Attribution of Persistent Precipitation in the Yangtze–Huaihe River Basin during February 2019

Zhixuan WANG, Jilin SUN*,2, Jiancheng WU, Fangyue NING, and Weiqi CHEN

1College of Oceanic and Atmospheric Sciences, Ocean University of China, Qingdao 266100, China

2Key laboratory of Physical Oceanography, Ocean University of China, Qingdao 266100, China

3Zhangzhou Meteorological Observatory, Zhangzhou Meteorological Bureau, Zhangzhou 363000, China

4Huaan Meteorological Observatory, Huaan Meteorological Bureau, Zhangzhou 363800, China

(Received 24 April 2020; revised 27 July 2020; accepted 10 August 2020)

ABSTRACT In February 2019, a month-long persistent precipitation event occurred in the Yangtze—Huaihe River basin. The geopotential height field that affected the duration of this frontal rainfall was divided into a high-latitude part and a lowlatitude part for analysis. In the high-latitude part, a two-wave structure led to quasi-stationary circulation, and the change of both the blocking high pressure and Arctic Oscillation phase caused cold air to invade South China continuously and changed the frontal position. In mid-to-low latitudes, the persistent precipitation showed quasi-biweekly oscillation characteristics. The so-called “subtropical high—precipitation—anticyclone” (SHPA) feedback mechanism blocked the circulation systems in the mid-to-low latitudes and provided a continuous supply of water vapor for precipitation. As for the effect of sea surface temperature, the western North Pacific anomalous anticyclone stimulated by El Ni?o strengthened the intensity of the southerly wind and provided support for the redevelopment of the anticyclone system in the SHPA feedback mechanism. The sea surface temperature anomaly in the South China Sea provided sensible heating for precipitation, and convergent rising airflow was conducive to the occurrence of precipitation. Additionally, the SHPA mechanism provides a reliable basis for the prediction of persistent precipitation in winter in the mid-to-low latitudes.

Key words: persistent precipitation, feedback mechanism, latent heat

1. Introduction

In February 2019, a month-long persistent rainfall event occurred in the Yangtze—Huaihe River basin (YHRB),China. During this period, the daily precipitation was substantially higher than in other years and the precipitation anomaly intensity in the central area of the rainfall reached a maximum of 6 mm d-1(Fig. 1a). Additionally, the number of rainy days and the total precipitation in February 2019 reached their highest level for 20 years (Fig. 1b). Persistent precipitation events are likely to cause adverse human effects, including damage to property. For example, persistent precipitation in the Yangtze River basin in 1998 caused around 3000 deaths and substantial economic losses of roughly 260 billion yuan (Ding and Hu, 2003).

In recent years, winter rainfall in the Yangtze River basin has become an important research topic (Zhou and Wu, 2010; Zhang et al., 2011). Wang and Feng (2011)found that winter precipitation over the YHRB shows large interannual variance, and Li et al. (2015) found that it may increase owing to warming in the tropical Indian Ocean.Jiang et al. (2011) investigated the transport of water vapor related to persistent precipitation over the YHRB in 2007.Some studies have shown that when rainy areas offset to the northern YHRB, precipitation will decrease in the Yangtze River basin (Wu et al., 2006). Furthermore, some researchers have used dual-polarization radar to analyze the microphysical structure and evolution characteristics of typical persistent precipitation (Shusse et al., 2009; Oue et al., 2010,2011; Kennedy and Rutledge, 2011).

Fig. 1. (a) Monthly mean precipitation distribution in the YHRB in February 2019 (contour interval: 1 mm d-1) and the anomaly distribution (shading; units: mm d-1). The red box indicates the study area. (b) Precipitation anomaly of YHRB in February during 2000—19 (histogram, units: mm d-1; left-hand y-axis) and the number of rainy days during this period (line,right-hand y-axis).

Previous studies have also revealed that winter precipitation bears a close relationship with the geopotential height field, especially the Arctic Oscillation (AO) (Shuai et al.,2010; Zhang et al., 2014). The South Asian high plays an essential role in causing persistent precipitation over South China (Qian et al., 2002; Wang et al., 2007). Moreover,some researchers have suggested that the subtropical high could have a significant influence on persistent extreme precipitation in the Yangtze River basin (Chen and Zhai,2015). An anomalous subtropical high directly determines the path of water vapor transport (Mao et al., 2010; Ren et al., 2013). Yu et al. (2004) pointed out that cooling of the tropopause in East Asia may lead to the subtropical high moving south, thus resulting in floods in Southeast China. He et al. (2001) found a high degree of correlation between the parameters of the subtropical high and precipitation in the Yangtze River basin.

Sea surface temperature anomalies (SSTAs) can exert significant impacts on circulation and rainfall at multiple time scales (Lau and Nath, 2009; Zhou et al., 2009). Nitta (1987)showed that the SSTA in the western Pacific warm pool area can affect the convective activity around the Philippines, thus modulating the intensity of the subtropical high pressure. A warm SSTA in the Indian Ocean can excite a warm Kelvin wave, which forms an anomalous anticyclonic circulation in the Northwest Pacific and thus enhances the strength of the subtropical high (Watanabe and Jin,2002; Yoo et al., 2006; Xie et al., 2009, 2010).

Recent studies have shown that the nonlinear interaction between El Ni?o and the western Pacific warm pool is an essential mechanism for the formation and maintenance of anticyclones in the Northwest Pacific (Stuecker et al.,2013; Ren et al., 2016). El Ni?o—Southern Oscillation(ENSO) affects the variations of precipitation patterns (Dai and Wigley, 2000; Sun et al., 2013). For instance, the interannual variability of wintertime rainfall over South China is significantly modulated by ENSO (Zhou et al., 2010; Yuan et al., 2014; Ren et al., 2013). From 1979/80 to 2012/13, the winter precipitation in South China was mainly controlled by the quasi-biweekly oscillation (QBWO) (Tong and Yao,2018). Additionally, some studies have shown that the QBWO of winter precipitation in South China could be associated with ENSO (Zhou et al., 2010; Yuan et al., 2014). During 1953—73, winter precipitation in the YHRB was often lower (higher) in La Ni?a (El Ni?o) years than in normal years (Chen et al., 2014). The relationship between winter precipitation in South China and ENSO weakened during 1974—94 (Chen et al., 2014).

Although considerable progress has been made in research on winter precipitation in the YHRB, studies on persistent precipitation in the YHRB have mostly focused on the summer season (Chen and Zhai, 2014; Wan et al., 2017;Li et al., 2018). This is because the northerly wind, which generally comprises cold and dry airflow and is not conducive to the production of persistent precipitation, prevails in the YHRB in February (Chen et al., 2019; Wang and Feng,2011). Thus, persistent precipitation in winter is rare in this region, and so its underlying mechanism remains unclear. In this paper, we take the month-long persistent precipitation event that took place in the YHRB in February 2019 as our research object to address the following questions: (1)Against the background of cold—dry winter monsoon circulation, where and how did the ample water vapor that led to the February 2019 persistent precipitation event originate?(2) Why was the duration of the precipitation so long? (3)Which factor (i.e., synoptic-scale baroclinic waves or extratropical cyclones and fronts) induced the persistent precipitation, and what was the physical linkage between these two factors? (4) What role did El Ni?o play in this persistent rainfall? (5) Did the SSTA of China’s coastal waters have an impact on the precipitation? We attempt to answer the above questions by analyzing the geopotential height field and external forcing field (i.e., SSTAs).

The remainder of this paper is organized as follows:The data and methods are described in section 2. The type of precipitation is analyzed in section 3. Section 4 introduces the 500-hPa geopotential height field, which is then divided into two parts (mid-to-high latitude and mid-to-low latitude) for analysis in section 5. Section 6 studies the influence of SSTAs on the rainfall. And finally, section 7 provides a discussion and a summary, including implications for the forecasting of persistent disastrous weather.

2. Data and methods

2.1. Data

Monthly/daily mean geopotential height, air temperature, wind speed, and outgoing longwave radiation (OLR)were obtained from the NCEP—NCAR reanalysis datasets,with 17 pressure levels in the vertical direction, for the same period (Kalnay et al., 1996; National Oceanic and Atmospheric Administration NOAA/OAR/ESRL PSD, 1996).The precipitation dataset is part of a product suite from the NOAA Climate Prediction Center (CPC) Unified Precipitation Project (Chen et al., 2008).

Additionally, monthly SST data with a 2° spatial resolution were obtained from NOAA’s ERSST dataset (Huang et al., 2015), and daily SST from NOAA’s OISST dataset(also known as Reynolds’ SST), which is a series of global analysis products on a 0.25° grid (Reynolds et al., 2007).

2.2. Methods

Harmonic analysis was performed by a combination of numerical and graphical procedures adapted from the methods (Daniel, 2011). With the development of the study of atmospheric circulation and ultra-longwave systems, it became recognized that using the average or sliding average can simplify the calculation of the filtering method;however, this is a form of low-pass filtering, which has strong limitations. Harmonic analysis is a type of function filtering that is not only low-pass filtering but also high-pass or bandpass filtering in the study of large-scale circulation.The harmonics are expressed in the form Ancos(nθ－φn),whereAnis the amplitude of thenth harmonic and φnis the phase angle. The first maximum or ridge of thenth harmonic is thus found at θ = φn/ndegrees of longitude.

The Butterworth filter, which is a maximally flat magnitude filter, has a flat frequency response in the bandpass filtering (Butterworth, 1930). In this study, we chose the bandpass Butterworth filter for our analysis.

Additionally, we also used some conventional statistical methods [empirical orthogonal function (EOF) analysis,correlation analysis, composite analysis,t-test] and combined the equation derivation to study the cause of this persistent precipitation.

3. Type of precipitation

The major types of rainfall include relief rainfall, convection rainfall, and frontal rainfall (Selase et al., 2015). Relief rainfall, also known as orographic rainfall, occurs in areas where the slope of the terrain increases (Selase et al., 2015).Convection rainfall is heavy but short in duration and always occurs in summer (Selase et al., 2015). The YHRB area is located on the Yangtze Plain. The duration of the persistent precipitation in February 2019 was long. Therefore,frontal waves may have been the main factor that produced this persistent precipitation. The frontogenesis function can reasonably be used to explain frontal rainfall (Cohen and Schultz, 2005) because the terms of the frontogenesis function can reveal the development and structure of fronts. The frontogenesis function can be written as follows (Petterssen,1936; Miller, 1948):

where |?hθ| refers to changes in the magnitude of the thermal gradient;Fh＞ 0 indicates frontogenesis andFh＜ 0 indicates frontolysis. Conceptually, frontogenesis is a local change in the horizontal temperature gradient near an existing front, baroclinic zone, or feature as it moves. Two main processes (parameters) make significant separate contributions to frontogenesis: divergence and deformation. When convergence is oriented nearly perpendicular to a thermal gradient and a confluent wind field is applied to the thermal gradient, frontogenesis (Fh＞ 0) occurs. Frontogenesis is common in the developing and mature stages of a low-pressure system when precipitation rates increase. The front can provide lift, instability, and moisture for precipitation(Meng et al., 2012). Frontolysis (Fh＜ 0) is the opposite.

According to the above formula, we calculated the frontogenesis function for February 2019 at latitudes of 25°N,30°N and 35°N. The results are shown in Fig. 2. As shown in Fig. 2a, the frontogenesis function is mainly positive (Fh＞ 0) in early February 2019, while it is mainly negative (Fh＜ 0) in late February 2019. As shown in Fig. 2b, although frontolysis occasionally appears at 30°N, frontogenesis is still predominant. The precipitation distribution in Fig. 2c is precisely the opposite of that in Fig. 2a, and frontogenesis mainly occurred in late February 2019. The daily precipitation distribution (Fig. 3d) combined with the frontogenesis function (Figs. 2a—c) shows that the precipitation process in February 2019 can be divided into two processes: Process 1,during 1—13 February 2019, in which the frontal position was located to the south of the YHRB; and Process 2, during 14—28 February 2019, in which the frontal position was located to the north of the YHRB. The daily precipitation distribution in Fig. 2d also reflects that the persistent precipitation in February 2019 had a periodicity at intraseasonal time scales.

Therefore, the front mainly appeared in the southern YHRB in early February and stabilized in the northern YHRB in late February. The front was mainly caused by the convergence of cold and warm air. A warm air mass rises to the condensation height over cold advection, and if it reaches the level of free convection, convective clouds are formed and may produce precipitation (Mazon and Pino,2013). Therefore, we calculated the temperature advection of 925 hPa (corresponding to the distribution of the frontal function). The results are shown in Fig. 3a. As shown in the figure, the cold advection invades the southern YHRB in early February and moves the front south. Meanwhile, in mid and late February, the warm advection intensity strengthens and blocks the cold air mass in the northern YHRB. Thus, this caused the front to move towards the northern YHRB.

Fig. 2. Distribution of frontogenesis function at (a) 25°N, (b) 30°N and (c) 35°N, and the (d) daily regional total precipitation of the YHRB in February 2019.

The temperature advection at 700 hPa (Fig. 3b) shows that the precipitation in the YHRB was mainly controlled by the anomalously cold airflow. This precipitation distribution(warm advection at 925 hPa and cold advection at 700 hPa)provided unstable stratification for precipitation. Based on the criterion of convective instability (?θse/?Z)＜0 and the results of dθsein Fig. 3c, the mid-to-low layer of the troposphere meets the conditions of convective instability. This situation is not only related to the convergence of cold and warm advection but also means that the low layer of the troposphere has abundant water vapor. The strong warm advection at 925 hPa complements the consumption of the convective available potential energy during persistent precipitation, resulting in the persistence of convective instability in February 2019. As for the frontal waves, the uplift of the warm air mass triggers the release of the convective instability energy, thus leading to frontal rainfall.

Therefore, the persistent precipitation in February 2019 can be identified as frontal rainfall. The long-term intersection between cold and warm air masses caused uplift airflow, the condensation of water vapor, and the instability of atmospheric stratification, which led to frontal rainfall. Additionally, at different stages of this persistent precipitation,the intensity of both cold advection and warm advection cannot be separated from the influence of the large-scale atmospheric circulation field. However, the causes of this rainfall need to be further analyzed using the geopotential height field.

Fig. 3. Latitude—time cross section of the temperature advection at (a) 925 hPa and (b) 700 hPa in February 2019(shading; units: 10-5 K s-1; lines represent 0 K s-1), and (c)isobaric—time cross section of dθse (shading; units: K), i.e.,dθse925hPa = θse925hPa - θse1000hPa) in February 2019.

Fig. 4. Monthly anomaly at 500-hPa geopotential height in February 2019 (shading; units: 10 gpm), and average of geopotential height at 500 hPa in February 2019 (contour interval: 10 gpm).

4. Geopotential height field

The geopotential height field at 500 hPa, which is the middle layer of the atmosphere, can reflect the changes of geopotential height in the lower or upper layers of the troposphere to varying degrees (Casola and Wallace, 2007). Therefore, we selected the geopotential height field at 500 hPa in February 2019 as our main research object. As shown in Fig. 4, the geopotential height field at high latitudes in the Eurasian region has a long span of latitude, and the geopotential height over the Bay of Bengal to the South China Sea is low. From eastern Lake Balkhash to Northeast China, northwest airflow steers the cold—dry air from the mid-to-high latitudes to South China. At low latitudes, the subtropical high is located near 25°N. The deep trough over the Caspian Sea continuously moves to the western Tibetan Plateau. Due to the the high terrain of the Tibetan Plateau, the deep trough divides into two parts. The northward part finally reaches North China, transporting cold air that contributes to the rainy weather, and the southward part moves along the southern edge of the Tibetan Plateau. After reaching the Bay of Bengal, the southward part enhances the development of the southern branch trough.

The 500-hPa anomaly of the geopotential height field in Fig. 4 shows that the intensity of the subtropical high in February 2019 was stronger than that in previous years. An anomalous high-pressure center exists in Europe, and the AO showed a positive phase in February 2019.

The anomalous geopotential height field may have caused the anomalous transportation of both water vapor and cold air during the persistent precipitation over the YHRB, and thus affected the position, intensity and duration of the front. We used the Hybrid Single-Particle Lagrangian Integrated Trajectory (HYSPLIT) model (Draxler and Hess, 1998) to calculate the backward trajectory of air parcels in February 2019. As most of the water vapor and dry-air transport was concentrated in the lower troposphere,we selected the height of 750 m MSL for our research. The backward trajectories of air parcels were output every six hours (0000, 0600, 1200 and 1800 UTC) between 27 January and 28 February 2019. The results are shown in Fig. 5.

In Fig. 5, there are two southerly paths (Channels 1 and 2) and one northerly path (Channel 3) during February 2019. The northerly path transports the cold—dry air mass to the YHRB through the ridge in the middle troposphere. The two southerly paths transported the water vapor to the YHRB in February 2019. The southwesterly path (Channel 1), which originates in the Bay of Bengal, may be affected by the activities of the India—Burma trough. Additionally,the path originating in South China (Channel 2) may be affected by the western Pacific subtropical high.

To further determine the source of the water vapor that caused the persistent precipitation in February 2019, we calculated the fractional contribution of water vapor (Qwv) carried by each considered channel. TheQwvcan be written as follows(Dirmeyer and Brubaker, 1999; Stohl and James, 2004):

whereqis the specific humidity,mis the number of clustered trajectories in each channel, andnis the total number of computed trajectories. The water vapor level may change during transportation owing to precipitation or evaporation; therefore, we calculated theQwvat the source and final locations of each patch.

Fig. 5. HYSPLIT Model trajectory frequencies (shading; 100× number of endpoints per grid square / number of trajectories)and cluster analysis of trajectories (black lines; the ratio of each channel is the percentage of trajectory clustering of air parcels) during 27 January 2019 to 28 February 2019(backward).

Fig. 6. Harmonic analysis parameters [sum of parameters of wavenumber (An) = 1-3] for the (a) monthly long-term mean (1981—2010) in February and (b) monthly mean in February 2019 of geopotential height at 500 hPa.

Channel 1 originates in the Bay of Bengal, with a water vapor contribution of 27.3% at the initial location.However, it decreases to 18.6% when it reaches the Tibetan Plateau because of the topography of this geographic feature. Channel 2, which provides the largest contribution ofQwvduring the persistent precipitation in February 2019,increases from 52.1% to 57.6% with transportation. Channel 3 originates in the Mongolian Plateau, which is the area of origin of the cold air affecting China. Thus, the cold and dry air mass was transported to the YHRB during February 2019 by Channel 3.

In short, the air parcel trajectory analysis (Fig. 5) shows the water vapor that led to the persistent precipitation event in February 2019 mainly came from the South China Sea,and little was transported from the Indian Ocean through the southern branch trough, and cold air invaded the YHRB from the Mongolian Plateau in February 2019.

Therefore, the circulation anomalies at low latitudes (subtropical high, south branch trough) may have affected the water vapor transport, while the anomalies of geopotential height at high latitudes (European high pressure, AO phase)affected the intensity of the cold air. These factors (subtropical high, southern branch trough, European high pressure,and AO phase) are likely to have been key in affecting this persistent precipitation anomaly, and are therefore worthy of separate analysis.

5. Factors affecting the duration of precipitation

Persistent precipitation is formed by the quasi-stationary wave system of the geopotential height field (Jin et al.,2013). As mentioned in section 2, harmonic analysis is commonly used to study the fluctuations or variations in a time series that has arisen from adding together a series of sine and cosine functions. Zhu and Lin (1992) noted that the geopotential height field is mainly a three-wave structure in winter in the YHRB. Thus, the first three waves (An= 1—3)were selected for harmonic analysis, and the total variance contribution of these first three waves (An= 1—3) reached 81.7% during February 2019.

As shown in Fig. 6a, the harmonic analysis of geopotential height is a three-wave structure in the long-term monthly mean. Meanwhile, Fig. 6b shows a significant twowave structure in February 2019. According to the formula for the Rossby wave velocity,

Cstands for Rossby wave velocity.uis westerly wind speed.Lis the length of waves, and β is Geostrophic parameters. Two-wave structure is easier for a quasi-stationary structure to form than a three-wave one. Thus, the geopotential height field at high latitudes in February 2019 formed quasistationary waves, thereby affecting the duration of precipitation.

Additionally, from Fig. 6b, it can be seen that, in Eurasia, the positive center of the two-wave structure corresponded to abnormally high pressure in Europe. Furthermore, the negative area was located in the Eurasian trough,which affected the transmission path and intensity of cold air. Based on the criterion of baroclinic instability(Emanuel, 2009), if the wavelength of the circulation in the mid-to-high latitudes exceeds 4000 km, baroclinic instability will occur in this area. From Fig. 6b, it can be seen that the wavelength of the two-wave structure exceeds the critical wavelength of the baroclinic instability. As a result,long waves in the middle and high latitudes will develop,thus strengthening the blocking situation in the high latitudes and bringing disturbances to the YHRB in the form of cyclones or synoptic eddies (Dacre et al., 2012).

According to the four key factors in the attribution of the persistent precipitation in February 2019 mentioned in section 4 (European high pressure, AO phase, subtropical high,and south branch trough), and the results of the harmonic analysis, we divided the geopotential height field into mid-tohigh latitudes and mid-to-low latitudes for analysis.

Fig. 7. Longitude—time cross section of the average geopotential height (contour interval: 1 gpm) and the anomaly geopotential height (shading; units: gpm) at 500 hPa over 50°—60°N.

5.1. Mid-to-high latitudes

During February 2019, the 500-hPa geopotential height in Europe was abnormally high (Fig. 2), and this anomalously high pressure provided the necessary cold air for precipitation in the YHRB. In the longitude—time cross section(Fig. 7), there are two visible high-pressure activities in February 2019. The first one occurred in early February and moved westward with time. The second one occurred in mid and late February and underwent a process of enhancement, followed by weakening, followed by re-enhancement,and the center of the anomalous high pressure remained at 10°E.

The existence of the two high-pressure activities in Europe may have provided the blocking situation that led to the persistent precipitation event in February 2019. From the perspective of nonlinear Rossby waves, blocking high pressure is a solitary wave that blocks the movement of long waves and makes the high-latitude weather situation quasi-stationary (Chopin and Malusk, 1980). Atmospheric blocking occurs when waviness in the jet stream causes congestion. In the mid-to-high latitudes, blocking situations cause the eastward meteorological systems to become quasi-stationary. In winter, a blocking situation always produces large and persistent cold waves (Chopin and Malusk,1980), which has an important impact on persistent precipitation (Higgins and Mo, 1997).

The blocking index, which was modified by Tibaldi and Molteni (1990), is associated with high-latitude blocks at 500 hPa. By calculating this index, we can further determine whether the anomalous activity of the European high pressure formed the blocking situation that led to the persistent precipitation event.

For each longitude, the southern 500-hPa geopotential height gradient (SGHG) and the northern 500-hPa geopotential height gradient (NGHG) were calculated as follows(Tibaldi and Molteni, 1990):

where φn=80+δ, φ0=60+δ , φs=40+δ , and δ = -5, 0, 5.(units:°)

If SGHG ＞ 0 m per degree of latitude or NGHG ＜ -10 m per degree of latitude, we can define this area as blocked at the given time, and the value of the blocking index is SGHG.

From the result in Fig. 8, the blocking high pressure is weaker in the early stage than the late stage. Additionally,the late process went through a process of enhancement, followed by collapse, followed by reconstruction, as shown in Fig. 7. In the late process, under the westerly wind-belts,blocking high pressure even westward-retreats instead of moving eastward. This movement path indicates that the existence of the blocking high pressure caused circulation to become quasi-stationary at high latitudes. This situation caused the cold air to continuously invade the YHRB.

According to the AO index of February 2019 provided by the CPC (https://www.cpc.ncep.noaa.gov/products/precip/CWlink/daily_ao_index/ao.shtml#curren), it was found that in early February, the AO index showed negative phases, and in late February it showed positive phases.Although the blocking high pressure was not strong in the early stage, the cold air still reached the southern part of the YHRB under the control of the negative phase of the AO index. As the AO index turned into positive phases in late February, the blocking high pressure became stronger.Then, the cold air was continuously maintained in the northern YHRB.

Fig. 8. Longitude—time cross section of the index of blocking high pressure in February 2019 (shading).

Therefore, in the high latitudes, the circulation field appeared to be quasi-stationary owing to the occurrence of ultra-long waves and the anomalously strong blocking system. This situation provided a steady cold air mass, leading to persistent precipitation in the YHRB. The AO phases and blocking system at high latitudes caused a sudden change of the intersection of cold and warm air, which then led to a change in the frontal position.

5.2. Mid-to-low latitudes

As can be seen from Fig. 2d, and as mentioned in section 3, precipitation in February 2019 may have had a periodicity at intraseasonal time scales. In order to investigate the dominant periodicities of precipitation, we calculated the power spectra of daily precipitation anomalies averaged over the YHRB during February 2019. The results showed a QBWO (10—20 days) with a peak at 12 days.

An EOF analysis was performed on the 10—20-day bandpass filtered precipitation anomalies over the YHRB to investigate the spatial patterns of the QBWO. The results show that the first two modes explain 47.5% and 29.3% of the variance, and the variance of the other modes are all less than 5%.

Therefore, the first two EOF modes can describe the most salient features of the 10—20-day bandpass filtered precipitation anomalies and the QBWO. Following Matthews(2000), the evolution of the QBWO can be divided into eight phases byZ(t) defined by the principal components(PCs) of the first two modes in the EOF, and the temporal evolution of the QBWO is an anticlockwise rotation:

As shown in Fig. 9, Phases 1—3 are the development stages of precipitation. The precipitation reaches a maximum during Phases 4—5 and then gradually weakens in Phases 6—8. Figure 9 also shows the 10—20-day filtered geopotential height field and wind field at 500 hPa for the eight phases. On the whole, each phase shows low-frequency wave trains of cyclones and anticyclones, and these wave trains move eastward over time. In the early stages, during the development of rainfall (Phases 1 and 2), and in the late stage, during the reduction of rainfall (Phases 7 and 8), the YHRB is controlled by anomalously high pressure and anticyclones. In these phases (Phases 1, 2, 7, 8), the cold—dry northerly airflow invaded the YHRB. Meanwhile, Phases 3—6 show that the YHRB is under the control of cyclones when the precipitation develops significantly. Thus, the extratropical cyclones have a good correspondence to the precipitation in February 2019. As mentioned above, in the development process, baroclinicity is the main forcing mechanism causing the development of the extratropical cyclone.

In Phase 2, the high-pressure center, which originally controlled the persistent precipitation that occurred in the YHRB in February 2019, moved eastward along the wave train. Then, the YHRB began to be affected by the southerly wind that was behind the high pressure. The wind provided water vapor for the continuous precipitation in the YHRB. In Phases 3—5, the high-pressure center enhanced again after moving eastward over the sea. This redeveloped high pressure blocked the eastward transmission of the waves. The low-pressure center and southerly winds, providing necessary conditions (uplift airflow, water vapor, etc.)for precipitation, were also blocked in the YHRB by this redeveloped high pressure. In Phases 6—8, as the blocking system (the high-pressure center) weakened and the low pressure moved eastward, the YHRB was re-affected by the cold—dry northerly airflow that was behind the low-pressure center, and precipitation began to decrease. Finally, the high-pressure center regained control of the YHRB and the precipitation process ended.

Fig. 9. Evolution of composite 10—20-day filtered 500-hPa winds (arrows) and 500-hPa geopotential height (shading; units: gpm)superimposed on the unfiltered 500-hPa geopotential height field (blue lines; only the contours of 5880 gpm are shown) and the unfiltered daily precipitation (green lines; only the contours of 10 mm d-1 are shown) over the YHRB (black box area) during a QBWO from phases 1 to 8 in February 2019.

Therefore, the blocking situation at low latitudes caused by the redevelopment of the high-pressure center may have been a key factor in the duration of precipitation in the YHRB. The intuitive reason for the redevelopment of the high-pressure center is the effect of the subtropical high.In Phases 1—8, the northern part of the subtropical high is located at a latitude of about 20°. In the latitude—time cross section of the monthly mean of geopotential height (figure omitted), the subtropical high in February 2019 was anomalously high. When the low-pressure center moved eastward to the YHRB, the gradient of northern to southern geopotential height increased, and the southerly wind strengthened.According to the Coriolis force, the southerly wind turned right into the high-pressure center, which resulted in the strengthening of the anticyclone. The redevelopment of the high-pressure center also led to an increase in the intensity of the subtropical high. Thus, the interaction between the high-pressure center and the subtropical high would have led to the formation of a blocking situation.

Latent heat is released by precipitation (Peixoto et al.,1991), and it may strengthen the anticyclone. The heating rate (Q) of latent heat can be written as follows (Ding,1989):

whereCpstands for specific heat at constant pressure,Lstands for latent heat, ω stands for vertical velocity, andqsstands for saturation specific humidity. The average latent heat of condensation of each tropospheric layer in February 2019 was calculated according to the above formula. The 3D model of the heating rate of latent heat in February 2019(figure omitted) shows that the heating rate of latent heat in the YHRB in February is mushroom-like and that the maximum value of the latent heating rate occurred at about 300 hPa and then gradually decreased.

The levels of 700, 500 and 300 hPa were selected to study the influence of the heating rate of latent heat on the subtropical high. The distribution of the latent heating rate and the distribution of anomalous stream function were plotted for each level (Fig. 10). From 700 to 300 hPa (Figs.10a—c), the heating rate of latent heat increased, and the anomalous anticyclone in the east of the latent heat source was enhanced due to the potential vorticity. The potential vorticity equation can be written as follows (Wu and Liu, 2000):

whereQstands for the heating rate of latent heat. ζ stands for relative vorticity, and other parameters are physical parameters commonly used in meteorology. The left part includes the advection term (I) and the β effect term (II),and the right part includes the ω term (III), the heating source term (IV), the zonal variation of heating rate term(V), the vertical variation of heating rate term (VI), and the meridional variation of heating rate term (VII). For longterm scales, the local variation of vorticity can be ignored.According to scale analysis (Wu and Liu, 2000), the vertical variation of heating term in the heat source zone is one order of magnitude larger than the other terms. Therefore,below the maximum heating layer (300 hPa), the potential vorticity equation can be written as:

Thus, southerly wind (v＞ 0) appears below 300 hPa.The β effect causes the anomalous anticyclone to appear to the east of the latent heat source below 300 hPa. This anomalous anticyclone will increase the intensity of the subtropical high.

According to the distribution of latent heating in Fig. 11,the heating rate of latent heat and the process of the precipitation have consistent changes. Additionally, when the anticyclone moved east to the sea, it was located on the right-hand side of the YHRB latent heat release area. Thus, the strength of the anticyclone was enhanced, which led to a blocking situation at low latitudes.

Fig. 10. Heating rate of latent heat (color shading; units: 10-3 K s-1) and stream function (contour interval: 2; solid lines represent anticyclones; dashed lines represent cyclones) at (a)700 hPa, (b) 500 hPa, and (c) 300 hPa in February 2019.

We name this blocking mechanism at low latitudes the“subtropical high—precipitation—anticyclone” (SHPA) feedback mechanism. Simply put, under the influence of the subtropical high and precipitation, the anticyclone redevelops and then forms the blocking situation. The anticyclone affects and enhances the anomalous subtropical high, and in turn affects and enhances the precipitation. Specifically,when the position of the subtropical high is farther north than in previous years, it increases the geopotential height gradient in the northern and southern part of the YHRB.The anomalous subtropical high also increases the speed of the southerly wind, which delivers water vapor, enhancing the precipitation. The latent heat released by the precipitation strengthens the anticyclone in the east of the YHRB,and the anticyclone maintains the intensity of the subtropical high. At the same time, the increase of the southerly wind caused by the subtropical high strengthens the anticyclone under the β effect and then blocks the low-latitude circulation situation and causes continuous rain.

Fig. 11. Evolution of composite 10—20-day filtered 500-hPa winds (arrows; units m s-1) and 500-hPa heating rate of latent heat (shading; units: 10-3 K s-1) during a QBWO from phases 1 to 8 in February 2019.

6. Impact of SSTAs

The circulation field of persistent precipitation was analyzed in detail in sections 4 and 5. As the external forcing field of the atmosphere, the SST is likely to have an impact on circulation and precipitation. In order to verify this view,we calculated the correlation coefficients between the winter (November—February) SSTA during 2000—19 (South China Sea, Bay of Bengal, the Philippines, and El Ni?o) and the February precipitation anomalies in the YHRB during 2000—19. The correlation coefficients between precipitation and (1) El Ni?o and (2) the South China Sea are 0.74 and 0.65, respectively. All of the correlations passed thet-test at the 95% confidence level. Therefore, we elaborate on the impact of the SSTA in these two areas on the persistent precipitation in the YHRB.

6.1. Impact of El Ni?o on precipitation in the YHRB

El Ni?o has a non-negligible influence on the global atmosphere (Meehl, 1990; Collins et al., 2010; Feng et al.,2019). Wang and Picaut (2004) noted that recent research findings about ENSO can be grouped into two parts: the first part shows that ENSO could be an irreplaceable part of the coupled ocean—atmosphere mode; and the second part shows that ENSO could be a damped mode that is mainly sustained by the atmospheric random “noise” forcing. The significant correlation between the tropical SSTA and the subtropical high anomaly has been confirmed in many previous studies (Yang et al., 2017; Qian et al., 2018). The above analyses show that the subtropical high anomaly is an important factor in this persistent precipitation that occurred in the YHRB in February 2019. Moreover, since 2019 was an El Ni?o year, it is necessary to analyze the role of El Ni?o in the persistent precipitation.

Fig. 12. SSTA (color shading; units: °C) and stream function (contour interval: 5; positive values represent anomalous anticyclones; red box indicates the area of the WNPAC; blue box is the Matsuno—Gill response) in February 2019.

Fig. 13. Anomalous ω (shading; units: -1 × 10-2 Pa s-1) and vertical wind profile (arrow lines; units: m s-1)in February 2019.

The SSTA in February is shown in Fig. 12. It has characteristics of both El Ni?o Modoki and an eastern Pacific El Ni?o event; however the latter may have been more prominent. The vertical wind (Fig. 13) shows that a strong updraft was located around the western dateline. This means that the Walker circulation in the tropical Pacific moved slightly eastward in response to the anomalous warming in the equatorial central Pacific. Thus, this response can also impact the circulation in the extratropical region by providing latent heat and stimulating teleconnection wave-trains under the tropical convective activity.

Moreover, the anomaly of the streamfunction distribution in Fig. 12 indicates that, in the winter of 2018/19, the typical response to El Ni?o events in East Asia may have been in the form of a western North Pacific anomalous anticyclone (WNPAC, red box in Fig. 12). The WNPAC causes anomalous southwesterly wind in the lower troposphere to be more prominent, with stronger water vapor transport.This anomalous southwesterly wind was also the direct cause of the extreme continuous rainy winter that occurred in the YHRB in February 2019. In combination with the distribution of anomalous OLR in February 2019, the WNPAC was caused by anomalous active convection (obvious negative OLR anomaly) in the equatorial central and eastern Pacific and was also forced by the strong latent heating. An obvious cyclonic circulation (blue box in Fig. 12), which was caused by the Matsuno—Gill response, occurred in the northwestern heating region. This wave train continued to propagate northwest, and then a downdraft appeared near the northwest coast of East Asia. Moreover, a response was stimulated by the convective cooling due to the weakened convective activity around the Philippines following the eastward-moving Walker circulation.

In February 2019, the anomalous ascending branch of the Pacific Walker circulation moved to the west side of the dateline, and then a cyclonic circulation at low altitudes was stimulated in the vicinity of the dateline by the Matsuno—Gill response. The anticyclonic circulation at low latitudes was also stimulated on the northwest side of the dateline by a sinking Rossby wave, which strengthened the subtropical high. Additionally, the WNPAC had a certain effect on the development of the anticyclone mentioned in section 4.

6.2. Impact of the SST anomaly in the South China Sea on precipitation in the YHRB

The South China Sea is located at the western boundary of the WNPAC. According to the wind—evaporation—SST feedback mechanism (Wang et al., 2000), the superposition of the WNPAC and the average northeast trade wind caused the trade wind on the west side of the anticyclone to weaken, and thus the heat loss caused by latent heat was reduced. Thus, the SST showed a warm anomaly.Moreover, the analyses of Barros and Silvestri (2002)showed that the subtropical high is basically consistent with a warming pool. Therefore, the SST warming in the South China Sea caused by the WNPAC was closely related to the strong intensity of the subtropical high and its position during the continuous rainy period in February 2019.

Therefore, research on the influence of the South China Sea SST on the subtropical high should consider the direct effect of sensible heating caused by the anomalous SST.

The anomalous SST in the South China Sea in winter 2019 increased the temperature gradient between the ocean and atmosphere. Sensible heating is caused by small-scale turbulent motion in the boundary layer, and its diffusion is limited to the lower troposphere (McIntosh et al., 1975). Simultaneously in the surface layer of the heating zone, the Burger number in the subtropical region is large (Hoskins,1991). The non-adiabatic heating is mainly balanced by adiabatic cooling in rising airflow (McIntosh et al., 1975). Therefore, the heating rate is rapidly reduced as the air rises to near 800 hPa.

The sensible heating effect causes positive vorticity at the bottom layer of the troposphere. Thus, the airflow converges in the surface layer and then causes a compensatory anomalous cyclone (Hoskins et al., 1985). At the same time,the converging airflow creates underlying friction. This applies negative vorticity to compensate for the positive vorticity generated by non-adiabatic heating in the gas column.Moreover, the developing southerly wind transports water vapor and results in the persistent precipitation observed in the YHRB.

In the middle layer, the vorticity turns negative and generates an anomalous anticyclone (development of the subtropical high) due to the diverging airflow.

7. Summary

In this paper, we address the following questions: (1) Persistent precipitation requires a lot of water vapor, but where is the water vapor coming from during the winter monsoon?(2) Why was the duration of the precipitation that occurred in the YHRB in February 2019 so long? (3) Which factor induced the enhancement of the persistent precipitation, and what was the physical linkage between this factor and the persistent precipitation? (4) What role did El Ni?o play in this precipitation? (5) Did the SSTAs of China’s coastal waters have an impact on the precipitation?

Through the analysis performed in this study, it was found that the persistent precipitation in February 2019 was frontal precipitation. The front provided conditions of lift,instability, and moisture for precipitation. Multiple systems(subtropical high, anticyclone, latent heating) enhanced the southerly wind, which transported water vapor for the precipitation. Additionally, a blocking system of both the high- and low-latitude circulation fields provided a continuous disturbance field for precipitation. As for the SSTA, El Ni?o influences the precipitation in the YHRB by stimulating a WNPAC and Matsuno—Gill response. The South China Sea SSTA provided sensible heating for precipitation and then caused the rising airflow that was conducive to the occurrence of the persistent precipitation in February 2019.

Furthermore, based on analysis of the QBWO of the persistent precipitation, we propose the SHPA feedback mechanism. This mechanism combines the subtropical high and precipitation anticyclone and expounds the physical linkage among the three (subtropical high, precipitation and anticyclone).

Moreover, in order to better demonstrate the applicability of the SHPA feedback mechanism, we performed a composite analysis of the anomaly wind fields and subtropical high fields during the persistent precipitation periods of February 2000—19 in the YHRB at 500 hPa. The results are shown in Fig. 14. It can be seen that the abnormal anticyclone and northerly airflow controlled the circulation field over the YHRB area before the occurrence of precipitation events in February 2000—19. By the time the continuous precipitation occurred, the abnormal anticyclone moved eastward over the sea, and the YHRB area was controlled by the southerly wind at the back of the anticyclone. Moreover,the subtropical high position was more northerly than in previous years.

The most important point in Fig. 14 is that the anticyclone moved eastward quickly (100°E to 120°E in three days) and then basically remained stationary in the eastern part of both the YHRB area and the latent heating area during the whole precipitation period (≥ 10 days). Additionally, the subtropical high strengthened the southerly wind that carried water vapor to the YHRB.

Fig. 14. Composite analysis of anomaly wind fields (arrow lines) and subtropical high fields (blue lines) related to persistent precipitation events (rainy days ＞ 10 d; precipitation＞ 10 mm d-1) in (a) 5—1 days before rainfall and (b) persistent rainfall periods over the YHRB at 500 hPa.

Therefore, the results of the composite analysis suggest that the SHPA feedback mechanism can provide new ideas for the forecasting of persistent precipitation in the YHRB area.

However, the feedback mechanism has some preconditions; that is, the area of precipitation needs to be located on the north side of the subtropical high, the anticyclone needs to be located on the west side of the precipitation, and the southerly wind can provide the necessary warm and moist conditions for the precipitation. The SHPA feedback mechanism reveals the physical mechanism of the blocking circulation and persistent precipitation at low latitudes in winter and can be used for the prediction of persistent weather disasters in low-latitude regions.

Acknowledgements.We gratefully acknowledge support from the State Key Program of the National Natural Science Foundation of China (Grant No. 41430963).