Introduction

On 2012/02/21 I decided to compare recordings of the local Mw=3.8 earthquake of 2012/02/21 08:58:43 on the SLU stations of the NM network. The comparison was made in order to test instrument performance and also to check the stored metadata. The presentation on the web page 2012_02_21_09_58_43 consists of an overlay of ground velocity (high pass filtered at 0.05) Hz obtained by deconvolving all broadband and/or accelerometer channels for each station and component. If the instruments work correctly and if the metadata is correct, then all> the ground motions must be the same.

After an analysis of the SLU stations, I decided to look at the CERI stations. I applied the following steps:

Metadata evaluation

To assess the responses from the dataless seed, the standard IRIS tools rdseed and evalresp are used. The dataless seed is unpacked using the command 
rdseed -f CERI.all.seed -p -R
where the -p flag indicates that Sac polezero files should be created and the -R indicates that the RESP file should be created. The RESP file is used with evalresp and contains the complete system specification including sensor, analog filters, analog-to-digical conversion and digital filters.

The SLU moodification of rdseed outputs extra information as the RESP files are created. Focusing on the HALT HHZ and HNZ channels, the following error messages appear: 

Setting the number of zeros to add to 0
Warning - couldn't find the abbrevation for the signal units code! Signal units code =0
For station: HALT; channel: HHZ

Setting the number of zeros to add to 0
Warning - couldn't find the abbrevation for the signal units code! Signal units code =0
For station: HALT; channel: HNZ
which indicates that something is not up to standard.

In the comparisons that follow, displacement sensitivities are computed using the polezero file, mdget.HHZ.pz obtained using mdget, the polezero file rdseed.HHZ.pz obtained using rdseed and the RESP file, rdseed.HHZ.resp obtained using rdseed and evalresp which gives the amplitude and phase resposne as a fucntion of frequency. This was also done for the accelerometer channel, HNZ.

All output plots are for COUNTS output and METERS input.

The mdget.HHZ.pz was adjust for the fact that mdget returned the response in terms of COUTS output for micrometer (UM) input.

The plots below compare the responses.

Evaluation of HALT HHZ response

Evaluation of HALT HNZ response

Discussion

The obvious difference is that the rdseed.HHZ.pz and rdseed.HNZ.pz do not agree with the other response specifications. The mdget.HHZ.pz response differs at high frequencies from the rdseed.HHZ.resp because evalresp includes the effect of the digital FIR filters.

The major problem is that the polezero file obtained using rdseed is actually the velocity sensitivity for the HHZ channel and the acceleration sensitivity for the HNZ channel, which is seen by the equivalence of the responses at a frequency of 0.16 Hz (ω = 1 ). This may be related to the rdseed error message indicated above.

The mdget.HHZ.pz files is appropriate for an accelerometer since the displacement sensitivity varies at high frequency as f 1.

The mdget.HNZ.pz files is appropriate for an accelerometer since the displacement sensitivity varies at high frequency as f 2.

Overlay

Using the polezero file obtained using mdget the HHZ and HNZ recordings were deconvolved to ground velocity (m/s) using the mdget polezero files. The result is shown on the enxt plot:

The agreement is excellent.

Conclusions