Comparison of regional seismic phases interpretation in REB and bulletins. Zlata I. Sinyova, Natalya N. Mikhailova Institute of Geophysical Research, Almaty, Kazakhstan
Abstracts. Three seismic arrays (MKAR, KURK, BVAR) and one three-component seismic station (AKTO), installed in Kazakhstan are included into IMS network. These four stations are participating both in REB compilation, and in a compilation of regional bulletin. In the work we analyzed the interpretation of the regional seismic arrivals, which were picked in both bulletins (Pn, Pg, Sn, Lg phases) by the following parameters: phase identification, difference in arrival time picks, difference in amplitudes. The following conclusions were made. The vast majority of phase names coincide in both bulletins. As a rule, Pn, Pg, Sn phases in REB bulletin are picked earlier, than in bulletin (in average on 0.6 sec), whereas Lg phase in REB bulletin is picked later, than in bulletin (in average on 0.88 sec). For some events the difference in arrival times of the same phases in REB and in bulletins reaches 8 and more seconds. Several such events were analyzed in more details. We observed decrease of the difference in arrival times with the increase of amplitude of signals. For Pg phase recorded on KURK array the correlation between distance to the event and difference in arrival times was observed. Measured amplitudes of seismic signals in bulletin are systematically high, than that in the REB bulletin. It can be explained by the difference in filters that are used during amplitude measurement in IDC and in.
Introduction. The correct determination of the arrival times and correct phase identification plays vital role during seismic events processing. Kazakhstan operates four IMS seismic stations. All this stations are actively used as in compilation of IDC REB bulletin, as well as in compilation of regional interactive bulletin. Three of these seismic stations are arrays. MKAR and BVAR arrays both have similar configuration, with aperture of 4-3 km, whereas KURK is a teleseismic array with an aperture of 22.5. The fourth one is three-component station AKTO. In the study we have made association of common arrivals from regional phases (Pn, Pg, Lg, Sn) from both bulletins, analyzed the difference in their arrival times, difference in amplitude measurements. Such approach allowed us to reveal an outliers, i.e., cases, when arrival times of the same phases in REB and bulletins are significantly differs. For some of such events we tried to find an explanation in the study. AKTO BVAR KURK MKAR Map of seismic events detected, processed and included into interactive bulletin during 2009 (20 340 events). Red stars IGR NNC seismic stations. In blue circles IMS stations in Kazakhstan: MKAR (PS23) Makanchi primary array; BVAR (AS57) Borovoye auxiliary seismic array; KURK (AS58) Kurchatov auxiliary seismic array; AKTO (AS59) Aktiubinsk auxiliary seismic 3-C station;
PS-23 (MKAR) primary Makanchi array. 15 April, 2009 Difference in arrival times of for MKAR array in dependence on distance. Difference in arrival times of for MKAR array in dependence on amplitude MKAR is a primary IMS array, located in a region with rather high seismic activity. In interactive bulletin for the 2009 there was 6683 Pn arrivals, detected by MKAR array. REB bulletin for the 2009 contains 970 Pn arrivals on MKAR. And 499 of them are included both into REB and bulletins. On the pictures above we plotted the difference in time picks of the same arrivals of in REB bulletin and in bulletin in dependence on distance to the event (on the left) and in dependence on the amplitude of the event (on the right). Amplitude measurements were taken from bulletin. As it could be expected, we see that with increase of the distance to the event or with decrease of the amplitudes, the spread in time picks became significantly bigger. More often analysts put Pn phase too late, if compare with IDC analysts. However, the opposite situations are also observed. One such event (marked in a circle on a pictures) was analyzed in more details.
Event of 15 April, 2009, 04:56:10, mb 3.3, Ml 3.0. For this event analyst picked two phases: Pn and Sn, whereas IDC analyst used only one phase Pn, picked about 9 seconds later. REB solution for this event has the following coordinates: lat 33.5480, lon 82.1231, solution is: lat 40.3051, lon 82.0472. These two solutions lie in about 750 km from each other. Date Time Latitude Longitude 2009/04/15 04:54:49.00 33.3531 81.8380 Sta Dist EvAz Phase Time TRes Azim AzRes MK31 13.44 1.3 Pn 04:57:50.371-10.4 181.0-0.6 MK31 13.44 1.3 Sn 04:59:08.955-83.1 181.9 0.2 MKAR 13.44 1.3 Pn 04:57:59.389-1.4 187.8 6.2 ZALV 20.70 5.0 P 04:59:29.914-0.6 186.0-1.1 CMAR 21.35 129.9 P 04:59:37.483-0.1 310.6-6.8 SONM 23.44 44.7 P 04:59:58.434-1.8 230.9-10.0 TORD 74.67 275.8 P 05:06:33.650 2.4 33.2-25.5 ASAR 75.54 131.6 P 05:06:35.330 1.0 304.1-12.9 YKA 83.24 7.6 P 05:07:16.759-1.9 336.5-9.7 Defining phases from REB bulletin (in bold) Date Time Latitude Longitude 2009/04/15 04:56:10.30 40.0629 82.0391 Sta Dist EvAz Phase Time TRes Azim AzRes MK31 6.73 1.5 Pn 04:57:50.371-0.0 181.0-0.6 MK31 6.73 1.5 Sn 04:59:08.955-0.0 181.9 0.2 MKAR 6.73 1.5 Pn 04:57:59.389 9.0 187.8 6.2 ZALV 14.01 6.8 P 04:59:29.914-0.1 186.0-2.8 SONM 19.08 58.0 P 04:59:58.434-36.5 230.9-24.1 CMAR 26.05 141.0 P 04:59:37.483-127.5 310.6-18.8 TORD 74.25 274.1 P 05:06:33.650-76.3 33.2-18.6 YKA 76.57 7.8 P 05:07:16.759-47.3 336.5-10.5 ASAR 79.95 132.9 P 05:06:35.330-104.7 304.1-18.1 Defining phases from bulletin (in bold) REB MKAR station, Filter 2 4 Hz applied. We made relocation of this event adding phase picks (picture on the left). Only ZALV station P phase (closest to MKAR array) comes with a very good agreement with solution. To our opinion, missing Sn phase and wrong associations of P phases from distant arrays lead to a significant shift in final event coordinates.
Lg phase Lg phase Difference in arrival times of Lg phase for MKAR array in dependence on distance. We compared the difference in time picks of the same arrivals of Lg phase in dependence on distance to the event (up on the left) and in dependence on the amplitude of the event (up on the right). Amplitude measurements were taken from bulletin. Only for local events (up to 3 degrees from the station) we observe good correspondence between and REB phase picks. On further distances the big scatter in phase pick times is observed even for the strong phases. Difference in arrival times of Lg phase for MKAR array in dependence on amplitude REB Filter 1 3 Hz applied. Event on 22 July, 2009 Picture on the right shows an example of erroneously picked Lg phase by analyst. Most probably analyst used vertical channel (MK31, bz) for time pick, whereas there is clear Lg arrival on horizontal channels 10 seconds earlier.
Comparison of phase amplitudes for MKAR array Lg phase a) b) Comparison of amplitudes, measured in and amplitudes, measured in IDC. MKAR array, a) -, b) Lg phase. We compared amplitudes of Pn and Lg arrivals, measured in and in IDC (see pictures above). Amplitudes of the same arrivals, computed in are systematically higher, than amplitudes computed by IDC algorithms. Especially large difference is observed for Lg phase. This difference can be explained by difference in filters, applied to a trace prior to amplitude's measurement. Amplitudes measurement in is based on a filter, which emulates short period seismometer SKM, with a central frequency about 1 Hz. The usage of this filter is a necessary stage, because regional magnitude (MPVA) and energy class (K) were calibrated for this seismometer characteristics. Apparently, IDC procedures use filters with higher frequencies diapason. Similar difference in amplitudes was observed and for three other IMS stations.
AS057 (BVAR) auxiliary seismic array. 24 September, 2009 Difference in arrival times of for BVAR array in dependence on distance. Difference in arrival times of for BVAR array in dependence on amplitude BVAR array is located in a region with low natural seismicity. Most of seismic events, located at local distance to the array are mining explosions. In REB bulletin for the 2009 there was 402 Pn arrivals on BVAR array, in bulletin for the year 2009 1510 Pn arrivals on BVAR array. And 262 of them are common for both bulletins. As it was observed for MKAR array, at average, the difference between arrival pick times is larger for small amplitudes and for larger distances. And again, analysts pick weak signals later, than IDC analysts. On BVAR array the largest outlier was observed for the event, occurred on 24 of September, 2009. This event was analyzed in more detail.
Event on 24 of September, 2009 In REB for the 24 of September, there are two events: first one (lat 41.8479 lon 77.4953, time 7:57.08) and the second one 8 seconds later with almost the same coordinates (lat 41.9662 lon 7.3531). Magnitude mb of the first event is 4.40, of the second 4.50. Accordingly, on BVAR array there are two Pn arrivals in REB bulletin (see picture on the right). In only one event (corresponding to the first REB event) was processed and only one was picked. The situation, when we see several P arrivals is quite common for strong mountain events. For example, on the picture below, we plotted BVAR records of the event, occurred on May 22, 2011 (lat 42.0615, lon 82.1861, mb=5),which looks very similar to the event of the 24 of September, 2009. KNDC station, about 250 km. from the epicenter. Red arrow points at the predicted time of the second REB event. Second arrival of P phase was not observed on the stations, close to epicenter (see picture above).
AS058 (KURK) auxiliary array KURK is a large teleseismic array. It has 21 sites and aperture of 22.5 kilometers. For the year 2009 we found only 5 common Pg arrivals on KURK array in REB and bulletins (see picture on the right). For Pg phase on KURK array there is clear dependence of the time difference between REB and picks and distance to the event. The largest outlier (placed in a circle on the picture) is the mining explosion on Kara-Zhyra mine (about 75 kilometers from the central point of the array) detonated on the, which turned out to be a NPE09 event. Pg pick in REB bulletin corresponds to the arrival time on the closest to the mine array site (this fact was demonstrated by presenters from Norway on NDC Evaluation Workshop 2010). Pg pick time corresponds to the arrival time on the central point of the array. It is interesting, that Lg phase picks coincides in both bulletins and correspond to the arrival time on the central point of the array. KURK array configuration REB Difference in arrival times of Pg phase for KURK array in dependence on distance. Lg
Difference in arrival times of for KURK array in dependence on distance. Difference in arrival times of for KURK array in dependence on amplitude For KURK array there is 527 s in REB bulletin in 2009, 1990 s on bulletin, and only 187 phases are common for both bulletins. Picture on the right explains the reason of the largest outlier. Apparently, analyst used filter from 1 2 Hz to pick. However, if we apply filter with larger frequency diapason, the real Pn arrival will be revealed. REB Filter 1 2 Hz, REB Filter 1 4 Hz
AS059 (AKTO) auxiliary three-component station. Difference in arrival times of for AKTO in dependence on distance. There was 352 s in REB bulletin for 2009, 935 s in bulletin. 270 s are common in both bulletins. Picture in the right demonstrate the event with the largest outlier. Apparently, there was an error, made by analyst during picking. By some reason, IDC analyst didn't pick clear Sn phase for this event (may be because it is much better seen on horizontal channels and IDC analyst looked on vertical channel only?). Difference in arrival times of for AKTO in dependence on amplitude REB only Event of 23 of March, 2009. Wrong Pn time in bulletin, and missing Sn phase in REB.
Sn phase Sn phase Difference in arrival times of Sn phase for AKTO in dependence on distance. Difference in arrival times of Sn phase for AKTO in dependence on amplitude bulletin for 2009 has 750 Sn phases from AKTO station, REB 176 Sn phases, and 123 Sn phases are common for both bulletins. As it was observed yet for s, the difference between arrival pick times of Sn phases is larger for small amplitudes and for larger distances. The picture on the right shows event with large difference in Sn pick time in and REB bulletins. Apparently, IDC analyst placed Sn phase looking only on vertical channel, whereas on horizontal channels the Sn phase appears several seconds earlier. REB Event of 18 May, 2009
Conclusions. The comparison of regional seismic phases interpretation between phases in REB bulletin and interactive bulletin was made for four IMS stations of Kazakhstan. The following conclusions can be made: - as a rule, arrival times of regional phases in both bulletins are consistent, however, weak signals and signals from distant events are often picked by analysts with some delay; - the decrease of the difference in arrival times with the increase of amplitude of signals was observed; - the systematic shift in Pg phase pick times on KURK array, which depends on distance to the event was found; - measured amplitudes of seismic signals in bulletin are systematically high, than that in the REB bulletin. We intend to continue such comparison of REB and bulletins. In a future it would be interesting to make systematic analysis of missed phases (phases included into one bulletin and not included into another bulletin); to analyze the reasons of phase misinterpretation (cases when, for example Lg phase in one bulletin is named Sn phase in another bulletin) and so on.