1. NEAREST WP3 meeting Berlin 8tth October 2008, Technische Fachhochschule Berlin Seismic data from GEOSTAR Stephen Monna and Francesco Frugoni I.N.G.V. - Rome

2. Presentation overview Data statistics Signal quality Disturbances Possibility to “cure” disturbances Data examples Conclusions

3. GURALP CMG-40T 3 component broadband seismometer (30 s-50 Hz)‏ GURALP CMG-5T 3 component broadband accelerometer, strong motion (DC-100 Hz)‏ OAS E-2PD- Hydrophone 0- 5KHz 100 Hz sampling, 7 channels 24 bit + 3 mass channels 16 bit GURALP DM-24 Digitizer, 7 channels at 24 bit + 3 auxiliary OBS mass channels at 16 bit synchronization from SERCEL rubidium clock- input in GPS DM-24 channel

4. MCU restarted day 53 (17 October 2007)‏ Number of bytes recorded/day for OBS, ACC, HYD For OBS: 2.22*10^4 over 2.34*10^4 Mbytes = 95 % of expected bytes recorded

5. Number of bytes recorded/hour for OBS, ACC, HYD Hour 20:00Hour 21:00 About 325 days of recording. 25 August 2007--> 14 July 2008

6. 21 August 2007- Port of Faro, Portugal OBS-E OBS-Z OBS-N ACC-Z ACC-N ACC-E HYD Time 200 s Recording 7 channels before deployment

7. First recording period after deployment 7 channels25 August 2007 OBS-E OBS-Z OBS-N ACC-Z ACC-N ACC-E HYD Time 200 s Hydrophone breakdown 145 s period disturbance

8. OBS-E OBS-Z OBS-N ACC-Z ACC-N ACC-E HYD 25 August 2007 First 80 s after deployment Time 10 s

9. Detail- hydrophone breakdown, note the corresponding step on ACC OBS-E OBS-Z OBS-N ACC-Z ACC-N ACC-E HYD Time 10 s 25 August 2007

10. Hydrophone- comb-style waveform starts25 August 2007 OBS-E OBS-Z OBS-N ACC-Z ACC-N ACC-E HYD OBS-E OBS-Z OBS-N ACC-Z ACC-N ACC-E HYD Time 100 s

11. Example 1 hour recording for all 7 channels OBS-E OBS-Z OBS-N ACC-Z ACC-N ACC-E HYD Time 500 s 06 March 2008

12. OBS-E OBS-Z OBS-N ACC-Z ACC-N ACC-E HYD Zoom on 145 s period disturbance Time 20 s This signal form strongly suggests leveling of the sensors is going on

13. Zoom on 145 period disturbance- onset OBS-Z Mass OBS-Z ACC-Z HYD Time 0.5 s

14. 1 hour FFT, OBS-Z component Fundamental frequency 7mHz ---> 145 s period

15. 2 time series, 1 hour long. OBS-Z component Can we “clean” the signal ?”

16. auto-correlation and cross-correlation on the 2 previous series 145 s amplitude Time AUTO CROSS

17. Example of 145 s time series which includes the disturbance, used to “clean” the signal

18. Effect of simple cleaning procedure, teleseismic event Mw6.9 N-mid atlantic ridge Onset first P-arrival 1 Hz, at 9:45:40 (no drift correction)‏ Arrival at Portuguese land station PFVI ( 10 0 km from GEOSTAR ), 9:45:47 7 s difference is in agreement with P traveltime tables for shallow events Clock drift for the whole mission is 184 ms in 359 days ---> ~ 0.51 ms/day time 10 s before after

19. time 5 s Signal cleaning example for a very local event (not present on local seismic bulletins)‏ before after Onset P arrival Bandpassed 4-20 Hz 4 poles, butterworth

20. 11 January 2008 OBS-Z OBS-N OBS-E local event (  ~ 50 km), Ml4.4, unfiltered traces Example data recorded

21. Possible explanation for the disturbances (Guralp)‏ - Solve the problem before next deployment - A posteriori cleaning of the data: how to do it ? Possilbility of running trigger algorithm for automatic event detection or, in any case we can comper to trigger by nearby OBS (MAP)‏ Possibility of using nearby OBS (same sensor ?) signal to clean GEOSTAR-OBS signal

22. - Atalante shot oceanographic cruise (info received from Grabriela Carrara CNR-ISMAR)‏ Start End Shot Signal Annotation occurrence Frequency period (s) (Hz)‏ Date Time (GMT) Date Time (GMT)‏ 25 Aug. ‘07 01:45 28 Aug. ‘07 9:46 19.4 s (50 m) 10-50 29 Aug. ‘07 03:00 09/09/2007 06:00 29.2 s (75 m) 10-40 Hz LON=6º 41.64468 'W LAT=36º 22.71906 ' N (shot 1686-línea MF22)‏