Deep learning categorization of infrasound array data.

We develop a deep learning-based infrasonic detection and categorization methodology that uses convolutional neural networks with self-attention layers to identify stationary and non-stationary signals in infrasound array processing results. Using features extracted from the coherence and direction-of-arrival information from beamforming at different infrasound arrays, our model more reliably detects signals compared with raw waveform data. Using three infrasound stations maintained as part of the International Monitoring System, we construct an analyst-reviewed data set for model training and evaluation. We construct models using a 4-category framework, a generalized noise vs non-noise detection scheme, and a signal-of-interest (SOI) categorization framework that merges short duration stationary and non-stationary categories into a single SOI category. We evaluate these models using a combination of k-fold cross-validation, comparison with an existing "state-of-the-art" detector, and a transportability analysis. Although results are mixed in distinguishing stationary and non-stationary short duration signals, f-scores for the noise vs non-noise and SOI analyses are consistently above 0.96, implying that deep learning-based infrasonic categorization is a highly accurate means of identifying signals-of-interest in infrasonic data records.

Generalization of serotonin and dopamine ligands to the discriminative stimulus effects of different doses of ±3,4-methylenedioxymethamphetamine.

Studies that have attributed the discriminative stimulus effects of ±3,4-methylenedioxymethamphetamine (MDMA) to serotonergic mechanisms typically use a relatively low training dose of 1.5 mg/kg. The role of serotonin in the discriminative stimulus effects of higher doses of MDMA is, however, unknown. Separate groups of rats were trained to discriminate MDMA (1.5 or 3.0 mg/kg) from saline using a two-lever, food-reinforced drug-discrimination procedure. Generalization tests were carried out with a range of serotonin and dopamine ligands. Fluoxetine (0.3-3 mg/kg), clomipramine (1-10 mg/kg) and meta-chlorophenylpiperazine (0.3-2 mg/kg) dose-dependently substituted for the 1.5 mg/kg MDMA stimulus, but not the 3.0 mg/kg MDMA stimulus. 8-OH-DPAT (0.03-0.3 mg/kg) and RU-24969 (0.3-3 mg/kg) substituted for both the low-dose and the high-dose MDMA stimulus. The generalization dose-effect curve produced by 2,5-dimethoxy-4-iodoamphetamine (0.3-3 mg/kg) was shifted to the right for the 3.0 mg/kg MDMA-trained group. Amphetamine (0.25 and 0.5 mg/kg) and apomorphine (0.125 and 0.25 mg/kg) substituted for the 3.0 mg/kg, but not the 1.5 mg/kg MDMA stimulus. The results suggest some differences in the role of serotonin and dopamine in the discriminative stimulus effects of a low versus a higher dose of MDMA.

Serotonin antagonists fail to alter MDMA self-administration in rats.

Acute exposure to ±3,4-methylenedioxymethamphetamine (MDMA) preferentially increases release of serotonin (5-HT), and a role of 5-HT in many of the behavioral effects of acute exposure to MDMA has been demonstrated. A role of 5-HT in MDMA self-administration in rats has not, however, been adequately determined. Therefore, the present study measured the effect of pharmacological manipulation of some 5-HT receptor subtypes on self-administration of MDMA. Rats received extensive experience with self-administered MDMA prior to tests with 5-HT ligands. Doses of the 5-HT1A antagonist, WAY 100635 (0.1-1.0mg/kg), 5-HT1B antagonist, GR 127935 (1.0-3.0mg/kg), and the 5-HT2A antagonist, ketanserin (1.0-3.0mg/kg) that have previously been shown to decrease self-administration of other psychostimulants and that decreased MDMA-produced hyperactivity in the present study did not alter MDMA self-administration. Experimenter-administered injections of MDMA (10.0mg/kg, ip) reinstated extinguished drug-taking behavior, but this also was not decreased by any of the antagonists. In contrast, both WAY 100635 and ketanserin, but not GR 127935, decreased cocaine-produced drug seeking in rats that had been trained to self-administered cocaine. The 5-HT1A agonist, 8-OH-DPAT (0.1-1.0mg/kg), but not the 5-HT1B/1A agonist, RU 24969 (0.3-3.0mg/kg), decreased drug-seeking produced by the reintroduction of a light stimulus that had been paired with self-administered MDMA infusions. These findings suggest a limited role of activation of 5-HT1A, 5-HT1B or 5-HT2 receptor mechanisms in MDMA self-administration or in MDMA-produced drug-seeking following extinction. The data suggest, however, that 5-HT1A agonists inhibit cue-induced drug-seeking following extinction of MDMA self-administration and might, therefore, be useful adjuncts to therapies to limit relapse to MDMA use.

Infrasonic wind noise under a deciduous tree canopy.

In a recent paper, the infrasonic wind noise measured at the floor of a pine forest was predicted from the measured wind velocity spectrum and profile within and above the trees [Raspet and Webster, J. Acoust. Soc. Am. 137, 651-659 (2015)]. This research studies the measured and predicted wind noise under a deciduous forest with and without leaves. A calculation of the turbulence-shear interaction pressures above the canopy predicts the low frequency peak in the wind noise spectrum. The calculated turbulence-turbulence interaction pressure due to the turbulence field near the ground predicts the measured wind noise spectrum in the higher frequency region. The low frequency peak displays little dependence on whether the trees have leaves or not. The high frequency contribution with leaves is approximately an order of magnitude smaller than the contribution without leaves. Wind noise levels with leaves are very similar to the wind noise levels in the pine forest. The calculated turbulence-shear contribution from the wind within the canopy is shown to be negligible in comparison to the turbulence-turbulence contribution in both cases. In addition, the effect of taller forests and smaller roughness lengths than those of the test forest on the turbulence-shear interaction is simulated based on measured meteorological parameters.

Wind fence enclosures for infrasonic wind noise reduction.

A large porous wind fence enclosure has been built and tested to optimize wind noise reduction at infrasonic frequencies between 0.01 and 10 Hz to develop a technology that is simple and cost effective and improves upon the limitations of spatial filter arrays for detecting nuclear explosions, wind turbine infrasound, and other sources of infrasound. Wind noise is reduced by minimizing the sum of the wind noise generated by the turbulence and velocity gradients inside the fence and by the area-averaging the decorrelated pressure fluctuations generated at the surface of the fence. The effects of varying the enclosure porosity, top condition, bottom gap, height, and diameter and adding a secondary windscreen were investigated. The wind fence enclosure achieved best reductions when the surface porosity was between 40% and 55% and was supplemented by a secondary windscreen. The most effective wind fence enclosure tested in this study achieved wind noise reductions of 20-27 dB over the 2-4 Hz frequency band, a minimum of 5 dB noise reduction for frequencies from 0.1 to 20 Hz, constant 3-6 dB noise reduction for frequencies with turbulence wavelengths larger than the fence, and sufficient wind noise reduction at high wind speeds (3-6 m/s) to detect microbaroms.

Wind noise under a pine tree canopy.

It is well known that infrasonic wind noise levels are lower for arrays placed in forests and under vegetation than for those in open areas. In this research, the wind noise levels, turbulence spectra, and wind velocity profiles are measured in a pine forest. A prediction of the wind noise spectra from the measured meteorological parameters is developed based on recent research on wind noise above a flat plane. The resulting wind noise spectrum is the sum of the low frequency wind noise generated by the turbulence-shear interaction near and above the tops of the trees and higher frequency wind noise generated by the turbulence-turbulence interaction near the ground within the tree layer. The convection velocity of the low frequency wind noise corresponds to the wind speed above the trees while the measurements showed that the wind noise generated by the turbulence-turbulence interaction is near stationary and is generated by the slow moving turbulence adjacent to the ground. Comparison of the predicted wind noise spectrum with the measured wind noise spectrum shows good agreement for four measurement sets. The prediction can be applied to meteorological estimates to predict the wind noise under other pine forests.

RU 24969-produced adipsia and hyperlocomotion: differential role of 5HT 1A and 5HT 1B receptor mechanisms.

RU 24969 is a widely used, but non-selective, 5-HT1B/1A agonist that decreases fluid consumption and increases forward locomotion. The mechanism underlying these behavioural responses is not, however, well understood. Accordingly, effects of the selective 5-HT1A and 5-HT1B antagonists, WAY 100635, and GR 127935, respectively, on these two responses to RU 24969 were determined. RU 24969 (0.03-3.0mg/kg, s.c.) dose-dependently decreased water consumption in water deprived rats. This effect was attenuated by GR 127935 (3.0mg/kg), but not by WAY 100635 (1.0mg/kg). RU 24969 (0.3-3.0mg/kg) dose-dependently increased forward locomotion but a higher dose was required to produce this response than the adipsic response. The increased locomotor response was attenuated by WAY 100635 (1.0mg/kg), but not GR 127935 (3.0mg/kg). These results suggest that RU 24969-induced adipsia is mediated by 5-HT1B mechanisms, while RU 24969-induced hyperlocomotion is mediated by 5-HT1A mechanisms.

Wind noise measured at the ground surface.

Measurements of the wind noise measured at the ground surface outdoors are analyzed using the mirror flow model of anisotropic turbulence by Kraichnan [J. Acoust. Soc. Am. 28(3), 378-390 (1956)]. Predictions of the resulting behavior of the turbulence spectrum with height are developed, as well as predictions of the turbulence-shear interaction pressure at the surface for different wind velocity profiles and microphone mounting geometries are developed. The theoretical results of the behavior of the velocity spectra with height are compared to measurements to demonstrate the applicability of the mirror flow model to outdoor turbulence. The use of a logarithmic wind velocity profile for analysis is tested using meteorological models for wind velocity profiles under different stability conditions. Next, calculations of the turbulence-shear interaction pressure are compared to flush microphone measurements at the surface and microphone measurements with a foam covering flush with the surface. The measurements underneath the thin layers of foam agree closely with the predictions, indicating that the turbulence-shear interaction pressure is the dominant source of wind noise at the surface. The flush microphones measurements are intermittently larger than the predictions which may indicate other contributions not accounted for by the turbulence-shear interaction pressure.

Improved prediction of the turbulence-shear contribution to wind noise pressure spectra.

In previous research [Raspet et al., J. Acoust. Soc. Am. 123(3), 1260-1269 (2008)], predictions of the low frequency turbulence-turbulence and turbulence-mean shear interaction pressure spectra measured by a large wind screen were developed and compared to the spectra measured using large spherical wind screens in the flow. The predictions and measurements agreed well except at very low frequencies where the turbulence-mean shear contribution dominated the turbulence-turbulence interaction pressure. In this region the predicted turbulence-mean shear interaction pressure did not show consistent agreement with microphone measurements. The predicted levels were often much larger than the measured results. This paper applies methods developed to predict the turbulence-shear interaction pressure measured at the ground [Yu et al., J. Acoust. Soc. Am. 129(2), 622-632 (2011)] to improve the prediction of the turbulence-shear interaction pressure above the ground surface by incorporating a realistic wind velocity profile and realistic turbulence anisotropy. The revised prediction of the turbulence-shear interaction pressure spectra compares favorably with wind-screen microphone measurements in large wind screens at low frequency.

Measurement of wind noise levels in streamlined probes.

This paper investigates the wind noise pressure spectra measured by aerodynamically designed devices in turbulent flow. Such measurement probes are often used in acoustic measurements in wind tunnels to reduce the pressure fluctuations generated by the interaction of the devices with the incident flow. When placed in an outdoor turbulent environment however, their performance declines noticeably. It is hypothesized that these devices are measuring the stagnation pressures generated by the cross flow components of the turbulence. Predictions for the cross flow contribution to the stagnation pressure spectra based on measured velocity spectra are developed, and are then compared to the measured pressure spectra in four different probe type devices in windy conditions outdoors. The predictions agree well with the measurements and show that the cross flow contamination coefficient is on the order of 0.5 in outdoor turbulent flows in contrast to the published value of 0.15 for measurements in a turbulent jet indoors.

Low frequency wind noise contributions in measurement microphones.

In a previous paper [R. Raspet, et al., J. Acoust. Soc. Am. 119, 834-843 (2006)], a method was introduced to predict upper and lower bounds for wind noise measured in spherical wind-screens from the measured incident velocity spectra. That paper was restricted in that the predictions were only valid within the inertial range of the incident turbulence, and the data were from a measurement not specifically designed to test the predictions. This paper extends the previous predictions into the source region of the atmospheric wind turbulence, and compares the predictions to measurements made with a large range of wind-screen sizes. Predictions for the turbulence-turbulence interaction pressure spectrum as well as the stagnation pressure fluctuation spectrum are calculated from a form fit to the velocity fluctuation spectrum. While the predictions for turbulence-turbulence interaction agree well with measurements made within large (1.0 m) wind-screens, and the stagnation pressure predictions agree well with unscreened gridded microphone measurements, the mean shear-turbulence interaction spectra do not consistently appear in measurements.

Role of regulator of G protein signaling 16 in inflammation-induced T lymphocyte migration and activation.

Chemokine-induced T lymphocyte recruitment to the lung is critical for allergic inflammation, but chemokine signaling pathways are incompletely understood. Regulator of G protein signaling (RGS)16, a GTPase accelerator (GTPase-activating protein) for Galpha subunits, attenuates signaling by chemokine receptors in T lymphocytes, suggesting a role in the regulation of lymphocyte trafficking. To explore the role of RGS16 in T lymphocyte-dependent immune responses in a whole-organism model, we generated transgenic (Tg) mice expressing RGS16 in CD4(+) and CD8(+) cells. rgs16 Tg T lymphocytes migrated to CC chemokine ligand 21 or CC chemokine ligand 12 injection sites in the peritoneum, but not to CXC chemokine ligand 12. In a Th2-dependent model of allergic pulmonary inflammation, CD4(+) lymphocytes bearing CCR3, CCR5, and CXCR4 trafficked in reduced numbers to the lung after acute inhalation challenge with allergen (OVA). In contrast, spleens of sensitized and challenged Tg mice contained increased numbers of CD4(+)CCR3(+) cells producing more Th2-type cytokines (IL-4, IL-5, and IL-13), which were associated with increased airway hyperreactivity. Migration of Tg lymphocytes to the lung parenchyma after adoptive transfer was significantly reduced compared with wild-type lymphocytes. Naive lymphocytes displayed normal CCR3 and CXCR4 expression and cytokine responses, and compartmentation in secondary lymphoid organs was normal without allergen challenge. These results suggest that RGS16 may regulate T lymphocyte activation in response to inflammatory stimuli and migration induced by CXCR4, CCR3, and CCR5, but not CCR2 or CCR7.