Strengthening effect of El Niño on the following spring Indian Ocean warming with implications for the seasonal prediction of the Asian summer monsoons

El Niño induces a southwest Indian Ocean (SWIO) warming in decaying springs by forcing the slow-propagating downwelling oceanic Rossby waves south of the equatorial Indian Ocean (IO), which could exert a great influence on the subsequent South and East Asian summer monsoons. This brings the seasonal predictability to the regional monsoons. Here we identify a strengthening effect of El Niño on the following spring SWIO warming during 1948–2020. This is owing to the enhancing intensity and lengthening duration of the El Niño-related warm sea surface temperature anomalies over the tropical central Pacific in recent decades. In particular, this strengthening lagged effect of El Niño on the SWIO warming further results in more significant correlations between El Niño and the subsequent South and East Asian summer monsoons. Conceivably, this enhances the regional monsoon predictability, with potentially tremendous benefits for the socio-economic livelihood of billions of people living in the Asian monsoons.


Introduction
El Niño events, featured by anomalous sea surface temperature (SST) warming over the central-eastern equatorial Pacific, often mature during boreal winters (December-February) and have profound effects on the subsequent climate worldwide [1][2][3]. For instance, El Niño excites an anomalous SST warming over the Indian Ocean (IO), particularly over the southwest IO (SWIO), with one season lag in decaying springs (March-May; figure S1; [4][5][6][7][8]). The spring SWIO warming could in turn exert great influences on the South and East Asian monsoons in the following summer (June-August), although the El Niño-related equatorial Pacific SST anomalies have largely dissipated [9].
The mechanisms of how El Niño affects the spring SWIO warming and subsequent Asian summer monsoons have been revealed in previous studies (e.g. [5,10,11]). The peaking winter El Niño induces an anomalous anticyclone (AAC) over the southeast IO (SEIO) via Walker circulation adjustments [12,13], which forces westward-propagating downwelling Rossby waves in the south IO (SIO; [14,15]). When they slowly propagate into the SWIO thermocline dome in the following spring, the SWIO SST becomes warmer (figure S1(a); [11]), which could sustain equatorially antisymmetric, C-shaped wind anomalies over the tropical IO until the following summer [11,16]. Such summer C-shaped wind anomalies against the mean monsoon can result in a North IO (NIO) SST warming [5,17], which further induces an AAC spanning the Northwest Pacific and NIO through a series of ocean-atmospheric interactions [9,18,19]. As a result, the large-scale AAC can Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI. affect the South Asian summer monsoon to the west [20][21][22] and the East Asian summer monsoon to the north [23][24][25][26]. Thus, the El Niño-induced spring SWIO warming is crucial for the Asian summer climate prediction.
Previous studies [27][28][29][30][31][32][33][34] suggested that, under the background of climate change [30], the El Niño-related warm SST anomalies over the central equatorial Pacific tend to feature an enhancing intensity with a lengthening duration during recent decades. Theoretically, we believe that stronger and longer-lasting SST anomalies over the central equatorial Pacific may tend to excite stronger remote responses. Indeed, this study finds that there are significantly strengthening effects of El Niño on both the following spring SWIO warming and the subsequent Asian summer monsoons during 1948-2020. This implies an enhancing predictability of the Asian summer monsoons, potentially favoring the socio-economic livelihood over the densely populated Asian monsoon regions.

Data and method
Multiple observed and reanalyzed datasets are used in this study, including the monthly SST data from the Hadley Centre Met Office Ice and Sea Surface Temperature (HadISST; [34]), the monthly sea surface height (SSH) data from the European Centre for Medium-Range Weather Forecasts (ECMWF) Ocean Reanalysis System 4 (ORAS4; [35]), and the monthly wind data from the National Centers for Environmental Prediction/ National Center for Atmospheric Research (NCEP/NCAR; [36]) and the European Centre for Medium-Range Weather Forecasts (ECMWF) reanalysis v5 (ERA5; [37]). Since the results from the two wind reanalysis datasets are very similar, only the results from the NCEP/NCAR dataset are shown here. The present study period is 1948-2020, except that the ORAS4 SSH (ERA5 wind) data analyses only adopt the period 1958-2017 (1958-2020) due to the data availability. To remove the long-term trend, all data are linearly detrended. Then, the anomalies are obtained by removing the entire period climatology.
Here, an El Niño event is determined when the Niño3.4 (170°W-120°W, 5°S-5°N) index ( [38]; hereinafter referred to as Niño3.4) exceeds 0.5°C for at least 5 months. Table S1 shows [39], the indices of South Asian summer monsoon (SASM) and East Asian summer monsoon (EASM) are defined as the differences of 850-hPa zonal wind anomalies between the two regions (70°E -90°E, 15°N-20°N and 50°E-80°E, 5°N-10°N) and between the two regions (110°E-140°E, 15°N-25°N and 100°E-130°E, 5°N-15°N), respectively. In this study, all indices are normalized by their respective standard deviations and the significance level p is estimated based on a two-tailed Student's t test, unless otherwise specified. As the samples in the neighboring time points are not independent, the degree of freedom of

21-year moving correlations/standard deviations is estimated by
where N is the sample number and r is the auto-correlation coefficient. Figure 1 shows the time series of the winter Niño3.4, IOBSI, and SWIOSI. The correlation coefficient between the Niño3.4 and IOBSI (SWIOSI) is 0.82 at p < 0.001 (0.84 at p < 0.001) for the entire study period 1948-2020, demonstrating a significant relationship between the winter El Niño and the following spring SST warming over the IO, particularly over the SWIO. Moreover, we find that the 21-year moving correlations between the Niño3.4 and IOBSI (SWIOSI) show a statistically significant enhancement (figure 1). The linear fitting of the 21year moving Niño3.4-IOBSI (SWIOSI) correlations exceeds the 0.001 (0.001) significance level and explains 51% (64%) of the total variance, demonstrating an evident strengthening effect of El Niño on the following spring IO warming. Indeed, while the correlation between the Niño3.4 and IOBSI (SWIOSI) is only 0.64 at p < 0.01 (0.68 at p < 0.01) during the first two decades , that reaches up to 0.86 at p < 0.001 (0.91 at p < 0.001) during the last two decades (2000-2020; figure S2). This increase of the correlations between the two periods is statistically significant at p < 0.05 (p < 0.05) using a single-tailed z test.

Physical mechanism
In this sub-section, we aim to reveal the possible cause for the strengthening effect of El Niño on the following spring SWIO warming. To better display the strengthening trend, we choose three 36-year sub-periods 1948-1983, 1967-2002, and 1985-2020 to represent the early, middle, and later epochs, respectively. Note that the results based on the shorter but non-overlapped 24-year sub-periods 1948-1972, 1973-1997, and 1998-2022 are consistent with those of the 36-year sub-periods (figure not shown). Figure 2 shows the composite seasonal evolutions of SST and 850-hPa wind anomalies for the El Niño events during the three epochs (table S1). The El Niño-related central Pacific SST warming presents an enhancing trend among the three epochs from the developing autumns to the decaying springs, while the eastern Pacific warming changes relatively little. For example, in April and May of the El Niño decaying years, the central Pacific warming is not significant in the early epoch but still statistically significant in the later epoch ( figure S3). Although the El Niñorelated warm SST anomalies in the eastern Pacific do not feature a significant decadal change, those in the central Pacific are significantly enhancing at p < 0.05 using a two-tailed Student's t test ( figure S4). This indicates the enhancing intensity and lengthening duration of the El Niño-related central Pacific SST warming during recent decades, consistent with previous studies [27,29].
Theoretically, stronger and longer-lasting central Pacific SST anomalies may excite more robust remote responses. Thus, we suppose that the strengthening effect of El Niño on the following spring SWIO warming may be associated with the changing intensity and duration of the El Niño-related warm SST anomalies over the equatorial central Pacific during recent decades. Indeed, accompanied by the strengthened intensity and longer duration of the central Pacific warming, the tropical Walker circulation adjustment becomes stronger [40], particularly the anomalous sinking convection over the western Pacific (figure S5), further leading to stronger and longer-lasting easterly wind anomalies over the equatorial IO ( figure 3). While the El Niño-induced equatorial IO easterly anomalies in the early epoch are weaker and only persist into January, those in the later epoch are stronger and persist longer until the following April. This would intensify the AAC over the SEIO in El Niño developing autumns and mature winters (figure 2) and the subsequent downwelling oceanic Rossby waves over the SIO from the early to later epochs (figure 3). In El Niño decaying springs, intensified Rossby waves propagate into the SWIO thermocline dome, inducing a stronger SWIO warming. We also check the role of surface heat fluxes in the SWIO warming during the El Niño decaying spring. The results show that the composite anomalies of surface shortwave radiation flux and the wind-evaporation-SST (WES) term of surface latent heat flux over the SWIO are not statistically significant (figure not shown). This highlights the important role of El Niño-inducing SIO Rossby waves in the decaying spring SWIO warming. Indeed, both the effects of El Niño on the following spring SWIO thermocline deepening and the SWIO thermocline deepening on the SST    figure S8). The variation of the 21-year moving Niño4-SWIOSI correlations is closely linked to that of the 21-year moving SDs of the Niño4 index, with the correlation of 0.68 at the significance level of p < 0.1. After removing the long term trends of the 21-year moving correlations and SDs, their correlation is still at the significance level of p < 0.1. All this suggests that the strengthening effect of El Niño on the spring SWIO warming is due of the enhancing intensity and lengthening duration of the El Niño-related central Pacific SST warming.
3.3. Impact on the relationship between El Niño and the Asian summer monsoons As explained in the Introduction, the El Niño-induced spring SWIO warming could sustain the C-shaped wind anomalies and IO warming until the following summer [4,11,16,17,41]. This further induces a large-scale AAC spanning the Northwest Pacific and NIO [9,18,19], which could in turn weaken (strengthen) the SASM (EASM) [20,21,23,24]. In other words, through modulating the spring SWIO warming, El Niño can exert an important effect on both the SASM and EASM [22,26]. Thus, the strengthening of the El Niño-SWIO warming relationship is very likely to induce an enhancement of the relationships between El Niño and the Asian summer monsoons. Indeed, as the SWIO warming and equatorially antisymmetric, C-shaped wind anomalies in El Niño decaying springs are featuring an enhancing trend from the early to later epochs, the NIO warming and Asian monsoon responses in El Niño decaying summers are also strengthening among the three epochs ( figure S9). As a result, the correlations of the winter Niño3.4 with the decaying year SASM index are −0.19 (p > 0.1; not significant), −0.41 (0.01 < p < 0.05), and −0.59 (p < 0.001) as well as those with the decaying year EASM index are 0.24 (p > 0.1; not significant), 0.36 (0.01 < p < 0.05) and 0.44 (p < 0.01) in the early, middle, and later epochs, respectively ( figure 4). This demonstrates a strengthening trend in both the El Niño-SASM and El Niño-EASM relationships among the three 36-year sub-periods. As the winter El Niño signals lead the Asian summer monsoons for two seasons, the relationships between El Niño and the Asian summer monsoons bring the seasonal predictability to the Asian summer climate. Thus, the present results imply an enhancing predictability of the Asian summer monsoons and also potentially benefit a better understanding of the interannual variations of the Asian summer climate.

Summary and discussions
El Niño induces a SWIO SST warming in decaying springs through forcing the slow-propagating downwelling oceanic Rossby waves in the SIO [9], with important impacts on the subsequent SASM and EASM [22,25]. This brings the seasonal predictability to the Asian summer monsoons and is of potentially great importance for the regional socio-economic livelihood, including agriculture, water resources, food security, human health, ecosystems, disaster mitigation, and infrastructure construction, and so on. This study highlights a strengthening effect of El Niño on the following spring SWIO warming during 1948-2020, owing to the enhancing intensity and lengthening duration of the El Niño-related warm sea surface temperature anomalies over the equatorial central Pacific in recent decades. In particular, this strengthening lagged effect of El Niño on the SWIO warming further results in more significant correlations between El Niño and the subsequent SASM and EASM. This implies a potentially enhancing predictability of the Asian summer monsoons at monthly to seasonal leads, with tremendous benefits for the socio-economic livelihood of billions of people living in the Asian monsoons.
Previous studies suggested that stronger intensity and longer duration of the El Niño-related SST warming over the equatorial central Pacific could be regarded as a result of more frequent occurrences of central Pacific El Niño events [27,29,42,43], which may be induced by global warming in recent decades [30]. Indeed, when using the central Pacific El Niño index instead of the Niño3.4 index, the El Niño-related spring SWIO SST warming features a similar decadal strengthening (figure not shown). In response to global warming, the Pacific Walker circulation and equatorial easterlies would be weakened [44][45][46], resulting in a flattening of the thermocline in the equatorial Pacific [47]. A shallower thermocline in the central Pacific would enhance both the vertical advection and the zonal advective feedback there [48] and thus promote more frequent occurrences of central Pacific El Niño events [30,42,43,49]. Overall, under global warming, the El Niño-related warm SST anomalies in the equatorial central Pacific would tend to be stronger and longer-lasting [27][28][29]. In particular, an enhanced El Niño-IO warming relationship is indeed projected by climate models under the near future greenhouse warming [50,51], in which the El Niño-related central Pacific warming would strengthen significantly although the El Niño-related eastern Pacific warming would change little [52]. To the extent that the strengthening El Niño-IO warming relationship could induce more significant correlations between El Niño and the subsequent Asian summer monsoons, this suggests a possible enhancement of the seasonal predictability of the SASM and EASM under the future global warming.
We also check the La Nina cases. However, unlike El Niño, La Nina do not feature a strengthening effect on the following spring SWIO cooling during the recent decades (figure not shown). One possible reason is that the strengthening effect of El Niño on the decaying spring SWIO warming is mainly attributed to more frequent occurrence of CP El Niño, but La Nina tends to feature a mixed EP and CP pattern with a weak diversity in the spatial pattern [53][54][55][56][57][58].
El Niño is the most important external factor to trigger enhance the positive Indian Ocean dipole (pIOD) through tropical Walker circulation adjustments [59,60]. As the El Niño-related central Pacific SST warming tends to induce a strengthening response of the tropical Indo-Pacific deep convection during the recent decades (figure S5). the enhancing El Niño-related central Pacific warming would also induce a strengthening effect on the pIOD (figure 2). The underlying processes and physical mechanisms have been detailedly investigated in our another paper [61].