Fresh marker with uncertainty

[HerrMuellerluedenscheid]

Improved, cleaned, simplified and fresh version of #234 (which I'm going to close right now)

Uncertainties only apply to phase markers. Select a Phase Marker, press Ctrl, Click + drag down/up.
Marker table fields (press 'm' and activate 'Uncertainty' from the right-click table header menu) receive live updates.
Let me know if you have a more natural, intuitive way in mind to adjust uncertainties.

[d-chambers]

I may have found a small bug. When using a windowed phase marker the uncertainty window always, at a minimum, fills the entire area between the start and end marks. The actual uncertainty displayed in the marker file when saved to disk, however, is much smaller and should probably only be plotted around the start of the window. Consider the following screen shot where the window duration is about 20 seconds (and so is the plotted uncertainty) but the uncertainty saved in the marker file is less than one second.

[grajh]

Hi!

To me everything looks allright. Your "start" time is 17:13:10.7543, "end" time 17:13:33.3218 and the difference among them corresponds to 22.5675. . .

[HerrMuellerluedenscheid]

Hi, I think the numbers are correct. The visual representation might be misleading for the extended markers. One alternative to filling the entire block between tmin and tmax of the extended marker would be to draw a block only to the left of tmin and right of tmax:

(The marker on 4th trace)

But that visually detaches both ends making the marker look like two individual ones. What do you think?

[grajh]

Hi! I do not find the first representation misleading at all. I propose to leave it like it is. As far as I know symmetrical filling between tmin and tmax is standard in other picking software.

[HerrMuellerluedenscheid]

Ok, that is good to know. I do, however, understand the point of @d-chambers and it is likely that others will also fall for that. How about a compromise: Adding another block in between tmin and tmax which is lighter in color than the actual uncertainty. I personally like that option the most. Opinions? @d-chambers @grajh

Talking of which: What is 'uncertainty' called in other programs? Also uncertainty or error or something else? Any experience?

[grajh]

Hi!

Yes, that should also be ok. As long as the case on the third trace above is also available, which is in my opinion the most correct representation of a pick with uncertainty. That is the "central" pick time surrounded by its uncertainty value on both sides.

It is called uncertainty, because its values represents a certainty with which user picked the arrival time of the phase.

[d-chambers]

I like the new representation, it does make it more clear what is going on. I do agree with @grajh that the third (blue) trace is the most conventional representation (the non -extended maker with uncertainty) and will likely be the most commonly used form of the uncertainty feature.

We should discuss the meaning/use of an extended phase marker, as we may not be thinking of it the same way. My thought was to use the start as a traditional phase pick, with uncertainty, and the end to denote a change in phase type. For example, a single extended maker might be used to pick the arrival time of P energy and to delineate where the P energy ends. The window between the start and the end could then be used for fitting source spectra to calculate moment. One might also use the start and end to define a window for particle motion analysis, in which case the end would be in a different place and have a different meaning than the first case. How have you envisioned the windowed phase makers being used?

I guess what I am getting at is that, depending on the use case, it might make sense to have an uncertainty with the start of the marker and a potentially different uncertainty associated with the end of the marker. What do you think?

[grajh]

I see. It really depends on the use case. The "extended" marker denoting uncertainty was implemented in Snuffler originally, but because this way it is hard to define symmetrical uncertainty I proposed to implement a single extended marker with symmetrical uncertainty being shown as filled area. Pyrocko team implemented this just as I imagined. Again thank you all for this.

Related to your need. I would propose you to define two phases in configuration file (e.g. Pstart and Pend) and then you can assign symmetrical uncertainty to both.

[emolch]

• the format of the marker files is already a too big mess. The new format should be YAML based.
• the new feature needs documentation in the help panel and in the tutorial web page.