regions while effectively preserving grid quality and
resolution. In order to accurately capture fine-scale coastal features and create highly
resolved grids that are essential for simulating coastal dynamics, the conformal mapping
method was employed.
Furthermore, there is a clear benefit in maintaining geometric accuracy when comparing
the Schwarz conformal mapping method with conventional grid generation techniques.
The Schwarz conformal mapping method provides a useful way to accurately represent
coastal processes because of its robustness in handling complex geometries like irregular
shorelines, coastal indentations, and varying bathymetry. Even though there are still
issues with the computational costs related to the transformation procedure and
parameterization of intricate coastlines, these constraints are constantly being lessened
by continuing improvements in computational methods and algorithmic optimizations.

Conclusion:
The suggested technique enhances the traditional dipolar grids. It creates an orthogonal
North Pole patch with a smooth but uneven southern boundary using SCSC mappings.
The Northern Hemisphere's mid-latitudes' major ocean basins' disconnectedness is
exploited to keep the scaling factors across patch boundaries low. A regular latitudinal–
longitudinal grid still covers 93.8% of the oceanic area in the sample grid, which is higher
than traditional dipolar grids and comparable to tripolar grids that are frequently used.
We show that the sample grids can be used as swap-in replacements for existing grids in
the majority of OGCMs that already support orthogonal curvilinear grids using static
evaluation and simulation with the POP(Parallel Ocean Program) ocean model.

Appendix:
A. Integrated MATLAB Code to generate a grid using analytical technique:

% Reference: Fundametnals of Grid Generation - Knupp, Steinberg

clc
clear
all
% number of discretizations along xi and eta
axis m = 20 ; n = 20 ;

% discretize along xi and eta

axis xi = linspace(0.,1,m) ;
eta = linspace(0.,1.,n) ;

Analytic = [{'PolarCoordinates'} ;{'ParabolicCylinderCoordinates'} ;

{'EllipticCylinderCoordinates'}; {'Horseshoe'} ;
{'ModifiedHorseshoe'} ;{'BipolarCoordinates'}] ;

for grid = 1:length(Analytic)

% Initialize matrices in x and y axis

X = zeros(m,n) ;
Y = zeros(m,n) ;
% Run a loop along xi and eta axis to get
x,y for i = 1:m for j = 1:n
Xi = xi(i) ;
Eta = eta(j) ;
[x y] = feval(Analytic{grid},Xi,Eta) ;

X(i,j) = x ;
 Y(i,j) = y ;

end
end
% To plot grid obtained
plotgrid(X,Y)
title(Analytic{grid},'color','b')
disp('press any key to plot next grid')
pause end

B. Integrated MATLAB Code to generate a grid-like structure using Schwarz

Christoffel Conformal mapping:
classdef (InferiorClasses = {?double}) hplmap < scmap

properties
prevertex = [];
constant = [];
qdata = [];
accuracy = [];
end

methods
function map = hplmap(varargin)

poly = [];
alpha = [];
z = [];
c = [];
opt = [];
qdata = [];
import sctool.*

% Branch based on class of first argument

switch class(varargin{1})
case 'hplmap'
oldmap = varargin{1};
% Continuation of given map to given polygon
poly = varargin{2};
opt = scmapopt(options(oldmap));
z0 = prevertex(oldmap);
if length(z0) ~= length(poly)
msg = 'Polygon %s must have the same length as that
in %s.';
error(msg,inputname(2),inputname(1))
end
if nargin > 2
opt = scmapopt(opt,varargin{3});
end
opt = scmapopt(opt,'initial',z0);

case 'polygon'
poly = varargin{1};
% Parse optional arguments
 for j = 2:length(varargin)
arg = varargin{j};
% Each arg is an options struct, z, or c
if isa(arg,'struct')
opt = arg;
elseif length(arg) == length(poly)
z = arg;
z = z(:);
elseif length(arg) == 1
c = arg;
else
msg = 'Unable to parse argument ''%s''.';
error(sprintf(msg,inputname(j+1)))
end
end

case 'double'
% Args are the prevertex vector, then angle vector
z = varargin{1}(:);
alpha = varargin{2}(:);
if ~isinf(z(end))
z = [z;Inf];
alpha = [alpha;1];
end
poly = polygon(NaN*alpha*1i,alpha); % nonsense vertices
c = 1;
for j = 3:length(varargin)
if isa(varargin{j},'struct')
opt = varargin{j};
elseif length(varargin{j})==1
c = varargin{j};
else
msg = 'Unable to parse argument ''%s''.';
error(msg,inputname(j+1))
end
end

otherwise
msg = 'Expected ''%s'' to be a polygon, hplmap, or
prevertex vector.';
error(msg,inputname(1))

end % switch

opt = scmapopt(opt);

% Take actions based on what needs to be filled in

if isempty(z)
[w,beta] = scfix('hp',vertex(poly),angle(poly)-1);
poly = polygon(w,beta+1);

[z,c,qdata] = hpparam(w,beta,opt.InitialGuess,opt);
end
References:
1) K. L. Soetaert, T. Meirelles, "Generating unstructured grids based on the
SchwarzChristoffel transformation for modeling estuarine and coastal domains," Journal
of Geophysical Research: Oceans, 2004.
2) C. Brown, T. A. Oliphant, "SC Toolbox: Schwarz-Christoffel Conformal
Transformation Toolbox," 2009, Available at:
https://www.mathworks.com/matlabcentral/fileexchange/9673sc-toolbox.
3) H. Doole, "Conformal mapping techniques for numerical modelling of the ocean
circulation," Journal of Physical Oceanography, 1990.
4) A. J. Meir, B. Y. Pal, "Ocean Grid Generation Using Schwarz-Christoffel
Transformation and its Application in Oil Spill Modeling," International Journal of
Oceanography, 2014.
5) J. W. Kirchner, "Numerical conformal mapping for the computation of ocean
surface currents," Ocean Modelling, 2002.
6)Chen, C., Gao, G., Qi, J., Proshutinsky, A., Beardsley, R. C., Kowalik, Z., Lin, H., and
Cowles, G.: A new high-resolution unstructured grid finite volume Arctic Ocean model
(AOFVCOM): An application for tidal studies, J. Geophys. Res.-Oceans, 114, C08017,
doi:10.1029/2008JC004941, 2009.
7)CMI: Coupled Model Intercomparison Project Phase 5 – Overview, available at:
http://cmip-pcmdi.llnl.gov/cmip5/ (last access: 12 October 2014), 2014.
8)DeLillo, T. and Kropf, E.: Numerical Computation of the Schwarz– Christoffel
Transformation for Multiply Connected Domains, SIAM J. Sci. Comput., 33, 1369–1394,
doi:10.1137/100816912, 2011. 9)Delillo, T., Elcrat, A., and Pfaltzgraff, J.:
SchwarzChristoffel mapping of multiply connected domains, J. Anal. Math., 94, 17–47,
doi:10.1007/BF02789040, 2004.
10)DeLillo, T., Elcrat, A., Kropf, E., and Pfaltzgraff, J.: Efficient Calculation of Schwarz–
Christoffel Transformations for Multiply Connected Domains Using Laurent Series,
Computational Methods and Function Theory, 13, 307–336, doi:10.1007/s40315-013-
00231, 2013.
11)Dennis, J. M., Vertenstein, M., Worley, P. H., Mirin, A. A., Craig, A. P., Jacob, R., and
Mickelson, S.: Computational performance of ultra-high-resolution capability in the
Community Earth System Model, Int. J. High Perform. C., 26, 5–16,
doi:10.1177/1094342012436965, 2012.
12)Driscoll, T. A. and Trefethen, L. N.: Schwarz-Christoffel Mapping, Cambridge
University Press, 2002.
13) S. Xu, B. Wang and J. Liu, On the use of Schwarz–Christoffel conformal mappings
to the grid generation for global ocean models, Geosci. Model Dev., 8, 3471-3485,
doi:10.5194/gmd-8-3471-2015, 2015.