of 8
Evidence
for
Unusually
Strong
Near-field
Ground
Motion
on
the
Hanging
Wall
of
the
San
Fernando
Fault
during
the
1971
Earthquake
Clarence
R.
Allen
Seismological
Laboratory,
California
Institute
of
Technology
James
N.
Brune
Seismological
Laboratory,
University
of
Nevada
at
Reno
Lloyd
S.
Cluff
Pacific
Gas
and
Electric
Company
Allan
G.
Barrows,
Jr.
California
Division
of
Mines
and
Geology
INTRODUCTION
Reports
of
unusually
intense
ground
motions
in
the
very
near
fields
of
faults
that
have
ruptured
during
earthquakes
are
becoming
more
common,
particularly
with
the
markedly
increased
worldwide
strong-motion
instrumentation
in
recent
years
(e.g.,
Heaton
and
Wald,
1994).
The
reported
ground
motions
are
sufficiently
strong
to
have
significant
potential
engineering
impact
(Hall
et
al.,
1995).
In
addition
to
fault
proximity,
two
other
factors
that
have
contributed
to
unusually
high
strong
motions
are
rupture
directivity
(e.g.,
Somerville
et
al.,
1997)
and
locations
on
the
hanging
walls
of
thrust
faults
(e.g.,
Nason,
1973;
Abrahamson
and
Somer-
ville,
1996;
Brune,
1996a;
Brune,
1996b).
Perhaps
nowhere
has
the
sharp
distinction
between
damage
on
the
hanging
wall
and
footwall
of
a
thrust
fault
been
more
dramatically
documented
than
during
the
1945
Mikawa
earthquake,
Japan
(Iida,
1985).
Earlier
reports
of
anomalous
strong
near-
field
vertical
acceleration,
such
as
those
of
Oldham
(1899)
describing
the
great
1897
Assam
earthquake,
have
often
heretofore
been
discounted
or
ignored,
but
they
now
look
more
reasonable
in
the
light
of
recent
well
recorded
events.
One
figure
of
Oldham's
report
was
reproduced
in
Richter
(1958,
p.
51)
and
has
resulted
in
considerable
analysis
and
criticism
relative
to
peak
vertical
accelerations
(e.g.,
Bolt
and
Hansen,
1977).
However,
another
section
of
Oldham's
report,
relating
to
horizontal
ground
velocities
and
displace-
ments,
is
more
important
to
the
phenomenon
reported
herein
and
will
be
discussed
later
in
detail.
This
second
part
of
the
report
has
been
overlooked
by
most
researchers.
We
here
describe
an
intriguing
phenomenon
which
sat-
isfied
all
of
the
above
three
criteria
(near
field,
directivity,
and
hanging
wall
location)
which
occurred
along
the
Tujunga
segment
of
the
San
Fernando
Fault
during
the
M
L
=
6.5
San
Fernando,
California,
earthquake
of
9
February
1971.
The
fault
rupture,
which
was
of
combined
thrust
and
strike-slip
displacement,
crossed
the
grounds
of
the
Blue
Star
Trailer
Park
near
the
mouth
of
Lopez
Canyon
(Figure
1)
and
dis-
placed
a
50
m-long
asphalt
roadway
in
a
manner
suggestive
of
very
high
vertical
accelerations
and
large
amplitudes
of
associated
rapid
horizontal
ground
displacement.
The
local-
ity
was
noted
and
photographed
by
Allen
and
Cluff
a
few
days
following
the
earthquake,
and
it
was
subsequently
mapped
and
trenched
by
Barrows
(1975)
and
his
colleagues.
The
physics
of
the
phenomenon,
here
and
elsewhere,
has
been
the
subject
of
ongoing
interest
by
Brune
(1996),
who
suggested
that
this
episode
be
more
fully
documented.
Earlier
reports
of
unusually
intense
shaking
during
the
San
Fernando
earthquake,
particularly
on
the
hanging
wall
of
the
fault,
include
descriptions
of
shattered
earth
on
ridge
crests
(Nason,
1973;
Barrows,
1975;
Kahle,
1975)
and
the
puzzling
movements
of
a
series
of
6-ton
wire
spools
close
to
the
fault
(Kamb
etal.,
1971).
OBSERVATIONS
The
geologic
and
geophysical
relationships
relevant
to
the
1971
San
Fernando
earthquake
have
been
described
in
detail,
probably
most
completely
in
Oakeshott
(1975).
Just
west
of
the
mouth
of
Lopez
Canyon
(Figure
1),
the
western-
524
Seismological
Research
Letters
Volume69,
Number6
November/December1998
J~P
Granada
Hills
Sylmar
,,~
Pacoima
Dam
/
!
/
/
/
/
/
/
/
/
San
/~
~
/
Little
Tujunga
C.
Fernando
//
"~
i
/
t
Blue
Star
Trailer
Park
~~
~..,
/
/
/
/
/
J
I
I
/
Pacoima
Wash
'~
Hanson
Dam
/
/
/
/
/
/
/
/
Sunland
0
2
4
6
8
km
I
!
I
I
I
9
Figure
1.
Index
map
of
the
San
Fernando
earthquake
area,
showing
the
Blue
Star
Trailer
Park
locality
near
the
mouth
of
Lopez
Canyon.
Heavy
line
is
the
surlace
trace
of
the
San
Fernando
Fault.
most
segment
of
the
Tujunga
segment
of
the
San
Fernando
Fault
passed
through
the
Blue
Star
Trailer
Park,
which
was
founded
on
a
series
of
excavated
benches
and
near-vertical
risers
cut
into
the
underlying
Miocene
Modelo
formation.
The
horizontal
benches
were
mosdy
covered
by
concrete
trailer
pads
and
asphalt
roads
and
driveways.
One
of
these
asphalt
roads
(Figure
2)
was
cut
by
the
fault
at
an
angle
of
about
45
~
leaving
the
asphalt-draped
scarp,
about
0.5
m
high,
shown
in
Figure
3.
The
locality
is
now
buried
by
a
roadfill
for
the
extension
of
Paxton
Street
north
of
Foothill
Boulevard.
The
quandary
presented
by
Figure
3
is
that,
although
the
asphalt
slab
is
relatively
unbroken
and
unrumpled,
the
pipe
railing
along
the
southeast
side
of
the
road
can
be
seen
in
the
photograph
(right)
to
be
broken
and
deformed,
repre-
senting
shortening
by
an
estimated
1.2
m
(Barrows,
1975).
But
this
shortening
occurred
over
the
same
interval
in
which
the
asphalt
clearly
shows
no
such
shortening.
Underlying
the
asphalt,
the
1971
rupture
is
represented
by
two
breaks
about
0.7
m
apart,
although
the
upper
(main)
break
flattens
sur-
faceward
so
that
both
breaks
intersect
the
ground
surface
near
the
base
of
the
scarp.
This
is
known
from
relationships
exposed
in
a
trench
subsequently
excavated
only
a
few
meters
northwest
(left)
of
the
camera
position
(Figure
4),
and
also
from
good
exposures
on
a
near-vertical
face
just
to
the
south-
east
(right)
of
the
camera
position,
where
the
pre-earthquake
rock
facing
collapsed.
The
1971
faults
here
dipped
about
30
~
northeast
(away
from
the
camera),
except
very
close
to
the
surface,
which
is
consistent
with
measurements
elsewhere
along
the
San
Fernando
Fault.
Independent
estimates
of
the
horizontal
shortening
across
the
scarp
in
this
vicinity
were
1.4
m
and
2.0
m
by
Sharp
(1975)
and
1.0
m
by
Kamb
etaL
(1971).
These
estimates
were
based
on
offsets
of
features
other
than
the
pipe
railing.
How
can
the
absence
of
horizontal
shortening
of
the
asphalt
slab
here
be
explained?
There
was
no
significant
deformation
of
the
asphalt
on
the
footwall
side
of
the
scarp
(i.e.,
behind
the
camera
position
of
Figure
3).
But
it
surpris-
ingly
turned
out
that
50
m
northeast
of
the
scarp,
near
the
tree
in
the
background
of
Figure
3,
the
same
asphalt
slab
was
"overthrust"
relatively
northeastward
about
1.75
m
over
the
asphalt
slab
of
a
perpendicular
road
(Figure
5).
Thus,
it
seems
that
the
entire
50
m-long
by
6
m-wide
asphalt
slab,
of
estimated
5
cm
thickness,
was
translated
from
the
rear
(i.e.,
from
the
area
of
the
scarp)
and
essentially
moved
as
a
coher-
ent
unit
1
to
2
m
relatively
northeastward.
The
physics
of
this
process
may
at
first
seem
to
be
a
real
puzzle.
It's
a
bit
analogous
to
the
problem
of
trying
to
push
a
tablecloth
across
a
long
table
solely
by
pushing
at
one
end.
The
fric-
tional
resistance
simply
does
not
allow
it,
and
instead
the
tablecloth
simply
rumples
near
where
one
is
pushing
it.
As
Seismological
Research
Letters
Volume69,
Number6
November/December
1998
525
+
1,75
m
~-"ove
rt h rust"
1.59
m
L.L.
offset
of
groove
cracks
f
f
Trench
~
f
o,~ '~ ~'c'e
0
10rn
L
__1
9
Fioure
2.
Detailed
map
of the displaced asphalt slab (heavy
solid
line), simplified from the map of E H. Weber,
Jr.,
in Barrows
(19/'5,
pl.
4).
Heavy
dashed line is
at
the base of the
1971
scarp. Points
"A"
and
"B"
are,
respectively, the camera locations of Figures 3 and
5.
9
Figure
3.
View
northeast
from
point
"A"
in Figure
2.
Scarp is about
0.5
m high. Note bent and broken pipe at right, which was formerly a straight railing
along
the edge of the road (as in the background)
but
was shortened about
1.2
m by movement
on
the fault, which intersected the ground surface near the
base
of
the scarp. Note
that
the
asphalt shows no such shortening here. Photo by C.
R.
Allen.
526
Seismological
Research Letters
Volume69,
Number6
November/December1998
asphalt
surface
debris
\\
~"~-
.
~--~,.~
~9
71
f a u,
t~,~,~
"~.
~<
~'~
\
0
1
2m
I
I
I
9
Figure
4.
Cross-section
of the
fault
and
scarp
in trench
excavated
near
the
locality
of Figure
3. See
Figure
2 for
exact
location
and
orientation.
Heavy
lines
show
1971
breaks.
Simplified
from
the
trench
log
by A. G. Barrows
and
J. E. Kehle
in Barrows
(1975,
pl.
4).
we
discuss later,
the
phenomenon
appears
less
puzzling
when
recent
phenomena
occurring
at the
hanging-wall
toes
of
thrust
faults
are
considered.
Several
other
observations
are
relevant.
About
midway
along
the
length
of
the
displaced
slab,
on
its
northwestern
margin,
a well
delineated
pre-earthquake
groove
that
crossed
the
slab
margin
nearly
at
right
angles
was
clearly
offset
left-
laterally
1.59
m
(Figure
2)
(Barrows,
1975).
This
is the
most
precise
and nondebatable measurement
anywhere
of
the
slab
offset.
It's
also
interesting
and
coincidental
that
the
fault
dis-
placement
was
a combination
of
thrust
and
left-lateral
strike
slip,
so
that
the
horizontal
component
of
the
motion
vector
(i.e.,
its
bearing)
was
almost
exactly
parallel
to
the
direction
of
the
street,
thus
allowing
the
slab
to
move
without
having
to
force
its
way
out
of
its
constraining
edges. Barrows
(1975)
estimates
this
direction
as
N
50
~ E
from
slicken
sides
exposed
in
the trench;
Sharp
(1975)
reports
N
55
~ E
and
N
56
~ E;
and
Kamb
et
al.
(1971)
calculated
N 61 ~ E
nearby.
The
road
itself,
as
shown
on
E
H.
Weber's
map
in
Barrows
(1975),
trends
N
59
~ E,
which
was
truly
a fortuitous
coinci-
dence.
Concerning
the
friction
along
the
base
of
the
asphalt
slab,
we
can
only
report
that
the
slab
lay
directly
on
the
exca-
vated
and
truncated
upturned
strata
of
mainly
sandstones
and
siltstones
of
the
Modelo
formation,
although
the
trench
revealed
pockets
of
surface
debris
that
presumably
remained
from
the
original
excavation
process.
INTERPRETATION
In
the
years
since
1971,
numerous
informal
discussions
have
taken
place
attempting
to
explain
the
puzzling
relationships
described
above.
Hypothesized
solutions
tend
to
fall
into
five
categories:
(1)
The
observations are
invalid,
misinterpreted,
or
grossly exag-
gerated.
We
rule
out
this
hypothesis
on
the
basis
of
our own
personal
observations
and
the
fact
that
many
people
inspected
the
locality
with
similar
reactions.
Furthermore,
we
feel
that
the
photographs
and
maps
speak
for
themselves.
(2)
Frictional
resistance
to
sliding
along
the
interface
of
the
asphalt
slab
and
the
underlying
materials
was
substantially
reduced
in
some
way
other
than
by
vertical
acceleration.
Of
rel-
evance
in
this
context
are
various
mechanisms
that
have
been
proposed
for
landslide
"lubrication."
Liquefaction
in
the
classical
soil-mechanics
sense
can
be
ruled
out
here
by
the
nature
of
the
underlying
materials
and
the
substantial
local
depth
of
the
water
table.
Movements
of
some
landslides
are
Seismological
Research
Letters
Volume69,
Number6
November/December1998
527