release-notes-v4.md

title	keywords	sidebar	permalink	summary	toc
SHTOOLS release notes: Version 4	spherical harmonics software package, spherical harmonic transform, legendre functions, multitaper spectral analysis, fortran, Python, gravity, magnetic field	mydoc_sidebar	release-notes-v4.html		true

Version 4.6

New extended grids

All grid formats now allow to compute the redundant values at 360 E longitude (GLQ and DH), as well as at 90 S (DH only). These extended grids are now the default in pyshtools, but remain optional in the Fortran 95 routines. The use of extended grids is controlled by the optional argument extend. The purpose of these extended grids is to better integrate with the plotting routines that require these points (i.e.., Cartopy and pygmt).

Improved plotting and map projections

The plotting routine SHGrid.plot() has been refactored to allow support for projections using Cartopy and pygmt.

• An incorrect 0.5 pixel offset was fixed when plotting grids via matplotlib, and grids now correctly plot both 0 and 360 degrees using the new "extended" grids of SHGrid.
• Support was added for Cartopy projections, by specifying: SHGrid.plot(projection=ccrs.ProjectionName()).
• The argument colorbar now takes the options 'top', 'bottom', 'left' or 'right'.
• Improved plotting and placement of colorbars. New optional arguments include cb_label for labels, cb_ylabel for a label on the y axis of the colorbar, cb_tick_interval for specifying the major tick interval, cb_minor_tick_interval for specifying minor tick intervals, cb_triangles for plotting upper/lower limit triangles at the ends of the colorbar, cb_width to specify the colorbar width, and cb_offset to override the default spacing between the map and colorbar.
• Improved colormap handling: New optional arguments include cmap_limits to specify the lower and upper bounds of the data, as well as an interval for constant color intervals, and cmap_reverse to reverse the colormap.
• Improved handling of ticks and annotations: The optional argument ticks specifies which ticks and annotations to show, using a syntax from the generic mapping tools (i.e., 'WSen').
• Experimental support for pygmt using the routine SHGrid.plotgmt(). This function takes nearly the same arguments as plot(). As soon as pygmt implements projection classes (GenericMappingTools/pygmt#379), this will be incorporated into the plot function in the same manner as Cartopy was.
• All gravity, magnetics, tensor, localization windows and slepian function plotting routines incorporate these changes.
• Added a new introductory notebook that shows how to use all features of the plot() function.

Improved integration with xarray DataArrays, xarray DataSets, and netcdf files

• Added the methods to_netcdf() and from_netcdf() to the SHCoeffs, SHGravCoeffs and SHMagCoeffs classes.
• Added the method SHGrid.from_xarray() to initialize a grid from an xarray DataArray.
• Added improved descriptive attributes for netcdf files that mirror these conventions.
• Added the method SHGeoid.to_xarray() to export an xarray DataArray and to_netcdf() to export a netcdf object readable by the generic mapping tools.
• Added the methods SHGravGrid.to_xarray() and SHMagGrid.to_xarray() to export all gridded data (radial, theta, phi, total, and potential) as an xarray DataSet.
• Added the methods SHGravTensor.to_xarray() and SHMagTensor.to_xarray()to export all gridded data (Vxx, invariants, eigenvalues) as an xarray DataSet.

Gravity routine improvements

• Added the method SHGravCoeffs.center_of_mass to calculate the center of mass of a body.
• Added the method SHGravCoeffs.inertia_tensor() to calculate the moment of inertia tensor.
• Added the Earth dynamical flattening constant H (IERS Conventions 2010) to the constant module.
• The read_icgem_gfc() function was extended with the option encoding as some models in ICGEM are not in UTF-8.
• Addded the method centroid() to the class SHCoeffs. The centroid is computed as the center of mass of a homogeneous object.

Other changes

• New methods SHGrid.to_real() and SHGrid.to_imag() return the real and imaginary components of a complex SHGrid instance.
• Added an optional argument copy to SHCoeffs.pad().
• Fixed bugs in the Fortran code of PlBar_d1 and PlON_d1 when calculating the Legendre polynomials at the north and south pole.
• Spherical harmonic coefficients can be read remotely by specifying a URL as the filename. This functionality uses requests.get(), and has been implemented in the function shread() and the SHCoeffs method from_file().
• Fixed a bug in the fortran code of Curve2Mask. As part of this fix, the optional parameter centralmeridian has been removed as it is no longer required. The longitudes of the curve can possess values from -360 to 720 degrees, and the routine searches for discontinuities that may occur between two successive points as the longitudes pass from 360 to 0, or -180 to 180 degrees.

Version 4.5

SHCoeffs

• Added cross_spectrum(), plot_cross_spectrum() and plot_cross_spectrum2d() methods to the SHCoeffs class.
• Added from_cap() constructor to create coefficients of a spherical cap.
• Fixed a small bug when rotating coefficients in the SHCoeffs, SHMagCoeffs and SHGravCoeffs classes when the input csphase is different from the default value.

SHGrid

• Added the method from_zeros() to initialize a grid with zeros.
• Added the method from_cap() to initial a grid with a spherical cap.
• Added support for saving gridded data to netcdf and xarray formats. to_netcdf() exports data to netcdf format, and when saved to file can be used directly with GMT (generic mapping tools) where they are known as 'grd' files. to_xarray() exports data to an xarray DataArray.
• Fixed a type error when expanding complex coefficients at arbitrary points.

Slepian functions

• Added the fortran function SHSCouplingMatrix and Slepain class method coupling matrix() for computing the coupling matrix that relates the Slepian expansion power spectrum to the global power spectrum.
• Added the fortran function SHSCouplingMatrixCap which is optimized for working with spherical cap Slepian functions.
• Improved the plotting capabilities of the method plot_coupling_matrix, such as the addition of colorbars, and the option to normalize to maximum value to unity.
• Added the option taper_degrees() and slepian_degrees() to the SHWindow and Slepian classes, respectively, to allow for the construction of Slepian functions that exclude certain spherical harmonic degrees.
• Added the fortran function SHMTVar and the SHWindow class method variance() to compute the variance of a multitaper spectral estimate (based on SHMTVarOpt).
• Added the fortran function SHSlepianVar and the Slepian class method variance() to compute the variance of a Slepian expansion spectral estimate.
• Improved the handing of the optional parameter weights in all methods of SHWindow.
• Changed the name of the optional parameter nwin of SHWindow.coupling_matrix to k for consistency with the localized spectral analysis routines.

Fortran 95

• The Fortran code was modified to be strictly compliant with the f95 standard (-std=f95 in the gfortran compiler). Double precision, double complex, and long integers are defined as real(dp), complex(dp), and integer(int8), and the types are defined in a new module ftypes.f95.
• A few intrinsic function calls have been renamed to conform with the standards, and all double precision constants are now defined by appending _dp to them.

FFTW

• The syntax of the fftw routines has been updated. In particular, the old call dfftw_execute(plan) statements now include all their dependent variables using the new syntax call fftw_execute_dft(plan, grid, coef). Importantly, the old syntax caused the GCC9 optimizer to break the spherical transform routines, generating meaningless output for large parts of the grids or coefficients.
• The fortran routines now access the FFTW library using Fortran 2003 standards. Though most compilers do not yet support all features of F2003, the few features used here (such as the module iso_c_binding) are claimed to be supported by the majority of modern compilers. This allows us to access the FFTW routines without use of fortran bindings, which are not always included in compiled versions of FFTW.
• All FFTW routines are explicitly defined in an interface block in FFTW3.f95. For simplicity, the use of the optional parameter FFTW_UNDERSCORE used in the Makefile has been deprecated.

M. A. Wieczorek, M. Meschede, E. Sales de Andrade, I. Oshchepkov, B. Xu, and A. Walker (2019). SHTOOLS: Version 4.5, Zenodo, doi:10.5281/zenodo.2350781

Version 4.4

New Slepian expansion routines

• Added the new Fortran 95 functions SHRotateTapers for rotating spherical cap Slepian functions, SlepianCoeffs for expanding a function in a Slepian basis, and SlepianCoeffsToSH to convert the Slepian coefficients into spherical harmonic coefficients.
• Added two python classes Slepian and SlepianCoeffs for managing Slepian basis functions and the Slepian expansion coefficients of a function.

Legendre and spherical harmonic convenience functions

• Added 4 python convenience functions for computing Legendre and spherical harmonic functions: legendre() to compute all the legendre functions, legendre_lm() to compute the legendre function for a specified degree and order, spharm() to compute all the spherical harmonic functions, and spharm_lm() to compute the spherical harmonic functions for a specific degree and order.

Other improvements

• Added the option to perform a weighted least squares inversion in the fortran routine SHExpandLSQ, and added a python wrapped function SHExpandWLSQ().
• Added min() and max() methods to the SHGrid class, to return the minimum and maximum value of the gridded data.
• Added the property mass to the SHGravCoeffs class, which is computed by the input GM and the codata value for G.
• Add the optional parameter omega to SHGravCoeffs.geoid() to override the value provided in the class instance.
• Fixed some minor typos and usability issues, and cleaned up the python wrapper and signature files.

M. A. Wieczorek, M. Meschede, E. Sales de Andrade, I. Oshchepkov, B. Xu, and A. Walker (2018). SHTOOLS: Version 4.4, Zenodo, doi:10.5281/zenodo.592762

Version 4.3

New Gravity and Magnetic field classes

• Added gravity classes SHGravCoeffs, SHGravGrid, SHGravTensor and SHGeoid.
• Added magnetic field classes SHMagCoeffs, SHMagGrid, and SHMagTensor.
• Added new fortran subroutine MakeMagGradGridDH, which is analogous to MakeGravGradGridDH.
• Reorded the arguments of CilmPlusRhoHDH to be consistent with CilmPlusDH.
• The python routine MakeMagGridDH now also outputs the magnetic potential as a grid.

Better figures

• Addition of the function pyshtools.utils.figstyle(), that sets several matplotlib parameters for better figures. This function takes as optional parmeters the maximum useable width of a journal page, the relative width of the figure with respect to this value, and the screen resolution in dpi.
• Most plotting routines have optional parameters to set minor tick intervals, grids, label font size, and tick font size.
• Degree symbols are plotted on tick labels for maps.
• examples/python/Common/FigStyle.py was removed from the examples.
• Added the options vmin and vmax to the plotting methods SHCoeffs.plot_spectrum2d() in order to specify the limits of the color scale.
• Added the option to plot colorbars on SHGrid plots, along with the option to specify their orientation and a text label.
• All notebooks have been updated.

New constant subpackage

• The constant subpackage has been completely rewritten and now makes use of the astropy Constant class. This class has attributes name, value, uncertainty, unit, and reference. The naming of the constants has changed in some cases for consistency. A few constants that are not necessary were removed. Many of the constants were updated with more recent values. Constants can be used in arithmetic operations with either other Constants or with objects of the astropy class Quantity.
• Constants are organized into modules for each of the planets (Mercury, Venus, Earth, Moon, and Mars), and for convenience, these are all added to the main namespace. The fundamental constants G and mu0 from the astropy constants package were added (as taken from CODATA 2014).

Other changes

• Fixed a bug in how the random coeffcients were determined for unnormalized coeffcients in SHCoeffs.
• Optional parameter seed added to SHCoeffs.from_random() to allow for reproducibility.
• One can now specify colat instead of lat for the method SHCoeffs.expand().
• Added __repr__ methods to all pyshtools classes.
• Changed the mathematical operators of SHCoeffs such that addition and subtraction of a constant only affects the degree 0 term.
• Added the optional parameter lmax to SHCoeffs.plot_spectrum() and SHCoeffs.plot_spectrum2d().
• Fixed a bug in SHCoeffs.pad() where the attribute mask was not similarly padded.
• For mathematical operations with SHCoeffs grids, it is now required that the two class instances have the same lmax.
• Clarified the documentation of SHRotateCoef to point out that this is only valid for intrinsically real functions that are expressed in complex harmonics.
• Added the method volume() to the class SHCoeffs, that calculates the volume of the object.
• Added the attributes area and shannon to SHWindow, which provides the area of the concentration domain and the shannon number.
• Removed python installation support from Makefile: use pip instead.

Citation:

M. A. Wieczorek, M. Meschede, E. Sales de Andrade, I. Oshchepkov, B. Xu, and A. Walker (2018). SHTOOLS: Version 4.3, Zenodo, doi:10.5281/zenodo.1346663

Version 4.2

Change log:

• Full support added for the use of unnormalized harmonics in the classes SHCoeffs and SHGrid. To make use of this normalization, just specify normalization='unnorm'.
• Added a new python native routine mag_spectrum() in the subpackage gravmag that replaces the original fortran wrapped routines. The old python wrapped functions have been removed from pyshtools. This new routine is nearly the same as spectrum(), and further allows one to compute the spectrum of the potential or magnetic intensity.
• The old fortran based SHRead function has been replaced with a python native version shread(). The functionality is nearly identical as before, and combines the previous routines SHRead, SHReadError, SHReadErrorH and SHReadH into one. Differences include: (1) It is no longer necessary to specify the lmax of the file: This is determined automatically by reading the file from the end, (2) both real and complex coefficients are supported, (3) a header line can be output, but it is a simple list of type str that will need to be converted to the correct format by the user, and (4) "comment" lines are read and ignored: A valid line is one where there are 4 or more words, and where the first two words are integers.
• A new python native function convert() was added in the subpackage shio that converts between arrays of spherical harmonic coefficients with different normalizations. The class SHCoeffs was then simplified by using this external function for all conversions involving SHCoeffs class instances.
• The optional parameter lmax was added to SHCoeffs.spectrum().
• When plotting grid from the class SHGrid, one can now specify the label to use for the x and y axes with xlabel and ylabel, as well as the interval to use when plotting ticks on both axes using tick_interval.
• The pyshtools rotation routines now allow you to specify the optional parameter convention to treat Euler angles in either the x or y conventions (i.e., which axes to use for the second rotation). Furthermore, the optional argument body allows you to specify if you want to rotate the body (True), or coordinate system (False, default). The tutorial number 3 was updated to clear up some inconsistencies in how the angles were defined.
• New optional parameters added to SHWindow.plot_windows() and SHWindow.plot_spectra() that include xlim and ylim for the limits when plotting spectra, maxcolumns for the number of columns to use when plotting several windows, and lmax which controls the grid spacing when plotting the windows.
• Added the optional argument lmax to SHCoeffs.from_random() that allows you to create coefficients with maximum bandwidths that are either greater or less than the bandwidth of the input power spectrum.
• Added a warning message when using SHCoeffs.rotate() with degrees greater than 1200, as the routine is not accurate beyond this value.
• Added an optional argument legend to SHWindow.plot_windows() to control whether the legend is plotted or not.
• Fixed a minor bug in ClassExample.py file concerning the use of SHCoeffs.rotate().
• Added support for plotting to an already existing figure by allowing the user to specify an existing matplotlib axes.
• Removed some non-standard ascii dashes in the documentation files, and forced all doc files to be opened as utf-8.
• HTML documentation was completely redone using Jekyll. The markdown source files are now located in pages/mydoc. A static html web site is built using jekyll, whose files are located in doc. Github will automatically create the static pages and serve them on shtools.github.io/SHTOOLS. To build the static pages yourself, it is only necessary to execute bundle exec jekyll build in the directory doc, which will build the site into _site in the same directory. Alternatively, make www in the main directory will create a static site in the top-level directory www that could be used to deploy on a different web server. The site is based on the template Jekyll documentation theme by @tomjoht.

Citation:

M. A. Wieczorek, M. Meschede, E. Sales de Andrade, I. Oshchepkov, B. Xu, and A. Walker (2018). SHTOOLS: Version 4.2, Zenodo, doi:10.5281/zenodo.1250054

Version 4.1

This version adds improved functionality to SHTOOLS and fixes a couple of minor bugs. In addition, this release will be the first where pre-built wheels for unix/macOS/windows will be distributed via PYPI.

Change log:

• Added an optional argument lmax to SHCoeffs.from_array().
• Coefficients are zero-padded when lmax is greater than the maximum degree in SHCoeffs.to_array().
• The method pad() was added to the SHCoeffs class that zero pads or truncates the coefficients to a different lmax.
• Fixed the method SHCoeffs.from_file() such that the maximum spherical harmonic degree of the class is the maximum spherical harmonic degree of the coeffs (and not lmaxin as before).
• Fixed formatting issues with error messages in SHCoeffs.
• Removed print statements from the fortran code in BAtoHilm and BAtoHilmRohH that served no purpose.
• Fixed a bug in the argument order of the python wrappers of CilmPlusRhoDH and BAtoHilmRhoDH.
• Fixed the makefile to remove the dist directory during clean.
• Fixed a bug in the python routine cross_spectrum(), where the numpy arange function was incorrectly called.
• Fixed theSHWindow plotting methods to work when the number of rows is equal to 1.
• Conditional tests in the routine Wigner3j were reordered to avoid a division by zero.
• Numpy's auto-configuration is now used to detect the LAPACK libraries.
• Many minor updates to the python documentation and unix man pages.

Citation:

M. A. Wieczorek, M. Meschede, E. Sales de Andrade, I. Oshchepkov, B. Xu, and A. Walker (2017). SHTOOLS: Version 4.1, Zenodo, doi:10.5281/zenodo.1067108

Version 4.0

This is a major update that fixes bugs, adds new functionality, and improves Python error handing. All users are requested to upgrade to 4.0.

Change log:

• Instead of executing a Fortran STOP, which kills the Python kernel, the Fortran subroutines now return an exitstatus that allows Python to raise an exception. This technique does not work with the few Fortran functions that pyshtools calls, but these functions are relatively benign, and will soon be phased out for Python native functions.
• The Fortran powerspectrum routines have been removed from pyshtools, and have been replaced with Python native routines spectrum and cross_spectrum. The Python routines allow to specify the normalization, whether the output should be power, energy or l2norm, and whether the spectrum is per degree, per coefficient, or per log bandwidth.
• The method plot_spectrum2d() was added to the class SHCoeffs to plot the power as a function of degree and order.
• All pyshtools modules have been converted into proper Python subpackages. The subpackage localizedpsectralanalysis has been merged into spectralanalysis, and the subpackage other has been renamed utils.
• The Python class method SHCoeffs.expand() now can evaluate the function either on an SHGrid or for a list of latitude and longitude points. As part of this change, a new fortran function MakeGridPointC was created for complex coefficients.
• The majority of the methods for the classes SHCoeffs, SHGrid and SHWindow have been rename for consistency (see documentation!). Also, the classes now give the option of reading or saving to files as numpy arrays.
• Added new Python function read_icgen_gfc for reading ICGEM-format gravity coefficient files.
• The operator pow was added to the class SHCoeffs.
• All methods in the pyshtools classes now return copies by default, which can be modified by the optional argument copy.
• Added pot as a mandatory return argument for the Python routine MakeGravGridDH.
• Several minor modifications and bug fixes were made to the makefiles to improve compatibility and to allow the use of make -j.
• The routines other.EigValSym, other.EigValVecSym, other.EigValVecSymTri, other.RandomGaussian, other.RandomN and other.PreGLQ were removed from pyshtools, as these can be found in other scipy packages.
• The SHTOOLS routine DHaj was added to the pyshtools subpackage utils.
• Python docstrings have been streamlined and standardized.
• ...plus, many minor changes and optimizations...

Citation:

M. A. Wieczorek, M. Meschede, I. Oshchepkov, E. Sales de Andrade, and heroxbd (2016). SHTOOLS: Version 4.0. Zenodo. doi:10.5281/zenodo.206114