Gravity Wave Activity and Stratosphere-Troposphere Exchange During Typhoon Molave (2020)

The troposphere and stratosphere have different chemical composition characteristics and thermal and dynamic mechanisms. The weather in the troposphere changes rapidly and violently, while the atmosphere in the stratosphere is slow and stable. Studies have shown multiple scales and forms of coupled processes between the troposphere and stratosphere (Brewer ^[1]; Fritts ^[2]; Wirth ^[3]). It includes stratospheric sudden warming (SSW), planetary wave activity, stratospheric-tropospheric exchange (STE), and so on. Among them, STE is one of the most important processes. STE can change atmospheric compositions by controlling both the natural and anthropogenic emissions of chemicals. These changes have a non-negligible impact on both short-term air quality changes and long-term climate change (Tian et al. ^[4]). Gravity waves (GWs), as atmospheric waves, are influenced by atmospheric laminarity, and these waves, characterized by meso- and small-scale perturbations, have a discernible impact on atmospheric circulation and structure. Recently, some research related to atmospheric waves has also been conducted on Mars (Zhang et al. ^[5]), heat waves (He et al. ^[6]), and stratospheric disturbance (He et al. ^[7]). Moreover, Chen et al. ^[8] analyze the wind field in the near space where there are multi-scale atmospheric fluctuations. Quasi-biennial oscillation (QBO) is the main model of the equatorial stratosphere, and can regulate the dynamic circulation and distribution of trace gases in the stratosphere (Wang et al. ^[9]). Some research has been carried out on GWs and QBO (He et al. ^[10]). The energy released by the break of tropospheric GWs during the uploading process could induce the stratospheric QBO phenomenon. Furthermore, if the gravity wave activity weakens, it will also weaken the Brewer-Dobson (B-D) circulation, resulting in the Antarctic ozone hole. With further research, the “mechanical oscillator effect” (Fovell et al. ^[11]), “obstacle effect” (Clark et al. ^[12]), and “thermal effect” (Salby and Garcia ^[13]) are regarded as the three main mechanisms responsible for inducing GWs related to convective activity.

Tropical cyclones (TCs) are strong convective weather scale systems, including various types based on their locations and intensity, such as hurricanes, typhoons, tropical storms, and cyclonic storms. Studies have shown a close relationship between typhoons and GWs. The latent heat release, topographic uplift, and the change of atmospheric stratification in typhoons can stimulate GWs. Typhoon-induced GWs (TGWs) are generated along the typhoon track in the troposphere and propagate in the horizontal and vertical directions. Moreover, the TGWs have a great impact on not only the dynamics, but also the distribution of chemical constituents of the Earth’s middle atmosphere (Cairo et al. ^[14]; Dhaka et al. ^[15]; Sato ^[16]). Typhoons in the Northwestern Pacific can induce a series of GWs with horizontal wavelengths up to several hundred kilometers long and lasting for many hours (Kim et al. ^[17]). In the early studies of typhoon-induced GWs (TGWs), many people measured the presence of TGWs by using instruments such as aircraft, and middle and upper atmosphere (MU) radar (Chane-Ming et al. ^[18]; Dhaka et al.^[15]; Pfister et al. ^[19]). Later, mesoscale models were gradually adopted to simulate TGWs (Kim et al. ^[20]; Kim et al. ^[21]; Kuester et al. ^[22]).

Tropical cyclones play an important role in STE (Emanuel ^[23]; Merrill ^[24]), which is demonstrated by many previous studies on typhoon-induced STE processes. As one of the most important trace components in the atmosphere, Ozone can be used as a tracer to reflect physical processes such as QBO, El Niño-Southern Oscillation (Chang et al. ^[25]; Li et al. ^[26]), and it is easy to be depleted by reacting with other substances (Feng et al. ^[27]). STE is of great significance to the change of O₃ content between the stratosphere and troposphere, so studying the variation of O₃ content in the stratosphere and troposphere can characterize the role of typhoons on STE. Bellevue et al. ^[28] found that air intrusions from the stratosphere to the troposphere during TC events can increase the content of O₃ in the upper troposphere. Fadnavis et al. ^[29] found that ozone in the upper troposphere over the Bay of Bengal was strongly enhanced during TC events. A typical TC can extend the vertical circulation of strong convection within an area radius of 100–2000 km to the upper troposphere and lower stratosphere (UTLS) region at an altitude of 10–18 km (Emanuel ^[30]) and also produce violent disturbance to the structure and chemical composition of the UTLS. Using the mesosphere-stratosphere (MST) radar, Das et al. ^[31] observed the invasion of the stratosphere into the troposphere during the passage of a cyclone. Through the In-Service Aircraft for a Global Observing System (IAGOS), Roux et al. ^[32] found that tropical cyclones were involved in the atmosphere transport from the stratosphere to the troposphere, and they speculated that this transport was a conventional feature of typhoons. Rossow and Pearl ^[33] counted 22 years of tropical convective events across the stratosphere, and the results show that the penetration events of hurricanes and typhoons generally last more than one day. Moreover, TCs also transport a large amount of water vapor, energy, and momentum to the UTLS region (Ray and Rosenlof ^[34]), and they are conducive to transporting water vapor-rich and ozone-deficient air from the troposphere to the lower stratosphere and moving dry and ozone-rich air to the upper troposphere, resulting in the occurrence of STE (Romps and Kuang ^[35]). STE includes stratosphere-troposphere transport and troposphere-to-stratosphere transport. Zhan and Wang ^[36] studied the typhoons in the Northwestern Pacific in 2003 and 2004 and found that the effect of strong typhoons on tropospheric to stratospheric transport was stronger than that of weak typhoons.

According to the latest atmospheric reanalysis data released by the European Centre for Medium-Range Weather Forecasts (ECMWF), EAR5 can provide a large number of grid products with high spatial and temporal resolution, which is good for analysis. However, EAR5 is not sufficient to understand the characteristics of GWs generated by typhoons. Predecessors such as Wang et al. ^[37] have achieved good results in using the WRF model to simulate mesoscale gravity wave signals. For example, Chen et al. ^[38] used the WRF model to simulate the process of TGWs on stratosphere and achieved good results. Moreover, the WRF model has been a useful tool in the study of TGWs, providing various high spatiotemporal resolution atmospheric variables. Therefore, we used the ERA5 reanalysis data and the WRF model to analyze the influence of GWs on STE induced by a typhoon and to provide the basis for understanding the specific process of material exchange in STE.

At present, some progress has been made on typhoon-induced STE and TGWs, but further study on the mechanism and specific processes of STE caused by typhoon-induced gravity wave breaking is required. GW’s breaking process is an important driving source of the small-scale turbulent mixing process in the middle and upper atmosphere (Miyazaki et al. ^[39]). The turbulent mixing can cause STE and the redistribution of atmospheric components in the middle and upper atmosphere. Deep convection associated with typhoons is considered to be an important source of GWs. Adiabatic forcing and overshoot in deep convective systems can excite a wide range of GWs on large temporal and spatial scales. The South China Sea is a region characterized by significant deep convection (Wu et al. ^[40]) and a remarkable level of atmospheric activity. Therefore, this study focused on Typhoon Molave, which occurred in the South China Sea in 2020 between October 24 and October 29, 2020, to analyze TGWs and the STE process caused by the breaking of TGWs.

In this study, we used the ERA5 reanalysis data, which are ECMWF’s fifth-generation atmospheric reanalysis dataset of global climate with a horizontal resolution of 0.25°×0.25°. The reanalysis dataset combines model data with observations from across the world. We incorporated a wide range of variables across 37 levels spanning from 1000 to 1 hPa (Available at: https://cds.climate.copernicus.eu/cdsapp#!/dataset/reanalysis-era5-pressure-levels?tab=overviewhttps://cds.climate.copernicus.eu/cdsapp#!/dataset/reanalysis-era5-pressure-levels?tab=overview). These variables include vertical velocity, potential vorticity, temperature, ozone mass mixing ratio, and wind (U, V, and W) in three dimensions of the atmosphere from October 27 to October 28, encompassing the transition of Typhoon Molave from strong to weak. The spatial dimension spanned from 0° to 30°N and from 90°E to 135°E, with the motion range of the whole process considered.

In order to observe GWs, we used the hyperspectral Atmospheric Infrared Sounder (AIRS). It is a continuously operating cross-track scanning detector that is carried by the Aqua satellite launched in 2002. Moreover, we also used typhoon data from the China Meteorological Administration (CMA). The data included the longitude, latitude, wind speed, pressure, and other parameters of the typhoon during its activity period.

In this study, we first used the ERA5 reanalysis data to obtain gravity wave signal and ozone distribution and then employed the WRF model to simulate the gravity wave process. Finally, we selected the ozone distribution anomaly area to analyze the changes in potential vorticity vertical flux after being affected by gravity wave breaking. The results will help us understand the STE process influenced by TGWs.

The WRF-ARW (WRF) V.4.3 mesoscale model selected in this study is a new generation of prediction model and assimilation system jointly developed by the National Center for Atmospheric Research (NCAR), the National Centers for Environmental Prediction (NCEP) and the Center for Analysis and Prediction of Storms (CAPS) at the University of Oklahoma. Because the mesoscale WRF model has high accuracy and includes a variety of Earth system processes, it can reflect the gravity wave process characterized by the meteorological variables simulated by the WRF model, during the occurrence of Typhoon Molave with a higher spatial and temporal resolution. The simulated duration of 121 hours was from 00:00 UTC on October 24 to 00:00 UTC on October 29, 2020. The simulation areas encompass the outer domain D01 (67°N–25°S, 173°W–30°E), the middle domain D02 (4°–23°N, 114°–134°E), and the inner domain D03 (7°–20°N, 116°–131°E) using the Mercator projection with a horizontal resolution of 27 km, 9 km, 3 km, respectively. With the moving nesting method employed, the nested grids of D02 and D03 moved dynamically with the typhoon center. To effectively analyze the temporal and spatial evolution of GWs and diagnose the stratospheric-tropospheric material and energy exchange caused by gravity wave processes, we set 28 layers in the vertical direction of the model and set the top of the model at 50 hPa. In the numerical simulation test, the physical schemes were as follows: the WRF single–moment 6–class microphysical scheme (Hong and Lim ^[41]), the Rapid Radiative Transfer Model longwave and shortwave radiation scheme (Iacono et al. ^[42]), the Yonsei University scheme boundary layer scheme (Hong et al. ^[43]), the Unified Noah Land Surface Model for land surface processes scheme (Mukul Tewari et al. ^[44]), and the MM5 Similarity Scheme for land surface processes (Zhang and Anthes ^[45]). In order to ensure the stability of the integral and prevent the reflection of GWs, we also used a damping layer scheme on the upper boundary, and the height of the damping layer was 5 km. The initial time of the model simulation was set at 00:00 UTC on October 24, 2020. The initial field of the model was generated by using the data provided by NCEP, with a resolution of 1°. During the simulation, results were output every 30 minutes with a time step of 54 s, and a total of 121 hours of the simulation was completed by 00:00 UTC on October 29, 2020.

Typhoon Molave was a strong typhoon that occurred in the South China Sea. It was generated on the eastern side of the Philippines, developed into a strong typhoon in the South China Sea, and finally landed on the coast of Vietnam. The typhoon had a large cloud circulation range and a wide influence range, causing a total of at least 115 deaths, 186 injuries, and 742 million dollars in economic losses in the Philippines, Vietnam, and the coastal areas of South China. Fig. 1 shows the characteristics of the center track and the intensity of the typhoon moving and changing with time. To intuitively reflect the changing process of Typhoon Molave, the maximum wind speed level standard set by the World Meteorological Organization was adopted, i.e., the central maximum wind speed V_max<18 m s^–1 as the tropical depression stage, 18 m s^–1≤V_max <33 m s^–1 as the tropical storm stage, and 33 m s^–1≤V_max <67 m s^–1 as the typhoon stage. Typhoon Molave first appeared as a tropical depression on the east side of the Philippines at 00:00 UTC on October 24 and was upgraded to a tropical storm at 00:00 UTC on October 25. Moreover, the typhoon continued to move towards the South China Sea and turned into a typhoon at 18:00 UTC on October 25, with a central pressure of 970 hPa and a central maximum wind speed of 37.5 m s^–1. Subsequently, it reached its strongest state at 06:00 UTC on October 27, with the center maximum wind speed V_max of 47.8 m s^–1, and finally landed on the coast of Vietnam at 06:00 UTC on October 28. As shown in Fig. 1, comparing the typhoon center position and typhoon intensity of the WRF model with the actual CMA typhoon data at each time point, we found that the simulation results matched well with the actual time point of intensity change and maximum wind speed (12:00 UTC on October 27). Moreover, the simulated and observed typhoon tracks were basically the same. However, at the time point of transition from a tropical storm to a typhoon (18:00 UTC on October 25), the WRF model had an error of about 12 hours from the actual time, and the simulated intensity was generally lower than the actual value. In conclusion, the WRF model successfully simulated the typhoon track and the change in typhoon intensity. It was necessary to confirm whether the numerical simulation was true by comparing it with the observed results, so the result shows that the WRF model can simulate the basic characteristics of Typhoon Molave, and can reflect the basic situation of its meteorological variables, which can be used to analyze the GWs and STE.

Figure 1.  (a) Track, (b) minimum sea level pressure, and (c) maximum wind speed of Typhoon Molave in observation (solid line) and simulation (dotted line).

To extract the gravity wave signal, we analyzed the horizontal distribution of vertical velocity to obtain the distribution and structure of the GWs generated by the typhoon (Fig. 2 and Fig. 3). In Fig. 2a, the typhoon produced strong disturbances and significant arc-shaped parallel waves in the coastal area of the Indo-China Peninsula. Fig. 2 and Fig. 3 show that the fluctuation area was mainly located on the east side of the typhoon center. In the fluctuation area, an obvious alternating distribution of positive and negative vertical velocity was observed, which was the gravity wave signal. Particularly, Figs. 3b–3d show a concentric ring structure around the typhoon center, which had strong vertical vibration energy and can provide strong power to generate GWs. According to the vertical velocity plane diagrams (Fig. 2 and Fig. 3), we preliminarily conclude that in the horizontal direction, the GWs were mainly concentrated on the east side of the typhoon center.

Figure 2.  Vertical velocity field (units: cm s^–1) at 100 hPa provided by the ERA5 reanalysis dataset from 00:00 UTC on October 27 to 18:00 UTC on October 28.

Figure 3.  Vertical velocity field (units: cm s^–1) at 400 hPa provided by the ERA5 reanalysis dataset from 00:00 UTC on October 27 to 18:00 UTC on October 28.

Wang et al. ^[46] summarized three methods for extracting gravity waves, including setting the average temperature within seven days as the background temperature, and then using the temperature disturbance map obtained by subtracting the background temperature from the original temperature to represent the GWs. In order to better show the GWs, this study adopted this method (Fig. 4 and Fig. 5). According to Fig. 4 and Fig. 5, there were also obvious positive and negative alternations at the center of the typhoon, and the GWs were mainly distributed on the east side of the typhoon center.

Figure 4.  Temperature disturbance field (units: K) at 100 hPa provided by the ERA5 reanalysis dataset from 00:00 UTC on October 27 to 18:00 UTC on October 28.

Figure 5.  The temperature disturbance field (units: K) at 400 hPa provided by the ERA5 reanalysis dataset from 00:00 UTC on October 27 to 18:00 UTC on October 28.

Gravity waves can be observed by hyperspectral Atmospheric Infrared Sounder (AIRS) (Chang et al. ^[47]; Hoffmann and Alexander ^[48]). According to Fig. 6 and Fig. 7, the existence of GWs can be directly observed.

Figure 6.  Brightness temperature perturbation (units: K) for ARIS at 05:48 UTC on October 27.

Figure 7.  Distribution of gravity waves and deep convection from ARIS at 05:48 UTC on October 27.

Figures 8a–8d show a strong vertical updraft at 150–700 hPa, which contained a lot of energy and can provide power for the rise of GWs (Hong et al. ^[49]). Similar to the vertical velocity plane diagrams, the east side of the typhoon center was the main distribution area of the GWs and exhibited the typical gravity wave signal featured by alternating updraft and downdraft. Compared with the plane vertical velocity (Fig. 2 and Fig. 3), the vertical profile can reflect GWs change with height during the upload process. In this gravity wave event, the distribution of GWs at 400 hPa was more symmetrical than that at 100 hPa, implying that the west side typhoon center fluctuated less than the east side at 100 hPa, while the distribution at 400 hPa was uniform.

Figure 8.  Vertical velocity zonal field (units: cm s^–1) of ERA5 reanalysis data along the typhoon center latitude from 00:00 UTC on October 27 to 18:00 UTC on October 28. The yellow triangle is the location of CMA typhoon center from 00:00 UTC on October 27 to 18:00 UTC on October 28; the black dotted line is the wind speed at each time point of ERA5 reanalysis data (units: m s^–1; the positive value is the westerly wind, and the negative value is the east wind) along the typhoon center latitude from 00:00 UTC on October 27 to 18:00 UTC on October 28.

Beres et al. ^[50] suggest that the background wind field, namely the zonal wind field, in the same direction as the phase velocity of the gravity wave can reduce its momentum flux along the direction of the background wind field and hinder the GWs from propagating upward. Moreover, when the background wind field was opposite to the propagation direction, the background wind field can enhance the inherent phase velocity of the GWs and cause the GWs to propagate upward. Therefore, the cause of the significant fluctuation on the east side of the typhoon center, and the semi-circular structure of the GWs were all related to the consumption of some east–west side fluctuations by the background wind field during the vertical rise of GWs. However, how the specific background wind field affected the distribution of GWs needs further analysis of the zonal winds. On the west side of the typhoon center, the GWs were generated at 500 hPa at 00:00 UTC on October 27 and gradually uploaded to about 150 hPa. Finally, they broke at about 150 hPa at 18:00 UTC on October 27. According to the zonal wind field in Fig. 8, the reason why GWs broke here was mainly due to the significant easterly jet.

Figure 9 shows the distribution of buoyancy frequency N² (N² = dlnθ/dz along the longitude-height profile of the typhoon center. In general, N² can be used to represent atmospheric stratification stability. The particles in the atmosphere will move vertically after being disturbed, return to the equilibrium position under the combined actions of gravity and buoyancy, and oscillate due to inertia. The frequency of oscillation is called buoyancy frequency. When N²>0, the atmospheric stratification is stable; otherwise, it is unstable. Zhao et al. ^[51] selected 1 potential vorticity unit (PVU, 1 PVU = 10⁻⁶ K m² s⁻¹ kg⁻¹) as the tropopause height when studying the impact of stratospheric intrusion on the ozone in the lower troposphere and achieved good simulation results. Therefore, we also set 1 PVU as the tropopause height in our study. Combined with the fluctuation of the previous vertical velocity, the atmosphere probably produced GWs due to the disturbance induced by the typhoon. Furthermore, at the upper end of the purple box (102°–116°E, 120–250 hPa) in Figs. 9a–9e, the buoyancy frequency values decreased as the typhoon moved, and the buoyancy frequency at the center was abnormally low. The smaller the buoyancy frequency, the weaker the atmospheric stability. The purple box is just near the top of the troposphere, and there is a low center of buoyancy frequency in the box, which could result in the gravity wave breaking. When the atmosphere is unstable, the GWs will break and generate turbulent mixing. Potential vorticity (PV) is a physical quantity that combines thermal and dynamic effects, which can reflect a certain physical process, so it is often used to analyze various phenomena and events (Zhang et al. ^[52]). The aircraft observation of Cho et al. ^[53] has shown that the GWs generated by convective activity will be accompanied by the folding of the tropopause during its breaking process, and the folding of the tropopause is often accompanied by the downward transmission of positive PV. In Figs. 9a–9e, there is a significant positive PV downward transmission in the red box area (107°–117°E, 72–117 hPa), which corresponds to the low buoyancy frequency in the purple box area at the lower end. This not only indicates the existence of tropopause folding but also proves the occurrence of gravity wave breaking. In the following section, we will analyze the STE process caused by gravity wave breaking in more detail.

Figure 9.  Buoyant frequency zonal field (units: s⁻²) of ERA5 reanalysis data along the typhoon center latitude from 00:00 UTC on October 27 to 18:00 UTC on October 28. The yellow triangle is the location of CMA typhoon center from 00:00 UTC on October 27 to 18:00 UTC on October 28, the black solid lines are the potential vorticity at each time point of ERA5 reanalysis data (PVU, 1 PVU = 10^–6 K m² s^–1 kg^–1) along the typhoon center latitude from 00:00 UTC on October 27 to 18:00 UTC on October 28, and the pink solid line is the tropopause (1 PVU). Red box: 107°–117°E, 72–117 hPa; purple box: 102°–116°E, 120–250 hPa.

The distribution of gravity waves observed by temperature perturbation field, vertical velocity field, and AIRS satellite data is closely related to the typhoon structure, so it can be explained that gravity waves are generated by this typhoon.

To intuitively show the influence of gravity wave breaking on STE, this study used ozone and potential vorticity as tracers of changes at the UTLS during the gravity wave process generated by Typhoon Molave (Fig. 10). Figs. 10a and 10b show that the ozone mixing ratios were striped around the typhoon center, gradually decreasing outwards until symmetrically distributed along the typhoon center. Ozone anomalies were obvious in the white box region (100°–108°E, 80–150 hPa) of Figs. 10a–10f. The low-value area of the ozone mixing ratio in the white box shows an upward convex shape, which continued to expand over time, indicating a decreasing trend of the ozone mixing ratio in this area. Meanwhile, the corresponding potential vorticity field showed fluctuations. When the contours of the potential vorticity field bulge upward, the potential vorticity value was smaller than that of the surrounding area, indicating poor static stability of the atmosphere in the white box, which was conducive to the exchange of material and energy between the upper and lower parts of the region. When the isoline of the potential vorticity field bulged downward, the tropopause folded, which was accompanied by gravity wave breaking. In the white box of Fig. 10, there is a significant fluctuation in the potential vorticity field. Therefore, the increasing ozone mixing ratio was probably caused by the gravity wave breaking and the weakening of static stability, which had a good response to the downward intrusion of the stratospheric atmosphere into the troposphere. The change in ozone concentration was affected by chemical, advection, and vertical mixing. In the white box, the circulation was dominated by the east wind, and the advection at the same latitude would not have a significant impact on the ozone concentration. The gravity wave process lasted for a short time, and the chemical effect within a few hours was not significant. Therefore, this further shows that the main reason for the distribution of ozone anomalies was gravity waves.

Figure 10.  Ozone mass mixing ratio zonal field (units: kg kg⁻¹) and wind field (U and V; units: m s^–1; 70–150 hPa) of ERA5 reanalysis data along the typhoon center latitude from 00:00 UTC on October 27 to 18:00 UTC on October 28. The yellow triangle is the location of the CMA typhoon center from 00:00 UTC on October 27 to 18:00 UTC on October 28, the black solid lines are the potential vorticity at each time point of ERA5 reanalysis data (PVU, 1 PVU = 10^–6 K m² s^–1 kg^–1) along the typhoon center latitude from 00:00 UTC on October 27 to 18:00 UTC on October 28, and the pink solid line is the tropopause (1 PVU). White box: 100°–108°E, 80–150 hPa; black box: 120°–130°E, 100–600 hPa.

Generally, the potential vorticity of air in the stratosphere is higher than that in the troposphere. According to Fig. 10, the potential vorticity contours were mostly distributed in the center of the typhoon, very similar to the strip distribution of ozone concentration mixing ratios. Moreover, high PVU contours in the troposphere over the eight periods show that the stratospheric atmosphere intruded the troposphere. Therefore, the potential vorticity distribution and the ozone concentration anomaly in Fig. 10 can show a good response to the STE effect, which is the increase of ozone concentration at the top of the troposphere and the appearance of stratospheric air near the center of the typhoon.

Similar to the white box area in Fig. 10, the abnormal increase of ozone also appeared in the black box area (120°–130°E, 100–600 hPa) in Fig. 10. The zone mixing ratio exhibited a horizontal strip distribution with a higher concentration in this area than that in the surrounding area, and the strip gradually moved to the bottom atmosphere as the typhoon moved westward. According to the vertical distribution of vertical velocity (Fig. 8), a large number of downdrafts were distributed at 150–550 hPa on the east side of the typhoon center. These downdrafts transported the ozone-rich atmosphere in the lower stratosphere downward to the troposphere, increasing the average ozone mass mixing ratio in the black box region. In particular, in the black box area (120°–130°E, 100–600 hPa) of Fig. 10, the lack of obvious positive PV downward transmission indicates that the ozone change in this area was not directly related to the material exchange induced by gravity wave breaking.

In order to diagnose the effect of GWs on matter and energy transport, it is necessary to investigate the change of momentum flux caused by turbulence after gravity wave breaking (Kawatani et al. ^[54]; Shapiro ^[55]). The breaking of GWs can cause turbulent activity in the UTLS region, resulting in the vertical mixing of matter and momentum. Therefore, the study of the change of momentum flux can better recognize the influence and effect of GWs caused by typhoons on the STE at UTLS. PVW, the average potential vorticity vertical flux can be used to quantitatively determine the turbulent mixing caused by gravity wave breaking (Shapiro ^[55]). The white box (Fig. 10, 100°–108°E, 80–150 hPa) was under the action of GWs and showed obvious ozone distribution anomalies. Accordingly, the analysis of the average potential vorticity vertical flux PVW and the average ozone mass mixing ratio in this region can reflect the STE process caused by gravity wave breaking.

Affected by gravity wave breaking, the ozone mass change in the white box area had a good response to turbulent mixing. Fig. 11 shows the PVW change (solid line) of the typhoon center (100°–108°E, 80–150 hPa) and the change of the average ozone mass mixing ratio (dotted line) in the same range. The rapid increase of PVW at 18: 00 UTC on October 27 indicates the strong vertical mixing, which also means the STE occurred here after the GWs breaking. Moreover, the average ozone mass mixing ratio increased with the rising PVW from 18:00 UTC on October 27 to 00:00 UTC on October 28. The changes of average ozone mass mixing ratio and PVW had a good consistency, indicating that the turbulent mixing in the UTLS region was the result of the high ozone concentration atmosphere in the stratosphere intruding into the troposphere and thus increasing the average ozone mixing ratio.

Figure 11.  Changes in the average potential vorticity vertical flux PVW (solid line, units: PVU m s^–1) and the average ozone mass mixing ratio (dotted line, units: kg kg^–1) of ERA5 reanalysis data with time along the typhoon center at 100°–108°E and 80–150 hPa from 00:00 UTC on October 27 to 18:00 UTC on October 28.

Figure 11 also shows the influence of GWs on STE well, but the time of gravity wave breaking lagged behind ozone change. It is because the ERA5 reanalysis data filters out some small-scale and mesoscale disturbances, making some GWs missing. The resolution of domain D03 in the WRF model is about 3 km, while ERA5 is about 27.75 km (0.25°), indicating that the WRF can simulate the meteorological elements with a higher resolution. Consequently, it is necessary to use the mesoscale WRF model to conduct a more in-depth analysis of the GWs and their impacts on STE.

Chang et al. ^[56] have used the WRF model to analyze GWs and STE, proving that the simulation results can be used to analyze GWs and STE. In this paper, we adopted the WRF model to simulate the meteorological elements of the event and used them to characterize the fluctuation process of GWs and others, rather than directly simulate GWs and processes of gravity wave breaking and turbulence mixing.

Background wind field plays an important role in gravity wave propagation. Convection can produce a large gravity wave spectrum, and these waves can propagate eastward and westward. However, the background wind field can determine which waves to filter out and prevent them from continuing to propagate upward (Beres et al. ^[50]). Based on the simulation result of the WRF model, Figs. 12a and 12b show the temperature field and wind field (U and V) conditions of 400 hPa and 100 hPa. At 400 hPa (Fig. 12a), the zonal wind was both westward and eastward, while there was only westward zonal wind at 100 hPa (Fig. 12b). The easterly zonal wind at 100 hPa and 400 hPa had an influence on the westward propagation of gravity wave trains, making the gravity wave propagate upward on the east side of the typhoon center. Likewise, the existence of westward zonal wind at 400 hPa also caused the upward propagation of the GWs to the west side of the typhoon center in this altitude layer. These lead to the change of gravity wave structure between different altitude layers. It can be concluded from the zonal wind fields of the WRF model that the GWs at 100 hPa were mainly distributed on the east side of the typhoon center, while at 400 hPa, they were more evenly distributed on the east and west sides. Moreover, it shows that the distribution of GWs reflected by the WRF model was in good agreement with the results of the ERA5 reanalysis data.

Figure 12.  Wind fields (units: m s^–1) and temperature field (units: K) (a) on 400 hPa and (b) 100 hPa simulated by the WRF model at 06:00 UTC on October 27. The diamonds are the typhoon track provided by CMA from 00:00 UTC on October 24 to 06:00 UTC on October 29.

Figures 13 and 14 are the vertical velocity fields on 100 hPa and 400 hPa simulated by the WRF model from 00:00 UTC on October 27 to 18:00 UTC on October 28, with a time interval of 6 hours. The yellow diamond is the typhoon track of CMA data from 00:00 UTC on October 24 to 06:00 UTC on October 29. Compared with those in Figs. 2–3, the gravity wave signals in Figs. 13–14 are more obvious, and the gravity wave distribution of the semi-annular structure is dense. This is because the WRF model data has higher resolution and can simulate the GWs with smaller wavelengths, which makes up for the deficiency of ERA5 reanalysis data in the spatial and temporal distribution of GWs. Furthermore, the gravity wave distribution areas of WRF model data and ERA5 reanalysis data are basically similar, indicating that the WRF model has simulated the basic characteristics of gravity wave signals.

Figure 13.  Vertical velocity field (units: cm s^–1) on 100 hPa simulated by the WRF model from 00:00 UTC on October 27 to 18:00 UTC on October 28. The yellow diamonds are the typhoon track provided by CMA from 00:00 UTC on October 24 to 06:00 UTC on October 29.

Figure 14.  Vertical velocity field (units: cm s^–1) on 400 hPa simulated by the WRF model from 00:00 UTC on October 27 to 18:00 UTC on October 28. The yellow diamonds are the typhoon track provided by CMA from 00:00 UTC on October 24 to 06:00 UTC on October 29.

Similarly, comparing the vertical velocity zonal field of the WRF model from 00:00 UTC on October 27 to 18:00 UTC on October 28 (Fig. 15) with ERA5 reanalysis data (Fig. 8), we found that they were consistent in the vertical direction of gravity wave upload height, which was mainly located below the tropopause, partially through the tropopause, and the gravity wave phases of the two images were also consistent. However, the WRF model further refined the wave train distribution of GWs in the vertical direction, expanding the gravity wave signal and making the gravity wave phenomenon more obvious. As shown in Fig. 15, from 00:00 UTC on October 28 to 06:00 UTC on October 28, the gravity wave signal reappeared on the west side of the typhoon center, which was missing from the ERA5 reanalysis data. Besides, Figs. 15b and 15c, within the green box (105°–120°E, 70–150 hPa), clearly show the folding phenomenon of the tropopause that reveals a wide range of gravity wave breaking in multiple regions. The potential vorticity of WRF model data (Fig. 15) fluctuated more significantly than that of ERA5 (Fig. 8). In order to better reflect this process, we made an average temperature anomaly distribution map (Fig. 16) from 00:00 UTC on October 27 to 18:00 UTC on October 28. All these more significantly show the turbulent mixing motion after gravity wave breaking, which provided favorable conditions for the occurrence of STE.

Figure 15.  Vertical velocity zonal field (units: cm s^–1) stimulated by the WRF model along the typhoon center latitude from 00:00 UTC on October 27 to 18:00 UTC on October 28. The yellow triangle is the location of CMA typhoon center from 00:00 UTC on October 27 to 18:00 UTC on October 28, the black solid lines are the potential vorticity at each time point of WRF model data (PVU, 1 PVU = 10^–6 K m² s^–1 kg^–1) along the typhoon center latitude from 00:00 UTC on October 27 to 18:00 UTC on October 28, and the pink solid line is the tropopause (1 PVU). Green box: 105°–120°E, 70–150 hPa.

Figure 16.  Vertical temperature zonal field (units: K) stimulated by the WRF model along the typhoon center latitude from 00:00 UTC on October 27 to 18:00 UTC on October 28. The yellow triangle is the location of CMA typhoon center from 00:00 UTC on October 27 to 18:00 UTC on October 28, the black solid line is the potential vorticity at each time point of WRF model data (PVU, 1 PVU = 10^–6 K m² s^–1 kg^–1) along the typhoon center latitude from 00:00 UTC on October 27 to 18:00 UTC on October 28, and the pink solid line is the tropopause (1 PVU). Green box: 105°–120°E, 70–150 hPa.

In order to analyze the spatial spectral structure of TGWs, we used the fast Fourier transform (FFT) method. The horizontal wavelength can be derived by $\widehat{\omega}$=ω–ku_TC–lv_TC, while the vertical wavelength can be derived by $\widehat{\omega}_i^2$=N²(k² + l²) + f²m²/k² + l² + m² = (ω−ku−lv)² (Chen et al.^[57]). They included zonal wave number (k), meridional wave number (l), frequency (ω), natural frequency ($\widehat{\omega}$), buoyant frequency (N), Coriolis parameter (f), the zonal and meridional average wind speed from 00:00 UTC on October 27 to 18:00 UTC on October 28 (u and v), and the U and V component of vertical velocity from 00:00 UTC on October 27 to 18:00 UTC on October 28 (u_TC and v_TC). In conclusion, the horizontal wavelength was about 800–1000 km in the center of GWs, and the vertical wavelength was around 5–8 km. Moreover, the wave number was equal to 2π divided by the wavelength. Therefore, vertical wave number was 6.9 × 10⁻⁶ m and horizontal wave number was 1.04 × 10⁻³ m.

The WRF model has basically simulated the STE process under the action of gravity wave breaking. Fig. 17 shows the variation of the average potential vorticity vertical flux PVW of WRF model data and the average ozone mass mixing ratio of ERA5 reanalysis data along the typhoon center at 100°–108°E and 80–150 hPa. Similar to Fig. 11, PVW (Fig. 17, solid line) can also explain the change in the average ozone mass mixing ratio (Fig. 17, dotted line). According to Fig. 17, the changes in ozone content caused by gravity wave breaking can be reflected from 06:00 UTC on October 27 to 00:00 UTC on October 28. The PVW increased rapidly at 06:00 UTC on October 27, which means the occurrence of gravity wave breaking. Therefore, the average ozone mass mixing ratio increased at 12:00 UTC on October 27 with a delay of 6 h, reflecting the relationship between STE and gravity wave breaking.

Figure 17.  Changes in the average potential vorticity vertical flux (PVW) (units: PVU m s^–1, solid line) of WRF model data and average ozone mass mixing ratio (units: kg kg^–1, dotted line) of ERA5 reanalysis data with time along the typhoon center at 100°–108°E and 80–150 hPa from 00:00 UTC on October 27 to 18:00 UTC on October 28.

This study combined the ERA5 reanalysis data and the WRF mesoscale model to investigate the gravity wave events caused by Typhoon Molave in 2020 and the stratosphere-troposphere exchange. The following results are obtained:

The GWs generated by Typhoon Molave were mainly distributed on the east side of the typhoon center (10°–20°N, 110°–120°E) below the tropopause, with a stronger gravity wave signal on the east side than that on the west side. The GWs were mainly distributed in the horizontal direction on the east side of the typhoon center, and the gravity wave fluctuation signal on the east side of the typhoon center was more intense than that on the west side at 06:00 UTC on October 27. In the vertical direction, because of the different zonal wind fields, the GWs were obviously inclined eastward with the increase of height from 06:00 UTC on October 27 to 06:00 UTC on October 28.

The turbulent mixing produced by gravity wave showed a good response to STE. In the UTLS on the west side of the typhoon center (100°–108°E, 80–150 hPa), PVW rose from −1.36 × 10⁻³ PVU m s^–1 to 2.79 × 10⁻³ PVU m s^–1 from 12:00 UTC on October 27 to 00:00 UTC on October 28, while the ozone mass mixing ratio increasing from 7.78 × 10⁻⁸ kg kg^–1 to 9.56 × 10⁻⁸ kg kg^–1 from 12:00 UTC on 27 to 00:00 UTC on 28. Moreover, because the resolution of ERA5 data was low, the time of gravity wave breaking was later than that of ozone change.

The results of the WRF simulation of Typhoon Molave show that this parameterization scheme can simulate the uneven distribution of GWs at the center of a typhoon, the characteristics of different signal intensities, and the turbulent mixing in the UTLS region under the action of gravity wave. Compared with ERA5 reanalysis data, in the horizontal direction, the distribution of GWs in the east side of the typhoon center of the WRF model was more intensive. In the vertical direction, the positive and negative alternation of vertical velocity on the east side of the typhoon center in the WRF model was more obvious, indicating more GWs and stronger signals on the east side of the typhoon center. Moreover, the change of potential vorticity field in the green box area (105°–120°E, 70–150 hPa) of the WRF model was more intense and frequent, reflecting the strong turbulent mixing movement caused by gravity wave breaking. Among them, the great increase in PVW at 06:00 UTC on October 27 explains the gravity wave breaking. The ozone change after 12:00 UTC on October 27 also exhibited a good response to the STE that the stratospheric atmosphere invaded the troposphere.

In this study, we analyzed the GWs induced by Typhoon Molave (2020) and obtained the spatial distribution and development of GWs. The research on ozone indicates that the turbulent mixing caused by the breaking of typhoon-induced GWs had a strong connection to the occurrence of STE. Although the gravity wave breaking would cause the STE, the scale of the turbulent mixing was small, so more evidence is needed to show its direct impact on air mixing. In 2016, Das et al. ^[31] found that there were inertial gravity waves generated by the wind system related to tropical cyclones in the UTLS area, and because of the downward airflow in the UTLS area, stratospheric air can further invade the middle and lower troposphere during tropical cyclone Nilam. Although both Nilam and Molave successfully landed and similar conclusions were reached, there is no direct evidence that the surface-landed typhoon is related to the STE process caused by TGWs. What kind of typhoon can produce such an impact process and what is the feedback effect of this process on typhoons in turn need further research. It is undeniable that the process mentioned in this paper needs to be further studied, but the present study is also innovative and can provide a reference for the study of TGWs and STE. In the future, we will continue to analyze the relationship between TGWs and STE, with the aim of refining our understanding of the underlying mechanisms.