Variations in the annual cycle of the East Asian monsoon and its phase-induced interseasonal rainfall anomalies in China

JIANG Song,ZHU Congwen and JIANG Ning

Institute of Climate System,Chinese Academy of Meteorological Sciences,Beijing,China

ABSTRACT The East Asian monsoon (EAM) exhibits a robust annual cycle with significant interannual variability. Here, the authors find that the EAM annual cycle can be decomposed into the equinoctial and solstitial modes in the combined sea level pressure,850-hPa low-level wind,and rainfall fields.The solstitial mode shows a zonal pressure contrast between the continental thermal low and the western Pacific subtropical high,reaching its peak in July and dominating the East Asian summer monsoon. The equinoctial mode shows an approximate zonal contrast between the low-level cyclone over the east of the Tibetan Plateau and the western Pacific anticyclone over the east of the Philippines. It prevails during the spring rainy season in South China and reaches its peak in April.The interannual variations of the lead-lag phase of the two modes may result in the negative correlation of rainfall anomalies in North China between spring and fall and in South China between winter and summer,which provides a potential basis for the across-seasonal prediction of rainfall.The warm phase of ENSO in winter could give rise to the reverse interseasonal rainfall anomalies in South China,while the SST anomaly in the Northwest Pacific Ocean may regulate the rainfall anomaly in North China.

KEYWORDS Annual cycle;East Asian monsoon;interannual variability

1. Introduction

The East Asian monsoon (EAM) shows a robust annual cycle, characterized by the alternation of dry and wet seasons with the seasonal reversal in the monsoon circulation between winter and summer (e.g., Webster et al. 1998; Ding and Chan 2005). This has been widely regarded as a direct response to the seasonal cycle of solar radiation with land-sea thermal contrast over the Asia-Pacific regions.The rain belt of the East Asian summer monsoon begins with the spring rainy season in South China, goes through the mei-yu season in the lower reaches of the Yangtze River, and finally reaches its peak during the rainy season in North China(He et al.2008; He and Liu 2016; Zhu et al. 2011), showing a stepwise northward shift from April to August during the annual cycle of the EAM (Ding 1992; Wu and Wang 2001; Zhu et al. 2011). Wang et al. (2008) analyzed the annual cycle of tropical precipitation and lower-level circulation and found the solstitial and equinoctial asymmetric modes are dominant in the annual cycle of global monsoon rainfall. Wu, Zhou, and Tim (2009) and Hao,Ding, and Ming (2012) analyzed the seasonal evolution of the EAM and discussed the impact of sea surface temperature(SST)patterns on the interannual variability of the EAM.However,the dominant modes of the annual cycle of the EAM and its interannual variability have not been clearly addressed yet.

The annual cycle of the EAM exhibits year-to-year variation, characterized by an earlier or later onset of the dry or rainy season with rainfall anomalies in a given region.Evidence shows that the phase anomaly of the annual cycle of the EAM exhibits a large-scale decadal oscillation (Chang and Li 2000; Liu and Ding 2012).It not only results in the seasonal rainfall anomaly,but also gives rise to the out-of-seasonal rainfall response (Sun and Sun 1995; Chen, Graf and Huang,2000). This interseasonal linkage of rainfall anomalies is possibly regulated by the SST anomaly (Sun and Sun 1996;Yang,Chen,and Sun 2005).

In the present study, we aim to reveal the dominant modes of the annual cycle of the EAM and discuss its interannual variation.We focus on the phase anomaly to explore the internal connection between the annual cycle and seasonal rainfall anomaly in East Asia,and try to provide a feasible approach to the seasonal prediction of rainfall in China.

2. Data and methods

The wind feild and sea level pressure (SLP) dataare from the daily NCEP-DOE Reanalysis dataset, with a horizontal resolution of 2.5°×2.5°and covering the period from 1979 to 2016(Kanamitsu et al.2002).The pentad-averaged precipitation is based on the Climate Prediction Center Merged Analysis of Precipitation,with a horizontal resolution of 2.5° × 2.5° and covering the period from 1979 to 2016(Xie and Arkin 1997).Monthly averaged SST data,with a horizontal resolution of 1°×1°and covering the period from 1979 to 2016,are provided by the Hadley Centre Sea Ice and Sea Surface Temperature dataset (Rayner et al.2003). The climatology is defined as the 30-yr arithmetic average between 1981 and 2010,for which the departures from this climatology are referred to as the climate variation of anomalies.

Following Stine, Huybers, and Fung (2009), we define the first harmonic component as the annual cycle relative to the solar insolation,the amplitude is the maximum value of the annual cycle, and the phase is the timing corresponding to the amplitude.Taking the rainfall at 115°E as an example,the first harmonic can basically resemble the annual cycle of the first three harmonic components(>90 days)(figure not shown).Therefore,the annual cycle of the EAM can be simplified as the first harmonic monsoon circulation regime.We apply multivariate empirical orthogonal function(MV-EOF)analysis to reveal the annual cycle modes of the EAM(Wang 1992;Wang and Ding 2008;Song et al.2016),and non-dimensional eigenvectors are used to standardize the zonal-mean removed meteorological fields before the analysis.

3. Dominant modes of the EAM annual cycle

Figure 1 shows the first two MV-EOF modes and their principle components(PCs)of the annual cycle component of the EAM in the combined SLP,850-hPa wind and rainfall fields. They are statistically significant according to the criterion of North et al. (1982), and account for 75.5%and 24.4% of the total variance, respectively. The first mode shows a zonal dipole of pressure gradient divided by 140°E and exhibits a large-scale cyclonic circulation with low pressure,centered over the Mongolian region(Figure 1(a)).In contrast,an anticyclonic circulation with high pressure locates over the North Pacific near the dateline.Rainfall enhances and prevails over most of continental East Asia,corresponding to the southerly summer monsoon winds at 850 hPa (Figure 1(c)). This mode reaches its positive and negative maximum in July and January(Figure 1(e)),revealing the seasonal reversal of the monsoon circulation regime between summer and winter.

The second MV-EOF mode also shows a zonal circulation contrast,but locates in the lower latitudes(Figure 1(b)).A cyclonic circulation with low-pressure nests in the eastern side of the Tibetan Plateau and Myanmar,while an anticyclonic circulation with high pressure locates in the east of the Philippines over the Northwest Pacific.The rain belt is mainly concentrated in southern China, extending eastwards to the south of Japan and then to the midlatitude western Pacific (Figure 1(d)). PC2 reaches its positive and negative maximum in April and November (Figure 1(f)),suggesting a seasonal reversal of the circulation and rainfall regime between spring and fall.

PC1 lags the phase of solar decline at the summer solstice by approximately 1 month, and occurs concurrently with the maximum warming of land surface air temperature (Trenberth 1983), corresponding to the peak phase of the solstitial mode in the tropical summer monsoon(Wang et al.2008).Therefore,the first mode,as a lagged response of the EAM to the solar radiation at the solstice,can be named as the solstitial mode.During boreal summer, the warming land surface air temperatures enhance the low-level continental low, the zonal pressure gradient, southwesterly flow, and rainfall over the whole of the Asian continent. The opposite is the case during the winter due to the lagged response of cooling land surface air temperatures.

PC2 leads the phase of solar decline at the summer solstice and reaches its maximum and minimum around the fourth pentad of April and fourth pentad of October(Figure 1(f)),concurrently with the equinoctial asymmetric mode between the two transitional seasons in tropical regions(Wang et al.2008).Here,we name this mode the equinoctial mode of the EAM.The Tibetan Plateau sensible heating pump in spring has been considered as the dominant driver of this equinoctial mode(Wu 2004;Duan et al.2012). The low-level cyclonic circulation over the east of the Tibetan Plateau and rainfall in South China are mainly caused by heating-induced atmospheric convergences under the effect of the Tibetan Plateau mechanical block.Chang et al. (2005) attributed it to a combination of asymmetric wind-terrain interaction and low-level divergence asymmetry, which are both induced by different land-ocean thermal inertia over the Asia-Pacific region.

Figure 1.MV-EOF analyses of the first harmonic component in the combined SLP,850-hPa wind,and rainfall fields over the East Asia-Pacific region during 1979-2016:first and second modes in(a,b)SLP(shading)and 850-hPa winds(units:m s?1)and(c,d)rainfall(units:mm d?1);and(e,f)the normalized year-to-year PC1 and PC2 and their long-term average(black solid curve).

4. Phase-induced interseasonal rainfall anomalies

The seasonal march of the East Asian summer monsoon can be regarded as the phase transition from the equinoctial to the solstice mode during spring to summer.However, PC1 and PC2 show lead-lag phase variation with distinct amplitudes relative to their long-term means(the trend and decadal oscillation have been removed),indicating a distinct interannual variability of the annual cycle of the EAM during 1979-2016 (Figure 1(e,f)). The phase (amplitude) anomalies of PC1 and PC2 show a positive(negative)correlation,with a statistically significant correlation coefficient of+0.65(?0.51)at the greater than 95% confidence level (figure not shown). Therefore,a lead-lag variation of the equinoctial mode may cause a similar change in the solstitial mode.However,a stronger(weaker)equinoctial mode is usually followed by a weaker(stronger)solstitial mode.Such a relationship implies that the annual phase anomaly may result in a seasonal climate anomaly and their interseasonal linkage.

Based on a threshold of ±0.5 pentads of the phase anomaly, we select significant lead years (1984, 1986,1989, 1994, 2001, 2002, 2004, 2012, and 2015) and lag years (1981, 1983, 1987, 1993, 1996, 1998, 2003, 2005,2007, and 2011) for both PC1 and PC2, and verify our hypothesis through composite analyses (Figure 2). The PC1 anomaly in lead (lag) years reaches its positive maximum in spring (fall), while the PC2 anomaly reaches its positive maximum in winter(summer).This implies that the phase anomaly can cause the seesaw variation in interseasonal climate anomalies, which provides a potential basis for the across-seasonal prediction of the EAM.

The rainfall anomaly in a given region is a common focus in seasonal climate prediction. However, whether the phase-induced seesaw relationship can be reproduced by the original rainfall anomalous fields needs to be determined.Correlation analysis shows there are significant negative correlations of rainfall anomalies between spring and fall in North China, and the significant opposite changes in rainfall anomalies between the winter and summer are mainly located in South China(Figure 3). Of note is that North and South China are respectively centered in the maximum loadings of the solstitial and equinoctial modes(Figure 1(c,d),and therefore the negative correlation of regional rainfall anomalies reflects their phase change-induced seasonal rainfall relationship (Figure 2). Figure 4 shows the temporal evolution of the annual cycle component of rainfall in North and South China(regions A and B)corresponding to the lead- and lag-phase years of PC1 and PC2. It is clear that opposite rainfall anomalies appear in spring and fall in North China(Figure 4(b)),but generally,opposite rainfall anomalies between winter and summer are found in South China(Figure 4(d)).Therefore,the phase anomaly in the annual cycle of the EAM provides a potential two-season-ahead rainfall forecast in North and South China.Due to phase changes,the proportion of the variance of anomalous rainfall in North China is 0.31(0.35)in spring(fall),and that in South China is 0.61(0.27) in winter (summer) (figure not shown). Also,around 60% (55%) of years in South China (North China) with opposite interseasonal rainfall anomalies can be explained by the changes of the annual phase.

Figure 2.Annual cycle of PC1 and PC2 and their anomalies composited in the lead and lag-phase years:(a,c)lead and lag PC1 and PC2 relative to the climatology; (b, d) anomaly of PC1 and PC2 in lead and lag-phase years. The red (blue) line represents the lag(lead)year,and the black line represents the climatology.

SST anomalies have been considered as the most important factor for the seasonal forecasting of the EAM.We extract the years in quadrant α,β,ψ,and θ for regions A and B(Figure 5(a,b))to conduct a composite analysis of SST anomalies in the winter.The winter SST anomalies in the tropical Pacific show a significant ENSO-like mode.For the years of quadrant α and β (Figure 5(c,e)), the SST anomalies in the central East Pacific both show warm anomalies, but they are opposite in the Northwest Pacific Ocean east of Japan, implying the reversal of the rainfall anomaly between spring and fall in North China is mainly regulated by SST anomalies in the Northwest Pacific Ocean in winter. For South China, the change in phase of ENSO and the Indian Ocean Basin-wide(IOBW)in winter cause the reversal of the seasonal rainfall anomaly between winter and summer(Figure 5(d,f)).Previous studies have suggested that the warming of the eastern Indian Ocean/South China Sea can be caused by an increase in solar fluxes associated with remote EI Ni?o forcing(Klein,Soden,and Lau 1999;Lau and Nath 2003),which can influence precipitation in southeastern China by affecting the western North Pacific circulation (Wu,Zhou, and Tim 2009). Notably, there is more rainfall in South China in winter when EI Ni?o events (red dots)occur(Figure 5(b)).

Figure 3.Correlation coefficients of raw seasonal rainfall anomalies between(a)spring and fall,and(b)winter and summer.The redframed region indicates the North and South China region discussed in the current study.The stippled points indicate a statistically significant correlation passing the 95%confidence level.

Figure 4.Annual cycle of rainfall and its anomaly corresponding to the lead and lag years in(a)North China in region A and(c)South China in region B.(b,d)Rainfall anomaly of regions A and B in lead and lag-phase years.

5. Summary

Based on the first harmonic components of SLP,850-hPa winds, and rainfall during 1979-2016, we have investigated the annual cycle of the EAM and its interannual variability.The results show that the annual cycle of the EAM can be decomposed into the equinoctial and solstitial modes. The equinoctial mode shows the zonal contrast between the low-level cyclonic low over the east of the Tibetan Plateau and the western Pacific anticyclonic high over the east of the Philippines.It prevails in the spring rainy season in South China and reaches its peak in April.The solstitial mode shows a zonal pressure gradient contrast between the mid-high-latitude lowlevel monsoon low and the western North Pacific subtropical high, reaching its peak in July and dominating the East Asian summer monsoon.

Figure 5.Scatterplots of rainfall anomalies between(a)spring and fall in North China(region A in Figure 3)and(b)winter and summer in South China(region B in Figure 3).The solid black line is the regression line,and red(blue)dots represent EI Ni?o(La Ni?a)years.(c-f)Composite SST anomalies in winter corresponding to years in the α, ψ, β and θ quadrant. The stippled points indicate a statistically significant correlation passing the 95%confidence level.

The interannual variability of the annual cycle of the EAM is characterized by the phase and amplitude anomalies of these two modes.The lead-lag annual phase of the EAM results in the negative correlation of rainfall anomalies in North(South)China between spring and fall(winter and summer).The phases of ENSO and the IBOW in winter give rise to a reversed relationship of the seasonal rainfall in winter and summer in South China.The SST anomaly in the midlatitude Northwest Pacific Ocean mainly gives rise to the opposite change in the precipitation anomaly in North China between spring and fall.

The current study highlights the possibility for the prediction of rainfall two seasons in advance based on the impact of the annual phase of the EAM. However,there are other external forcings of the EAM,such as soil moisture,Eurasian snow cover,sea ice,and so on.These factors can affect the annual phase of the EAM and result in the seasonal rainfall anomalies in China.In addition,the physical mechanism behind the reversed change in interseasonal climate anomalies needs to be further studied.

Disclosure statement

No potential conflict of interest was reported by the authors.

Funding

This work was jointly supported by the National Natural Science Foundation of China [grant numbers 41830969 and 41775052], the National Key R&D Program [grant number 2018YFC1505904] and the Basic Scientific Research and Operation Foundation of CAMS [2018Z006 and 2018Y003]. It was also supported by the Jiangsu Collaborative Innovation Center for Climate Change.