Pick aggregation and filtering

Automatic pickers often produce several picks of the same phase at one station within a short time, for example when several pickers run in parallel or a picker triggers repeatedly on one arrival. Such duplicates distort nucleation, because one arrival counts several times, and they confuse the locator, which sees two different times for the same station and phase.

Aggregation reduces each group of duplicate picks to one pick. Pick filters hide picks that should not take part in nucleation.

How picks are grouped

Picks are grouped per station and phase. Within one station and phase, the picks are sorted by time, and a gap larger than Dup. win starts a new group. Each group becomes one pick.

With the profile key aggAcrossPhases = true, picks of different phases at one station are grouped together.

Press Aggregate in Pick processing → Aggregation to run it. With dbAggregate = true in the profile, or aggregate ticked in the database bar, the picks are aggregated automatically after every load.

Modes

Generate virtual pick

One new virtual pick is created per group, and the original picks are kept and tagged. The virtual pick gets:

• a time chosen by Time: the first pick, the 5th percentile, the median, the 95th percentile or the last pick of the group. The percentiles are robust against a single outlier.

• the measurements ticked under Carry:

  • Backazimuth: the circular mean over the group,

  • Slowness: the mean horizontal slowness,

  • Onset: the majority onset (impulsive or emergent),

  • Polarity: the majority polarity.

Virtual picks have a methodID starting with aggregated, and the pick table can show them alone or hide them (see Pick filters).

Virtual picks exist only in catalyst. When origins are sent to the messaging system, arrivals that refer to virtual picks are removed first, because the rest of SeisComP does not know these picks.

With Consider virtual picks in the Relocate group, the locator also sees the source picks of each virtual pick, as arrivals with weight 0: they are shown but do not pull the solution.

Keep preferred pick

One real pick per group is kept, and the others are deleted from the working set. This cannot be undone in catalyst; reload the data to get the picks back.

The pick to keep is chosen by the priority list under Advanced parameters. The rules are compared from top to bottom, and a tie falls through to the next rule:

Rule	Preference
Manual	Manual picks before automatic ones.
Status	The more vetted evaluation status.
Time uncertainty	The smaller time uncertainty.
Time	The earliest or the latest pick.
Author	A named author, for example a preferred picker.
Method	A named methodID.

The list is saved in the profile as pickPrefRules.

Pick filters

The pick filters are opened with the filter button of the Picks panel. Hidden picks are not used for nucleation, but they can still be associated to origins.

Filter	Description
Show associated / unassociated	Show the picks that belong to an origin, the picks that belong to none, or both. These are view filters only: nucleation always works on the unassociated picks.
Only the selected origin	Show only the picks of the origin selected in the origin table.
Network, Station, Phase	Comma-separated codes. Empty = any.
Station lat / lon	Show only picks from stations inside the range.
Aggregation	All picks, Aggregated only, Originals only, Not isolated or Isolated only.
Time	Show only picks within a time window.
Slow/Baz	The sectors selected in the Slow/Baz chart (see Slow/Baz), including the Is band, shown here as a summary that can be switched off.

Histogram bins

Clicking a bar of the pick histogram shows only the picks of that time bin. Ctrl+click adds or removes single bins, and Shift+click selects the range of bins from the last clicked one, so several separate time ranges can be selected. The selected bars are highlighted.

Isolated picks

A pick is isolated when too few other stations picked at about the same time. Isolated picks are usually noise. The limits are profile settings:

Profile key	Default	Meaning
pickIsoMinStations	2	Minimum number of other stations picking within the window.
pickIsoWindow	10	Half-width of the time window [s].
pickIsoOverdetectingOnly		true: only picks of over-detecting stations can count as isolated.

Infrasound trace velocity

An infrasound array pick with a slowness has a trace velocity of 111.19 km/deg divided by the slowness. A signal from the atmosphere crosses the array at about 0.30–0.45 km/s. Slower detections are usually wind noise; faster ones are seismic or coupled signals. This range is the Is band of the Slow/Baz chart (see Slow/Baz): select the band to keep only atmospheric infrasound, and right-click it to change the limits. They follow the European Infrasound Bulletin and are saved in the profile as pickIsTraceVelMin and pickIsTraceVelMax. Picks without a slowness and picks of other phases are not affected.