1.1.1 A new Nordic Format, Nordic2

The Nordic2 format for parametric data is implemented from version 12 to overcome limitations with the previous Nordic format. The old Nordic format does not included network and location codes that are part of the data stream identification. More and more station use these codes and two different stations (digitizer and/or seismometer) can have the same station code and then have to be differentiated by network code and possibly location code. A significant weakness with the old Nordic format is that component codes are 2 letters, which in the new format (Nordic2) are 3. Another problem when data is merged from different sources is to know where the data come from. We have therefore, for each phase, included the possibility to include both agency and author of the phase.

To see how to migrate from Nordic to Nordic2 format, see end of this section.

The philosophy of the new format is that each type of observation used in S-files (phase, amplitude, coda, back azimuth) must occupy one line with a corresponding phase. This has been an optional case also before for amplitudes and back azimuth, now it is a must. The amplitude phase has the same name as before like IAML, the back azimuth (and apparent velocity) are BAZ, possibly with a phase name like BAZ-P to indicate the phase. The coda is END (IASPEI standard). Below is an example, the Appendix has the full description.

STAT COM NTLO IPHASE   W HHMM SS.SSS   PAR1  PAR2 AGA OPE  AIN  RES W  DIS CAZ7
BER  SHZ NS00 IP         1325 38.120      C       BER lo      -.9801061.00  11
ASK  SHZ NS00 ES         1325 48.070              BER jh      -1.021071.10   3 
ASK  SHZ NS00  AMP       1325 48.070  333.3  2.20                    71.10   3 
ASK  SHZ NS00  IAML      1325 50.900  111.0  0.30                    71.10   3 
NRA0 SHZ BB01 IPn       A1326 19.090              NAO hhh 50.0-.05010368.0  72 
NRA0 S Z       END       1326 19.090  333.0                     0.2  368.0  72 
NRA0 S Z       BAZ-P     1326 19.090  256.9   6.9               0.1  368.0  72 
MUD  BHZ DK00 IP         1326 38.120      C       DNK pv      -.9801061.00  11
MUD  BHE DK00 ES         1326 48.070              DNK pv      -1.021071.10   3

The new columns are NT and LO for network and location respectively. AGA and OPE are agency and operators. PAR1 and PAR2 are the observation associated with the phase which are:

Seismic phase: polarity
Amplitude: amplitude and period
END: coda length
BAZ: back azimuth and apparent velocity

The residual (RES) is now residual for the observation if any, degree for BAZ, seconds for phase, magnitude for amplitude and coda.

The user can select the format to use in SEISAN.DEF, parameter NORDIC_FORMAT. The default, for now, in the distribution is that the format is set to the old Nordic format (0) and everything will work as before. Setting to 1 old and new format can be read but new will be written (also in programs such as SLEECT, SPLIT, ...) and setting to 2 only new for a can be used. It is recommended to use 0 or 2 to avoid mixing the format in one database, but 1 is needed for example to compare data in Nordic and Nordic2 format.

SEISAN can be hardcoded to work with new nordic format only; in LIB/libsei.for uncomment the two lines given at the end of the textttget_seisan_def subroutine and recmpile SEISAN.

It is possible to convert from new to old and vise versa by program NOR2NOR2.

The majority of programs have been changed to be able to work with the new format, if not it will be noted.

There are several programs that work with the SEISAN 4 character component names, particularly conversion programs. These have not been changed to 3 character components since they will also work with the 3 character components, leaving the 3rd character blank. When converting from 2 character to 3 character component codes, the 2nd character is left blank. Also many programs can only work with station and component and will not, at the moment, be converted to also use network and locations (e.g. CODAQ).

Old to new components, NT and LO in MULPLT
When old data is transferred to new format there is only component e.g. S Z and no location and network code. If the original data has this information, MULPLT is still able to plot the phases since:

-  if component code 2 is blank, no check for 2nd component charcter.
-  check for location and network is disabled if they are both blank.

If new phases are read and NT or LO are present, the component code must then have 3 characters.

Response files. The SEISAN naming of response files does not have NTLO so they cannot distinguish same station names and different NTLO. This will probably also be fixed but for now the problem can be solved by using RESP response files.

As the Nordic2 format allows more complete mapping between the data streams and observations, it is recommended to convert complete databases. However, it is planned to continue to distribute SEISAN, which is able to use the old format to be backward compatible.

To convert from Nordic to Nordic2, the user should carefully check that no significant changes are introduced. This can be checked by comparing locations and magnitudes between the orignal Nordic and recomputed results after conversion to Nordic2 format. An additional complication is changes in HYPOCENTER, especially compared to quite old distributions. The user should carefully consider what needs to be checked, the following list provides a suggestion:

-  make sure you have backup of the original database
-  keep NORDIC_FORMAT set to 0, and use SELECT or COLLECT to create Nordic file
-  check for changes because of HYPOCENTER, by relocating, then use program COMPARE_HYP or COMP_HYP to check for changes; if changes are significant this needs to be investigated before moving on
-  next convert file from Nordic to Nordic2, using program NOR2NOR2, check with COMPARE_HYP or COMP_HYP
-  set NORDIC_FORMAT to 1, and use HYPOCENTER to relocate
-  compare to Nordic format data using COMPARE_HYP or COMP_HYP
-  if significant changes, this needs to be investigated
-  when it is verified that there are no significant differences (unlikely in first round), the output from HYPOCENTER in Nordic2 format can be split into an empty new database (see MAKEREA), which then replaces the original database (be careful about changing database folder name when replacing)
-  set NORDIC_FORMAT to 2, which makes Nordic2 a must, Nordic will no longer be read; in case of different databases, different SEISAN.DEF files can be used; NORDIC_FORMAT=1 is not recommended

Peter Voss : Tue Jun 8 13:38:42 UTC 2021