WORLD METEOROLOGICAL ORGANIZATION =====================================

WORLD METEOROLOGICAL ORGANIZATION ===================================== WORLD METEOROLOGICAL ORGANIZATION (WMO) IN CLOSE COLLABORATION WITH THE INTERNATIONAL CIVIL AVIATION ORGANIZATION (ICAO) AND THE CIVIL AVIATION AUTHORITY OF NEW ZEALAND FOURTH INTERNATIONAL WORKSHOP ON VOLCANIC ASH Rotorua, New Zealand, 26-30 March 2007 Agenda Item 3: Latent State and Predictions Title of Paper: Overview of Volcano Monitoring for Eruption Forecasting and Alerting Authors: Marianne Guffanti, John W. Ewert 1

Overview of Volcano Monitoring for Eruption Forecasting and Alerting Marianne Guffanti 1 and John W. Ewert 2 1 U.S. Geological Survey, Reston, Virginia USA 2 U.S. Geological Survey, Cascades Volcano Observatory, Vancouver, WA USA ABSTRACT Volcano monitoring is conducted in two general modes: a forecasting mode before and between eruptions and an alerting mode when eruptive activity is detected. For the aviation sector, reliable eruption detection and rapid alerting are paramount to mitigate risks to en-route aircraft from encounters with airborne volcanic ash. While similar methods for monitoring seismicity, deformation, gas flux, and thermal changes are used for both forecasting and alerting, there are some differences between the two modes. Additional techniques used in the alerting mode include video surveillance, near-field infrasonic pressure sensors, lightning detectors, airborne infrared cameras, visual observations, satellite-based multi-spectral sensors, and weather radar. Ground-based weather radar can significantly improve the chances of meeting the fiveminute alert benchmark for aviation, particularly at night and in poor weather conditions, and the combination of seismic and radar data can provide unambiguous evidence of an ash-producing eruption. In evaluating monitoring methods, it is important to recognize that no single technique works at all volcanoes for the purposes of forecasting and alerting. The optimal approach is to have multiple monitoring data streams from different sensor types received and analyzed together at a Volcano Observatory. INTRODUCTION Monitoring volcanoes is not a routine process. During unrest (anomalous behavior that may be precursory to eruptive activity), volcanoes exhibit a wide range in the behavioral style and duration. Some restless volcanoes progress to eruption very quickly (days to weeks), others take months to years, or do not erupt even after exhibiting heightened unrest. Eruption styles can vary from relatively mild events that produce small lava flows or phreatic emissions to extremely explosive events, and eruption magnitudes can vary from erupted magma volumes of 0.001 km 3 to (rarely) >100 km 3. Generally, an eruption involves episodes of eruptive activity separated by non-eruptive intervals of hours to months. The duration of a single eruptive episode usually ranges from a few minutes to tens of hours, whereas an entire eruption can last for a day to decades. Volcano monitoring is conducted in two general modes, forecasting and alerting (Ewert and Guffanti, 2006). The forecasting mode occurs before and between eruptions as magma moves to the surface, a process that typically causes earthquakes and tremor, deforms the volcano s surface, emits magmatic gases, and changes the thermal regime of the volcano. Forecasting of expected hazardous events is especially important when time is needed for preparedness, such as when people living close to 2 volcanoes may need to evacuate. In the alerting mode, the objective is detection or confirmation of eruption onset and timely notification of actual hazards, which is paramount to mitigate risks to en-route aircraft from encounters with airborne volcanic ash. Furthermore, Volcano Observatories sometimes receive mistaken reports that a volcano is erupting (for example, normal steaming becomes more noticeable in certain weather conditions), and the alerting capability will also provide the important confirmation that a volcano is not erupting. Volcano monitoring uses a variety of ground-based, airborne, and satellite-based techniques. Transmission of monitoring data occurs via radios, phone lines, Internet, and/or satellites to scientific facilities sometimes quite distant from the monitored volcanoes for processing and analysis. Automatic, computer-based data processing systems make most data available in real to near-real time for analysis by scientists who may be located in different facilities. Interpreting monitoring data and forecasting the future behavior and eruptive potential of a restless volcano, however, are far from automatic and require complex analysis by a variety of volcanological experts as soon as the data are received. A primer of the various monitoring methods is online at http://volcanoes.usgs.gov/about/what/monitor/monitor.ht ml (last visited 8 March 2007). While similar monitoring methods are used in both modes, there are some differences. MONITORING IN THE FORECASTING MODE The main monitoring methods for forecasting volcanic activity are local seismic and geodetic networks on volcanoes, gas-emission measurements, and multi-spectral satellite surveillance. Seismic monitoring is the most widely employed, best understood, and most reliable monitoring technique in either mode. Seismic networks are used to locate earthquakes in three dimensions as well as determine their timing and magnitudes, track any migration of zones of seismicity, detect tremor associated with fluid (gas or magma) moving through a constriction, and characterize other physical processes at the source of the seismic signal. Digital broadband seismometers, which can resolve a wide range of ground shaking and usually do not go off scale or clip during intense activity, provide more data about the physical processes at the source of the seismic signal than traditional short-period analog seismometers (Figure 1). However, broadband instruments are more expensive and require more power and higherbandwidth telemetry. A seismic network with a mix of broadband and short-period instruments is a good solution in terms of overall cost, utility, and reliability.

Measurement of the displacement, or deformation, of the ground surface of a volcanic edifice is used to model the geometry and depth of the subsurface magmatic source and to track the rise of magma toward the surface. Deformation also is known to precede earthquake activity at some volcanoes and thus can provide some of the earliest indication of anomalous activity (see Wicks et al., 2002). The primary techniques for monitoring volcanic deformation (e.g., uplift, subsidence, fracturing) are GPS, interferometric synthetic aperture radar (InSAR), borehole strainmeters, and electronic tiltmeters. Continuously telemetered data from permanent GPS arrays provide good temporal coverage of volcanic deformation, and such arrays are increasingly being deployed at volcanoes worldwide. InSAR is a satellite-based technique with good spatial coverage of a volcanic edifice; however, repeat observations are available only weeks or months apart. Electronic tiltmeters and borehole strainmeters are sensitive indicators of short-term volcanic processes at shallow depths and provide data in real-time. The optimal approach for deformation monitoring in the forecasting mode is to use a combination of GPS, InSAR, and tiltmeters or borehole strainmeters at a volcano. Dzurisin (2007) provides a comprehensive overview of modern volcano deformation monitoring. Magma contains significant quantities of dissolved gases, which can separate from the magma at depth and be released into the atmosphere before, during, and after eruptions. The primary objective of gas monitoring is to determine changes in the type and rate of released gases, primarily carbon dioxide and sulfur dioxide. Various airborne and ground-based methods are used (see http://volcanoes.usgs.gov/about/what/monitor/gas/gasm onitor.html; last visited 8 March 2007). A promising new satellite-based tool for monitoring volcanic degassing in the forecasting mode is the Ozone Monitoring Instrument (OMI), an ultraviolet sensor on a polar-orbiting NASA satellite launched in 2004. In addition to delineating the distinctive sulfur dioxide clouds that are produced by explosive eruptions, OMI has the sensitivity and resolution to reveal the subtle degassing of sulfur dioxide that can occur before eruptions (Carn et al., 2006). As more volcanoes are studied with OMI, important new insights about the amount and timing of magmatic degassing can be expected. Changes in thermal features often accompany other signs of unrest. Existing thermal features at a volcano may increase their thermal output and/or expand in size prior to eruptive activity. Ground-based, airborne, and satellitebased sensors are employed to monitor and quantify thermal signals. Thus far, satellite systems have been most widely employed (Harris et al., 2000). Temporal and spatial resolutions, as well as meteorological factors, can limit the utility of these observing systems. Volcano monitoring is best started well before unrest begins to escalate at a volcano. Waiting to deploy a proper monitoring effort until a hazardous volcano awakens and an unrest crisis is building means that scientists, civil authorities, businesses, and citizens are caught in a reactive stance, trying to get instruments and mitigation measures in place before the situation worsens. Precious data as well as time are lost in the weeks it can take to deploy a response to a reawakening volcano. Ewert et al. (2005) present a methodology for systematically determining what level of monitoring coverage should be in place at a volcano before unrest escalates, based on an assessment of the threat posed by that volcano. In the United States, many threatening volcanoes lack sufficient monitoring for useful hazard forecasts. MONITORING IN THE ALERTING MODE In the alerting mode, the objective is detection of eruption onset and timely notification of actual hazards produced by the eruption. As with eruption forecasting, seismic monitoring is the primary method in the alerting mode. Given enough telemetered stations around a volcano, seismic data alone sometimes can be enough to confirm that an eruption is in progress. For example, at Augustine Volcano, Alaska, in 2006, seismic data were the basis for raising the color-coded alert level to Orange (volcano is exhibiting heightened unrest with increased likelihood of eruption) on 10 January 2006 and then to Red 8.5 hr later on 11 January 2006, which alerted the public and the aviation sector that an explosive eruption was in progress. However, at some frequently active volcanoes, seismic indicators of the onset of an eruptive phase may be difficult to distinguish from a high level of background seismicity (e.g., at Redoubt Volcano in 1989; Power et al., 1994). Having other corroborating observations is critical for full confidence. Other techniques useful for detecting and confirming eruptions include real-time geodetic methods (tiltmeters, GPS, borehole strainmeters), near-field (<20 km) infrasonic pressure sensors, lightning detectors, video surveillance (in fair weather), airborne infrared cameras, visual observation (e.g., pilot reports), weather radar, and satellite-based multi-spectral sensors (visible, infrared, and ultraviolet wavelengths). A recent example of the use of many of these techniques for eruption alerting is the response by the Alaska Volcano Observatory to the 2006 eruption of Augustine Volcano (Power et al., 2006). Although satellite remote-sensing techniques typically do not have data-download rates rapid enough for real-time alerting of eruption onset, they are useful for confirming the occurrence and size of an eruption, particularly at volcanoes with no ground-based monitoring (for example, at Anatahan volcano in the Northern Mariana Islands in 2003; Guffanti et al., 2005). Weather Radar Of the various methods useful for detecting the onset of eruptive activity, ground-based weather radar is particularly effective for confirming ash hazards. Radar has proven useful in detecting and characterizing volcanicash plumes from numerous eruptions, beginning with the 1980 eruption of Mount St. Helens (Harris et al., 1981). Radar can be used to estimate plume heights, ash-particle sizes, and the direction and speed of ash clouds; radar observations of the eruptive phenomenon are available instantly and can be made at night and in poor weather conditions. A specific example of radar s utility in providing rapid warnings of ash hazards to aviation is given by Hoblitt and 3

Quaas Weppen (1999). In 1997-1998, Mexico s National Center for the Prevention of Disasters (CENAPRED) and the U.S. Geological Survey deployed an experimental ground-based Doppler radar in Mexico City to track the height, direction, and speed of ash plumes from Popocatepetl Volcano ~60 km distant. Seismic data were collected and analyzed in conjunction with the radar data. Simultaneous occurrence of a seismic event and a strong radar reflector over the volcano provided incontrovertible evidence that an explosive event had occurred and that ash was in the air. When the combined real-time seismic and radar data confirmed an ash-producing eruption had occurred, CENAPRED would quickly call Mexico City International Airport to warn air-traffic controllers of imminent ash hazards. In response to stated needs of the aviation industry, U.S. Volcano Observatories strive to notify regional air-traffic control centers by telephone within five minutes of an eruption. To significantly improve the chances of meeting the five-minute alert benchmark, the U.S. Geological Survey s Volcano Hazard Program is acquiring a groundbased, portable, Doppler radar (C-band) for deployment where explosive eruptions pose hazards to aviation. Radar operated for detecting meteorological phenomena is useful but not ideal for detecting ash clouds; 360-degreevolumetric scans typical for meteorological purposes can take up to 10 minutes to complete, and various settings may not be optimized for ash. By operating its own radar, the USGS will have full operational control of the unit with the benefits of virtually no delay between data acquisition and image availability, full access to archived data, ability to move the system to different volcanic sites, and ability to merge radar data with other geophysical data streams such as seismic data. CONCLUSION In evaluating monitoring methods, it is important to recognize that no single technique works at all volcanoes for the purposes of forecasting and alerting. Volcanoes are complex systems, and scientists need to have multiple sources of data available as quickly as possible on which to base forecasts and eruption alerts. The optimal approach is to have multiple monitoring data streams from different sensor types received and analyzed together by volcano specialists. ACKNOWLEDGMENTS Reference herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States Government or any agency thereof. REFERENCES Carn, S., N.A. Krotkov, and A.J. Krueger, 2006: Monitoring global degassing with OMI: Abstracts Volume, Fourth Conference Cities on Volcanoes. International Association of Volcanology and Chemistry of the Earth s Interior, 23-27 January 2006, Quito, Ecuador, p. 86. Dzurisin, D., 2007: Volcano deformation: Geodetic monitoring techniques: Springer-Verlag, New York, 441 p. Ewert, J.W., M. Guffanti, and T.L. Murray, 2005: An assessment of volcanic threat and monitoring capabilities in the United States framework for a National Volcanic Early Warning System (NVEWS). U.S. Geological Survey Open-File Report 2005-1164, 62 p. [online at http://pubs.usgs.gov/of/2005/1164 Last visited 6 Mar. 2007] Ewert, J.W., and Guffanti, M., 2006: Assessing volcanic threat and prioritizing volcano monitoring in the United States. Abstracts Volume, Fourth Conference of Cities on Volcanoes, International Association of Volcanology and Chemistry of the Earth s Interior, 23-27 January 2006, Quito, Ecuador, p. 91. Guffanti, M., J.W. Ewert, G. Swanson, G. Gallina, G., and G. Bluth, 2005: The volcanic-ash hazard to aviation during the 2003-2004 eruptive activity of Anatahan Volcano, Commonwealth of the Northern Mariana Islands. Journal of Volcanology and Geothermal Research 146, 241-255. Harris, D.M., W.I. Rose, R. Roe, and M.R. Thompson, 1981: Radar observations of ash eruptions. U.S. Geological Survey Professional Paper 1250, 323-333. Harris, A.J.L., L.P. Flynn, K. Dean, E. Pilger, M. Wooster, C. Okubo, P. Mouginis-Mark, H. Garbeil, C. Thornber, S. De la Cruz, D. Rothery, and R. Wright, 2000: Real-time satellite monitoring of volcanic hot spots. Remote Sensing of Active Volcanism, edited by P.J. Mouginis-Mark, J.A. Crisp, and J.H. Fink, American Geophysical Union Monograph 116, 139-160. Hoblitt, R.P., and R. Quaas Weppen, 1999: Doppler radar as a volcano monitoring tool (abstract). International Symposium on Popocatepetl Volcano, 22-24 March, 1999, Mexico City, p. 19. Power, J.A., J.C. Lahr, R.A. Page, B.A. Chouet, C.D. Stephens, D.H. Harlow, T.L. Murray, and J.N. Davies, 1994: Seismic evolution of the 1989-1990 eruption sequence of Redoubt Volcano, Alaska. Journal of Volcanology and Geothermal Research, 62, 69-94. Power, J. A., C. J. Nye, M. L. Coombs, R. L. Wessels, P. F. Cervelli, J. Dehn, K. L. Wallace, J. T. Freymueller, and M. P. Doukas, 2006: The reawakening of Alaska's Augustine Volcano, Eos Trans. AGU, 87(37), 373. Wicks, C.W. Jr., D. Dzurisin, S. Ingebritsen, W. Thatcher Z. Lu, and J. Iverson, 2002: Magmatic activity beneath the quiescent Three Sisters volcanic center, central Oregon Cascade Range, USA. Geophysical Research Letters, 29 doi:10.1029/2001gl014205, 2002 4

Figure 1. Two seismograms from the same earthquake measured by different instruments. Top panel is a clipped seismogram from a traditional short-period instrument where the amount of ground shaking has exceeded the range of the instrument. Bottom panel is from a modern broadband instrument (at a slightly greater distance from the earthquake) which is responsive to a much larger range of shaking and thus less likely to clip. The signal in the bottom panel can be better characterized as to whether it results from a major eruption, small explosion, rock fall, snow avalanche, mudflow, or other common event at a volcano (from Ewert et al., 2005). ORANGE issued at 9:10 PM AST on 10Jan2006 8.5 hr later, RED issued at 5:50 AM AST on 11Jan2006 Figure 2. Seismic record from station AUH at Augustine Volcano, Alaska, showing when the Alaska Volcano Observatory issued color-coded alert levels on 10-11 January 2006. The Observatory had raised the color code from Green to Yellow on 29 November 2005. AST is Alaska Standard Time. 5