Characterizing the Background Noise Level

of Rotational Ground Motions on Earth

Andreas Brotzer*1 , Heiner Igel1 , Eléonore Stutzmann2 , Jean-Paul Montagner2 ,
Felix Bernauer1 , Joachim Wassermann1 , Rudolf Widmer-Schnidrig3 , Chin-Jen Lin4 ,
Sergey Kiselev5 , Frank Vernon6 , and Karl Ulrich Schreiber7

Abstract
The development of high-sensitive ground-motion instrumentation for Earth and
planetary exploration is governed by so-called low-noise models, which characterize
the minimum level of physical ground motions, observed across a very broad frequency
range (0.1 mHz–100 Hz). For decades, broadband instruments for seismic translational
ground-motion sensing allowed for observations down to the Earth’s low-noise model.
Knowing the lowermost noise level distribution across frequencies enabled not only to
infer characteristics of Earth such as the ocean microseismic noise (microseisms) and seis-
mic hum, but also to develop highly successful ambient seismic noise analysis techniques
in seismology. Such a low-noise model currently does not exist for rotational ground
motions. In the absence of a substantial observational database, we propose a prelimi-
nary rotational low-noise model (RLNM) for transverse rotations based on two main
wavefield assumptions: the frequency range under investigation is dominated by sur-
face-wave energy, and the employed phase velocity models for surface waves are rep-
resentative. These assumptions hold, in particular, for a period range of about 2–50 s and
lose validity towards long periods when constituents produced by atmospheric pressure
dominate. Because noise levels of vertical and horizontal accelerations differ, we expect
also different noise levels for transverse and vertical rotations. However, at this moment,
we propose a common model for both types of rotations based on the transverse RLNM. Cite this article as Brotzer, A., H. Igel,
We test our RLNM against available direct observations provided by two large-scale ring E. Stutzmann, J.-P. Montagner, F. Bernauer,
J. Wassermann, R. Widmer-Schnidrig,
lasers (G-ring and ROMY) and array-derived rotations (Piñon Flats Observatory array, C.-J. Lin, S. Kiselev, F. Vernon, et al. (2023).
Gräfenberg array, and ROMY array). We propose this RLNM to be useful as guidance Characterizing the Background Noise
Level of Rotational Ground Motions on
for the development of high-performance rotation instrumentation for seismic applica- Earth, Seismol. Res. Lett. XX, 1–12,
tions in a range of 2–50 s. Achieving broadband sensitivity below such a RLNM remains a doi: 10.1785/0220230202.
challenging task, but one that has to be achieved. Supplemental Material

Introduction on the envelope of averaged PSDs computed for recordings of

The concept of low-noise models is known in many scientific ground acceleration for a set of 75 global stations (Peterson,
fields and characterizes the minimum background noise
level for a certain measurement in a given environment.
1. Department of Earth and Environmental Sciences, Ludwig-Maximilians-Universität
Noncoherent noise levels, thus nontransient signals, are com- München, Munich, Germany, https://orcid.org/0000-0003-2422-7907 (AB);
monly described by computing a power spectral density (PSD) https://orcid.org/0000-0002-7242-6399 (HI); https://orcid.org/0000-0002-
9509-4905 (FB); https://orcid.org/0000-0002-4088-1792 (JW); 2. Université Paris
across the relevant frequency bandwidth. Applied on seismic, Cité, Institut de physique du globe de Paris, CNRS, Paris, France, https://orcid.org/
or more general, physical ground motions, this reflects the 0000-0002-4348-7475 (ES); https://orcid.org/0000-0001-9958-3012 (J-PM);
3. Black Forest Observatory (BFO), Institut für Geodäsie, Universität Stuttgart,
minimal power of seismic or physical background motions
Stuttgart, Germany, https://orcid.org/0000-0001-9698-2739 (RW-S); 4. Institute of
expected on a planet per frequency band under the best obser- Earth Sciences, Academia Sinica, Taipei, Taiwan, https://orcid.org/0000-0003-
vational circumstances. Therefore, a low-noise model sets the 4374-4256 (C-JL); 5. Institute of Physics of the Earth, Moscow, Russian Federation,
https://orcid.org/0000-0003-2285-5862 (SK); 6. Institute of Geophysics and
minimum threshold for observations of signals of interest. For Planetary Physics, Scripps Institution of Oceanography, San Diego, California, U.S.A.,
translational seismic ground motions, more precisely acceler- https://orcid.org/0000-0002-9379-4000 (FV); 7. Fundamental Station Wettzell,
Technical University of Munich, Bad Kötzting Germany, https://orcid.org/0000-
ation, a new low- and high-noise model (NLNM and NHNM) 0002-3775-5058 (KUS)
has been presented by Peterson (1993) replacing the previous *Corresponding author: [email protected]
models by Brune and Oliver (1959). These models were based © Seismological Society of America

Volume XX • Number XX • – 2023 • www.srl-online.org Seismological Research Letters 1

Downloaded from http://pubs.geoscienceworld.org/ssa/srl/article-pdf/doi/10.1785/0220230202/6122979/srl-2023202.1.pdf

by CNRS_INSU user
 1993). The NLNM still serves as a benchmark for instrumental Methodology
self-noise level of translational seismic sensors to resolve The NLNM by Peterson (1993) for vertical acceleration is
observations of the lowermost expected signals. With the inferred from global observations. Because of a current short-
emergence of rotational observations in seismology due to age of permanently operated rotational sensors with high sen-
new instrumentation developments, such as large-scale Sagnac sitivity, we chose a theoretical approach to convert the NLNM
interferometers, also referred to as ring lasers (Stedman et al., for vertical accelerations to obtain a theoretical RLNM for
1995; McLeod et al., 1998; Schreiber and Wells, 2023), and transverse rotations on Earth.
optical fibre gyroscopes (Bernauer et al., 2012, 2018), at the
beginning of this century, a constant drive for improved instru- Translation to rotation
mentation and application in seismology could be observed To convert translations to rotations, we use known relations
(e.g., Pancha et al., 2000; Igel et al., 2005; Wassermann et al., for plane waves in linear elasticity, introducing rotations at
2009; Schmelzbach et al., 2018; Kislov and Gravirov, 2021). the
̣ free surface (e.g., Cochard et al., 2006). The rotation rate
Observations of three components of rotational ground Ω~ is generally defined as follows:
motions complement classic observations of three components ̣
of translational ground motions. In general, a deformable ~
~r,t  1 ∂
∇
Ω ~ × u~
~r ,t,
1
2 ∂t
elastic medium has 12 independent degrees of freedom (three
for translation, three for rotation and six for strain; e.g. Aki in which u~
~r,t is the seismic displacement, at the location ~r
and Richards, 2002). Colocated seismometers and rotational and time t.
sensors, with three components each (6C station), enable For
̣
SH-polarized waves (Love waves), the vertical rotation
the application of new processing techniques (e.g., Igel rate ΩZ is related via the horizontal Love phase velocity cL with
et al., 2007; Sollberger et al., 2020), in particular, for field the transverse acceleration üT ,
deployment (Yuan et al., 2020), to better understand seismic ̣ 1 üT
wavefields and the subsurface. Resulting benefits comprise, for ΩZ  − :
2
2 cL
instance, general wavefield decomposition (Sollberger et al.,
2020), ambient noise analysis (Hadziioannou et al., 2012; For SV-polarized
̣
waves (Rayleigh waves), the transverse
Tanimoto et al., 2015), site characterization (Singh et al., rotation rate ΩT is related via the horizontal Rayleigh phase
2020; Keil et al., 2021), local measurement of anisotropy velocity cR with the vertical acceleration üZ ,
(Noe et al., 2022; Tang et al., 2023), improved structural ̣ üZ
health monitoring (Zembaty et al., 2021; Liao et al., 2022), tele- ΩT  :
3
cR
seismic and eigenmode observations (e.g., Igel et al., 2011; Lin
et al., 2011; Nader et al., 2015), enhanced moment tensor Equations (2) and (3) have been described in the previous
inversion (Donner et al., 2016), dynamic tilt correction of literature (e.g., McLeod et al., 1998; Pancha et al., 2000; Igel
horizontal components of seismometers (e.g., Bernauer et al., 2005; Lin et al., 2011; Belfi et al., 2012). A more detailed
et al., 2020), especially for ocean-bottom seismometers derivation is provided in the supplemental material available to
(Lindner et al., 2017), as well as structural health monitoring this article (Appendix 6).
(Guéguen and Astorga, 2021). Currently, a variety of rotational The existing NLNM for vertical translations as PSD
sensors are operated and developed, such as large-scale (Peterson, 1993), commonly encountered as relative decibel
ring lasers (e.g., Schreiber et al., 2006; Ortolan et al., 2016; units, enables an estimate of a low-noise model for (transverse)
Igel et al., 2021), fibre-optic gyroscopes (e.g., Bernauer rotations using equation (3) according to the following equation:
et al., 2018), mechanical beam balances (e.g., McCann et al., p

























PSDΩ
T 
10PSDa
T=10 =cph,R
T2 with
dB
2021; Venkateswara et al., 2021), or liquid-based sensors
(Leugoud and Kharlamov, 2012). However, to fully exploit
rad2
the aforementioned variety of benefits for seismological appli- PSDΩ
T  ,
4
s2 Hz
cations, portable rotational sensors with improved sensitivity
across a wide frequency range are required. To benchmark this in which T denotes the period, and cph,R
T is the dispersive
instrumentation development, knowing the rotational low- Rayleigh phase velocity. See supplemental material for more
noise model (RLNM) for Earth (and eventually other extrater- details. However, this approach relies on two essential
restrial bodies) is crucial. assumptions:
We describe the employed methodology in the Methodology
section, all data used for comparison with the theoretical 1. The power of the vertical acceleration in the NLNM is
low-noise model for rotations in the Observational Data section, attributed dominantly to fundamental mode Rayleigh
and our results in the Results section. Supporting materials are waves, at least for a range of 2–200 s in which this
appended. assumption approximately holds; and

2 Seismological Research Letters www.srl-online.org • Volume XX • Number XX • – 2023

Downloaded from http://pubs.geoscienceworld.org/ssa/srl/article-pdf/doi/10.1785/0220230202/6122979/srl-2023202.1.pdf

by CNRS_INSU user
 2. the global, dispersive Rayleigh phase velocity structure been neglected and dominating seismic noise is assumed across
(for the fundamental mode) is known or modeled as a the long observation period justifying PSD computation.
distribution or maximum-likelihood curve.
Statistical single-station validation
These assumptions and their consequences are discussed in To evaluate the model in a first step, we chose a statistical
detail in the Discussion. In this article, figures are consistently approach to infer many local rotational models using vertical
2
discussed in terms of power or rather power density
srad 2 Hz of acceleration recordings and the local Rayleigh-wave phase veloc-
s  observations and across periods instead of
rotational rate
rad ities computed for the HUM2 model (Haned et al., 2016). Data
frequencies. of the entire year of 2019 of 120 globally distributed seismic sta-
tions (Fig. S6a) are used to infer mean and median PSDs for
Global phase velocities each station. An estimate of the local PSD for rotational ground
We used the global S-wave velocity model obtained by Haned motions is obtained using the local Rayleigh-wave phase veloc-
et al. (2016), hereafter called HUM2. This model was derived ities, which are interpolated at each station location (Fig. 3e).
from the inversion of path average Rayleigh-wave group veloc-
ities measured on empirical Green functions in the period Array-derived rotations
range from 32 to 246 s. The crustal part of the model is based Rotational ground motions can be inferred from well-designed
on Crust1.0 (Laske et al., 2013) laterally smoothed with a cor- seismic arrays (Huang, 2003; Suryanto et al., 2006; Spudich
relation length of 800 km, which corresponds to the resolution and Fletcher, 2008). Three-component array-derived rotations
of the mantle part of the model. The lateral resolution of this (ADR) are computed for 2019 for three different seismic
model is sufficient for investigating the RLNM. Phase velocities arrays: (1) the ROMY array (RMY) surrounding the site of
were then computed on a grid of 1° × 1°, following Saito (1988) the Geophysical Observatory in Fürstenfeldbruck, Germany;
and using the model HUM2. To infer the global phase veloc- (2) a subset of the Piñon Flats Observatory array (PFO array)
ities for fundamental mode, dispersive Rayleigh waves, two in California, U.S.A.; and (3) a subset of the Gräfenberg array
processing steps were applied: (GRF) located in southeastern Germany.
For RMY, an inner array comprising four stations (FUR,
1. the sampling density at the pole regions has been reduced FFB1, FBB2, and FFB3) and an outer array with six stations,
by introducing the minimum distance (4° ≈ 444 km) using FUR as a central station, is deployed (see Fig. S1b). The
between sampling locations along each latitude, to avoid station distribution of the selected seismic arrays is shown in
a regional bias. The resampled locations are shown in Figure S1 with the arrays’ aperture indicated. The frequency
Figure S6b; and range for array derived rotations is limited,
2. sampling locations in the oceans are neglected using a geo-
graphic continent-ocean mask (Todd, 2020) to satisfy the 0:03c 0:25c
<f < ,
5
assumption of observing dominantly fundamental mode a a
Rayleigh-wave energy and use comparable velocity profiles in which the lower limit depends on the array’s aperture a and
for continental crust only. the local apparent horizontal seismic phase velocity c (Spudich
and Fletcher, 2008; Poppeliers and Evans, 2015). Poppeliers
The sampled locations to extract phase velocities, based on and Evans (2015) set the value of 0.03 for the lower frequency
the HUM2 model (Haned et al., 2016), are shown in Figure limitation based on amplitude measurement uncertainty and
S6a. A probability density distribution of all 1542 extracted interstation distances, both being crucial for gradient calcula-
phase velocities within 2–200 s are shown in Figure 1b as a tion. For the PFO array, Donner et al. (2017) adjusted the
probability density distribution. As expected, high variance lower frequency limit,
of Rayleigh phase velocities is observed for shorter periods
0:00238c 0:25c
(below 10 s), which are mostly sensitive to the large lateral var- <f < ,
6
iations of the crust and lithosphere. Toward longer periods, a a
phase velocities display less lateral variations. based on a re-evaluation of the amplitude uncertainty of the
arrays’ sensors. Table 1 lists the applied limits for the three
Observational Data arrays. Hourly PSDs are computed for each of the three com-
To validate the proposed theoretical RLNM, we compare it ponents (Figs. S2–S5). As a statistical representation, a median
with available direct observations of large-scale ring lasers PSD is generally compared with the proposed RLNM.
as well as array-derived rotational ground motions based on
three seismic arrays with different apertures, thus period Ring laser observations
ranges. All data cover the entire year of 2019 and are described Direct measurements of vertical ground rotations are provided
in detail hereafter. An explicit exclusion of seismic events has by permanent, large-scale ring lasers (= optical Sagnac

Volume XX • Number XX • – 2023 • www.srl-online.org Seismological Research Letters 3

Downloaded from http://pubs.geoscienceworld.org/ssa/srl/article-pdf/doi/10.1785/0220230202/6122979/srl-2023202.1.pdf

by CNRS_INSU user
 (a)

(b)

(c)

Figure 1. (a) The distribution of yearly median power spectral Figure S6b. The black line indicates the maximum of the distri-
densities (PPSDs) for all stations shown in Figure. S6A color coded bution. (c) The estimated rotational low-noise models (RLNMs)
as a probability density with a computed median PSD, and the new based on the NLNM in panel (a) and the phase velocity curves in
low-noise model (NLNM) and new high-noise model (NHNM; panel (b). The black solid line indicates the median RLNM. The
black lines) by Peterson (1993). (b) The probability density of color version of this figure is available only in the electronic edition.
Rayleigh phase velocity curves at sampling locations shown in

4 Seismological Research Letters www.srl-online.org • Volume XX • Number XX • – 2023

Downloaded from http://pubs.geoscienceworld.org/ssa/srl/article-pdf/doi/10.1785/0220230202/6122979/srl-2023202.1.pdf

by CNRS_INSU user
 (a)

(b)

interferometers). For this study, data of 2019 of the G-ring, Figure 2. The left panels show hourly power spectral densities for
located at the Geodetic Observatory in Wettzell (Germany; 2019 for the vertical component of (a) the ROMY ring laser in
e.g., Schreiber et al., 2006) and the vertical component of Fürstenfeldbruck, Germany; and (b) G-ring in Wettzell, Germany.
The right panel represents the median of all PSDs (red solid line)
rotational motions in seismology (ROMY) ring laser array
with a 95% confidence interval (red shade). The black dashed
(Igel et al., 2021), located at the Geophysical Observatory in line represents the RLNM. The color version of this figure is
Fürstenfeldbruck (Germany), are incorporated. Lasing in a available only in the electronic edition.
horizontal plane, both the G-ring as well as the horizontal ring
of ROMY are sensitive to rotations around a vertical axis.
Hence, these rotational motions are induced by SH-polarized for a selected bandwidth of 2–200 s. The median of the distri-
seismic waves (Love waves). The computed hourly PSDs for bution is selected and referred to hereafter as a theoretical
ROMY and G-ring of 2019 are shown in Figure 2. Gaps result RLNM. For the secondary microseism peak at ∼4–5 s, the
from nonoperational periods and maintenance work. The sec- RLNM yields a level of about 10−21 rad2 s2 Hz−1 and for the
ondary microseism is the dominating signal (2–10 s), with primary microseism peak (∼14 s) a level of about
expected seasonal variation, yielding more energy in the winter 5 × 10−24 rad2 s2 Hz−1 . The minimum rotational noise level is
months compared with the summer months (e.g., Tanimoto defined at 50–70 s with 10−26 rad2 s2 Hz−1 . For the secondary
et al., 2015). microseism, all three components of ADR of the PFO array
show a smooth increase in power toward the peak with a steep
Results drop toward 10 s (Fig. 3a). The peak for the secondary micro-
We convert the NLNM for vertical acceleration to a low-noise seism shows an offset of about 2 s toward longer periods com-
model for transverse rotations using equation (3). Figure 1c pared with the RLNM (Fig. 3a,b). Detailed investigations of this
shows the obtained RLNMs as a probability density distribution shift, observed at the PFO, is not within the scope of this article.

Volume XX • Number XX • – 2023 • www.srl-online.org Seismological Research Letters 5

Downloaded from http://pubs.geoscienceworld.org/ssa/srl/article-pdf/doi/10.1785/0220230202/6122979/srl-2023202.1.pdf

by CNRS_INSU user
 Figure 3. Inferred RLNMs are shown as a probability density distri- (dotted) and medians of selected days (solid) for vertical rotations of
bution, and the median RLNM is plotted as a black, solid line. the large ring laser ROMY and G-Ring both located in Germany. The
Compared are median PSDs of 2019 for three components of array- 90% confidence interval is shown as colored area, respectively. (e) A
derived rotations (ADR) for (a) the Piñon Flats Observatory (PFO) array single-station model is based on median PSDs of 2019 of global
in California, U.S.A. (UC San Diego, 2014); (b) the outer (solid) and seismic stations converted to rotations using local phase velocity
inner (dotted) ROMY (RMY) array in Fürstenfeldbruck, Germany; as profiles. (f) Compares self-noise levels of existing portable sensors
well as (c) a subarray of the German Gräfenberg array (GRF; Federal BS3A (Bernauer et al., 2018), QRS (Venkateswara et al., 2021), BRS
Institute for Geosciences and Natural Resources, 1976), where (Venkateswara et al., 2017), ALFRA (McCann et al., 2021; solid
fainted colors are outside the estimated usable frequency range of lines), a theoretical limit based on geometrical design only for large-
this array. The dotted lines toward longer periods for ADR indicate scale ring lasers (ROMY and G-ring; dashed–dotted lines) and
contamination by tilt predominantly by horizontal components, thus planned BS1C (Guattari, personal comm.; dashed line). The color
affecting vertical rotations the most. (d) The median PSDs of 2019 version of this figure is available only in the electronic edition.

6 Seismological Research Letters www.srl-online.org • Volume XX • Number XX • – 2023

Downloaded from http://pubs.geoscienceworld.org/ssa/srl/article-pdf/doi/10.1785/0220230202/6122979/srl-2023202.1.pdf

by CNRS_INSU user
 TABLE 1
Frequency Limits (f min and f max ) and Set Period Limits (pmin and pmax ) according to Equations (5) and (6) and
Required Quantities (a = Aperture; c = Apparent Phase Velocity) for the Seismic Arrays: Piñon Flats Observatory
Array (PFO), Gräfenberg Array (GRF), and ROMY Array (RMY; see Fig S1), Are Listed

Seismic Array a (m) c (km/s) f min (Hz) f max (Hz) pmin (s) pmax (s)

PFO 730 3.8 0.01 1.3 1.0 50.0

GRF 17851 3.8 0.006 0.05 18.8 156.6

RMYo 2628 3.4 0.04 0.3 3.1 25.8

RMYi 230 3.4 0.4 3.7 1.0 3.3

The RMY array is divided into an inner (RMYi ) and outer (RMYo ) subarray.

For the horizontal components, the peak of primary microseism accelerations using globally distributed phase velocities of
is present with about one order of magnitude in power above the Rayleigh waves. The underlying assumptions are (1) that
model (Fig. 3a). The ADR power spectra of the inner and outer Rayleigh waves are the most energetic waves contributing
RMY array show a broad peak for the secondary microseism to the seismic noise for periods between 2 and 200 s; and
band with approximately one order of magnitude in power (2) the model-based phase velocities are representative.
higher (Fig. 3b). For the horizontal components, a signature Concerning the latter assumption, the largest variations are
of the primary microseism is present. The horizontal compo- at short period, and related to the crust and lithospheric lateral
nents of ADR for the GRF array follow the RLNM for periods heterogeneities. We are using phase velocities derived from the
within 18–70 s (Fig. 3c), supporting the estimated low-noise recent global models. To convert vertical acceleration into
level for these periods. A continuation of the PSD outside transverse rotation with equation (3), it is assumed that most
the sensitive period band of the array (see Table 1) is indicated of the seismic energy corresponds to the Rayleigh waves as
with dotted and dashed lines in Figure 3c. Median PSDs and a stated at the end of the Translation to rotation section as
95% confidence interval based on all PSDs of 2019 for G-ring assumption 1. Ambient seismic noise in the 3–300 s period
and ROMY are shown in Figure 3d. For periods longer than range is generated by ocean wave interactions with specific
10 s, the optical Sagnac interferometers G-ring and ROMY mechanisms depending on the period range (Hasselmann,
approach a power level at ∼2 × 10−22 rad2 s2 Hz−1 and 1963; Ardhuin et al., 2015). Secondary microseisms (period
3 × 10−22 rad2 s2 Hz−1 , respectively. This limit is mainly con- band: 3–10 s) are generated by ocean gravity wave–wave inter-
trolled by optical losses of the optical resonator at the coated actions and have been successfully modeled by considering the
mirrors. Generally, more energy in the period band for secon- propagation of Rayleigh waves from the source regions to the
dary microseisms for ROMY compared with G-ring is most stations (Stutzmann et al., 2012). Primary microseisms (period
likely attributed to the geological setting of ROMY inside the band: 10–20 s) are generated by ocean gravity waves interact-
pre-Alpine Molasse basin and its amplification characteristics. ing with the shallow ocean floor at the coast and can also be
effectively modeled by considering only Rayleigh wave propa-
Discussion gation (Gualtieri et al., 2019). Finally, the seismic hum with
Validation of the RLNM periods between 50 and 300 s was successfully modeled by con-
The NLNM and NHNM by Peterson (1993) have been influ- sidering ocean infragravity waves hitting continental shelves as
encing seismology in terms of better understanding Earth source mechanisms and the propagation of Rayleigh waves
by quantifying its continuously excited noise level across (Deen et al., 2018). Most sources of seismic noise in the period
frequencies, as well as a benchmark for instrument developers range of 3–300 s originate from the ocean. Based on excitation
to provide instruments able to observe the smallest signals. If by pressure loading at the ocean floor, Rayleigh waves are
taken at face value, the RLNM is three to four orders of mag- excited predominantly and polarized in the plane of propaga-
nitude in power below the self-noise level of currently operat- tion. However, Love waves, which have transverse polarization,
ing (portable) ground rotation sensors (see Fig. 3f), whereas at have also been observed (e.g., Friedrich et al., 1998). Recently,
the same time self-noise level of today’s broadband seismom- Gualtieri et al. (2021) and Le Pape et al. (2021) showed that
eters are below the NLNM in the studied period band. To close Love-wave constituents of secondary microseisms can be gen-
this gap and observe rotational ground motions at the level of erated by Rayleigh-to-Love-wave conversion on a 3D interface
the RLNM, major technical developments are still required. close to the source region. An analysis of the average energy
We provide the first estimate of the transverse RLNM for ratio of Rayleigh-to-Love waves for the secondary microseism
Earth based on a conversion of the NLNM for vertical including rotational measurements for the G-ring site in

Volume XX • Number XX • – 2023 • www.srl-online.org Seismological Research Letters 7

Downloaded from http://pubs.geoscienceworld.org/ssa/srl/article-pdf/doi/10.1785/0220230202/6122979/srl-2023202.1.pdf

by CNRS_INSU user
 Wettzell yields a range of 0.9–1.0 (Tanimoto et al., 2015). For microseisms range (2–10 s), a reduction of the self-noise level
the site of PFO using seismic array data, energy ratios of for BRS, QRS, and ALFRA of about one to two orders of mag-
2.0–2.5 were obtained (Tanimoto et al., 2016). For Japan, a nitude in power would be sufficient to resolve signals at the
ratio estimate of ∼2 was reported by Nishida et al. (2008). RLNM. Still two to three orders of magnitude in power of reduc-
Although the Rayleigh-to-Love ratio is highly variable tion is required for periods above about 20 s.
(Juretzek and Hadziioannou, 2016), it seems that considering To gain access to the full dynamic range of rotational sig-
only Rayleigh waves propagation serves a valid first-order nals, the self-noise levels of these sensors have to be further
approximation to support assumption 1 in the Translation reduced below the RLNM for the seismically active frequency
to rotation section. To account for assumption 2, stated in range between 2 and 100 s (10 mHz–0.5 Hz).
the Translation to rotation section, on dispersive, global phase
velocities for Love and Rayleigh waves, we relied on up-to-date Comparison of NLNM and GSN low-noise models
crustal and upper-mantle models and employed median veloc- We decided to focus on the NLNM by Peterson (1993) as a well-
ities to suppress local effects of the globally sampled velocities. cited benchmark for vertical seismic acceleration observations.
The median and confidence interval of the single-station Other low-noise models, based on data of the Global Seismic
estimate based on local rotational models is compared with Network (GSN) for vertical (GSN-Z) and also horizontal
the RLNM in Figure 3e. The overall characteristics of the (GSN-H) accelerations, introduced by Berger et al. (2004),
model are reflected, however with an overall higher power are converted to rotational models in Figure 4 for comparison
level, that we attribute mostly to station quality. purposes. The GSN-Z model only shows minor deviations
ADR data of three arrays support the RLNM for their from the NLNM, thus the converted transverse RLNMs of
respective frequency range. An increase in power toward RLNM − TNLNM and RLNM − TGSN−Z , respectively, match well.
longer periods is observed for all ADR data, which can be For the conversion, an assumption of dominantly first-
attributed to a dominating influence of amplitude noise for order Rayleigh wave energy for vertical component for the
longer periods affecting derived gradients. For all arrays, the investigated frequency bandwidth has been made and dis-
vertical ADR spectra show higher noise levels compared with cussed. To convert the GSN-H noise model to rotational rates,
the horizontal ones. Generally, horizontal components of an assumption of dominantly Love-wave energy would be
translational records are more noisy. required. This would be a very strong assumption, when
Large-scale ring lasers currently provide the best direct, assuming equipartitioning of Love-wave and horizontal
high-sensitive measurement of rotational round motions. Rayleigh-wave energy. Nevertheless, we include the conversion
However, a nonportable ring laser is not directly comparable of the GSN-H model for horizontal accelerations using Love
with portable sensors, neither in design, nor operation effort or phase velocities (see Fig. 4b) according to equation (2). The
operation costs. A theoretical sensitivity limit, for the large- resulting model (RLNM − VGSN−H ) shows a noise level for ver-
scale ring lasers G-ring and ROMY, respectively, is shown tical rotations of about one order of magnitude in power lower
in Figure 3f. This resolution or sensitivity limit is merely taking compared with transverse rotations while converging toward
into account the geometrical design of the ring, therefore, longer periods (Fig. 4c). This resembles the characteristics
defining a pure theoretical limit. We neglect limitations due for horizontal acceleration noise levels, being lower for the
to noise sources (e.g., lasing process, electronic self-noise, or microseisms bands and higher toward longer periods, in com-
cavity losses at the mirrors). The dominant limitation defining parison with vertical acceleration noise models (Fig. 4a). For
in the current observation levels (see Fig. 3d) is dominated by the range of 2–20 s, the GSN-H model suggests a lower noise
the scattering, transmission, and absorption losses at the floor for horizontal accelerations with respect to vertical accel-
coated mirrors of the resonator cavity. erations. Because vertical rotations are induced by horizontal
Fibre-optic rotational rate sensors, such as the blueSeis-3A motions, we assume the noise floor of a vertical RLNM to be
sensor (BS3A; Bernauer et al., 2018) or the planned blueSeis- rather lower than higher with regard to the transverse RLNM
1C sensor (BS1C; personal comm.), already offer many field for the respective period band. Although we consider the
applications based on rotational sensing; however, their instru- assumptions for the derivation of the transverse RLNM to
mental self-noise levels are still 2–4 orders of magnitude in be on solid ground, we acknowledge that the assumptions that
power above the presented RLNM. Self-noise characteristics went into the construction of the vertical RLNM are not.
of three mechanical beam balances under development for hori- Therefore, we simply propose that the former model should
zontal components, namely, ALFRA (McCann et al., 2021), QRS be taken as a benchmark for both types of rotations until better
(Venkateswara et al., 2021), and BRS (Venkateswara et al., constrained vertical RLNM models become available.
2017), are shown in Figure 3f. With regard to the QRS sensor,
the presented self-noise level is inferred from observations in a Toward long periods
vault (Venkateswara et al., 2021), which seemed to reveal some The proposed low-noise model for rotational ground motions
power in the secondary microseism band. For the secondary is restricted by the discussed assumptions, in particular,

8 Seismological Research Letters www.srl-online.org • Volume XX • Number XX • – 2023

Downloaded from http://pubs.geoscienceworld.org/ssa/srl/article-pdf/doi/10.1785/0220230202/6122979/srl-2023202.1.pdf

by CNRS_INSU user
 (a)

(b)

(c)

concerning the period range. Toward long periods above 200– Figure 4. (a) Translational low-noise models for ground acceler-
300 s locally generated noise resulting from atmospheric move- ation: the new low- and high-noise models (NLNM and NHNM)
ment above the station starts to dominate seismic far-field sig- by Peterson (1993), as well as the horizontal (GSN-H) and the
vertical (GSN-Z) low-noise model based on data of the Global
nals, such as spheroidal and toroidal hum introduced rotations.
Seismic Network (GSN) by Berger et al. (2004). (b) The phase
Those influences are likely pronounced differently for vertical velocities for Rayleigh and Love waves are shown, including a
and transversal rotational components. For vertical and hori- 95% confidence interval. (c) The corresponding low-noise
zontal rotational observations at long periods, we expect a dif- models for transverse (-T) and vertical (-V) rotational motions
ferent level in analogy to translational observations, because according to equations (2) and (3). Here we assume that the
GSN-Z noise is predominantly composed of Rayleigh waves,
local processes, such as atmospheric pressure-induced ground
whereas the GSN-H noise is predominantly made up of Love
tilts, influence vertical and horizontal components differently. waves, being aware that the latter assumption of equiparti-
tioning is a very strong assumption. The color version of this
Conclusions figure is available only in the electronic edition.
We propose a (transverse) RLNM based on the NLNM by
Peterson (1993) for vertical accelerations, assuming domi-
nantly Rayleigh-wave type energy across a period range of To validate the RLNM with currently available observa-
2–200 s. The conversion is based on surface-wave velocities tions, we used array-derived rotations (ADR) from three differ-
extracted from state of the art lithospheric seismic velocity ently sized seismic arrays and direct rotational measurements
models. A comparison with GSN low-noise models for accel- of the large-scale ring lasers, G-ring and ROMY, over the entire
erations by Berger et al. (2004) reveals a lower noise level for year of 2019. All observational data are compatible with the
vertical rotations of about one magnitude in power, however, inferred background noise level for rotations of the RLNM
requiring a strong assumption of equipartitioning for the con- for their representative period bands.
version. We expect different levels for transverse and vertical Self-noise levels of state-of-the-art rotational sensors, either
rotation noise floors, but we propose that the transverse RLNM already operational or still under development, are compared
is taken as a benchmark for both types of rotations until better against the RLNM. Large-scale, high-sensitive ring lasers, such
constrained vertical RLNM models become available. as G-ring and ROMY, approach the RLNM, in particular, at

Volume XX • Number XX • – 2023 • www.srl-online.org Seismological Research Letters 9

Downloaded from http://pubs.geoscienceworld.org/ssa/srl/article-pdf/doi/10.1785/0220230202/6122979/srl-2023202.1.pdf

by CNRS_INSU user
 the secondary microseism peak. Lowering the currently limit- Maximilians University (LMU) Munich. Heiner Igel is grateful
ing self-noise levels for these ring lasers requires technical for support from the Cecil and Ida Green Foundation supporting
improvements, especially for the coated corner mirrors to his visit to the Institute of Physics and Planetary Physics,
reduce losses of the optical cavity. University of Southern California, in 2022 and 2023. Special thanks
to Céline Hadziioannou for fruitful discussions on the ocean micro-
Available portable rotational sensors (e.g., blueSeis-3A,
seism generation and energy partitioning. Finally, the authors
ALFRA, and QRS) already provide a sensitivity level sufficient
acknowledge the suggestions made by the reviewers and the editor
for a wide range of geophysical applications or seismic isolation to enhance our article.
(e.g., gravitational wave detectors). However, none of these sen-
sors has currently a self-noise level below the proposed RLNM. References
Although lowering self-noise levels of rotational sensors below Aki, K., and P. G. Richards (2002). Quantitative Seismology,
the RLNM remains technically challenging, it is nevertheless University Science Books, Sausalito, California, U.S.A.
essential to achieve this goal to extend the domain of applica- Albuquerque Seismological Laboratory/U.S. Geological Survey
tions to that currently possible with classic seismometers. (USGS) (2014). Global seismograph network, GSN - IRIS/USGS,
doi: 10.7914/SN/IU.
Data and Resources Ardhuin, F., L. Gualtieri, and E. Stutzmann (2015). How ocean
Data of the Ring laser ROMY and its seismic array were provided by the waves rock the earth: Two mechanisms explain microseisms with
Geophysical observatory in Fürstenfeldbruck (Department of Earth periods 3 to 300 s, Geophys. Res. Lett. 42, 765–772, doi: 10.1002/
and Environmental Sciences, Geophysical Observatory, University of 2014GL062782.
Munich, 2001). Data of the G-ring were provided by the Geodetic Belfi, J., N. Beverini, G. Carelli, A. Di Virgilio, E. Maccioni, G. Saccorotti,
observatory in Wettzell. Data of the Piñon Flats Observatory array F. Stefani, and A. Velikoseltsev (2012). Horizontal rotation signals
(UC San Diego, 2014) and the Gräfenberg array (Federal Institute detected by “g-pisa” ring laser for the M w 9.0, March 2011, Japan
for Geosciences and Natural Resources, 1976) are openly accessible earthquake, J. Seismol. 16, 767–776, doi: 10.1007/s10950-012-9276-9.
via Incorporated Research Institutions for Seismology (IRIS) and Berger, J., P. Davis, and G. Ekström (2004). Ambient earth noise: A
BGR International Federation of Digital Seismograph Networks survey of the global seismographic network, J. Geophys. Res. 109,
(FDSN) services, respectively. Data of the Global Seismograph doi: 10.1029/2004JB003408.
Network II (Scripps Institution of Oceanography, 1986), Global Bernauer, F., J. Wassermann, F. Guattari, A. Frenois, A. Bigueur, A.
Seismograph Network (GSN) IU (Albuquerque Seismological Gaillot, E. de Toldi, D. Ponceau, U. Schreiber, and H. Igel (2018).
Laboratory/U.S. Geological Survey [USGS], 2014), and GEOSCOPE BlueSeis3A: Full characterization of a 3C broadband rotational seis-
network G (Institut de physique du globe de Paris [IPGP] and mometer, Seismol. Res. Lett. 89, 620–629, doi: 10.1785/0220170143.
École et Observatoire des Sciences de la Terre de Strasbourg Bernauer, F., J. Wassermann, and H. Igel (2012). Rotational sensors—
[EOST], 1982) went into the analysis, and the free availability is A comparison of different sensor types, J. Seismol. 16, 595–602,
acknowledged. Extensive parts of data processing were done using doi: 10.1007/s10950-012-9286-7.
modules of the ObsPy Python package (Megies et al., 2011). Phase Bernauer, F., J. Wassermann, and H. Igel (2020). Dynamic tilt correc-
velocities are computed using the model from Haned et al. (2016). tion using direct rotational motion measurements, Seismol. Res.
GSN noise model data are based on Berger et al. (2004). All other data Lett. 91, 2872–2880, doi: 10.1785/0220200132.
used in this article came from the published sources listed in the refer- Brune, J. N., and J. Oliver (1959). The seismic noise of the Earth’s surface,
ences. The supplemental material includes a more detailed background Bull. Seismol. Soc. Am. 49, 349–353, doi: 10.1785/BSSA0490040349.
for the equations presented in the Methodology section and additional Cochard, A., H. Igel, B. Schuberth, W. Suryanto, A. Velikoseltsev, U.
figures related to the observational data used for comparison with Schreiber, J. Wassermann, F. Scherbaum, and D. Vollmer (2006).
the rotational low-noise model (RLNM). Data repository: Data files Rotational motions in seismology: Theory, observation, simulation,
of the transverse RLNM as well as Jupyter notebooks to recreate the in Earthquake Source Asymmetry, Structural Media and Rotation
figures of this article are available at https://github.com/andbrocode/ Effects, R. Teisseyre, E. Majewski, and M. Takeo (Editors),
RotationalLowNoiseModel.git; doi: 10.5281/zenodo.10027880; and Springer, Berlin, Heidelberg, 391–411.
https://syncandshare.lrz.de/getlink/fiGtpiKTn9cmUtHixKqP1A/SRL- Deen, M., E. Stutzmann, and F. Ardhuin (2018). The earth’s hum var-
RLNM-2023. All websites were last accessed in December 2023. For iations from a global model and seismic recordings around the
more information, please contact the first author. Indian Ocean, Geochem. Geophys. Geosys. 19, 4006–4020, doi:
10.1029/2018GC007478.
Department of Earth and Environmental Sciences, Geophysical
Declaration of Competing Interests
Observatory, University of Munich (2001). BayernNetz, available at
The authors acknowledge that there are no conflicts of interest https://www.fdsn.org/networks/detail/BW/ (last accessed December
recorded. 2023).
Donner, S., M. Bernauer, and H. Igel (2016). Inversion for seismic
Acknowledgments moment tensors combining translational and rotational ground
The authors gratefully acknowledge support from the European motions, Geophys. J. Int. 207, 562–570, doi: 10.1093/gji/ggw298.
Research Council for funding the ERC-Advanced ROMY Project Donner, S., C. Lin, C. Hadziioannou, A. Gebauer, F. Vernon, D. C.
(Grant Number 339991) and additional support from the Ludwig- Agnew, H. Igel, U. Schreiber, and J. Wassermann (2017).

10 Seismological Research Letters www.srl-online.org • Volume XX • Number XX • – 2023

Downloaded from http://pubs.geoscienceworld.org/ssa/srl/article-pdf/doi/10.1785/0220230202/6122979/srl-2023202.1.pdf

by CNRS_INSU user
 Comparing direct observation of strain, rotation, and displacement Keil, S., J. Wassermann, and H. Igel (2021). Single-station seismic
with array estimates at Piñon flat observatory, California, Seismol. microzonation using 6C measurements. J. Seismol. 25, 103–114,
Res. Lett. 88, 1107–1116, doi: 10.1785/0220160216. doi: 10.1007/s10950-020-09944-1.
Federal Institute for Geosciences and Natural Resources (1976). Kislov, K., and V. Gravirov (2021). Rotational seismology: Review of
German Regional Seismic Network (GRSN), available at https:// achievements and outlooks, Seismic Instrum. 57, 187–202, doi:
www.seismologie.bgr.de/doi/grsn/, doi: 10.25928/MBX6-HR74. 10.3103/S0747923921020262.
Friedrich, A., F. Krüger, and K. Klinge (1998). Ocean-generated Laske, G., G. Masters, Z. Ma, and M. Pasyanos. (2013). Update on
microseismic noise located with the Gräfenberg array, J. CRUST1.0 - A 1-degree global model of Earth’s crust, Geophys.
Seismol. 2, 47–64, doi: 10.1023/A:1009788904007. Res. Abstracts, 7–12 April 2013, Vienna, Austria, Abstract
Gualtieri, L., E. Bachmann, F. J. Simons, and J. Tromp (2021). EGU2013-2658, available at http://igppweb.ucsd.edu/∼gabi/
Generation of secondary microseism love waves: effects of crust1.html.
bathymetry, 3-d structure and source seasonality, Geophys. J. Le Pape, F., D. raig, and C. J. Bean (2021). How deep ocean-land cou-
Int. 226, 192–219, doi: 10.1093/gji/ggab095. pling controls the generation of secondary microseism love waves,
Gualtieri, L., E. Stutzmann, C. Juretzek, C. Hadziioannou, and F. Nat. Commun. 12, 2332, doi: 10.1038/s41467-021-22591-5.
Ardhuin (2019). Global scale analysis and modelling of primary Leugoud, R., and A. Kharlamov (2012). Second generation of a rotational
microseisms, Geophys. J. Int. 218, 560–572, doi: 10.1093/gji/ electrochemical seismometer using magnetohydrodynamic technol-
ggz161. ogy, J. Seismol. 16, 587–593, doi: 10.1007/s10950-012-9290-y.
Guéguen, P., and A. Astorga (2021). The torsional response of civil Liao, C. M., K. Hicke, F. Bernauer, H. Igel, C. Hadziioannou, and E.
engineering structures during earthquake from an observational Niederleithinger (2022). Multi-sensor measurements on a large-
point of view, Sensors 21, 342, doi: 10.3390/s21020342. scale bridge model, in 5. Brückenkolloquium: Fachtagung für
Hadziioannou, C., P. Gaebler, U. Schreiber, J. Wassermann, and H. Beurteilung, Planung, Bau, Instandhaltung und Betrieb von
Igel (2012). Examining ambient noise using colocated measure- Brücken, 6–7 September 2022, Ostfildern, Germany, 223.
ments of rotational and translational motion, J. Seismol. 16, Lin, C. J., H. P. Huang, N. D. Pham, C. C. Liu, W. C. Chi, and W. H.
787–796, doi: 10.1007/s10950-012-9288-5. Lee (2011). Rotational motions for teleseismic surface waves,
Haned, A., E. Stutzmann, M. Schimmel, S. Kiselev, A. Davaille, and A. Geophys. Res. Lett. 38, doi: 10.1029/2011GL047959.
Yelles-Chaouche (2016). Global tomography using seismic hum, Lindner, F., J. Wassermann, M. C. AurschSchmidt-, K. U. Schreiber,
Geophys. J. Int. 204, 1222–1236, doi: 10.1093/gji/ggv516. and H. Igel (2017). Seafloor ground rotation observations:
Hasselmann, K. (1963). A statistical analysis of the generation of micro- Potential for improving signal-to-noise ratio on horizontal OBS
seisms, Rev. Geophys. 1, 177–210, doi: 10.1029/RG001i002p00177. components, Seismol. Res. Lett. 88, 32–38, doi: 10.1785/0220160051.
Huang, B. S. (2003). Ground rotational motions of the 1999 Chi-Chi, McCann, J. J., J. Winterflood, L. Ju, and C. Zhao (2021). A multi-ori-
Taiwan earthquake as inferred from dense array observations, entation low-frequency rotational accelerometer, Rev. Sci. Instrum.
Geophys. Res. Lett. 30, doi: 10.1029/2002GL015157. 92, 064,503, doi: 10.1063/5.0047069.
Igel, H., A. Cochard, J. Wassermann, A. Flaws, U. Schreiber, A. McLeod, D., G. Stedman, T. Webb, and U. Schreiber (1998). Comparison
Velikoseltsev, and N. Dinh Pham (2007). Broad-band observations of standard and ring laser rotational seismograms, Bull. Seismol. Soc.
of earthquake-induced rotational ground motions, Geophys. J. Int. Am. 88, 1495–1503, doi: 10.1785/BSSA0880061495.
168, 182–196, doi: 10.1111/j.1365-246X.2006.03146.x. Megies, T., M. Beyreuther, R. Barsch, L. Krischer, and J. Wassermann
Igel, H., M. F. Nader, D. Kurrle, A. M. Ferreira, J. Wassermann, and (2011). ObsPy–What can it do for data centers and observatories?
K. U. Schreiber (2011). Observations of Earth’s toroidal free Ann. Geophys. 54, 47–58.
oscillations with a rotation sensor: The 2011 magnitude 9.0 Nader, M., H. Igel, A. Ferreira, D. Al-Attar, J. Wassermann, and K.
Tohoku-Oki earthquake, Geophys. Res. Lett. 38, doi: 10.1029/ Schreiber (2015). Normal mode coupling observations with a rota-
2011GL049045. tion sensor, Geophys. J. Int. 201, 1482–1490, doi: 10.1093/gji/ggv082.
Igel, H., K. U. Schreiber, A. Gebauer, F. Bernauer, S. Egdorf, A. Nishida, K., H. Kawakatsu, Y. Fukao, and K. Obara (2008). Background
Simonelli, C. J. Lin, J. Wassermann, S. Donner, C. Hadziioannou, Love and Rayleigh waves simultaneously generated at the Pacific
et al. (2021). ROMY: A multicomponent ring laser for geodesy ocean floors, Geophys. Res. Lett. 35, doi: 10.1029/2008GL034753.
and geophysics, Geophys. Res. Lett. 225, 684–698, doi: 10.1093/ Noe, S., S. Yuan, J. P. Montagner, and H. Igel (2022). Anisotropic elastic
gji/ggaa614. parameter estimation from multicomponent ground-motion obser-
Igel, H., U. Schreiber, A. Flaws, B. Schuberth, A. Velikoseltsev, and A. vations: A theoretical study, Geophys. J. Int. 229, 1462–1473,
Cochard (2005). Rotational motions induced by the M 8. 1 doi: 10.1093/gji/ggac006.
Tokachi-oki earthquake, September 25, 2003, Geophys. Res. Lett. Ortolan, A., J. Belfi, F. Bosi, A. Di Virgilio, N. Beverini, G. Carelli, E.
32, doi: 10.1029/2004GL022336. Maccioni, R. Santagata, A. Simonelli, A. Beghi, et al. (2016). The
Institut de physique du globe de Paris (IPGP), École et Observatoire GINGER project and status of the GINGERino prototype at LNGS,
des Sciences de la Terre de Strasbourg (EOST) (1982). Geoscope, in Journal of Physics: Conference Series, IOP Publishing, 072,003
French global network of broad band seismic stations, doi: pp., doi: 10.1088/1742-6596/718/7/072003.
10.18715/GEOSCOPE.G. Pancha, A., T. H. Webb, G. E. Stedman, D. P. McLeod, and K. U.
Juretzek, C., and C. Hadziioannou (2016). Where do ocean micro- Schreiber (2000). Ring laser detection of rotations from teleseismic
seisms come from? A study of love-to-Rayleigh wave ratios, J. waves, Geophys. Res. Lett. 27, 3553–3556, doi: 10.1029/
Geophys. Res. 121, 6741–6756, doi: 10.1002/2016JB013017. 2000GL011734.

Volume XX • Number XX • – 2023 • www.srl-online.org Seismological Research Letters 11

Downloaded from http://pubs.geoscienceworld.org/ssa/srl/article-pdf/doi/10.1785/0220230202/6122979/srl-2023202.1.pdf

by CNRS_INSU user
 Peterson, J. R. (1993). Observations and modeling of seismic back- Suryanto, W., H. Igel, J. Wassermann, A. Cochard, B. Schuberth, D.
ground noise, U.S. Geol. Surv. Open-File Rept. 93–322, doi: Vollmer, F. Scherbaum, U. Schreiber, and A. Velikoseltsev (2006).
10.3133/ofr93322. First comparison of array-derived rotational ground motions
Poppeliers, C., and E. V. Evans (2015). The effects of measurement with direct ring laser measurements, Bull. Seismol. Soc. Am. 96,
uncertainties in seismic-wave gradiometry, Bull. Seismol. Soc. 2059–2071, doi: 10.1785/0120060004.
Am. 105, 3143–3155, doi: 10.1785/0120150043. Tang, L., H. Igel, and J.-P. Montagner (2023). Single-point dispersion
Saito, M. (1988). DISPER80: A subroutine package for the calculation measurement of surface waves combining translation, rotation and
of seismic normal mode solutions, in Seismological Algorithms: strain in weakly anisotropic media: Theory, Geophys. J. Int. 235,
Computational Methods and Computer Programs, D. J. Doornbos 24–47, doi: 10.1093/gji/ggad199.
(Editor), Academic Press, New York, 293–319, doi: 10.3124/ Tanimoto, T., C. Hadziioannou, H. Igel, J. Wasserman, U. Schreiber, and
segj.64.127. A. Gebauer (2015). Estimate of Rayleigh-to-Love wave ratio in the
Schmelzbach, C., S. Donner, H. Igel, D. Sollberger, T. secondary microseism by colocated ring laser and seismograph,
Taufiqurrahman, F. Bernauer, M. Häusler, C. Van Renterghem, Geophys. Res. Lett. 42, 2650–2655, doi: 10.1002/2015GL063637.
J. Wassermann, and J. Robertsson (2018). Advances in 6c seismol- Tanimoto, T., C. J. Lin, C. Hadziioannou, H. Igel, and F. Vernon
ogy: Applications of combined translational and rotational motion (2016). Estimate of Rayleigh-to-Love wave ratio in the secondary
measurements in global and exploration seismology, Geophysics microseism by a small array at Piñon Flats Observatory, California,
83, 53–69, doi: 10.1190/geo2017-0492.1. Geophys. Res. Lett. 43, 11–173, doi: 10.1002/2016GL071133.
Schreiber, U., and J. P. Wells (2023). Rotation Sensing with Large Ring Todd, K. (2020). toddkarin/global-land-mask: Release of version 1.0.0,
Lasers: Applications in Geophysics and Geodesy, Cambridge University doi: 10.5281/zenodo.4066722.
Press, Cambridge, United Kingdom, doi: 10.1017/9781108524933. UC San Diego (2014). Piñon flats observatory array, available at
Schreiber, K. U., G. E. Stedman, H. Igel, and A. Flaws (2006). Ring https://www.fdsn.org/networks/detail/PY/ (last accessed December
laser gyroscopes as rotation sensors for seismic wave studies, in 2023).
Earthquake Source Asymmetry, Structural Media and Rotation Venkateswara, K., C. A. Hagedorn, J. H. Gundlach, J. Kissel, J.
Effects, Teisseyre, R., E. Majewski, and M. Takeo (Editors), Warner, H. Radkins, T. Shaffer, B. Lantz, R. Mittleman, F.
Springer, 377–390. Matichard, et al. (2017). Subtracting tilt from a horizontal seis-
Scripps Institution of Oceanography (1986). Global seismograph mometer using a ground-rotation sensor, Bull. Seismol. Soc.
network, IRIS/IDA, doi: 10.7914/SN/II. Am. 107, 709–717, doi: 10.1785/0120160310.
Singh, S., Y. Capdeville, and H. Igel (2020). Correcting wavefield Venkateswara, K., J. Paros, P. Bodin, W. Wilcock, and H. J. Tobin
gradients for the effects of local small-scale heterogeneities, (2021). Rotational seismology with a quartz rotation sensor,
Geophys. J. Int. 220, 996–1011, doi: 10.1093/gji/ggz479. Seismol. Res. Lett. 93, 173–180, doi: 10.1785/0220210171.
Sollberger, D., H. Igel, C. Schmelzbach, P. Edme, D. J. Van Manen, F. Wassermann, J., S. Lehndorfer, H. Igel, and U. Schreiber (2009).
Bernauer, S. Yuan, J. Wassermann, U. Schreiber, and J. O. Performance test of a commercial rotational motions sensor,
Robertsson (2020). Seismological processing of six degree-of-free- Bull. Seismol. Soc. Am. 99, 1449–1456, doi: 10.1785/0120080157.
dom ground-motion data, Sensors 20, 6904, doi: 10.3390/s20236904. Yuan, S., A. Simonelli, C. J. Lin, F. Bernauer, S. Donner, T. Braun, J.
Spudich, P., and J. B. Fletcher (2008). Observation and prediction of Wassermann, and H. Igel (2020). Six degree-of-freedom broad-
dynamic ground strains, tilts, and torsions caused by the Mw6.0 band ground-motion observations with portable sensors:
2004 Parkfield, California, earthquake and aftershocks, derived Validation, local earthquakes, and signal processing, Bull.
from UPSAR array observations, Bull. Seismol. Soc. Am. 98, Seismol. Soc. Am. 110, 953–969, doi: 10.1785/0120190277.
1898–1914, doi: 10.1785/0120070157. Zembaty, Z., F. Bernauer, H. Igel, and K. U. Schreiber (2021). Rotation
Stedman, G., Z. Li, and H. Bilger (1995). Sideband analysis and rate sensors and their applications, Sensors 21, no. 16, 5344, doi:
seismic detection in a large ring laser, Appl. Opt. 34, 5375–5385, 10.3390/s21165344.
doi: 10.1364/AO.34.005375.
Stutzmann, E., F. Ardhuin, M. Schimmel, A. Mangeney, and G. Patau
(2012). Modelling long-term seismic noise in various environments, Manuscript received 22 June 2023
Geophys. J. Int. 191, 707–722, doi: 10.1111/j.1365-246X.2012.05638.x. Published online 22 December 2023

12 Seismological Research Letters www.srl-online.org • Volume XX • Number XX • – 2023

Downloaded from http://pubs.geoscienceworld.org/ssa/srl/article-pdf/doi/10.1785/0220230202/6122979/srl-2023202.1.pdf

by CNRS_INSU user