Analysis of the Mechanism Underlying Tibetan Plateau Vortex Frequency Difference between Strong and Weak MJO Periods

1.   Introduction

There are many unique weather systems over the Tibetan Plateau (TP). Among them, the TP vortex (TPV, also known as Qinghai–Xizang Plateau vortex, or QXP-vortex for short) is representative of the low air pressure weather systems. The generation, development, and eastward propagation of the TPV are usually accompanied by precipitation, strong winds, thunderstorms, and other weather processes, resulting in a series of extreme weather events over the TP and the Sichuan–Chongqing regions downstream of the plateau. Some TPVs, which move from the plateau to the east and develop strongly, even affect a wide range of eastern region of China. Thus, research on the mechanisms underlying the occurrence, development and eastward propagation of TPVs, and their predictability, is highly important.

Madden and Julian (1971, 1972) first reported the phenomenon of atmospheric low-frequency oscillation in the tropics. Since then, Chinese and international scholars have continued to research in this field, making many important discoveries. In recent years, research has increasingly focused on the weather systems and processes in the midlatitudes affected by this tropical atmospheric low-frequency oscillation. For instance, Pan et al. (2010) found that the MJO plays a significant role in the modulation of typhoons in the western Pacific, and the frequency of typhoons generated in strong MJO is twice that in weak MJO periods. Moreover, the number of typhoon generated in phases 5 and 6 of strong MJO periods is much more than in phases 2 and 3. Ma et al. (2011) analyzed the underlying mechanism of the snow disaster affected by the MJO in southern China in 2008, arguing that the MJO convective activity was highly active in the equatorial regions of the Indian Ocean, meaning that warm and humid airflows in the lower troposphere transported large amounts of water vapor to southern China, and under the continuing impact of the cold air, the disastrous weather occurred. Lyu et al. (2012) explored the mechanism responsible for the Yunnan extreme drought in 2009–2010, as influenced by MJO activity. They suggested that continuous positive anomalies of the MJO over the tropics in the central and eastern Indian Ocean weakened the vertical circulation anomalies of the South Asian monsoon, leading to a reduction in precipitation in the fall of 2009, which intensified the drought in Yunnan. International scholars have also carried out a large amount of work on the weather and climate of the midlatitude regions affected by the MJO (e.g., Jones, 2000; Bond and Vecchi, 2003; Donald et al., 2006; Wheeler et al., 2009).

There has also been a considerable amount of research conducted specifically on the atmospheric low-frequency oscillation over the TP. For example, Sun and Chen (1988) found that atmospheric low-frequency waves over the TP have an oscillation cycle within 30–40 days, and that the north–south propagation phenomenon of cyclones and anticyclones generated over the TP at the 500-hPa isobaric surface is significant. Sun and Chen (1994) revealed the mass characteristics of TPVs and explored the relationship between these mass characteristics and the atmospheric low-frequency oscillation over the plateau, pointing out that the mass period of TPVs relates to the conversion of phases of atmospheric low-frequency oscillation, atmospheric vertical structure, and the strength of high-frequency disturbance. Xu and Zhu (2000) analyzed the structural features of the atmospheric low-frequency oscillation over the TP in 1998, and pointed out that atmospheric low-frequency oscillation over the plateau has an equivalent barotropic structure, and low-frequency precipitation mainly occurs in the convergence zone of low-frequency cyclones. Xu and Zhu (2002) then studied the sources and sinks of the atmospheric low-frequency oscillation in the summer of 1998, and pointed out that the characteristics of the sources and sinks have significant differences in different regions of the plateau. Wang et al. (2011) analyzed the characteristics of spring atmospheric low-frequency oscillation over the TP between strong and weak summer monsoon years in the South China Sea, and pointed out that the northward propagation of low-frequency oscillation formed in the northern part of the plateau in strong monsoon years is significant; however, the atmospheric low-frequency oscillation possesses in-situ oscillation characteristic in weak monsoon years. Zhang et al. (2010) further studied the relationship between the mass characteristics of TPVs and the atmospheric low-frequency oscillation over the plateau, and pointed out that TPVs mainly emerge in positive phases of the 10–30-day atmospheric low-frequency oscillation and the negative phases of convective disturbances. Furthermore, the mass period of TPVs mainly overlaps with the cyclonic phase of relative vorticity of the 10–30-day oscillation.

In spite of the above-mentioned findings, research on the influence of tropical atmospheric low-frequency oscillation on the TPV is limited. In particular, what is the impact of tropical atmospheric low-frequency oscillation on the generation of TPVs? This question forms the main focus of the present paper, in which we examine the impact of tropical atmospheric low-frequency oscillation on the occurrence frequency of TPVs to reveal the mechanisms involved in the modulation of the TPV by tropical atmospheric low-frequency oscillation, and provide new ideas for short-term TPV forecasting.

2.   Data and methodology

2.1   Data

This paper uses the reanalysis data acquired from the NCEP–NCAR (Kanamitsu et al., 2002); the real-time multivariate MJO (RMM) index from the Australian Bureau of Meteorology; daily satellite-observed outgoing longwave radiation (OLR) data from the NOAA; and statistical TPV data from the Chengdu Institute of Plateau Meteorology of the China Meteorological Administration.

Combined fields of near-equatorially averaged 850-hPa zonal wind, 200-hPa zonal wind, and OLR data (each averaged over the latitudes within 15°S–15°N), are used for EOF analysis. The resulting time series pairs are referred to as the real-time multivariate MJO series 1 (RMM1) and 2 (RMM2), with an amplitude of $\sqrt {{\rm{RMM}}1{}^2 + {\rm{RMM}}{2^2}}$ (e.g., Wheeler et al., 2004, 2009).

2.2   Methodology

The statistical method employed in this study of the TPV is to use the RMM index to examine the strength of the MJO. Two different metrics are determined by using amplitudes of 1 and 0.8, where a value greater than 1 or 0.8 corresponds to an active MJO period and a value of less than 1 or 0.8 corresponds to a suppressed MJO period. In this paper, we count the frequency of TPVs in active and suppressed periods of the MJO separately, in addition to the frequency distribution in different phases of active MJO periods.

Wavelet analysis is also used, to determine the atmospheric oscillation cycle of the various meteorological elements over the TP, and to obtain the wavelet power spectrum and wavelet variance by analyzing time series of these meteorological elements. The wavelet power spectrum is the norm of complex wavelet coefficients, while wavelet variance is the integral of all wavelet power spectra in the time domain (Torrence and Compo, 1998). We obtain the oscillation cycle from the latter, and the intensity distribution and tendency change of different periodicities of various meteorological elements’ time series from the distribution of the former (Ji et al., 1999). Then, the Lanczos band-pass filter is used for oscillation cycle filtering. The TP region is defined as 27.5°–40°N, 75°–105°E.

3.   Results

3.1   Atmospheric oscillation cycle analysis over the plateau

Figure 1 illustrates the wavelet analysis of the normalized zonal wind component at 200 and 500 hPa over the TP from 1998 to 2010. The results show an obvious 30–60-day cycle in the time series of the zonal wind component at 200 hPa, and the cycle is quite strong in winter and spring but relatively weak in autumn and summer (Figs. 1a₁, a₂). A similar result can be seen in Figs. 1b₁, b₂ for the 500-hPa zonal wind field, as well as for the OLR field and geopotential height field at 500 hPa (figures omitted) (Zhao et al., 2016).

Download

Larger image

PowerPoint slide

Fig  1.  Wavelet analysis of daily normalized variables over the TP from 1998 to 2010: (a₁, b₁) the wavelet standard power spectrum and (a₂, b₂) wavelet variance for the (a₁, a₂) 200 and (b₁, b₂) 500-hPa zonal wind field. Dotted areas exceed the 95% confidence level (adopted from Zhao et al., 2016).

3.2   Statistical characteristics of the plateau vortex

The onset period refers to the time when a TPV generates, while duration periods are the periods during which the TPV continuously develops. Strong periods are periods when the MJO amplitude is greater than the two chosen thresholds of 1 or 0.8, and weak periods are periods when the MJO amplitude is less than the thresholds of 1 or 0.8.

Table 1 shows the statistical results from the comparison of the frequency of TPVs between strong and weak MJO periods, and the distribution of the frequency of TPVs in different MJO phases of strong MJO periods. Taking the amplitude of 1 as the standard, the frequency ratio between strong and weak MJO periods is 314:193; whereas, taking the amplitude of 0.8 as the standard, the frequency ratio is 381:126. Under both MJO measurement standards, the number of TPVs generated during strong MJO periods is obviously larger than that during weak MJO periods. Upon examination of the TPV frequency distribution in each strong MJO phase during the outlet period, under both MJO measurement standards, it can be seen that there is a significant increase (above normal) in TPV frequency in phases 1 and 2, while there is a significant reduced frequency (below normal) in phases 3 and 7. In addition, examination of the TPV frequency distribution in each strong MJO phase throughout the duration period shows that there is also a significantly increased frequency (above normal) in TPV frequency in phases 1 and 2. It should be noted that the different measurement standards used for the MJO correspond to different lower frequency phases of the MJO. The measurement standard of amplitude 1 corresponds to phases 3 and 5, whereas the amplitude 0.8 corresponds to phases 3 and 7 (Zhao et al., 2016). The above conclusion is significantly different from what has been found for typhoons. The number of typhoons generated in phases 5 and 6 of the MJO is significantly greater than that in phases 2 and 3 (e.g., Pan et al., 2010; Li et al., 2012). This indicates that the generation mechanism of TPVs may be different from that of typhoons, but the influence of the MJO on TPVs and typhoons is equally important.

Table  1.  Statistical analysis of TPV frequency in different MJO phases and the ratio between strong and weak MJO periods (from Zhao et al., 2016)

Download as CSV

	MJO amplitude	TPV frequency in different phases of the MJO								Strong/weak amplitude = 1.0	Strong/weak amplitude = 0.8
		1	2	3	4	5	6	7	8
Onset	> 0	99	82	48	54	61	57	41	67	314:193	381:126
	> 1.0	71	49	25	32	34	31	27	46
	> 0.8	84	63	31	40	40	40	33	51
Duration	> 0	154	143	73	86	79	86	80	94	491:301	588:204
	> 1.0	104	89	43	51	46	55	44	60
	> 0.8	126	109	50	63	50	64	61	70

 | Show Table

DownLoad: CSV

But why does the frequency distribution of TPVs differ so significantly between strong and weak MJO periods (the ratio being as pronounced as 381:126)? What mechanisms are involved? The next section discusses these aspects.

4.   Mechanistic analysis

4.1   Atmospheric circulation

Figure 2 shows the atmospheric circulation filtered in 30–60-day bands. In the composites of low-frequency atmospheric circulation at 500 hPa in Fig. 2a, a low-frequency, low-pressure system and cyclonic wind anomalies are apparent over the TP, which contributes to the high occurrence of TPVs in strong MJO periods. The Korean Peninsula is in control of a low-frequency, low-pressure system and cyclonic wind anomalies too, while Lake Baikal, the Iranian Plateau, the Arabian Sea, and the South China Sea are controlled by a low-frequency, high-pressure anticyclone system, and the south branch trough develops intensely. Under the combined effect of the south branch trough and the low-frequency high-pressure anticyclone, the southwest of China, southern China, and the area south of the Yangtze River are under the control of strong southwesterly flow. This flow is highly favorable for the transportation of water vapor to the TP and South China, which can provide good water vapor conditions for TPV generation. Figure 2c presents composites of the low-frequency atmospheric circulation field at 100 hPa in the strong MJO periods, revealing a low-frequency, high-pressure anticyclone circulation over the TP corresponding to low-frequency cyclonic circulation at low altitude, and the TP being controlled by a deep low-pressure cyclonic system. Figure 2b shows composites of the low-frequency atmospheric circulation field at 500 hPa in the weak MJO periods, illustrating a low-frequency, high-pressure anticyclonic circulation over the TP and, at the same time, the eastern part of China being controlled by northerly flow in the front of this system, with insufficient water vapor being unfavorable for the generation of a TPV. Figure 2d presents the composite results of the atmospheric circulation field at 100 hPa, showing that the TP is under the influence of a low-frequency, low-pressure cyclonic system, which corresponds to the low-frequency, high-pressure anticyclonic system at low altitude. These results indicate that the atmospheric circulation conditions over the TP are conducive to the generation of TPVs in strong MJO periods, and vice versa in weak MJO periods.

Download

Larger image

PowerPoint slide

Fig  2.  Composites of the low-frequency atmospheric circulation field: (a, c) strong MJO and (b, d) weak MJO. (a, b) 500-hPa low-frequency field (m s^–1; gpm) and (c, d) 100-hPa low-frequency field (m s^–1; gpm).

Figure 3 shows the latitude–height profile of the zonal wind difference between strong and weak MJO periods. There is a notable positive anomaly zone over the TP, which denotes strong westerly flow over the plateau. The high-value region is in the upper part of the southern plateau; plus, there is a high-value center at 200 hPa, which indicates that westerly flow is significantly stronger in the south of the plateau in strong MJO periods. At lower levels, in the range of 20°–40°N, $- \displaystyle\frac{{\partial u}}{{\partial y}} > 0$, corresponding to cyclonic shear, which could be conducive to the generation of the cyclonic vorticity ($\zeta = \displaystyle\frac{{\partial v}}{{\partial x}} - \displaystyle\frac{{\partial u}}{{\partial y}} > 0$).

Download

Larger image

PowerPoint slide

Fig  3.  Latitude–height cross-section of the zonal wind (m s^–1; zonal mean between 75° and 105°E) difference between strong and weak MJO periods.

Figure 4 displays the longitude–height cross-section of the meridional wind difference between strong and weak MJO periods. In Fig. 4, the upper level over the east of the plateau is dominated by positive anomalies, and there is a positive-value center at 300 hPa, meaning that southerly flow is particularly strong there. However, the upper level over the west of the plateau below 250 hPa is dominated by negative anomalies, and a negative center is found at 400 hPa; whereas, the upper levels above 250 hPa are dominated by positive anomalies and, within the range of 75°–105°E at low levels, $\displaystyle\frac{{\partial v}}{{\partial x}} > 0$, corresponding to cyclonic shear, which is also conducive to the cyclonic vorticity.

Download

Larger image

PowerPoint slide

Fig  4.  Longitude–height cross-section of the meridional wind (m s^–1; meridional mean between 27.5° and 40°N) difference between strong and weak MJO periods.

The comparative analyses above indicate that, during strong (weak) MJO periods, southerly (northerly) flow prevails over the east of the plateau, while northerly (southerly) flow prevails over the west. The northern plateau is dominated by easterly (westerly) flow, while the southern plateau is dominated by westerly (easterly) flow. Therefore, the situation is conducive (inconducive) to the generation of cyclonic vorticity and cyclonic circulation at low altitude over the plateau, and thus beneficial for the generation of TPVs.

Figure 5 shows the vertical velocity filtered through 30–60-day bands at 300 hPa in strong and weak MJO periods. As we can see, in strong MJO periods, the TP, southern China, and Indochina are primarily dominated by updrafts, which provides good dynamic conditions for TPVs. However, in weak MJO periods, the TP, southern China, and Indochina are mainly dominated by downdrafts, which is not beneficial for the generation of convective activity, and thus the generation of TPVs.

Download

Larger image

PowerPoint slide

Fig  5.  Composites of atmospheric vertical velocity at 300 hPa (10³ Pa s^–1): (a) strong MJO and (b) weak MJO. Shaded areas exceed the 95% confidence level.

Figure 6 illustrates the vertical shear difference of the zonal wind between strong and weak MJO periods. It shows that the TP is in a significant negative area, that is, the vertical shear of westerly wind is smaller over the TP, and thus water vapor and heat assemble over the plateau region, convections form easily, and thus TPVs generate favorably. This result is consistent with the findings of Qian et al. (1984).

Download

Larger image

PowerPoint slide

Fig  6.  Difference in vertical wind shear (m s^–1; 1000 hPa minus 500 hPa) between strong and weak MJO periods. Shaded areas exceed the 95% confidence level.

4.2   Atmospheric energy and temperature conditions

The OLR reflects the temperature of the top of clouds, so it is often used to represent the strength of convective activity. Here, we compare the low-frequency OLR difference distribution between strong and weak MJO periods to obtain their convective differences (Fig. 7). Figure 7 indicates a convection belt extending to the northwest from the equatorial region over 110°–120°E, and it is connected to the east–west convection belt south of China. Combined with the 500-hPa low-frequency circulation characteristics shown in Fig. 2a, in strong MJO periods, the Indian Ocean, Indochina Peninsula, and southern TP are controlled by a low-frequency trough from the southern TP to the Indian Ocean, and southerly flow between a low-frequency South China Sea anticyclone and low-frequency TP anticyclone. From the OLR distribution characteristics, high energy flow from tropical regions arrives northward in the TP and then migrates to southern China along the convection belt. As the eastern TP, Southwest China, southern China, and the area south of the Yangtze River are all located in this high energy flow belt, the convection zone can provide plenty of energy for the generation and development of TPVs.

Download

Larger image

PowerPoint slide

Fig  7.  Difference in the low-frequency OLR field (W s^–1) between strong and weak MJO periods. Shaded areas exceed the 95% confidence level.

Figure 8 shows the difference in atmospheric low-frequency kinetic energy between strong and weak MJO periods. It demonstrates that the vast majority of the TP is in a positive area, indicating that greater kinetic energy is generated in strong MJO periods. Both atmospheric low-frequency kinetic energy and low-frequency eddy kinetic energy are higher, which provides adequate rotational kinetic energy for the generation of TPVs.

Download

Larger image

PowerPoint slide

Fig  8.  Difference in the low-frequency kinetic energy field (m² s^–2) between strong and weak MJO periods. Shaded areas exceed the 95% confidence level.

Figure 9 presents the longitude–height profiles of the atmospheric temperature difference between strong and weak MJO periods. As we can see, the TP is primarily dominated by negative anomalies below 200 hPa, and by positive anomalies above 200 hPa. The atmospheric temperature in the east of the plateau is higher than that in the west. Similar to Fig. 9, Fig. 10 presents the latitude–height cross-section of the atmospheric temperature difference between strong and weak MJO periods, in which negative anomalies are situated below 200 hPa and positive anomalies above 200 hPa. The air temperature over the south of the plateau is higher than that over the north. These results indicate that, in strong MJO periods, low temperature at the low atmospheric levels over the plateau corresponds to strong cold advection, while high temperature at the high atmospheric levels corresponds to strong warm advection. The low temperatures of the low atmospheric levels over the west of the plateau and the north of the plateau correspond to strong cold advection. Such a vertical air temperature distribution structure is conducive to baroclinic instability and stratification instability, and thus beneficial for convective activity.

Download

Larger image

PowerPoint slide

Fig  9.  Longitude–height cross-section of the air temperature (°C; meridional mean between 27.5° and 40°N) difference between strong and weak MJO periods.

Download

Larger image

PowerPoint slide

Fig  10.  Latitude–height cross-section of the air temperature (°C; zonal mean between 75° and 105°E) difference between strong and weak MJO periods.

4.3   Water vapor conditions

A warm and humid atmosphere contains plenty of latent heat, and plays an important role in the generation and development of TPVs. However, how the moisture conditions differ between strong and weak MJO periods? Figure 11 shows the latitude–altitude profiles of the relative humidity difference between strong and weak MJO periods, and Fig. 12 shows the longitude–altitude profiles of the relative humidity difference. As can be seen from these two figures, the TP is in a significant negative area, in both cases, which indicates that the atmosphere over the plateau is relatively drier in strong MJO periods and, indirectly, that the role sensible heat plays in the generation of TPVs is more important than that of latent heat. This is similar to the conclusions of Ding et al. (1994) and Chen et al. (1996).

Download

Larger image

PowerPoint slide

Fig  11.  Latitude–height cross-section of the specific humidity (g kg^–1; zonal mean between 75° and 105°E) difference between strong and weak MJO periods.

Download

Larger image

PowerPoint slide

Fig  12.  Longitude–height cross-section of the specific humidity (g kg^–1; meridional mean between 27.5° and 40°N) difference between strong and weak MJO periods.

However, a number of other studies have shown that water vapor and latent heat are particularly important for the development of TPVs (Chen et al., 1996, Yu and He, 2001). But what about the distribution of water vapor over the TP and its surroundings, and how does water vapor transport to the plateau? Figure 13 shows the difference of the 30–60-day filtered atmospheric specific humidity between strong and weak MJO periods at 600 hPa. In Fig. 13, most of the plateau regions are in a negative area of the low-frequency field of specific humidity, showing that the atmosphere over the plateau is relatively drier, while there is a vast positive-value zone north of the equator from 80° to 110°E. From Fig. 2a, there is a low-frequency trough in the central Indian Ocean and a low-frequency, high-pressure anticyclone in the South China Sea, with low-frequency southerly flow between them, which indicates that the region mentioned above has a relatively wetter atmosphere and abundant latent heat. Thus, under the delivery of the southerly winds, high humidity and energy air flow can be transported northward to southeastern plateau, and energy is replenished for the further development of TPVs.

Download

Larger image

PowerPoint slide

Fig  13.  Difference in the low-frequency specific humidity field (10^–5 kg kg^–1) between strong and weak MJO periods at 600 hPa. Shaded areas exceed the 95% confidence level.

Figure 14 shows the difference in the low-frequency water vapor field between strong and weak MJO periods. As we can see, under the combined influence of the low-frequency south branch trough and the low-frequency, high-pressure anticyclone over the South China Sea, water vapor is continuously transported to the plateau and eastern China, with the convergence of water vapor in the west of the plateau at the same time.

Download

Larger image

PowerPoint slide

Fig  14.  Difference in the low-frequency water vapor field between strong and weak MJO periods at 600 hPa (shaded areas exceed the 95% confidence level): (a) moisture transport fluxes (2 × 10^–4 kg m^–1 hPa^–1 s^–1) and (b) moisture transport flux divergence (10^–10 kg m^–2 hPa^–1 s^–1).

In brief, water vapor over the TP is not abundant in strong MJO periods, and the water vapor needed for the generation of TPVs mainly comes from the Bay of Bengal, which is transported by the south branch trough and the low-frequency, high-pressure anticyclone system over the South China Sea. As the water vapor content of the plateau atmosphere itself is not high, a TPV in its formative period is mainly dependent on sensible heat. However, the delivery of water vapor is much stronger, which provides latent heat for the eastward propagation and development of the TPV. Therefore, the water vapor conditions are a key factor contributing to the difference in the frequency of TPVs between strong and weak MJO periods.

5.   Summary and discussion

This paper studies the possible effects of the tropical MJO on TPVs, revealing that the MJO plays an important modulating role in the formative process of TPVs over the TP. The conclusions are drawn as follows.

(1) TPVs mainly generate in strong MJO periods. The number of TPVs generated in the strong MJO periods, within the current study’s timeframe is 381, while the number generated in the weak MJO periods is only 126. Thus, the frequency ratio between strong and weak MJO periods is approximately 3:1, demonstrating that the modulation of TPV generation by the MJO is significant.

(2) During strong (weak) MJO periods, the TP is in control of a low-frequency, low-pressure cyclone (high-pressure anticyclone) system, and thus the atmospheric circulation conditions over the plateau are conducive (inconducive) to the generation of TPVs.

(3) During strong (weak) MJO periods, southerly (northerly) flow is prevalent over the east of the plateau, but northerly (southerly) flow is prevalent over the west. Over the north part of the plateau, easterly (westerly) flow predominates, while westerly (easterly) prevails over the south. Thus, the situation is conducive (inconducive) to the formation of cyclonic circulation (i.e., a TPV) at low altitude over the plateau.

(4) In strong MJO periods, water vapor is relatively less abundant over most of the plateau, which is not beneficial for the generation of TPVs. However, moisture transported by the south branch trough and low-frequency, high-pressure anticyclone system from the Bay of Bengal, is very important for the development of TPVs.

As the strength of the MJO changes continuously during its eastward propagation, the intensity of the tropical convection and vertical circulation structure of the tropical atmosphere also change accordingly. The change in the distribution of the baroclinic instability, available potential energy, and eddy available potential energy under the interaction between the midlatitude atmosphere and the low-latitude atmosphere will affect atmospheric circulation, temperature, the structure of energy and moisture transport over the TP and its surrounding areas. Alternation between favorable and unfavorable conditions for the generation of TPVs occurs, thus resulting in a significant frequency difference of TPVs between strong and weak MJO periods.

This paper demonstrates the possible mechanisms involved in the frequency difference of TPVs between strong and weak MJO periods from a statistical perspective. However, these mechanisms also need to be proved through numerical simulation and kinetic analysis. In addition, what role does the MJO play in whether or not a TPV shifts out of the TP region? This is worthy of further investigation, too.