Python 3-D coordinate conversions
Pure Python (no prerequistes beyond Python itself) 3-D geographic coordinate conversions and geodesy. API similar to popular $1000 Matlab Mapping Toolbox routines for Python PyMap3D is intended for non-interactive use on massively parallel (HPC) and embedded systems.
Thanks to our contributors.
Similar toolboxes in other code languages
Prerequisites
Pymap3d is compatible with Python ≥ 3.7 including PyPy. Numpy and AstroPy are optional; algorithms from Vallado and Meeus are used if AstroPy is not present.
Install
python3 -m pip install pymap3d
or for the latest development code:
git clone https://github.com/geospace-code/pymap3d
pip install -e pymap3d
One can verify Python functionality after installation by:
pytest pymap3d -r a -v
Usage
Where consistent with the definition of the functions, all arguments may be arbitrarily shaped (scalar, N-D array).
import pymap3d as pm
x,y,z = pm.geodetic2ecef(lat,lon,alt)
az,el,range = pm.geodetic2aer(lat, lon, alt, observer_lat, observer_lon, 0)
Python argument unpacking can be used for compact function arguments with scalars or arbitrarily shaped N-D arrays:
aer = (az,el,slantrange)
obslla = (obs_lat,obs_lon,obs_alt)
lla = pm.aer2geodetic(*aer,*obslla)
where tuple lla
is comprised of scalar or N-D arrays (lat,lon,alt)
.
Example scripts are in the examples directory.
Functions
Popular mapping toolbox functions ported to Python include the following, where the source coordinate system (before the "2") is converted to the desired coordinate system:
aer2ecef aer2enu aer2geodetic aer2ned
ecef2aer ecef2enu ecef2enuv ecef2geodetic ecef2ned ecef2nedv
ecef2eci eci2ecef eci2aer aer2eci geodetic2eci eci2geodetic
enu2aer enu2ecef enu2geodetic
geodetic2aer geodetic2ecef geodetic2enu geodetic2ned
ned2aer ned2ecef ned2geodetic
azel2radec radec2azel
vreckon vdist
lookAtSpheroid
track2 departure meanm
rcurve rsphere
geod2geoc geoc2geod
Additional functions:
- loxodrome_inverse: rhumb line distance and azimuth between ellipsoid points (lat,lon) akin to Matlab
distance('rh', ...)
andazimuth('rh', ...)
- loxodrome_direct
- geodetic latitude transforms to/from: parametric, authalic, isometric, and more in pymap3d.latitude
Abbreviations:
- AER: Azimuth, Elevation, Range
- ECEF: Earth-centered, Earth-fixed
- ECI: Earth-centered Inertial using IERS via
astropy
- ENU: East North Up
- NED: North East Down
- radec: right ascension, declination
array vs scalar
Use of pymap3d on embedded systems or other streaming data applications often deal with scalar position data. These data are handled efficiently with the Python math stdlib module. Vector data can be handled via list comprehension.
Those needing multidimensional data with SIMD and other Numpy and/or PyPy accelerated performance can do so automatically by installing Numpy. pymap3d seamlessly falls back to Python's math module if Numpy isn't present. To keep the code clean, only scalar data can be used without Numpy. As noted above, use list comprehension if you need vector data without Numpy.
Caveats
- Atmospheric effects neglected in all functions not invoking AstroPy. Would need to update code to add these input parameters (just start a GitHub Issue to request).
- Planetary perturbations and nutation etc. not fully considered.
Notes
As compared to PyProj:
- PyMap3D does not require anything beyond pure Python for most transforms
- Astronomical conversions are done using (optional) AstroPy for established accuracy
- PyMap3D API is similar to Matlab Mapping Toolbox, while PyProj's interface is quite distinct
- PyMap3D intrinsically handles local coordinate systems such as ENU, while PyProj ENU requires some additional effort.
- PyProj is oriented towards points on the planet surface, while PyMap3D handles points on or above the planet surface equally well, particularly important for airborne vehicles and remote sensing.
AstroPy.Units.Quantity
At this time, AstroPy.Units.Quantity is not supported. Let us know if this is of interest. Impacts on performance would have to be considered before making Quantity a first-class citizen. For now, you can workaround by passing in the .value
of the variable.