Southwesterly Water Vapor Transport Induced by Tropical Cyclones over the Bay of Bengal during the South Asian Monsoon Transition Period

Beiyao LIU and Ying LI

1 State Key Laboratory of Severe Weather, Chinese Academy of Meteorological Sciences, China Meteorological Administration, Beijing 100081

2 University of Chinese Academy of Sciences, Beijing 100049

ABSTRACT

Key words: tropical cyclones, Bay of Bengal, South Asian summer monsoon, southwesterly water vapor transport,anomalous characteristics

1. Introduction

The Bay of Bengal (BoB) is the area in which the activity of the South Asian monsoon (SAM) occurs and is an essential source of water vapor transport (WVT) for precipitation in China. Tropical cyclones (TCs) are active in this area. They can transport water vapor from the Indian Ocean northeastwards to the Tibetan Plateau,Southwest China, and even the middle and lower reaches of the Yangtze River under certain atmospheric circulation conditions, resulting in wide ranging precipitation in China. Research has indicated that BoB TCs are one of the critical systems influencing the dryness and wetness of the Yangtze River basin (Zhou and Li, 2005).

The monthly variation in BoB TC activity demonstrates a unique bimodal pattern, with peaks in May and October–November, respectively (Gray, 1968; Singh et al., 2000). This is mainly attributable to the smaller vertical shear of the ambient wind during the summer monsoon transition period, which is favorable for TC development (Chen and Ding, 1979). In spring and autumn,the average vertical wind shear between 850 and 200 hPa over the BoB is less than 10 m s−1(Fan, 1990), providing favorable conditions for TC genesis and development. The general coincidence of the peak periods of BoB TC activity with the onset and retreat of the South Asian summer monsoon (SASM) has a significant influence on the activities of the monsoon and the related southwesterly WVT (Wang and Wang, 1988, 1989; Duan et al., 2009; Islam and Peterson, 2009). The northward movement of the BoB branch of the Southwest Indian monsoon and the establishment of the Indo–China summer monsoon are both closely related to the northward movement of BoB TCs in early summer. When the tracks of BoB TCs are anomalous, the summer monsoon may establish late (Li, 1981). Chen and Zhu (1985) indicated that the BoB’s sea surface is usually dominated by an anticyclonic circulation in winter, which then turns into a cyclonic circulation in summer, and the emergence of TCs marks the beginning of the transition from the winter monsoon to the summer monsoon. Moreover, Ren et al. (2016) revealed that the activity of BoB TCs is accompanied by the onset of the BoB summer monsoon,and the peak period of the TC activity generally occurs around the summer monsoon onset. Therefore, BoB TCs are closely related to the activity of the SASM.

Numerous studies have investigated the physical mechanisms through which BoB TCs impact the weather in China, indicating that the water vapor brought by the TCs plays an important role in causing rainstorms in the low-latitude plateau region (Xu et al., 2007) and snowstorms on the Tibetan Plateau (Phurbu and Zhou,1998; Zhang et al., 2017). The southwesterly WVT caused by BoB TCs is an important factor influencing the determination of the start dates of the rainy season in Southwest China (Yan et al., 2013), and is also one of the critical water vapor sources for precipitation in the middle and lower reaches of the Yangtze River (Lyu et al., 2013; Duan and Zhang, 2015). The influence of TCs on precipitation in China primarily takes place through a southwesterly low-level jet established by its cooperation with systems such as the western Pacific subtropical high or the southern branch trough (Yang et al., 2000;Zhou et al., 2006; De et al., 2015). The jet stream from the tropical BoB to Southwest China and the middle and lower reaches of the Yangtze River can provide abundant water vapor and heat energy for precipitation in these areas. Fu et al. (2009) analyzed a persistent heavy rainfall process in Tibet, noting that the southwesterly flow in front of the Indo–Bangladesh low pressure trough steered the water vapor from the BoB to the Tibetan Plateau, providing sufficient moisture for rainfall. These results suggest that the activities of BoB TCs can significantly influence the southwesterly WVT to China.

Previous diagnostics-based case studies and qualitative analyses have emphasized the role of BoB TC activity in the southwesterly WVT during the SASM transition period. In this paper, based on statistical and composite analyses of multiple cases, the anomalous circulation for the WVT induced by BoB TCs during the monsoon transition period and its contribution to southwesterly WVT are investigated, with the aim being to further understand the role played by TCs in the WVT from the northern Indian Ocean and its important impact on precipitation in China. Section 2 provides a description of the data and methods used in the present study. Section 3 examines the connection between TCs in the BoB and the activities of the SASM. Section 4 examines the anomalous characteristics of the WVT induced by TCs in the BoB during the two peak periods. The contribution of TC-induced WVT in the bimodal periods is also analyzed. A summary and concluding remarks are provided in Section 5.

2. Data and methods

The TC data used in the paper are the best-track data for North Indian Ocean TCs from the Joint Typhoon Warning Center (JTWC). Based on these data, Duan et al. (2009) found that the annual frequency of BoB TCs has been decreasing significantly since 1977, which may be attributable to global warming or inconsistent means of observation in different periods. Initially, TC monitoring mainly relied on manual observations, whereas meteorological satellites and radars accompanied by the Dvorak TC intensity estimation technique (Chu et al.,2002) have been widely used since the 1970s. Therefore,the JTWC best-track data after 1977 are utilized for statistical analysis in this paper, covering the area of 5.0°–22.6°N, 80.0°–98.8°E. Based on the criteria of the JTWC, TCs are classified into seven intensity classes according to the maximum 1-min sustained surface wind near their centers—namely, tropical depression, tropical storm, and five grades of hurricanes.

The inter-pentad evolution of outgoing longwave radiation (OLR) was calculated by using the 1° × 1° daily OLR data during 1979–2018 from the NOAA. The seasonal variation of sea surface temperature (SST) was calculated by adopting the 2° × 2° monthly average SST data from NOAA’s Extended Reconstructed Sea Surface Temperature dataset, version 4 (https://psl.noaa.gov/thre dds/catalog/Datasets/noaa.ersst/catalog.html). The interpentad evolution of zonal winds at 850 and 100 hPa was obtained from the 2.5° × 2.5° daily reanalysis data during 1979–2018 from NCEP/NCAR. Moreover, monthly average zonal wind data were used to calculate the seasonal variation in the vertical shear of zonal wind. The climatological average hereafter refers to the average during 1979–2018. Since the tropical region around 90°–95°E in Asia is the junction between the two monsoon subsystems of the East Asian monsoon and the SAM (Chen, 2006), the area of (10°–20°N, 70°–85°E) is defined as the SAM region in this study (red box in Fig.1).

The OLR data and the daily reanalysis zonal wind data were pentad-averaged, and then the multi-year means were calculated to obtain the 73 pentad averages. The multi-year mean SST and vertical shear of zonal wind were regionally averaged in the area with frequent TCs(5°–20°N, 85°–95°E; Zhang et al., 2016) to obtain their seasonal variations in this area. The vertical shear of zonal wind is the absolute value of the difference between the zonal winds at 200 and 850 hPa (|u200hPa−u850hPa|).

According to the JTWC data during 1979–2018, there were 21 and 70 BoB TCs generated in the peak periods of May and October–November, respectively. The TC impact period is defined as the period from the record start to the positioning end referring to the JTWC data. It should be noted that three BoB TCs generated in May but ended in June, and six TCs generated in November but ended in December. These nine TCs are also analyzed.

The vectorQof the WVT flux in the whole layer of a unit air column was calculated is follows:

wheregdenotes gravitational acceleration,Vis the wind vector in each layer of the unit air column,qis the specific humidity, andpsis the surface pressure. There are 8 layers vertically—namely, 1000, 925, 850, 700, 600, 500,400, and 300 hPa.

According to the locations of the WVT channels, the zonal–vertical section at 25°N and the meridional–vertical section at 100°E were chosen to calculate and analyze the vertical circulations for the WVT of TCs, as well as the relative contribution rate of the southwesterly WVT.Only the southerly and westerly WVTs are considered in the calculation, that is,

whereFsandFwdenote the southerly and westerly WVT fluxes, respectively.

The relative contribution rate of the WVT induced by BoB TCs is calculated as follows:

whereFdenotesQ,Fs, orFw[refer to notation in Eqs.(1), (2), and (3)] at a given moment,mis the total number of 6-h observation records during the 21 TCs in May or the 70 TCs in October–November, andnis the total number of 6-h observation records in the whole month of May or months of October–November. The relative contribution rate of the TC-induced WVT is the proportion of the cumulative WVT during the TC impact period in May (October–November) to the total WVT in the whole month of May (months of October–November), which was calculated to investigate the contribution of the TCinduced southwesterly WVT during the two peak periods (May and October–November) to the annual average WVT (Duan and Zhang, 2015).

Fig. 1. Locations of the SAM region (red box; 10°–20°N, 70°–85°E) and the BoB (yellow box; 5.0°–22.6°N, 80.0°–98.8°E). The red solid line denotes the zonal–vertical section at 25°N (80°–110°E), and the black solid line denotes the meridional–vertical section at 100°E (5°–35°N).

3. Relationship between BoB TCs and SASM activity

3.1 Main factors of influence for the bimodal pattern of TC activity in the BoB

The monthly average number of BoB TCs during 1977–2018 demonstrates a bimodal pattern, with the peak values in May and October–November (Fig. 2). The primary peak is in October and November, with the number of TCs reaching 38 and 39, respectively, and the number of TCs in the secondary peak in May is 23. In March and August, only one TC is generated on average,and no TCs occur in February. Seasonally, the TCs in spring (March–May), summer (June–August), autumn(September–November), and winter (December–February) account for 20.9%, 5.9%, 55.6%, and 17.6%, respectively. Therefore, BoB TCs occur mainly in autumn and spring. However, there are differences between the atmospheric circulations in autumn and spring. In autumn (spring), the low-level circulation over the North Indian Ocean is similar to that in summer (winter), which is dominated by southwesterly (northerly) flow. This results in differences in water vapor content in the BoB (Li et al., 2013). Therefore, the relative humidity in the middle and low levels over the BoB is higher after the retreat of the summer monsoon than that before its onset,which is a possible reason for the higher TC genesis number in October–November than in May (Zhang et al.,2016).

Fig. 2. Monthly variations of TC genesis (blue bars, left-hand axis),regional average SST (orange line, right-hand axis; °C), and vertical shear of zonal wind (black line, left-hand axis; m s−1) in the BoB during 1977–2018.

The bimodal monthly variation of BoB TCs can be mainly attributed to the combined effect of the variations in SST and ambient vertical wind shear. Figure 2 shows the monthly variations of the regional average SST and vertical shear of zonal wind in the BoB, indicating that they are out of phase. During January–May, the vertical shear of zonal wind is below 12 m s−1, while the SST gradually increases in this period, reaching a maximum of 30°C in May, which is related to the heating effect of the heat flux term (Liu et al., 2013). The vertical wind shear suddenly increases to 24 m s−1in June and achieves its annual maximum in July and August, which is associated with the onset of the SASM (Fig. 3b). Due to the decrease in shortwave radiation heating and the large release of latent heat at the sea surface in summer, the surface net heat flux reduces significantly (Liu et al., 2013)and the SST gradually decreases and minimizes in August. The vertical wind shear decreases rapidly to below 10 m s−1around the time of the monsoon’s retreat in October. In November, it reaches a minimum of nearly 6 m s−1, while the SST rises to another extreme value of 29°C. In May and October–November, the vertical wind shear shows annually low values of less than 10 m s−1,while the SST is high, showing a bimodal monthly variation. Such a configuration is favorable for the genesis of BoB TCs. Although the SST is above 28.5°C in the summer monsoon period during June–September, the vertical wind shear exceeds 20 m s−1, which is not conducive to TC genesis (Li et al., 2013). In winter, the SST is less than 28°C, limiting the genesis and development of BoB TCs. Thus, the TC activity shows a bimodal pattern.

3.2 TC activity in the BoB and the onset and retreat of the SASM

The previous analysis indicates that BoB TCs are active in the transition period between the winter and summer monsoons. The abrupt development of lower-level southwesterly and upper-level northeasterly flow indicates the onset of the summer monsoon. Therefore, the abrupt transition of the upper- and lower-level zonal wind directions can be chosen as a signal of the summer monsoon’s onset and retreat. Moreover, the OLR can be used to characterize the strength of convective activity in the lower latitudes and reflect the monsoonal precipitation intensity. It can also be used to determine the onset, advance, and retreat of the summer monsoon. Chen (2006)conducted empirical orthogonal function decompositions for different summer monsoon onset indicators and found that the variance contributions of the first eigenvectors of the OLR, the 850-hPa zonal wind, and the 100-hPa zonal wind were all above 75%. Besides, the variance contribution of the 100-hPa zonal wind reached 88.9%, and was therefore the indicator most representative of the transition from winter to summer circulation in Asia.

The OLR in the SAM region (marked by the red box in Fig. 1) shows considerable fluctuation with time (Fig.3a). The OLR is large in winter and spring and decreases gradually during the transition period from spring to summer. It begins to decline significantly from the 29th pentad until it reaches its minimum value. From late June to early August, the OLR is continually less than 210 W m−2, staying at a value around the minimum. This indicates that there is deep convective activity in the SAM region during the summer monsoon. After autumn, as the summer monsoon weakens and retreats, the OLR gradually increases, indicating a weakening of convective activity. OLR less than 230 W m−2is defined as the indicator of the SASM (Lyu et al., 2006), and then the onset and retreat of the summer monsoon are in the first pentad of June and the second pentad of October, respectively. The TC genesis frequency is high before and after the onset and retreat of the summer monsoon. The number of TCs generated is highest four pentads before the onset of the summer monsoon (3rd pentad in May),which is six. The TC genesis frequency then decreases significantly after the summer monsoon’s onset. TC activity becomes more frequent three pentads before the retreat of the summer monsoon (5th pentad in September), but reaches its maximum of 11 two pentads after the summer monsoon’s retreat. These results indicate that the two peaks of TC genesis occur in the transition period of the summer monsoon (before or after the summer monsoon’s onset and retreat).

In terms of the large-scale circulation field, the 850-hPa zonal wind in the lower troposphere demonstrates a significant fluctuation with time (Fig. 3b; blue line).Easterly flow is prevalent in winter and spring, which then turns to westerly during the transition period from spring to summer. Such an abrupt change in the dominant wind direction indicates the onset of the summer monsoon. The westerlies are dominant in the SAM region from the 22nd pentad, and the wind speed gradually increases. The westerly wind speed maintains at around 11 m s−1during the period when the summer monsoon prevails from late June to early August, which is the period with the strongest westerly winds. As the summer monsoon weakens and retreats, the westerlies gradually weaken and change to easterlies. The zonal wind component at 850 hPa greater than 0 m s−1is defined as the signal of the SASM’s onset (Chen, 2006). The onset and retreat of the SASM take place in the 4th pentad of April and the 3rd pentad of October, respectively. The time of onset is different from that defined by using the OLR, but the time of retreat is approximately the same. The TC genesis frequency increases significantly during the periods before or after the summer monsoon’s onset and retreat. TCs become active one pentad before the summer monsoon’s onset, and reach a maximum five pentads after the onset, followed by a gradual decline. TC genesis becomes active again two pentads before the summer monsoon’s retreat, which then maintains until the first pentad of December. The maximum number of TCs generated appears in the first pentad after the retreat.

The variation of the 100-hPa zonal wind in the upper troposphere (Fig. 3b; orange line) is in opposite phase to that at 850 hPa. The westerly flow changes into an easterly one in the 25th pentad, and the easterly wind speed remains at around 33 m s−1during the prevailing period of the summer monsoon, which is triple that at 850 hPa.The zonal wind component at 100 hPa less than 0 m s−1is defined as the signal of the SASM’s onset (Chen,2006). Thus, the onset and retreat of the summer monsoon occurs in the first pentad of May and the third pentad of November, respectively. The time of onset is roughly the same as that defined by using the 850-hPa zonal wind, but the time of retreat is one month later. The TC genesis frequency reaches its maximum two pentads after the summer monsoon’s onset and five pentads before its retreat, respectively. Thus, BoB TCs mainly occur around the time of onset and retreat of the SASM.

Fig. 3. Inter-pentad variations of regional average (a) OLR (blue line, left-hand axis; W m−2), and (b) zonal winds at 850 hPa (blue line, lefthand axis; m s−1) and 100 hPa (orange line, left-hand axis; m s−1) and the frequency of BoB TC genesis (gray bars, right-hand axis) in the SAM region.

In summary, the two peak periods of BoB TC activity coincide with the critical periods of SASM onset and retreat. There are some differences among the dates of onset and retreat of the SASM when defined using the OLR and 850 or 100 hPa zonal winds. However, all three definitions indicate that the peak periods of TC activity in early summer and autumn mainly appear in the transition period of the summer monsoon’s onset and retreat.Since the WVT for precipitation in China is closely related to the SASM’s activity, the anomalous characteristics of the WVT induced by BoB TCs during the two peak periods of TC activity and its contribution to the monsoonal WVT need to be understood.

4. WVT anomalies induced by BoB TCs

4.1 Horizontal circulation anomalies

Figure 4 shows the average water vapor fluxes at 700 and 500 hPa during the period of impact of BoB TCs in May, the climatological mean in May and the anomaly fields. During the impact period in May, southwesterly WVT is evident, with strong northward WVT at 700 hPa(Fig. 4a). The southwesterly flow carries water vapor from the east−central BoB to the southeastern side of the Tibetan Plateau, and then it turns eastwards after being blocked until it reaches Southwest China and South China via the westerly jet. The two large-value centers of WVT are located on the southeast side of the Tibetan Plateau near 25°N and the border region of Guangdong,Hunan, and Jiangxi in China. The climatological water vapor flux in May demonstrates no southwesterly jet axis in the eastern BoB, with weaker northward transport and smaller flux values (Fig. 4c). It is dominated by eastward transport between 20° and 25°N, with smaller transport volume than that during the TC impact period.The large values of WVT are located from the eastern Indian Peninsula to Guangdong and Guangxi in China. The anomaly field (Fig. 4e) shows that the southwesterly WVT anomaly is quite significant (the anomalies are significantly different from 0 at the 0.05 level using at-test)in the eastern BoB during the TC impact period in May.There is an anomalous WVT channel from the east−central BoB to the southeastern side of the Tibetan Plateau.

The characteristics of the southwesterly WVT at 500 hPa during the TC impact period are similar to those at 700 hPa, with the water vapor first transported northwards and then eastwards. The magnitude of WVT at 500 hPa is significantly smaller than that at 700 hPa, but the area of WVT is situated more to the north, up to nearly 35°N. This can be attributed to the weakened blocking effect of the Tibetan Plateau at upper levels(Fig. 4b). In terms of the climatological average (Fig.4d), the southwesterly WVT in the BoB is smaller than that during the TC impact period at 500 hPa, with the large WVT values located from southeastern Tibet to Guangdong and Guangxi in China. The anomaly field is similar to that at 700 hPa (Fig. 4f). There is a significant southwesterly WVT anomaly over the BoB (the anomalies are significantly different from 0 at the 0.05 level using at-test) and the northward transport is more northerly over southeastern Tibet.

The mid-to-low-level WVT during the TC impact period in October–November at both 500 and 700 hPa is relatively smaller than that in May (Figs. 5a, b), which is related to the weakening and retreat of the summer monsoon. The strong southwesterly airflow retreats eastwards from the BoB to the northern part of the Indo–China Peninsula with the southwest monsoon weakening sharply, meaning that there is not such a strong southwesterly wind to transport water vapor in the area of the BoB, which leads to a weaker southwesterly WVT than that in May (Duan and Zhang, 2015). The westward WVT from the South China Sea region reaches the BoB in October–November, which converges with the water vapor from the BoB TCs and is transported northwards and then eastwards. The Indo–China Peninsula and the east–central BoB are controlled by the strong easterly and southeasterly airflows on the south and southwest side of the subtropical high during the TC impact period, and weak southwesterly WVT is found only in the northern BoB. This is associated with the climatological state of the subtropical high (Figs. 5c, d).The anomaly field in October–November (Fig. 5e) shows that the BoB is dominated by a positive anomalous cyclonic circulation of water vapor flux. In addition, anomalous westward WVT from the South China Sea region is found at 700 hPa. The WVT anomalies at 500 hPa are less significant than those at 700 hPa, with weaker northward WVT anomalies on the northeast side of TCs (Fig.5f). The results indicate that the early-summer TCs have a more pronounced effect on the water vapor circulation in the area around the BoB.

Fig. 4. (a, b) Average water vapor fluxes (shading and contours) and vector (arrows) fields during the period of impact of BoB TCs in May 1979–2018, (c, d) climatological mean water vapor fluxes in May, and (e, f) anomaly fields at (a, c, e) 700 hPa and (b, d, f) 500 hPa (unit: g hPa−1 s−1 cm−1). In (e, f), anomaly fields with stippled small black dots and fields with heavily plotted vectors of water vapor fluxes are significantly different from 0 at the 0.05 level using a t-test.

In the difference field of the whole-layer-average water vapor flux between the TC impact period and the climatological mean (Fig. 6), there is a clear cyclonic circulation during the TC impact period in May over the BoB,with an abnormal southwesterly water vapor channel in its east–central part, indicating more northward WVT induced by TCs. The southeastern Tibetan Plateau and the area from Southwest China to the middle and lower reaches of the Yangtze River are also areas of positive WVT anomalies, while the east of the Indo–China Peninsula and Indian Peninsula are home to negative anomalies. The abnormal northward WVT in the cyclonic circulation is situated more to the south and is weaker during the TC impact period in October–November compared to that in May.

In summary, TC activity has a significant influence on the northward WVT from the BoB. The WVT in May is more widespread and stronger than that in October–November. The anomalous WVT volume in the whole layer in May is 1.7 times larger than that in October–November.

4.2 Vertical circulation anomalies

Fig. 5. As in Fig. 4, but for October–November.

The meridional and zonal WVT during the peak periods of TCs are investigated by using vertical sections taken at 25°N, 100°E. Since the WVT during the TC impact periods is dominated by southwesterly flow, it was assumed that the value of northerly and easterly WVT were zero and only the southerly and westerly components of water vapor flux were considered. The vertical profiles of the meridional and zonal water vapor fluxes at 25°N, 100°E during the TC impact period and their anomalies in May 1979‒2018 are shown in Fig. 7. The primary large-value center of the WVT induced by the southerly component is located roughly in the lower troposphere below 800 hPa in the region around 110°E (Fig.7a). The secondary large-value center, meanwhile, is located around 95°E at 850 hPa. For the climatological mean (figure omitted), the position of the primary largevalue center is approximately the same as that during the TC impact period. However, the secondary large-value center is situated around 90°E at 900 hPa, which is more to the west than that during the TC impact period. The anomaly plot (Fig. 7c) indicates that the largest anomalous values are in the region of 96°‒98°E from 800 to 700 hPa (the anomalies are significantly different from 0 at the 0.05 level using at-test), indicating that the TCs increase the southerly WVT mainly in the lower troposphere in the eastern area of the BoB.

In the 100°E vertical section during the TC impact period in May, the mean WVT induced by the westerly component demonstrates two large-value centers near 10°‒13°N at 900 hPa and 22°‒28°N at 800‒600 hPa, respectively (Fig. 7b), which is generally consistent with the climatological mean (figure omitted). Figure 7d shows that the anomalies north of 20°N are positive, indicating a positive contribution of BoB TCs to the eastward WVT. Affected by the strong southwesterly airflow in May, the water vapor channel of TCs will move northwards and eastwards, resulting in the positive-value center of the difference being more to the east and north,while there is a correspondingly negative difference in the regions more to the west and south. The positive anomaly center is in the region of 23°‒30°N at 650‒400 hPa (the anomalies are significantly different from 0 at the 0.05 level using at-test), which is the area of the southeastern Tibetan Plateau and western Yunnan. This indicates that abundant water vapor is transported eastwards into China under the influence of BoB TCs.

Fig. 6. Anomalies of the whole-layer-average water vapor flux (shading and contours), and the anomalous water vapor flux vectors (arrows; kg s−1 m−1), in (a) May and (b) October–November during 1979–2018. The anomaly fields with stippled small black dots and the fields with heavily plotted vectors of water vapor fluxes are significantly different from 0 at the 0.05 level using a t-test.

Fig. 7. Vertical profiles of (a, b) the mean water vapor flux during the period of impact of BoB TCs (shading and contours; g hPa−1 s−1 cm−1)and (c, d) difference between the TC impact period and the climatological mean at (a, c) 25°N and (b, d) 100°E in May 1979–2018. In (c, d), the anomaly fields with stippled small black dots are significantly different from 0 at the 0.05 level using a t-test.

The northward WVT anomalies are larger than the eastward WVT anomalies, indicating more WVT induced by the southerly component during the TC impact period. It is worth noting that there is a strong westerly moisture jet on the north side of the subtropical high both in the climatological mean and the periods of TC circulation, which is conducive to the southerly WVT to be transported continuously to the east during the TC period(refer to Figs. 4, 5).

Figure 8 shows the vertical profiles of the average WVT fluxes during the TC impact period and the anomalies in October–November at 25°N, 100°E. Compared with the condition in May (Fig. 8a), the large center of meridional water vapor flux in the 25°N vertical section is unique within 98°–110°E below 700 hPa. In terms of the climatological mean (figure omitted), the vertical profile is generally consistent with the TC impact period.The location of the largest positive anomalies is within 90°–100°E from 700 to 450 hPa (Fig. 8c). However, the anomalies of the northward WVT are negative in most areas of 80°–90°E below 800 hPa.

The water vapor fluxes in both the zonal and meridional vertical sections are smaller during the TC impact period in October–November than those in May (refer to Figs. 7a, b and Figs. 8a, b). The water vapor fluxes in the meridional vertical sections during the TC impact period in May can reach more than 4.5 g hPa−1s−1cm−1(Fig.7b), while the maximum water vapor flux in October–November is only about 2.5 g hPa−1s−1cm−1(Fig. 8b), which is about half of that in May.

In the 100°E vertical section (Fig. 8b), the large-value center of the westerly WVT is located higher around 600 hPa and 26°N, which is generally consistent with that of the climatological mean (figure omitted). As shown in the anomaly plot (Fig. 8d), positive anomalies are found within 15°–28°N above 800 hPa, indicating anomalous westerly WVT to western Yunnan and parts of southwestern Sichuan on the southeastern side of the Tibetan Plateau during the TC impact period. However, negative anomalies are dominant within 10°–20°N below 900 hPa,which results from the decreased westerly WVT due to the strong easterly WVT from the South China Sea(shown in Fig. 5a). Moreover, the eastward zonal WVT is slightly larger than the northward meridional WVT during the TC impact period in October–November(Figs. 8a, b). The eastward WVT anomalies are larger compared to the northward WVT anomalies, indicating more WVT induced by the westerly component during the TC impact period, which is different from that in May.

In general, the large-value area of the southerly WVT anomalies during the TC impact periods is located in the lower troposphere in low-latitude areas, whereas that of the westerly WVT anomalies is located in the middle troposphere in high-latitude areas. The results indicate a water vapor channel in which the southwesterly jet advances northwards from the BoB to the southeastern Tibetan Plateau and then turns eastwards to East China via the westerly jet in the periphery of the subtropical high (shown in Figs. 4a, 5a). It is an important water vapor channel affecting the occurrence of droughts and floods in the middle and lower reaches of the Yangtze River basin (Xu and Chen, 2006), and the TC activity can help to enhance and maintain the water vapor channel.

4.3 Differences in the circulation situation for the TCinduced WVT during the two peak periods

The above analysis shows that the anomalous characteristics of the WVT differ between the two TC impact periods of May and October–November. Although more TCs are generated in October–November than in May,the average WVT flux is greater in the TC impact period of May.

Fig. 8. As in Fig. 7, but for October–November.

In the 500-hPa average wind field during the TC impact period in May during 1979–2018 (Fig. 9a), the southern branch trough on the southern side of the Tibetan Plateau is active, and the western North Pacific subtropical high ridge is around 15°N. Meanwhile, the BoB is located in front of the deep southern branch trough and west of the subtropical high, so the strong southwesterly flows between the trough and the subtropical high may transport a large amount of water vapor northeastwards from the BoB TCs to China. In October–November (Fig. 9b), the subtropical high weakens and retreats, with the position being more to the south and east, and the southern branch trough is less active, resulting in weak southwesterly flow in the BoB. Accordingly,there is not such a strong southwesterly wind to transport water vapor. The west–central BoB and Indian Peninsula are controlled by cyclonic circulation, and southeasterly winds from the southwest side of the subtropical high prevail in the east–central BoB. There is only a weak southwesterly flow in front of the shallow trough in the northern BoB. The results indicate that the southwesterly WVT during the TC impact period in October–November is weaker, and situated more to the south and east than that in May.

4.4 Contributions of TCs in the BoB to the southwesterly WVT

4.4.1Spatial distribution

The relative contribution rate of the southwesterly WVT anomaly under the influence of TCs to the total WVT was calculated by grid point in May and October–November, separately. Figure 10a shows the distribution of the relative contribution rate of TC-induced WVT in the whole layer in May during 1979–2018. Note that the southwesterly WVT induced by BoB TCs accounts for 9%–18% of the total in most regions. The large-rate center is in the central BoB, which extends northeastwards to 35°N, covering the western Indo–China Peninsula,eastern Tibetan Plateau, western Sichuan, and most of Yunnan with a value of 12%. The large-rate center in October–November is located in the central BoB, with the largest contribution rate being 26%, which is more to the north and larger than that in May (Fig. 10b). The contribution rate in the southeastern Tibetan Plateau, Sichuan,and most of Yunnan is about 16%–19%, indicating that BoB TCs play a more important role in transporting WVT to China in October–November than they do in May.

4.4.2Vertical distribution

Figure 11 shows the contribution rates of TC-induced vertical WVT at 25°N, 100°E in May and October–November during 1979–2018. The contribution rate of the southerly WVT can reach up to 28% in May (Fig.11a). The large-value center is situated at around 500 hPa and 93°E, and the large-value area tilts to the west with height. The maximum contribution rate of the westerly WVT in May is 12% (Fig. 11b), with the large-value area located in the middle troposphere at around 600–400 hPa and 20°–30°N. That is, the contribution rate of the southerly WVT is larger than that of the westerly WVT.

Fig. 9. Distributions of the mean geopotential height field (contours; dagpm) and wind field (arrows; m s−1) at 500 hPa during the TC impact period of (a) May and (b) October–November during 1979–2018.

Fig. 10. Relative contribution rates of TC-induced WVT in the whole layer in (a) May and (b) October–November during 1979–2018.

Fig. 11. Contribution rates of (a, c) the southerly vertical WVT at 25°N and (b, d) the westerly vertical WVT at 100°E in the TC impact periods of (a, b) May and (c, d) October–November during 1979–2018.

The contribution rate of the southerly component is 10%–24% in most regions in October–November. The large-value center is located within 80°–95°E at around 500 hPa, which is more to the west than that in May (Fig.11c). The largest contribution rate of the westerly component can reach 24% in October–November (Fig. 11d),which is twice that in May. The large-value area is located within 17°–22°N at around 400 hPa, with the center being 3°–5° of latitude farther south than that in May. In comparison, the proportion of the southerly WVT in October–November is less than that in May, with a more westward position, but the proportion of the westerly WVT is greater and with a more southward position. In addition, the large values of the southerly and westerly WVT contribution are located in the middle troposphere during the TC periods, implying a deepened wet layer under the impact of TCs.

5. Conclusions

TCs in the BoB occur mostly in May and October–November, representing a bimodal pattern. The pattern is related to the high SST and the vertical shear of zonal wind being less than 10 m s−1in the BoB during the two peak periods. By using different monsoon indicators to analyze the onset, advance and retreat of the SASM, it was found that the two peak periods of BoB TC activity coincide with the critical periods of the summer monsoon’s onset and retreat. In general, the TC activity peak in May mainly occurs a few days before the onset of the monsoon, whereas that in October–November mainly occurs 5–10 days after the monsoon’s retreat, implying an important relationship between BoB TCs and southwesterly monsoonal activity.

The southwesterly WVT is significant in the middle and lower layers during the bimodal periods of BoB TC activity, with large values situated in the east–central BoB, southeastern Tibetan Plateau, Southwest China, and South China. The southwesterly WVT from the Indian Ocean is dominant in May. The meridional WVT at 500 hPa is situated more to the north than that at 700 hPa,reaching nearly 35°N and covering the southeastern Tibetan Plateau owing to the weakening of the plateau’s blocking effect at upper levels. However, the southwesterly WVT, including the water vapor from the South China Sea, is much weaker in October–November. The southwesterly WVT during the TC period in May is stronger and more widespread than that in October–November, with about twice the WVT capacity.

The TC activity mainly increases the northward WVT anomalies compared to the whole-layer mean, and there is a clear cyclonic circulation anomaly in the BoB. More southerly/westerly component WVT anomalies are generated during the TC impact period in May than in October–November. Meanwhile, the large values of the southerly WVT anomalies are located in the lower troposphere in low-latitude areas, and those of the westerly WVT anomalies are mainly located in the middle troposphere in high-latitude areas. This indicates that a WVT channel is formed under the influence of TCs in which the southwesterly water vapor first climbs northwards to the southeastern Tibetan Plateau and then turns eastwards before reaching South China via the westerly zone jet.

The large-relative-contribution-rate center of the TCinduced WVT in the whole layer is situated in the central BoB in the bimodal periods, with the largest contribution rates being 18% and 26%, respectively. The larger contribution rate in southwestern Yunnan reaches 12% in May and in the southeastern Tibetan Plateau, Sichuan,and most of Yunnan it is about 16%–19% in October–November, indicating a larger WVT contribution in October–November than in May. The large values of the southerly and westerly WVT contribution are located in the middle troposphere.

The results of this study reveal the anomalous characteristics of the WVT during the TC impact periods in the transition season of the SASM, which is important because of the effect on precipitation in China. However,the influence of the interaction between TCs and SASM activity still needs to be further investigated.