PyGeopack

A Python wrapper for Geopack-2008. This includes the T89, T96, T01 and TS05 magnetic field models for Earth's magnetosphere. See https://ccmc.gsfc.nasa.gov/modelweb/magnetos/tsygan.html and http://geo.phys.spbu.ru/~tsyganenko/modeling.html for more information.

1. Requirements

The following Python packages will be installed automatically:

• numpy
• PyFileIO
• RecarrayTools
• DateTimeTools

The following packages will also be required for running the PyGeopack.UpdateParameters() routine:

• kpindex
• pyomnidata

This package is primarily a Python 3 wrapper for a mixture of C/C++/FORTRAN code which is compiled during installation. For building to work, the following are needed:

• g++

• binutils

• make

• gfortran

2. Installation

2.1 Linux

Firstly a few environment variables need setting up: $KPDATA_PATH, $OMNIDATA_PATH and $GEOPACK_PATH, which will point to the Kp index data, the omni data and the Geopack data, respectively. This can be done by including the following in your ~/.bashrc file, or by running it in the terminal before starting Python:

export KPDATA_PATH=/path/to/kp
export OMNIDATA_PATH=/path/to/omni
export GEOPACK_PATH=/path/to/geopack/data

where both of those directories must be writable by the current user, unless the data already exist in them.

Then simply install using pip3:

pip3 install PyGeopack --user

or by downloading the latest release on GitHub and running:

pip3 install PyGeopack-1.2.2.tar.gz --user

NOTE: You should uninstall any previous versions before installing this. If you had a version installed before 0.0.12 - you will need to remove the old shared object files - they are likely to be contained somewhere like (depending on the Python version used):

~/.local/lib/python3.10/site-packages/PyGeopack/

It's best just to remove everything within that folder!

2.2 Windows

Install TDM-GCC (make sure to select g++ and gfortran), install anaconda and then install the package in an anaconda power shell session using pip3 as in the Linux instructions. The same environment variables will need to be set as those in the Linux instructions. This package has only briefly been tested in Windows 10 x64, mileage may vary with other versions of Windows. This code has also been successfully built using the Strawberry Perl compilers.

It appears that ctypes doesn't necessarily search the directories listed within the $PATH environment variable for the C++ dependencies on some versions of Python. To hack myself around this, I have added a function which with scan all of the directories within $PATH for libstdc++*.dll; if found then the paths which it is found in will be added using os.add_dll_directory(). If the directory of your compiler's libraries is not listed within $PATH, this may be the cause of any failures at build time or importing the library. If you have multiple compilers installed, then I am unsure what will happen! Feel free to post an issue if you encounter any errors, or if you have any suggestions for fixes.

2.3 Mac

This module should work on MacOS as it does on Linux, with the same requirements. I have tested it using MacOS Big Sur.

3 Post-install

After installation, the PyGeopack module will attempt to locate the OMNI data required for the models. If these data exist already in $GEOPACK_PATH then it will load into memory. If they don't exist, then the user will be shown a warning - the next section explains how to fix this.

4. Usage

There are three main uses for this Python package:

1. Calculating the model magnetic field at any given point in space.
2. Tracing along the magnetic field.
3. Coordinate conversions (usually for the purposes of 1 and 2).

Before doing any of the above, it's recommended that you grab the up to date omni parameters - the UpdateParameters routine will download and update the Kp index and OMNI parameters, then calculate the G and W parameters required for the models:

import PyGeopack as gp
gp.Params.UpdateParameters(SkipWParameters=True)

The SkipWParameters keyword (set to True by default) can be used to skip the lengthy process of calculating the six W parameters for the TS05 magnetic field model - if True then these will be filled with zeros. Apparently they aren't all that important anyway. The code included in this module can calculate them and is Tsyganenko's own code, but it produces different numbers to those given in the files on Tsyganenko's website! No idea why, so use them with caution!

If the above parameters are not loaded, a warning will appear when the module is imported. To suppress warnings, set:

export GEOPACK_NOWARN=1

4.1 Calculating the model field.

To calculate the model field, use the ModelField function:

Bx,By,Bz = gp.ModelField(x,y,z,Date,ut,Model='T96',CoordIn='GSM',CoordOut='GSM',**kwargs)

where the output field components Bx, By and Bz are in units of nT.

The ModelField function accepts the following arguments and keywords:

4.2 Tracing the magnetic field

The TraceField object will accept either one argument or 5 arguements, alongside a range of keyword arguments:

TraceField(*args,**kwargs)

Trace along the magnetic field from 1 or more starting positions in the magnetosphere x, y and z using the TraceField object, e.g.:

T = gp.TraceField(x,y,z,Date,ut)

where x, y, z, Date and ut are defined in the previous section.

Or loading from file:

T = gp.TraceField(filename)

where filename is the full path and file name of a file where a previous TraceField object was saved using TraceField.Save, e.g.:

T.Save(filename)

The keyword arguments accepted by the TraceField object include one discussed in the model parameters section and ones specific to this object:

The TraceField object, T in the above code snippet, contains the following arrays:

4.3 Coordinate conversion

Coordinate conversion code is now within the PyGeopack.Coords submodule. A simple function exists to convert between the a bunch of Cartesian coordinate systems:

x1,y1,z1 = gp.Coords.ConvCoords(x0,y0,z0,Date,ut,CoordIn,CoordOut,V=V)

where x0, y0 and z0 are either scalars or arrays of positions to be transformed. Dateis an integer in the format yyyymmdd, ut is the time in hours (i.e. ut = hours + minutes/60.0) and V can be used to set a custom solar wind velocity. CoordIn and CoordOut are strings defining the input and output systems, which can be any of the following: 'GEI'|'GEO'|'GSE'|'GSM'|'MAG'|'SM'. x1, y1 and z1 are the transformed coordinates.

There are other Cartesian coordinate conversion routines:

x1,y1,z1 = gp.Coords.GEItoGEO(x0,y0,z0,Date,ut)
x1,y1,z1 = gp.Coords.GEItoGSE(x0,y0,z0,Date,ut)
x1,y1,z1 = gp.Coords.GEItoGSM(x0,y0,z0,Date,ut)
x1,y1,z1 = gp.Coords.GEItoMAG(x0,y0,z0,Date,ut)
x1,y1,z1 = gp.Coords.GEItoSM(x0,y0,z0,Date,ut)

x1,y1,z1 = gp.Coords.GEOtoGEI(x0,y0,z0,Date,ut)
x1,y1,z1 = gp.Coords.GEOtoGEO(x0,y0,z0,Date,ut)
x1,y1,z1 = gp.Coords.GEOtoGSE(x0,y0,z0,Date,ut)
x1,y1,z1 = gp.Coords.GEOtoGSM(x0,y0,z0,Date,ut)
x1,y1,z1 = gp.Coords.GEOtoMAG(x0,y0,z0,Date,ut)

x1,y1,z1 = gp.Coords.GSEtoGEI(x0,y0,z0,Date,ut)
x1,y1,z1 = gp.Coords.GSEtoGEO(x0,y0,z0,Date,ut)
x1,y1,z1 = gp.Coords.GSEtoGSE(x0,y0,z0,Date,ut)
x1,y1,z1 = gp.Coords.GSEtoGSM(x0,y0,z0,Date,ut)
x1,y1,z1 = gp.Coords.GSEtoMAG(x0,y0,z0,Date,ut)

x1,y1,z1 = gp.Coords.GSMtoGEI(x0,y0,z0,Date,ut)
x1,y1,z1 = gp.Coords.GSMtoGEO(x0,y0,z0,Date,ut)
x1,y1,z1 = gp.Coords.GSMtoGSE(x0,y0,z0,Date,ut)
x1,y1,z1 = gp.Coords.GSMtoGSM(x0,y0,z0,Date,ut)
x1,y1,z1 = gp.Coords.GSMtoMAG(x0,y0,z0,Date,ut)

x1,y1,z1 = gp.Coords.MAGtoGEI(x0,y0,z0,Date,ut)
x1,y1,z1 = gp.Coords.MAGtoGEO(x0,y0,z0,Date,ut)
x1,y1,z1 = gp.Coords.MAGtoGSE(x0,y0,z0,Date,ut)
x1,y1,z1 = gp.Coords.MAGtoGSM(x0,y0,z0,Date,ut)
x1,y1,z1 = gp.Coords.MAGtoMAG(x0,y0,z0,Date,ut)

x1,y1,z1 = gp.Coords.SMtoGEI(x0,y0,z0,Date,ut)
x1,y1,z1 = gp.Coords.SMtoGEO(x0,y0,z0,Date,ut)
x1,y1,z1 = gp.Coords.SMtoGSE(x0,y0,z0,Date,ut)
x1,y1,z1 = gp.Coords.SMtoGSM(x0,y0,z0,Date,ut)
x1,y1,z1 = gp.Coords.SMtoMAG(x0,y0,z0,Date,ut)

Also included are the following routines:

MLon,MLat = gp.Coords.GEOtoMAG(Lat,Lon,Date,ut)
Lon,Lat = gp.Coords.MAGtoGEO(MLat,MLon,Date,ut)

which convert between geographic (Lat and Lon) and magnetic (MLat and MLon) latitude and longitudes.

And for converting between magnetic longitude (MLon) and magnetic local time (MLT):

MLT = gp.Coords.MLONtoMLT(MLon,Date,ut)
MLon = gp.Coords.MLTtoMLON(MLT,Date,ut)

Descriptions of the coordinate systems:

NOTE: By "fixed", I mean that they do not rotate with the Earth's spin, they are not really fixed.

4.4 Model Parameters

In this section, the parameters and relevant **kwargs are discussed for each model. If **kwargs aren't used, then the relevant parameters are found autmoatically for the date and time provided when using the models. Individual parameters amy be altered without affecting the others - if only a single parameter is changed, then the others are still automatically calculated. The **kwargs accepted by ModelField and TraceField are:

A fucntion exists which will return a dict object containing the parameters for a given model, date and time:

params = gp.Params.GetModelParams(Date,ut,Model)

We can also return the dipole tilt:

tilt = gp.Params.GetDipoleDilt(Date,ut)

All models can be affected by the Vx, Vy, and Vz parameters as these are used to aberrate the coordinates into the GSW frame, where GSW is equivalent to GSM in the situation where Vy=0 and Vz=0. tilt is calculated automatically for all models based on the date, time and the IGRF magnetic field model.

T89

The only parameter used here is iopt which can be controlled with either setting iopt or Kp keywords to an integer. Valid values for iopt are integers in the range 1-7, if Kp is set, then iopt is set automatically equal to Kp+1. For Kp>=6 iopt=7.

T96

The first four elements of the parmod array are used for this model where parmod[0] is the dynamic pressure, parmod[1] is the SymH, parmod[2] is the y component of the interplanetary magnetic field (IMF) and parmod[3] is the z component of the IMF. All other elements of this array are ignored. The entire parmod array can be set using the parmod keyword, otherwise individual elements can be edited using the Pdyn, SymH, By and Bz keywords, where other unchanged parameters will be calculated automatically.

T01

This model uses the first six elements of the parmod array, where the first four are set in exactly the same way as in the T96 model. parmod[4] and parmod[5] correspond to the G1 and G2 parameters calculated in Tsyganenko, 2002b. These can, I believe, be set to 0.

TS05

This model uses all of the parmod array, where the first four are as in the T96 model. The last 6 elements are the W1-W6 parameters described in Tsyganenko and Sitnov, 2005.

References

1. N.A. Tsyganenko, A Magnetospheric Magnetic Field Model with a Warped Tail Current Sheet, Planet. Space Sci. 37, 5-20, 1989.
2. N.A. Tsyganenko, Modeling the Earth's Magnetospheric Magnetic Field Confined Within a Realistic Magnetopause, J.Geophys.Res., 100, 5599-5612, 1995.
3. N.A. Tsyganenko and D.P. Stern, Modeling the Global Magnetic Field of the Large-Scale Birkeland Current Systems, J. Geophys.Res., 101, 27187-27198, 1996.
4. N.A. Tsyganenko, A model of the near magnetosphere with a dawn-dusk asymmetry - 1. Mathematical Structure, J. Geophys.Res., 107, A8, 10.1029/2001JA000219, 2002.
5. N.A. Tsyganenko, A model of the near magnetosphere with a dawn-dusk asymmetry - 2. Parameterization and fitting to observations, J. Geophys.Res., 107, A7, 10.1029/2001JA000220, 2002.
6. N.A. Tsyganenko and M. I. Sitnov, Modeling the dynamics of the inner magnetosphere during strong geomagnetic storms, J. Geophys.Res., 110, A3, 10.1029/2004JA010798, 2005.
7. Singer, H. & Southwood, D. & Walker, Raymond & Kivelson, M.. (1981). Alfven wave resonances in a realistic magnetospheric magnetic field geometry. Journal of Geophysical Research. 86. 4589-4596. 10.1029/JA086iA06p04589.