EMagPy

Python API for inversion/modelling of frequency domain electromagnetic data (FDEM).

EMagPy is divided into a Python API and a standalone graphical user interface (GUI). It aims to be a powerfull but simple tool for inverting EMI data obtain from conductimeters. EMagPy documentation can be viewed at https://hkex.gitlab.io/emagpy. Please also check the FAQ below.

Graphical User Interface (GUI)

Getting started

Clone the repository:

git clone https://gitlab.com/hkex/emagpy

Change to the src directory and run ui.py to start the GUI.

cd emgapy/src
python ui.py # this will start the GUI

The python API is available by simply importing the emagpy module from the python shell:

import emagpy
k = Problem()
k.createSurvey('examples/cover-crop/coverCrop.csv')
k.invert(forwardModel='CS') # specify the forward model (here the Cumulative Sensitivty of McNeil1980)
k.showResults() # display the section
k.showMisfit() # display predicted and observed apparent EC
k.showOne2one() # 1:1 line of misfit of apparent EC

For more example, check out the jupyter-notebook.

Downloads

as a self-extractable executable:

as a zip file:

Citing EMagPy

If you use EMagPy for you work, please cite this paper as:

McLachlan, Paul, Guillaume Blanchy, and Andrew Binley. 2020. 
‘EMagPy: Open-Source Standalone Software for Processing, Forward Modeling 
and Inversion of Electromagnetic Induction Data’.
Computers & Geosciences, August, 104561. https://doi.org/10.1016/j.cageo.2020.104561.

BibTex code:

@article{mclachlan_emagpy_2020,
	title = {{EMagPy}: open-source standalone software for processing, forward modeling and inversion of electromagnetic induction data},
	issn = {0098-3004},
	shorttitle = {{EMagPy}},
	url = {http://www.sciencedirect.com/science/article/pii/S0098300420305513},
	doi = {10.1016/j.cageo.2020.104561},
	language = {en},
	urldate = {2020-08-30},
	journal = {Computers \& Geosciences},
	author = {McLachlan, Paul and Blanchy, Guillaume and Binley, Andrew},
	month = aug,
	year = {2020},
	pages = {104561}
}

Frequently Asked Question (FAQ)

How to format the input files?

EMagPy requires input file to be formated as .csv with one header row. Headers include x, y and elevation for the locations. Other headers are formatted as <coilOrientation><coilSpacing>f<frequency>h<heightAboveGround><suffix>. e.g.: VCP0.32f30000h1 means the coils are in vertical coplanar (VCP) orientation, spaced by 0.32 meters, operated at a frequency of 300000 Hz and are at 1 meter above the ground surface.

With:

• <coilOrientation> could be: VCP (=HMD), HCP (=VMD) or PRP
• <coilSpacing> is the distance between the transmitter coil and the receiving coil expressed in meters
• <frequency> is th operating frequency in Hz
• <heightaboveGround> is the height at which the device was operated above the surface in meters
• <suffix> is optional and be set to _inph for inphase value in ppt or _quad for quadrature value in ppt

example:

x	y	elevation	VCP0.71	VCP1.18
0	0	0	34.67	38.32
1	0	0	36.69	40.16
2	0	0	35.79	39.21

What is ERT calibration and do I need it?

Electrical Resistivity Tomography (ERT) is an electrical method that measure the ground resistivity using electrodes in contact with the subsurface. Given its larger number of measurements compared to FDEM, it usually gives a better representation of the subsurface. FDEM instruments usually provide qualitative values. In order to convert these qualitative value to quantitative values, co-located ERT and FDEM measurement can be taken. FDEM measurements are then calibrated according to the ERT data.

When to convert to LIN ECa?

To the best of our knowledge, this is specific to GF-Instruments (CMD Mini-Explorer, CMD Explorer). GF Instruments used a custom linear relationship to convert quadrature values to ECa (F-Ground, F-0m or F-1m) while other instruments usually rely on the Low Induction Number (LIN) approximation (McNeil, 1980). For 'F-1m' for instance, the ECa values displayed by the GF instruments are gained by this relationship to appear closer to the actual ground EC. However, this relationship comes in the way for proper inversion as it is not physically based (while LIN is). It is then recommended that if the data comes from a GF-Instruments, such conversion is applied. While this correction has minimal effect on the F-Ground and F-0m calibrations, it has big implication for the F-1m of the CMD Explorer. More information is provided in the EMagPy paper.

Building the software

Note that you can run EMagPy fully from source. However, if you want to distribute it to other clients without needing them to install Python, it might be more convenient to bundle the app.

Bundles are build using pyinstaller (pip install pyinstaller). Different types of bundles can be produced:

• .zip bundle, the user unzip it and inside can run EMagPy.exe (Windows) or just EMagPy (linux), or extractthe .app (mac)
• .exe (Windows), no extension (linux), .dmg (mac). Windows and Linux are self-extractable archives which auto-extract themselves to a temporary directory at run-time (while the splashscreen is loading). Mac dmg image can be mounted and the .app can be drag and dropped to the Applications folder.

Building is automated by the build.sh (linux), build.bat (windows), build-app.sh (mac) scripts. An additional build-pypi.sh is supplied to build PyPi packages.

Note that you need to setup a python environment pyenv, as referenced by the scripts, which contains the necessary packages to run EMagPy (see requirements.txt and requirments-gui.txt). pyenv should be located at the same level as the repository folder. The scripts should be run from withing the src folder.

e.g.

pyenv/
resipy/
    doc/
    jupyter-notebook/
    src/
        build.sh
        build.bat
        build-app.sh
        build-pypi.sh

To build the Windows bundle, do:

cd emagpy/src
build.bat

The bundles produced will be located in emagpy/src/dist/.

Current version

v 1.1.0

Older versions

v 1.0.0

(alpha) EMagPy online based on Voila: replace tree with voila/tree once the binder url is loaded, then load ui.ipynb.