Migrating sites from streams.csv to polarities.csv

[Thomas-Benson]

Found six sites marked as reversed in install/streams.csv:

BFZ | 10 | 2003-07-30T05:00:00Z | 2008-11-09T22:13:00Z
BKZ | 10 | 2004-02-11T02:00:00Z | 2013-10-09T00:00:00Z
HIZ | 10 | 2004-02-03T02:00:00Z | 2014-01-30T01:00:00Z
KHZ | 10 | 2003-08-06T00:00:00Z | 2011-02-23T04:00:00Z
ODZ | 10 | 2003-09-11T06:00:00Z | 2008-07-08T00:00:00Z
OUZ | 10 | 2003-08-26T05:00:00Z | 2010-02-24T00:00:00Z

Transfered into install/polarities.csv
Assuming Primary = true since every other entry uses it?

[elidana]

thanks @Thomas-Benson !
the changes looks pretty good to me, but I think the code might need some changes?
(@ozym to confirm)

I did the following:

• downloaded your branch/pull request, built a stationXML file for BFZ
• used the main GeoNet branch, built a stationXML file for BFZ

the dip and azimuth info do not match, i.e.

./stationxml-build -station BFZ -version 1.0 -location 10 -base /home/elidana/git/delta/ -output main.xml
./stationxml-build -station BFZ -version 1.0 -location 10 -base /home/tmp/delta/ -output thomas.xml

here's some examples of differences that I've found:

[Thomas-Benson]

@elidana I had a look at that, and it seem like the reversal flag in streams.csv is not implemented because I do not see a difference when building the stationxml with only the modifications to streams.csv. Only reversals listed in polarities.csv appear to have any effect. (if you want to test, you can use this branch: https://github.com/Thomas-Benson/delta/tree/Polarities-only-modified-streams)

If this is right, the difference we are seeing is the change in polarity being correctly applied to the stations now that they are listed in polarities.csv

[elidana]

oh, nice, good call to also test by only applying the change to streams.csv.
I also see that the current FDSN service is providing the info that match the status prior to your pull request, i.e.

[elidana@hutl24201 install]$ curl 'https://service.geonet.org.nz/fdsnws/station/1/query?network=NZ&station=BFZ&level=channel&format=text' | grep HH | grep 2003

#Network | Station | Location | Channel | Latitude | Longitude | Elevation | Depth | Azimuth | Dip | SensorDescription | Scale | ScaleFreq | ScaleUnits | SampleRate | StartTime | EndTime
NZ|BFZ|10|HHE|-40.679647|176.246245|283.000000|0.000000|90.000000|0.000000|Broadband Seismometer|629145600.000000|1.000000|m/s|100.000000|2003-07-30T05:00:00|2007-11-01T00:00:00
NZ|BFZ|10|HHN|-40.679647|176.246245|283.000000|0.000000|0.000000|0.000000|Broadband Seismometer|629145600.000000|1.000000|m/s|100.000000|2003-07-30T05:00:00|2007-11-01T00:00:00
NZ|BFZ|10|HHZ|-40.679647|176.246245|283.000000|0.000000|0.000000|-90.000000|Broadband Seismometer|629145600.000000|1.000000|m/s|100.000000|2003-07-30T05:00:00|2007-11-01T00:00:00

I have another question now.
By looking at differences, it seems to me that the reversal is applied to all 3 components now (i.e. when data are moved to polarities.csv). Is that what we want?

For BFZ, I was assuming that we only wanted to revers the vertical component. I'm not sure that's what is expected/reflecting what we know

based on stationXML definition for Dip

Dip of the component in degrees, positive is down from horizontal. For horizontal dip=0, for vertical upwards dip=-90
and for vertical downwards dip=+90.

and if I get it right, if dip=-90 , for HHZ component , then the component is marked as reversed?

Or am I getting this totally wrong?

[salichon]

Hi there,
As pickedup by @elidana
This may need some clarification perhaps to me

• Polarities.csv "polarities" is for one single component. Mostly the Z component that is for some reason inverted (Z sign changes) - other components being correct (or not checked)
• in streams.csv "reversed" is for the entire 3 components to be inverted with regards to the configuration. (all signs are changed - instrument orientation changed from standard)

The migration therefore does not look straight forward as this

As far as i remember, the polarity correction for the Z component can be informed by the metadata ("comment" or else) in addition to the cartesian system orientation, though the Z (in general) sign correction itself is generally operated by the user.
There is room for improvement in that matter that affects instruments in many places.

[elidana]

thanks @salichon ! whatever we end up with, I think might be better if we only use one table for the reversal information (i.e. move the info out of streams). Perhaps we can add a new column in polarities.csv to indicate the vertical reversal only, or one field per component? also not 100% sure how streams.csv is currently used, might be worth a double check from expert eyes