Regarding the azimuth of major semi-axis

[Matif009]

In my case, in the twistpy code, the azimuth will divide in two parts, such as (1) azi1 =[0,30] (2) azi1=[170,180]

So, is this possible during filtering we can provide both cases of azimuth together, for ex either we can do->

data_filtered_majorazimuth = analysis.filter(
plot_filtered_attributes=True,dop=[.8,1],elli=[.2,1],azi1=[0,30],inc1=[20,90]
)
plt.show()

or, we can do->

data_filtered_majorazimuth = analysis.filter(
plot_filtered_attributes=True,dop=[.8,1],elli=[.2,1],azi1=[170,180],inc1=[20,90]
)
plt.show()

So, is this possible we can filter both cases of azimuth together and then plot? Please help asap. Thank you

[solldavid]

Hi @Matif009

It is true that there is a bit of a shortcoming with the current filtering implementation for filters that wrap around from 360 to 0 degrees. I am currently not further developing twistpy, but I could think of a simple workaround for your application that would allow you to filter your data with the current version of the code as desired:

1. Rotate your data into a new coordinate frame by rotating around the vertical axis (e.g., by 90 degrees),
2. Filter the data in the rotated coordinate frame.
3. Finally, rotate back into the original coordinate frame.

Admittedly not a super clean solution but this should give you the desired output.

[Matif009]

Dear @solldavid,

Thank you for your response and for developing the TwistPy tool.

Although I understand that you are no longer actively developing it, I’d like to offer a small suggestion that could further enhance its applicability. Based on my experience using your code, I believe that refining the azimuth calculation to output values from 0° to 360° measured clockwise from North (station to source) could significantly expand its utility—not only for earthquake pulse identification but also for applications in fluvial seismology. In long riverine systems, where pinpointing the directional source of sediment transport or turbulence is critical, such an enhancement would be invaluable.

While exploring your code, I noticed that the azimuth calculation could be improved by modifying a specific section. Here’s what I suggest for TimeFrequencyAnalysis3C:

#Remove this block:
self.azi1[self.azi1 < 0] += 180
self.azi1[self.azi1 > 180] -= 180

#Replace with this logic:
if self.azi1 < 0:
self.azi1 = (np.degrees(2 * np.pi)) - self.azi1

With these changes, the code would provide the azimuth of the major semi-axis in a full 0–360° range from North in the clockwise direction. This is a standard convention in many geophysical and hydrological studies.

Following that, the filtering logic should also be updated to properly handle azimuth ranges spanning 0–360°. These enhancements would make TwistPy a more versatile tool for polarization analysis across multiple disciplines.

Thanks again for your great work on TwistPy. I’ve really enjoyed exploring the code!

[solldavid]

Dear @Matif009

Please note that in 3C polarization analysis, there will always be a 180-degree ambiguity in the azimuth. The major semi-axis can indicate two directions of propagation and, using just a single 3C station, it is not possible to determine whether a wave propagates in the positive or negative direction of the major semi axis. Hopefully, this explains why the logic above is in place.

The ambiguity can only be resolved when using an array of 3C sensors (or a single 6C station that also records rotational motion as demonstrated in other tutorials in TwistPy).

Best,

David