Under the influence of the East Asian winter monsoon (EAWM), extreme winter precipitation events, such as heavy rainfall, snowstorms, and severe freezing rain, occasionally occur over South China (SC), causing a catastrophic impact on agriculture, transportation, and human lives (Wen et al., 2009). Although the winter SC precipitation (SCP) accounts only for approximately 10% of the annual total precipitation, it experiences large year-to-year variability (Wang and Feng, 2011; Ge et al., 2016). Therefore, a deeper understanding and accurate prediction of such events are of great importance. The interannual variation of winter SCP had become a subject of great concern, but the underlying physical mechanisms are still not fully understood and require further investigation.

Many previous studies had revealed that the El Niño–Southern Oscillation (ENSO) and the EAWM are two important factors that affect the SCP (Wang et al., 2000; Huang et al., 2003; Zhou and Wu, 2010; Jia and Ge, 2017). Wet (dry) anomalies tend to appear in SC during the El Niño (La Niña) winters (Wu et al., 2003). The ENSO impacts SCP mainly through modulating the Walker circulation and low-level circulation anomalies over the western North Pacific (WNP) region. In an El Niño winter, the sea surface temperature (SST) warming anomalies in the equatorial central/eastern Pacific (CP/EP) and in the tropical Indian Ocean (TIO), as well as cooling in the equatorial western Pacific (WP), jointly force a WNP anticyclone (WNPAC) anomaly through chains of ocean–atmosphere coupling and atmospheric dynamical processes (Wu et al., 2010; Tim Li et al., 2017). The WNPAC anomaly then induces anomalous southwesterly flow and transports more moisture to the SC, resulting in increased precipitation (Zhang et al., 1996; Wang et al., 2000; Chung et al., 2011). In La Niña, the nearly opposite SST anomalies tend to force a WNP cyclone (WNPC) anomaly and thus cause SCP deficiencies. It had also been noted that different flavors of ENSO may have different types of impact on SCP. The EP El Niño tends to exert a stronger influence on SCP than the CP El Niño (Feng and Li, 2011; Jiang et al., 2019). Recent studies had revealed that the EP and CP types of La Niña would also exert different types of climate impact over SC (Yuan et al., 2014; Yu and Sun, 2018).

The EAWM is another important system that affects winter SCP (e.g., Wang and Feng, 2011; Lee et al., 2013). A weak EAWM is associated with anomalous south-westerlies over the East Asian coast that can induce anomalous moisture transport and upward motion, thus enhancing SCP. It should be pointed out that the EAWM and ENSO are not independent. A weak EAWM is usually accompanied by the mature phase of an El Niño event (Kim et al., 2017). But Zhou and Wu, (2010) noted that the EAWM-associated wet anomaly extends more northward than that of the ENSO, due to the anomalous southerlies penetrating more northward over eastern China. It had also been noted that the ENSO–EAWM relationship is unstable, which can be modulated by the Pacific Decadal Oscillation, the Atlantic Multidecadal Oscillation, and global warming (He and Wang, 2013; Jia et al., 2020). In addition to ENSO and EAWM, some recent studies highlighted that a Rossby wave train propagating along the subtropical jet over South Asia can also significantly impact the winter SCP (Xiuzhen Li et al., 2017; Hu et al., 2018; Ma et al., 2019; Shen et al., 2019; Li et al., 2020). This wave train can be captured as the leading empirical orthogonal function (EOF) mode of the monthly meridional wind (v) over the North Africa–Asia region in the upper troposphere (Hu et al., 2018). It deepens the India–Burma trough (IBT) over the northern Bay of Bengal (BoB), enhancing water vapor transport from the BoB to SC, thus enhancing SCP.

As had been previously discussed , wet winters over SC are most likely to be associated with El Niño and weak EAWM, such as the extreme positive SCP anomalies occurred in winter 1982/1983, 1997/1998, and 2015/2016. Actually, although La Niña tends to increase the probability of dry anomalies and persistent cold, and wet weather occasionally occurred in La Niña winters, for example, winter 2007/2008 and 2017/2018 (Wu et al., 2011; Wang et al., 2020). But these La Niña winters are usually primarily featured by severe snowfall or freezing rain, with the seasonal mean precipitation amounts still near or even below normal (Figure 1B). However, the La Niña winter of 2021/2022 is a unique case. It is surprising to see that SC experienced extreme wet conditions throughout this winter (Figure 1A). The SCP amount in JF 2022 even exceeds that in El Niño winters of 1998 and 2016 (Figure 1D). So far, the underlying mechanisms driving the 2022 wet anomalies over SC remain unclear. In this study, we use observations and re-analysis to investigate which role tropical thermal forcings and extra-tropical circulation anomalies played in the extreme SCP in JF 2022. Particularly, whether the extreme winter precipitation in 2022 could be ascribed to the wave train along the South Asian jet or high latitude circulation patterns is an interesting question. In the following sections, Section 2 describes the data and methods. In Section 3, anomalous SCP in 2022 and its linkages with the atmospheric/oceanic conditions are analyzed. In Section 4, a multiple linear regression (MLR)-based reconstruction of the winter rainfall of JF 2022 is presented, in order to estimate the relative contribution from several identified influence factors. Finally, a summary and discussion of the results are given in Section 5.

The monthly precipitation dataset from the Global Precipitation Climatology Project (GPCP) with a horizontal resolution of 2.5° × 2.5° is used in this study (Adler et al., 2003). The monthly SST dataset is the Centennial in situ Observation-Based Estimates of SST (COBE SST; Ishii et al., 2005), which has a horizontal resolution of 1.0° × 1.0°. For the circulation variables, we use the National Centers for Environmental Prediction-Department of Energy (NCEP-DOE) atmospheric reanalysis dataset (Kanamitsu et al., 2002), with a horizontal resolution of 2.5° × 2.5°. Considering that the observations and reanalysis before the satellite era (around 1979) exhibit larger uncertainties, the analysis period of the present study ranges from 1979 to 2022. In this article, the anomalies in JF 2022 are computed relative to the climatology of 1991–2020.

The Niño3 (Niño4) index is defined as the area-averaged SST anomaly over 150–90°W, 5°S–5°N (160°E–150°W, 5°S–5°N). The Niño3.4 index is defined as the averaged SST anomaly over 170–120°W, 5°S–5°N. The La Niña events are defined based on a threshold of −0.5°C of the Niño3.4 index for five consecutive months. According to this definition, 14 La Niña winters had been identified, including 1985, 1989, 1996, 1999, 2000, 2001, 2006, 2008, 2009, 2011, 2012, 2018, 2021, and 2022. Similar to the definition of Kug et al. (2009), we further classify these events into EP and CP types according to the absolute ratio of the Niño3 index to the Niño4 index. If the ratio is greater (less) than 1, then an EP (CP) La Niña event is identified. The EP La Niña events include 1985, 1996, 2000, 2006, 2008, 2018, and 2022, while the CP La Niña events include 1989, 1999, 2001, 2009, 2011, 2012, and 2021.

To describe the propagation of Rossby wave energy in the upper troposphere, the phase-independent wave activity flux (WAF; Takaya and Nakamura, 2001; see equation below) was calculated. The phase-independent flux was derived from combined perturbation energy and pseudo-momentum terms. Mathematically, the WAF can be written as follows: W = 1 2 | U | ( u ¯ ( ψ x ′ 2 − ψ ′ ψ x x ′ ) + v ¯ ( ψ x ′ ψ y ′ − ψ ′ ψ x y ′ ) u ¯ ( ψ x ′ ψ y ′ − ψ ′ ψ x y ′ ) + v ¯ ( ψ y ′ 2 − ψ ′ ψ y y ′ ) f 2 R σ / P { u ¯ ( ψ x ′ ψ p ′ − ψ ′ ψ x p ′ ) + v ¯ ( ψ y ′ ψ p ′ − ψ ′ ψ y p ′ ) } ) .

Here, ψ denotes the stream function, f is the Coriolis parameter, R is the gas constant, U =(u, v) represents the horizontal wind velocity, and σ = ( R T ¯ / C p p ) - d T ¯ / d p , with temperature T, and the specific heat at constant pressure C p . Overbars and primes represent the climatology and anomalies, respectively. The so-derived WAF is parallel to the group velocity of local Rossby waves and is suitable for a snapshot analysis of either stationary or migratory waves on a zonally varying basic flow. Thus, the WAF represents the propagation of the wave packet.

The aforementioned sections discussed several climate drivers that may impact the SCP anomaly in JF 2022. These factors include the ENSO phase, the warming of TIO, the wave train along the South Asian jet, and the GPH anomaly over eastern Siberia. To further identify the relative importance of the aforementioned four factors, the MLR method is employed to reconstruct the precipitation anomaly in JF 2022. First, the MLR analysis is performed using the Niño4 index (quantify ENSO), the SST averaged over 20°S–20°N, 50–100°E (quantify TIO SST), the vPC1 (quantify the South Asian jet wave train activity), and the ESHI (quantify GPH anomaly over East Siberia), based on monthly data from winter 1979/1980 to 2020/2021. Here, considering an EP La Niña occurred in 2022 which exerts a weaker impact on SCP than its CP counterpart, instead of the Niño3 index, we use the Niño4 index as a predictor to avoid overestimating La Niña effects on SCP in 2022. The correlations between the Niño4 index and the ESHI, as well as between vPC1 and ESHI, are near zero, indicating they are nearly independent. However, we note that there is a moderate correlation between Niño4 and vPC1 (r = 0.23) and a strong correlation between Niño4 and TIO (r = 0.55). Therefore, before performing the MLR analysis, we had removed the ENSO-related variability from the vPC1 and TIO SST to obtain an ENSO-independent vPC1 (i.e., vPC1res) and an ENSO-independent TIO SST (i.e., TIOres SST), respectively. These are performed by means of linear regression. After this, the four indices are almost orthogonal to each other, making it more reliable to superimpose their individual climate effects. Then, the MLR coefficients are multiplied by the corresponding values of each index in January and February. The reconstructed JF precipitation anomalies are obtained by adding the reconstructed January and February values. Our examination suggests that this MLR model can well reconstruct the historical SCP variability. The observed JF SCPI has a significant correlation coefficient (r = 0.84) with the MLR-reconstructed SCPI during 1980–2021; thus, over 70% of the total variance of the SCPI can be explained by the selected factors. Therefore, it is reliable to use this MLR model to reconstruct SCP.

Figures 10A–D show the patterns of contributions of each factor to precipitation anomalies in JF 2022; it is found that the signals of different factors show distinct spatial distributions. We primarily focus on the reconstructed precipitation anomalies over the SC. The contributions of each factor to SCP are displayed in Figure 11. The imprint of the Niño4 index, indicative of the impact from the La Niña event, exhibits a weak dry anomaly (−0.16 mm/day). Thus, La Niña exerts a slightly negative contribution to the SCP anomalies in JF 2022. The TIO warming causes a slight increase in SCP (0.29 mm/day) that accounts for 16% of the observed rainfall anomaly, generally consistent with the previous findings (Zhang et al., 2015). In contrast, the signal of vPC1res, representing the effect of the South Asian jet wave train, is featured by stronger wet anomalies over SC. The SCPI response is 0.86 mm/day, accounting for almost 50% of the observed rainfall anomaly. The ESHI also produces an evident positive SCP, with an SCPI response of 0.47 mm/day, accounting for ∼25% of the observed rainfall anomaly. Therefore, the two extra-tropical circulation patterns, characterized by vPC1res and the ESHI, dominated the observed SCP anomalies in JF 2022, with a larger contribution from the former (∼50%). The sum of these two factors accounts for ∼75% of the observed SCP anomaly. On the other hand, the tropical oceanic forcings, including the La Niña and the TIO warming, exert a relatively weak impact on SCP in 2022.

Next, we examined the spatial pattern of the reconstructed precipitation. Figure 10E depicts the reconstructed 2022 precipitation anomalies using all four factors. It is found that the main features of the rainfall pattern over the Asian–Pacific region are well captured. The wet anomalies over SC, BoB, SCS, and equatorial WP, as well as dry anomalies over Japan and equatorial Indian Ocean, are all consistent with observations. In addition, the reconstructed H500 pattern bears high resemblance with observation (Figure 10F and Figure 6A). The key atmospheric circulation anomalies associated with the 2022 SC wet anomaly are also well reproduced, including the wave train along the South Asian jet and positive GPH anomalies over eastern Siberia. These results indicate that the MLR reconstruction may have captured the main physical drivers of SCP anomaly in JF 2022.

There comes another interesting question: are these four identified factors also responsible for the sub-seasonal change in SCP anomalies from early (December 2021) to late winter (JF 2022)? In order to reveal this, we repeated the MLR reconstruction in Figure 11A for the SCP anomaly in December 2021, as shown in Figure 11B. It is seen that for December 2021, the sum of these four factors, which represent the MLR reconstruction of SCP, is nearly identical to the observed SCP anomaly, demonstrating that the sub-seasonal change in the precipitation anomaly in winter 2021/2022 can be well captured by this MLR reconstruction. It is also found that the contributions from tropical oceanic anomalies, including the La Niña event and the TIO warming, are very close between early and late winter. Therefore, the sub-seasonal change in the precipitation anomaly in this winter mainly arises from the extra-tropical circulation anomalies, including the changes in the South Asian jet wave train pattern and the ESHI anomaly.

The SC region had experienced long-lasting rainy weather throughout the winter of 2021/2022. The wet anomalies were particularly strong during JF 2022. The SCPI reaches +1.78 mm/day, the second largest value since 1980, only next to 1983. At the same time, there was a La Niña event. As suggested by the previous studies, canonical anomalies associated with a mature La Niña would have the WNPC anomaly bring a dry winter to the SC in 2022. But in reality, an extreme wet event occurred instead. This promotes us to ask why was the La Niña winter of 2021/2022 extremely wet over SC? In this study, we use observations and reanalysis to investigate which role tropical SST anomalies and extra-tropical circulation anomalies played in this wet event.

Our analysis suggests the La Niña event had not effectively forced circulation anomalies over the WNP region in JF 2022 as expected. Particularly, the WNPC anomaly at lower levels was not obvious, which is at least unfavorable for SCP deficiencies. The absence of a WNPC anomaly may be related to the SST distributions across the tropical oceans. Over the tropical Pacific, the EP type of a La Niña event shifted the center of the negative SST anomaly more eastward than the average of historical La Niña events. This would slacken the La Niña-associated anomalous zonal SST gradient between CP and WP, thus suppressing the development of anomalous Walker circulation and the WNPC. On the other hand, there was an unexpected SST warming over the TIO, which would favor the anticyclonic anomaly over the WNP region. Thus, the EP type of La Niña and the SST warming in the TIO act in concert to dampen the WNPC anomaly and the SCP deficiencies. However, these tropical SST anomalies, along with the corresponding WNP circulation pattern, could only explain why there was no evident dry anomaly over the SC. They are still unable to explain the observed wet extreme, as the WNPAC anomaly was absent. Thus, the tropical oceanic condition may not be the primary cause for the extreme wet condition over SC in JF 2022.

Furthermore, the investigation identifies two extra-tropical circulation anomalies over Eurasia that play dominant roles in the 2022 SC wet extreme. One is the wave train propagating along the South Asian jet. This wave train pattern results in a strong cyclonic anomaly over TP and the northern BoB region, which intensifies the IBT. In JF 2022, the IBTI anomaly exceeds +1.2 standard deviation. The intensified IBT can enhance SCP through exciting anomalous moisture transport from BoB and ascending motion. A positive GPH anomaly over eastern Siberia is another important factor influencing the winter SCP. In JF 2022, the GPH anomalies at high latitudes show a west–east dipole pattern, and negative anomalies center over Scandinavia, while the positive anomalies dominate eastern Siberia. The circulation anomaly over eastern Siberia usually exhibits a quasi-barotropic structure, which is physically consistent with the anticyclonic anomaly at 850 hPa and positive SLP anomaly at the surface. The pronounced anomalous cold northeasterly from this anomalous anticyclone encounters warm and moist air over the SC, which would support strong convergence and upward motion and consequently provide a favorable dynamical condition for SCP. The MLR reconstruction suggests that the South Asian wave train explained ∼50% of the observed SCP anomaly, while ESHI explained ∼25% of the observed SCP anomaly in JF 2022. Therefore, the unexpected extreme wet condition over the SC in JF 2022 is attributed mainly to the enhanced IBT maintained by the South Asian jet wave train, as well as the high-pressure anomalies over eastern Siberia.

Canonical anomalies associated with a mature La Niña would have the WNPC anomaly bring a dry winter to SC in 2022. In reality, the South Asian jet wave train, as well as the high-pressure anomaly over eastern Siberia, overwhelmed this tendency, bringing an extremely wet winter instead. Particularly, the South Asian jet wave train that intensifies IBT plays the most important role. This promotes us to ask what are the potential causes for this pattern? Previous studies had identified two kinds of forcing factors that may influence the phase of this wave train: the tropical SST anomalies and the North Atlantic Oscillation (NAO). For SST anomalies, previous studies suggested a moderate impact of an El Niño-like SST pattern across the tropics on the South Asian jet wave train. According to Hu et al. (2018), warm SST anomalies and positive rainfall anomalies in the equatorial EP would excite a wave-like pattern from the Northeast Pacific across the North Atlantic through Europe to East Asia, thus projecting onto the positive phase of vPC1. But considering a La Niña event that occurred, the tropical SST pattern is not a plausible cause for the observed wave train pattern in JF 2022.

Some studies had emphasized the role of the NAO in triggering wave trains along the South Asian jet stream, and the positive NAO anomaly tends to cause the positive phase of vPC1 (Branstator, 2002; Watanabe, 2004). In JF 2022, the NAO index reaches +1.1, which may favor the observed South Asian jet wave train anomaly to some extent. However, the impact of NAO on vPC1 is not deterministic, as a comprehensive analysis by Huang et al. (2020) had shown that both positive and negative phases of NAO can excite the same phase of the South Asian wave train, although with different orientations. In fact, the wave train is an atmospheric internal mode, which can be developed without external forcing. Internal atmospheric dynamic processes, such as the conversion between barotropic and baroclinic energy, are also important in maintaining and strengthening the wave train along the westerly jet stream (Li et al., 2020). Therefore, the wave train anomaly in JF 2022 may be largely due to atmospheric internal dynamical processes, which indicates a low predictability season ahead. But the exciting mechanisms of the South Asian jet wave train remain controversial. In addition, although the present study highlights the influence on the SCP anomalies of both tropical SST/circulation anomalies and atmospheric inner dynamic processes in the extra-tropics, other climatic factors, for example, the Arctic sea ice concentration, may also play a role. Previous studies revealed that the reduction in the Arctic sea ice may excite anomalous upward air motion due to strong near-surface thermal forcing, which further triggers a meridional overturning wave-like pattern extending to mid-latitudes of Eurasia, thus, in turn, inducing the zonally oriented Rossby wave train along the mid-latitudes of jet stream (He et al., 2018; He et al., 2020). Whether the sea ice anomaly affects the 2022 extreme SCP should be further investigated in future studies.