ALBERT

Description: We present a methodology for infrasonic remote sensing of winds in the stratosphere that does not require discrete ground-truth events. Our method uses measured time delays between arrays of sensors to provide group velocities (referred to here as celerities) and then minimizes the difference between observed and predicted celerities by perturbing an initial atmospheric specification. Because we focus on interarray propagation effects, it is not necessary to simulate the full propagation path from source to receiver. This feature allows us to use a relatively simple forward model that is applicable over short-regional distances. By focusing on stratospheric returns, we show that our non-linear inversion scheme converges much better if the starting model contains a strong stratospheric duct. Using the Horizontal Wind Model (HWM)/Mass Spectrometer Incoherent Scatter (MSISE) empirical climatology as a starting model, we demonstrate that the inversion scheme is robust to large uncertainties in backazimuth, but that uncertainties in the measured trace velocity and celerity require the use of prior constraints to ensure suitable convergence. The inversion of synthetic data, using realistic estimates of measurement error, shows that our scheme will nevertheless improve upon a starting model under most scenarios. The inversion scheme is applied to infrasound data recorded from a large event on 2010 December 25, which is presumed to be a bolide, using data from a nine-element infrasound network in Utah. We show that our recorded data require a stronger zonal wind speed in the stratosphere than is present in the HWM profile, and are more consistent with the Ground-to-Space (G2S) profile.

Description: The mathematical framework used in the Bayesian Infrasonic Source Localization (BISL) methodology is examined and simplified providing a generalized method of estimating the source location and time for an infrasonic event. The likelihood function describing an infrasonic detection used in BISL has been redefined to include the von Mises distribution developed in directional statistics and propagation-based, physically derived celerity-range and azimuth deviation models. Frameworks for constructing propagation-based celerity-range and azimuth deviation statistics are presented to demonstrate how stochastic propagation modelling methods can be used to improve the precision and accuracy of the posterior probability density function describing the source localization. Infrasonic signals recorded at a number of arrays in the western United States produced by rocket motor detonations at the Utah Test and Training Range are used to demonstrate the application of the new mathematical framework and to quantify the improvement obtained by using the stochastic propagation modelling methods. Using propagation-based priors, the spatial and temporal confidence bounds of the source decreased by more than 40 per cent in all cases and by as much as 80 per cent in one case. Further, the accuracy of the estimates remained high, keeping the ground truth within the 99 per cent confidence bounds for all cases.

Description: We use the Rayleigh integral (RI) as an approximation to the Helmholtz–Kirchoff integral to model infrasound generation and propagation from underground chemical explosions at distances of 250 m out to 5 km as part of the Source Physics Experiment (SPE). Using a sparse network of surface accelerometers installed above ground zero, we are able to accurately create synthetic acoustic waveforms and compare them to the observed data. Although the underground explosive sources were designed to be symmetric, the resulting seismic wave at the surface shows an asymmetric propagation pattern that is stronger to the northeast of the borehole. This asymmetric bias may be attributed to the subsurface geology and faulting of the area and is observed in the acoustic waveforms. We compare observed and modelled results from two of the underground SPE tests with a sensitivity study to evaluate the asymmetry observed in the data. This work shows that it is possible to model infrasound signals from underground explosive sources using the RI and that asymmetries observed in the data can be modelled with this technique.

Description: Celerity-range models, where celerity is defined as the epicentral distance divided by the total traveltime (similar to the definition of group velocity for dispersed seismic surface waves), can be used for the association of infrasound automatic detections, for event location and for the validation of acoustic propagation simulations. Signals recorded from ground truth events are used to establish celerity-range models, but data coverage is uneven in both space and time. To achieve a high density of regional recordings we use data from USArray seismic stations recording air-to-ground coupled waves from explosions during the summers of 2004–2008 at the Utah Training and Test Range, in the western United States, together with data from five microbarograph arrays at regional distances (〈1000 km). We have developed a consistent methodology for analysing the infrasound and seismic data, including choosing filter characteristics from a limited group of two-octave wide filter bands and picking the maximum peak-to-peak arrival. We clearly observe tropospheric, thermospheric and stratospheric arrivals, in agreement with regional ray tracing models. Due to data availability and the dependence of infrasound propagation on the season, we develop three regional celerity-range models for the U.S. summer, with a total of 2211 data picks. The new models suggest event locations using the Geiger method could be improved in terms of both accuracy (up to 80 per cent closer to ground truth) and precision (error ellipse area reduced by 〉90 per cent) when compared to those estimated using the global International Data Center model, particularly for events where stations detect arrivals at ranges 〈350 km. Whilst adding data-based prior information into the Bayesian Infrasound Source Localization (BISL) method is also shown to increase precision, to increase accuracy, the parameter space must be expanded to include station-specific celerity distributions.

Description: To assess infrasound detector performance, automated detections by the progressive multichannel correlation method ( Cansi, 1995 ) and the adaptive F -detector (AFD; Arrowsmith et al. , 2009 ) are compared with signals identified by five independent analysts. Each detector was applied to a 4-hr time sequence recorded by the Korean seismoacoustic array, CHNAR, composed of small (〈100 m) and large (~1000 m) aperture subarrays. Detector effectiveness was estimated for a selection of array elements and detection thresholds under low- and high-noise conditions. Estimated receiver operating characteristic based on events identified by analysts evaluates the change in detection probability ( P d ) and false-alarm probability ( P f ) for various detector parameters. This empirical study documents that the use of smaller aperture subarrays by both detectors increases P d with smaller p -values recommended for AFD to minimize P f . P d is impacted most by noise level, as shown by an increase in detections for average root mean square amplitudes from 1.2 to 3.2 MPa. Critical to this assessment is the identification of the source of the noise, constrained by signal characteristics, complementary seismic observations, and realistic atmospheric modeling. Based on signal characteristics (correlation value, phase velocity, and detection azimuth) and raytracing using global and local weather datasets, we conclude that during low-noise conditions some detections from local distances (10–50 km) are affected by surface wind direction, and a second set is affected by tropospheric winds. This illustrates the role that surface and higher-atmosphere winds play in array performance when assessing signals from regional infrasound sources in which local detections may be considered as noise or clutter. Electronic Supplement: Figures showing summary of detection results and polar plots of correlation estimates and phase velocity with respective to azimuth.

Description: We extend a time-frequency discrimination algorithm, developed in an earlier article (Arrowsmith et al., 2006), for application to seismic-array data. Spectrograms evaluated at each component of an array are stacked and then converted into binary form for computation of discriminants. Because noise can bias the discriminants, we develop a procedure for removing the effect of noise on the discriminants. The binary spectrograms are randomized where the spectral amplitude of the signal is similar to the mean spectral amplitude of the pre-event noise at that frequency. The formulism of Arrowsmith et al. (2006) is further extended by modifying the objective function used to optimize the values of input parameters and by removing high-frequency and low-frequency spectral content. We apply the method to a dataset of regional recordings of earthquakes and delay-fired mine blasts recorded at the Pinedale seismic array in Wyoming. Our results show that the utilization of array data improves the success rate for source identification. Furthermore, we find that incorporating the noise-correction procedure increases the separation between earthquakes and cast overburden blasts (the largest type of delay-fired mine blasts). In total, the algorithm successfully identifies 97.4% of the events (74 of a total of 76 events, which comprise earthquakes and cast overburden blasts).

Description: This short article explores and extends the adaptive detection algorithm recently developed by Arrowsmith, Whitaker, et al. (2008). In particular, this article highlights its application for seismic data, compares results for collocated seismic and infrasonic data, and assesses detector performance through comparison with analyst picks. We assess the adaptive detector by generating receiver-operating characteristic (ROC) curves, illustrating the trade-off between detection probability and false-alarm probability, and comparing the results with the conventional F-detector. The results show that the adaptive detector performs much better than the conventional detector for both seismic and infrasound data by maintaining high detection probabilities while significantly decreasing false-alarm probabilities, illustrating that correlated noise is ubiquitous for both types of data. The effect of the adaptation window is illustrated and shown to be especially important for infrasound data where diurnal variations in ambient noise levels are pronounced. A window choice of 1 hr (i.e., significantly less than 24 hr) is shown to be adequate for representing variations in ambient noise levels.