Abstract
Personnel of the U.S. Geological Survey's Cascades Volcano Observatory
established trilateration networks at Mount Baker, Mount Rainier, Mount Hood,
Crater Lake, Mount Shasta, and Lassen Peak in 1980-1984. These networks are
capable of detecting changes in slope distance of several centimeters or more.
The networks were established to provide baseline information on potentially
active volcanoes and were designed along guidelines found useful at Mount St.
Helens. Periodic reoccupation of the networks is planned as part of the
overall monitoring program of Cascades volcanoes. Methodology, slope distance
and vertical angle data, maps of the networks, and benchmark descriptions are
presented in this report. Written benchmark descriptions are augmented by
photographs, which we have found by experience to very useful in relocating
the marks. All repeat measurements at the six volcanoes are probably within
measurement error.
Introduction
In response to renewed
eruptive activity at Mount St. Helens in 1980, the
Cascades Volcano Observatory (CVO) was established in Vancouver, Washington,
with the support of the U.S. Geological Survey's
Volcano Hazards Program.
In addition to continued monitoring at Mount St. Helens,
CVO scientists have
initiated geodetic and geochemical monitoring at other potentially active
volcanoes in the Cascade Range (fig. 1).
Field work in 1980-84 included the
acquisition of baseline geochemical and geodetic information at six Cascade
volcanoes other than Mount St. Helens. Geochemical studies include temperature
measurements and gas sampling of fumaroles at
Mount Baker,
Mount Hood,
Mount Shasta, and
Lassen Peak (T.J. Casadevall, oral commun., 1983).
Geodetic studies
consist of: (1) slope distance and vertical angle measurements at
Mount Baker,
Mount Rainier,
Mount Hood,
Crater Lake,
Mount Shasta, and
Lassen Peak;
(2) tilt surveys at
Mount Baker (Frank and others, 1975; Daniel Dzurisin, unpublished data),
Mount Rainier (Dzurisin and others, 1983),
Mount Shasta and Lassen Peak (Dzurisin and others, 1982), and
Mount Hood (Daniel Dzurisin, unpublished data);
and
(3) precision gravity surveys at
Mount Shasta and Lassen Peak (Jachens and others, 1983).
Further work is planned at these and other Cascade volcanoes as
funds permit. The geodetic studies supplement a program of continuous seismic
monitoring at Cascade volcanoes funded by the U.S.Geological Survey in
cooperation with the University of Washington. These investigations provide
useful baseline data for evaluating future deformation related to subvolcanic
activity.
Geodetic Monitoring of Volcanoes
Magma rising beneath a volcano forcefully displaces the surrounding rock, and
the resulting deformation can be measured at the ground surface. Since about
1910, attempts have been made to measure this deformation in order to help
understand magmatic processes and predict eruptions at active volcanoes (F.
Omori, in K.Mogi, 1958). The vertical component of deformation has generally
been determined by leveling techniques and by the less precise measurement of
vertical angles. Early attempts to measure the horizontal component of
deformation utilized relatively imprecise triangulation methods. In the
mid-1960's, the development of electronic distance meters (EDM) spurred
personnel of the U.S. Geological Survey's Hawaiian Volcano Observatory to
initiate studies of horizontal strain at Kilauea Volcano. These measurements
have helped document repeated episodes of inflation and deflation of the shield
volcano related to intrusions and eruptions (Kinoshita and others, 1974).
Few such detailed and precise monitoring efforts had been attempted on active
stratovolcanoes or composite volcanoes before 1980 (a notable exception is the
study of Usu volcano, Japan [Yokoyama and others, 1981]), when the reawakening
of Mount St. Helens provided an ideal opportunity to test the utility of such
measurements. In late April 1980, distance and angle measurements were
initiated from instrument sites on the apron of the volcano to targets on its
flanks. This monitoring documented a remarkable bulging of the north flank as
magma intruded the volcano before the large landslide and explosion of May 18,
1980 (Lipman and others, 1981). The need to monitor all sides of a volcano was
indicated by the localized development of the bulge, which covered 3-4 square
kilometers, extended nearly 2 kilometers downslope from the summit area, and was
mostly confined to a 60 degree radial sector of the cone. Points on the bulge
moved tens of meters northward, whereas points just off the bulge and elsewhere
on the one were nearly immobile. After the May 18 events, geodetic monitoring
of the volcano's flanks suggested slight horizontal expansion before other
explosions in 1980 and slight contraction afterward (Swanson and others, 1981).
Since 1980, distance measurements inside the crater of Mount St. Helens have
been used to predict dome-building extrusions of dacite a few days to 3 weeks in
advance (Swanson and others, 1983; Chadwick and others, 1983). The
unprecedented success of horizontal strain monitoring at Mount St. Helens
suggests that this technique can be used for surveillance of other composite
volcanoes, such as those of the Cascade Range.
Trilateration Networks at Cascade Volcanoes
We describe in this report the trilateration networks installed and occupied in
1980 to 1984 at Mount Baker, Mount Rainier, Mount Hood, Crater Lake, Mount
Shasta, and Lassen Peak (table 1), including installation procedures,
benchmark locations, and baseline measurements. We intend the report to be a
reference to facilitate reoccupation of the networks and analysis of the data
obtained, as well as a guide to those interested in making similar measurements
at other volcanoes.
Equipment and procedures
-
All benchmarks are I0 cm diameter die-cast brass monuments with 7 cm
stems cemented into holes bored into bedrock or large boulders using a
masonry (star) drill. In general, we tried to place benchmarks within
each 60 degree sector of the volcano and at two different elevations within each sector. These goals were not often realized. Site selection was
governed largely by the distribution of bedrock outcrops accessible by
helicopter. Ease of ground access was not a strong factor in site selection, because helicopter support was considered logistically essential for
adequate monitoring. A resurvey 1-2 yrs after the first survey was planned for each volcano to check the initial data and provide an estimate of
expected errors. Subsequent reoccupations are scheduled every 3-5 years,
unless increased seismicity, phreatic activity, or reports of increased
snow melt or other visible changes are reported.
Slope distances were measured with an EDM (Hewlett-Packard 3808A) and
vertical angles with a theodolite (Wild T-2, both old and new styles) from
benchmarks generally at low elevations around each volcano to reflectors
at benchmarks high on the cones. Vertical angles were measured primarily
to establish station elevations and to make mark-to-mark distance reductions; they were not measured reciprocally (table 2). Temperature, pressure, and humidity corrections were applied to EDM data. Temperature was
measured with a thermistor or thermometer generally about 2.5 m above
ground at each benchmark. Pressure was measured with a pressure transducer at each reflector site and a transducer backed up by a precision aneroid barometer at the instrument site. In 1983 and 1984, humidity was read
at the instrument site with a sling psychrometer; no end-point humidity
measurements were made in 1981 and 1982, when a nominal correction of
-0.5 ppm was assumed.
Measured distances several kilometers long commonly cross 1 km or more
of elevation on the steep volcanoes; consequently atmospheric properties
change significantly along the line path. Determining the atmospheric
refractivity using end-point measurements of temperature, pressure, and
humidity provides some correction for the EDM measurements, but temperature and humidity do not necessarily change linearly along such steep
lines with large elevation differences.
We attempted to improve the precision of the slope-distance measurements by using a helicopter to take semi-continuous temperature and.humidity readings along most line paths while distance measurements were being
made. (This procedure was not followed at Crater Lake, where the lines
are nearly horizontal and the noise of a helicopter would be particularly
disturbing to tourists.) Temperatures and humidities were determined
with a thermistor and hygrister mounted on the front of the left helicopter skid and recorded on a data logger inside the aircraft. The helicopter generally flew downslope at a nearly constant air speed and was guided
by radio from the reflector end in order to stay within a few meters of
the line path. Most lines were flown at an air speed of about 45 mph, but
some were flown as slowly as 25 mph and others as rapidlv as 60 mph.
Programs for CVO's VAX 11/750 computer (Endo and others, 1985), adapted from similar programs used in the U. S. Geological Survey's Tectonophysics Branch in Menlo Park, California, use the flight-line data to calculate an average refractive index for the entire line and correct the
measured slope distances accordingly. Savage and Prescott (1973) and Bornford (1980) describe in detail the measurement techniques and related procedures. Slope distances for 1982-1984 were calculated using flightline
atmospheric data (table 4). Most distance measurements in 1981 were accompanied by flightline readings, but equipment malfunctions led to so
many obviously incorrect temperature measurements that we decided to discard all of the 1981 flightline data.
The manufacturer's stated precision for the EDM is +(5 mm+l ppm) in
the temperature range of interest. If two measurements of the same line
differ bv twice this value or less, the difference cannot be considered
significant. This is an instrumental precision only, however, and does
not take into account inaccuracies in measuring the atmospheric index of
refraction. An overall precision of +3 ppm, including errors resulting
from end-point temperature, pressure, and humidity measurements, was found
for relatively flat lines measured with the same instrument model at Long
Valley, California (R. P. Denlinger, personal commun., 1984). By assuming
no strain in our networks, we calculate an overall precision of +3.9 ppm
from the differences in 142 end-point line lengths given in tabl" 3 (exclusive of the long lines at Crater Lake) (fig. 2). This value may be
larger than that at Long Valley owing to the differences in terrain, to
some strain or benchmark instability in our networks, or to problems in
our method of temperature measurement. In addition, the figure was calculated using some data from Mount Shasta and Lassen Peak in 1984 that we
believe are of poorer quality than normal owing to windy conditions.
Lacking objective evidence of this, however, we use the figure of +3.9
ppm as a guide for evaluating apparent changes.
A significant improvement in precision was apparently not obtained by
flying the lines. We had expected a precision of perhaps about +2 ppm
(a't Long Valley it is +1.5 ppm), but instead we find a precision of about
+ 3.4 ppm for the 72 measurements in table 4 under the assumption of no
train (fig. 3). This result is addressed in the discussion section.
Results
-
Table 3 gives slope distances computed on the basis of end-point temperature and pressure measurements and an assumed humidity correction of
-0.5 ppm. Table 4 lists the slope distances calculated using flightline
data. All distances listed are mark-to-mark, corrected for instrument
heights, and not reduced to sea level. Maps of the survey networks and
descriptions and photographs of benchmark sites are in Appendices A-F.
Surveys at Mount Baker (table 3A) show no evidence of deformation between 1981 and 1983 (fig. 4). All repeat measurements agree within twice
the assumed error of one end-point measurement (3.9 ppm), consistent with
the lack of seismicity at the volcano during the same time. Measurement
conditions were excellent during both surveys: light winds, clear air, and
moderate day and night temperatures. Such conditions probably contribute
much toward the relatively high quality of the surveys. In 1983, distances calculated using flightline data are longer (mean=+3.1 ppm, s.d.=l.l)
than those using end-point data (tables 3A, 4A, and 5A). This difference
could be explained by an average end-point temperature about 3oc lower
than that obtained by the aircraft. Pressure and humidity have relatively
little effect on the calculations, and all other variables--instrument
height, uncorrected slope distance readings, station elevations, etc.--are
the same for both sets of calculations.
At Mount Rainier, most repeat measurements agree within twice the
expected error of a single measurement between 1982 and 1983 (tables 3B
and 4B). Some lines could not be measured in 1983 owing to poor weather.
As at Mount Baker, distances calculated from flightline data are generally
longer than those calculated from end-point data, in 1982 by a mean of
2.3 ppm (s.d.=l.5) and in 1983 by a mean of 2.0 ppm (s.d.--l.8) (table 5B).
These comparisons suggest that the average end-point temperatures were
about 2oC lower than the flightline temperatures. Conditions were poor
during measurements of several lines. Particularly strong, gusty winds
30 knots) badly vibrated both the EDM and reflector during measurement
of line 2 in both years. Past experience at Mount St. Helens with a
Rangemaster 3 has shown that such windy and gusty conditions shake the
instrument and reflector out of plumb an often accompany air instability,
both factors adversely affecting measurements. The windy conditions may
account for the large apparent change in length of line 2, which is much
above the expected error and was excluded from the precision analysis for
that reason (this is the only measurement excluded from any statistical
treatment in this report). Other shots involving the end points (McClure
Tilt and Camp Hazard) of line 2 are within expected error, so that both
benchmarks are probably stable. In addition, the relative elevation of
McClure Tilt mark was determined in both years by precise levelling and
showed no undue change (Daniel Dzurisin, unpublished data). The apparent
length change on line 29 (Iron Mountain to St. Andrews Rock) is beyond
that of expected error for the end-point calculation and barely within" expected error for the flightline calculation. Local site stability of St.
Andrews Rock cannot be checked because it is not sighted from another
station. The apparent change in flightline distance for line 8 is above
expected error, but the end-point calculation is acceptable; the reason
for this discrepancy is unknown. A longer history of measurements will
be necessary before such changes can be evaluated adequately. The inconsistent changes on adjacent lines (fig. 5) argues against but cannot exclude the possibility of deformation of the cone.
At Mount Hood, several apparent changes between 1980 and 1983 are relatively large (table 3C; fig. 6), but little stock can be placed in them
owing to the lack of adequate temperature and pressure equipment and. the
use of a different EDM (Rangemaster 3) in 1980, when a limited number of
PK masonry, nails were installed (the nails were left in place when benchmarks were emplaced in 1983). The apparent changes are small by comparison with those expected should the volcano begin to swell in rsponse to
magma intrusion at depth.
Apparent changes in line length at Mount Hood from 1983 to 1984 are
within expected error for flightline calculations and, except for line
20, also for end-point calculations (tables 3D and 4C; fig. 7). Flightline data yield longer distances than end-point data; in 1983, the mean
difference is 3.6 ppm (s.d.=l.5), and in 1984, 2.8 ppm (s.d.=0.9) (table
5C). This is consistent with end-point temperatures being about 3.5oc
and 3o C lower respectively than flightline temperatures. Three of the
four distances measured from Cathedral in 1984 (lines 17, 18, and 20) are
longer by 4.5, 6.6, and 10.7 ppm than in 1983, but the fourth (line 19) is
nearly the same length. Flightline data for these lines are not available
owing to equipment malfunction. Perhaps the Cathedral benchmark is unstable or the setup was not properly centered, although neither of these possibilities by itself can account for the small change on line 19. Random
error is possibly the best explanation for the apparent changes.
At Crater Lake, repeat measurements are well within expected precision limits (table 3E). The presence of the lake beneath virtually the
entire length of each line may help stabilize air density. Moreover, the
lines are nearly flat (table 2D) and are high enough above lake surface
to be unaffected by scintillation and enhanced humidity. End-point
measurements should suffice across the caldera under most conditions.
Most line lengths at Mount Shasta were similar in 1981 and 1982
(table 3F; fig. 8)). Most of the apparent changes were small extensions,
generally within expected error. However, lines 12 and 13 show apparent
changes greater than expected; no reason is evident. The apparent 1981-1982 extensions were cancelled or changed to contractions by measurements
in 1984 (fig. 9), and the net apparent change on most lines between 1981
and 1984 is contraction (tables 3F and 4D). The small 1981-1982 extensions and somewhat larger 1982-1984 contractions probably reflect slightly
different atmospheric conditions. In 1982, the comparison between calculated line lengths using end-point and flightline data (table 5D) is
closer flightline data slightly longer, mean=0.9 ppm, s.d.=l.5) than in
1984 at Mount Shasta mean=3.1 ppm, s.d.=l.3) and at other volcanoes in
1982-1984 (Mount Baker, Mount Rainier, Mount Hood, and Lassen Peak). This
suggests that end-point temperatures were closer to average air temperatures in 1982 than normal. In 1984, conditions at Mount Shasta were very
windy, and the overall quality of the survey was probably less than in
other years. This may account for the large apparent changes on lines 9,
12, and 20 end-point calculation only), although downslope movement of
Shastina (line 9) and Wishbone (line 12) is also possible. On balance,
the data probably should be considered as reflecting measurement errors
rather than real strains, although we cannot eliminate the possibility of
slight areal contraction around the mountain between 1981 and 1984.
The pattern of apparent changes at Lassen Peak is very similar to that
at Mount Shasta. Generally small extensions within expected error were
recorded between 1981 and 1982 (fig. 10) and larger contractions, shown
both by end-point and flightline data, accrued between 1982 and 1984
(fig. II), with a net overall contraction between 1981 and 1984 (tables
3G and 4E). Calculated distances based on flightline data (table 4E) are
slightly longer than those based on end-point data in 1982 (mean=l.3 ppm,
s.d.=0.9) and significantly longer in 1984 (mean=3.2, s.d.=2.1) (table
5E), a pattern similar to that at Mount Shasta. The 1984 survey was made
during strong winds, just as at Mount Shasta, and the quality of the data
is probably less than in previous years and likely accounts for large ap-
parent changes on many of the lines beyond those expected from our estimates of overall precision. The 1982 weather at Lassen was favorable except for one day of strong winds ending with a thunderstorm. Mornings
were cooler than during other surveys at Lassen, however, and this apparently caused the great disagreement 7 ppm) between flightline and end-point calculations for line I0, which was measured early in the morning
when the ground was cold and the average end-point temperature was 9.4oC
cooler than in 1982. We conclude that there is no strong evidence that
deformation is occurring at Lassen Peak and that the spread in the line
lengths probably reflects measurement error. We cannot, however, rule
out the possibility of slight areal contraction between 1982 and 1984.
Discussion and Summary
The distances calculated on the basis of flightline temperature and humidity
are of unexpectedly poor quality compared with what we expected; those
calculated from end-point measurements along ear nearly equivalent. Why?
We have no good reason but can suggest several possibilities. Our
temperature-measuring setup on the slid of the helicopter may need improvement,
and we are investigating this now. To date, however, we have found nothing
that leads us to believe that the thermistor is inferior. Perhaps the air speed
of the helicopter is too slow to enable proper corrections for frictional
effects of airspeed to be made to the raw thermister readings. Perhaps rotor
wash affects the readings in a way unaccounted for. Possibly inherent but
unrecognized errors exist when using a helicopter on short, generally steep
lines that are absent when using an airplane on long, relatively flat lines. We
will investigate and test means to improve the quality of the flightline data,
but if they cannot be improved, we will probably revert to making only end-point
measurements of temperature and humidity.
The end-point results differ from flightline results in a predictable manner, as
described above for each volcano. Combining all 1980 differences between
flightline and end-point results yields a mean of 2.4 ppm (s.d=1.7), with
flightline calculations consistently longer than end-point calculations
(fig. 12). This suggests that end-point temperatures are
systematically about 2.4 degrees C +/- 1.7 cooler than the effective flightline
temperatures, regardless of time of day. This was surprising to us, for we had
thought that the end-point temperatures would generally be warmer than the air
temperatures or at least cooler in only early morning hours. If the flightline
temperatures were incorrect, this conclusion is of course invalid.
The distance measurements, although of less than desired quality, indicate no
significant deformation of any of the monitored volcanoes. We cannot rule out
the possibility of slight areal contraction at Mount Shasta and Lassen Peak
between 1982 and 1984 but prefer an alternative interpretation that the data
from 1984 are adversely affected by the windy conditions during the survey.
The data presented in this report should define adequate baselines for detecting
changes of a few centimeters but no less. We will attempt to improve the
quality of these baselines, but already they are far superior to those existing
at Mount St. Helens before 1980. We now have a way to recognize and interpret
the early stage of deformation at the monitored volcanoes that may be precursory
to future eruptions, although we must realize that during a typical winter and
spring snow cover will make reoccupation of most of each network impractical.
Acknowledgments
We thank Steve Brantley, Ed Brown, Tom Casadevall, Dan Dzurisin,
Christina Heliker, Dan Johnson, Bobble Myers, John Power, Ben Talai
(Rabaul Volcanological Observatory, Papua New Guinea), Lyn Topinka, and
Richard Waitt for assistance in the field at one or more volcanoes. We
also thank Terry Leighley and Lyn Topinka for many hours of darkroom work
and Elliot Endo for warping his busy schedule to provide yet another service to CVO, writing the requisite computer programs for reducing flightline data. Roger Denlinger helped with the statistics. Special thanks
go to our helicopter pilots, Bob Edwards, Anthony Reece, Gary Trailor,
and Steve Wistrand, who not only adeptly got us in and out of one tight
spot after another but also operated the recording equipment as they flew
along the line of sight. We are grateful for the assistance and cooperation of personnel of the U.S. Forest Service at Mount Baker, Mount Hood,
and Shasta-Trinity National Forests, and of the National Park Service at
Mount Rainier, Crater Lake, and Lassen Volcanic National Parks.
References
(not online)
|