Appendix D. Monitoring Networks and Instrumentation

This appendix summarizes the instrumentation currently operating in Long Valley Caldera that fulfills the monitoring requirement for prompt identification of changes in activity. The USGS continues to upgrade the monitoring networks as funding permits, and this summary applies to the instrumentation configurations as of early 2002. The telemetered networks provide continuous data on seismicity, deformation, water-well levels, and CO₂-gas concentrations and flux rates for realtime review and analysis on computer systems in the LVO Field Center in Mammoth Lakes, the USGS in Menlo Park, California, and the Cascades Volcano Observatory in Vancouver, Washington. Routine measurements of the twocolor electronic distance meter (EDM) network provide data on deformation changes on time scales of a week or more. The regional, survey-mode geodetic networks (GPS and leveling) are normally measured just once a year, and they will generally not contribute to rapid evaluation of activity levels. In the case of condition YELLOW or above, however, field crews may be mobilized to measure critical parts of these networks on a frequent basis to help better define the areal extent, magnitude, and rate of the associated ground deformation. Each of these monitoring networks is summarized below. For descriptions of some of the more experimental networks for tracking variations in gravity, hydrological, electromagnetic, and geochemical parameters measured at infrequent or irregular intervals, see (Battaglia and others, 1999; Gerlach and others, 1999; Hill, 1984; Sorey and others, 1998). Near-real-time summaries of most of these monitoring data are available through the Long Valley Observatory web page at http://lvo.wr.usgs.gov.

Seismic Network

The seismic network monitoring Long Valley Caldera and vicinity (fig. D1) was being upgraded as of early 2002. The network in its current (early 2002) form was established in the summer of 1982. From the beginning, the seismic network was operated as part of the Northern California Seismic Network (NCSN of the USGS) in cooperation with the University of Nevada Reno (UNR). In 1995, the NCSN became a cooperative effort between the USGS and U.C. Berkeley. Data from the Long Valley seismic network are analyzed, catalogued, and archived as part of routine NCSN processing. The archived data reside on a mass-storage device maintained jointly by U.C. Berkeley and the USGS. The archived data are available through the Internet at http://quake.geo.berkeley.edu/ncedc/catalog-search.html.

Figure D1. Map showing seismic stations operating in Long Valley Caldera and vicinity. Stars are stations of the Northern California Seismic Network (NCSN). Triangles are six-component borehole packages colocated with borehole dilatometers. The circle is the deep borehole three-component package installed at a depth of 3 km (1.8 miles) in the Long Valley exploratory well (LVEW). The square is station MLAC in the Terrascope network operated by the California Institute of Technology. (Click for large version of Figure D1.)

The existing network also includes three, six-component (three velocity and three acceleration components) digital stations colocated with the borehole dilatometers POPA, BSP, and MCX (fig. D2). The signals from these stations, digitized at 200 sps, are also transmitted to Menlo Park by the 128 Kbps frame relay. The 3-km-deep (1.8 mile deep) borhole in the center of the resurgent dome (see LVEW in fig. D4) is currently being developed as a borehole observatory that will include a 6-component seismic package installed at a depth of between 2 and 3 km (1.2 and 1.8 miles), as well as a borehole dilatometer and tiltmeter plus a variety of hydrological monitoring instruments.

Figure D2.A, Map of borehole dilatometer (triangles) and tiltmeter (circles) locations. The cluster “LB” represents the long-base Michelson tiltmeter installed and maintained by Roger Bilham of the University of Colorado (Behr, and others, 1992). B, Map of differential magnetometer locations in Long Valley Caldera. The reference magnetometer station, MG, is located 20 km (12 miles) southeast of the caldera on the Sherwin grade near Highway 395.

The station MLAC co-located with the long-base tiltmeter (figs. D1 and D2A) is a broadband, wide-dynamic range seismometer that is operated as part of the California Institute of Technology TERRAscope network (see http://www.gps.caltech.edu/seismo/earthquake/ for more information on the Caltech TERRAscope stations). The station OMM is a broadband seismometer (CMG-3) installed by the University of Nevada, Reno, in the summer of 2000 in a mine adit located 4 km (2.5 miles) southeast of Mammoth Mountain.
The existing analog network routinely detects and locates earthquakes with magnitudes as small as M=0.3 (the detection threshold), and it systematically detects and locates 90 percent or more of all earthquakes with magnitudes M=1.2 and greater (the completeness threshold) occurring within the network. These thresholds increase somewhat during periods of intense earthquake activity when the waves from earthquakes occurring in rapid succession overlap. The detection and completeness thresholds are somewhat higher for events along the InyoMono Craters chain, because the network is less dense to the north of the caldera (see fig. D1). The Earlybird component of this computer system is programmed to initiate pages alerting the LVO SIC and the seismologist on duty in the USGS Menlo Park office of significant increases in the level of seismic activity within LVC/MC. The threshold for initiating a page is nominally set for a single M"3 earthquake and (or) a swarm of 20 or more smaller events per hour. Once a recognized earthquake swarm is under way, these thresholds may temporarily be raised to suppress needless frequently recurring pages.

The real-time capability of the Earthworm system (and its predecessor, the Real-Time-Processing (RTP) system) has been particularly useful in (1) tracking in real-time the spatial/temporal evolution of earthquake swarms within and adjacent to the caldera during episodes of elevated activity; (2) providing high-resolution seismicity maps (for example figs. A3, A5A, and A6) and first-motion focal mechanisms; and (3) providing P-wave arrival time data used in tomographic inversions for P-wave velocity models of the subsurface caldera structure. Hardware/software modules on the Earthworm system provide the capability to view waveforms, spectrograms, and continuous signal levels in the form of real-time seismic amplitude measurements (RSAM amplitudes; see Murray and Endo, 1989) in near real-time for earthquakes occurring within the Long Valley Caldera network and several other sub-networks within NCSN. This capability greatly enhances seismologists ability to recognize signals from unusual seismic sources such as LP or hybrid "volcanic" earthquakes and volcanic tremor.

Borehole Strainmeter (Dilatometer) Network

Four Sacks-Evertson borehole volumetric strainmeters (dilatometers) installed at depths of 200 m (650 ft) are operated in the Long Valley Caldera-Mono Craters region in a cooperative effort with the Carnegie Institution of Washington (POPA, PLV, BSP, and MCX in figure D2A). Biaxial borehole tiltmeters are installed with the BSP and MXC dilatometers. These dilatometers are sampled automatically every 10 minutes and the data are transmitted to USGS offices in Menlo Park by the Geostationary Operational Environmental Satellite (GOES). The dilatational strain averaged over the last 60 minutes is automatically computed and updated every 10 minutes by a computer in Menlo Park. The raw online data are then corrected for Earth tides and atmospheric pressure loading, determined from a theoretical Earth-tide model and onsite pressure transducers at each site. These instruments clearly resolve changes in strain to better than 0.602 ppm over several days and 0.01 ppm for periods less than a day. At still shorter periods corresponding to the seismic band (100 sec to 0.1 sec), resolution of signals below 0.001 ppm is routine.

Tiltmeter Network

Data from seven shallow (2 to10 m or 6 to 30 ft deep) borehole tiltmeter sites and a longbase Michelson tiltmeter (fig. D2oA) are sampled every 10 minutes and telemetered to USGS offices in Menlo Park at 10 minute intervals by the GEOS satellite. The borehole tiltmeter array is capable of discriminating rapid changes in tilt at the level of 5 to 10 microradians (ppm) occurring within a period of a few days to one week (Mortensen and Hopkins, 1987). The Michelson long-base tiltmeter, which is jointly operated by Roger Bilham at the University of Colorado and the USGS, is an Lshaped, water-tube instrument 0.5 km (0.3 mile) on a side. It has both greater accuracy and stability than the shallow borehole instruments, and is capable of resolving tilt changes on the order of 0.1 microradian over periods of a week. In principal, it has the capability of resolving tilt changes of 1 microradian over a period of years or more (Behr and others, 1992).

Differential Magnetic Field and Magnetotelluric Network

Magnetotelluric sites LAMT and LBMT (fig. D2B) use orthogonal, 0.5-km legs for electric field measurements and fluxgate magnetometers to monitor the electrical resistivity beneath the caldera. These sites were installed as part of a cooperative program with J. Zlotnicki of the Institut de Physique du Globe Paris (IPG) in Paris, France. Data from these instruments are telemetered by GOES satellite to USGS computers in Menlo Park.

Geodetic Networks

Two-Color EDM Network

Distance measurements for eight baselines extending from the monument CASA on the southern part of the resurgent dome (fig. D3A) using a two-color Electronic Distance Meter (EDM) have been collected two to three times a week, weather permitting, since the summer of 1983. With this measurement frequency, a week or 10 days is the minimum time interval for which meaningful displacements can be resolved under normal circumstances. Approximately 70 other baselines are measured with sampling intervals between one month and one year depending on activity levels and deformation rates. These lines generally are not used in the short-term evaluation of the Color-Code status unless the center of activity shifts away from the central part of the caldera. If necessary, measurements on a single baseline can be repeated as frequently as once every 10 seconds (this will not be an option if evolving activity in the Long Valley region might place the on-site operator at risk).

Figure D3. Monuments and baselines in the two-color electronic distance meter (EDM) (geodimeter) net work in the Long Valley region. Large dots are instrument si tes; small dots are reflector si tes. The baselines radiating from the instrument site, CASA, to the labeled refl ector sites are measured several times a week, weather permit ting. B, Locations of Global Positioning System (GPS) stations operated in a continuous mode. GPS stations operated in real-time mode enclosed in triangles.

Based on the analysis of the measurement error, Langbein and Johnson (1997) found two distinct error sources (1) a white noise source due to error in the measurement of the index of refraction of the atmosphere, and (2) a random-walk component due to localized motion of individual monuments. The white-noise error in millimeters (mm) is:

where L is the length of the line in mm. The random-walk component is:

where has the units of mm/yr. To estimate the 95 percent confidence interval in rate, the above can be combined as:

where N is the number of measurements made over the time interval, T years. Thus, for a 10-day interval on a 5-km-long baseline with three measurements, the 95 percent confidence interval in deformation rate is roughly 100 mm/yr. In the case of a 200-day interval with 70 observations, however, the 95 percent confidence interval in deformation rate drops to 4.4 mm/yr (0.2 inches/year).

These manpower-intensive, two-color EDM measurements will be phased out once the continuous GPS network (see following section) becomes fully operational and after sufficient temporal overlap (6 months to a year) in measurements has accumulated with both systems to insure continuity in the deformation time history within the caldera.

GPS Networks

Continuous and Real-Time Stations

Deformation in the region around Long Valley Caldera from 1980 to 1989 was monitored by annual trilateration surveys using Geodolite laser-ranging instruments (Savage, 1988). These annual Geodolite surveys were phased out in 1989 and replaced with an expanded network of monuments surveyed by the Global Positioning System (GPS) (fig. D4). In 1998, the USGS began a 4-year project to update geodetic monitoring in the Long Valley Caldera area with a 16 station permanent GPS network. Four stations (MWTP, LINC, KNOL, RDOM) were installed in 1998, an additional four (CAS2, DEAD, BALD, PMTN) in 1999, two (SAWC, TILC) in the summer of 2000, and an additional four (HOTC, WATC, DECH, and KDAKC) in 2001. Prior to 1998, the USGS operated permanent GPS stations at Minaret (MINS) and at June Lake (JNPR) and the Jet Propulsion Laboratory (JPL) operated stations Krakatoa (KRAK) and CASA. GPS data are collected by NT workstations in Mammoth Lakes and transferred daily to a server at the USGS Cascades Volcano Observatory located in Vancouver, Washington. Twenty four-hour RINEX files for stations installed by the project are available for download. Daily solutions for RINEX files are available at the USGS web site http://quake.wr.usgs.gov/QUAKES/crustaldef/longv.html

Data from a subset of stations are processed epoch-by-epoch in real-time on an NT workstation. The program is available from a commercial source. It offers processing features not available in what is commonly known as real-time kinematic (RTK) solutions. Current real-time processes compute solutions every 5 sec, create daily files, and plot results in real-time. A web module is used to transfer plot image files for web browser viewing.

The GPS technique has the following advantages over conventional trilateration or EDM techniques (1) line-of-sight is not required between GPS receiver sites (monuments), (2) measurements can be completed in poor weather, (3) measurements resolve both horizontal and vertical displacements, and (4) data acquisition and analysis processes are easily automated with the use of computers. Real-time GPS or epoch-by-epoch processing gives several mm accuracy over 24-hour period and cm-level accuracy in near real time.

Regional Geodetic Networks

The regional geodetic networks involve arrays of monuments that are normally surveyed on an annual basis using leveling, GPS, gravity, and two-color EDM techniques (fig. D4). These regional networks provide longterm definition of the regional deformation field and a regional context for more localized deformation within Long Valley Caldera or along the Mono-Inyo volcanic chain (Marshall, and others, 1997). Because these networks are measured infrequently, they will not normally contribute to the determination of the status for short-term fluctuations in moderate to strong unrest under condition GREEN. In the case of intense unrest and condition YELLOW (or EVENT RESPONSE), however, the USGS may initiate more frequent surveys of appropriate sections of these regional networks and incorporate the results with the two-color EDM network subject to the above error equations. Typical values for w for GPS data are 3mm and 10 mm for horizontal and vertical components, respectively. We expect the random-walk component to be approximately 2 mm/yr, although other temporally correlated errors may mask this monument noise over time intervals spanning one month to a few years.

Figure D4. Map of locations for survey-mode Global Positioning System (GPS) monuments in the Long Valley region.

Leveling Network

Leveling lines used to track vertical ground deformation in the Long Valley Caldera- Mono Craters region are shown in Figure D5. The earliest leveling measurements in the Long Valley region date back to a 1905 survey along Highway 395. Subsequent surveys along this section of the highway were conducted in 1914, 1932, 1957, 1975, 1980, annually from 1982 to 1988, and in 1992, 1995, and 1997 (Castle and others, 1984; Langbein and others, 1995; Savage, 1988). Lines 3 and 6 within the caldera were established in 1975, line 2 in 1982, lines 4 and 5 in 1983, and the Mary Lake Road line in 1989. All of these lines except the Mary Lake Road line were measured annually from 1983 to 1986 and 1988, and 1992. The Mary Lake Road line was measured in 1989, 1995, and 1997.

Figure D5. Map of level lines in the Long Valley Caldera region surveyed to track vertical deformation. Numbers identify individual leveling loops following the convention used by Savage and others (1987). The circles at Lee Vining and Tom’s Place indicate reference benchmarks.

The random error in leveling surveys can be approximated by s=aL1/2 where s is the standard deviation, a is a constant for each order, class, and epoch of leveling, and L is line length in km. (Castle, and others, 1984) estimated a‰6 for the 1905 and 1914 surveys, and a‰1.4 for all subsequent surveys through 1983. By extension, a=1.4 was used through 1991. Since 1992, the surveys have been of first order and with a=1.0. Experience shows that these values adequately account for random errors and so can be used to characterize the uncertainty in most leveling surveys, but there is also a possibility of systematic errors. Except for the 1984 survey, which seems to suffer from a serious systematic error (Savage and others, 1987), the leveling results are sufficiently accurate to characterize vertical ground movements throughout the area and in fact are the most accurate information of this type available. However, leveling surveys are time consuming, personnel intensive, and therefore expensive.

The USGS intends to replace the annual leveling surveys by survey-mode GPS measurements following a campaign planned for the summer of 2002 to jointly occupy the network monuments with leveling, GPS, and gravity measurements. Although the GPS results are likely to be less precise than leveling for vertical ground movements, GPS measurements are easier to make, are therefore more cost-effective, and provide information about horizontal as well as vertical ground movements.

Geochemical Monitoring

CO₂-Soil-Gas Concentrations

The carbon dioxide (CO₂) gas-concentration monitoring network in the Long Valley Caldera-Mono Craters region, which was established in 1995, currently consists of seven monitoring stations (Figure C6; also see McGee and Gerlach, 1998). Each station consists of a collection chamber buried in the soil, the air from which is pumped to nearby CO₂ sensors housed in U.S. Forest Service structures or culverts. Local barometric pressure is also measured at the site HS1. All monitoring sites have backup data loggers that also record ambient temperature. The data are sampled every hour and telemetered every 3 hours to the USGS Cascades Volcano Observatory in Vancouver, Washington, by GOES satellite. Stations HS1 and HS2 are located near the central portion of the tree-kill area near Horseshoe Lake with high CO₂ concentrations, whereas HS3 is located in a group campground on the margin of the area of high CO₂ concentrations. Stations located away from the Horseshoe Lake tree-kill area include SKI near Chair 19 in the Mammoth Mountain ski area, SRC in Shady Rest Campground adjacent to the U.S. Forest Service Visitors Center in Mammoth Lakes, EQF near the "Earthquake Fault" midway between Mammoth Lakes and the Ski Area, and LSP near Laurel Spring on the southern margin of the caldera.

Figure C6. Map showing stations in the Long Valley region that continuously monitor carbon dioxide (CO₂) soil-gas concentrations (circles) and a typical flight path for airborne measurements of CO₂ flux from Mammoth Mountain (dashed line centered on Mammoth Mountain). Open square indicates location of grid for periodic spot measurements of CO₂ flux.

CO₂-Flux Monitoring

The rate of discharge of CO₂ from areas of diffuse gas emission on Mammoth Mountain is monitored both by periodic gas-flux measurements over grids of stations established at several tree-kill areas and continuous flux measurements at a single automated instrument stationed near Horseshoe Lake during the snow-free portion of each year. The grid determinations of total flux at various tree-kill areas involve some 50 to 100 measurement points and contour-integration techniques applied to the measurements made over a 1 to 3 hour period. Such determinations repeated at daily intervals indicate that the total gas discharge at these areas can vary by as much as +/-50 percent in response to changes in atmospheric parameters such as barometric pressure and wind speed. A continuously recording flux instrument has been deployed at the Horseshoe Lake tree-kill area for portions of the past 3 years. Data from this site are transmitted by radio to a base station in Mammoth Lakes were a computer analyzes and stores flux measurements made at hourly intervals. The resultant record can be obtained remotely through telephone-modem connection.

Airborne CO₂ Monitoring

Airborne CO₂ surveillance is routinely carried out once or twice a year at Mammoth Mountain to establish baseline CO₂ emission rates for comparison in the event of future increases in unrest. The method as applied to Mammoth Mountain is fully documented in Gerlach and others (1999). If necessary, the method can be readily adapted for unrest at other locations in the LVCMC region, such as the resurgent dome or the Mono-Inyo Craters. The Mammoth Mountain airborne surveys are accomplished by flying a series of circular orbits with diameters of 6 to 7 km centered about the mountain's summit (fig. D6) at altitudes ranging from 2,895 m (9,500 ft) to 3,657 m (12,000 ft) and with a vertical separation between individual orbits averaging 61 m (200 ft). The measurements are made with a nondispersive infrared (NDIR) CO₂ analyzer and flow control unit interfaced with a laptop computer and chart recorder. The system is mounted in a twin-engine aircraft and configured for open-flow sampling of external air. The analyzer produces average CO₂ concentrations at 1-sec intervals. A pressure transducer produces corresponding 1-sec averages of barometric pressures of external ambient air. Outside air temperature is sampled at 1-sec intervals. Aircraft location during the survey is tracked using a GPS receiver. The concurrent data streams from all instruments are recorded in a Campbell data logger. CO₂ emission rates are calculated from these data and wind velocity data. Gas emissions at Mammoth Mountain are presently free of SO₂ (sulfur dioxide), but the airborne system can also accommodate correlation-spectrometer (COSPEC) instrumentation for SO₂ measurements should they become necessary.

Helium Isotopic Ratio in Fumarole Gas

The helium isotopic ratio (³He/⁴He) in gas discharged from a fumarole on the north side of Mammoth Mountain has been shown to vary with the degree of crustal unrest in the vicinity of the mountain. Data obtained from periodic sampling of this vent, referred to as Mammoth Mountain Fumarole (MMF), during the years 1982 to 2000 show that a period marked by a significantly increased component of magmatic helium began during the summer of 1989. This increase in the ³He/⁴He ratio corresponded with the onset of seismic swarms beneath Mammoth Mountain in May 1989 that appear to have resulted from the emplacement of a magmatic dike. Since 1989, MMF has been sampled several times each year during site visits in which its flow rate and temperature are also determined.

Hydrologic Monitoring

A network dedicated to hydrologic monitoring in Long Valley Caldera tracks changes in ground-water levels and the discharge of hot springs (Howle and Farrar, 1996). Automated measurements of the free-surface water level in five wells (LKT, CW-3, LVEW, SF, and CH10B; fig. D7A) are recorded at least every 30 minutes and telemetered every 4 hours by GOES satellite to computers in USGS offices in Sacramento, California. Similar measurements are made in two additional wells (HSL-1 and ESN) and recorded by on-site data loggers but not telemetered. Data from these data loggers are recovered during monthly site visits. Water-levels are measured monthly in a network of nine other, mostly shallow wells (CD-2, SC-1, SC-2, AP, LV-2, LV-19, SQ, SS-2, MW-14, RDO-8) at sites throughout the caldera moat (fig. D7B); these measurements provide information on seasonal changes in water-table elevations. Another 20 wells in the caldera region are measured three times a year. The combined discharge of hot springs in Hot Creek gorge is determined monthly from measurements of the differences in chemical flux in the creek at sites both upstream and downstream from the hot springs. At the downstream site (HCF), instruments in a concrete flume continuously track streamflow, temperature, and electrical conductivity. These data are recorded every 4 hours and transmitted by GOES satellite to USGS offices in Sacramento.

Figure D7. Map showing locations in the Long Valley region of continuously monitored hydrological sites (A) and shallow water wells that are monitored periodically (B).

The monitoring wells instrumented for automated water-level measurements range in depth from about 96 to 2,997 m (315 to 9,800 feet). Ground-water levels are measured using pressure transducers that are either submersed below the water surface or that measure the pressure in a gas-filled tube open below the water surface. Ground-water level data are recorded at least every 30 minutes and at a higher frequency in LKT and CW-3 if the water level changes more than 6 mm in 30 seconds. Any high-frequency data collected are not telemetered, but are stored on site and retrieved during monthly site visits. Atmospheric pressure is measured at CW3, CH10B, LVEW, LKT, and HSL and recorded every 30 minutes. Telemetered data are automatically processed every 24 hours at midnight and appended to plots of the previous 15 days data posted on a public USGS web page at http://lvo.wr.usgs.gov/HydroStudies.html. Each month, the previous month's data are filtered to remove the effects of atmospheric loading and earth-tide strain from the measured water levels. The measured and filtered data for the past month are plotted and posted on an internal USGS web page at http://eratos.wr.usgs.gov/lv/.

Water levels in each well respond with a unique sensitivity to atmospheric loading, earth tides, and volumeteric crustal strain. The measureable water-level responses in some wells to crustal strain make these wells potentially useful for detecting crustal deformation, although the water-level response to crustal strain is susceptible to complicating factors such as ground-water flow to or from the deformed, saturated rocks tapped by the well. Nevertheless, the wells being monitored with telemetery do show responses to caldera-scale and larger strain events. For three of these wells, LVEW, LKT, and CW-3, the effects of drainage are negligible for variations with periods as much as 14 days. For example, under undrained conditions, the expected water-level change in response to a step volume strain of 0.01 ppm is about 24, 5.5, and 2.5 mm, respectively in LVEW, LKT, and CW-3. Such changes are routinely measured, making these wells sensitive to the same small strain changes detected by the borehole dilatometers (Galloway and others, 1999; Rojstaczer and others, 1985).

The streamflow data collected at HCF include stage, specific conductance, and water temperature. Data are recorded every 15 minutes and are processed and posted to the Website (http://lvo.wr.usgs.gov/HydroStudies.html) every 24 hours. The processing uses an established stage-discharge relation to convert stage to discharge. Data from this site can be used to determine if significant changes in thermal-water discharge have occurred in Hot Creek Gorge.

URL: https://pubs.usgs.gov/bul/b2185/appendd.html
Maintained by: Michael Diggles
Last modified: May 18, 2005