of 11
Potential
Advantages
of
a
Strong-motion
Velocity
Meter
over
a
Strong-motion
Accelerometer
John
F.
Clinton
and
Thomas
H.
Heaton
Department
of
Civil
Engineering,
California
Institute
of
Technology
INTRODUCTION
This
study
examines
whether
it would
be
better
to deploy a
velocity-recording
strong-motion
instrument
in place
of
existing force-balance accelerometers.
The
proposed
instru-
ment
would
be
comparable
to a low-gain version
of
existing
broadband
seismometers.
Using a large suite
of
Earth
signals,
we
compare
such a
hypothetical
long-period
low-gain veloc-
ity
seismometer
(with a
clipping
level set to ±5
m/s)
with
the
existing ±2 g
clipping
Kinemetrics
FBA-23 accelerometer.
We show
that
there are significant advantages in the
deployment
of
the
proposed
instrument
over an accelerometer:
• the velocity
instrument
would
have several orders
of
magnitude
greater
sensitivity
in the
period
band
from
2
s
to several
hundred
seconds.
This
would
allow:
- the
recording
of
long-period
basin response from
regional
earthquakes
as small as
M 3.0
the
recording
of
teleseismic
ground
motions
from
earthquakes
as small as
M 6
which
could
potentially
lead to dense spatial
recording
of
small-amplitude
motions
that
are
not
recorded
by tradi-
tional
strong-motion
networks.
• furthermore,
as well as allowing full recovery
of
ground
acceleration, recovery
of
ground
displacement
is likely to
be more stable from such a
long-period
low-gain
broad-
band
seismometer.
The
most
critical role
of
strong-motion
networks is to pro-
vide on-scale recordings
of
potentially
damaging
motions
over a
broad
frequency
band.
Because
continuous
analog
recording
is extremely expensive
and
strong
shaking
is infre-
quent,
strong-motion
seismographs
were designed to record
only
during
strong
ground
shaking.
Furthermore,
because
of
the
limited
dynamic
range
of
recording
devices, it was
most
efficient to record
ground
acceleration,
since
near-source
strong
ground
motions
have relatively flat
acceleration
spec-
tra in the
band
from 0.3 to 3.0 Hz.
Seismometry
has seen huge advances in the past 30 years.
The
dynamic
range
of
typical
seismometers
has increased
from less
than
5 orders
of
magnitude
to more
than
7,
prima-
rily because
of
the
development
of
force feedback systems
(Iwan
et al.,
1985;
Wielandt
and
Steim, 1986;
Wielandt
and
Streckeisen, 1982). Advances in
recording
systems have been
even more
dramatic;
current
24-bit
digitizers record over 7
orders
of
magnitude
compared
to the 3 orders
of
magnitude
achievable by analog
recording
devices (Trifunac and
Todor-
ovska,
2001a).
The
past 30 years have also seen the
dramatic
development
of
digital
data
communication,
processing,
and
storage,
which
has
prompted
the
development
of
a plan for a
proposed
Advanced
National
Seismic Plan (Benz
and
Filson,
1998;
Heaton
et al.,
1989).
These
new capabilities allow us to
devise new strategies to record
ground
motions.
In this
paper
we discuss
potential
advantages
of
one such strategy, deploy-
ing
continuously
telemetered
strong-motion
velocity seis-
mometers
in place
of
existing triggered
strong-motion
accelerometers. We
anticipate
that
such a
strong-motion
velocity
seismometer
would
essentially be a low-gain version
of
existing
broadband
seismometers,
such as the
Wielandt-
Streckeisen STS-2. We suggest its cost
would
also be similar
to
other
broadband
seismometers.
A
strong-motion
velocity
seismometer
could
record a
broader
swath
of
Earth
motions
than
are
currently
recorded
by existing
strong-motion
accelerometers.
With
a clipping
level
of
±5
mls
(a velocity
magnitude
greater
than
that
of
any
seismic
ground
motion
measured
to date) it
would
recover
on-scale all
motions
relevant to
structural
engineering,
and
it
would
record
long-period
motions
with
accelerations too
small to be
recorded
by
traditional
accelerometers. We esti-
mate
that
direct
recording
of
velocity
with
a dynamic
range
of
140 dB
would
permit
recording
of
broadband
motions
from
regional
and
near-source
events as small as
M
3.0.
These
broadband
motions
could
be used to
study
path
effects such
as the
amplification
of
long-period
motions
by basins. In
addition,
a strong-motion
velocity array
should
be capable
of
recording
teleseisms as small as
M
6.
This
could
potentially
lead to dense spatial
recording
of
small-amplitude
motions
that
are
not
recorded
by
traditional
strong-motion
networks.
332
Seismological
Research
Letters
Volume
73,
Number
3
May/June
2002
Single
differentiation
of
the raw velocity
output
would
produce
the
acceleration
records
currently
used by engineers,
with
the
significant
advantage
that
displacement
estimates
of
ground
motion
would
be
obtained
from a single
integration
of
the raw data.
Assuming
a similar error in the true signal
of
velocity
and
acceleration, single
integration
would
result in
smaller
long-period
error
than
does a
double
integration.
Sin-
gle
integration
thus leads to a more stable
and
conclusive set
of
displacement
time series, especially
with
regard to static
offsets. Even very small baselines or linear trends,
which
are
difficult to isolate and remove in
current
strong-motion
records, can seriously
distort
the
resultant
displacement
after
double
integration,
often leaving its
estimation
more a
matter
of
judgment
rather
than
science (Boore,
2001).
We
note
that
although
resolving
long-period
accelera-
tions associated
with
static
displacements
will be more stable
using a
long-period
instrument,
even small changes in tilt can
significantly
affect the
derivation
of
ground
displacement,
as
pointed
out
by
Trifunac
and
Todorovsky (2001 b).
This
is a
problem
common
to any
inertial
device
and
so affects all
instruments
discussed here. To derive the
translational
dis-
placements
fully in the presence
of
tilt requires
additional
information
from a
colocated
rotational
meter, or a nearby
true
displacement
meter
such as GPS.
To
demonstrate
the increased range
of
Earth
recordings
from the
proposed
device, we
compare
this
strong-motion
velocity
seismometer
with
both
a typical accelerometer, the
Kinemetrics
FBA-23, and a
broadband
velocity
instrument,
the
Wielandt-Streckeisen
STS-2.
The
performance
of
each
device is
illustrated
by showing how
their
dynamic
character-
istics relate to a wide range
of
seismic
motions,
in terms
of
frequency
content
and
acceleration
amplitude.
To
compare
the
instruments,
we first need to describe the
seismographic
system,
which
includes a digital
recording
device as well as
the seismometer.
THE
SEISMOGRAPHIC
SYSTEM
The
range
of
amplitude
and
frequency
recorded
by a
modern
seismographic
system is
controlled
by
both
the
seismometer
and the digital recorder.
Dynamic
Range
of
the
Digital
Recorder
Current
state-of-the-art
digital recorders
employ
24-bit
digi-
tizers.
The
nominal
dynamic
range
of
such a device is
about
140 dB.
Theoretically,
the
dynamic
range can exceed 140 dB
at low frequencies, since
low-frequency
signals are oversam-
pled
and
each
point
is the average
of
many
samples.
This
dynamic
range
enhancement
does
not
occur
where the noise
is characterized
by a power
density
that
increases as frequency
decreases,
i.
e.,
some form
of
1
If
noise.
This
type
of
noise has
a constant
power in frequency bands
of
equal relative
width
(Wielandt
and
Streckeisen, 1982).
Most
electronic
systems,
in fact, are
characterized
by 1
If
noise below 1 Hz,
and
hence
no
resolution
enhancement
occurs (Joe Steim, personal com-
munication,
2001).
In practice, we find
that
under
normal
operating
temperatures
the
dynamic
range can
indeed
increase. For example, the
Quanterra
Q330,
with
135 dB
nominal
dynamic
range, at
26°C
records 136 dB at
10Hz,
up to 142 dB at 0.5 Hz, before
dropping
slightly at lower fre-
quencies (Joe Steim,
personal
communication,
2001).
As this
is
not
a very large difference, we will assume, for the
purposes
of
this paper, a
frequency-independent
constant
dynamic
range
of
140 dB,
approximately
7 orders
of
magnitude.
Dynamic
Range
of
Each
Seismometer
FBA-23
The
clipping
limit
of
the FBA-23
seismometer
is ±19.6
m/s
2
(±2 g) up to its
corner
frequency
of
50 Hz. By
comparing
ground
motions
recorded
simultaneously
with
the FBA-23
and STS-2, we were able to establish
that
the FBA-23 can
resolve
acceleration
above the noise level
of
the
instrument
down to
about
3
X
10-
6
m/s
2
across a
broad
band
of
frequen-
cies (0.01 to 10 Hz).
This
is illustrated
in
FiguresA2
andA3
in the
Appendix,
which
show the
band-passed
records
of
a
M
8.1 event at
2,900
km
epicentral
distance.
The
FBA-23
noise at periods
of
about
100
sand
50 s are
both
of
this level.
For example, in Figure A3, the noise level is
approximated
as
a sine wave
with
a 100 s
period
and
amplitude
5
X
10-
5
mis,
which
is
equivalent
to an
amplitude
of
3.14
X
10-
6
m/s
2
in
acceleration, 136 dB below the clip level
of
±19.8
m/s
2
This
is less
than
the
published
145 dB for the
frequency
range
0.01 to 20 Hz
(http://www.kinemetrics.com)
but
could
also be
due to
limitations
of
the digitizer.
STS-2
Broadband
seismometers
such as the STS-2 have more com-
plex characteristics. For seismic signals
with
periods
shorter
than
the
corner
frequency
(120 s for an STS-2), they typically
have clip levels
that
are given in
both
velocity
and
accelera-
tion; in the case
of
the STS-2 this is ±13
mmls
and ±3.3
m/s
2
(±0.34 g). (Velocity clip levels are often given as
peak-to-peak
values; for the STS-2 this
would
be 26
mmls
peak to peak,
hence the 1/2
peak-to-peak
value is 13
mm/s.)
The
mini-
mum
resolved
motion
for the STS-2 is
published
in the
STS-2
manual
and is shown in Figure 1.
Strong-motion
Velocity
Meter
The
hypothetical
long-period,
low-gain velocity
instrument
would
have a similar type
of
response as the STS-2,
with
a
corner
frequency
at 120 s and clip level at ±5
ml
sand
±49
m/s
2
(±5 g).
Minimum
resolution
is
assumed
to be
140 dB below the clip level, a similar value to
both
the
STS-2
and
FBA-23.
The
dynamic
characteristics
of
these three
seismometers
are
summarized
in Table 1 and
illustrated
in Figure 1.
The
final response
of
the
seismographic
system is similar
to the
instrument
response,
but
the system
dynamic
range at
any
frequency
does
not
exceed 140 dB due to the
limitations
Seismological
Research
Letters
May/June
2002
Volume
73,
Number
3
333