A zonal and segmentation analysis of the relationships between variations in sea-surface temperatures of the western Pacific Ocean and Southeast Asian rainfall

Embodying the warmest pool of the ocean, the western Pacific Ocean is regarded as crucial chiefly due to its powerful role in influencing Earth's climate, including that in Southeast Asia (SEA). One of the mechanisms that makes the warm pool¹ of the western Pacific so powerful is its intense evaporation, also known as convection, which serves as the incessant, dominant supply of moisture into the atmosphere (De Garidel-Thoron et al. 2005; De Deckker 2016; Jones & Ricketts 2019). Because of this strong evaporation process, the warm pool of the western Pacific Ocean can lead to intense precipitation (sometimes in excess of 2,000–3,000 mm/year), particularly in nearby lands (Delcroix et al. 1996; Chen 2004). By the same token, changes of sea-surface temperature (SST) in the western Pacific warm pool can also result in changes in large-scale precipitation patterns (Palmer & Mansfield 1984; Cronin & McPhaden 1997). This is evidenced by the occurrences of the El Niño-Southern Oscillation (ENSO) (Philander 1985; Ropelewski & Halpert 1987; Halpert & Ropelewski 1992; Linsley et al. 2010). Specifically, during the incidents of the El Niño, the warm pool expands further eastward following the relaxation of the trade winds in the eastern and central Pacific, whereas during the times of the La Niña events, the warm pool is limited to the western margin of the equatorial Pacific following the reinforcement of the trade winds (Yan et al. 1992; Neelin & Latif 1998). These changes in SST, manifested in terms of the relocations of the western Pacific warm pool, bring about changes in rainfall patterns in areas of lands close to the western Pacific Ocean. In case of SEA, the El Niño tends to bring to drought to the region, while the La Niña is likely to bring heavy rain to SEA (Aldrian & Dwi Susanto 2013; Juneng & Tangang 2005; Lau & Wang 2006). Therefore, although there are other factors that can affect precipitation in SEA, for instance, terrain heights and characteristics (Wang & Chang 2012; Mandapaka et al. 2017), the SEA monsoon (IPCC 2013; Misra & DiNapoli 2014; Loo et al. 2015), the SST of the western Pacific Ocean remains crucial. This is also pointed out in some studies. Dayem et al. (2007) indicated that SEA areas of high precipitation rates correlate with high SST. Another study (Niedermeyer et al. 2007) described that the Maritime Continent (approximately 10 °S–20 °N and 90 °E–130 °E) – largely within the SEA boundary — is situated within the Indo-Pacific Warm Pool. Its hydrological characteristics, in particular considered from the seasonal perspective, are therefore influenced by the relocation of the Intertropical Convergence Zone (ITCZ) as well as changes in the monsoonal wind flow. Soman & Slingo (1997) pointed out that warm SST anomalies in the west Pacific Ocean enhanced the tropical convection over Indonesia, and this in situ response to the convection leads to an early occurrence of and more intense monsoon.

However, among all known potential factors that could affect SEA precipitation, it is still unclear that rainfall in which particular areas within SEA are significantly impacted by the warmer SST of the western Pacific Ocean. The answer to this question is very important as it is unlikely that rainfall in every spot of SEA is equally affected by warmer SST of the western Pacific. Likewise, since it is also unlikely that SST of the entire area of the western Pacific Ocean equally affects SEA rainfall, due largely to the influence of the trade winds, it would be very important to determine the approximate scope of the western Pacific area whose SST would play a significant role in SEA precipitation. Even though some studies (Meehl et al. 2007; Diffenbaugh & Scherer 2011; IPCC 2011; Anderson 2012; Mora et al. 2013) suggested that tropical systems will face unusual warming and significant alteration in the timing and amount of rainfall within a few decades, those studies were too generalized to gain an insight into the impact of SST in the western Pacific on SEA rainfall. Moreover, while Singh & Xiaosheng (2019a, 2019b) suggested that over the period between 1951 and 2014, there was a significant increase in rainfall amount in most of SEA regions, it is unclear how or whether SST, particularly in which areas of the western Pacific Ocean, plays a role.

Knowing rainfall in which area of the region is likely to be affected by the warmer SST of which zone within the western Pacific Ocean would be crucial for the impact assessment of how future warming of western Pacific SST could affect SEA rainfall and would also serve as crucial information for the managements of water resources in the region. With this premise, we applied the highly localized correlation analysis to the data sets to identify the boundary of both the land (SEA region) and the ocean (the western Pacific) in which its rainfall and SST closely interact. The objectives of the study are to (i) determine the boundary of the western Pacific area whose SST variations can significantly affect precipitation in the SEA region and (ii) to identify areas within the SEA region, whose precipitation could be sensitive to the warmer SST of the impactful zones of the western Pacific.

There were two sets of data used in the analysis: the Group for High Resolution Sea-Surface Temperature (GHRSST) Level 4 Multiscale Ultrahigh Resolution (MUR) (JPL 2015) for SST² and GPM Integrated Multi-satellitE Retrievals for Global precipitation measurement (IMERG) Final Precipitation L3 (Huffman et al. 2019) for monthly rainfall³. Both of the datasets were gathered by the US National Aeronautics and Space Administration (NASA). The GHRSST Level 4 MUR has a spatial resolution of approximately 1 km × 1 km. The entire set of the SST data used in the analysis extended from June 1, 2002 to November 30, 2020, accounting for the total of 6,758 files to be processed (all the files in the SST dataset were originally available in a daily basis). In order to make the data temporally consistent with rainfall data, the monthly-average SST was used, covering 222 months. Furthermore, for each month, the original data, which had a global coverage, were clipped to cover only the western Pacific Ocean, which is defined by the longitude of 120 °E–150 °W and the latitude of 20 °S–20 °N.

Aside from dividing the western part of the Pacific Ocean into the North and the South, it was also crucial that each part of the longitude of 120 °E–150 °W be thoroughly analyzed. This zonal analysis along the longitude line could enable us to see the spatial variations of SST within the western Pacific Ocean and also their respective roles in SEA rainfall. Therefore, in this study, the longitude of 120 °E–150 °W, which determines the boundary of the western Pacific, was further divided into nine zones along the longitude line. As a result, 18 zones were obtained (Figure 1).

Based on Equation (2), in order to determine the trend, seasonal and noise components were removed from the observed SST time-series. This study used the classical decomposition presented by Brockwell & Davis (2002) to solve Equation (2). After the trend had been obtained, the linear regression was applied to derive the linear equation that best fitted the trend line. Then, the warming rate was finally estimated from the slope value of the linear equation. This trend extraction process was applied for all ocean zones. Additionally, this analysis was performed for three statistical parameters, namely mean, minimum, and maximum SST. All of these, therefore, allowed us to compare yearly warming rates in different zones of the western Pacific Ocean.

This North–South variation pattern might serve as one of the factors that contribute to the SEA precipitation variability, in particular that between the northern SEA and its southern counterpart. Moreover, this North–South oscillation is significant because it indicates that the impact of SST of the western Pacific Ocean on the SEA rainfall cannot focus merely on the ‘horizontal’ proximity — or the longitudinal distance from the SEA land — as in cases of the El Niño and La Niña events. It is well-established that the movements or spreading of the western Pacific warm pools along the longitudinal positions can have an important impact on the SEA rainfall. However, the results from Figure 3 also suggest that the western Pacific warm pool can re-locate along the latitudinal line, and this latitudinal movement is also marked by seasonal variations. Hence, in order to gain more complete picture into how SST of the western Pacific Ocean affects SEA precipitation, the existence of the North–South SST variation must be acknowledged and taken into consideration.

On the other hand, despite the seasonal variability ranging between approximately 30.6 and 25.5 °C, the SST in each zone of the South is different from that in its North counterpart. The most noticeable is that the warm-pool conditions in the South zones adjacent to SEA were not sustained. From Figure 4, after the SST peaks in the ZONE 1–3S, they were followed by the significant decrease in SST to the point that could no longer be considered as warm pool. This after-peak decrease could bring SST in the ZONE 1–3S down to as low as around 25.8 °C. This makes the warm pool in the ZONE 1–3S less sustained, compared to that in the North. Also, unlike that in the North, the minimum SST in the South occurred in the ocean zones close to the land (ZONE 2S or 3S). Therefore, in the South, it is not true that the more distant the ocean zones from the SEA continent, the lower SST.

Additionally, we discovered that the minimum SST in the ZONEs 1, 2, and 3 South shows negative rates, which suggests that these sea areas have been getting even lower minimum SST since 2013, while the same ZONEs in the North have a distinct pattern, for their minimum SST increased (Figure 6(b)). Still, as the sea areas move further away from the SEA continent, the warming trends in the minimum SST in the South were higher than that in the North as seen in ZONEs 6–9 (Figure 6(b)). Moreover, we found that all the ocean zones showed the positive rates for their maximum SST, which have been higher since 2013 (Figure 6(c)). All these results show different warming rates in different parts of the western Pacific, some of which was not even warmer; instead, it was colder. This implies that the assessment of its SST impact on SEA rainfall should be highly discrete or zone-specific.

Although the sea–land proximity can influence the precipitation in some parts of SEA as described above, there is rainfall in other part of SEA that is not correlated with SST of the ocean zones close to them. In Figure 7, the SEA area that is located between approximately 120 °E–127 °E and 5 °N–18 °N, despite its closeness to the ocean, did not show a strong correlation with SST. Not only that, rainfall in the small islands (123 °E–127 °E and 5 °N–18 °N) had very weak negative correlation with SST even though these islands are located next to the ocean. Moreover, the SEA landmass between 0 °N and 5 °N, albeit situated close to the ocean, had very weak correlation with SST. Furthermore, the influence of the SST in ocean ZONE 2 North was less strong than that in ZONE 3 North although ZONE 2 is closer to the SEA region. Likewise, in Figure 8, the SEA land of roughly between 125 °E–140 °E and 0 °S–5 °S, in spite of its close proximity to the sea, did not have strong correlation with SST. Also, the SEA area of about 0 °N–5 °N also had weak correlation with SST in all the ocean zones in the South.

Even in the SEA area whose precipitation tends to fall under the influence of the sea–land proximity, its extent of rainfall could be also be affected by the sustainability of the warm pool formed. Based on the result in Figure 4, the warm pool formed in the North is more sustained than that in the South. This also could affect the rainfall amount in the northern and southern SEA as rainfall in northern SEA would be subject to less variability than that in southern SEA because of the more sustained warm-pool condition. Additionally, the result from Figure 6 revealed that mean SST in the ZONE 1 South and minimum SST in the ZONE 1–3 South were subject to cooling, suggesting that if these cooling rates continue, some southern parts of SEA would likely have less precipitation.

The relationship between SEA precipitation and SST of the western Pacific is complex. The North–South and zonal SST variation analyses indicated that SST in the North and South peaked in different times. This suggests that the warm pool of the western Pacific Ocean, which usually occurs near the SEA landmass, did not form simultaneously across the North and South. There was seasonal variability involved in the peak-SST times and thus the period and location in which the warm pool forms. Furthermore, the zonal SST variation analysis suggested that the warm pool in the North of western Pacific zones close to SEA (120 °E–150 °E) was more sustained and less varied than that in the South zones. In terms of the western Pacific SST impact on rainfall in SEA, the correlation analysis confirmed the well-established concept that the SST of the ocean part close to the SEA land, in general, has more influence on SEA rainfall than does the SST of the ocean zones far away from SEA. This is particularly true when considered from the latitudinal proximity standpoint. The SST in the North and South of the western Pacific tends to affect the precipitation in the northern and southern parts of SEA, respectively, although such influence on the northern and southern SEA rainfall happens at different times. However, the significance of sea–land longitudinal proximity is less definitive in determining how much rain SEA would have. The analysis showed that rainfall in some areas of the SEA region, albeit longitudinally close to the western Pacific, did not have the strongest correlation with SST. Moreover, rainfall in some SEA areas, in particular those are located between 0 °N and 5 °N, had a weak correlation with the SST of the ocean part close to it. Therefore, the idea that closer longitudinal proximity means stronger correlation is not true in some areas of SEA. The result from the correlation analysis, designed to be highly localized in terrestrial rainfall pixels, leads to the possibility that change in SST in the western Pacific areas, in particular those situated close to the SEA region, might not bring about the uniform changes in precipitation throughout the entire region. And even in SEA areas whose rainfall is governed by the influences of the sea–land juxtaposition, rainfall can also be influenced by how sustained the warm pool is. From the zonal SST variation analysis and correlation results, it is believed that precipitation in the northern SEA might be subjected to less variability because of the more sustained warm pool in the North of the western Pacific Ocean than in the South.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.