Intraseasonal variability of summer monsoon rainfall over the lower reaches of the Yangtze River basin

OUYANG Yu and LIU Fei

Earth System Modeling Center and Climate Dynamics Research Center,Nanjing University of Information Science&Technology,Nanjing,China

ABSTRACT This work investigates the boreal-summer intraseasonal variability (ISV) of the precipitation over the lower reaches of the Yangtze River basin (LYRB) during 1979-2016, based on daily Climate Prediction Center global precipitation data.The ISV of the summer monsoon rainfall over the LYRB is mainly dominated by the lower-frequency 12-20-day variability and the higher-frequency 8-12-day variability. The lower-frequency variability is found to be related to the northwestwardpropagating quasi-biweekly oscillation (QBWO) over the western North Pacific spanning the South China Sea(SCS) and Philippine Sea, while the higher-frequency variability is related to the southeastward propagating midlatitude wave train (MLWT). Moreover, not each active QBWO(MLWT) in the SCS (East Asia) can generate ISV components of the precipitation anomaly over the LYRB. The QBWO can change the rainfall significantly with the modulation of mean state precipitation, while the quasi-11-day mode mainly depends on the intensity of the MLWT rather than the mean precipitation change.These findings should enrich our understanding of the ISV of the East Asian summer monsoon and improve its predictability.

KEYWORDS Yangtze River basin;intraseasonal variability;quasi-biweekly oscillation;midlatitude wave train;mean state change

1. Introduction

The Yangtze River basin (YRB), located in one of the world’s most well-known monsoon regions, is a central area of subtropical East Asia (EA). Heavy flooding often occurs over the YRB during boreal summer, causing serious damage to the economic and social development of EA(Zhu et al.2003;Mao and Chan 2005).Lowfrequency oscillation(LFO)has significant effects on the weather and climate in China at the seasonal time scale(Li 2014),through affecting the outbreak and disruption of the Asian monsoon, which is closely related to lowfrequency meteorological disasters affecting the YRB area, such as persistent drought and flooding, snow and ice, and heat waves (Ding and Wang 2008; Wang et al. 2014; Gao et al. 2017; Ren et al. 2018). Therefore,studying the intraseasonal variability (ISV) of East Asian summer monsoon rainfall provides a useful guidance for extended-range (10-30 days) forecasting of summer monsoon rainfall over the lower reaches of the YRB(LYRB).

There is significant ISV over the LYRB region, which has been confirmed by many studies (Wang and Xu 1997; Mao and Wu 2006; Ding and Wang 2008; Liang and Ding 2012; Yin, Zhu, and Yuan 2014). The climatological intraseasonal oscillation was proposed to be very significant over the LYRB (Wang and Xu 1997). The 10-90-day climatic ISO was found to be related to the advancement and maintenance of main rain belts in East China(Ding and Wang 2008;Wang and Ding 2008).The influences of northward-propagating 25-90-day ISO on East China summer rainfall was also investigated (Chen et al. 2014). The probability of extreme rainfall over southeastern China was found to be increased by about 35%-45% in the wet phase of the first boreal summer intraseasonal oscillation mode(Ren et al.2018).The relationship between the north-south antiphase distribution of rainfall during the mei-yu period and the quasi-biweekly oscillation (QBWO) in the atmosphere has been shown to be significant (Yin, Zhu, and Yuan 2014).

Furthermore, the physical mechanism of the ISV on the precipitation over the LYRB has been documented.During boreal summer, the westward-propagating QBWO can be considered as equatorial Rossby waves coupled with convection (Kikuchi and Wang 2009;Ortega et al. 2017). As reviewed by Yang, Wang, and Bao (2010), there is a QBW mode and a 21-30-day mode of the ISV of rainfall over the LYRB. For the QBW mode,the extreme wet phase over the LYRB is primarily initiated by a midlatitude southeastward-propagating jet stream and enhanced by northwestward movement of a low-level anticyclonic anomaly. For the 21-30-day mode, the extreme wet phase is primarily associated with the westward extension of the western North Pacific (WNP) subtropical high and strengthened by an upper-level trough anomaly moving from Lake Baikal to the far east of Russia.Chen et al.(2014)investigated the influences of the northwestward-propagating QBWO on East China summer rainfall.The QBWO propagates from the western tropical Pacific northwestward into the YRB.Enhanced rainfall over East China results directly from enhanced convection propagating northward from the WNP-South China Sea (SCS), strengthens northward water vapor transport, and intensifies upper-level divergence.

In contrast to the QBW variability, the influence of the midlatitude wave train (MLWT, quasi-11-day period) variability over the LYRB has rarely been studied.By analyzing the effect of the QBWO (8-21-day period)on heat waves over the YRB region, Gao et al. (2017)found three types of QBW modes, one of which was associated with an eastward/southeastward migrating wave train from eastern Europe to the WNP in the upper troposphere. When a significant low-level anticyclonic anomaly associated with the QBWO appears over the YRB, temperatures rise sharply due to adiabatic heating caused by subsidence and to enhanced downward solar radiation caused by reduced cloud cover.

Some previous studies on the ISV rainfall over the LYRB have either been case studies (with a focus on 1991 and 1998) or based on short data records. Yang,Wang, and Bao (2010) researched the QBW and 21-30-day modes of summer monsoon rainfall over the LYRB in which a dataset of 24 years (1979-2002) was used.However, the ISV of rainfall of the LYRB region features interannual change (Yin and Wang 2011; Chen et al.2014), and whether or not each QBWO or MLWT can significantly change the rainfall is uncertain despite the large volume of literature concerning the ISV over the LYRB.

Following the study of Yang,Wang,and Bao(2010),we further explore the effects of the QBW and quasi-11-day modes over the LYRB during a longer period(1979-2016),and investigate the relationship between the LFO and mean state change on the ISV.This paper is organized as follows:The data and methods are described in section 2.In section 3,we describe the dominant periods and evolution of the ISV over the LYRB.In section 4,two ISV modes over the LYRB are investigated in detail. The influence of mean state change and the intensity of the LFO on the ISV of rainfall over the LYRB are also presented.Finally,conclusions and discussion are summarized in section 5.

2. Data and methods

We use the daily Climate Prediction Center global precipitation data with a high resolution of 0.5° × 0.5°provided by the National Oceanic and Atmospheric Administration (NOAA) (Xie et al. 2007). Daily outgoing longwave radiation (OLR) data (resolution: 2.5° × 2.5°),also from the NOAA(Liebmann and Smith 1996),are also used in this study, as a proxy for organized deep convection in the tropics.Daily circulation variables,including wind and geopotential height (GPH), are extracted from the National Centers for Environmental Prediction reanalysis data (Kalnay et al. 1996). The period of each dataset covers 38 years(1979-2016)from May to August(MJJA).

The methods used in this study include the Lanczos filter (Duchon 1979), power spectral analysis, Student’st-test, and composite analysis. To extract the ISV component, an 8-80-day bandpass filtering is applied to remove the climatology, seasonal change, and synoptic fluctuation. The dominant ISV periodicity over the LYRB can be identified by calculating the average of individual power spectra during MJJA for the 38 summers with significance testing. Composite analysis is used to show the spatial structure and temporal evolution of ISV modes, and significance is tested via the Student’st-test.

3. Dominant ISV over the LYRB

To obtain the dominant periods of the precipitation over the LYRB, we analyze the mean power spectra of the 38 summer seasons (MJJA) of area-averaged daily rainfall within the LYRB region (29°-34°N, 115°-120°E)from 1979 to 2016. As shown in Figure 1, three bands are clearly identifiable during MJJA in the LYRB region, which respectively occur at 20-40 days (peaking on day 25), 12-20 days (peaking on day 15), and 8-12 days (peaking on day 11). However, only the latter two modes can be well separated based on the 99% confidence level. In addition, we calculate the percentage variance of the three ISV rainfall components against the total ISV (8-80 days) over the LYRB region. The contributions are 26.5% from the 8-12-day mode, 21.2% from the 21-40-day mode,and 13.4% from the 21-40-day mode. The 21-40-day mode plays a lesser role than the other two modes and does not pass the significance test. Therefore, in this study, we focus on the 12-20-day and 8-12-day variabilities of summer rainfall over the LYRB.

Figure 1.Mean power spectra of the rainfall ISV component over the LYRB region based on 38 summers(MJJA),with a Markov red noise spectrum(red line)and a priori 99%confidence bound(blue line).The x abscissa has been rescaled into the natural logarithm of the frequency,and the corresponding y abscissa is multiplied by frequency.

The evolution of the significant ISV of rainfall over the LYRB during boreal summer is described in this section by compositing the low-frequency strong precipitation events. The criterion of case selection is one standard deviation of the time series of area-averaged bandpassfiltered rainfall over the LYRB. Based on this criterion,a total of 167 QBW and 244 quasi-11-day events are selected from the 38 years (1979-2016). Day(0) is defined as the day with the maximum rainfall anomaly over the LYRB region.

Figure 2 presents successive composite maps of the QBW and quasi-11-day modes of OLR, 200-hPa wind anomalies, and 200-hPa GPH, from day(?8) to day(0).For the QBW mode shown in Figure 2(a), the most remarkable feature is that the low-level northwestward propagation of an anticyclonic circulation anomaly is coupled with a positive OLR anomaly extending from the WNP to the SCS,which is associated with the Rossby wave on the QBW time scale proposed in previous studies(Kikuchi and Wang 2009).The upper-level anticyclonic anomaly, which cannot be defined as a wave train,moves southeastward from a midlatitude area to the LYRB. The vertical baroclinic configuration, which has upper-level atmospheric divergence and low-level convergence, enhances the local upward motion over the core region, and the southwesterly wind of low-level cyclonic activity conveys more moisture to the LYRB; as a result, the precipitation of the LYRB reaches its maximum at day(0)(not shown).

For the quasi-11-day mode (Figure 2(b)), from day(?8)to day(0),there is a midlatitude GPH wave train propagating southeastward, which originates from the Ural Mountains, migrates through the West Siberian Plain and southern Lake Baikal, before then arriving in the LYRB region and finally moving toward the WNP.Meanwhile,the OLR migrates southeastward,transforming from a positive to negative phase over the LYRB region in the low level.The rainfall moves southeastward to the LYRB at day(0)and reaches about 5 mm d?1(not shown). Furthermore, the quasi-11-day variability of summer rainfall is unrelated to the synoptic-scale disturbance.The period of the MLWT is about 11 days,while the synoptic-scale disturbance is only 3-8 days.Also,the MLWT is a large-scale spatial circulation rather than a local disturbance.In summary,the ISV of the summer monsoon rainfall over the LYRB is mainly dominated by the QBW variability and quasi-11-day variability.The lower-frequency variability is found to be related to the northwestward-propagating QBWO over the WNP spanning the SCS and Philippine Sea, while the quasi-11-day variability is related to the southeastward-propagating MLWT.

Figure 2.Temporal evolution of composite 200-hPa wind anomalies(vectors;units:m s?1),OLR anomalies(shading;units:W m?2),and GPH anomalies(contours;units:m)for the(a)QBW mode and(b)quasi-11-day mode.Cyan,green,and blue boxes denote the LYRB(29°-34°N,115°-120°E),SCS(14°-24°N,110°-130°E),and EA(30°-40°N,116°-132°E),respectively.Only locally statistically significant(>95%)variables are shown.

Figure 3.Composites of rainfall anomalies(shading over land;units:mm d?1),OLR anomalies(shading over ocean;units:W m?2),and 200-hPa GPH anomalies(contours;units:m)at day(0)for(a,b)SP events and(c,d)WP events in the(a,c)QBW mode and(b,d)quasi-11-day mode.Only locally statistically significant(>95%)variables are shown.Solid(dashed)contours represent positive(negative)GPH anomalies.

4. Different influences of the QBW and quasi-11-day mode on LYRB rainfall

In this section,we explore the important factors of the QBW or quasi-11-day modes that can influence the LYRB rainfall.As with the criterion for defining heavy ISV rainfall events over the LYRB in section 3.2,we select 190 QBW cases and 271 quasi-11-day cases by defining the OLR intensity as the QBW index in its extreme wet phase over the SCS(14°-24°N,110°-130°E)and the GPH as the quasi-11-day index over EA (30°-40°N, 116°-132°E). The synchronous ISV component of the precipitation anomaly of the LYRB is a threshold to divide these ISV events into two categories.If the rainfall value over the LYRB in the QBW(quasi-11-day)extreme dry(wet)phase over the SCS(EA)exceeds the average of ISV precipitation anomaly of all QBW (quasi-11-day) events over the LYRB, this event is selected as a QBW (quasi-11-day) event that can result in a strong precipitation (SP)anomaly in the LYRB;otherwise, it is a weak precipitation(WP)anomaly case.A total of 190 QBW and 271 quasi-11-day events are found according to the definition,and the QBW events are divided into 89 (46.8%) SP and 101 WP anomaly events, while the quasi-11 days are divided into 129(47.6%)SP and 142 WP anomaly events.

By applying composite analysis to the selected events, different spatial features between the SP and WP anomaly events over the two regions are presented in Figure 3,separately.Day(0)is defined as the day with the maximum OLR (GPH) anomaly over the SCS (EA)region for QBW (quasi-11-day) events. At day(0),a negative low-level OLR anomaly and an upper-level anticyclonic anomaly migrate to the LYRB when the maximum rainfall anomaly reaches 4 mm d?1. By comparison, with the same QBW signal moving from the Philippine Sea to the SCS, the upper-level high for the WP is weaker than that in the SP anomaly events, and there is almost no precipitation over the LYRB(Figure 3(c)).Similarly,with the same MLWT propagating from the Eastern European Plain to the LYRB in the upper troposphere, the low-level OLR wave train also migrates southeastward for both SP and WP anomaly events(Figure 3(b)).However,it is obvious that rainfall is different between these two types of events, which is about 4 mm d?1for the SP anomaly events, while there is no rainfall over the LYRB for the WP anomaly events.

To explore whether the intensity of the two ISVs or other factors can influence the ISV of rainfall over the LYRB,the correlation coefficients of the OLR(GPH) over the SCS (EA) against the intensity of rainfall over the LYRB are calculated for the QBW (quasi-11-day) mode(Figure 4). The relationship between the mean state precipitation (MPR), by averaging the rainfall from day(?15)to day(15),and the ISV of rainfall over the LYRB is also analyzed. For the QBW mode (Figure 4(a,c)), the correlation coefficient between the precipitation ISV and the intensity of the QBWO is only 0.04, while the MPR has a significant correlation of 0.32 with the ISV of rainfall. For the quasi-11-day mode, the intensity of the MLWT makes a significant difference to the ISV of rainfall over the LYRB,while the effect of the MPR is weaker than the intensity of the MLWT.

In summary, not each active QBWO over the SCS or MLWT over EA can change the precipitation significantly over the LYRB.For the QBW mode,the interaction of the QBWO with heavy MPR may induce a strong ISV of rainfall over the LYRB,while the quasi-11-day mode is mainly associated with the intensity of the MLWT over EA.

Figure 4.Relationships between rainfall and impact factors over the LYRB:scatterplots of rainfall anomalies(units:mm d?1)and(a)OLR(units:W m?2)over the SCS,(b)GPH(units:m)over EA,and(c,d)mean precipitation(units:mm d?1)over the LYRB for the(c)QBW mode and(d)MLWT.

5. Summary and discussion

Subseasonal prediction (2-4 weeks) has in recent decades become an important research area for operational weather forecasting and climate prediction. LFO is a major predictability source for subseasonal forecasts over the LYRB region. To improve the quality of lowfrequency prediction over the LYRB region, two dominant ISV modes are identified by composite analysis in this study;and the genesis,evolution,and the influence of mean state change are investigated. The ISV of the summer monsoon rainfall over the LYRB displays quasi-15-day and quasi-11-day variability(Figure 1).The quasi-11-day mode is characterized by a northwestwardpropagating QBWO over the WNP spanning the SCS and Philippine Sea, while the QBW variability is related to the southeastward-propagating MLWT (Figure 2). In addition,not each active QBWO(MLWT)in the SCS(EA)can induce precipitation over the LYRB(Figure 3).For the QBW mode,the occurrence of low-frequency heavy rainfall needs the cooperation of the tropical QBWO and MPR, while the quasi-11-day mode is associated with the intensity of the MLWT rather than the mean state change(Figure 4).

The findings of this work should enrich our understanding of the ISV of rainfall and the impact factors of the two ISV modes on the ISV of rainfall over the LYRB.These can help to predict whether or not the QBWO or MLWT can influence the ISV of rainfall of the LYRB,which is crucial for improving the extended-range predictability of East Asian summer monsoon rainfall over the LYRB.

Disclosure statement

No potential conflict of interest was reported by the authors.

Funding

This work was supported by the National Natural Science Foundation of China [grant number 41420104002] and the Natural Science Foundation of Jiangsu Province [grant numbers BK20150907 and 14KJA170002].