Quality control

Catalyst computes a set of quality metrics for every origin and shows them as columns of the origin table. On the REFINE → QC tab, conditions on these metrics are combined into rules that select, star, confirm, reject or remove origins. Rules can be saved as named rule sets, so the same checks are applied the same way on every run.

Quality metrics

Metrics marked 0 = good are badness scores between 0 (good) and 1 (bad).

Code	Metric	Description
0	Depth [km]	Hypocentre depth.
1	Phase count	Number of arrivals.
2	Azimuthal gap	Largest azimuthal gap between the stations [deg].
3	RMS [s]	Root mean square of the travel-time residuals.
4	Mismatch	Weighted fraction of the nearby stations that were available but not used by the origin. 0 = used all of them, 1 = used none. A high value points to a mislocation, unless the event is far away and the nearby stations could not see it.
5	Move-out 0-1	How badly the arrivals follow a straight move-out: the scatter of the apparent-velocity fit and an unrealistically high apparent velocity, per P and S. 0 = good.
6	DD ref ties	Number of accepted differential times in the last DD relocation.
7	DD residual [s]	Residual of those differential times.
8	DD qual	DD quality from ties and residual. 0 = good.
9	S cov	Missing S coverage: of all stations that recorded the event, the fraction without an S pick near the predicted S time. A real local event is usually seen in S at most stations; a spurious P-only coincidence is not. 0 = good.
10	Fault Q	How well the hypocentre fits the loaded fault model. 0 = good. Needs faults loaded in the 3D view.
11	Min sta dist [km]	Distance to the nearest station used.
12	Mode	Evaluation mode: 0 = automatic, 1 = manual.
13	Status	Evaluation status: 0 = preliminary, 1 = confirmed, 2 = reviewed, 3 = final, 4 = rejected, 5 = reported.
14	Zone	Number of the QC zone that contains the epicentre. 0 = outside all zones.
15	Clean Q	Weakness of the pick evidence. Each arrival at a station that is not over-detecting counts 1/(1 + d/100 km), where d is the station distance; four or more such arrivals count as good. 0 = good.
16	Score	Origin score of the scevent OriginMultiFeature score processor. Higher is better.

Mode and Status use the SeisComP enumeration values, so they are best tested with = or ≠, for example Status = 4 for rejected origins.

The code is the number used for the metric in saved rule sets and in the profile.

Score

The score is computed automatically for every origin when the scoremf plugin is loaded (see Setup and configuration). Calculate score re-reads the score configuration and recomputes all scores, for example after the score.mf.* parameters were changed.

Conditions and rules

A rule is a chain of conditions. Each condition compares one metric with a threshold, using <, ≤, >, ≥, = or ≠.

• + condition adds a condition.

• The AND/OR box joins a condition to the one above. AND binds more tightly than OR.

• The indent buttons group conditions in brackets. Consecutive indented conditions form one bracket, joined to the condition above by the AND/OR of the first indented one. For example, A, then AND indented B, then OR indented C, reads A AND (B OR C). Up to four levels are possible.

• Clear conditions removes all conditions.

Match shows how many visible origins fulfil the rule. It updates when the conditions, the origin filter or the origins change.

Only origins visible in the origin table are tested. Use the origin filter to limit a rule to a subset.

Actions

Button	Effect on the matching origins
Select matching	Select them in the origin table.
Star matching	Add a star. Clear stars removes all stars.
Confirm matching	Set the evaluation status to confirmed and the evaluation mode to manual.
Reject matching	Set the evaluation status to rejected. The picks are kept. Relocating or accepting an origin clears the rejection.
Remove matching	Delete them from the working set. This cannot be undone; check them with Select matching or Star matching first.

Reject next to mismatch > rejects every visible origin whose mismatch is above the threshold. Clear origins removes all origins after a confirmation; the picks are kept.

Rule sets

A rule set is a saved rule with a name.

• Select a rule set in Rule set to load its conditions.

• + Save as… saves the current conditions under a new name.

• Delete removes the selected rule set. Origins are not changed.

• (custom) is a scratch rule that is not saved.

Rule sets are stored in the current processing profile (key qcProfiles) when the profile is saved, so each network or region can have its own checks.

Example rule set that flags poorly constrained origins in zone 2, with the second and third conditions indented one level:

     Zone           =  2
AND      Azimuthal gap  >  180
OR       Phase count    <  6

It reads Zone = 2 AND (Azimuthal gap > 180 OR Phase count < 6). Without the indentation it would read (Zone = 2 AND Azimuthal gap > 180) OR Phase count < 6, because AND binds more tightly than OR.

QC zones

Zones divide the region into areas with different quality expectations, for example the network interior, the coast and offshore areas with poor coverage. Load zones (GeoJSON)… loads a GeoJSON FeatureCollection of polygons:

• the integer property zone of each feature is the value of the Zone metric,

• the property name is shown in the zone legend.

{
  "type": "FeatureCollection",
  "features": [
    {
      "type": "Feature",
      "properties": { "zone": 1, "name": "Network interior" },
      "geometry": {
        "type": "Polygon",
        "coordinates": [[[-22.0, 63.8], [-18.0, 63.8], [-18.0, 66.0],
                         [-22.0, 66.0], [-22.0, 63.8]]]
      }
    }
  ]
}

Loading assigns a zone to all origins at once. Origins loaded or relocated later get their zone automatically. The path of the zone file is saved in the profile (qcZones), and the zones are loaded with the profile.