The Yushu fireball occurred at 23:23:33 UTC on December 22, 2020, over Qinghai Province, China. The coordinates of this event were about \(31.9^\circ \mathrm{N}\), \(96.2^\circ \mathrm{E}\), and the meteor shone brightly across the sky. It is suspected to have landed near Yushu County according to China’s Earthquake Network Centre (Qinghai Fireball Incident-Investigation 2020). The distance between the infrasound array in Yunnan Province which was deployed to forewarn earthquake and this event was about 1100 km (Fig. 1). The array (\(25.9^\circ \mathrm{N}\), \(102.3^\circ \mathrm{E}\)) consists of four CDC-2B infrasound sensors which ranges from 0.02 to 20 Hz in frequency band. The sample rate of data acquisition unit (DAU) is 100 Hz. On December 23, 2020, the array detected an infrasound event at 00:14:03 UTC. The PMCC method was used to estimate the wave parameters using 15 logarithmically spaced frequency bands ranging from 0.02 to 50 Hz with second order Chebyshev filters which is more stable than third and fourth order Chebyshev filters here (Garces 2013). The time-window length maintained a constant period of 50 s and was time-shifted by 25% of the window length.
As shown in Fig. 2, two arrivals were detected at 00:14:03–00:17:52 and 00:18:44–00:28:57, which are marked as A and B, respectively. The back-azimuth of arrival A is \(337^\circ\) and that of arrival B is \(326^\circ\), which correspond to the position reported by JPL. In the lower panel of Fig. 2, there seems to be an event at about 00:35:00. The signal at 00:35:00 has no corresponding PMCC family detected. This may be an interfering sound source in the array geometry, because the amplitude of signal is obviously different.
In addition to the Yunnan infrasound array, the Xinjiang infrasound array (\(40.7^\circ \mathrm{N}\), \(84.4^\circ \mathrm{E}\)) deployed for forewarning earthquake is located 1300 km from the fireball event. However, this array failed to observe the signal from the fireball with PMCC method (Fig. 3). Simulation of the infrasound in the atmosphere was utilized to explain this. To demonstrate the propagation traces of the infrasound signal, the nonlinear progressive wave equation (NPE) was used to simulate the variation in the transmission loss of the infrasound signal with distance (McDonald et al. 2011). The influence of atmospheric winds was approximated by adding the horizontal wind in the direction from the source toward the receiver and the sound speed in the stratified atmosphere. The result labeled as the effective sound speed was used to simulate the infrasound propagation in an atmosphere with sufficiently low wind speeds. Models Mass Spectrometer and Incoherent Scatter Radar Extended from the Ground through Exosphere (MSISE00) provides temperature profile which can be utilized to calculate the sound speed profile (Picone et al. 2002). Horizontal Wind Model (HWM14) can provide horizontal wind profile from the ground to an altitude of 150 km (Hedin et al. 1998). The effective sound speed profile can be obtained by adding them.
The effective sound speed profiles in the direction toward the infrasound array in Yunnan Province and Xinjiang Province are shown in Fig. 4. The maximum value of the profile between altitude 30–60 km in the direction toward Yunnan Station is bigger than that on the ground, which means the waveguide between stratosphere and the ground can be formed (Yang et al. 2007). The situation in the direction toward Xinjiang Station is different. The maximum value of effective sound speed in altitude 30–60 km is smaller than that on the ground. This means the energy propagated through the waveguide between stratosphere and the ground leaked into the area between stratosphere and thermosphere.
As shown in Fig. 5a, the most of infrasound energy is reflected to ground, which facilitates to propagate far in this waveguide between stratosphere and the ground. In contrast, a large part of infrasound energy propagated upward into thermosphere in Fig. 5b, which causes the large attenuation of the signal. The attenuation is mainly caused by the classical absorption and relaxation absorption (Sutherland et al. 2004).