1 Introduction

The South China Sea summer monsoon (SCSSM) onset can be viewed as a combination of the eastward retreat of west Pacific subtropical high (WPSH), the intrusion of low-level southwesterly monsoonal winds into the South China Sea (SCS) from the Indian Ocean, and abrupt increase in rainfall. The SCSSM onset usually plays a critical role in the East Asian summer monsoon (EASM) system. The SCSSM onset is considered as a precursor for the commencement of EASM (Tao and Chen 1987; Lau and Yang 1997) and is also regarded as starting of the rainy season over China (Wang and LinHo 2002; Wang et al. 2004). The monsoon variability plays typically a significant source of extra-tropical climate predictability (Wang and Coauthors 2009; Lee and Coauthors 2010, 2011). An extensive study has already been conducted to understand the inter-annual, decadal, and long-term monsoon variability. Wang et al. (2012a) reported that the mean state changes over the Pacific Ocean significantly affect the recent global monsoon variability. In the monsoon onset, Kajikawa and Wang (2012) reported a clear advancement of the Asian summer monsoon (ASM) onset in recent decades. Regionally, the ASM onset has advanced considerably after 1997 in the Arabian Sea due to tropical storm intensification (Wang et al. 2012b) and in the Bay of Bengal (BOB) (Yu et al. 2012), as well. There is also a strong decadal variability after 1993/1994 with the advancement in SCSSM onset (Kajikawa and Wang 2012). The monsoon onset variability has always been sensitive to the agriculture, economy, and global climate system. The early or late onset of the SCSSM may be connected to the occurrence of drought/flood during the boreal summer months (Ding and Chan 2005; Huang et al. 2006).

Most of the previous studies have emphasized the importance of sea surface temperature (SST) warming over the Western Pacific (WP) and the Indian Ocean as the possible potential factors behind the advancement of the ASM onset or monsoon variability (Wei-Dong et al. 2012; Xiang and Wang, 2013; Choudhury et al. 2019). Yu et al. (2012) reported that SST changes play the driving role for the progress of monsoon over BOB. Kajikawa and Wang (2012) suggested that warmer WP triggers the SCSSM onset in advance. Moreover, it is known that the advancement of the SCSSM onset was driven by the interdecadal warming of the warm pool during spring, which triggered the stronger convection anomalies and a weaker WPSH. In a separate study, Wang et al. (2012b) highlighted that the thermal contrast between the Asian continent and the tropical Indian Ocean might play an essential role in governing the ASM onset.

This study aims to investigate the composite monthly mean state of May and the daily evolution of major meteorological variables associated with some extremely early and late SCSSM onset years. The main objective of this study is to find the key features associated with some early and late onset years based on the available observation and reanalysis data. Also, we attempt to address the reason behind the late SCSSM onset. The rest of the paper is organized as follows. “Section 2” describes the data and method used in this study. The results are described in “Section 3.” Possible mechanisms responsible for onset variability are provided in “Section 4.” “Section 5” describes the summary with concluding remarks.

2 Data and methodology

Here, in this study, we incorporate the monthly and daily data sets, which are as follows: Daily SST data from National Oceanic and Atmospheric Administration (NOAA) optimum interpolated SST (Smith et al. 2008), the daily mean interpolated outgoing long-wave radiation (OLR) data from NOAA (Liebmann and Smith 1996), daily and monthly mean sea level pressure (SLP), wind fields, air temperature from the National Centers for Environmental Prediction (NCEP)-National Center for Atmospheric Research (NCAR) reanalysis data set (Kalnay and Coauthors 1996), and the monthly mean Hadley Center SST data set (Rayner et al. 2003).

The SCSSM onset dates are taken from the list of onset dates provided by Hu et al. (2018) from 1979 to 2016. The mean date of the SCSSM onset is May 21, as Hu et al. (2018) determined. Hu et al. (2018) showed that the SCSSM onset date’s standard deviation is approximately 2 weeks. Therefore, the early onset years (hereafter it will be called as “early onset”) are the years when onset occurred as early as 14 days or more and similarly the late onset years (hereafter, it will be called as “late onset”) are the years when onset occurred as late as 14 days or more from the mean onset date (May 21). Thus, from 1979 to 2016, 5 early onset and 9 late onset years are selected from the list which is given in Table 1. The anomalies of different meteorological variables are calculated as total mean minus the climatology in May during the entire study periods, i.e., from 1979 to 2016. Since the overall ASM onset occurs in May, this paper focuses on the changes in mean fields in May. Also, it includes the composite daily evolution of vital meteorological variables to find the difference and evolution pattern 10 days before the SCSSM onset for the early and late onset years. All the differences are tested with a two-tailed t-test.

Table 1 List of SCSSM onset dates
Full size table

3 Results

3.1 Atmospheric mean circulations during May

To understand the monthly mean features during the early and late onset years, we analyze the composite mean pattern of SST, SLP, wind at 850 hPa and 200 hPa during May. Figure 1 shows the composite of SST and SLP anomalies in the left and right panels, respectively. The May composite SST pattern during the early onset, late onset, and difference plots (Fig. 1a–c) show that during the early onset years, cold anomalies are dominant over the Indian Ocean, BOB, Sea of Japan, whereas warm anomalies are dominant over the North Pacific Ocean and SCS. On the other hand, the May composite SLP anomalies during the early onset, late onset, and difference plots are shown in Fig. 1d–f, respectively. The difference plot (Fig. 1f) shows that during the early onset years, SLP anomalies are significantly higher over the Indian Ocean and SLP anomalies are considerably lower over East Asia, the North Pacific Ocean, SCS, and the Sea of Japan. The relatively lower SST anomalies over the Indian Ocean and higher SST anomalies over Northern China, East Asia, SCS, and Sea of Japan increase and decrease the SLP anomalies over the Indian Ocean and Northern China, East Asia, and SCS, respectively.

Fig. 1
figure 1

Composite monthly mean of a, b, c SST anomalies (°C) and d, e, f SLP anomalies (hPa) for early onset years, late onset years and its difference (early–late) during May, respectively. Dots in the composite differences indicate values higher than 95% confidence level (two-tailed t-test)

Full size image

The wind at 850 hPa and 200 hPa during the early onset years, late onset years, and the differences between them are shown in Fig. 2 (a,b,c) and 2 (d,e,f), respectively. During the early onset years, the cross-equatorial low-level jet (LLJ) over the Indian Ocean is stronger. It converges SCS and East Asia, whereas, during the late onset years, the wind from the Indian Ocean to the SCS is weaker. Secondly, over the western Pacific Ocean, the anticyclone is centered away from the SCS during the early onset years. During the late onset years, the anticyclone stretches more towards East Asia and takes the moisture away from SCS towards the North Pacific Ocean. At 200 hPa (Fig. 2d–f), the subtropical jet (STJ) is weaker but shifts more southward over South and East Asian landmass during the early onset years.

Fig. 2
figure 2

Composite monthly mean of wind at a, b, c 850 hPa (ms−1) and at d, e, f 200 hPa (ms−1) for early onset years, late onset years and its difference (early–late) during May, respectively

Full size image

Therefore, in May composite analyses, during the early onset years, SLP is higher over the Indian Ocean and lower over the North Pacific Ocean, SCS, and East Asia, which strengthens the LLJ from the Indian Ocean towards the SCS. In the next section, we investigate the daily evolution of different meteorological variables during some extremely early and late onset years to examine the robustness of the onset variation processes.

3.2 Daily evolution of meteorological variables

To cast further light on the evolution pattern before the early and late SCSSM onset, we further discuss the composite patterns of OLR, SLP, SST, and surface air temperature (SAT), respectively. Here, the daily evolution is plotted for 10, 8, 6, 4, 2 days (hereafter, it will read as day -10, day -8, day -6, day -4, day -2, respectively) before the corresponding onset dates.

Figure 3 contains the daily evolution of OLR anomalies. It can be seen that across the entire evolution process from day -10 to -2, during early SCSSM onset years, the convective activities are mostly confined over East Asia, the North Pacific Ocean, and the Sea of Japan (Fig. 3a–e). On the other hand, convective anomalies are weaker and shift westward over the Indian subcontinent, Central Asia, and the Indian Ocean during the late onset evolution period (Fig. 3f–j). In the composite difference (Fig. 3k–o), the early onset years are accompanied by higher convective activities over East Asia, the North Pacific Ocean, the Sea of Japan, and the SCS. This increase in convective activities during the early SCSSM onset years is consistent with the strengthening of equatorial LLJ towards the SCS and East Asia.

Fig. 3
figure 3

Composite daily evolution of OLR anomalies (Wm−2) from day -10 to day -2 prior to the a, b, c, d, e early onset years, f, g, h, i, j late onset years and k, l, m, n, o its differences (early–late), respectively. Dots in the composite differences indicate values higher than 95% confidence level (two-tailed t-test)

Full size image

The daily evolution pattern of SLP anomalies during the early and late onset years is shown in Fig. 4. It indicates that negative SLP anomalies develop over the North Pacific Ocean, BOB, and East Asia during the early onset years. It can be seen that the band of negative SLP anomalies during early onset years extend from the North Pacific Ocean to the East China Sea before the onset. On the other hand, during the late onset years, the band of positive SLP anomalies can be observed over the Indian subcontinent and East Asia, which stretches to the East China Sea and North-West Pacific Ocean before the onset. In the difference plot (Fig. 4k–o), right from day -10 to day -2, the SLP anomalies are significantly lower over the North-West Pacific Ocean, East Asia, North Indian landmass, BOB, and SCS, while SLP anomalies are higher over the West Indian Ocean and the Arabian Sea during the early onset years. This difference in SLP anomalies between the West Indian Ocean and SCS-East Asia strengthens the LLJ from the Indian Ocean to the SCS and East Asian landmasses.

Fig. 4
figure 4

Composite daily evolution of SLP anomalies (hPa) from day -10 to day -2 prior to the a, b, c, d, e early onset years, f, g, h, i, j late onset years and k, l, m, n, o its differences (early–late), respectively. Dots in the composite differences indicate values higher than 95% confidence level (two-tailed t-test)

Full size image

To investigate the factors driving the anomalous SLP changes over the Indian and North Pacific Oceans during early and late onset years, we analyzed the changes in composite mean SAT anomalies at 1000 hPa, shown in Fig. 5. It can be seen that the Northern part of China, East Asia, and the North Pacific Ocean was significantly warmer during the entire evolution period of early SCSSM onset years as compared to late onset years, which coincides approximately with low SLP anomalies over these regions. These high SAT anomalies lower the SLP anomalies over Northern China and the North Pacific Ocean, which gradually strengthens the Tibetan heat low. On the other hand, the SAT anomalies are colder over the Indian Ocean. Therefore, a strong land-sea thermal contrast over the East Asian region-SCS and the Indian Ocean is associated with early SCSSM onset, which eventually triggers the SLP variation between them.

Fig. 5
figure 5

Composite daily evolution of SAT anomalies (hPa) from day -10 to day -2 prior to the a, b, c, d, e early onset years, f, g, h, i, j late onset years and k, l, m, n, o its differences (early–late), respectively. Dots in the composite differences indicate values higher than 95% confidence level (two-tailed t-test)

Full size image

4 Factors responsible for early and late SCSSM onset

To analyze the physical mechanism driving the early SCSSM onset, we plot the May temperature advection and wind anomalies at 1000 hPa during the early onset years, late onset years, and the difference between them in Fig. 6a, b, c, respectively. During the early onset years, the surface temperature advection is higher over East Asia and stretched over the Indian subcontinent, Northern China, SCS, and South of Japan. Simultaneously, the wind vectors are pointing more from the North Pacific Ocean towards the East Asian landmasses. The difference plot (Fig. 6c) exhibits a significantly higher temperature advection over Northern China, the North Pacific Ocean, along the coastline of Southern China, SCS, and the Western part of the Indian subcontinent. Simultaneously, the surface wind strengthens from the North Pacific Ocean to North China, SCS, and the coastline bordering South China, along the pathways of higher temperature advection. It indicates that during the early onset years, warmer SAT anomalies over Northern China and East Asia are due to the advection of warm anomalies from the North Pacific Ocean, which decrease the SLP anomalies over Northern China, SCS, and East Asian landmasses. It strengthens the moisture-laden wind from the Indian Ocean to SCS and East Asian continent, which advances the SCSSM monsoon onset at least by 2 weeks. Thus, the onset variability is primarily due to the SLP variation between the Indian and SCS-North Pacific Ocean.

Fig. 6
figure 6

Composite monthly mean of surface temperature advection (°C ms−1) over-plotted with wind at 1000 hPa for a early onset years, b late onset years and c its difference (early–late) during May, respectively

Full size image

5 Summary

In this study, we investigate the key features associated with the late and early SCSSM onset. To carry out the investigation, we select five extreme early and nine extreme late onset years (after and before 14 days or more from the regular onset date is called “late” and “early” onset year, respectively). May 21 is chosen as the regular onset date as determined by Hu et al. (2018). Then, we perform composite analyses of different variables during May, both for early and late onset years. Next, examine the differences in daily evolution (from 10 to 2 days before the onset date) before the SCSSM onset, the composite of OLR, wind at 850 hPa and 200 hPa, SLP, SST, and SAT anomalies are investigated. The preliminary results can be summarized as follows:

In May, the early onset years are associated with lower SLP anomalies over the North Pacific Ocean, SCS, East Asian landmasses, and higher SLP anomalies over the Indian Ocean, which strengthens the LLJ from the Indian Ocean to the SCS and East Asia.

In the daily evolution, the composite difference of OLR clearly shows that the convective activities are higher over East Asia, SCS, and the Sea of Japan during early onset years. Like May composite analysis, daily SLP anomalies decrease significantly over the North-West Pacific Ocean, East Asia, North Indian landmass, BOB, and SCS, while SLP anomalies increase over the West Indian Ocean and the Arabian Sea during the early onset years. This difference in SLP anomalies between the West Indian Ocean and SCS-East Asia strengthens the LLJ from the Indian Ocean to the SCS and East Asian landmasses.

The anomalous changes in SLP anomalies over the Indian Ocean and North Pacific Ocean during early onset years are due to warmer SAT anomalies in the Northern part of China, East Asia, and North Pacific Ocean and colder SAT anomalies in the Indian Ocean. These SAT differences coincide with the lower and higher SLP anomalies over East Asia and the Indian Ocean. Therefore, during the early SCSSM onset years, a strong land-sea thermal contrast between East Asia-SCS and the Indian Ocean is associated with the SLP variation between the two regions.

Finally, we examine the physical mechanism driving the early SCSSM onset. During the early onset years, the May temperature advection is higher over Northern China, the North Pacific Ocean, along the coastline of Southern China, SCS, and the Western part of the Indian subcontinent. These warmer SAT anomalies advect from the North Pacific Ocean, which causes the SLP anomalies to decrease over Northern China, SCS, and East Asian continents. These differences in SLP anomalies strengthen the moisture-laden wind from the Indian Ocean to SCS and East Asian landmasses, which advances the SCSSM monsoon onset at least by 2 weeks.