The interdecadal variations and causes of the relationship between Autumn Precipitation Anomalies in Eastern China and SSTA over the Southeastern tropical Indian Ocean

Eastern China has a large population with rapid development of the economy, where is the important crop producing region. In this region, the spatial and temporal distribution of autumn rainfall in Eastern China is uneven, which has important societal impact. Using the NCEP–NCAR reanalysis and other observational datasets, it is found that the spatial distribution of the first EOF mode of autumn rainfall anomalies in eastern China is consistent across the region, with significant interannual variabilities. Pronounced interdecadal variations are presented in the relationship between autumn rainfall anomalies in eastern China and sea-surface temperature anomaly (SSTA) over the southeastern tropical Indian Ocean (SETIO). The interdecadal changes have been analyzed by considering two epochs: one during 1979–2004 and the other during 2005–2019. It shows weak and insignificant correlations between the autumn rainfall anomalies in eastern China and SSTA over SETIO during the first epoch. On the other hand, they are remarkable and positively correlated with each other during the second epoch. The inter-decadal changes of the above relationship are related to the warming of SST over SETIO during the second epoch. It causes stronger low-level convergence and ascending motion over SETIO, with the co-occurrence of enhanced western Pacific subtropical high and anomalous abundant moisture over eastern China carried by a low-level southerly anomaly originating from the South China Sea. Simultaneously, the local Hadley circulation over eastern China becomes weak, corresponding to the anomalous ascending motion. The collaboration of anomalous water vapour transport and ascending motion strengthens the connection between the SETIO SSTA and the autumn precipitation anomalies in eastern China, and vice versa. In the boreal autumn of 2019, entire eastern China suffered extreme drought. It suggests that this drought event in eastern China is strongly affected by the negative SSTA over SETIO, which is consistent with the statistical results.


Introduction
Eastern China is adjacent to the Northwest Pacific Ocean, and locates in the East Asian monsoon region, with a vast territory and complex climatic fluctuations (Tao and Chen 1987;Ding 2004). As is well known, eastern China is the most economically developed and densely populated area of China, with a geographical concentration of large cities. Within this large area, the rainy season is concentrated in summer (Ding 2004;Jin et al. 2015). The temporal and spatial variation of summer precipitation in eastern China and its causes have received great attention in recent years (e.g., Tao and Chen 1987;Huang and Wu 1989;Ding 2004;Ding et al. 2008;Huang et al. 2012;Jin et al. 2015). Likewise, there are significant interannual and interdecadal variability in the autumn precipitation anomalies in eastern China (Ying et al. 2015;Yuan and Wang 2019), which could cause severe climate impacts. The extreme drought event in eastern China during the autumn in 2019 produced damaging impacts on local residential life as well as societal and economic development.
The precipitation anomalies in eastern China is regulated by many factors, such as El Niño-Southern Oscillation (ENSO) and sea surface temperature anomaly (SSTA) in the Indian Ocean (Wang et al. 2000(Wang et al. , 2013Saji and Yamagata 2003;Guan and Yamagata 2003;Xie et al. 2009;Jin et al. 2017). ENSO is the leading mode of climate variability in tropical regions. Its warm phase, namely the El Niño event, can affect the western Pacific subtropical high through local air-sea interactions (Wang et al. 2000(Wang et al. , 2013, which in turn affects precipitation anomalies in eastern China (e.g., Wu and Wang 2002;Jin et al. 2016). The Indian Ocean basin-scale warming (Chambers et al. 1999;Xie et al. 2009) can also affect the atmospheric circulation anomalies in East Asia through the "capacitor" effect (Xie et al. 2009) or the "two-stage thermal adaptation" mechanism , and further affects the climate anomalies in East Asia (Wu et al. 2010a, b;He et al. 2015;Xie et al. 2016).
The Indian Ocean Dipole (IOD; Saji et al. 1999) is one of the important air-sea coupling modes in the tropical Indian Ocean and has a significant impact on the climate anomalies in the neighboring regions (e.g., Ashok et al. 2001Ashok et al. , 2004Saji and Yamgata 2003;Preethi et al. 2015;Hameed 2018). Southeastern Tropical Indian Ocean (SETIO)  is the eastern pole of IOD, where is close to the tropical western Pacific Ocean. The SSTA over SETIO appears to affect the climate anomaly of the adjacent area (e.g., Wu et al. 2010a, b;Chen et al. 2016;Huo and Jin 2016). The SETIO SSTA could influence the anomalous precipitation over southern China by adjusting the anomalous water vapour transport and the local Hadley circulation . The above relationship has significant interdecadal changes (Chen et al. 2016;Wu et al. 2010a, b).
Previous studies documented the remarkable interdecadal variations of the anomalous autumn precipitation over eastern China. Before the 1990s, the autumn precipitation anomalies in eastern China presented a quasi-meridional dipole mode, which appeared to be not obvious after that. Yuan and Wang (2019) suggested that this change is closely associated with the interdecadal variations of ENSO. The above conclusions motivated this study. If the ENSO signal is filtered out, is there any possible connection can be found between the autumn precipitation anomalies in eastern China and SSTA in SETIO? If the answer is true, what are the possible mechanisms behind this relationship? It is also unclear whether the link between the two is stable. In the present paper, we investigated these possibilities by looking at the autumn rainfall anomaly over eastern China and its association with the SETIO SSTA on the interdecadal timescale. We further analyzed the possible mechanisms of SETIO SSTA influencing the extreme drought event in eastern China during the autumn of 2019.
This paper is organized as follows. Section 2 introduces the data and method that are used. Section 3 explores the temporal and spatial variations of autumn precipitation in eastern China. In Sect. 4, the interdecadal variations of the relationship between autumn rainfall anomaly over eastern China and the SETIO SSTA are investigated. The mechanism behind the above relationship is discussed in Sect. 5. In Sect. 6, we discuss the influence of the negative SETIO SSTA on the extreme drought event in eastern China during the autumn of 2019. The conclusions are given in the last section.

Data and methodology
We obtained rainfall observations of 2479 stations in China at daily resolution from the National Meteorological Information Center of China Meteorological Administration. The station data have been interpolated by using iterative improvement objective analysis (Cressman 1959), and the influencing radius array is set up to (/1°, 0.5°, 0.25°, 0.1°/). From this we chose 604 stations located in eastern China for our analysis (Fig. 1a), after excluding those with data missing. The NCEP-NCAR reanalysis data (Kalnay et al. 1996) with the horizontal resolution of 2.5°×2.5°, and the Extended Reconstructed Sea Surface Temperature (ERSST v3b, Smith et al. 2008) data from the National Oceanic and Atmospheric Administration (NOAA) with a horizontal resolution of 2°×2° are also used in this study. The COBE (Ishii et al. 2005) and HadISST (Rayner et al. 2003) SST data are also used to verify the interdecadal variations in the relationship between autumn precipitation in eastern China and the SETIO SSTA. All these datasets cover the period of 1979-2019. Boreal autumn months in the present work are September-November (SON). Eastern China [110-124 °E, 24-35 °N] is the mainland China area as shown in Fig. 1a.
To focus on the interannual variations of SSTA over the tropical Indian Ocean, long-term trends are removed Red and Blue thick lines represent the Yangtze River in (a, b) and d, respectively. The value on the upper right corner of (c) indicates the correlation coefficient between the time coefficient of the first EOF mode and precipitation anomaly averaged over the eastern China through least squares method, and 9 years running mean is also utilized to exclude the decadal and longer signals but in Figs. 1a and 8a, and c. The Nino3.4 index used in this paper is the regional average SSTA in the geographic region [170-120 °W, 5 °S-5 °N]. The ENSO signal can affect the Indian Ocean SSTA (e.g., Saji and Yamagata 2003;Xie et al. 2009;Doi et al. 2020), and rainfall anomaly over eastern China Yuan and Wang 2019). To explore the relationship between rainfall anomaly over eastern China and Indian Ocean SSTA, ENSO signal indicated by the Nino3.4 index is excluded by using the linear regression method as follows: In Eq. (1), Y stands for the time series of variables of interest. I Nino3.4 stands for the normalized time series of the autumn ENSO index. Y r stands for the residual value of the variable Y after the ENSO signal is removed. Coefficient α is the covariance of Y and I Nino3.4 .

Temporal and spatial variations of autumn rainfall over eastern China
Climatologically, the distribution of autumn precipitation over eastern China is very uneven in spatial. It decreases in amplitude from the south to the north (Fig. 1a). Coastal areas of southern and southeastern China receive the most abundant rainfall with values exceeding 300 mm. The maximum value is larger than 360 mm. Markedly low rainfall is found to the north of 33 °N, with minimum values smaller than 150 mm. It is noted that autumn precipitation in eastern China accounts for about 15-25% of the annual precipitation, and the region with the largest proportion is located in the northern part of eastern China, which is over 25%. This also suggests that in addition to the large share of summer precipitation in the annual precipitation in eastern China (Jin et al. 2015), autumn precipitation also makes an important contribution to the annual precipitation in the region. Based on the empirical orthogonal function (EOF) analysis, the first EOF mode (EOF1) shows spatially consistent autumn precipitation anomalies (Fig. 1b), consistent with the previous study (Yuan and Wang 2019). EOF1 explains about 29.8% of the total variances, variance contribution of the second EOF mode (EOF2) is 15.9% (figure not shown). And the EOF1 is well separated from the EOF2 (North 1982). Higher loadings are located in central and southern South China, whereas lower loadings are distributed to the north of the Yangtze River. The time series of EOF1 ( Fig. 1c) presents significant interdecadal variability. The correlation coefficient between the time sequence of EOF1 and mean autumn precipitation anomalies over eastern China is as high as 0.93. It further demonstrates that the main mode of the autumn precipitation anomalies over eastern China is consistent across the whole region.
Noted that normalized time series of EOF1 and mean autumn precipitation anomalies over eastern China in 2019 are both smaller than their minus two standard deviations (Fig. 1c), which are their minimum values during the recent 40 years. Figure 1d shows that the mean autumn precipitation anomaly in 2019 and its percentage are negative over eastern China. The minimum rainfall areas are located in the central and adjacent areas of Jiangxi Province, southern China, and the coastal areas of Jiangsu and Zhejiang Province. The minimum precipitation value is less than − 250 mm, with a minimum percentage of less than − 90%.

Relationship between the autumn precipitation anomalies over eastern China and southeastern tropical Indian Ocean SSTA
As concluded by Saji et al. (1999), the positive phase of IOD is described as the negative SSTA over SETIO and positive SSTA over the western tropical Indian Ocean. The IOD presents an extreme positive phase during the autumn in 2019 (Doi et al. 2020), corresponding to the negative anomalous autumn precipitation over eastern China as discussed above. It inspires us to explore the relationship between the extreme autumn drought over eastern China and the SSTA over the tropical Indian Ocean in 2019.
To answer this question, we calculate the correlation coefficients of the precipitation anomaly averaged over eastern China (P EC ) and SSTA averaged over SETIO (I SETIO ), the western pole of IOD and IOD index, which is 0.39, 0.06 and − 0.22, respectively. Since ENSO is closely associated with IOD (Saji and Yamagata 2003;Doi et al. 2020) and also the precipitation anomaly over eastern China Yuan and Wang 2019), the above correlation coefficients are calculated by further excluding the ENSO signal. After filtering the ENSO signal, the correlation coefficients between P EC and the western pole of IOD, P EC and IOD index are still not strong with the value of 0.21 and − 0.15, respectively. The correlation coefficient between P EC and I SETIO is 0.36, indicating their intrinsic relations. Time series of I SETIO (Fig. 2a) suggests that I SETIO is dramatically anomalous negative in 2019, with the value of -1.79 standard deviations.
To further clarify the relationship between P EC and I SETIO , we calculate the 11-years sliding correlation coefficient between the two, showing a significant interdecadal variation between P EC and I SETIO (Fig. 2b). Before 2005, the correlation between P EC and I SETIO gradually evolves from a weak positive correlation to a weak negative correlation. After that, P EC is closely related to I SETIO with the correlation coefficient during 2010-2014, exceeding the 95% significance level. Thus, we consider 2005-2019 as the closely related period, while 1979-2004 as the unrelated period. We also calculated the 21-year sliding correlation between P EC and I SETIO (Fig. 2b). It can be seen that the interannual relationship between the two enhances in the early 21st century, suggesting that this interdecadal enhancement of the relationship is not dependent on the window length. The correlation coefficients between P EC and I SETIO during 1979SETIO during -2004SETIO during and 2005SETIO during -2019 are 0.04 and 0.74, respectively. It indicates that precipitation anomaly over eastern China and SETIO SSTA has an enhanced interdecadal relationship. During 1979-2004, P EC is positively related to SSTA over the central Tropical Indian Ocean with values smaller than 0.4 in most areas (Fig. 2c). But P EC is significantly associated with the SETIO SSTA during 2005-2019 (Fig. 2d), with values larger than 0.4 in most areas. The results obtained using COBE and HadISST data are consistent with the above conclusions (figure omitted).
As shown in Fig. 3a, SETIO SSTA is weakly correlated with the P EC over limited areas of coastal southeastern China and to the north of the Yangtze River during 1979-2004. During 2005-2019, evident positive correlations between P EC and I SETIO are broadly distributed over most areas of eastern China (Fig. 3b), especially over the middle and lower reaches of the Yangtze River and South China. The scatter plot of I SETIO and P EC during 1979-2019 (Fig. 4a) suggests that even positive correlations between I SETIO and P EC are evident, most of I SETIO and P EC pairs are away from the linear fitted line indicating their weak linear relation. As further shown in Fig. 4b, c, P EC is not related to I SETIO during 1979-2004, but is highly related to I SETIO during 2005-2019 with a linear fitted index of 0.83.

Possible mechanisms
To discover the possible mechanisms resulting in the interdecadal variability of the relationship between P EC and SETIO SSTA, especially the closer relationship after 2005, the atmospheric circulation anomalies associated with the SETIO SSTA are analyzed in the following.

Anomalous stream function and rotational wind
During 1979-2004, the 850-hPa stream function and rotational wind regressed from I SETIO (Fig. 5a) show that there exists a pair of anomalous cyclonic circulations and anomalous westerly flows over the equatorial Indian Ocean stretching from the tropical to subtropical regions, due to the Mastuno-Gill response to the SETIO SSTA (Mastuno 1966;Gill 1980). The anomalous cyclonic circulation in the northern hemisphere is more zonally-elongated (40-140 °E).  Eastern China is under the influence of this anomalous cyclonic circulation. Vice versa. As shown in Fig. 5b, the anomalous cyclonic circulation is located from the Bay of Bengal to the Indochina Peninsula and adjacent areas during 2005-2019, corresponding to the anomalous cyclonic circulation in the southern hemisphere. Due to the Kelvin wave response to SST warming over SETIO (Mastuno 1966;Gill 1980), the easterly flow anomalies over the western Pacific Ocean could enhance the western Pacific subtropical high and contribute to the anomalous southerly flows over eastern China.
Comparing the anomalous circulation associated with I SETIO during the two periods (1979-2004 vs. 2005-2019), there both exist the response of Rossby wave associated with the positive SETIO SSTA. The anomalous cyclonic circulation is more zonally-elongated during 1979-2004, whereas is confined in relatively small areas over South China and Indochina Peninsula during 2005-2019, associated with the intensification and westward extension of the western Pacific subtropical high.

Anomalous water vapour transport
From 1979 to 2004, the integrated water vapour flux and its divergence regressed from I SETIO (Fig. 6a) show that there is anomalous water vapour transport from the west to the east over the tropical Indian Ocean, resulting in the anomalous water vapour convergence over SETIO. Part of the water vapour is transported from the tropical Indian Ocean, passing through Java Islands, the South China Sea and the Kuroshio area, then transported westward to eastern China, and finally to South Asia along the southern foot of the Tibetan Plateau. Anomalous water vapour divergence is distributed in some areas of eastern China, which is consistent with the anomalous circulation in the lower troposphere (Fig. 5a).
As for the period of 2005-2019, a pair of cyclonic circulations is located along the equatorial Indian Ocean, corresponding to the eastward water vapour transport anomaly (Fig. 6b). Anomalous water vapour flux over the tropical western Pacific Ocean is transported to eastern China along the edge of the western Pacific subtropical high. The water vapour is anomalous convergence over eastern China, favoring the above-normal autumn precipitation over eastern China. Vice versa. During 1979During -2004, the positive SETIO SSTA corresponds to the anomalous lower-level convergence (Fig. 7a), upperlevel divergence (Fig. 7b) and associated ascending motion (Fig. 7c) over the Maritime Continent. To the opposite, the tropical western Indian Ocean and the African continent are regulated by anomalous lower-level divergence (Fig. 7a), upper-level convergence (Fig. 7b) and associated descending motion (Fig. 7c). Whereas eastern China has no anomalous vertical motions (Fig. 7c).

Vertical motion analysis
Similar to the period of 1979-2004, SETIO is featured with the anomalous lower-level convergence (Fig. 7d), upper-level divergence (Fig. 7e) and associated ascending   (Fig. 7d). Noted that the anomalous convergence, divergence and ascending motions over SETIO during 2005-2019 are all stronger than those during 1979-2004. Eastern China is controlled by the anomalous convergence at the lower troposphere, divergence at the upper troposphere, and ascending motions, which is favorable for the above-normal autumn precipitation over eastern China.
The climatological mean equatorial zonal-vertical circulation over the tropical eastern Indian Ocean and Pacific Ocean (Fig. 8a) indicates that areas from the tropical eastern Indian Ocean to the central Pacific Ocean are controlled by anomalous ascending motion, whereas the tropical eastern Pacific Ocean is dominated by anomalous descending motion, that is known as the Walker Circulation. The difference of the unfiltered SSTA averaged over SETIO between the two periods, i.e. 2005-2019  (Fig. 8b) present the strengthened anomalous ascending motion over tropical eastern Indian Ocean and anomalous descending motion over tropical western Pacific Ocean. This would produce the anomalous low-level divergence over the western Pacific Ocean, and strengthen the western Pacific subtropical high (Figs. 5 and 7a, b). The integrated water vapour transported from the western Pacific Ocean to eastern China can subsequently be strengthened, providing abundant water vapour for the enhancement of the connection between SETIO SSTA and the anomalous autumn precipitation in eastern China.
As shown in Fig. 8c, the ascending branch of eastern China is distributed between the equator to 20 °N, with two descending branches locating at its north and south. The difference of regressed zonal-vertical circulation between the two periods demonstrates that the warming SETIO SSTA is corresponding to the enhancement of anomalous ascending motion over SETIO, weakening of the ascending motion at 5-20 °N and subsequently strengthened ascending motion over eastern China (25-30 °N). It shows a circulation pattern of strengthened local Hadley circulation over eastern China, providing favorable vertical motion conditions for the enhanced connection between SETIO SSTA and the anomalous autumn precipitation in eastern China.

Analysis of the extreme drought in 2019
The spatial distribution of SSTA over the tropical Indian Ocean in 2019 presents the negative SSTA over SETIO with a minimum of less than − 1 °C (Fig. 9a). A pair of anti-cyclonic circulations on 850 hPa (Fig. 9b) is induced across the equator to the west of SETIO, and is corresponding to the anomalous westerlies along the equator (Mastuno 1966;Gill, 1980). The anomalous cyclonic circulation over the northwest Pacific Ocean is indicative of a weakening northwest Pacific subtropical high (Fig. 9b). SETIO is characterized by the anomalous divergence in the lower troposphere (Fig. 9c) and anomalous convergence in the upper troposphere (Fig. 9d). On the contrary, the northwest Pacific Ocean is controlled by the anomalous convergence in the lower troposphere (Fig. 9c) and anomalous divergence in the upper troposphere (Fig. 9d). The enhanced locally Hadley circulation over eastern China consists of the anomalous subsidence over SETIO and eastern China, and the anomalous ascending motion over the South China Sea (Fig. 9e). It is also noted that a water vapour transport anomaly toward the equator over eastern China (Fig. 9f) and reducing the

Discussion and conclusions
This study investigates the interdecadal variations of the relationship between autumn precipitation anomaly over eastern China and SSTA over SETIO during the past 40 years . Possible mechanisms of the negative anomaly of SETIO SSTA on the autumn drought in eastern China in 2019 are also discussed. The conclusions are as follows: (1) Autumn precipitation over eastern China is unevenly distributed in spatial and temporal perspective, with significant interannual variability. The first EOF mode of autumn precipitation anomaly over eastern China is consistent across the whole region. The autumn precipitation anomaly over eastern China in 2019 is smaller than minus two standard deviations.
(2) The relationship between autumn precipitation anomalies in eastern China and SETIO SSTA has significant interdecadal changes. Before 2005, there is no signifi-  We reveal that the link between the autumn precipitation in eastern China and the SETIO SSTA intensified during 2005-2019, which is related to the anomalous atmospheric circulations that are more favorable to the autumn precipitation in eastern China caused by the SETIO SST warming during this period. The Indian Ocean SSTA, such as IOD (Han et al. 2014;Hameed 2018) has significant interdecadal variability. The summer IOD can excite downstream propagating quasi-stationary wave trains in the Mediterranean (Enomoto et al. 2003) through the monsoon-desert mechanism (Rodwell and Hoskins 1996) and affect climate anomalies in East Asia . Could the autumn SETIO SSTA regulate climate anomalies in East Asia through the above-mentioned pathway and thus influence the interdecadal variabilities of autumn precipitation in eastern China? If the above hypothesis is true, what are the involving dynamical and thermodynamic mechanisms? That's an interesting question.
The results in this paper are based on observations and reanalysis data. However, the instability of the relationship between two series might be due to the internal variability in either of the correlated drivers (Gershunov et al. 2001). Due to the limited observational data sample, we will use the historical simulations outputs from Phase 6 of the Coupled Model Intercomparison Project (CMIP6; Eyring et al. 2016) to verify the results obtained in this paper. In addition, sensitivity experiments linked to SETIO SSTA are planned to analyze the response of autumn precipitation anomalies in eastern China by using the atmospheric general circulation model with the 1979-2004 and 2005-2019 climate-averaged SST fields as background fields, respectively. This will verify the stability of the results based on observations and reanalysis data. The above issues will be further investigated in the future. It should be noted that the EOF1 of the autumn precipitation anomaly in eastern China is consistent across the region, and the EOF2 presents a north-south dipole pattern. Such dipole mode disappeared in the mid-to-late 1980s, and its relationship with ENSO has interdecadal variations. (Yuan and Wang 2019). Studies have shown that ENSO and summer precipitation in eastern China, ENSO and Indian monsoon precipitation have significant interdecadal variations (Ashok et al. 2014;Jin et al. 2016). And there is also a link between summer precipitation in eastern China and monsoon precipitation in India (Zhang et al. 2021a, b). Is this relationship also subject to interdecadal variations due to ENSO regulation? It requires further research to explore whether the disappearance of the dipole mode of China autumn precipitation is related to the warming SST over SETIO. It is still unknown why the autumn precipitation is related to the SSTA over SETIO where is the eastern pole of IOD (Saji et al. 1999), but shows little correlations with the SSTA over the western pole of IOD. For the period of 2005-2019, the correlation coefficient between the P EC and Niño4 index is -0.38 after filtering out the SETIO signals. And the correlation coefficient between P EC and I SETIO is 0.68 after eliminating Niño4 effects,. That is, ENSO and SETIO could independently explain 14.44% and 46.24% of the variance contribution of autumn precipitation variability in eastern China, respectively. The cooperative influences of ENSO and SETIO on autumn precipitation anomalies in eastern China are also worth studying. The influences of the SSTA over the tropical Atlantic Ocean on the autumn precipitation also need further research in the future, considering the close relationship between the SSTA over the tropical Atlantic Ocean and eastern Asian climate anomalies (Ham et al. 2013;Chen et al. 2015;Huo et al. 2015;Chen et al. 2018;Jin and Huo 2018;Zuo et al. 2019a, b).

Declarations
Conflict of interest The authors declare that they have no conflict of interest.