0. sacplot.py -d -C myevent.pkl

1. Arbitrary event selection: SNEP 2007.150.20.22.12, data are of mixed quality

2. Zoom-to-rectangle button and function

3. Arrow button goes back to previous view

4. Unselect the zoom-to-rectangle function

5. Click and drag to select the red window – let go to zoom

6. Press z to go back to previous view - and repeat 5 & 6

7. Select the matplotlib span selector function

8. Click left to fix point, click right to zoom in x and y direction

9. Click and drag to move view

10. Lastly, click back in steps through previous views.

1. (1)Plot SAC files (sacplot.py)

(2) Pick SAC time headers manually (sacppk.py)

(3) Pick SAC time headers automatically (ttpick.py)

0. sacppk.py -r0 myevent.pkl

1. Click on "bad" traces to deselect them

2. Page to the next page of seismograms

3. Zoom by (left) click and drag to inspect details of the waveforms and to make sure that the P wave indeed dominates the -15 to 15 s window around T0 (used for initial, coarse alignment of traces through cross correlation)

4. Press "z" to unzoom

5. Go back to previous page

6. Repeat zoom by "click and drag" and unzoom by "z". Be very careful to click in between traces because otherwise you could accidentally deselect a trace

7. Save the selections!

0. sacppk.py -r0 myevent.pkl

1. Page through the - now three - pages to see how the traces are sorted: page 0 has "bad" and good traces; page 1 has good traces, and page -1 has "bad" traces.

2. Use the span selector to zoom to remaining data mysteries. Click the left mouse button to fix a point in the plot and zoom in both x and y directions with the right mouse button. Left click and drag will translate the traces.

3. Adjust your trace selections, in this case we are deselecting one additional trace.

4. Save.

0. ttpick.py -pP -s1 myevent.pkl

1. Launching ttpick as above - with the default [-15, 15] s time window around t0 - will start with an initial cross correlation for coarse alignment of the P waves.

    The time shifts needed for this alignment are stored in headers T1. Also, an Array Stack is calculated.

2 You can do more trace de-selection or re-selection by clicking the traces.

3. Press ICCS-A to update the Array Stack and sort order of the traces.

4. Zoom in -- in this case I am using the "zoom to rectangle" function. Don't forget to unclick the zoom button.

5. Pick the absolute arrival of the P wave (the onset) by pressing keys "t", then "2". This is stored in variable T2, and marked by a red line.

6. Press Sync(hronize) to shift all T1 markers in all (good) traces by a constant amount (derived from your T2 pick in the Array Stack) to T2 markers at the onset.

7. Zoom in more if necessary - in this case by click and drag - to set the new time window for the fine cross correlation by pressing "w". Do this very carefully.  

0. ttpick.py -pP -r3 -saz myevent.pkl

1. Note how the seismograms are nicely aligned (by the best pick (T3, -r3)) - they are sorted by azimuth (-saz)

2. Zoom in by click n Drag. (note that red T2 is beneath green T3)

3. Use pan & zoom to scroll through the arrivals.

4. computer had a sticky mouse, and is slow because it was being backed up at the time.

You want to capture the onset of the P wave and just enough of its waveform for it to be characteristic, but not too much so the geometrical ray approximation remains largely valid.

8. Press Sync to copy this narrower time window to all traces. Note how predicted arrival times (black lines, T0) are all over the place. The next instance of ttpick we will align the traces by delay time (-r3), so we get an idea of the mean and variance of the delays, and their regional characteristics (if we are able to affiliate stations names with geography).

9. Zoom out, in this case by pressing the back arrow button, to verify that the time window was copied correctly.

10. Press ICCS-B, for the actual, fine cross correlation. You may re-pick the absolute arrival in the array stack, then rerun the ICCS-B, to iterate and converge onto robust delay times.

11. Zoom in – the zoom to rectangle function precludes accidentally deselecting a trace - and examine the alignment. Make adjustments in trace selection and correlation window, running ICCS-B after each adjustment, until you are happy with the waveform alignment.

12. Note how the "back of the camel bumps" is slanted compared to the onset of the camel bumps (P wave), which is a vertical lead line, illustrating the complexities involved in the results of cross correlation for P wave forms longer than the onset.

13. Press MCCC: this runs a final cross correlation, between each and all traces. This outputs a *.mcp file with arrival time differences for all traces. The arrival times are relative to one central, absolute time.  It is up to the user to difference these times with those expected from a reference model.

14. Save all work.