Skip to main content
U.S. flag

An official website of the United States government

22-17. Advancing automated infrasound signal detection and classification

We seek a Fellow to advance USGS infrasound monitoring capabilities such as automatic signal categorization (e.g., volcanic eruptions, landslides, rock falls, lahars, and debris flows), development of detection and characterization algorithms, and building a robust infrastructure for signal processing and event documentation to support real-time monitoring and scientific analyses of sources.

Description of the Research Opportunity

Low-frequency sound in the atmosphere (infrasound) is generated by many natural processes that have the potential to be hazardous to life and property. This has spurred the recent, rapid expansion in the use of infrasound as a tool for detecting and monitoring sources including volcanic eruptions, landslides, rock falls, lahars, and debris flows. The result of the expansion of infrasound in hazard monitoring has been an exponential growth in the volume of high-quality infrasound data from regional and local sensors in areas of naturals hazards such as eruptions and mass movements. Much of this data comes from arrays of closely spaced sensors that are capable of distinguishing coherent signals from noise through processing methods that have been in use for decades. However, efforts to discriminate and classify coherent signals detected on monitoring arrays are typically done by analysts, if at all. Real-time monitoring and response operations are limited by the lack of routine classification of infrasound signals, and the growing volume of data and study areas containing multiple sources (e.g., explosive volcanic eruptions that trigger lahars, or rock falls near calving glaciers) creates additional challenges. Routine infrasound classification has significant potential to support hazard monitoring and provide critical data to better understand the fundamental physics of the underlying phenomena. 

Automated infrasound detection and classification is of broad interest across the USGS Natural Hazards Mission Area and would have immediate benefits to at least two program areas. The Volcano Hazards Program (VHP) utilizes infrasound to monitor for explosive eruptions and large mass movements, including real-time detection of lahars (e.g., monitoring lahars at Mount Rainier). These are major hazards at many volcanoes in the US, making the rapid detection and characterization of such events of utmost importance; however, mass movements and explosions detected on the more than 30 existing VHP arrays are not currently systematically cataloged or characterized. The Landslide Hazards Program (LHP) collects data on landslide occurrence and processes, conducts fundamental research on landslide dynamics, and seeks to improve methods for situational awareness and early warning. The LHP has a specific requirement to monitor landslides in Prince William Sound, AK (Barry Arm, Alaska landslide and tsunami monitoring). The Barry Arm landslide in Prince William Sound is monitored with a local infrasound array that has captured numerous small mass movements and rock falls, as well as abundant glacial calving and snow avalanche events, thus providing a wide variety of signal types from multiple sources of interest in a geographically compact area. Nearby instrumentation and camera systems provide known locations, times, and durations for many of these event types. These data provide a robust training and calibration dataset to develop, refine, and test algorithms for automated and potential real-time detection and classification of infrasound signals at this unique location. 

We solicit proposals that will advance efforts to systematically detect and classify infrasound signals recorded in monitoring data, with focuses on 1) automatic signal categorization, 2) development of tools or algorithms that can improve detection and characterization capabilities, and 3) building a robust infrastructure for signal processing and event documentation that can aid in magnitude-frequency analyses of signal sources and constrain physical properties (e.g., energy released, source mass, volume flow rate). These are complicated problems that will require novel, imaginative solutions. Signals of interest span a wide range of duration, amplitude, and frequency content while occurring in settings with varying coherent and incoherent noise sources. The scope of this challenge is large, and no one proposal is likely to be able to address classification of all potential signals of interest. We support proposals that will focus on addressing tractable portions of this challenge, while considering how products would be expanded or generalized for operational use. Applicants may consider building on existing array-based signal detection used in the observatory setting (e.g., Lyons et al., 2019; Iezzi et al., 2022) since they provide a starting library of detections with which to test new classification algorithms.  

We strongly encourage applicants to consider including cutting-edge data mining and processing approaches coming out of artificial intelligence research since early studies are showing promise for aiding in the identification and categorization of infrasound signals (e.g., Thüring et al., 2015). However, these artificial intelligence approaches must be grounded by real-world observations. Therefore, it will benefit proposals to include complementary data streams to ground-truth infrasound event classification. The USGS collects multiparametric data as part of volcano and landslide monitoring efforts, including other geophysical data and a suite of remote sensing data. 

Interested applicants are strongly encouraged to contact the Research Advisor(s) early in the application process to discuss project ideas. 



Lyons, J.J., Iezzi, A.M, Fee, D., Schwager, H.F., Wech, A.W., & Haney, M.M. (2020). Infrasound generated by the 2016–2017 shallow submarine eruption of Bogoslof volcano, Alaska. Bulletin of Volcanology 82 (19). doi: 

Iezzi, A.M, Matoza, R.S., Bishop, J.W., Bhetanabhotla, S., & Fee, D. (2022) Narrow‐Band Least‐Squares Infrasound Array Processing. Seismological Research Letters, 93 (5): 2818–2833. doi: 

Thüring, T., Schoch, M., van Herwijnen, A., & Schweizer, J. (2015). Robust snow avalanche detection using supervised machine learning with infrasonic sensor arrays. Cold Regions Science and Technology, 111, 60-66. doi: 


Proposed Duty Station(s)

Anchorage, Alaska        


Areas of PhD

Geophysics, seismology, computer science, or related fields (candidates holding a Ph.D. in other disciplines, but with extensive knowledge and skills relevant to the Research Opportunity may be considered). 



Applicants must meet one of the following qualifications:  Research Geophysicist, Research Geologist, Research Computer Scientist, Research Mathematician, Research Physical Scientist 

(This type of research is performed by those who have backgrounds for the occupations stated above. However, other titles may be applicable depending on the applicant's background, education, and research proposal. The final classification of the position will be made by the Human Resources specialist.)