**3. Autonomous Fibre-Optic Rotational Seismograph**

A detailed description of the AFORS system was published previously (Jaroszewicz at al., 2011a, 2011b) hence here we summarized the above data regarding its construction, calibration and management. Now we present two examples of these devices - AFORS-1 located in the Książ (Poland) seismological laboratory for the investigation of the rotational events connected to earthquakes, and AFORS-2 located in Warsaw (Poland) used for initial works connected to the investigation of the irregular engineering construction torsional response and the interstory drift (Jaroszewicz et al., 2011c). Before the end of 2011 the next system AFORS-3 will be available as the replacement to the older version FORS-II mounted in the Ojców (Poland) seismological laboratory (Jaroszewicz & Krajewski, 2008).

The optical head of the constructed AFORS devices uses a fibre interferometer in a minimum optical gyro configuration (Jaroszewicz et al. 2006a), as it is shown in the upper part of Fig. 3a.

The application of the broadband low coherence superluminescent diode – SLD (*EXALOS* - Switzerland with optical power – 20.87 mW, operation wavelength – 1326.9 nm, and spectral radiation band – 31.2 nm,) gives possibility for a minimisation of polarization influence on

Fibre-Optic Sagnac Interferometer as Seismograph for Direct Monitoring of Rotational Events 341

where: So – optical and Se – electronic constants, related to the parameters of used optical and electronic components. The digital form of a signal processing enables the application of the 32-bit signal processor TMS320F283535 (*Texas Instruments*) working with a frequency of 150 MHz as an optimal DSP unit, for a calculation and monitoring of the rotation rate

the basis of signal frames having 1024 length of 16-bit samples. Finally, the obtained results are stored on a CD card and transmitted by a GSM/GPS module to a special WEB FORS -

The evaluation of the optical and electronic constans needs a sensor calibration process, which is based on the measurement of the Earth rotation for Warsaw, Poland i.e. E = 9.18 deg/h ≡ 4.4510-5 rad/s (Krajewski et al., 2005; Jaroszewicz et al., 2011a). During the calibration the AFORS is mounted vertically on a rotation table. For the sensor loop directed in the East-West direction, the measured rotation equals zero because in this direction, its plane is collinear with the Earth rotation axis, whereas for the North-South direction, the measured signal obtains the maximum plus or minus values because the plane of a sensor loop is perpendicular to the vector component of the Earth rotation E. For two existing systems we obtained the following values for the above constants: *S0*= 0.0043 s-1, *Se*=0.0144

After the practical construction of the AFORS devices, their accuracy has been checked. However, this work made in MUT located in Warsaw city, could give limited information on the system accuracy because of urban noises. Figure 4 summarizes these measurements. Since the ASPU allows for step changes of the detection frequency band in the range from 0.83 Hz to 106.15 Hz (Jaroszewicz et al., 2011b), the obtained accuracy is at the level of 5.0710-9/4.8110-9 rad/s - 5,5110-8/6.1110-8 rad/s (for AFORS-1/AFORS-2), respectively for the lower and higher working frequency band. As one can see, the obtained values are well correlated with Ωmin in quantum noise limitation. It should be noticed that the linear dependence of AFORS sensitivity and accuracy in the detection frequency range is the advantage of this system, taking into account the expected frequency characteristics of the rotational seismic waves (Teisseyre et al., 2006). For comparison, Fig. 4 includes also the measured accuracy of the older system FORS-II; which was 4.310-8 rad/s (Jaroszewicz et al.,

2006; Jaroszewicz & Krajewski, 2008), at 20 Hz - the fixed detection frequency band.

Fig. 4. The accuracy measured in Warsaw, Poland for the chosen detection band for three

Telemetric Server.

for AFORS-1 and *S0*=0.059 s-1, *Se*=0.0134 for AFORS-2.

devices: AFORS-1, AFORS-2 and FORS-II

on

Fig. 3. General schema of the AFORS (a): upper – the optical head (generation of the Sagnac phase shift proportional to measured rotation rate ), bottom – Autonomous Signal Processing Unit (rotation calculation and recording), (b): general view of all AFORS (top) and ASPU (bottom)

the system operation by achieving light depolarization in a sensor loop (Krajewski et al., 2005). Next the set of cascade fibre polarizers (with total extinction above 100 dB) enables a true single mode operation of the whole system and guarantees that the only nonreciprocal effect in system is the Sagnac effect. Moreover, a 0.63 m diameter sensor loop has been made from a special composite material with permalloy particles for shielding from the magnetic field. A long length of SMF-28 fibre has been winded in a double-quadrupole mode (Dai et al., 2002) with a 0.2 mm Teflon insulation between each fibre layers which is for the thermal stabilization of the sensor's work, for expected 2-4 degree per day temperature fluctuation in seismic observatories. The system optimization made for AFORSs (15000 m length of fibre with attenuation equal to 0.436 dB/km in sensor loop for AFORS-1 and respectively, 15056 m and 0.450 dB/km for AFORS-2) allows for theoretical sensitivity in quantum noise limit (Jaroszewicz & Wiszniowski, 2008) equal to 1.97·10-9 rad/s/Hz1/2 and 2.46·10-9 rad/s/Hz1/2, respectively for AFORS-1 and AFORS-2. The above mentioned difference between two constructed devices is connected to their total optical loss which is equal to 13.33 dB and 14.47, respectively for AFORS-1 and AFORS-2 (Jaroszewicz et al., 2011b).

The optimisation for a detection of rotation rate is made on the basis of special detection units and utilizes a synchronous detection with properly chosen PM that operates according to the principles presented in part 2 (Krajewski, 2005). For AFORS, a new Autonomous Signal Processing Unit - ASPU (*ELPROMA Ltd*), according to the scheme shown in the lower part of Fig. 1a, has been developped. The ASPU enables the detection of a rotation rate from proper selection (special low-pass filters) and processing (in digital form) the first *A*<sup>1</sup> and the second *A*<sup>2</sup> amplitude of the harmonic output signal, on the basis of the following relation (Jaroszewicz et al., 2011a):

$$\mathbf{Q} = \mathbf{S}\_o \arctan\left[\mathbf{S}\_e \cdot \boldsymbol{\mu}(t)\right]; \quad \mathbf{S}\_0 = (\mathbf{A} \bullet \mathbf{c})(\mathbf{2} \bullet \boldsymbol{\pi} \bullet \mathbf{L} \bullet \mathbf{D}), \quad \mathbf{u}(t) = \mathbf{A}\_{1\alpha} / A\_{2\alpha} \tag{3}$$

(a) (b)

Fig. 3. General schema of the AFORS (a): upper – the optical head (generation of the Sagnac

the system operation by achieving light depolarization in a sensor loop (Krajewski et al., 2005). Next the set of cascade fibre polarizers (with total extinction above 100 dB) enables a true single mode operation of the whole system and guarantees that the only nonreciprocal effect in system is the Sagnac effect. Moreover, a 0.63 m diameter sensor loop has been made from a special composite material with permalloy particles for shielding from the magnetic field. A long length of SMF-28 fibre has been winded in a double-quadrupole mode (Dai et al., 2002) with a 0.2 mm Teflon insulation between each fibre layers which is for the thermal stabilization of the sensor's work, for expected 2-4 degree per day temperature fluctuation in seismic observatories. The system optimization made for AFORSs (15000 m length of fibre with attenuation equal to 0.436 dB/km in sensor loop for AFORS-1 and respectively, 15056 m and 0.450 dB/km for AFORS-2) allows for theoretical sensitivity in quantum noise limit (Jaroszewicz & Wiszniowski, 2008) equal to 1.97·10-9 rad/s/Hz1/2 and 2.46·10-9 rad/s/Hz1/2, respectively for AFORS-1 and AFORS-2. The above mentioned difference between two constructed devices is connected to their total optical loss which is equal to 13.33 dB and

The optimisation for a detection of rotation rate is made on the basis of special detection units and utilizes a synchronous detection with properly chosen PM that operates according to the principles presented in part 2 (Krajewski, 2005). For AFORS, a new Autonomous Signal Processing Unit - ASPU (*ELPROMA Ltd*), according to the scheme shown in the lower part of Fig. 1a, has been developped. The ASPU enables the detection of a rotation

 from proper selection (special low-pass filters) and processing (in digital form) the first *A*<sup>1</sup> and the second *A*<sup>2</sup> amplitude of the harmonic output signal, on the basis of the

•*c*••*L*•*D*), *u(t)*=A1 /A2 (3)

phase shift proportional to measured rotation rate ), bottom – Autonomous Signal Processing Unit (rotation calculation and recording), (b): general view of all AFORS (top)

14.47, respectively for AFORS-1 and AFORS-2 (Jaroszewicz et al., 2011b).

following relation (Jaroszewicz et al., 2011a):

arctan ( ) *S S ut o e* ; *S0*=(

and ASPU (bottom)

rate  where: So – optical and Se – electronic constants, related to the parameters of used optical and electronic components. The digital form of a signal processing enables the application of the 32-bit signal processor TMS320F283535 (*Texas Instruments*) working with a frequency of 150 MHz as an optimal DSP unit, for a calculation and monitoring of the rotation rate on the basis of signal frames having 1024 length of 16-bit samples. Finally, the obtained results are stored on a CD card and transmitted by a GSM/GPS module to a special WEB FORS - Telemetric Server.

The evaluation of the optical and electronic constans needs a sensor calibration process, which is based on the measurement of the Earth rotation for Warsaw, Poland i.e. E = 9.18 deg/h ≡ 4.4510-5 rad/s (Krajewski et al., 2005; Jaroszewicz et al., 2011a). During the calibration the AFORS is mounted vertically on a rotation table. For the sensor loop directed in the East-West direction, the measured rotation equals zero because in this direction, its plane is collinear with the Earth rotation axis, whereas for the North-South direction, the measured signal obtains the maximum plus or minus values because the plane of a sensor loop is perpendicular to the vector component of the Earth rotation E. For two existing systems we obtained the following values for the above constants: *S0*= 0.0043 s-1, *Se*=0.0144 for AFORS-1 and *S0*=0.059 s-1, *Se*=0.0134 for AFORS-2.

After the practical construction of the AFORS devices, their accuracy has been checked. However, this work made in MUT located in Warsaw city, could give limited information on the system accuracy because of urban noises. Figure 4 summarizes these measurements. Since the ASPU allows for step changes of the detection frequency band in the range from 0.83 Hz to 106.15 Hz (Jaroszewicz et al., 2011b), the obtained accuracy is at the level of 5.0710-9/4.8110-9 rad/s - 5,5110-8/6.1110-8 rad/s (for AFORS-1/AFORS-2), respectively for the lower and higher working frequency band. As one can see, the obtained values are well correlated with Ωmin in quantum noise limitation. It should be noticed that the linear dependence of AFORS sensitivity and accuracy in the detection frequency range is the advantage of this system, taking into account the expected frequency characteristics of the rotational seismic waves (Teisseyre et al., 2006). For comparison, Fig. 4 includes also the measured accuracy of the older system FORS-II; which was 4.310-8 rad/s (Jaroszewicz et al., 2006; Jaroszewicz & Krajewski, 2008), at 20 Hz - the fixed detection frequency band.

Fig. 4. The accuracy measured in Warsaw, Poland for the chosen detection band for three devices: AFORS-1, AFORS-2 and FORS-II

Fibre-Optic Sagnac Interferometer as Seismograph for Direct Monitoring of Rotational Events 343

The applied technology gives possibility for the remote (via Internet) controlling and changing of all electronic parameters of the ASPU for a given sensor made according to the AFORS technology, as presents, for example, the bookmark *Config* for the AFORS-1 in Fig. 6a. This remote control may comprise a software upgrade. Moreover, the bookmark *Data&Variables* (Fig. 6b) monitors, for given AFORS, in real time the main data and variables with possibility for the remote changing of the threshold – the level of signals which initialize automatic data storing and its GSM transfer. Additionally, the top right corner of bookmarks for the given system on server contains the information on a current date and time and the four main AFORS's parts of state of work (good – as green, partially good as

(a)

(b) Fig. 6. The view of two main bookmarks for AFORS-1at the *FORS - Telemetric Server*: (a)

*Config,* and (b) *Data&Variables*

A *FORS-Telemetric Server* with its main page shown in Fig. 5a (reader can use http://fors.m2s.pl with login and password - AFORSbook for free access to the system) is used for data storing and for monitoring the work of the FORS-II and the AFORSs. Because ASPU of the AFORS contains a GSM/GPS module and an independent power supply for all electronic components of the system, hence the AFORS is fully autonomous and mobile system. In this moment, the 3 devices are managed via server: FORS-II, AFORS-1, AFORS-2 located in Ojców, Książ and Warsaw (all in Poland), respectively (Jaroszewicz et al., 2011b) as it is shown in Fig. 5b.

Fig. 5. Elements of WEB page for AFORS managing: (a) the main page of *FORS - Telemetric Server,* (b) the GOOGLE map with devices localization

A *FORS-Telemetric Server* with its main page shown in Fig. 5a (reader can use http://fors.m2s.pl with login and password - AFORSbook for free access to the system) is used for data storing and for monitoring the work of the FORS-II and the AFORSs. Because ASPU of the AFORS contains a GSM/GPS module and an independent power supply for all electronic components of the system, hence the AFORS is fully autonomous and mobile system. In this moment, the 3 devices are managed via server: FORS-II, AFORS-1, AFORS-2 located in Ojców, Książ and Warsaw (all in Poland), respectively (Jaroszewicz et al., 2011b)

(a)

(b) Fig. 5. Elements of WEB page for AFORS managing: (a) the main page of *FORS - Telemetric* 

*Server,* (b) the GOOGLE map with devices localization

as it is shown in Fig. 5b.

The applied technology gives possibility for the remote (via Internet) controlling and changing of all electronic parameters of the ASPU for a given sensor made according to the AFORS technology, as presents, for example, the bookmark *Config* for the AFORS-1 in Fig. 6a. This remote control may comprise a software upgrade. Moreover, the bookmark *Data&Variables* (Fig. 6b) monitors, for given AFORS, in real time the main data and variables with possibility for the remote changing of the threshold – the level of signals which initialize automatic data storing and its GSM transfer. Additionally, the top right corner of bookmarks for the given system on server contains the information on a current date and time and the four main AFORS's parts of state of work (good – as green, partially good as


(b)

Fig. 6. The view of two main bookmarks for AFORS-1at the *FORS - Telemetric Server*: (a) *Config,* and (b) *Data&Variables*

Fibre-Optic Sagnac Interferometer as Seismograph for Direct Monitoring of Rotational Events 345

to the commercially available FOG instruments such as FORS-1, the proposed system is designed for a direct measurement of a rotational rate, whereas any FOG measures change the angle which is written in their inner electronic system and difficult to direct changes. Additionally, our system prepared according to the AFORS technology has developed the software designed for the Internet system monitoring as well as the remote control which

can manage a large numbers of such devices in a useful way for the operator.

Fig. 7. The data recorded on second (left) and first (right) floor as response for ground

Fig. 8. The data recorded on the second (left) and the first (right) floor as a response for the ground moves generated by the street morning intensity within a distance of about 50 m

At beginning of July, 2010 the AFORS-1 has been installed in the Książ (Poland) seismological observatory together with a set of the Two Antiparallel Pendulum Seismometers (*TAPS-1* and *TAPS-2*) constructed by the Institute of Geophysics (Teisseyre et

As the first example we repeat here (Fig. 9) the histogram of the previously analyzed data (Jaroszewicz et al., 2011b) collected in Książ on March 11th, 2011 at 6 h 58 min. (after Honshu earthquake, M=9.0 on 11 March 2011 at 5 h 46 min. 23 s, recorded in Książ on 11 March 2011 at 5 h 58 min. 35 s.). The above data were obtained from the common for AFORS

al., 2003). As usually, TAPSes are placed perpendicularly, in directions N-S and E-W.

moves after tram pass through street

from and parallel to the long building wall

yellow or no work – as black, respectively). The bookmark *GSM/GPS* (not shown in figure, see for example Jaroszewicz et al., 2011a) monitors in real time the GSM parameters as well as the GPS parameters which include the AFORSes' global localization (see Fig. 5b). Yet another bookmark named *Measurement* presents the collection of data recorded by different devices connected to the server. These data are stored with the main parameters of AFORS in the recording time: ADEV – rotation rate average deviation in rad/s, Omega Offset – rotation rate offset in rad/s, GS Level/Before/After - adjusted level of signal for data stored, and B - adjusted detection band. In this way, in our opinion, the AFORSes with their management via *FORS - Telemetric Server* are fully adopted for monitoring of rotational phenomena connected to earthquakes as well as torsional response and interstory drift of the irregular structures in-plane existing during any ground moves.

### **4. Examples of the experimental data obtained by AFORS**

The previously obtained data from an older system FORS-II have been wide discussed as well as summarized (Jaroszewicz et al., 2006, 2008). For the above reason here we present the summarized data obtained with regard to the AFORS application where AFORS-1 is installed in the Książ (Poland) seismological laboratory for the recording of the rotational phenomena connected to earthquakes, whereas AFORS-2 has been used in the initial experiments for monitoring building rotational moving.

The main source of disturbance during an investigation by AFORS-2 of a building rotation moving was an urban ground motion generated by tram moves within a 50 m distance from a building wall parallel to it. The investigated building is a light construction (five floors of aluminium structure with sandwich walls and ceilings), and the AFORS-2 has been installed, subsequently on the second and first floors in the hall, in the same vertical position (with accuracy of about 10 cm). Since it is an old building with asbestos used as an inner wall isolation, now it is not in use by the academy anymore, so we expected that the recorder signals will be connected to an external perturbation. Figure 7 presents the building moves recorded on the first and the second floors (difference about 3 m of height) for relatively the same ground motion generated by tram moves nearly by midnight on July 13th (AFORS-2 on the first floor) and July 14th (AFORS-2 on the second floor). Since it was a middle of the night during summer holidays, the academy area was empty which had a direct influence on recorded signals and they were very clear. As one can see in the above experiment the accuracy for the AFORS-2 was 3.1510-6 rad/s and 7.9110-6 rad/s (see ADEV parameter in the left down corner of two pictures in Fig. 7), for the chosen detection band equal to 21.23 Hz. The amplitude of the detected rotation rate was about twice higher for the second floor, and was much higher than the system accuracy (more than ten orders).

The urban noise influence on the recorded signals can be observed on the data presented in Fig. 8, which have been obtained in the morning when the Academy opened for work. As one can see the higher amplitude as well as frequency were observed in this time. However, again the much higher amplitude of torsional moves of the building is observed on the higher floor of building.

The above initial results show that the device type AFORS can be useful for a continuous monitoring of an engineering structure of, for example, multi-storey buildings with regard to the investigation of their torsional rotations as well as measuring interstory drifts. These measurements are made without any reference frame which is very important during earthquakes and may be made only by a system based on the Sagnac effect. In comparison

yellow or no work – as black, respectively). The bookmark *GSM/GPS* (not shown in figure, see for example Jaroszewicz et al., 2011a) monitors in real time the GSM parameters as well as the GPS parameters which include the AFORSes' global localization (see Fig. 5b). Yet another bookmark named *Measurement* presents the collection of data recorded by different devices connected to the server. These data are stored with the main parameters of AFORS in the recording time: ADEV – rotation rate average deviation in rad/s, Omega Offset – rotation rate offset in rad/s, GS Level/Before/After - adjusted level of signal for data stored, and B - adjusted detection band. In this way, in our opinion, the AFORSes with their management via *FORS - Telemetric Server* are fully adopted for monitoring of rotational phenomena connected to earthquakes as well as torsional response and interstory drift of

The previously obtained data from an older system FORS-II have been wide discussed as well as summarized (Jaroszewicz et al., 2006, 2008). For the above reason here we present the summarized data obtained with regard to the AFORS application where AFORS-1 is installed in the Książ (Poland) seismological laboratory for the recording of the rotational phenomena connected to earthquakes, whereas AFORS-2 has been used in the initial

The main source of disturbance during an investigation by AFORS-2 of a building rotation moving was an urban ground motion generated by tram moves within a 50 m distance from a building wall parallel to it. The investigated building is a light construction (five floors of aluminium structure with sandwich walls and ceilings), and the AFORS-2 has been installed, subsequently on the second and first floors in the hall, in the same vertical position (with accuracy of about 10 cm). Since it is an old building with asbestos used as an inner wall isolation, now it is not in use by the academy anymore, so we expected that the recorder signals will be connected to an external perturbation. Figure 7 presents the building moves recorded on the first and the second floors (difference about 3 m of height) for relatively the same ground motion generated by tram moves nearly by midnight on July 13th (AFORS-2 on the first floor) and July 14th (AFORS-2 on the second floor). Since it was a middle of the night during summer holidays, the academy area was empty which had a direct influence on recorded signals and they were very clear. As one can see in the above experiment the accuracy for the AFORS-2 was 3.1510-6 rad/s and 7.9110-6 rad/s (see ADEV parameter in the left down corner of two pictures in Fig. 7), for the chosen detection band equal to 21.23 Hz. The amplitude of the detected rotation rate was about twice higher for the second floor, and was much higher than the system accuracy (more than ten orders). The urban noise influence on the recorded signals can be observed on the data presented in Fig. 8, which have been obtained in the morning when the Academy opened for work. As one can see the higher amplitude as well as frequency were observed in this time. However, again the much higher amplitude of torsional moves of the building is observed on the

The above initial results show that the device type AFORS can be useful for a continuous monitoring of an engineering structure of, for example, multi-storey buildings with regard to the investigation of their torsional rotations as well as measuring interstory drifts. These measurements are made without any reference frame which is very important during earthquakes and may be made only by a system based on the Sagnac effect. In comparison

the irregular structures in-plane existing during any ground moves.

**4. Examples of the experimental data obtained by AFORS** 

experiments for monitoring building rotational moving.

higher floor of building.

to the commercially available FOG instruments such as FORS-1, the proposed system is designed for a direct measurement of a rotational rate, whereas any FOG measures change the angle which is written in their inner electronic system and difficult to direct changes. Additionally, our system prepared according to the AFORS technology has developed the software designed for the Internet system monitoring as well as the remote control which can manage a large numbers of such devices in a useful way for the operator.

Fig. 7. The data recorded on second (left) and first (right) floor as response for ground moves after tram pass through street

Fig. 8. The data recorded on the second (left) and the first (right) floor as a response for the ground moves generated by the street morning intensity within a distance of about 50 m from and parallel to the long building wall

At beginning of July, 2010 the AFORS-1 has been installed in the Książ (Poland) seismological observatory together with a set of the Two Antiparallel Pendulum Seismometers (*TAPS-1* and *TAPS-2*) constructed by the Institute of Geophysics (Teisseyre et al., 2003). As usually, TAPSes are placed perpendicularly, in directions N-S and E-W.

As the first example we repeat here (Fig. 9) the histogram of the previously analyzed data (Jaroszewicz et al., 2011b) collected in Książ on March 11th, 2011 at 6 h 58 min. (after Honshu earthquake, M=9.0 on 11 March 2011 at 5 h 46 min. 23 s, recorded in Książ on 11 March 2011 at 5 h 58 min. 35 s.). The above data were obtained from the common for AFORS

Fig. 9. The plots of the seismic events recorded in Książ on March 11th, 2011, starting from 6 h 58 min, after the Honshu M=9.0 earthquake, all times UTC (Jaroszewicz et al., 2011b)

and TAPSes standard seismic recording system named KSPROT with the samples of a signal with the frequency 12.8 kHz and after re-sampling stores with the frequency 100 Hz. Fibre-Optic Sagnac Interferometer as Seismograph for Direct Monitoring of Rotational Events 347

It should be noticed that in the previously paper (Jaroszewicz et al., 2006), we used a wrong name for this station, KST. We underline that here only the AFORS-1 shows the rotational component in a direct way (plots marked as AFORS-1). The rotational component is obtained also from the TAPS system, calculated from the recordings of linear motions in four channels (data named TAPS-1 channel 1, TAPS-1 channel 2, TAPS-2 channel 1 and TAPS-2 Channel 2) with the application of a suitable mathematical procedure, which has been widely described in the previous paper (Solarz et al., 2004). Since AFORS records rotation in a direct way, we use this recording as the reference source, despite that the rotations calculated from TAPSes are generally poorly correlated with it. In the results presented in Fig. 9, a good correlation has been found mainly in short bursts of seismic oscillations, marked here as "events 3", "events 1" and "events 2". These were clusters of short peaks, found when the traces of the great Honshu Earthquake were studied (without much success in the domain of rotational field component, but this is not strange concerning the distance to the earthquake focus and the characteristics of our instruments). After the preliminary analysis, from each group only one event was chosen, and these we call event 3,

For event 1, the data from AFORS-1 were identified by the FORS-Telemetric Server (see Fig. 10a) as the rotational event of an amplitude of about 15·10-6 rad/s with AFORS-1 accuracy equal to 4·10-8 rad/s for the given frequency bandpass which was about 10.6 Hz. The rotation calculated from the linear motions recorded in TAPS system has similar characteristic, as is seen in Fig. 10b (with arbitrary units of amplitude). Nevertheless, we observed some time advance in a relation to the AFORS-1 registration as well as apparent disturbances, visible before and after the event; these may result from a limited accuracy of

Fig. 10. The rotational event 1 from Fig. 5 on 11 March 2011 – a): recorded from AFORS-1 in the FORS -Telemetric Server at 8 h 46 min; – b): calculated from TAPSes four channels

For further study, the recordings of several mining shocks have been chosen; these shocks occurred in the Legnica-Głogów Copper Mining District - LGOM in Western Poland. Some typical results of the analysis of strong shocks are presented here; for the event which occurred in April 30th at 03:32 UTC - the magnitude 2.9 had been found, for the event from June 28th, 23:16 UTC – magnitude 3.2. For viewing the data, for the normalization of the sampling to the first channel and for writing selected time-periods as ASCII, we used the

TAPSes as was previously mentioned (Jaroszewicz & Krajewski, 2008).

event 1 and event 2 accordingly.

(Jaroszewicz et al., 2011b)

Fig. 9. The plots of the seismic events recorded in Książ on March 11th, 2011, starting from 6 h 58 min, after the Honshu M=9.0 earthquake, all times UTC (Jaroszewicz et al., 2011b)

and TAPSes standard seismic recording system named KSPROT with the samples of a signal with the frequency 12.8 kHz and after re-sampling stores with the frequency 100 Hz. It should be noticed that in the previously paper (Jaroszewicz et al., 2006), we used a wrong name for this station, KST. We underline that here only the AFORS-1 shows the rotational component in a direct way (plots marked as AFORS-1). The rotational component is obtained also from the TAPS system, calculated from the recordings of linear motions in four channels (data named TAPS-1 channel 1, TAPS-1 channel 2, TAPS-2 channel 1 and TAPS-2 Channel 2) with the application of a suitable mathematical procedure, which has been widely described in the previous paper (Solarz et al., 2004). Since AFORS records rotation in a direct way, we use this recording as the reference source, despite that the rotations calculated from TAPSes are generally poorly correlated with it. In the results presented in Fig. 9, a good correlation has been found mainly in short bursts of seismic oscillations, marked here as "events 3", "events 1" and "events 2". These were clusters of short peaks, found when the traces of the great Honshu Earthquake were studied (without much success in the domain of rotational field component, but this is not strange concerning the distance to the earthquake focus and the characteristics of our instruments). After the preliminary analysis, from each group only one event was chosen, and these we call event 3, event 1 and event 2 accordingly.

For event 1, the data from AFORS-1 were identified by the FORS-Telemetric Server (see Fig. 10a) as the rotational event of an amplitude of about 15·10-6 rad/s with AFORS-1 accuracy equal to 4·10-8 rad/s for the given frequency bandpass which was about 10.6 Hz. The rotation calculated from the linear motions recorded in TAPS system has similar characteristic, as is seen in Fig. 10b (with arbitrary units of amplitude). Nevertheless, we observed some time advance in a relation to the AFORS-1 registration as well as apparent disturbances, visible before and after the event; these may result from a limited accuracy of TAPSes as was previously mentioned (Jaroszewicz & Krajewski, 2008).

Fig. 10. The rotational event 1 from Fig. 5 on 11 March 2011 – a): recorded from AFORS-1 in the FORS -Telemetric Server at 8 h 46 min; – b): calculated from TAPSes four channels (Jaroszewicz et al., 2011b)

For further study, the recordings of several mining shocks have been chosen; these shocks occurred in the Legnica-Głogów Copper Mining District - LGOM in Western Poland. Some typical results of the analysis of strong shocks are presented here; for the event which occurred in April 30th at 03:32 UTC - the magnitude 2.9 had been found, for the event from June 28th, 23:16 UTC – magnitude 3.2. For viewing the data, for the normalization of the sampling to the first channel and for writing selected time-periods as ASCII, we used the

Fibre-Optic Sagnac Interferometer as Seismograph for Direct Monitoring of Rotational Events 349

For the analysis, we divided each signal into several frequency bands, then searched for similar – and dissimilar! – rotational motions. As it is seen in the Fig. 12, even this method did not reveal great similarity between the rotations obtained from AFORS-1 and from TAPSes. Here, the time-period when the P waves have arrived is shown. For all the transformations, we applied the same procedures, including the digital filters which we

Fig. 12. Rotation motions in the recording of a mining seismic event, during P waves arrivals, as seen in three frequency bands: 12.5-18 Hz, 10-12.5 Hz and 7.5-10 Hz. In each of the paired diagrams, the upper one shows the rotation calculated from the system of

TAPSes, the lower one – the rotation sensed by the AFORS-1

constructed in Matlab®.

programs written by dr Jan Wiszniowski of the Institute of Geophysics, P.A.Sci. Subsequent analysis was done in Matlab®.

The rotational motions in the trace of a seismic event, recorded from AFORS-1 and indirectly obtained from TAPSes, differ substantially, as it is seen in Fig. 11. The signal obtained from AFORS-1, here – the middle plot, has much more peaks and indentations than both the channel 1 and channel 4, which are two of the four channels of the TAPSes system. The signals from TAPSes are shown in m/s, while the rotational signal from AFORS-1 is in conventional units. In the following Figures, the rotations found from TAPSes system are plotted in rad/s.

Fig. 11. The example of the seismic event traces, prepared for analysis. Upper channel – channel 1, first of two from the first electromechanical rotational seismometer TAPS. Middle channel – rotational seismogram from AFORS-1, sampling-normalized to the channel 1. Lower channel – first from the second TAPS, sampling-normalized to the channel 1. The mining seismic event in LGOM (western Poland) on 2011.04.30, at 03:32 UTC

programs written by dr Jan Wiszniowski of the Institute of Geophysics, P.A.Sci. Subsequent

The rotational motions in the trace of a seismic event, recorded from AFORS-1 and indirectly obtained from TAPSes, differ substantially, as it is seen in Fig. 11. The signal obtained from AFORS-1, here – the middle plot, has much more peaks and indentations than both the channel 1 and channel 4, which are two of the four channels of the TAPSes system. The signals from TAPSes are shown in m/s, while the rotational signal from AFORS-1 is in conventional units. In the following Figures, the rotations found from TAPSes

Fig. 11. The example of the seismic event traces, prepared for analysis. Upper channel – channel 1, first of two from the first electromechanical rotational seismometer TAPS. Middle channel – rotational seismogram from AFORS-1, sampling-normalized to the channel 1. Lower channel – first from the second TAPS, sampling-normalized to the channel 1.

The mining seismic event in LGOM (western Poland) on 2011.04.30, at 03:32 UTC

analysis was done in Matlab®.

system are plotted in rad/s.

For the analysis, we divided each signal into several frequency bands, then searched for similar – and dissimilar! – rotational motions. As it is seen in the Fig. 12, even this method did not reveal great similarity between the rotations obtained from AFORS-1 and from TAPSes. Here, the time-period when the P waves have arrived is shown. For all the transformations, we applied the same procedures, including the digital filters which we constructed in Matlab®.

Fig. 12. Rotation motions in the recording of a mining seismic event, during P waves arrivals, as seen in three frequency bands: 12.5-18 Hz, 10-12.5 Hz and 7.5-10 Hz. In each of the paired diagrams, the upper one shows the rotation calculated from the system of TAPSes, the lower one – the rotation sensed by the AFORS-1

Fibre-Optic Sagnac Interferometer as Seismograph for Direct Monitoring of Rotational Events 351

Searching for a similarity between peaks, indentations and their sequences, placed in the same points in the both plots of rotations' average values, is the searching for hidden similarities between the rotations obtained from electromechanical seismometers system and these obtained from AFORS-1. As it is seen in the Figs. 13 – 16, only partial similarity may be found, in the specific frequency bands only. In the two cases presented here, these were the frequency bands 3 - 4.8 Hz and 10 - 12.5 Hz, and to lesser extent – 12.5 - 18 Hz.

Fig. 14. Analysis as in Fig. 13; the same seismic event. The lower frequency bands: 4.8-7.5

It is often hard to find the beginning of the seismic event's trace in the diagram of rotational motions; dividing the signals into frequency bands usually helps in the case of recordings from electromechanical seismometers (TAPSes), but for recordings from AFORS-1 the improvement is smaller. For the frequencies in the band 12.5 - 18 Hz, and higher, the rotational trace of a mining seismic event is often obscured by the noise. This is more clearly seen in the readings from AFORS-1 because of a high sensitivity of this

Hz, 3-4.8 Hz and 1.7-3 Hz

Further, we tried to compare both rotational signals after averaging each of them for consecutive time-periods of 0.5 s, that is – 50 samples. The results, for six chosen frequency bands, are shown in Figs. 13 – 16. The averages of the absolute values of rotations are plotted as thick line. Additionally, two other averages are added. The thin continuous is for acceleration (in fact – the difference between neighbouring samples) – again, the absolute values are averaged for consecutive stages. Finally, thin dashed line shows the average of the squared rotational signal. These additional plots were normalized to the rotational signals, so their maxima coincide. The Figs. 13 and 14 are obtained from the analysis of the shock from April 30th; the Figs. 15 and 16 – from the analysis of the stronger shock, that of June 28th.

Fig. 13. The rotations, their squares and time-differentials equivalent to rotational acceleration, averaged in consecutive stages, 50 or 0.5 second long. Before averaging, the rotations and their differentials were transformed into the absolute values. The seismic event in LGOM, 2011.04.30, at 03:32 UTC. The upper frequency bands: 12.5-18 Hz, 10-12.5 Hz and 7.5-10 Hz. In each of the paired diagrams, the upper one shows signals obtained from the system of TAPSes, the lower one – from the AFORS-1

Further, we tried to compare both rotational signals after averaging each of them for consecutive time-periods of 0.5 s, that is – 50 samples. The results, for six chosen frequency bands, are shown in Figs. 13 – 16. The averages of the absolute values of rotations are plotted as thick line. Additionally, two other averages are added. The thin continuous is for acceleration (in fact – the difference between neighbouring samples) – again, the absolute values are averaged for consecutive stages. Finally, thin dashed line shows the average of the squared rotational signal. These additional plots were normalized to the rotational signals, so their maxima coincide. The Figs. 13 and 14 are obtained from the analysis of the shock from April 30th; the Figs. 15 and 16 – from the analysis of the stronger shock, that of June 28th.

Fig. 13. The rotations, their squares and time-differentials equivalent to rotational acceleration, averaged in consecutive stages, 50 or 0.5 second long. Before averaging, the rotations and their differentials were transformed into the absolute values. The seismic event in LGOM, 2011.04.30, at 03:32 UTC. The upper frequency bands: 12.5-18 Hz, 10-12.5 Hz and 7.5-10 Hz. In each of the paired diagrams, the upper one shows signals obtained

from the system of TAPSes, the lower one – from the AFORS-1

Searching for a similarity between peaks, indentations and their sequences, placed in the same points in the both plots of rotations' average values, is the searching for hidden similarities between the rotations obtained from electromechanical seismometers system and these obtained from AFORS-1. As it is seen in the Figs. 13 – 16, only partial similarity may be found, in the specific frequency bands only. In the two cases presented here, these were the frequency bands 3 - 4.8 Hz and 10 - 12.5 Hz, and to lesser extent – 12.5 - 18 Hz.

Fig. 14. Analysis as in Fig. 13; the same seismic event. The lower frequency bands: 4.8-7.5 Hz, 3-4.8 Hz and 1.7-3 Hz

It is often hard to find the beginning of the seismic event's trace in the diagram of rotational motions; dividing the signals into frequency bands usually helps in the case of recordings from electromechanical seismometers (TAPSes), but for recordings from AFORS-1 the improvement is smaller. For the frequencies in the band 12.5 - 18 Hz, and higher, the rotational trace of a mining seismic event is often obscured by the noise. This is more clearly seen in the readings from AFORS-1 because of a high sensitivity of this

Fibre-Optic Sagnac Interferometer as Seismograph for Direct Monitoring of Rotational Events 353

frequencies. The share of low frequency rotations varies between local events, as was found for several shocks recorded in Ojców, Poland (K. P. Teisseyre, 2006) and in l'Aquila, central Italy (K. P. Teisseyre, 2007). In each of these studies, the rotational motions were derived from the set of two TAPSes, in other words – from two pairs of electromechanical horizontal

Fig. 16. Analysis as in Fig. 13. The seismic event in LGOM, 2011.06.28, at 23:16 UTC. The

The agreement between the plots of the stage-averaged rotation and the stage-averaged squared rotation is not surprising; it shows that in the seismic energy received at the station, higher amplitudes motions are more important than the smaller ones. On the other hand, in several time-periods (stages), a disagreement between rotational acceleration and the appropriate rotational signal is often found. When the acceleration plot is below that of the adequate rotational velocity plot - see Fig. 14 - this may be explained by a shift from

lower frequency bands: 4.8-7.5 Hz, 3-4.8 Hz and 1.7-3 Hz

seismometers.

equipment to high frequency rotations. The mentioned noise might be of an instrumental or external origin, the latter is more probable. Nevertheless, the rotational traces of the seismic events in LGOM area also bear high frequency rotations, in various portions. These high frequency rotations may originate in the focus, or in the vicinity of the seismic station, as a response to the arriving seismic waves. We relate the latter explanation also to the short bursts of high frequency rotational motions, which quite often accompany the recording of an earthquake or other seismic event (Jaroszewicz et al., 2011b).

Fig. 15. Analysis as in Fig. 13. The seismic event in LGOM, 2011.06.28, at 23:16 UTC. The upper frequency bands: 12.5-18 Hz, 10-12.5 Hz and 7.5-10 Hz

When a recording from AFORS-1 is analyzed in low frequencies, below 1.7 Hz, a trace of a seismic event is almost not visible, and it is barely discernible also in the frequency band 1.7 – 3 Hz. That's unfortunate, as a substantial part of rotational oscillations comes in low

equipment to high frequency rotations. The mentioned noise might be of an instrumental or external origin, the latter is more probable. Nevertheless, the rotational traces of the seismic events in LGOM area also bear high frequency rotations, in various portions. These high frequency rotations may originate in the focus, or in the vicinity of the seismic station, as a response to the arriving seismic waves. We relate the latter explanation also to the short bursts of high frequency rotational motions, which quite often accompany the recording of an earthquake or other seismic event (Jaroszewicz et

Fig. 15. Analysis as in Fig. 13. The seismic event in LGOM, 2011.06.28, at 23:16 UTC. The

When a recording from AFORS-1 is analyzed in low frequencies, below 1.7 Hz, a trace of a seismic event is almost not visible, and it is barely discernible also in the frequency band 1.7 – 3 Hz. That's unfortunate, as a substantial part of rotational oscillations comes in low

upper frequency bands: 12.5-18 Hz, 10-12.5 Hz and 7.5-10 Hz

al., 2011b).

frequencies. The share of low frequency rotations varies between local events, as was found for several shocks recorded in Ojców, Poland (K. P. Teisseyre, 2006) and in l'Aquila, central Italy (K. P. Teisseyre, 2007). In each of these studies, the rotational motions were derived from the set of two TAPSes, in other words – from two pairs of electromechanical horizontal seismometers.

Fig. 16. Analysis as in Fig. 13. The seismic event in LGOM, 2011.06.28, at 23:16 UTC. The lower frequency bands: 4.8-7.5 Hz, 3-4.8 Hz and 1.7-3 Hz

The agreement between the plots of the stage-averaged rotation and the stage-averaged squared rotation is not surprising; it shows that in the seismic energy received at the station, higher amplitudes motions are more important than the smaller ones. On the other hand, in several time-periods (stages), a disagreement between rotational acceleration and the appropriate rotational signal is often found. When the acceleration plot is below that of the adequate rotational velocity plot - see Fig. 14 - this may be explained by a shift from

Fibre-Optic Sagnac Interferometer as Seismograph for Direct Monitoring of Rotational Events 355

Bergh, R. A., Lefèvre, H. C. & Shaw, H. J. (1981). All-Single-Mode Fiber-Optic Gyroscope

Burns, W. K. (1986). Phase-Error Bounds of Fiber Gyro With Polarization-Holding Fiber, *J. Lightwave Tech.*, Vol. LT4, No.1, (January 1986), pp. 8-14, ISSN 0733-8724 Cowsik, R, Madziwa-Nussinov T., Wagoner K., Wiens D., Wysession, M. (2009).

Dai, X., Zhao, X., Cai, B., Yang, G., Zhou, K. & Liu, C. (2002). Quantitative Analysis of the

Droste, Z. & Teisseyre, R. (1976). Rotational and Displacemental Components of Ground

Eringen, A. C. (1999). *Mirocontinuum field theories. Vol. 1 Foundations and Solids*, Springer-

Ezekiel, S., Davis, J. L. & Hellwarth, R. W. (1982). Intensity Dependent Nonreciprocal Phase

Ferrari, G. (2006). Note on the Historical Rotation Seismographs, in: *Earthquake Source* 

Franco-Anaya, R., Carr, A. J. & Schreiber. K. U. (2008). Qualification of Fibre-Optic

Fredricks, R. J. & Ulrich, R. (1984). Phase-Error Bounds of Fibre Gyro with Imperfect Polariser/Depolarizer, *Electron. Lett.*, Vol.20, No.8, pp. 330-332, ISSN 0013-5194 Harress, F. (1912). Die geschwindigkeit des lichtes in bewegten korper, *Dissertation,* Jena,

Jaroszewicz, L. R. & Krajewski Z. (2008). Application of the Fibre-Optic Rotational

Jaroszewicz, L. R. & Wiszniowski J. (2008). Measurement of Short-Period Weak Rotation

Jaroszewicz, L. R., Krajewski, Z. & Teisseyre, K. P. (2011b). Usefulness of AFORS –

Seismometer in Investigation of the Seismic Rotational Waves, *Opto-Electron. Rev.,* 

Signals, in: *Physics of Asymmetric Continuum: Extreme and Fracture Processes*, R. Teisseyre, H. Nagahama & E. Majewski, (Eds.), pp.17-47, ISBN 978-3-540-68354-4,

Autonomous Fibre-Optic Rotational Seismograph for Investigation of Rotational Phenomena, *Journal of Seismology,* Special issue: Rotational Ground Motions, in

(June 2002), pp. 1155-1156, ISSN 0091-3286

Verlag, ISNB 0-387-95275-6, New York

Harzer, P. (1914). *Astron, Nachr.*, Vol.198, pp.377-378

*Geophys. Pol. Acad. Sc., Vol.*97, pp. 157-167, ISSN 0139-0109

Majewski, (Eds.), 367-376, ISBN 3-540-31336-2, Springer, Berlin

Vol.16, No.3, (September 2008), pp. 314-320, ISSN 1230-3402

ISSN 0146-9592

0037-1106

336

Zealand, 2008

Germany

Springer, Berlin

press, ISSN 1383-4649

with Long-Term Stability, *Optic. Lett.,* Vol. 6, No.10, (October 1981), pp. 502-504,

Performance Characteristics of a Rotational Seismometer for Near-Field and Engineering Applications, *BSSA*, Vol.99, No.2B, (May 2009), pp. 1181-1189, ISSN

Shupe Reduction in a Fiber-Optic Sagnac Interferometer, *Opt. Eng.,* Vol.41, No.6,

Motion as Deduced from Data of the Azimuth System of Seismograph, *Publs Inst.* 

Shift in a Fiberoptic Gyroscope, *Springer Series in Optical Sciences*, Vol.32, pp. 332-

*Asymmetry, Structural Media and Rotation Effects*, R. Teisseyre, M. Takeo & E.

Gyroscopes for Civil Engineering Applications, *Proc. of the New Zealand Society of Earthquake Engineering (NZSEE) Conf.,* (available on CD-ROM), Wairakei, New

symmetry of average signal amplitudes in a relation to the zero level. But when we take into account that the acceleration has been amplitude-normalized to the rotation, it is clear that we cannot estimate the size of this shift. Such shifts may be present always, or they may be absent only in some cases and frequency bands, for example - this depicted in Fig. 15, two lowermost panels; here – the acceleration coincide with the rotation or is plotted higher except for the summit.

Some methods of cleaning the signals from possible contaminations are under development, and we are going to record rotational motions with the use of two Sagnac-type interferometers placed at some distance. Nevertheless, the studies outlined here show that the field experiments with AFORS-1 bring an important information concerning the seismic field of mining seismic events.
