Atmospheric Gravity Wave Derived from the Neutral Wind with 5-Minute Resolution Routinely Retrieved by the Meteor Radar at Mohe

Abstract

Atmospheric gravity waves (GWs) in the mesosphere-lower thermosphere (MLT) are crucial for the understanding of general circulation. However, their dynamical characteristics are hardly retrieved due to the difficulty in the high-resolution observation of wind. Therefore, this paper uses eight years (2013–2020) of meteor radar measurements in the MLT region at Mohe station (53.5°N, 122.3°E), China, to retrieve high-temporal-resolution mesospheric wind data and further evaluate the temporal variation of GW kinetic energy. As the detected meteor trails exceed 6, the wind velocity is recalculated using the least square algorithm method, significantly increasing the temporal resolution of wind from 1 h up to 5 min. This resolution is sufficiently high for the investigation of GW kinetic energy, which exhibits a high spatial-temporal variability. For instance, it is enhanced in the winter season during the period of 0200–1400 UT and in the spring season during the period of 0800–1300 UT. The similarity between the climatological characteristics of GWs in MLT and the seasonal variation of GW total energy in the troposphere, determined from high-resolution radiosondes near to Mohe station, suggests that the meteorology in the lower atmosphere could be an important source of GWs in the MLT region.

1. Introduction

Gravity waves (GWs) play a remarkable role in defining the large-scale circulation and thermal structure of the mesosphere-lower thermosphere (MLT) region, and their spatial-temporal variabilities are largely determined by global-scale or synoptic-scale meteorology in the troposphere and stratosphere [1]. Although the instantaneous motion fields at MLT heights are dominated by planetary waves and atmospheric tides for their large horizontal and vertical amplitudes, the largest systemic effects on the MLT region mainly result from small-scale GWs that can transport energy and momentum from the sources in the lower atmosphere to the MLT region [2,3,4].

A variety of techniques have examined the characteristics of GWs in the MLT over a wide range of locations, including ground-based instruments, such as meteor radar [5,6,7], medium-frequency radar [8,9], lidar and airglow imager [10,11,12,13,14,15,16], and instruments on board satellites, such as Thermosphere, Ionosphere, Mesosphere Energetics Dynamics/Sounding of the Atmosphere using Broadband Emission Radiometry (TIMED/SABER) [17,18,19]. However, their working principles and temporal–spatial detection capabilities are somewhat different. By detecting the disturbance in the intensity of OH airglow at nighttime, the airglow imager can identify high-frequency GWs with periods shorter than 1 h in the imaging range [12]. Lidar can measure neutral temperature and wind vertical profiles to further obtain the GWs, but it lacks information in the horizontal direction [10]. TIMED/SABER is capable of providing a global picture of the entire height region from the stratosphere to the MLT region [19], but the detection of specific regions is inefficient. The ability of meteor radar to measure GW momentum fluxes has been investigated [6]. Importantly, meteor radar is an economic instrument that can provide continuous observation of wind fields in the MLT.

The meteor radar has made considerable progress in understanding the large-scale dynamic processes of the MLT region. The meteor radar was first designed in the 1950s [20]. It has been extensively used in investigating neutral winds and wave motions in the height range of 80–100 km [21]. Based on a developed SkiYMET radar system, Hocking and Thayaparan [22] first utilized the least square fitting algorithm to obtain the horizontal hourly winds. Then, numerous research on long period oscillations with a period range from a few hours to a few days has been carried out with hourly meteor winds over the last decades, including atmospheric tides, planetary waves, and the nonlinear interactions between waves [23,24,25]. Hocking [26] showed that the meteor radar observations can obtain the GW momentum fluxes and wind variances with a time scale of 2–3 h by proposing a new method. The winds derived from the radar observations show the seasonal and geographical distributions of GW momentum fluxes and variances [6,27,28].

Investigating the characteristics of GWs is difficult for the meteor radar due to the limitation of the temporal resolution of the meteor winds. Some previous studies optimized the temporary resolution of meteor winds. To investigate the horizontal propagation characteristics of GWs and their specific propagating parameters, Yamamoto et al. [29] averaged the radial drift velocity of meteor echo over 10 min and converted it into horizontal wind velocity. Using a similar method, Suzuki et al. [30] used meteor winds with a 20 min resolution to observe the horizontal wind perturbations caused by GWs. Kumar et al. [31] developed an algorithm to obtain the 15 min resolution meteor wind, which agreed with the hourly wind. The high-resolution winds mentioned above were all estimated during the particular periods with high meteor counts, such as during the meteor shower periods of 0000–0400 UT and 1800–2300 UT.

The meteor radar operated at Mohe, China, has a good measurement capability in detecting meteor echoes [32,33]. Nearly 25,000 meteor counts can be detected per day over Mohe, and the hourly meteor counts of the annual average range from 500 to 1200. The meteor counts are sufficient for continuously estimating the wind with a higher temporal resolution. In this paper, the Mohe meteor radar data are utilized to obtain a 5 min high-resolution wind in the MLT region and then to calculate the GW kinetic energy ($\mathrm{Ek})$. The similarity between the climatological characteristics of GWs in MLT and the variability of GW total energy in the troposphere, which is determined from high-resolution radiosondes close to Mohe station, is discussed. The relationship between GW activity and medium-scale traveling ionospheric disturbances (MSTID) is also discussed based on the statistical results from the Ek distribution obtained from the eight-year meteor data from 2013 to 2020.

2. Data and Methodology

2.1. Meteor Data Description

The Institute of Geology and Geophysics of the Chinese Academy of Sciences (IGGCAS) established a meteor radar chain to record the continuous wind in the MLT region [32,33]. The IGGCAS meteor radar chain, consisting of Mohe, Beijing (BJ, 40.3$°\mathrm{N}$, 116.2$°\mathrm{E}$), Wuhan (WH, 30.5$°\mathrm{N}$, 114.6$°\mathrm{E}$), and Sanya (SY, 18.3$°\mathrm{N}$, 109.6$°\mathrm{E}$), was situated at the 120$°\mathrm{E}$ meridian from north to south [34]. The radar located at Mohe was manufactured by the Atmospheric Radar Systems of Australia, and it has a peak power of 7.5 kW and a duty cycle of 10% at a frequency of 38.9 MHz. The antenna array consists of six crossed dipoles. A two-element Yagi antenna serves as the transmitter, and the five Yagi antennas serve as the receivers. This system, utilizing the interferometric technique to determine the arrival angles and range of the meteor echoes, is almost identical to the Buckland Park meteor radar operating in Australia and the SKiYMET radar operating in Canada [5,35]. The Mohe meteor radar is able to operate for both daytime and nighttime, continuously, and is generally undisrupted by severe weather conditions.

In addition, the radiosonde measurements (50 m vertical resolution) during the years from 2016 to 2020 at Hailar station (49.25°N; 119.70°E; 650.0 m), near to the meteor radar station, were applied to investigate the variation in GW total energy in the troposphere. The measurements include air temperature, pressure, relative humidity using specially designed hygristors, and wind speed and direction by tracking the position of a drifting balloon using an L-band radar [36,37].

The results in Figure 1 serve as a suitable basis for selecting meteor data. First, the meteor echo candidates considered as being under dense ones in the following analysis are selected. The detailed rejection criteria are shown in Holdsworth et al. [5]. Figure 1a shows that the daily meteor counts are mostly over 20,000, and the maxima exceed 30,000. Day 266 of 2013 is one of the days with a large number of meteor echoes. The meteor peak height (${\mathrm{H}}_{\mathrm{peak}}$) represents the mean height of the echoes detected in a day. Figure 1b shows the variation of daily ${\mathrm{H}}_{\mathrm{peak}}$ over the whole year. The red fitting curve reveals that the variation of daily ${\mathrm{H}}_{\mathrm{peak}}$ follows a semi-annual variation, and two peaks of the ${\mathrm{H}}_{\mathrm{peak}}$appear in April and September. The results generally agree with the previous research [33]. Figure 1c schematically illustrates the vertical distribution of meteor echoes on day 266 of 2013. The height distribution peaks at above 90 km and extends from 70 km to 110 km. The red curve presents the fitting of the meteor altitude distribution with a normal distribution, which verifies the assumption that the height distribution of the meteor echoes follows a Gaussian distribution [21]. The fitting is centered at 89.73 ± 0.15 km and has a full width at half maximum of 6.82 km. Figure 1d shows that the echo number per hour on the particular day is clearly above the annual average, and a diurnal variation of daily meteor counts is visible [35,38].

2.2. Least Square Algorithm Method

After grouping a sufficient amount of meteor echo information, including echo radial drift velocities (${\mathrm{V}}_{\mathrm{r}}$) and angle of arrival (AOA) direction cosines ($\mathrm{l}$,$\mathrm{m}$,$\mathrm{n}$), into a specific height/time bin, a least square algorithm can be utilized to estimate the wind velocity ($\mathrm{u}$, $\mathrm{v}$, $\mathrm{w}$). The equation is described as follows:

$${\mathrm{V}}_{\mathrm{r}}=\mathrm{ul}+\mathrm{vm}+\mathrm{wn}$$

(1)

The details of this method were illustrated by Hocking and Thayaparan [22] and Holdsworth and Reid [38]. The present work fully considers the limitations and implicit assumptions of the previous studies on the algorithm to secure against suspect wind estimates. Specifically, it requires a minimum of six echoes in each height/time bin, considers the assumption that $\mathrm{w}=0$, and rejects meteor echoes whose absolute difference between the actual and projected radial wind velocities $\mathsf{\Delta }{\mathrm{v}}_{\mathrm{r}}$ exceeds 30 $\mathrm{m}/\mathrm{s}$.

This paper primarily focuses on meteor winds with a 3 km height resolution. The daily${\mathrm{H}}_{\mathrm{peak}}$is the center height of the specific height region.

2.3. Error $\sigma$ for Estimating the Meteor Wind

The error $\mathsf{\sigma }$ for estimating the meteor wind is estimated by the following equations:

$${\mathsf{\Sigma }}_1=\sqrt{\frac{{{\displaystyle \sum }}_{\mathrm{i}=1}^{\mathrm{N}}{\left(\mathrm{C}-\mathrm{A}\ast \mathrm{B}\right)}^2}{\mathrm{N}-\mathrm{t}}},$$

(2)

$$\mathrm{D}=\mathrm{inv}\left({\mathrm{A}}^{\mathrm{T}}\mathrm{A}\right);{\mathrm{D}}_{\mathrm{u}}=\mathrm{D}\left(1,1\right),$$

(3)

$$\mathsf{\Sigma }={\mathsf{\sigma }}_1\ast \sqrt{{\mathrm{D}}_{\mathrm{u}}},$$

(4)

where C and A represent the matrices of meteor echo radial drift velocities (${\mathrm{V}}_{\mathrm{r}}$) and AOA direction cosines ($\mathrm{l}$,$\mathrm{m}$), respectively. B represents the matrix of wind velocity ($\mathrm{u}$, $\mathrm{v}$). N is the number of echoes in each height/time bin and represents the number of equations in the window. t is the degree of freedom of the equations.

2.4. GW Kinetic Energy in the MLT Region

The wind deviations from a background state of the atmosphere can be utilized to analyze wave structures in the meteor winds [39,40]. The mean values of zonal and meridional wind are estimated using the polynomial fit method [41,42,43,44]. The polynomial fits contain a superposition of various gravity and tidal waves because averaging over a full cycle of the diurnal tide is not included in the estimated background state. Therefore, the method mentioned above is not suitable to separate the GWs and the tides.

Suppressing the unwanted wave structures is essential to separate the GWs from the tides and planetary waves. Cai et al. [10] employed the full diurnal cycle data (temperature, wind) in a short period (38 h), and used harmonic fitting to fit the 6, 8, 12, and 24 h tide to obtain a background. Baumgarten et al. [40] and Kopp et al. [45] applied a harmonic fitting with a superposition of diurnal, semidiurnal, and terdiurnal tidal components to the retrieved sounding deviations from the background. The resulting deviations were assumed due to tides because of their stable phases, and the GWs were generated with random phases. Then, removing these fitted deviations induced by the tides (retrieved from a monthly average) from the total deviations, the residuals should only contain the GW information when assuming constant tidal amplitudes during the month. However, the particular tidal modes were not constant over the averaged time period. Hence, this approach is not suitable to separate the GWs from tides because the 1 month averaging results in a complete loss of the tides with short-term variability.

In this paper, the GW extraction method proposed by Kopp et al. [45] and Baumgarten et al. [40] is employed. A Butterworth filter of fifth order as a high pass in the temporal domain is used. The filter properties are selected such that for temporal filtering, waves with a period larger than 8 h are removed. According to the study of Baumgarten et al. [40] and Kopp et al. [45], a cutoff period of 8 h is sufficient to suppress tides because the amplitudes of the tidal waves are relatively smaller than the GWs. The Coriolis frequency $\mathrm{f}$ over Mohe is $1.17\ast {10}^{-4}$ rad/s (0.4212 rad/h$; \mathrm{f}=2\ast \mathsf{\Omega }\ast \sin \mathsf{\phi }$, $\mathsf{\phi }$ is the latitude in radians, $\mathsf{\phi }=53.5°$ and $\mathsf{\Omega }$ is the rotation rate of the Earth, Ω = 7.2921 × ${10}^{-5}$ rad/s).

The averaged GW kinetic energy per unit mass ($\mathrm{Ek})$ is approximated as follows [36,41,42,43]:

$$\mathrm{Ek}=\frac{1}{2}\left[\overline{{{\mathrm{u}}^{\prime }}^2}+\overline{{{\mathrm{v}}^{\prime }}^2}\right],$$

(5)

where ${\mathrm{u}}^{\prime }$, ${\mathrm{v}}^{\prime }$ are the GW perturbation components of the zonal and meridional wind velocity, respectively.

2.5. GW Total Energy in the Troposphere

As described in our recent studies in Zhang et al. [36], the GW total energy in the troposphere is extracted based on the broad spectral method, which will be introduced in greater detail below. The total perturbations of wind field and temperature (${\mathrm{u}}^{\prime },{\mathrm{v}}^{\prime },{\mathrm{T}}^{\prime }$) are derived by removing the background states (${\mathrm{u}}_0$, ${\mathrm{v}}_0$, and ${\mathrm{T}}_0$) that are estimated by a second-order polynomial fit [41] from the total measurements. The total perturbations could consist of measurement noises, tidal waves, planetary waves, and GWs. By applying a band-pass filter with vertical wavelengths of 0.3–6.9 km, the perturbation induced by GWs can be filtered. Thereby, the average GW total energy per unit mass (energy density) can be expressed as:

$$\mathrm{Et}=\frac{1}{2}\left[\overline{{{\mathrm{u}}^{\prime }}^2}+\overline{{{\mathrm{v}}^{\prime }}^2}\right]+\frac{1}{2}\frac{{\mathrm{g}}^2\overline{{\widehat{{\mathrm{T}}^{\prime }}}^2}}{{\mathrm{N}}^2}$$

(6)

where $\mathrm{g}$ is the gravitational constant, ${\hat{\mathrm{T}}}^{\prime }={\mathrm{T}}^{\prime }/\overline{\mathrm{T}}$ the normalized perturbation temperature, and the overbar indicates an averaging over all study heights. The study height interval is selected as 2–8.9 km, following Wang and Geller [41].

3. Results

Figure 2 displays an example of the zonal winds estimated by the least squares with different temporal resolutions on 23 September 2013. The red line represents the meteor wind, and the black bar represents the error of the wind velocity. The detailed calculation of the error is shown in Section 2.3. The wind with 60 min resolution shown in Figure 2a is used to analyze the large-scale atmospheric oscillations in the MLT region, such as tidal or planetary waves. The winds maintain a similar variation trend with the increase in the time resolution. The error increases, which indicates that the high temporal resolution winds become more variable. The 5 min wind in Figure 2d reveals some wave fluctuation information. In the previous research, Yamamoto et al. [29] and Suzuki et al. [30] extracted GWs from high-resolution winds. As for the amplitude of neutral wind perturbations, they are possibly related to the estimation error of the neutral wind velocity. Figure 3 shows the variation of error of the zonal winds retrieved by the least square fit with meteor counts’ height/time bin under different temporal resolutions on 23 September 2013. The slashed black line is a regression line of the wind error as a function of the meteor counts. From the slashed black lines in Figure 3, the error of the wind velocity becomes smaller (or larger) as the meteor counts increase (or decrease).

Then, the meteor data observed in 2013 are selected to retrieve the zonal winds under different resolutions, and the results are displayed in Figure 4. As for the high-resolution meteor winds, Figure 4b−e preserves the primary wind structure shown in Figure 4a, and the structure is more definite. Taking Figure 4a as an example, Figure 4a shows that the wind field reveals the tidal variation. The seasonal variation of the tidal signature shows the maximum value during the fall months and winter months. The tidal signature is relatively weaker in the summer months and not obvious in the spring months. Additionally, the strength of the tidal signature is strong at 0300 UT and 1500 UT during the winter months, and 00 UT and 12 UT during the fall months, respectively. The variations of the tidal signature described here are similar to the previous results of Yu et al. [34]. Similarity in the seasonal variation is also found in the nearby sites, as studied by Jacobi [46]. The meteor wind with 5 min temporal resolution also shows the tidal variation mentioned above.

Figure 5 shows an example of 5 min zonal wind $\mathrm{u}$ and 5 min meridional wind $\mathrm{v}$ on day 266 of 2013. The solid lines are the retrieved winds, and the dotted lines are the polynomial fits that represent the background mean values. Section 2.4 shows the details. The dU (dV) is the difference between the retrieved zonal (meridional) wind and the mean values. Wave-like perturbations can be seen in dU and dV (Figure 5). The maximum amplitude of dU occurs at 1000 and 1930 UT, and has a value of$~20 \mathrm{m}/\mathrm{s}$. During the period from 1700 UT to 2400 UT, the dU shows atmospheric oscillations with an hourly period. The maximum amplitude of dV also occurs at 1000 and 1930 UT, and has a value of$~20 \mathrm{m}/\mathrm{s}$. The long-period changes suggest that the meteor radar records the tidal waves in the MLT region, which can be expected.

To extract the wave perturbations related to GWs, a high-pass filter with a cut-off frequency at 8 h is applied to remove the long-period signal, which is described in Section 2.4. Figure 6a,b shows the zonal (meridional) velocity perturbation caused by GWs on day 266 in 2013. Mostly, the value zonal component of GW perturbation varies steadily from −10 m/s to 10 m/s, while at 2000–2100 UT, the value zonal component of GW perturbation changes strongly, reaching 25 m/s. The variation above can also be observed in Figure 2e at 2000–2100 UT, when the wind velocity become more variable. The variation of the meridional component of GW perturbation (Figure 6b) shows two maxima at 0200 UT and 2000–2100 UT. The value of the meridional component of GW perturbation is smaller than that in Figure 6a.

Figure 7 displays the comparison between the error wind velocity (red dots) and absolute value of GW perturbation of 5 min natural wind (black dots) on day 266 in 2013. In general, both the error wind velocity and GW perturbation become smaller (or larger) as the meteor counts increase (or decrease). The value distribution of the absolute value of the GW perturbation (from 0 m/s to 25 m/s) is much wider than the error wind velocity (from 2 m/s to 10 m/s).

The GW $\mathrm{Ek}$ is calculated using Equation (5) in Section 2.4. Figure 8 shows the variations of the $\mathrm{Ek}$ strength as functions of universal time and season in the monthly and hourly bins. The Ek here is calculated using the eight-year meteor data from 2013 to 2020. Cai et al. [11] suggests that the value of GW potential energy, Ep, at ~90 km in the MLT region is about 50 J/kg. Based on the linear theory of GWs, the ratio $\mathrm{Ek}/\mathrm{Ep}$ ($\mathrm{Ep}$ stands for the GW potential energy) is a constant ranging from 5/3 to 2.0 [47]. This suggests the $\mathrm{Ek}$ ~90 km in the MLT region is about 80–100 J/kg. Therefore, the value of GW kinetic energy in Figure 8 is close to the result above. The results in Figure 8 suggest that the strength of GW $\mathrm{Ek}$ has clear dependences on season and universal time. The strong GW activity appears in spring (March, April, and May) and winter (December, January, and February). Specifically, the maximum value of Ek occurs during 0200–1400 UT in winter and 0800–1300 UT in spring. The observational results in Jia et al. [48] suggest that the GW variances at ~90 km over Mohe are generally positively strong in the summer months. The variation of zonal GW variances shows a maximum reaching 180 ${\mathrm{m}}^2/{\mathrm{s}}^2$ in May, and the meridional GW variances show a maximum reaching 180 ${\mathrm{m}}^2/{\mathrm{s}}^2$ in July, respectively.

The hourly (seasonal) variation of GW kinetic energy $\mathrm{Ek}$ calculated using the eight-year meteor data from 2013 to 2020 in the MLT region shows a maximum at 1100 UT (spring) (Figure 9). Figure 10a displays the GW kinetic energy $\mathrm{Ek}$ at ~90 km in the MLT region at 1200 UT. In each year of 2016–2020, The monthly variation of $\mathrm{Ek}$ here shows semiannual variation with two maxima in Dec−Jan and May. The variability of GW total energy, Et, in the troposphere estimated by the radiosonde measurements close to the Mohe meteor radar station clearly shows the strong GW activities in the spring months, especially in May (Figure 10b). The similarity between the mesospheric GW kinetic energy and GW total energy in troposphere is that both of these GW energies are enhanced in May.

The local Kelvin-Helmholtz instability could be the major generation source for GWs in the troposphere [36]. In addition, the jet stream in the upper troposphere could be a secondary source.

4. Discussion and Conclusions

Atmospheric GWs propagating from the surface into the MLT region are important due to their efficient vertical transport of energy and momentum from sources at lower altitudes into the stratosphere and the MLT [49]. However, the wave structures, especially the dynamical characteristics in the MLT region and its link with the lower atmosphere, still warrant further observational investigation. This paper analyzes the eight-year meteor data over Mohe from 2013 to 2020 to investigate the high-temporal-resolution mesospheric wind and the GW activity in the MLT region. In addition, the variability of GW activities in the troposphere near to the radar station is simultaneously analyzed.

Using least square fitting, the high-temporal-resolution mesospheric winds are calculated over Mohe station. The wind field with the temporal resolution ranging from 60 min to 5 min reveals the tidal signature and seasonal variation in the MLT region. The seasonal variation of the tidal signature shows the maximum value during the fall months and the winter months. Additionally, the strength of the tidal signature is strong at 0300 UT and 1500 UT during the winter months, and 0000 UT and 1200 UT during the fall months. The variations of the tidal signature are consistent with the previous results. The meteor wind with 5 min temporal resolution also shows the tidal variation mentioned above.

The GW kinetic energy, $\mathrm{Ek},$ in the MLT region contains clear seasonal and local time dependences. The seasonal variation of $\mathrm{Ek}$ shows a maximum in May (reaching 120 J/kg), and a second maximum in Dec−Jan. The hourly variation of $\mathrm{Ek}$ shows a maximum at 1100 UT. The maximum values of averaged Ek, composed by the eight-year data from 2013 to 2020, occur in winter during 0200–1400 UT and in spring during 0800–1300 UT.

The variability of GW total energy, Et, in the troposphere estimated by the radiosonde measurements close to the Mohe meteor radar station clearly shows the strong GW activities in the spring months, especially in May. Simultaneously, the variation of GW kinetic energy at ~90 km also shows a maximum in the spring months, especially in May. The similarity between the GW kinetic energy in the MLT region and the GW total energy in the troposphere is that both GW energies are enhanced in May. The similarity between the climatological characteristics of GWs in MLT and the variability of the GW total energy in the troposphere suggest that the meteorology in the lower atmosphere could be an important source of GWs in the MLT region. The local Kelvin-Helmholtz instability could be the major generation source for GWs in the troposphere [36].

Moreover, the upward propagation GWs in the MLT could be closely associated with the occurrence rates of middle-scale traveling ionospheric disturbance (MSTID) in the ionosphere. MSTIDs are considered the manifestation of atmospheric GWs of lower atmospheric origins in the ionosphere [50,51,52,53]. As is known, the upward propagating GWs can largely increase the amplitude to maintain kinetic energy, which is due to the decrease in the neutral density in the thermosphere [52]. Eventually, the upward propagating GWs break up and generate secondary GWs to accelerate Perkins instability, causing the generation of MSTIDs [52,53]. In addition, the coupling effects between the sporadic E layer (Es) and Perkins instability can be responsible for generating the MSTIDs, especially in the nighttime ionosphere. However, Cheng et al. [53] reported that the AGWs may have a significant contribution to MSTIDs in both daytime and nighttime in summer. The results in Cheng et al. [53] show a high occurrence rate of the southward (equatorward) MSTIDs during 0000–1300 UT in winter in Taiwan (14.5$°\mathrm{N}$ geomagnetic latitude; 32.5$°$ inclination). Southward MSTIDs mainly occur around 1300–1900 UT during summer. Otsuka et al. [52] suggest that the MSTID occurrence rate reached a peak at 1300–1400 UT in nighttime during summer in Japan (133.0$°$−137.0$°\mathrm{E}$, 33.0$°$−37.0$°\mathrm{N}$). In nighttime during the winter, the MSTID occurrence rate peaks at 1400 and ~1000 UT under low and high solar activity conditions, respectively. Huang et al. [54] suggest that the MSTIDs had a high occurrence rate during 1100–1900 UT at night in May at midlatitudes over central China. Daytime MSTID frequently occurred during 0000–0900 UT in winter. In the current paper, strong GW activity in the MLT region occurs during 0200–1400 UT in winter and 0800–1300 UT in spring. The temporal and monthly distributions of the strength of the GW kinetic energy $\mathrm{Ek}$ agree with the occurrence rates of MSTID, while the similarity between the climatological characteristics of GWs in MLT and MSTID still needs further investigation.

The neutral wind estimated from the meteor data using the traditional least square algorithm is efficient, and the mathematical process can still be improved. The new algorithm in deriving the high-resolution meteor wind and its application in the GW observation will be investigated in the future.