Infrasound, low frequency sound less than 20 Hz, is generated by both natural and anthropogenic sources. Infrasound sensors measure pressure fluctuations only in the vertical plane and are single channel. However, the most robust infrasound signal detection methods rely on stations with multiple sensors (arrays), despite the fact that these are sparse. Automated methods developed for seismic data, such as short-term average to long-term average ratio (STA/LTA), often have a high false alarm rate when applied to infrasound data. Leveraging single channel infrasound stations has the potential to decrease signal detection limits, though this cannot be done without a reliable detection method. Therefore, this report presents initial results using (1) a convolutional neural network (CNN) to detect infrasound signals and (2) unsupervised learning to gain insight into source type.
Accurately locating seismoacoustic sources with geophysical observations helps to monitor natural and anthropogenic phenomena. Sparsely deployed infrasound arrays can readily locate large sources thousands of kms away, but small events typically produce signals observable at only local to regional distances. At such distances, accurate location efforts rely on observations across smaller regional or temporary deployments which often consist of single-channel infrasound sensors that cannot record direction of arrival. Event locations can also be aided by inclusion of ground coupled airwaves (GCA). This study demonstrates how we can robustly locate a catalog of seismoacoustic events using infrasound, GCA, and seismic arrival times at local to near-regional distances. We employ a probabilistic location framework using simplified forward models. Our results indicate that both single-channel infrasound and GCA arrival times can provide accurate estimates of event location in the absence of array-based observations even when using simple models. However, one must carefully choose model uncertainty bounds to avoid underestimation of confidence intervals.
The Sound Fixing and Ranging (SOFAR) channel in the ocean allows for low frequency sound to travel thousands of kilometers, making it particularly useful for detecting underwater nuclear explosions. Suggestions that an elevated SOFAR-like channel should exist in the stratosphere date back over half a century and imply that sources within this region can be reliably sensed at vast distances. However, this theory has not been supported with evidence of direct observations from sound within this channel. Here we show that an infrasound sensor on a solar hot air balloon recorded the first infrasound detection of a ground truth airborne source while within this acoustic channel, which we refer to as the AtmoSOFAR channel. Our results support the existence of the AtmoSOFAR channel, demonstrate that acoustic signals can be recorded within it, and provide insight into the characteristics of recorded signals. Results also show a lack of detections on ground-based stations, highlighting the advantages of using balloon-borne infrasound sensors to detect impulsive sources at altitude.
Several sources of interest often generate both low-frequency acoustic and seismic signals due to energy propagation through the atmosphere and the solid Earth. Seismic and acoustic observations are associated with a wide range of sources, including earthquakes, volcanoes, bolides, chemical and nuclear explosions, ocean noise, and others. The fusion of seismic and acoustic observations contributes to a better understanding of the source, both in terms of constraining source location and physics, as well as the seismic to acoustic coupling of energy. In this review, we summarize progress in seismoacoustic data processing, including recent developments in open-source data availability, low-cost seismic and acoustic sensors, and large-scale deployments of collocated sensors from 2010 to 2022. Similarly, we outline the recent advancements in modeling efforts for both source characteristics and propagation dynamics. Finally, we highlight the advantages of fusing multiphenomenological signals, focusing on current and future techniques to improve source detection, localization, and characterization efforts. This review aims to serve as a reference for seismologists, acousticians, and others within the growing field of seismoacoustics and multiphenomenology research.
Infrasound, characterized by low-frequency sound inaudible to humans (<20 Hz), emanates from natural and anthropogenic sources. Its efficacy for monitoring phenomena necessitates robust sensing networks. Traditional ground-based infrasound sensors have limitations due to atmospheric dynamics and noise interference. Balloon-bore sensors have emerged as an alternative, offering reduced noise and improved capabilities. This study bridges clustering algorithms with balloon borne infrasound data, a domain yet to be explored. Employing K-Means, DBSCAN, and GMM algorithms on normalized and reshaped data and only normalized data from a New Zealand-based NASA balloon flight, insights into background noise at stratospheric altitudes were revealed. Despite challenges arising from distinguishing signals amid unique background noise, this research provides vital reference material for noise analysis and calibration. Beyond infrasound event capture, the dataset enriches comprehension of background noise characteristics in the southern hemisphere.
Low frequency sound below 20 Hz, also known as infrasound, is generated by both natural and anthropogenic sources. Local surface winds also generate signals within this frequency band and can dominate signals. Effectively monitoring sources of interest requires filtering out the influence of wind. Recently, the National Center for Physical Acoustics developed a 1 m fabric dome made from trampoline material that can serve as a wind filter for temporary field deployments. We assess the performance of this new dome by quantifying its overall noise reduction and show that it is an acceptable wind filter for temporary infrasound field deployments.
Natural and anthropogenic sources such as volcanoes, earthquakes, auroral processes, chemical and nuclear explosions, rocket launches, and aircraft can generate infrasound, sound with frequencies less than 20 Hz. Both the availability of infrasound data and interest in machine learning (ML) applications have grown in recent years. Large, open-source datasets are essential to solving complex ML problems, but the field of infrasound is lacking in this arena. To increase the utility of ML for infrasound, here we present new tables for infrasound event catalogs. It is the aim that these tables be incorporated into both existing and future infrasound processing pipelines to generate large datasets ripe for use with ML/DL methods.
Infrasound, with frequencies less than 20 Hz, is generated by both natural and anthropogenic sources. When one of these sources exerts a force on the atmosphere, infrasonic waves are generated. The propagation of these waves largely depends on temperature, wind speed, and wind direction. Previous work has used deep learning to accurately predict atmospheric specifications to altitudes of ~40 km. However, this model breaks down for local distances because it is too low resolution. Here we use a high-resolution meteorological dataset collected in Las Vegas, Nevada, USA to develop a deep learning model that can predict temperature, wind speed, and wind direction. Predictions are compared to ground truth observations to show that the model performs well at predicting temperature and wind direction but struggles with prediction wind speed. Model limitations and improvements are also discussed.
Here we investigate the application of ground-coupled airwaves observed by seismoacoustic stations at local to near-regional scales to detect signals of interest and determine back-azimuth information. Ground-coupled airwaves are created from incident pressure waves traveling through the atmosphere that couple to the earth and transmit as a seismic wave with retrograde elliptical motion. Previous studies at sub-local scales (<10 km from a source of interest) found the back-azimuth to the source could be accurately determined from seismoacoustic signals recorded by acoustic and 3-component seismic sensors spatially separated on the order of 10 to 150 m. The potential back-azimuth directions are estimated from the coherent signals between the acoustic and vertical seismic data, via a propagation-induced phase shift of the seismoacoustic signal. A unique solution is then informed by the particle motion of the 3-component seismic station, which was previously found to be less accurate than the seismoacoustic-sensor method. We investigate the applicability of this technique to greater source-receiver distances, from 50-100 km and up to 400 km, which contains pressure waves with tropospheric and stratospheric ray paths, respectively. Specifically, we analyze seismoacoustic sources with ground truth from rocket motor fuel elimination events at the Utah Test and Training Range (UTTR) as well as a 2020 rocket launch in Southern California. From these sources we observe evidence that while coherent signals can be seen from both sources on multiple seismoacoustic station pairs, the determined ground-coupled airwave back-azimuths are more complicated than results at more local scales. Our findings suggest more complex factors including incidence angle, coupling location, subsurface material, and atmospheric propagation effects need to be fully investigated before the ground-coupled airwave back-azimuth determination method can be applied or assessed at these further distances.
Infrasound is generated by a variety of natural and anthropogenic sources. Infrasonic waves travel through the dynamic atmosphere, which can change on the order of minutes to hours. Infrasound propagation largely depends on the wind and temperature structure of the atmosphere. Numerical weather prediction models are available to provide atmospheric specifications, but uncertainties in these models exist and they are computationally expensive to run. Machine learning has proven useful in predicting tropospheric weather using Long Short-Term Memory (LSTM) networks. An LSTM network is utilized to make atmospheric specification predictions up to ~30 km for three different training and testing scenarios: (a) the model is trained and tested using only radiosonde data from the Albuquerque, NM, USA station, (b) the model is trained on radiosonde stations across the contiguous US, excluding the Albuquerque, NM, USA station, which was reserved for testing, and (c) the model is trained and tested on radiosonde stations across the contiguous US. Long Short-Term Memory predictions are compared to a state-of-the-art reanalysis model and show cases where the LSTM outperforms, performs equally as well, or underperforms in comparison to the state-of-the-art. Regional and temporal trends in model performance across the US are also discussed. Results suggest that the LSTM model is a viable tool for predicting atmospheric specifications for infrasound propagation modeling.
The Dynamic Networks Experiment 2018 (DNE18) was a collaborative effort between Los Alamos National Laboratory (LANL), Sandia National Laboratories (SNL), Lawrence Livermore National Laboratory (LLNL) and Pacific Northwest National Laboratory (PNNL) designed to evaluate methodologies for multi-modal data ingestion and processing. One component of this virtual experiment was a quantitative assessment of current capabilities for infrasound data processing, beginning with the establishment of a baseline for infrasound signal detection. To produce such baselines, SNL and LANL exploited a common dataset of infrasound data recorded across a regional network in Utah from December 2010 through February 2011. We utilize two automated signal detectors, the Adaptive F-Detector (AFD) and the Multivariate Adaptive Learning Detector (MALD) to produce automated signal detection catalogs and an analyst-produced catalog. Comparisons indicate that automatic detectors may be able to identify small amplitude, low SNR events that cannot be identified by analyst review. We document detector performance in terms of precision and recall, demonstrating that the AFD is more precise, but the MALD has higher recall. We use a synthetic dataset of signals embedded in pink noise in order to highlight shortcomings in assessing detection algorithms for low signal to noise ratio signals which are commonly of interest to the nuclear monitoring community. For comparisons utilizing the synthetic dataset, the AFD has higher recall while precision is equal for both detectors. These results indicate that both detectors perform well across a variety of background noise environments; however, both detectors fail to identify repetitive, short duration signals arriving from similar backazimuths. These failures represent specific scenarios that could be targeted for further detector development.
Infrasound, or low frequency sound 20 Hz, is produced by a variety of natural and anthropogenic sources. Wind also generates signals within this frequency band and serves as a persistent source of infrasonic noise. Infrasound sensors measure pressure fluctuations, which scale with the ambient density and velocity fluctuations of ground winds. Here we compare four different wind noise reduction systems, or "filters", and make recommendations for their use in temporary infrasound deployments. Our results show that there are two filters that are especially effective at reducing wind noise: (1) a Hyperion high frequency (HF) shroud with a 1 m diameter metal mesh dome placed on top and (2) a Hyperion Four Port Garden Hose shroud with 4 Miracle-Gro Soaker System garden hoses. We also find that placing a 5-gallon bucket over the HF wind shroud should not be done as it provides a negligible decrease in noise up to ~ 1 Hz and then an increase in noise. We conclude that it is up to the researcher to determine which of the other filters is best for their needs based on location and expense. We anticipate this study will be used as a resource for future deployments when a wind noise reduction method is necessary but only needed for a limited time period.
Standard meteorological balloons can deliver small scientific payloads to the stratosphere for a few tens of minutes, but achieving multihour level flight in this region is more difficult. We have developed a solarpowered hot-air balloon named the heliotrope that can maintain a nearly constant altitude in the upper troposphere–lower stratosphere as long as the sun is above the horizon. It can accommodate scientific payloads ranging from hundreds of grams to several kilograms. The balloon can achieve float altitudes exceeding 24 km and fly for days in the Arctic summer, although sunset provides a convenient flight termination mechanism at lower latitudes. Two people can build an envelope in about 3.5 h, and the materials cost about $30. The low cost and simplicity of the heliotrope enables a class of missions that is generally out of reach of institutions lacking specialized balloon expertise. Here, we discuss the design history, construction techniques, trajectory characteristics, and flight prediction of the heliotrope balloon. We conclude with a discussion of the physics of solar hot-air balloon flight.
Low-frequency sound ≤20 Hz, known as infrasound, is generated by a variety of natural and anthropogenic sources. Following an event, infrasonic waves travel through a dynamic atmosphere that can change on the order of minutes. This makes infrasound event classification a difficult problem, as waveforms from the same source type can look drastically different. Event classification usually requires ground-truth information from seismic or other methods. This is time consuming, inefficient, and does not allow for classification if the event locates somewhere other than a known source, the location accuracy is poor, or ground truth from seismic data is lacking. Here,we compare the performance of the state of the art for infrasound event classification, support vector machine (SVM) to the performance of a convolutional neural network (CNN), a method that has been proven in tangential fields such as seismology. For a 2-class catalog of only volcanic activity and earthquake events, the fourfold average SVM classification accuracy is 75%, whereas it is 74% when using a CNN. Classification accuracies from the 4-class catalog consisting of the most common infrasound events detected at the global scale are 55% and 56% for the SVM and CNN architectures, respectively. These results demonstrate that using a CNN does not increase performance for infrasound event classification. This suggests that SVM should be the preferred classification method, as it is a simpler and more trustworthy architecture and can be tied to the physical properties of the waveforms. The SVM and CNN algorithms described in this article are not yet generalizable to other infrasound event catalogs. We anticipate this study to be a starting point for development of large and comprehensive, systematically labeled, infrasound event catalogs, as such catalogs will be necessary to provide an increase in the value of deep learning on event classification.
Low frequency sound ≤ 20 Hz, known as infrasound, is generated by a variety of natural and anthropogenic sources. Following an event, infrasonic waves travel through a dynamic atmosphere that can change on the order of minutes. This makes infrasound event classification a difficult problem as waveforms from the same source type can look drastically different. Event classification usually requires ground truth information from seismic or other methods. This is time consuming, inefficient, and does not allow for a classification if the event locates somewhere other than a known source, the location accuracy is poor, or ground truth from seismic data is lacking. Here we compare the performance of the state of the art for infrasound event classification, support vector machine (SVM), to the performance of a convolutional neural network (CNN), a method that has been proven in tangential fields such as seismology. For a 1-class catalog consisting of only volcanic activity and earthquake events, the 4-fold average SVM classification accuracy is 75%, while it is 74% when using a CNN. Classification accuracies from the 4-class catalog consisting of the most common infrasound events detected at the global scale are 55% and 56% for the SVM and CNN architectures, respectively. These results demonstrate that using a CNN does not increase performance for infrasound event classification. This suggests that SVM should be the preferred classification method as it is a simpler and more trustworthy architecture and can be tied to the physical properties of the waveforms. The SVM and CNN algorithms described in this paper are not yet generalizable to other infrasound event catalogs. We anticipate this study to be a starting point for the development of large and comprehensive, systematically labeled, infrasound event catalogs as such catalogs will be necessary to provide an increase in the value of deep learning on event classification.
We outline a method using gradient flow independent component analysis (ICA) to separate signals comprising the coda in a topographically complex setting.We also identify the sources of scattered signals by tracking signal backazimuths over time. The gradient flow ICA method is shown to effectively separate signals in the acoustic coda. The method correctly identifies the backazimuth of the first arrival from two 800 kg TNT equivalent explosions as well as subsequent signals scattered by the surrounding topography. Circular statistics is used to analyse the variance, mean and uniformity of calculated backazimuths. These results have strong implications for understanding the acoustic wavefield by identifying scatterers and inverting for atmospheric conditions.
This report provides details of the algorithms in the Bloodhound package for infrasound data analysis. The report provides a detailed description of the algorithms, general instructions on tuning Bloodhound for different signal types, and a complete listing of all input parameters and the complete output schema. Several Jupyter notebooks are provided with the distribution for illustrating how to use Bloodhound for different workflows.
Bowman, Daniel; Young, Eliot F.; Krishnamoorthy, Siddharth; Lees, Jonathan M.; Albert, Sarah; Komjathy, Attila; Cutts, James
Balloon-borne infrasound research began again in 2014 with a small payload launched as part of the High Altitude Student Platform (HASP; Bowman and Lees(2015)). A larger payload was deployed through the same program in 2015. These proof of concept experiments demonstrated that balloon-borne microbarometers can capture the ocean microbarom (a pervasive infrasound signal generated by ocean waves) even when nearby ground sensors are not able to resolve them (Bowman and Lees, 2017). The following year saw infrasound sensors as secondary payloads on the 2016 Ultra Long Duration Balloon flight from Wanaka, New Zealand (Bowman and Lees, 2018; Lamb et al., 2018) and the WASP 2016 balloon flight from Ft. Sumner, New Mexico (Young et al., 2018). Another payload was included on the HASP 2016 flight as well. In 2017, the Heliotrope project included a four element microbarometer network drifting at altitudes of 20-24 km on solar hot air balloons (Bowman and Albert, 2018). At the time of this writing the Trans-Atlantic Infrasound Payload (TAIP, operated by Sandia National Laboratories) and the Payload for Infrasound Measurement in the Arctic (PIMA, operated by Jet Propulsion Laboratory) are preparing to fly from Sweden to Canada aboard the PMC-Turbo balloon. The purpose of this experiment is to cross-calibrate several different infrasound sensing systems and test whether wind noise events occur in the stratosphere.