Interdecadal change in autumn rainfall over Southeast China and its association with tropical Pacific SST

In this paper, Japan Meteorological Agency (JRA-55) reanalysis and observational rainfall datasets from the National Climate Center (NCC) of China, as well as satellite datasets from the Tropical Rainfall Measuring Mission (TRMM), Global Precipitation Climatology Project (GPCP), and International Satellite Cloud Climatology Project (ISCCP), are used. The correlation coefficient and fast Fourier transform (FFT) low-pass filter are also used, in order to reveal the interdecadal decrease in autumn rainfall in Southeast China (SEC) after 1990. The close and robust relationship between the interdecadal variation in autumn rainfall in SEC and sea surface temperature (SST) in the tropical Pacific is investigated. The most significant and stable region of correlation is located in 10° S–10° N, 160° E–160° W, in which there also exists interdecadal warming after 1990. Furthermore, the interdecadal warming of SST can induce Gill responses of the atmosphere: a cyclone anomaly is produced on each side of the equator in the lower troposphere, with a westerly anomaly to the west of the dateline, and an anticyclone anomaly is produced in the upper troposphere. In particular, the cyclone anomaly on the northern side of the equator is located in the Northwest Pacific (NWP), and its ambient northerly airflow weakens meridional water vapor transport, as well as the local descending motion and low-troposphere divergence, in favor of the interdecadal decrease in SEC rainfall after 1990. In addition, the sensitive experiments with ECHAM-5.4 model also confirm that the interdecadal warming in the region (10° S–10° N, 160° E–160° W) would motivate the atmospheric Gill response and thereby cause the sinking motion in SEC and support the interdecadal decrease in autumn rainfall in SEC.


Introduction
The main meteorological disasters in China include droughts, floods, typhoons, high temperatures, cold waves, low-temperature freezing, sandstorms, hail, and tornadoes. Among them, drought and flood are the primary agricultural meteorological disasters (Zhang et al. 2008). Especially in the context of global warming, extreme weather and climate events and associated meteorological drought and flood disasters have resulted in some new changes. With the continuous development of agricultural resources in China, the ecosystem is relatively fragile, and drought affects the vegetation in the corresponding areas , thus affecting the stability of the ecosystem. Analysis of drought and flood changes is conducive to the ecological construction of China. Second, China is a large agricultural country with a large population. To ensure food security and reduce economic losses, the study of the variability of regional precipitation and associated drought and flood causes has become an important topic in China's climate research.
Southeast China (SEC) includes Fujian, Guangdong, East Guangxi, Jiangxi, Hunan, South Hubei, and South Zhejiang, which is economically booming and a crucial tropical and subtropical agricultural production base in China (Liu et al. 2019). Therefore, the climatology and variation features of rainfall in SEC, as well as associated droughts and floods, have received much attention (Xin et al. 2006;Qiu et al. 2009;Wan et al. 2009;Li et al. 2016aLi et al. , 2016bJin et al. 2017;Liu et al. 2019).
Previous studies have provided some explanation for the interdecadal shift of rainfall in all seasons in SEC, but among them, studies linked to autumn is relatively less (Xin et al., 2006;Wu et al. 2010;Feng and Li 2011;Huang et al. 2013). In the past 50 years, the spring in South China has shown a marked drought trend (Zuo et al. 2012), which is attributed to the interdecadal variation in the North Atlantic Oscillation (NAO) in the preceding winter (Xin et al. 2006), increase in long-lasting El Niño events, reduction in La Niña episodes (Qiu et al. 2009), increase in anthropogenic aerosol emissions (Hu and Liu 2013), interdecadal shift of Eurasian spring snow after the 1980s (Zuo et al. 2012), and Pacific Decadal Oscillation (PDO) (Wu and Mao 2016). The summer precipitation in South China shows a remarkable interdecadal variation accompanied by increasing flood events (Nitta and Hu 1996;Weng et al. 1999;Ding et al. 2008;Xing et al. 2016), which has been attributed to the increased heating in the tropical Pacific and the tropical Indian Ocean (Chang et al. 2000;Gong and Ho 2002;Yang and Lau 2004), internal dynamics of the atmosphere (Si et al. 2016), weakening of the Asian monsoon (Ding et al. 2008), interdecadal reduction in Eurasian snow water equivalent and increase in tropical cyclones in the South China Sea (SCS) in approximately 1993 Chen et al. 2012), and rising emissions of black carbon aerosols (Menon et al. 2002). For winter, the rainfall over South China has also increased significantly in recent decades, with increased moisture over SEC and decreasing snowfall totals (Zhai et al. 2005;Gill 1980;Zhang et al. 2014), which is attributed to the interdecadal shift of the El Niño-Southern Oscillation (ENSO), East Asian winter monsoon (EAWM), Arctic Oscillation (AO), West Pacific sea surface temperature (SST), and SCS SST (Zhou and Wu 2010;Zhou, 2011a, b;Yang and Huang 2021). For example, a weak EAWM is accompanied by anomalous southwesterly winds over the SCS, anomalous moisture convergence, and upward motion, which enhances precipitation in South China (Zhou and Wu 2010;Zhou 2011a, b;Yang and Huang 2021;Zhou 2011a, b).
Autumn is the transition period of the atmospheric circulation pattern from summer to winter, and the atmospheric circulation in SEC changes significantly, such as the withdrawal of the subtropical high from land to ocean and the gradual development of the winter monsoon. During these changes in atmospheric circulation, the autumn precipitation in SEC shows salient uniqueness. As shown in Fig. 1, the shading regions, mainly concentrating in SEC, were obtained through a significant t test at the 95% confidence level. It indicates that the autumn precipitation in SEC is weakly related to precipitation in the rest of China. Therefore, this study focuses on the rainfall in SEC to highlight its uniqueness.
In this region, autumn droughts occur more frequently than in other seasons (Jian et al. 2008), and in particular, severe droughts are most likely to occur in southeastern SEC (Chen et al. 2014a, b). The autumn rainfall in South China is a positive anomaly, accompanied by the ridge of the Western Pacific subtropical high located in southern South China, a relatively strong subtropical high, a relatively warm SST in the Western Pacific and North Pacific, and a relatively cold SST in the South Indian Ocean (Niu and Li 2008). In recent decades, the increase in aerosols has benefited the interdecadal decrease in autumn rainfall in SEC by raising atmospheric stability, lessening convective available potential energy, and reducing cloud effective particle size . Chen et al. (2014a, b) analyzed the variation in autumn drought in South China after the late 1980s. Autumn droughts occurred frequently, and the most strikingly sudden change was after 2002, with an aggravation of drought in this region (Chen et al. 2014b). Lin et al. (2012) reported that the years with autumn droughts are linked with slow adjustment of sea level pressure and inactive cold air, and The thick line represents the zero line, and the contour interval is 0.1. The shaded areas were obtained from a significance t test at the 95% confidence level, the rectangle denotes SEC (22.5° N-30° N, 110° E-120° E), and the dots represent the meteorological stations the acceleration of flash droughts in SEC is closely related to anthropogenic climate change (Wang and Yuan 2021). Jia et al. (2015) pointed out that the reduction in typhoon activity in SEC during autumn was an important factor for the decrease in autumn precipitation in SEC. Autumn precipitation anomalies in SEC has a stronger linkage with SSTA over the southeastern tropical Indian Ocean after 2005 (Huo et al. 2022), and especially, the late autumn rainfall also connects with the process of South China Sea summer monsoon withdrawal (Hu et al. 2020). Moreover, India-Burma Trough, eastern trough of the plateau, and Tibetan Plateau ground heat source could also exert an influence on the autumn rainfall over SEC (Niu and Li 2008;Gu et al. 2012).
Therefore, the factors affecting autumn precipitation over SEC are complex and diverse, but SST in the tropical Pacific could be an important factor. Some previous studies have revealed that the SST in the tropical Pacific has a significant impact on the precipitation in SEC by affecting the atmospheric circulation (Zhang et al. 2011;Wang et al. 2000;Jiang et al., 2014). Zhang et al. (2011) found that CP/EP El Niño events have a significant remote impact on autumn rainfall in SEC through anomalous cyclones and anomalous anticyclones in the NWP (Zhang et al. 2011). Jiang et al. (2014 considered that to the east of the Tibetan Plateau, September rainfall had an interdecadal reduction after the mid-1980s, which may have been related to the variation in SST in the equatorial Central Pacific (Jiang et al. 2014). According to former studies, we note that the relationship between SST in the tropical Pacific and autumn precipitation in SEC is focused on interannual scales. Does the variation in SST in the tropical Pacific cause the interdecadal change in autumn rainfall in SEC? If so, what is the physical linkage? This work explores this issue.
In this paper, the interdecadal change in rainfall in boreal autumn over SEC and its potential cause are investigated in depth. The data and methods used in this study are shown in Section 2. Section 3 confirms the interdecadal shift of autumn precipitation in SEC and associated circulation features. Section 4 investigates the association of the interdecadal warming in the tropical Central Pacific with interdecadal reduction of autumn rainfall in SEC, and Section 5 further explores the physical process of interdecadal warming of the tropical Central Pacific influencing the interdecadal reduction of autumn rainfall in SEC. The summary and conclusions are given in Section 6.

Data and method
SEC is selected as the research region marked by a rectangle in Fig. 1, which spans from 22.5° N-30° N to 110° E-120° E. All stations (74 meteorological stations) in the SEC region labeled by dots in Fig. 1 are chosen from 553-station rainfall dataset provided by the National Climate Center (NCC) of China Meteorological Administration (CMA). The monthly reanalysis data JRA-55 are used (Kobayashi et al. 2015), with a horizontal resolution of 1.25° × 1.25°. The monthly average SST is from HadISST version 1.1 from the Hadley Center, with a horizontal resolution of 1° × 1° (Rayner et al. 2003). Simple Ocean Data Assimilation (SODA) sea temperature is also used in this study. Version 3B43 of the Tropical Rainfall Measuring Mission (TRMM) datasets is provided by the National Space Development Agency of Japan (NASDA) and the National Aeronautics and Space Administration (NASA), and version 2.1 of the Global Precipitation Climatology Project (GPCP) datasets is provided by the Global Precipitation Climatology Center. Monthly total cloud amount and shortwave and longwave radiation datasets are provided by the International Satellite Cloud Climatology Project (ISCCP), referred to as the ISCCP-D2 dataset, with a horizontal resolution of 2.5° × 2.5° (Klein and Hartmann 1993;Marchand et al. 2010). Autumn refers to the mean of September, October, and November. The study focuses on the period from 1971 to 2018.
Correlation, fast Fourier transform (FFT) low-pass filter (remaining 9-year-and-above signals), Le Page mutation analysis (Yonetani 1992), and numerical simulation are all applied in this paper. It is worth noting that the degrees of freedom of the sequence after the FFT low-pass filter decrease significantly. Using the theoretical proof and Monte Carlo method, Yan et al. (2004) designed a scheme for determining the degrees of freedom of a time series after a FFT low-pass filter: , where DOF represents the degrees of freedom, ΔT is the sampling interval, equal to 1 here; T c is the cutoff period for low-pass filtering, equal to 8; and N that is equal to 38 is the data sample number (Yan et al. 2004). By using this formula, the degree of freedom of the filtered sequence is approximately 10 during the research period.
This work makes use of the ECHAM-5.4 model, with T63L19 resolution (with a horizontal resolution of T63 and vertical resolution of 19 layers), to conduct simulation experiments. The corresponding grid datasets have a 1.875° latitudinal step and meridional Gauss lattice, totaling up to 192 (in the meridional direction) × 96 (in the latitudinal direction) grid points.

Interdecadal shift of autumn precipitation in SEC
The standardized rainfall averaged over SEC from 1971 to 2018 is shown in Fig. 2a, and the most notable characteristic of autumn precipitation in SEC is a decreasing trend after 1990 and an increasing trend after 2009. It is clear that the precipitation during 1991-2009 is significantly lower than that during the 1971-1990 period, as seen from the 9-year running average curve. As shown in Fig. 2b, the standardized rainfall sequence is calculated using the TRMM during 1998-2018 and GPCP datasets during 1979-2018, and their correlations with the sequence in Fig. 2a during the same period are 0.95 and 0.96, respectively. Therefore, the interdecadal change of rainfall in SEC also can be discerned in other datasets. The Le Page mutation analysis of this sequence in Fig. 2a further confirms that the interdecadal change years of autumn precipitation in SEC are approximately 1990 and 2009, respectively (Fig. 2c), and the same conclusion can also be obtained by using another mutation analysis method, such as the sliding t test. Since the data spanning 2010-2018 are too short, the present study mainly focuses on the first interdecadal change at approximately 1990. This change can also be confirmed by the difference in autumn rainfall in SEC between 1991-2009 and 1971-1990. As shown in Fig. 2d, the rainfall decreases uniformly over SEC, and the most obvious decrease is up to 30 mm per month in Guangdong and Fujian Provinces, mainly located in the dotted regions. As shown in Fig. 3a, flow fields at 850 hPa differ obviously during 1991-2009 compared to those in the 1971-1990 periods, with a significant cyclonic anomaly and positive vorticity anomaly appearing in the Bay of Bengal (BOB). In addition, a cyclonic anomaly appears obviously in the NWP, and the middle-high latitude circulation over the Western Pacific also changes strikingly.
An easterly anomaly appears in the middle-latitude westerly belt, and it is transported from the ocean to the land and being opposite to the East Asian trough (EAT) climatological circulation pattern, which would cause the weakening of the EAT and the appearance of a negative vorticity anomaly. In general, the most significant circulation changes are in the BOB, NWP, and EAT regions, along with the interdecadal decrease in rainfall in SEC after 1990.
At 200 hPa, there is an anticyclonic anomaly located in 10° N-20° N, 120° E-140° E; a significantly cyclonic anomaly in 20° N-40° N, 100° E-120° E; and a significant weakening EAT, which can affect SEC (Fig. 3b). Because of the weakening EAT, the temperature increases significantly, mainly north of 30° N, and the geopotential height anomaly is positive (Fig. 3c). The divergent wind diverges at low troposphere values, especially in SEC (Fig. 3d). The water vapor flux and its divergence show that the enhanced BOB cyclonic circulation can only transport anomalous water vapor to Northwest China and southwestern Southwest China after 1990, and it cannot transport water vapor into SEC (Fig. 3e). The NWP cyclonic anomaly transports water vapor southward from the ocean into SEC, but water vapor in this region does not have obvious convergence (Fig. 3e), which is consistent with Gu et al. (2012). Anomalous sinking motion prevails in SEC, in which cloud liquid water content decreases significantly at and below 400 hPa (Fig. 3f), which is unfavorable to water vapor condensation and supports the decrease in rainfall. All of these atmospheric changes support the reduction in rainfall over SEC after 1990.

Tropical Central Pacific warming associated with the interdecadal change in autumn rainfall in SEC
The SST anomaly in the tropical Pacific is closely related to autumn rainfall in SEC (Liu et al. 2019;Jia et al. 2015;Gu Fig. 3 The differences in various meteorological variables between 1991-2009 and 1971-1990. Wind (vector, m s −1 ) and vorticity (shading, 10 −5 s −1 ) at a 850 hPa and b 200 hPa, c geopotential height (contour, gpm) and temperature (shading, °C) at 850 hPa, d divergent wind (vector, m s −1 ) and velocity potential (shading, 10 6 m 2 s −1 ) at 850 hPa, e column-integrated water vapor flux from 1000 to 200 hPa (vector, g kg −1 m s −1 ) and its divergence (shading, g kg −1 m 2 s −1 ), and f meridional wind, vertical motion (vector, omega multiplied by − 100), and cloud liquid water content (shading, g kg −1 m 2 s −1 ) cross section along 110° E-120° E. Red arrows are at the 95% confidence level for the vector variable, and the dotted (blue contour) area indicates the 95% confidence level for the variable represented by shading (contour). All datasets are analyzed by a FFT low-pass filter before calculating the difference. SEC is marked by the rectangle, and the violet area indicates the plateau topography et al. 2012); therefore, we also calculate the difference in SST and mixed layer sea temperature between 1991-2009 and 1971-1990 (Fig. 4). The results show that the tropical Indian Ocean and the western and Central Pacific undergo salient interdecadal warming, which occurs at the 95% confidence level but with uneven amplitudes, along with the interdecadal decrease in autumn rainfall in SEC. In particular, the most significant warming area is located in the tropical Pacific, with more than 0.3 °C warming during 1991-2009, compared to the 1971-1990 period, which is marked by a rectangle. Figure 4b also shows that the tropical Central Pacific warming extent in the mixing layer is also the largest, which is at the 95% confidence level. Therefore, the interdecadal warming in the tropical Central Pacific appears not only in the sea surface but also in the sea mixing layer.
To ensure whether the key SST region is located in the tropical Central Pacific, we also calculate the correlation between rainfall over SEC and SST, as displayed in Fig. 5. As shown in Fig. 5a, the correlation between the interdecadal sequence of autumn rainfall in SEC and SST is negative, which indicates that when rainfall in SEC is deficient, significant SST warms up in the tropical and subtropical Pacific (20° S-20° N, 140° E-140° W), SCS, and tropical eastern Indian Ocean. The warm SST anomalies, located in the Indian Ocean and SCS, gradually disappear from the preceding summer to simultaneous fall (Fig. 5a-d). However, the anomaly in the tropical Pacific is still maintained, but the central area shrinks to the region 10° S-10° N, 160° E-160° W (Fig. 5d), the same as the region marked in Fig. 4.
Consequently, we believe that the SST anomaly in the tropical Central Pacific (10° S-10° N, 160° E-160° W) is closely related to the interdecadal decrease in autumn precipitation in SEC. Therefore, a tropical Central Pacific SST index (referred to STI) is defined (Fig. 6a). The STI is calculated as the normalized sequence of SST averaged over 10° S-10° N, 160° E-160° W, which is filtered by a FFT lowpass filter. As displayed in Fig. 6b, the STI also experiences an interdecadal change at approximately 1990, consistent with the autumn rainfall in SEC, as displayed in Fig. 2b. Its correlation with the interdecadal sequence of autumn precipitation in SEC is − 0.60, which is significant at the 95% confidence level. We further confirm that the interdecadal reduction in autumn rainfall in SEC has a significant connection with the interdecadal enhancement of tropical Pacific SST (mainly at 10° S-10° N, 160° E-160° W).

Physical linkage between tropical Pacific warming and the interdecadal decrease in precipitation in SEC
To explore how the STI affects the interdecadal decline of precipitation in SEC, Fig. 7 is shown. As shown in Fig. 7a, with the interdecadal warming in the tropical Central Pacific, an anomalous cyclone is generated on the southern and northern sides of the equator in the low troposphere at 850 hPa, in step with the appearance of a westerly anomaly west of the dateline. However, the anomalous cyclonic circulation at both sides to the equator at 850 hPa is transformed into an anomalous anticyclonic circulation at 200 hPa (Fig. 7b). The configurations of the anomalous circulation in lower and higher layers show that the interdecadal warming of the key sea region in the tropical Central Pacific induces a Gill-type response in the atmosphere (Gillies et al. 2012;Zhang and Krishnamurti 1996;Wu et al. 2000).
Previous studies have emphasized that the anomalous cyclone or anticyclone of the NWP in the lower troposphere is an important atmospheric bridge, conveying the direct and indirect influence of tropical sea temperature on the atmospheric circulation of East Asia. As shown in Fig. 7, the Fig. 4 a The difference in SST between 1991SST between -2009SST between and 1971SST between -1990; b same as a, but for the weighted-mean mixed layer sea temperature (including 5 m, 15 m, 25 m, 35 m, and 46 m) (unit: °C); the dotted areas denote the 95% confidence level. Datasets are processed by a FFT low-pass filter before calculating the difference cyclone anomaly in the NWP in the lower layer at 850 hPa can provide an ambient current to influence SEC. To demonstrate the physical process clearly and how the interdecadal warming in the tropical Pacific influences the interdecadal reduction in rainfall in SEC, Fig. 8 is shown.
The anomalous pattern in Fig. 8a is consistent with that in Fig. 3c, only with more significance in the temperature field at 850 hPa. The positive geopotential height anomaly associated with the STI helps to reduce rainfall in SEC (Fig. 8a). The anomaly of divergent wind diverges in the low troposphere, especially in SEC (Fig. 8b), consistent with that displayed in Fig. 3d. The NWP anomalous cyclone transports southward water vapor from the ocean into SEC, but water vapor in this region is divergent (Fig. 8c), which is consistent with Gu et al. (2012) and Fig. 3e. Anomalous sinking motion prevails in SEC (Fig. 8d), which is unfavorable to water vapor condensation and supports the decrease in rainfall. All of these atmospheric changes associated with the STI support the reduction in rainfall over SEC after 1990. As displayed in Fig. 9, the tropical Central Pacific (10° S-10° N, 160° E-160° W) experiences an interdecadal warming after 1990. In the meantime, an anomalous cyclone is induced in the NWP, accompanied by a local meridional circulation with a subsiding branch in SEC, and these anomalous circulations help to lead to the interdecadal reduction in rainfall in SEC.
To confirm that the interdecadal warming in the tropical Pacific exerts an impact on autumn rainfall in SEC from the atmospheric Gill response, the numerical simulation experiments are designed as follows: the SST anomaly (− 1 ℃ for experiment 1, referred to as EXP1, and 1 ℃ for EXP2) is SST is operated by a FFT low-pass filter, and areas with light (dark) shading indicate a 90% (95%) confidence level. The area marked by the rectangle is 10° S-10° N, 160° E-160° W Fig. 6 a Normalized SST sequence averaged over the area (10° S-10° N, 160° E-160° W) marked by the rectangle in Fig. 5, called the STI and b corresponding to the Le Page mutation analysis. The black line with a value of 5.99 denotes significance at the 95% confidence level. SST is operated by a FFT low-pass filter before calculating the average added over the tropical Central Pacific (10° S-10° N, 160° E-160° W) based on the climatological mean states of SST, and then atmosphere mode ECHAM-5.4.00 is integrated for 22 years.
When the tropical Central Pacific region, 10° S-10° N, 160° E-160° W, has a warming anomaly, an anomalous cyclone is generated at both sides of the equator at 850 hPa in the low troposphere, as shown in Fig. 10a. To the west of the heat source is a westerly anomaly, and to the east of the heat source is an easterly anomaly. In Fig. 10b, in the upper troposphere at 200 hPa, there exists an anomalous anticyclone on both sides of the equator. These results from simulations are consistent with those from the statistical diagnoses in Fig. 7; therefore, it is confirmed that the warming anomaly in the tropical Central Pacific induces a Gill response in the atmosphere and further impacts autumn rainfall in SEC.
As shown in Fig. 10c, anomalous descending motion in SEC (at latitudes of 17.5° N-40° N) prevails when a warming anomaly appears in the tropical Central Pacific, and the descending motion in the statistical diagnoses is mainly located at 15° N-30° N (Fig. 3f), both of which have little difference, but both support the rainfall decrease in SEC. As shown in Fig. 10d, the decrease of rainfall in the northwestern part in SEC can reappear well, only with some deviation about the location of decreasing rainfall, but the southern part of the simulated rainfall in SEC has some inconsistencies with the observations shown in Fig. 2d. Generally, the wind anomalies in the sensitivity experiments verify the results from statistical diagnoses: warming anomalies in the tropical Central Pacific could exert an influence on autumn rainfall in SEC by exciting the atmospheric Gill response, as well as an anomalous local descending motion. However, the simulated precipitation has some differences from the observations, which indicates that the autumn rainfall in SEC may also be affected by some other factors, requiring further investigation in future work.

Conclusions and discussion
The interdecadal decrease in autumn precipitation in SEC occurred in approximately 1990, accompanied by interdecadal warming in tropical Pacific SST, and the correlation Fig. 7 Regression of horizontal wind at a 850 hPa and b 200 hPa on the STI in Fig. 6a during 1971-2009 (vector, m s −1 ); red arrows are at the 95% confidence level. C represents cyclone, and A represents anticyclone. SEC is marked by rectangle coefficient of the two interdecadal sequences is − 0.6, exceeding the 95% significance confidence level. The main conclusions are as follows.
The interdecadal warming of the tropical Pacific has caused an atmospheric Gill response, triggering an anomalous cyclone in the lower troposphere of the NWP. The ambient air current of the abnormal cyclone weakens the northward water vapor transport, which supports deficit precipitation in SEC. In addition, weak cold advection, higher tropospheric temperature, and increased geopotential Fig. 8 Regression of a geopotential height (contour, gpm) and temperature (shading, °C) and b divergent wind (vector, m s −1 ) and velocity potential (shading, 10 6 m 2 s −1 ) on the STI during 1971-2009 at 850 hPa. Regression of c column-integrated water vapor flux from 1000 to 200 hPa (vector, g kg −1 m s −1 ) and its divergence (shading, g kg −1 m 2 s −1 ) and d meridional wind, vertical motion (vector, omega multiplied by − 100), and cloud liquid water content (shading, g kg −1 m 2 s −1 ) cross section along 110° E-120° E on the STI. Red arrows are at the 95% confidence level for the vector variable, and the dotted area indicates the 95% confidence level for the variable represented by shading. All datasets are subjected to a FFT low-pass filter before calculating the regression, and the violet area indicates the plateau topography Fig. 9 The diagrammatic sketch of the influence of the STI on the interdecadal decrease in autumn rainfall in SEC. The shading represents the shading in Fig. 4a, black arrows represent horizontal wind anomalies, and red arrows represent local meridional circulation Fig. 10 The difference in wind in the sensitivity experiments at a 850 hPa and b 200 hPa (vector, m s −1 ; red vectors represent exceeding the 99% confidence level). c The difference in meridional wind and vertical motion cross section along 110° E-120° E (omega is multiplied by − 100, and the shading indicates exceeding the 99% confidence level). d The difference in rainfall over SEC (unit: 10 −6 kg m −2 s − . 1 ). The difference indicates EXP2-EXP1 Fig. 11 The difference in a total cloud amount in autumn (unit: %), b shortwave radiation obtained by the sea surface (positive value indicating downward transmission), and c the net longwave radiation released (positive value indicating upward transmission; unit: 0.1 W m −2 ) between 1991-1998 and 1983-1990. Dotted parts are obtained through a reliability t test at the 90% confidence level, and the box area represents the key area of interdecadal warming in the tropical Pacific, as shown in Fig. 4 height associated with the weakened EAT, as well as the local abnormal downdraft, decreased significant cloud liquid water content in the middle and lower troposphere, as well as a significant divergence wind anomaly at 850 hPa, which is conducive to the interdecadal decrease in autumn rainfall in SEC.
Numerical simulations testify to the results of statistical diagnosis that the anomalous warming of sea temperature in the tropical Pacific exerts an influence on autumn precipitation in SEC by stimulating the atmospheric Gill response.
It is emphasized that the interdecadal warming in the tropical Central Pacific is an important reason for the interdecadal decrease in autumn precipitation in SEC in the above analyses; therefore, we conduct further discussion about the reason for interdecadal warming. As shown in Fig. 11a, the total cloud amount in the region is beneficial to the increase in shortwave radiation (Fig. 11b), which is beneficial to the interdecadal warming of SST in the tropical Pacific. In addition, the interdecadal variation in the longwave radiation in this region is also beneficial to the interdecadal warming of SST in the region (Fig. 11c), but this paper does not analyze the reasons for this in depth. In view of the fact that the data of ISCCP began in 1983, and 1990 is referred to as the dividing year of the difference, we only choose the data spanning from 1983 to 1998 to keep the same time span between 1983-1990 and 1991 to the end. In addition to the effects of atmospheric cloud radiation, the impact of marine dynamic processes and other factors on tropical Central Pacific SST will be further explored in future work.
However, other relevant studies about SEC rainfall have some different focuses. Zhang et al. (2014) considered that the interdecadal variation of winter rainfall in SEC is not only connected with tropical Western Pacific SST, but also linked to AO. Xin et al. (2006) deemed that the interdecadal variation of North Atlantic Oscillation in the preceding winter causes the interdecadal reduction of spring rainfall in SEC. Hu et al. (2020) emphasized that the process of South China Sea summer monsoon withdrawal is significantly linked to the late autumn rainfall in SEC. Huo et al. (2022) considered that the interdecadal enhancement of the relationship between autumn precipitation anomalies in Eastern China and SSTA over the southeastern tropical Indian Ocean has occurred in the recent decades, which implies tropical Indian Ocean plays a bigger role in the variation of autumn rainfall there. In the present paper, we mainly focus on the influence from the tropical Central Pacific. So, about the interdecadal rainfall of SEC, other key sea areas and atmospheric circulation should be paid more attention in the future investigation.
Author contribution All the authors contributed to the conceptualization and design of the study. Data were gathered by Kui Liu. An initial draft of the paper was prepared by Kui Liu. The article was repeatedly revised to generate the final version by Lian-Tong Zhou, Jilong Chen, Zhibiao Wang, and Yong Liu. Funding This study was funded by the National Natural Science Foundation of China (grant no. 42105063).

Code availability
The codes used for the processing of data can be provided on request to the corresponding author.

Declarations
Ethics approval and consent to participate Not applicable.

Consent for publication
All the authors consented to publish the paper.

Competing interests
The authors declare no competing interests.