Advanced usage of the SHWindow class interface

The SHWindow class provides an interface to compute window functions that concentrate energy spatially in a given region and within a given spherical harmonic bandwidth. This allows to calculate spectral estimates localized to specific regions with the best possible spectral resolution.

To demonstrate, in this example, we will determine the spectrum of the bathymetry of Earth and compare it with that of the surface topography. To this end, we will need to generate window functions that are separately concentrated over the land and the oceans. To start, we read the topography of Earth from the example files, create masks for the oceans and land, and plot the masks:

In [1]:
%matplotlib inline

import matplotlib.pyplot as plt
import numpy as np
import pyshtools as pysh
In [2]:
%config InlineBackend.figure_format = 'retina'  # if you are not using a retina display, comment this line
In [3]:
clm = pysh.datasets.Earth.Earth2014.tbi(lmax=360)
topo = clm.expand(grid='DH2')

land_mask = pysh.SHGrid.from_array( > 0)
ocean_mask = pysh.SHGrid.from_array( <= 0)

fig, (col1, col2) = plt.subplots(1, 2)
land_mask.plot(ax=col1, tick_interval=(45, 45), minor_tick_interval=None)
col1.set(title='Land mask')
ocean_mask.plot(ax=col2, tick_interval=(45, 45), minor_tick_interval=None)
col2.set(title='Ocean mask');

Next, let's plot the masked topography over the oceans and land:

In [4]:
fig, ax = (topo * ocean_mask *(-1)).plot(colorbar='right', cb_label='Bathymetry', show=False)    # show=False is used to avoid a warning when plotting in inline mode
fig2, ax2 = (topo * land_mask).plot(colorbar='right', cb_label='Elevation', show=False)

Optimally concentrated window functions

SHTOOLS provides a method to solve the so called concentration problem, that is, to find windows with a specified maximum spherical harmonic degree lmax that are optimally concentrated in the region of interest. The smaller the region, the larger lmax will need to be to allow for a good concentration. Because the oceans span a larger area than the land mass, the spherical harmonic bandwidth of the ocean windows lmax_ocean can be smaller than the spherical harmonic bandwidth of the land windows lmax_land while still providing the same number of well concentrated windows.

Creating localization windows is easy using the SHWwindow method from_mask(). All we need to do is provide the mask, the maximum spherical harmonic bandwidth, an