Seasonal and Sub-Seasonal Circulation Anomalies Associated with Persistent Rainy Days in 2018/2019 Winter in Shanghai, China

Shanghai experienced the longest rainy days in 2018/2019 winter since 1988. The physical cause of such an unusual climate condition was investigated through the diagnosis of observational data. From a seasonal perspective, a long persistent rainy winter was often associated with an El Niño condition in the equatorial Pacific. This abnormal oceanic condition induces a remote teleconnection pattern with pronounced low-level southerly anomalies over East China. The wind anomalies transported moisture from tropical oceans and caused persistent rainfall in East Asia. Meanwhile, the local rainfall time series exhibited a strong quasi-biweekly oscillation (QBWO). Three persistent rainy events were identified in the 2018/2019 winter and they all occurred during the active phase of the QBWO. The first two events were associated with a low pressure anomaly west of Shanghai. Southerly anomalies associated with the low pressure system advected high mean moisture into central eastern China, leading to the persistent rainfall there. The third event was associated with a high pressure anomaly in lower troposphere to the east of Shanghai, which induced anomalous southerlies to its west, favoring the occurrence of rainfall in Shanghai. The result suggests the importance of high-frequency variability in affecting seasonal rainfall anomalies.


Introduction
Climatological seasonal wind in central East China in boreal winter is characterized by pronounced northerlies (Li, 2010). Such winds advect dry and cold air southward, leading to a minimum in seasonal rainfall in the region. The 2018/2019 winter, however, experienced an unusual wet season in Shanghai, which was characterized by persistent rain throughout the season. This motivates us to investigate the physical cause of this unusual seasonal climate condition.
One possible cause of the unusual rainy season is attributed to the change of winter mean circulation associated with the El Niño-Southern Oscillation (ENSO). For instance, it is well known that during El Niño mature winter, an anomalous large-scale anticyclone develops over the tropical western North Pacific (Chang et al., 2000a, b;Wang et al., 2000;Li et al., 2017). This anomalous anticyclone is possibly caused by a combined effect of local air-sea interaction mechanism (Wang et al., 2000) and moist enthalpy advection mechanism (Wu et al., 2017a, b) during its initiation stage [see Li and Hsu (2018) for a thorough review on this subject]. Anomalous southerlies to the west of the anomalous anticyclone transport high mean moisture northward, causing positive rainfall anomalies over South China during El Niño mature winter (Wang et al., 2003;Wu et al., 2010). This remote forcing effect is asymmetric between El Niño and La Niña, due to the intensity and pattern asymmetry between the anomalous anticyclone during El Niño and the anomalous cyclone during La Niña (Wang et al., 2019).
In addition to the seasonal mean circulation change, sub-seasonal modes may also alter large-scale circulation and contribute, to a certain extent, to rainfall activity Julian, 1971, 1972;Wang and Rui, 1990;Chatterjee and Goswami, 2004;Jiang et al., 2004;Jiang and Li, 2005;Li and Wang, 2005;Wen et al., 2010;Li, 2014;Xu et al., 2017;Zhu et al., 2017;Qi et al., 2019). It has been shown that there are marked quasi-biweekly and higher-frequency intraseasonal oscillations over East Asia in boreal winter (e.g., Yao et al., 2014;Li, 2016, 2017). For example, Yao et al. (2014) noted that strong intraseasonal variation of precipitation occurred in South China in boreal winter, and the rainfall variability was associated with the southeastward propagation of the quasi-biweekly mode that originated from the high latitudes. The surface air temperature in mid-and high-latitude Eurasia in boreal winter also exhibited a strong intraseasonal fluctuation (Yang and Li, 2016). Coupled with upper-tropospheric anomalous geopotential height, the signal of near surface temperature propagated southeastward over East Asia. A wave activity flux (WAF) analysis showed that Rossby wave energy dispersion caused the wave train to propagate southeastward. The source of the Rossby wave train appeared over the highlatitude Europe/Atlantic sector, where there was the maximum convergence of WAF. There was barotropic energy conversion from the mean flow to the intraseasonal perturbation during its southeastward journey.
The objective of the current study is to examine the physical cause of persistent rainy season in Shanghai during the 2018/2019 winter from both the seasonal and sub-seasonal circulation change perspective. The remaining part of this paper is organized as follows. In Section 2, the data and method are introduced. In Section 3, we examine seasonal abnormal conditions associated with the persistent rainy day in Shanghai during the past 32 years and compare the conditions with those in the 2018/2019 winter. In Section 4, we further examine subseasonal circulation patterns and their evolutions associ-ated with the persistent rainfall in the 2018/2019 winter. Finally, the conclusions and discussion are given in Section 5.

Data and analysis method
The primary observational data used in the current study include 1) daily station rainfall data in Xujiahui Station, a station that holds the longest meteorological record in Shanghai, 2) the NCEP/Department of Energy (DOE) Reanalysis II (Kanamitsu et al., 2002), 3) NOAA Climate Prediction Center (CPC) Merged Analysis of Precipitation (CMAP) data, and 4) NOAA Extended Reconstructed Sea Surface Temperature (ERSST) v4 dataset. Both the reanalysis and CMAP products have a horizontal resolution of 2.5° × 2.5°, whereas the ERSST has a horizontal resolution of 2° × 2°. The analysis focuses on boreal winter season (i.e., from December to February) for the period of 1988-2019.
A persistent rainy day index is defined as the number of rainy days within a 3-month period (from 1 December to 28 February) with daily rainfall amount no less than 0.1 mm. According to this definition, 2018/2019 winter became the year of longest rainy days in Shanghai in the past 32 years (Fig. 1). There were a total of 49 rainy days within the three months.
A regression analysis is further employed, to derive the seasonal mean circulation and rainfall anomaly patterns associated with the rainy day index. This analysis will illustrate favorable tropical SST and circulation patterns associated with the persistent rainfall in Shanghai.
A power spectrum analysis is applied to illustrate the APRIL 2020 dominant period of sub-seasonal rainfall variation in Shanghai Station. Once the dominant period is known, a Lanczos bandpass filtering is further applied to all datasets to retain the intraseasonal signals.
To illustrate Rossby wave energy propagation in the upper troposphere, we adopt phase-independent wave activity formula (Takaya and Nakamura, 2001). The Rossby wave activity flux may be written as where , , and represent the climatological seasonal mean horizontal, zonal, and meridional winds, denotes intraseasonal perturbation stream function, and subscripts denote partial derivatives at zonal or meridional directions. The wave activity flux represents Rossby wave energy propagation direction. Figure 1 shows that the Shanghai persistent rainy day index exhibits a strong interannual fluctuation. In some years, the number of total rainy day exceeds 35 days, while in other years the number of total rainy day is about 20 days or lower. The maximum rainy day index occurred in 2018/2019 winter, followed by 1997/1998 winter.

Seasonal mean circulation patterns associated with the rainy day index
To illustrate large-scale atmospheric and oceanic conditions associated with the large rainy day index, we conducted a regression analysis by regressing SST, precipitation, specific humidity, relative humidity, and wind fields onto the index during 1988-2019. Figure 2 shows the regression result. It is interesting to note that an El Niño like SST pattern appears over the tropical Pacific (Fig. 2a). Associated with this SST pattern is pronounced westerly anomaly and enhanced rainfall over the equatorial central Pacific. In the East Asia-western North Pacific sector, there is a low-level anomalous anticyclone . Anomalous northerly to the east of the anticyclone co-locates with cold SST anomaly (SSTA) and negative precipitation anomaly (Figs. 2a,b), confirming the local air-sea interaction mechanism (Wang et al., 2000) and moist enthalpy advection mechanism (Wu et al., 2017a, b). To the west of the anticyclone is pronounced southerly anomaly (Fig. 2a), which advects tropical moist and warm air northward, providing moisture and latent energy source over South and central China (Fig. 2c).
The circulation patterns shown in Fig. 2 were derived from the regression analysis. They illustrate general large-scale circulation conditions for a long persistent rainy winter in Shanghai during the past 32 years. Next, we compare the circulation feature above to those in 2018/2019 winter. Figure 3 shows the anomalous seasonal Contour from −0.3 to 0.5 by 0.1 (c) Fig. 2. The horizontal patterns of (a) anomalous SST (shaded; °C; black dots denote 90% confidence level) and 850-hPa wind fields (vector; m s −1 ), (b) anomalous precipitation (shaded; mm day −1 ; black dots denote 90% confidence level) and 200-hPa wind fields (vector; m s −1 ), and (c) anomalous 850-hPa relative humidity (shaded; %) and specific humidity (contour; g kg −1 ; black dots denote 90% confidence level) fields regressed onto the persistent rainy day index for 1988-2019.
circulation patterns during that particular winter. Note that the SST, wind, moisture, and precipitation patterns are in general similar to those derived from the regression. The most robust common features that appeared in both figures are an El Niño like warming pattern and pronounced southerly anomalies over East China. There are also some notable differences. For instance, equatorial SSTA in 2018/2019 winter was stronger than the re-gressed SSTA amplitude, and the associated equatorial precipitation anomaly was also stronger and shifted further westward (Figs. 3a, b). The area of enhanced moisture anomalies appeared confined more over central and South China (Fig. 3c). The result above describes the favorable conditions of seasonal SST and circulation anomalies for persistent rainfall in Shanghai in 2018/2019 winter. In the next section, we further examine the sub-seasonal circulation patterns and their evolutions in association with the persistent rainfall in Shanghai.

Role of sub-seasonal events in affecting the persistent rainfall in Shanghai
A power spectrum analysis was carried out to reveal the dominant period of daily rainfall in Shanghai Station during the 2018/2019 winter. Figure 4a shows the analysis result. A most significant peak appears in 10-25 days, indicating that the sub-seasonal variability is dominated by the quasi-biweekly oscillation (QBWO) during the period.
A persistent rainy event is defined as follows: there are at least 4 rainy days within a 5-day interval or at least 7 rainy days within a 10-day interval. Here, a rainy day is defined when daily amount of rain is no less than 0.1 mm per day. According to this definition, three persistent rainy events were derived during the 2018/2019 winter: Event 1 for 2-11 December, Event 2 for 4-12 January, and Event 3 for 7-22 February.
To illustrate the relationship between the persistent rainy events and the QBWO mode, we plotted the daily rain amount and the quasi-biweekly rain time series together in Fig. 4b. Note that the identified persistent rainy events happened during the active phase of the QBWO mode, implying that the persistent rainy events were closely related to the QBWO. Figure 5 shows the vertical profiles of vertical velocity and specific humidity anomalies averaged over the three rainy events. Note that there was a marked ascending motion anomaly in Shanghai associated with each of the rainy events. Accompanied with the ascending motion were positive specific humidity anomalies throughout the most of the troposphere. Maximum specific humidity appeared at 700 hPa. In addition, relative humidity anomalies were also positive throughout the troposphere. These large-scale moisture conditions favored the development of persistent rainy events in the region.
To reveal the origin of the large-scale ascending motion and moisture sources, we examine the evolution of the circulation pattern for each event. Figure 6 shows the     December, leading to northward moisture transport and thus the first persistent rainy event. Contour from −0.14 to 0.14 by 0.02

Associated with the low-level circulation evolution is the development of a wave train in the upper tropo-
Contour from −0.14 to 0.18 by 0.02 Contour from −0.14 to 0.14 by 0.02 Contour from −0.12 to 0.14 by 0.02  Fig. 6. Evolutions of (left) 700-hPa relative humidity (shaded; %) and 500-hPa vertical p-velocity (contour; Pa s −1 ) anomaly fields and (right) 850-hPa geopotential height (shaded; gpm) and wind (vector; m s −1 ) anomaly fields associated with the QBWO from 30 November to 6 December 2018. Letter "C" denotes the anomalous cyclonic center that affected rainfall in Shanghai.
APRIL 2020 sphere (Fig. 7). A positive geopotential height center at 200 hPa appeared around 60°N, 40°E on 30 November, and then moved eastward. Meanwhile, a negative height center appeared at 70°N, 100°E on 30 November, and then moved southeastward. A northwest-southeast oriented wave train was clearly seen on 2 December 2018. The wave train intensified as it propagated eastward. The formation of the upper-tropospheric wave train was caused by Rossby wave energy dispersion, as seen from the upper-tropospheric wave activity flux field (Fig.  8a). Given the equivalent barotropic vertical structure of the wave train (Fig. 8b), the low-frequency development and evolution of the upper-tropospheric wave train were at least partially responsible for the maintenance of the low pressure system at 850 hPa west of Shanghai.
The second persistent rainy event was also associated with the southward progression of ascending motion anomaly from 40°N on 4 January to near 30°N on 10 January (Fig. 9). The geopotential height anomaly at 850 hPa showed a similar southward migration, even though the low pressure center appeared west of the anomalous ascending motion center. Physically it is reasonable because ascending motion usually appears to the east of a trough, according to the quasi-geostrophic dynamics.
Again, pronounced southerly anomalies to the east of the low-level low pressure system persisted for a few     days, from 4 to 10 January 2019 (Fig. 9). These winds were against the winter mean climatological flow, and caused the increase of moisture in Shanghai. Such lowfrequency circulation and moisture patterns favored the Contour from −0.12 to 0.14 by 0.02 Contour from −0.12 to 0.16 by 0.02 Fig. 9. Evolutions of (left) 850-hPa relative humidity (shaded; %) and 500-hPa vertical p-velocity (contour; Pa s −1 ) and (right) 850-hPa geopotential height (shaded; gpm) and wind (vector; m s −1 ) anomalies during the second persistent rainy event from 4 to 10 January 2019. Letter "C" denotes the anomalous cyclonic center that affected Shanghai rainfall.  Contour from −0.28 to 0.14 by 0.02 Contour from −0.24 to 0.12 by 0.02 Fig. 10. Evolutions of (left) 925-hPa relative humidity (shaded; %) and 600-hPa vertical p-velocity (contour; Pa s −1 ) and (right) 850-hPa geopotential height (shaded; gpm) and wind (vector; m s −1 ) anomalies during the third persistent rainy event from 9 to 15 February 2019. Letter "A" denotes the anomalous anticyclonic center. the second event was primarily caused by low-frequency wind anomalies in lower troposphere.
Anomalous circulation patterns associated with the third rainy event were quite distinctive from the first two events in that a high pressure anomaly, rather than a low pressure anomaly, appeared in lower troposphere (Fig.  10, right panel). The high pressure anomaly was located along the coast of China. It induced a low-level anticyclonic flow, with pronounced southerly anomalies appearing to the west of the high pressure center. Because of the persistence of the high pressure system, the southerly anomalies also persisted. They were responsible for the increase of local moisture in Shanghai (Fig. 10, left panel) and were thus responsible for the occurrence of the third rainy event.

Conclusions and discussion
The observational analysis of daily rainfall in Shanghai Station shows that the 2018/2019 winter possessed the longest rainy days since 1988. What caused such a distinctive characteristic was investigated from both the seasonal and sub-seasonal circulation change perspective. A regression analysis using the past 32-yr data indicates that statistically a long persistent rainy winter appeared when an El Niño happened in the equatorial Pacific. The positive SST anomaly in the eastern equatorial Pacific may induce a teleconnection pattern in the East Asian coast and western North Pacific, causing southerly anomalies over East China. The moisture transport associated with the southerly anomalies favored the occurrence and maintenance of rainfall over Shanghai region. The SST, circulation, and moisture patterns in 2018/2019 winter in general resembled those derived from the regression analysis.
A further analysis of power spectrum of daily rainfall in Shanghai Station reveals that the rainfall time series was dominated by a 10-25-day period, and each of the persistent rainy events was associated with the QBWO mode. In total, three persistent rainy events were identified for the 2018/2019 winter. The first two events were associated with a low pressure anomaly system in the upstream, which gradually propagated southeastward. Lowlevel southerly anomalies associated with the low pressure system advected abundant moisture from tropical oceans to central China, leading to the persistent rainfall events. The third event, on the other hand, was associated with a high pressure anomaly system over and to the east of Shanghai region. Anomalous southerlies to the west of the positive pressure anomaly favored low-level moisture transport and thus the occurrence of rainfall.
The observational analysis above suggests that the exceptional longest rainy days in 2018/2019 winter were attributed to both the favorable seasonal mean flow and the strong sub-seasonal variability. Therefore, there is urgent need to conduct sub-seasonal to seasonal predictions in order to forecast exceptional climate events such as that over Shanghai in 2018/2019 winter.
While the current analysis points out the importance of upscale feedback of the sub-seasonal mode on seasonal mean climate anomalies, it is not clear how the sub-seasonal mode variability was influenced by the seasonal circulation anomaly and what specific processes through which the sub-seasonal mode could feed back to the interannual anomaly were. For example, in the feedback of higher-frequency modes to lower-frequency variability, does a higher intensity of the QBWO correspond to a greater seasonal mean precipitation anomaly (Qi et al., 2008;Li et al., 2015)? If this is the case, then a simple linear accumulation would work. On the other hand, it would be interesting to understand the nonlinear rectification process . For example, a positive (negative) boundary layer convergence anomaly and a positive (negative) moisture anomaly in situ associated with the QBWO mode would lead to a strengthening (weakening) of seasonal mean precipitation through an upscale feedback process similar to that proposed by Zhu et al. (2019). If both the linear and nonlinear processes were operated, one may further address their relative importance. These issues require further in-depth diagnoses and theoretical and modeling investigations.