Using the WVTransform

At the time of this writing there are five WVTransform subclasses with different capabilities,

  • WVTransformBoussinesq Non-hydrostatic, variable stratification
  • WVTransformHydrostatic Hydrostatic, variable stratification
  • WVTransformConstantStratification Non-hydrostatic, constant stratification
  • WVTransformStratifiedQG Quasigeostrophic dynamics, variable stratification
  • WVTransformBarotropicQG Quasigeostrophic dynamics, equivalent barotropic

These WVTransform subclasses are used in the same way, but may have different capabilities and requirements for initialization.

Initialization

All transforms require you specify the domain size, number of grid points and latitude.

The Boussinesq, hydrostatic and stratified QG transformations require you pass a function handle describing the stratification, e.g.,

wvt = WVTransformHydrostatic([Lx, Ly, Lz], [Nx, Ny, Nz], N2=@(z) N0*N0*exp(2*z/L_gm),latitude=33);

whereas constant stratification transform requies the buoyancy frequency \(N_0\)

wvt = WVTransformConstantStratification([Lx, Ly, Lz], [Nx, Ny, Nz], N0=5.2e-3,latitude=33);

The equivalent barotropic transform requires that you specify an equivent depth \(h\),

wvt = WVTransformBarotropicQG([Lx, Ly], [Nx, Ny], h=0.80,latitude=33);

where a typical oceanic value for the first barolinic mode would be around 80 cm.

Grids

Note that you do not specify the grids, only the domain size—the grids are determined by the transforms.

The dimensions that result can be accessed with, e.g., wvt.x or wvt.z. This will be array of size [Nx 1] and [Nz 1], respectively. More useful perhaps are the grids, e.g., wvt.X or wvt.Z which are of size [Nx Ny Nz].

For the current set of transforms \(x\) and \(y\) grids will always be evenly spaced grids appropriate for Fourier transforms, while the \(z\) grid will be evenly spaced for constant stratification (sine and cosine transforms), but will have variable spacing when using variable stratification. Never create your own grids—always use the built-in functions.

Initial conditions

Once a tranform is instantiated, it is often the case that one would like to add initial conditions. The WVTransform class has numerous methods for adding initial conditions—including very general initialization from any fluid state, as well as initialization specific to waves, inertial oscillations, and geostrophic motions.

Initializing from \((u,v,\eta)\)

The most direct methods for initializing the model are initWithUVEta and initWithUVRho which take either \((u,v,\eta)\) or \((u,v,\rho)\). As a simple example, let’s intialize with an inertial oscillation initial condition, \((u_0 \exp(z/100),0,0)\). In code, this is

wvt.initWithUVEta( 0.2*exp(wvt.Z/100), 0*wvt.X, 0*wvt.X );

These methods can be used to initialize from any flow fields that use the same stratification and boundary conditions as the WVTransform that is being used. For example, you might use output from another model and use initWithUVRho to initialize the WaveVortexMode with that output.

Initializing waves, inertial oscillations, and geostrophic motions

The wave-vortex model provides methods for initializing the fluid with specific dynamical solutions, including inertial oscillations, internal gravity waves, and geostrophic (vortex) flows.

To initialize individual waves, use initWithWaveModes and the related methods, e.g.,

[omega, k, l] = wvt.initWithWaveModes(kMode=10,lMode=0,j=1,phi=0,u=0.2,sign=1);
period = 2*pi/omega;

will initialize a first baroclinic mode wave with a wavenumber of \(k = 10(2\pi)/L_x\).

To initialize with a geostrophic stream function, use initWithGeostrophicStreamfunction and the related methods, e.g.,

Le = 35e3;
z0 = -wvt.Lz/4;
He = wvt.Lz/10;
U = 0.25; % m/s
psi = @(x,y,z) U*(Le/sqrt(2))*exp(1/2)*exp(-((x-x0)/Le).^2 -((y-y0)/Le).^2) .* (erf((z-z0)/He)+1)/2;

wvt.setGeostrophicStreamfunction(psi);

creates a deep eddy.

The initialization methods for the WVTransform all use the same nomenclature,

  • init—clears ALL variables Ap, Am, A0, then sets/adds
  • set—clears only the component requested, and sets with new value.
  • add—adds to existing component
  • removeAll—remove all features of given type

Other initialization methods

It can also be useful to initWithRandomFlow. Also check out the section on reading and writing to file for how to use initFromNetCDFFile to quickly read in the ocean state from a saved file.

Examining the fluid state

Once you have a WVTransform instance, you can now query it to return different state variables. The term “state” is used because these variables tell you something about the state of the fluid at the time in question.

Standard variables that are available are documented with the WVTransform and include \((u,v,w,\rho,p)\), the sea-surface height and velocities, the quasi-geostrophic potential vorticity (qgpv), the nonlinear fluxes \((F_p, F_m, F_0)\) and others. The transform has numerous built-in state variables

wvt.summarizeVariables

will enumerate all current variables.

There are two ways to access state variables. In the first way, you can query the transform using Matlab’s dot syntax, e.g.,

u = wvt.u;

will return a matrix of the \(u\) velocity component. This can then be plotted in the usual way, e.g.,

figure
pcolor(wvt.y,wvt.z,squeeze(wvt.u(1,:,:)).'), shading interp

where the properties of the WVTransform wvt.y and wvt.z refer to those dimensions.

The other way to access variables is to query them in a list using the variables method,

[u,v,w] = wvt.variableWithName('u','v','w');

this returns all the variables requested at the grid points. However, you can also the variableAtPositionWithName method to return the value of any variable at any set of points in the domain. This method is used to do particle advection by the WVModel.

This site uses Just the Docs, a documentation theme for Jekyll.