Bull Volcanol (1991) 53:533-545

Volc a ology
9 Springer-Verlag1991

Real-time Seismic Amplitude Measurement (RSAM):

a volcano monitoring and prediction tool
Elliot T Endo I and Tom Murray 2
1 U.S. Geological Survey, APO New York, NY 09697-7002, USA
2 Cascades Volcano Observatory, 5400 MacArthur Blvd., Vancouver, WA 98661, USA

Received April 20, 1990/Accepted February 2, 1991

Abstract. Seismicity is one of the most commonly mon- available for the three dome-building episodes, is com-
itored phenomena used to determine the state of a vol- pared to RSAM data to show that the increase in
cano and for the prediction of volcanic eruptions. Al- RSAM data was related to the transport of magma to
though several real-time earthquake-detection and data the surface. Thus, if tiltmeter data is not available,
acquisition systems exist, few continuously measure RSAM data can be used to predict future magmatic
seismic amplitude in circumstances where individual eruptions at Mount St. Helens. We also recognize the
events are difficult to recognize or where volcanic tre- limitations of RSAM data. Two examples of RSAM
mor is prevalent. Analog seismic records provide a data associated with phreatic or shallow phreatomag-
quick visual overview of activity; however, continuous matic explosions were not preceded by the same in-
rapid quantitative analysis to define the intensity of creases in RSAM data or changes in tilt associated with
seismic activity for the purpose of predicting volcanic the three dome-building eruptions.
eruptions is not always possible because of clipping
that results from the limited dynamic range of analog
recorders. At the Cascades Volcano Observatory, an
inexpensive 8-bit analog-to-digital system controlled by Introduction
a laptop computer is used to provide 1-min average-
amplitude information from eight telemetered seismic Premonitory earthquake activity is a key indicator of
stations. The absolute voltage level for each station is the state of a volcano and the potential for volcanic
digitized, averaged, and appended in near real-time to a eruptions (Decker 1973, 1986; Malone et al. 1983; Mi-
data file on a multiuser computer system. Raw real- nakami 1960, 1974; Minakami et al. 1969; Swanson et
time seismic amplitude measurement (RSAM) data or al. 1985). Provided discrete earthquakes can be recog-
transformed RSAM data are then plotted on a common nized, the type and intensity of seismic activity is often
time base with other available volcano-monitoring in- expressed by the number of earthquakes per unit time,
formation such as tilt. Changes in earthquake activity by earthquake magnitude, seismic-energy release, b-val-
associated with dome-building episodes, weather, and ue, or as earthquake source parameters for the purpose
instrumental difficulties are recognized as distinct pat- of predicting volcanic eruptions.
terns in the RSAM data set. RSAM data for dome- For Mount St. Helens the Cascades Volcano Observ-
building episodes gradually develop into exponential atory (CVO) relies on the Geophysics Program, Univer-
increases that terminate just before the time of magma sity of Washington for routine earthquake information
extrusion. Mount St. Helens crater earthquakes show such as hypocenter locations, magnitudes, and focal
up as isolated spikes on amplitude plots for crater seis- mechanism solutions to determine the state of the vol-
mic stations but seldom for more distant stations. cano. Although highly desirable, access to near real-
Weather-related noise shows up as low-level, long-term time (within a few minutes) seismic information was
disturbances on all seismic stations, regardless of dis- not available prior to 1985. An earthquake count of
tance from the volcano. Implemented in mid-1985, the CVO analog seismic records that were changed every
RSAM system has proved valuable in providing up-to- 3 h provided the most up-to-date information on seis-
date information on seismic activity for three Mount St. mic activity. Manual counts were appended to a file on
Helens eruptive episodes from 1985 to 1986 (May 1985, a multiuser computer, where data could be compared
May 1986, and October 1986). Tiltmeter data, the only with other volcano-monitoring measurements such as
other telemetered geophysical information that was tilt, sulphur dioxide output, total magnetic field, and
EDM (electronic distance measurement) line changes.
Offprint requests to: E T Endo Such an earthquake count provided only minimal in-
534

formation on the current state of seismic activity be- succession of earthquakes that could not be counted in-
cause it was at least several hours out of date at all dividually (Fig. lb). The inability to count earthquakes
times. While the cumulative earthquake count for any and to measure coda lengths resulted in an inaccurate
swarm associated with an eruption is an approximate estimate of seismic energy released before the eruption.
measure of intensity (proportional to total seismic en- One method used to reduce inaccuracies was to rely on
ergy provided earthquakes are similar in magnitude) of more distant seismic stations not affected by the in-
earthquake activity, the count is subject to inconsisten- crease in smaller magnitude earthquakes. However a
cies of analog seismic record interpretation, questiona- potential problem with using another seismic record for
ble gain changes, and loss of data due to overlapping earthquake counts and coda measurements was the bias
traces, clipped seismic signals, missed record changes toward counting only larger magnitude events and in-
or telemetry failure. Thus, earthquake counts prior to troducing a systematic error in seismic energy estimates
and during an eruption are not always quantitatively re- utilizing coda-magnitudes. What was needed was a
liable. monitoring technique that was capable of providing an
Premonitory seismicity associated with several easy-to-access and plot continuous time history of rela-
Mount St. Helens eruptive episodes during 1984 was tive seismic intensity related to eruptions.
characterized by intense earthquake swarms that began This paper describes a real-time seismic amplitude
with discrete earthquakes recorded on seismic stations measurement or RSAM system that provides the con-
located less than 1.5 km from the center of the dome, tinuous high-resolution real-time measurements that
then progressed to continuous high-amplitude signals. met our requirement to produce easily repeated meas-
The intense seismicity associated with the eruptive epi- urements of relative seismic intensity from one volcanic
sode of 10 September 1984 was a motivating factor in event to the next. Furthermore, in digital form the data
implementation of a real-time seismic amplitude meas- are easy to plot with the aid of a computer and com-
urement system. In less than 24 h seismicity progressed pare with other real-time data, such as tilt or geochemi-
from discrete events (Fig. la) to an almost continuous cal-monitoring data. The amplitude measurements ac-
curately define the relative intensities of eruptive epi-
sodes regardless of the type of seismic activity (discrete
earthquakes, closely spaced earthquakes, volcanic tre-
1 MINUTE mor, or mudflows). However, the present system does
not provide spectral information like the system de-
scribed by Hurst (1985) or Nishi (1987). For event dis-
crimination, the lack of spectral information with
RSAM data is a limiting factor in identification (with-
out an analog seismic record) of volcanic tremor, fuma-
rolic noise, wind noise, or noise generated by other
sources.
The idea of monitoring the mean amplitudes of seis-
mic records is not new. Sassa (1936) utilized daily mean
amplitudes to describe the state of volcano Aso in Ja-
pan more than 50 years ago. Nakamura (1977) briefly
described a similar technique for measuring acoustic
emissions associated with fracturing using an RMS
(root-mean square) voltmeter. Although the initial ap-
plication of the RSAM system did not consider moni-
toring of harmonic or volcanic tremor, RSAM is ideally
suited for continuous measurement of mean amplitude
as suggested by Schick et al. (1982) and Schick (1988) to
document the seismic time history associated with vol-
canic eruptions. Schick obtained an analog recording
of the mean envelope amplitude by feeding the demo-
dulated output from a magnetic tape recording to a
bridge rectifier and a low pass filter. Whatever the tech-
nique used to measure seismic amplitudes, it is impor-
tant to understand how the measurements are related to
the process being monitored.
To clearly show that RSAM data can be used for
monitoring and predicting dome-building episodes at
Fig. 1. A A section of 8 September 1984 analog seismic record for Mount St. Helens, we compare RSAM data with tele-
the seismic station GDN located 700 m N of the geometric center
of the lava dome. Lines are spaced 15 min apart and the section is metered tiltmeter data to show changes in RSAM data
about 3 min wide. B A section of 9 September 1984 analog seis- are related to the migration of magma from shallow
mic record for GDN showingthe intense seismic activitythat pre- depth to the surface of the earth. Because of the impor-
ceded the 10 September 1984 dome-buildingeruption tance of tiltmeter data to our interpretation of RSAM
 535

•PAK o

Fig. 2. The physiography of Mount St.

Helens, Washington, and locations of
seismic stations (open triangles) and tele-
metered tiltmeter sites (closed circles).
Seismic station ERT, located 16 km to
the W, has not been plotted. SHW, SOS,
JUN, HSR, and EDM are maintained
by the University of Washington. All
other seismometer stations and tilt meter
sites are maintained by CVO. Tiltmeter
-46~10' ~ L ~ r176176176 S I DIg~'Yk-'~ MS stations EMU, TNKR, and RUBY, were
operational in May of 1985. SAUN and
RUBY were operating in May 1986.
MOUNT ST. HELENS ' ' b y . swG'~uue ~ ,~ FAMS, OOPS, RAZA, and S W L P were
operating in October 1986. LONE is the
%. I, ~176 current location of a gas sensor and dis-
placement meter. The Mount St. Helens
, 5 KM , [ "'l. 1 KM /I lave dome is indicated by the 2000 m to-
pographic contour

data, we give a detailed description of tiltmeter instru- At CVO, audio carrier outputs from radio receivers
mentation and tiltmeter data. Two examples of RSAM are demodulated by a bank of discriminators. Discrimi-
data associated with shallow explosions are included in nator outputs were adjusted for a maximum of +2.5
this paper to point out the limitation of RSAM data in- volts output for + 125 Hz deviation of carrier signals.
predicting phreatic or phreatomagmatic explosions. Discriminator outputs are fed to an absolute value cir-
cuit that in turn is interfaced with the 8-bit analog-to-
digital (A/D) system described below. Unlike record-
Seismic network and instrumentation ers, where pen driver gains are easily adjusted by a
switch, inputs to the A / D converter are fixed and can-
Seismic information at the Cascades Volcano Observa- not be adjusted without an electronic circuit modifica-
tory is provided by analog signals from three stations tion of either the discriminator or the input-signal con-
located in the 1980 crater, station NSP (destroyed in ditioning of the A / D converter, thus preventing undo-
April 1986, replaced by DIO in September 1986 and cumented changes in recording sensitivity.
later by REM in January 1987), GDN, and YEL (Fig.
2). Five additional stations that provide information
(HOA, SHW, EDM, PAK, and ERT) are located out- RSAM instrumentation, data acquisition, analysis, and
side the crater at distances of as much as 16 km from interpretation
the vent. Six stations are maintained by the Cascades
Volcano Observatory and the remainder are maintained R S A M instrumentation
by the University of Washington.
Each seismic station consists of a l-Hz vertical seis- The real-time seismic-amplitude measurement system
mometer, a preamp/voltage-controlled oscillator, and a consists of a Tandy (Radio Shack) model 100 laptop
low-output-power VHF or U H F radio frequency trans- computer, an Elexor data-acquisition unit, and an in-
mitter (100-150 milliwatts). External resistance for each house designed signal conditioner/multiplexor board
seismometer station was adjusted for 0.7-0.8 critical (Fig. 3). Mounting the in-house board inside the Elexor
damping. Typical loaded-motor constants are approxi- unit keeps the total size of the analyzer to a minimum
mately 1.0 V/cm/s. The total response of the system is (32 • 24 x 12 cm). The model 100 and the Elexor units
similar to that used by the US Geological Survey in are designed to run on either internal or external batter-
California (Lee and Stewart 1981) and Hawaii (Koya- ies. With a total current draw of less than 100 ma at
nagi et al. 1975). The 3-db points for the system elec- 12 V, the entire unit can be powered easily by a single
tronics are 0.1 Hz and 25 Hz. The displacement re- 12 V car battery. The signal conditioner/multiplexor
sponse peaks at 15 Hz. Frequency-multiplexed seismic- buffers the eight seismic signals, removes any dc com-
carrier signals are transmitted from Mount St. Helens ponents, and amplifies the signals to take full advan-
to CVO in Vancouver, Washington ( > 75 kin) via sev- tage of the range (0-5 V) of the analog-to-digital con-
eral radio repeater links. verter (A/D). A CMOS (complementary metal oxide
536

I SEISMIC
DISCRIMINATOR I
2.0 SEC I
SOFTWARE
SAMPL'N~AT
50 SAMPLES
PER SECOND
FLOW
[-1 CHART

RESET COUNTERS I
BEGIN SAMPLING
OUTPUTS RSAM !
BLOCK DIAGRAM
0.2 SEC RUNNING TOTALS,
CHECK THRESHOLDS

ooFILTER
F . I J oLL w,vEI 1
I ] I NO
DATA ACQUISITION BOARD

SYSTEM BUS

I
&&J.I,J.I,J,
MODEL 100 LAPTOP COMPUTER
I AVERAGES
co.oTe H DATA I
Fig. 3. Blockdiagram of RSAM systemfrom the analog inputs to Fig. 4. Flow chart indicating software actions in the data acquisi-
laptop computer system tion process and the preprocessing of data prior to transmissionto
a minicomputer

semi-conductor) multiplexor selects the desired seismic The sums returned by the data-acquisition program
signal for digitizing. The negative component of the sig- are checked against thresholds set by the user. If any of
nal is converted to a positive voltage by a full-wave rec- the thresholds are exceeded, a buzzer is activated and
tifier before it is digitized by the 8-bit A/D. Control of LEDs corresponding to the station or stations that have
the multiplexor and acquisition of the data are accom- exceeded their thresholds are lit. They remain lit and
plished by using the model 100's system bus, freeing the the buzzer continues to sound until the sums fall below
computer's printer, modem, cassette, and RS-232C their thresholds. A running total of the returned sums
ports for connection to other peripherals. An 8-bit digi- for each channel is kept throughout each minute. At the
tal-output port of the Elexor unit controls the eight end of the minute, these running total are divided by
LEDs and buzzer for the threshold alarms. the number of samples taken in the minute. The result-
ing averages and the time are transmitted to the mini-
computer via the RS-232C port of the model 100. The
R S A M data acquisition and analysis running totals and the number of samples are reset to
zero and data acquisition for the next minute is re-
The model 100 computes the average absolute signal sumed. When 1-min averages have been computed and
amplitude once per minute for each input channel by sent to the minicomputer, stored data have been con-
dividing the sum of each channel's digitized samples by verted from counts which have a maximum of 255 for
the number of samples. Taking an average once per the 8-bit A / D system, to average volts. Digital counts
minute allows the cessation of data acquisition for were multiplied by 10 to allow plotting at an optimal
short periods in order to process the data. This greatly scale. Discriminator outputs are +2.5 volts peak-to-
simplifies programming because data acquisition and peak, but only absolute values are recorded. The maxi-
processing do not have to be performed concurrently. mum voltage range of 0 to 2.5 volts for discriminator
The analyzer's software can be divided into four pro- output and a 0 to 5 volts input range for the A/D, result
grams: data acquisition, threshold detection, average in a measurement resolution of about 20 my, or one
computation, and data transmission. The speed neces- half the typical peak-to-peak background noise level
sary to sample eight channels at 50 times per second for each seismic station. Figure 5 shows a block dia-
required the data acquisition program to be written in gram of the current configuration of the telemetered
machine code. The resident language (BASIC) is used volcano-monitoring system at CVO. Network receivers,
in the other programs where speed is not a requirement. the real-time seismic amplitude monitor, and the mini-
These programs take about 7 s/min., leaving the other computer are located at the observatory in Vancouver,
53 s/min, for data acquisition (Fig. 4). When the data Washington. The minicomputer functions as the pri-
acquisition program is called from BASIC, it collects mary data storage device and also provides a multiuser
data for 2 s (50 samples/channel) and returns the sum environment for real-time data analysis of all teleme-
of the samples for each channel. Because it is written in tered data.
machine code, it is very efficient and takes only about
1 ms to get one sample for each of the 8 channels. At 50
samples per s, 95% of the sampling time is spent waiting
to take the next group of samples. This time is available
for additional processing.
 537

DATA ACQUZSTTZON
CENTRAL-
ANALYSES
COMPUTER
GAS S E N S O R
DISPLACEMENT
METER

LO-t I I
TZLTMETER
GROUND-TEMP
SENSOR
DATA-RATE
NETWORK 1
KEY- dl EDN
RECEIVER SERIAL BOARD
TZLTMETER
ENTRY SPIRIT
TZLTMETER LEVEL
SEISMIC- TILT
AMPLITUDE
METER
RADIO LINKS
EARTHQUAKE
SEZSNONETER
COUNTS

SEISMIC- COSPEC
SEZSMOMETER NETWORK
RECEIVERS

SEESNOMETER

Fig. 5. Block diagram of Mount St. Helens telemetry monitoring system operated by CVO. The keyboard entry of monitoring data into a
common data base is no longer done on a routine basis

R S A M data interpretation (Norwack and Aki 1984). A formal approach is to look

at the statistical change in the average-amplitude ratios
Each seismic station uses a velocity transducer (seis- of stations located at different distances from the con-
mometer) with a free period of 1 Hz; thus the measured duit system (amplitude ratios for a number of stations
output is proportional to velocity or V/cm/s for fre- and perhaps the average ratio over time). A 2-d gaus-
quencies above 1 Hz. If the absolute response of each sian beam synthetic model (Norwack and Aki 1984) for
transducer were known, then each 1-min average would Mount St. Helens predicts a significant increase in am-
represent the average ground velocity detected by the plitude ratio for earthquakes that occur at a 1.0-km
seismometer at that seismic station for that minute. depth compared to earthquakes that occur at a 3.5-km
The average ground velocity of seismic waves has no depth. For earthquakes that cannot be located accurate-
simple relationship to commonly measured earthquake ly, alternate interpretations of increasing seismic ampli-
parameters such as magnitude, seismic energy, and tude are: (1) the source was getting closer to the seismic
seismic moment. At this time, we do not know the pre- station in a horizontal plane and (2) there was an in-
cise relationship of electrical energy to seismic energy crease in the magnitude of earthquakes from a fixed
as defined by Gutenberg and Richter (1956) or Richter source. At Mount St. Helens, eruption-related earth-
(1958), or how it is related to seismic moment (Aki and quake sources are mostly confined to a region close to a
Richards 1980). However, the output voltage of the dis- small magma chamber and a narrow conduit system
criminator is an easily understood quantity that can be (Scandone and Malone 1985). Thus the source of seis-
accurately measured. The square of the measured vol- mic energy prior to eruptions is approximately fixed in
tage is proportional to electrical power (Korneff 1966). space horizontally. For more distant seismic stations,
It is a simple measurement that can be repeated, pro- small changes in vertical position are of little conse-
vided there are no significant changes of instrumenta- quence.
tion. The physical significance of this power is that it is The seismic data collected from 1985 to 1986 for
proportional to the electrical power generated by the three dome-building episodes showed similar exponen-
geophone responding to the shaking of the Earth at the tial increases in RSAM data. Rapid changes in teleme-
seismometer site. Total electrical energy is obtained by tered continuous tiltmeter measurements indicated that
integrating measured power over time. significant deformation of the dome and (or) nearby
The relative changes in seismic wave amplitude (av- crater floor coincided with the rapid increases in seis-
erage ground velocity) with distance may give an indi- mic activity in the 24-48 h preceding extrusion. By con-
cation of average depth changes for earthquake sources trast to the nearly identical increases in RSAM data just
538

prior to the extrusion, RSAM data for the surface phase sites have been lost to rockfalls or avalanches resulting
of dome-building extrusions vary considerably for the from deformation of the dome. Still others have been
three examples presented. Two dome-building erup- buried by newly extruded lava lobes. Figure 2 shows
tions were accompanied by thrust faulting of the crater the locations of tiltmeters in operation during the three
floor and one resulted in large rockfalls and avalanches eruptive episodes monitored by the RSAM system.
from the lava dome. Avalanches and rockfalls asso- Tiltmeter data are telemetered to the Cascades Vol-
ciated with individual eruptions depended on the mor- cano Observatory via a USGS low-data-rate telemetry
phology of the dome (where extrusion occurred) and system. At programmed intervals (usually 10 min), 13-
the volume of extrusion. Unlike earthquakes, seismic bit temperature and tilt data from each site are teleme-
signals generated by avalanches and rockfalls were tered in digital format. Digital information collected by
usually not recorded at stations located more than a set of laptop computers is appended to a file on a
10 km from the event. multiuser minicomputer. Like RSAM data, tilt data are
Plots of RSAM data provide additional details of available on the multiuser minicomputer for routine
the time history of eruptive episodes previously not de- plotting with other volcano-monitoring data on a com-
tected in routine earthquake counts at station YEL. En- mon time base.
ergy plots of earthquakes and surface events suggested Since mid-1981, there have been 12 extrusive epi-
that an eruptive cycle has two possible phases (Malone sodes at Mount St. Helens during which tiltmeters have
et al. 1983). The first phase was thought to be asso- been in operation. Early installations were intended to
ciated with increasing stress in country rock due to the determine the best locations to monitor tilt changes.
movement of magma toward the surface. The second During the eruptive episodes described in this paper,
phase consisted of surface events associated with dis- outward tilt of the flanks of the dome accelerated in the
ruption of the dome (avalanches, rockfalls) and the last 24 to 48 h prior to extrusion. During 1981-1983,
growth of a new lobe on the dome. Although useful in- real-time plotting of telemetered tilt data was possible
ferences can be drawn from seismic data and its rela- with some difficulty using a minicomputer dedicated to
tionship to processes, seismicity itself does not neces- logging the tilt data. In mid-1984, repeated hardware
sarily indicate mass transfer from depth to the surface. failures spurred the writing and implementation of soft-
Surface deformation, such as changes in EDM lines ware to write tilt data directly to files on the CVO VAX
and tilt provided stronger evidence for the effects of 11/750 computer system.
magma movement. In 1985 a program called BOB (Murray and Endo
1986) was written to plot equally spaced time-series
data utilizing more efficient direct-file access available
Tiltmeter instrumentation and data on the VAX computer system. Plot parameters for mul-
tiple-frame plots can be easily changed interactively.
Electronic tiltmeters provide a convenient method to Since 1985, it has been relatively easy to plot various
measure surface deformation associated with the extru- geophysical data on a common time axis using BOB
sion of magma. For this paper, electronic tiltmeters and a plot-command file. The VAX computer system is
have provided data to support interpretation of RSAM also accessible via dial-in ports or any one of the many
data. Several types of platform tiltmeters were used at terminals linked to the computer system.
field sites in the 1980 crater. In most cases, single-axis
tiltmeters were used to detect tilt radial to the center of
the dome. A few installations had dual-axis tiltmeters Real-time Seismic Amplitude Measurement for three
that were oriented to detect radial and tangential tilt. dome-building eruptions at Mount St. Helens and
Current practice at Mount St. Helens is to install tilt- comparison to tilt measurements
meters on the dome or within 1 km of the center of the
dome. The proximity of tiltmeters to the point of extru- The May-June 1985 eruptive episode
sion commonly yields tilts in excess of 10000 microra-
dians. These large tilts reduce concern over diurnal ef- Narrative. From February 1985 to spring 1987, the
fects, such as those caused by temperature changes, RSAM system recorded three dome-building episodes
which preculde early detection of much smaller tilts at Mount St. Helens. The first, which occurred in late
(less than a few microradians) associated with early May 1985, caused substantial disruption of the dome
pre-eruption deformation. Sites on the dome have incli- due to endogenous dome growth. A 250 • 500 m graben
nometers to measure tilt beyond the range of tiltmeters. oriented east-west formed in the south summit area of
A typical installation includes a single-axis tiltmeter the dome accompanied by intense crater floor deforma-
with a 5 microradian sensitivity and 5000 microradian tion (thrust faulting, transverse fractures, etc.) of the
dynamic range, and an inclinometer with a 300 micro- south crater floor. Although the total volume of in-
radian sensitivity and 30000 microradian dynamic truded and extruded magma was about to 7 • 106 m 3,
range. less than 1 • 106 m 3 was extruded. Formation of the
A negative aspect of dome installations is the fre- graben structure and associated crater floor deforma-
quent loss of tiltmeters to the effects of eruptions. De- tion accounted for an extended period of seismicity
bris from small explosions prior to or accompanying (more than one week) following the initial phase of in-
eruptions have damaged or destroyed equipment. Other tense earthquake activity.
 539

O3
E--, F r ~ r r ~ ~ 1 5ool- . . . . . . ) ; " ; ~: " " ;: "-.., . . . . . . . . . . . . .
S0~- GDN EARTHQUAKE COUNT ~,I ~4~176 A f i \YET.EARTaQUAKEcousT
3ooF / : \
/ i\
d looF- --'-/~ "
o'l'/i . . . . . . . . . . . . i . . . . ,-T'~ ~ ~ +-.-,, .....
Z o 15oo~- ............. ,. . . . . . . . . . . . . . . . J
2500 IYEL 10 MIN RSAM DATAI~ ~ WIND NOISE[
i000 ~{LE, '~ -I
150o[ ~ h l I. . . . . I , } , I [-{
,oool- , I ,l ,l, l ,I,,H
o o
r..) 250 ,/ ~1500 SHW I0 MIN RSAM DATA l DOME INTRUSION
,.~ 200t GDN HOUR AVERAGE RSAM DATA r162
,000 ,,,/ START
r.~ 150F r~ r~ / ' C iV--- ..
loo~ ,. . . . AA A , J~/

o ~5o o ,:'-: .;,.~,,~.-!'..~.:'~':"~.

ERT 10 MIN RSAM DATA
,'-e'(, ! , , , ,' ,'r'~. : .-~.'....~....-;, |
139 140 141 142 143 144 145 146 147
TIME - JULIAN DAY 1985
iooo , j
Fig. 6. A Plot of hourly earthquake counts from station G D N
analog seismic records from 19 May (Julian day 139) to 27 May
500 D ~
(Julian day 147), 1985. B Plot of 10-min RSAM data for station
G D N or raw data for the same time period as A. Maximum pos- 135 138 141 144 147 150 153 156 159 162 165
sible value is 2550 counts. C Plot of the hourly average of 1-min TIME - JULIAN DAY 1985
RSAM data during the same time period as Fig. 6a and b
Fig. 7. A Plot of manual daily counts for seismic events recog-
nized as earthquakes on station YEL analog seismogram from 15
May (Julian day 135) to 14 June (Julian day 165), 1985. B Plot of
RSAM d a t a . During mid-May 1985, crater seismic sta- 10-min RSAM data for station YEL from 15 May (Julian day 135)
to 14 June (Julian day 165), 1985. C Plot of 10-min RSAM data
tions GND and YEL showed an increase in the total for station SHW. Although SHW is located about 3 x the YEL
number of earthquakes recorded daily. On 16 May (Ju- distance from the dome, the SHW 10-min plot mimics the YEL
lian day 136), the earthquake count for seismic station 10-min plot. D The bottom frame is a plot of 10-min RSAM data
YEL exceeded ten events per day compared to less than for the ERT seismic station. Like SHW, ERT 10-min average
one event per day during March-April. By 1800 UT 19 mimics RSAM plot for station YEL at a lower amplitude. The
May (Julian day 139), the count at station GDN was as period from 25 May (Julian day 145) to about 0000 U T 28 May
(Julian day 148, 1700 PDT) was the period of intense seismicity
much as 9-10 events per hour (Fig. 6a). Criteria for an associated with the magma-migration phase of the May 1985
event that was counted consisted of a minimum peak- eruption. The period from 0000 U T 28 May (Julian day 148) to
to-peak amplitude of 8 mm and-or a minimum coda 5-6 June (Julian days 156-157), and labeled dome intrusion was a
duration of 10 s. GDN is located about 700 m north of period of extended seismic activity thought to be associated with
the geometric center of the dome. Figure 6b and c illus- the dome's graben formation
trate just two of several types of plots possible with
RSAM data. Each average value for 10 min is plotted {
as a bar in the middle frame. In the lower frame, 1-h clipping amplitudes, the maximum number of individ-
average RSAM data was plotted as a line graph. During ual earthquakes that could be identified on the seismo-
19-25 May (Julian days 139-145), the hourly event grams was about 500 per day (Fig. 7a). YEL is located
count from the RSAM data mimicked the manual about 1200 m north of the geometric center of the
hourly count of the seismogram for station GDN. dome. While overlapping seismic signatures limited the
Hourly earthquake counts were made on seismograms maximum daily earthquake count to about 500 events,
with a rotation rate of 120-240 mm/min and a transla- RSAM data continued to provide meaningful informa-
tion rate of 8 mm per revolution. The decrease shown tion on seismic activity throughout the eruptive epi-
in the manual earthquake counts at GDN seismic sta- sode. Plots of 10-min RSAM data for the period brack-
tion does not show up as strongly in the RSAM count eting the eruption are shown in Figs. 7b-d. Data are
because of the increasing maximum amplitudes of indi- from seismic stations YEL, SHW, and ERT. Although
vidual earthquakes. Inclement weather prevented field these stations are spaced almost 16 km apart, the plot
observations of the start of extrusion, which occurred for the most distant station (ERT) mimics the plot for
between 27 May and 29 May (Julian days 147 and the nearest station (YEL). On 7 June 1985 (Julian days
149). 158-159), a strong weather front moved through the
Figure 7 illustrates the advantage of using RSAM area and produced increased average amplitudes at all
data instead of manual earthquake counts during peri- seismic stations.
ods of intense earthquake activity. Daily counts were
made on seismograms with a rotation rate of 60 m m / T i l t d a t a . The May 1985 eruptive episode was the first
min and translation rate of 2.5 mm each 15 min. Due to to be recorded by the RSAM system. There were three
the rotation rate and the translation rate, and the near tiltmeters (Fig. 2) operating at the time on or near the
540

3000 determined by field observation. The period of intense

Z
< * YEL RSAM DATA seismicity thought to be associated with rapid magma
[]EMU TILT -~ 10 migration lasted less than 24h (Endo et al. 1987).
Within 24 h this was followed by a period of rockfall-
o
2000 avalanche activity from the growing lobe of lava.

RSAM data. Although the eruption of May 1986 was

0 preceded by almost a month of small explosions and
other precursor seismic activity, much of the migration
Z I000
of magma from a depth of 2-3 km to the surface took
o
place in the last 24 h before extrusion began (Endo et
al. 1987). Earthquake counts for seismic station YEL
exceeded 10 per day starting on 25 April (Julian day
0
++I+ i~" ~i + ~ + +" II " ff . . . . I I
115). Seismic station SHW showed a significant in-
140 141 142 143 144 145 146 147 148 149 150 crease in seismicity on 30 April (Julian day 120), 10
TIME - J U L I A N DAY 1 9 8 5 days prior to extrusion (Fig. 9). The magma migration-
Fig. 8. Hourly average RSAM digital counts for seismic station extrusion phase (from rapid increase in tilt and begin-
YEL and tilt for EMU tiltmeter station for the May 1985 dome- ning of migration of mean hypocenter depth to extru-
building eruption plotted on a commonhorizontal scale. Tilt data sion) showed up as a well-defined peak in the RSAM
scaled for plotting adjacent to the RSAM curve plot (Fig. 10) separated from the surface phase of the
eruption by a relatively quiet period of 24 h. The sur-
face phase was characterized by continuous rockfall-
dome. Both RUBY, on the north flank of the dome, and avalanche activity associated with the growing lobe and
EMU, in the south summit area of the dome, showed a decrease in size and number of earthquakes. Another
rapid increase in tilt at the same time as the RSAM data characteristic of the surface phase was a swift decrease
increased in absolute average amplitude. Figure 8 in RSAM amplitude when the lobe stopped growing
shows tilt at the EMU site and RSAM data from seis- rapidly on 10 May (Julian day 130). The 2-3 day dura-
mic station YEL plotted on the same time scale. The tion for both the migration-extrusion phase and the sur-
tiltmeter at the TNKR site, located on the crater floor face phase contrasts with the more than 10 days of seis-
about 750 m north of the center of the dome, was unre- mic activity associated with the May 1985 eruption.
liable during this period. During 26-27 May (Julian
days 146-147), both YEL RSAM data and EMU tilt Tilt data. By May 1986, RSAM hardware had been
measurements showed rapid increases in data values wired for eight seismic stations located around Mount
that we assume were associated with the rapid migra- St. Helens. Tiltmeters were operating near Sauna on the
tion of magma up the conduit system. By 28 May (Ju- crater floor north of the dome (SAUN), and at tilt site
lian day 148), EMU showed additional rapid tilt asso- RUBY on the north flank of the dome (Fig. 2). The pro-
ciated with the development of the graben. Early on 29 gress of the changes in seismicity and tilt (coseismic de-
May (Julian day 149), telemetry from the EMU site ter- formation) were easily tracked by repeated plotting. A
minated due to intense disruption of the ground around plot of RSAM data for seismic station YEL and SAUN
the station. The net tilt at EMU as about 33 000 micro- radial tilt from the SAUN tiltmeter site showed coinci-
radians, north down.

2000 I

The May 1986 eruptive ep&ode A YEL H O U R AVERAGE R S A M DATA [

, IJ 1
Narrative. The early May 1986 eruptive episode pro- r.,ooo F 11 - A 1
duced a dacite lobe about 250 m wide and 500 m long
near the center of the dome. The estimated volume of ~

intruded and extruded material is about 9 x 106 m 3. The F .O R AVERAGE , t

episode was preceded by a month of almost daily small
explosion events. One explosion event (accompanied
by discharge of gas, ash, and other vent debris; see
cover photo of Science, vol. 221, no. 4618 for the pho-
tograph of an example) on 16 April (Julian day 106) 104 107 110 113 116 119 122 125 128 131 134
T I M E - J U L I A N DAY 1 9 8 6
disabled the first radio-telemetered seismic station NSP
on the dome. A subsequent explosion on 19 April (Ju- Fig. 9A-B. RSAM plots of hourly average data for stations YEL
lian day 109) tossed lava blocks more than 1 km north and SHW from 14 April (Julian day 104) to 14 May (Julian day
of the dome and was accompanied by three small de- 134), 1986. The magma migrationphase of the May eruption is the
interval indicated by M and the surface phase of the eruption was
bris flows from the dome; resulting hot rocks were dis- associated with the RSAMpeak after the interval indicated by M.
lodged from the dome by the explosion. As in the May Spikes on the left side of the YEL plot were associated with tele-
1985 episode, the exact time of extrusion could not be metry noise on the YEL radio link
 541

2000 west sides of the dome, more or less along the extension
X
< * YEL RSAM DATA of the May 1985 graben, were notable features of the
A SAUN TILT episode (Swanson and Holcomb 1989).
1500 EXTRUSION
START RSAM data. The first indication of an impending erup-
u
tive episode was a sequence of earthquakes that began
on 13 September (Julian day 256), 1986. While station
o i000

z
~~}~~~,.,~j
~ { + L//AVALPAHNACHSE YEL showed a count of about 20 earthquakes that day,
the dome seismic station (DIO) recorded 750-1000
small magnitude earthquakes. Within 5 days, activity
0
500 was back down to a few earthquakes per day. Less in-
tense but otherwise similar sequences occurred on 22
< September (Julian day 265), 30 September (Julian day
273), and 5 October (Julian day 278). One unconfirmed
o
124 125 126 127 128 129 130 131 132 133 134 explosion event occurred on 9 September (Julian day
TIME - JULIAN DAY 1986 252). After 9 October (Julian day 282), earthquake
counts at station YEL remained above 10 per day and
Fig. 10. May 1986 plot of hourly average RSAM digital counts for increased to over 100 per day by 20 October (Julian day
station YEL and radial tilt for SAUN tiltmeter station on a com- 293), less than 2 days before extrusion began.
mon time scale. RSAM for station YEL peaked at 1900 UT 8 May
(Julian day 128), 1986. Tilt for SAUN tiltmeter station reversed RSAM plots (Fig. 11) of seismic activity in Septem-
direction, or south down several hours later. The extrusion of lava b e r - O c t o b e r 1986 were not as distinctive as those for
was assumed to have taken place at this time. The SAUN tiltmeter A p r i l - M a y 1986, due to less-than-ideal conditions
installation was destroyed during the peak of the avalanche within the crater: avalanches and rockfalls from the
phase crater walls resulted in a more complex seismic record.
However, the plots (Fig. 12) did show the characteristic
buildup in average amplitude during the last 2-3 days
dent exponential increases in data values (Fig. 10). prior to extrusion. The first peak in RSAM data was
There was rapid increase in outward tilt at tilt site followed by a second peak at 0600 UT. We can only
RUBY at the same time as tilt site SAUN, but R U B Y speculate that the saddle in the station G D N RSAM
went off scale (readings exceeded 10000 microradians) plot is related to the change in seismicity accompanying
at 1400 U T 8 May (Julian day 128). Photographic evi- the transition from Crater floor to dome deformation.
d e n c e during the October 1986 eruption indicates that We also speculate that the second increase in average
extrusion began 12 h after the peak in RSAM data, at amplitudes was a result of earthquakes associated with
the beginning of slow inward tilt at the SAUN tilt site. intrusion of magma into the dome and subsequent
The SAUN tilt site was destroyed at 0240 U T on 9 May thrust faulting beneath the west side of the lava dome.
(Julian day 129) by an avalanche from the new lobe. RSAM data for station G D N initially peaked shortly
Slow inward tilt during the second increase in RSAM after 0000 U T on 22 October (Julian day 295). Timed
data supported the conclusion that seismicity following
the initial peak was related to rockfalls and avalanches
associated with the breakup of the dome and extrusion 4000 ]- . . . . . . . . . . . . . . . l. . . . . . . . . . . . . .
of the new lobe. A net outward tilt of about 500 micro- 30001YEL HOUR AVERAGE RSAM DATA M
radians was recorded at the SAUN tilt site.
zooo~ A 1 . . . . . l'l

The O c t o b e r 1986 eruptive episode o ~ ~, : ~ , ,-~--~, , , : : .~,~,~, ,~,~.~,-~, ,, : ,~.~, ,-%~:

~3000I SHW HOUR AVERAGE RSAM DATA 1
N a r r a t i v e . The late October 1986 eruptive episode re- r.D 2000 ~- t~
sulted in a lava lobe slightly smaller in areal extent than
the May 1986 lobe. Preliminary volume estimates for
the October episode are about 9 • 10 6 m 3. That estimate
o,ooof o|:.,., : : : ,.,..,, ,.,,
J/
takes into account lateral displacement of parts of the
dome to accommodate new magma, particularly the
September 1984 lobe, which was intensely disrupted.
cA5
1000~.~ ~['I(-~ TELEMETRYNOISE A
The initial peak in seismicity occurred at 0000 U T on 22
October (Julian day 295), and by 1200-1300 UT, timed
photography of the dome showed that extrusion had 272 275 278 281 284 287 290 293 296 299 302
begun. There was no rockfall-avalanche phase like that TIME - J U L I A N DAY 1 9 8 6
during the May 1986 episode, probably because the ex- Fig. 11 A-C. RSAM plots of hourly average data for seismic sta-
trusion was inherently more stable during the October tions YEL, SHW, and PAK from 29 September (Julian day 272) to
episode. The development of tear and thrust faults on 29 October (Julian day 302), 1986. The magma migration phase
the crater floor and talus slopes on the west and south- for the October eruption is indicated by the interval M
542

5000
a0001GDN '10MIN'RSAMDATA A~w~ ' ' 1
Z
<
~ 4000
* YEL RSAM DATA
A RAZA TILT
t
0
EXTRUSION 25oo ~ RAZA TILT ~_ ~ " ~ " - - - ~
aooo START
0 ooo r / -
2000 500 -,.--." ..~ - ' ~

Z 0 I- _ ~ - - ' - - ' . '''~ "~ . . . . . -

[ . . . . EXTI~USION ' ' __.--:
O T,LT TART
lO00

0 O~ , ; ..... q.=--~"~ , , L
290 291 292 293 294 295 296 297 298 299 300 294 295 296
TIME - JULIAN DAY 1 9 8 6 TIME - JULIAN DAY 1986

Fig. 12. S e p t e m b e r - O c t o b e r 1986 plot o f R S A M digital c o u n t s for Fig. 13. A Plot o f h o u r l y a v e r a g e G N D R S A M d a t a for two d a y s
Y E L s e i s m i c station a n d radial tilt for R A Z A tiltmeter s t a t i o n o n b e g i n n i n g at 0000 U T 21 O c t o b e r (Julian d a y 294), 1986. B Plot o f
the s a m e t i m e axis tilt for s t a t i o n R A Z A t h e s a m e t i m e w i n d o w as Fig. l l a . C Plot o f
F A M S tilt for t h e s a m e t i m e w i n d o w as Fig. l l A a n d B

photographs showed that extrusion in the summit area trusion phase was similar to that of the May 1986 erup-
of the dome began between 1200 and 1300 UT. tive episode. The October 1986 eruptive episode was
not accompanied by an extended surface phase like
Tilt data. All four tilt stations (FAMS, OOPS, RAZA that of May 1986.
and SWLP, Fig. 2) operating on the dome during Octo-
ber 1986 showed rapid increases i n outward tilt that
corresponded to the rapid increase in RSAM data at all R S A M data for two explosion events at Mount St.
seismic stations. In addition, the October 1986 RSAM Helens
and tilt data provided evidence for different tilt re-
sponses, depending on the location of the tiltmeter site 31 December 1985
relative to the center of deformation and whether or not
the site was located on the dome or the crater floor. The first explosion (gas/ash emission) event recorded
Two tilt stations (RAZA and SWLP) showed significant on the RSAM system occurred on 31 December 1985,
outward tilt (inflation of about 200 microradians per at 0000 UT. About 100 high-amplitude, short-coda-
day) beginning 7-8 days prior to extrusion. More rapid length events (<35 s) were recorded by the seismo-
inflation (5000 to 10000 microradians per day) began graphs for seismic stations NSP and G D N between
48-72 h prior to extrusion. The inclinometer at the 0000 UT and 0145 UT. Of these, only the largest events
SWLP tilt site recorded more than 10 ~ of tilt in the last (shown in Fig. 14) were visible on the analog seismo-
24 h before it was destroyed. Rapid inflation starting gram for station SHW (4 km to the SW). The largest
18-20 h later was recorded at tilt stations OOPS and event, 30 mm peak-to-peak on the CVO helicorder re-
FAMS. Tilt data from the RAZA tilt site have been cord, was assigned a preliminary coda magnitude of 1.5
plotted with RSAM data from seismic station YEL by the University of Washington. The earthquakes were
(Fig. 12) to show that tilt was coseismic, as in pre- accompanied by a 2-3 times increase in the background
viously described eruptive episodes. RSAM data noise level (Fig. 15) of the dome seismic station NSP
peaked at 0600 UT 22 October (Julian day 295), 1986. from 0041 to 0200 UT. The high-frequency (peak un-
Timed photography confirmed extrusion between 1200 corrected spectra near 12-13 Hz) tremor-like signal be-
and 1300 UT. came strong enough to be recorded at the seismic sta-
The first peak in RSAM data and a corresponding tion G D N at 0108 UT. The high-frequency tremor-like
decrease in outward tilt on the crater floor represent the signal was assumed to be fumarolic noise generated by
breakthrough of magma into the dome and the relaxa- degassing through a vent located a few hundred meters
tion of stress between the magma and conduit walls south of station NSP. Dominant frequencies occurred
near the crater floor-dome base interface (Fig. 13). The above 10 Hz in contrast to typical harmonic tremor
second phase of dome tilt observed at three tilt stations, which is commonly dominated by frequencies below
and the corresponding drop in seismicity, were very 4 Hz. The high-frequency noise is very similar to geyser
likely associated with the intrusion of magma into the noise studied by Kieffer (1984).
dome and thrust faulting beneath the west side of the The three largest earthquakes were recorded by
dome. The 2-day duration of the magma migration-ex- RSAM as spikes well above the background noise level
 543

2 MINUTES
I

-~ z g--
s

~ . ~_ ~ _

Fig. 14. A 6.5-min wide section of the

analog seismic record for GDN on 30-
r
31 December 1985. The largest ampli-
tude event (0108 UT on 31 December)
had an estimated coda magnitude of
1.8. Translation rate for the seismic re-
<- c- cord is 15 min per line

2 MINUTES 500

4O0 GDNRSAM DATA. . . . . |

300 I<--EARTHQUAKE ]
-! ~--~ 200
A
q f NIOISENa
i ~ 1 ~
9
i.
n
~,..~! I ,
~ .........
,
~'' 100
e o
.~ 40 DOME AIR H2
~ ao
B
~ zo
to
0 -~ - ~ J"'~-'~'~
364 365

n m : ~ .m n] m ~ m l TIME - JULIAN DAY 1985

n 0 " " r
Fig. 16. A RSAM plot of the highest 1-min value in a 10-min in-
[
I
terval for seismic station GDN from 30 to 31 December (Julian
~lt
day 364 and 365), 1985. Spike to the left of the earthquake identif-
ied as noise is a result of connection of GDN seismic signal to the
input of RSAM system. Two spikes to the right identified as noise
were a result of wiring modification to RSAM system. B Plot of
output from hydrogen sensor operating on the dome at the time of
the explosion event of 31 December (Julian day 365), 1985. The
Fig. 15. A 5-min wide section of the NSP (first telemetry station hydrogen event occurred shortly after the magnitude 1.8 earth-
installed on the dome) analog seismic record. Increase in high- quake at 0108 UT
frequency background noise level is indicated by the arrow. The
helicorder (or heat sensitive) record did not clearly reproduce
earthquake signatures
April 1986

The next recorded sequence of explosion events oc-

(Fig. 16a). R S A M data from station N S P was lost in curred during the m o n t h preceding the eruptive episode
D e c e m b e r due to a p o o r electronic connection in the of 9 M a y 1987. About 50 possible explosion events
system. The smaller increase in b a c k g r o u n d noise at were identified from the RSAM data. C o n f i r m a t i o n o f
station G D N can be identified by a small change in the each was not possible due to failure of telemetry f r o m
base level of the station's RSAM plot. The sequence of the gas sensor, but field observations o f ash over fresh
earthquakes and a c c o m p a n y i n g high-frequency tremor- snow confirmed the larger events. Ideal crater condi-
like signals at stations N S P and G D N suggested a small tions, low snow fall, freezing temperatures, and low
explosion ( g a s / a s h emission as a result of a phreatic or wind noise resulted in relatively low b a c k g r o u n d noise
p h r e a t o m a g m a t i c explosion) similar to m a n y others level at most seismic stations. As in late D e c e m b e r
which had been observed earlier. Telemetered data 1985, individual earthquakes a c c o m p a n y i n g explosion
from a volcanic gas sensor on the d o m e confirmed the activity showed as spikes on R S A M plots (Fig. 17a and
increased emission of gas. The gas data (Fig. 16b) b), and lower level high frequency noise showed as
peaked about 1 h after the onset of earthquake activi- slightly elevated base levels. At 0115 U T on 17 April
ty. (Julian day 107), N S P seismic station was disabled by
ballistic debris (Fig. 17a) from a relatively large explo-
sion event. There were no significant changes in tilt as-
sociated with any of the explosion events in 1985 and
1986. It was not possible to clearly identify the explo-
544
i -1- T r n- 7 T r | from the crater wall. Telemetry failures also introduced
3000 NSP I-MIN R S A M DATA
problems in interpretation. Ambiguity in interpretation
~3 ~/NSP DISABLED can be eliminated by comparison of R S A M data with
other real-time data, such as tilt, gas, displacement, and
Z 1ooo various meteorological information. Visual examina-
o o tion of seismographs usually eliminates possible misin-
~r')
200030GDNB
0011MIN~RSAMDATA terpretation.
R S A M is an easy-to-implement and convenient-to-
use m e t h o d of tracking the intensity of seismic activity
associated with volcanic eruptions in near real time.
The value of R S A M data is best demonstrated by a re-
106 107 108
TIME - JULIAN DAY 1986 cent application of an R S A M system in the monitoring
of R e d o u b t volcano (Alaska Volcano Observatory Staff
Fig. 17. A RSAM plot of seismic station NSP 1-min data from 16 1990). In addition to using R S A M data for predicting a
April (Julian day 106) to 18 April (Julian day 108), 1986. Explo- n u m b e r of m a j o r eruptions, RSAM data played a cru-
sion debris associated with the event (at 0115-0120 UT on 17
cial role in the decision to evacuate the Drift River Oil
April (Julian day 107) disabled the NSP seismic station. B RSAM
plot of seismic station GDN l-rain data from 16 April to 18 April Terminal hours before a major explosive eruption on 2
1987. The largest events assumed to be associated with explosion J a n u a r y 1990.
events are indicated by arrows
Acknowledgements. The authors credit Dr. R. Schick for the origin
of the idea of looking at the amplitude envelope of a seismic sig-
nal to document the time history of seismicity associated with vol-
sign events in D e c e m b e r 1985 and spring of 1986 as canic eruptions. We thank L. Topinka for her effort in obtaining
phreatic or p h r e a t o m a g m a t i c explosions. photographic evidence for the time of extrusion for the October
1986 dome-building episode. We also gratefully acknowledge the
early technical reviews by D. W. Swanson, P. Okubo, D. Dzurisin.
Conclusions Journal reviews by P. W. Lipman, S. D. Malone, and S. R.
McNutt significantly improved the paper. G. Varanas and P.
Johnson provided additional helpful comments.
The CVO R S A M system has been in operation since
the spring o f 1985. F r o m 1985 to 1986 there were three
dome-building eruptive episodes at M o u n t St. Helens;
from each episode we have learned how to better inter- References
p r e t R S A M plots together with seismic and other tilt
data. During all three episodes, similar rapid increases Aki K, Richards PG (1980) Quantitative seismology, theory and
methods. WH Freeman, San Francisco, Calif, 932 p
in seismicity began less than 48 h prior to the p e a k in
Alaska Volcano Observatory Staff (1990) The 1989-1990 Eruption
RSAM data. Large changes in tilt, indicating inflation of Redoubt volcano. EOS: 71 : 265 272-273 275
and the u p w a r d m o v e m e n t of magma, a c c o m p a n i e d Decker RW (1973) State-of-the-art in volcano forecasting. Bull
rapid increases in RSAM data. Clearly, R S A M data can Volcanol 37: 372-393
be used for the prediction of dome-building eruptions Decker RW (1986) Forecasting volcanic eruptions. Ann Rev Earth
at M o u n t St. Helens. The shape of RSAM plots follow- Planet Sci 14:267-291
Endo ET, Dzurisin D, Murray T, Syverson K (1987) The rate of
ing the initial p e a k in seismicity appears to depend on
magma ascent during dome building at Mount St Helens (ab-
the volume of m a g m a involved and events related to stract). Abstract volume Hawaii Symposium on How Volca-
the intrusi.on of m a g m a into the d o m e aiad its subse- noes Work, Hilo, Hawaii, p 64
quent extrusion. Thrust faulting on the crater floor as- Gutenberg B, Richter CF (1956) Magnitude and energy of earth-
sociated with" lateral displacement of the crater floor quakes. Estratto da Annali di Geofisica 9:1-15
and d o m e resulted in extended periods of seismicity Hurst AW (1985) A volcanic tremor monitoring system. J Volca-
during the M a y 1985 and October 1986 episodes. Insta- nol Geotherm ReD 26:181-187
Kieffer SW (1984) Seismicity at Old Faithful Geyser: an isolated
bilities a c c o m p a n y i n g the disruption of the d o m e by in- source of geothermal noise and possible analogue of volcanic
truded m a g m a - g e n e r a t e d rockfalls and a v a l a n c h e s ' p r o - seismieity. J Volcanol Geotherm ReD 22:59-95
duced large-amplitude signals on seismic stations Korneff T (1966) Introduction to electronics. Academic Press,
within the crater. Similarly, mass m o v e m e n t o f newly NY, 545 p
extruded m a g m a o v e r the sides of the d o m e also gener- Koyanagi RY, Meagher K, Klein FW, Okamura AT (1975) Ha-
ated r o c k f a l l / a v a l a n c h e signals on crater seismomet- waiian Volcano Observatory Summary 75, 117 p
Lee WHK, Stewart SW (1981) Principles and applications of mi-
ers. croearthquake networks. In: Advances in Geophysics, Supple-
Our experience a t M o u n t St. Helens has shown that ment 2. Academic Press, NY, 293 p
RSAM data cannot be used to predict phreatic or Malone SD, Boyko C, Weaver CS (1983) Seismic precursors to the
p h r e a t o m a g m a t i c explosions. We also recognize the Mount St Helens eruptions in 1981 and 1982. Science
limitations of RSAM data during late-spring and sum- 221 : 1376-1378
mer crater conditions, when rockfall and avalanche Minakami T (1960) Fundamental research for predicting volcanic
eruptions, pt. 1. Bull Earthquake ReD Inst 38:497-544
conditions are o p t i m u m within the crater at M o u n t St. Minakami T (1974) Prediction of volcanic eruptions. In: Civetta
Helens. Melting snow in late spring and dry conditions L, Gasparini P, Luongo G, Rapolla A (eds) Physical volcano-
during the s u m m e r p r o m o t e rockfalls and avalanches logy. Amsterdam, Elsevier, pp 313-333
 545

Minakami T, Hiraga S, Miyazaki T, Uchibori S (1969) Fundamen- Aso). Memoirs of the College of Science, Kyoto Imperial Uni-
tal research for predicting volcanic eruptions, pt. 2. Bull EaCh- versity, vol XIX, 1: 11-56
quake Res Inst 47 : 893-949 Scandone R, Malone SD (1985) Magma supply, magma discharge
Murray TL, Endo ET (1986) BOB, a computer graphics tool for and readjustment of the feeding system of Mount St Helens
real-time integrated volcano monitoring. EOS 67:397 during 1980. J Volcanol Geotherm Res 23:239-262
Nakamura Y (1977) Detection and analysis of acoustic emission Schick R (1988) Volcanic tremor-source mechanisms and correla-
signal. In: HR Hardy Jr, FW Leighton (eds) Proceedings First tion with eruptive activity. Natural Hazards 1:125-144
Conference on Acoustic Emission/Microseismic Activity in Schick R, Lombardo G, Patane G (1982) Volcanic tremors and
Geologic Structures and Materials 2:445-457 shocks associated with eruptions at Etna (Sicily), September
1980. J Volcanoi Geotherm Res 14:261-79
Nishi K (1987) Automatic data processing system for volcanic
earthquakes and tremors, Annuals Disaster Prevention Res. Swanson DA, Holcomb RT (1989) Regularities in growth of the
lnst Kyoto Univ 30B-1 : 1-18 Mount St Helens dacite dome, 1980-1986. In; Fink J (ed) Lava
flows and domes. Springer, Berlin Heidelberg New York
Norwack R, Aki K (1984) The two-dimensional gaussian beam 24 p
synthetic method:testing and application. J Geophysical Res Swanson DA, Casadevall TJ, Dzurisin D, Holcomb RT, Newhall
89:7797-7819 CG, Malone SD, Weaver CS (1985) Forecasts and predictions
Richter CF (1958) Elementary Seismology. WH Freeman and of eruptive activity at Mount St Helens, USA: 1975-1984. J
Company, San Francisco, 768 p Geodynamics 3:397-423
Sassa K (1936) Micro-Seismometric study on eruptions of the Vol-
cano Aso (Part II of the Geophysical studies on the Volcano Editorial responsibility: P. W. Lipman