Figure 1 shows the partial correlations between seasonal SST anomalies and normalized DJF CT ENSO index (left panel) and WP ENSO index (right panel) from El Niño developing summer (JJA0) to the decaying summer (JJA1), respectively. For the CT El Niño, significant warming is always concentrated along the equatorial eastern Pacific from the developing summer to the decaying summer. The warming patterns extend from the west coast of the Americas up to the dateline with warming centers located at the equatorial eastern Pacific. Meanwhile, negative SST anomalies develop in the western Pacific to the east of the Philippines and extend both northeastward and southeastward (Figures 1A–E) forming a cold ‘‘boomerang’’ (Trenberth and Stepaniak 2001). The CT El Niño is characterized by such a dipole SSTA pattern over the tropical Pacific throughout its life cycle. As for the WP El Niño, the center of the warming SSTA pattern is located persistently at the equatorial central Pacific from the developing to the decaying phase, with a weak cold SSTA to the equatorial eastern Pacific around 90W (Figures 1F–J). On the other hand, the cold ‘‘boomerang’’ SSTA pattern is less significant than that of the CT El Niño. This triple pattern along the tropical Pacific denotes a WP El Niño. Overall, the CT and WP ENSO indices are capable of capturing the SSTA progression patterns for the two types of El Niño events.

Chen et al. (2018) concluded that ENSO has significantly positive impact on the precipitation of Central Asia in during 1951–2013. Further, we found differences in the spatial patterns and intensity of the effects of the two types of El Niño on monthly precipitation in Central Asia. Figure 2 shows the regression coefficients of the monthly precipitation anomalies averaged through SON0 to MAM1 in Central Asia on the Niño3.4 index (Figure 2A), CT ENSO index (Figure 2B), and WP ENSO index (Figure 2C) during 1891–2016, respectively. Note that only regression coefficients that passed the 95% significance test are shown in these figures. It is shown that there are significant differences in the impact of the two types of El Niño on precipitation in Central Asia. The positive precipitation anomalies associated with CT El Niño are mainly located in the western Pamir Plateau (magnitudes about 2–4 mm/°C), with no significant impact on most of the rest parts of Central Asia (Figure 2B). In contrast, the positive precipitation anomalies associated with WP El Niño cover most of Central Asia except Xinjiang, China. The positive precipitation anomaly has a maximum center on the Pamir Plateau and in the Tian Shan Mountains, with magnitude of 10–12 mm/°C. In the hills, plains and desert areas of central and western Central Asia, the magnitude of the positive precipitation anomaly is about 2–4 mm/°C (Figure 2C). When the two types of El Niño events are considered together, the positive precipitation anomalies associated with the Niño3.4 index are found in the Pamir Plateau and Tian Shan Mountains regions (Figure 2A).

Moreover, we further explore the seasonal precipitation anomalies patterns associated with the two types of El Niño during their life cycle. Chen et al. (2018) found that ENSO has the largest impact on the precipitation in Central Asia in boreal winter, followed by spring and autumn and there is no significant correlation between ENSO and the summer precipitation. Therefore, we mainly focus on the precipitation anomalies in the autumn before the El Niño peaks (SON0), the El Niño peaking winter (DJF0), and the next spring (MAM1). Composite analyses are conducted here to illustrate these patterns. First, we define the two types of El Niño events according to Ren and Jin (2011), that is, when the WP (CT) index is positive greater than one standard deviation of its own, then the warm SST anomaly is regarded as a WP (CT) El Niño state. Based on this criterion, the two types of El Niño events that were determined using detrended monthly HadISST from 1891-2020 are as follows:

CT El Niño: 1896, 1899, 1902, 1905, 1911, 1913, 1918, 1925, 1930, 1940, 1941, 1951, 1957, 1965, 1972, 1976, 1982, 1986, 1991, 1997, 2009, 2015

It can be seen that several events (1940, 1941, 1957, 1986, 1991, 2009, 2015) are defined as both CT and WP El Niño events based on this determination criterion, which is due to the fact that these events show significant warming in both Niño3 and Niño4 regions and are basin-wide warm events. Therefore, some studies also refer to these events as “strong basin-wide” events and do not regard them as a regular CT or WP El Niño (Wang et al., 2019; Wang et al., 2020). In following analyses, we still keep these events in their respective categories.

The composites of seasonal precipitation anomaly associated with the CT El Niño (left panel) and WP El Niño (right panel) years during 1891–2020 from SON0 to MAM1 are shown in Figure 3. Only anomalies that passed the 90% significance test are shown in these figures. Consistent with the result in Figure 2, the impacts of WP El Niño on precipitation anomalies in Central Asia is stronger than CT El Niño in terms of intensity and spatial extent. In addition, there are prominent discrepancies between the precipitation anomalies associated with the CT and WP El Niño during every stage of the El Niño lifetime. In the CT El Niño developing autumn, positive precipitation anomalies are found in the plains of southwestern Central Asia and the northwestern Pamir Plateau (Figure 3A). In the following winter, as the El Niño peaks, the distribution of positive precipitation anomalies shifts northward, with more scattered wet conditions occurring in Kazakhstan (Figure 3B). In the El Niño decaying spring, a continuous positive precipitation anomaly appears in the Kazakh Hills and northern Xinjiang along with more precipitation in the northwestern Pamir Plateau (Figure 3C).

There are prominent discrepancies between the precipitation anomalies associated with the CT and WP El Niño during every stage of the El Niño lifetime. More precipitation appears in the Tian Shan Mountains and the Turan Plain in the WP El Niño developing autumn (Figure 3D). The positive precipitation anomalies in the mountainous region along Pamir Plateau and Tian Shan Mountains become conspicuous in the WP El Niño peaking winter along with some relatively weak positive rainfall anomalies in Xinjiang and plains around the Aral Sea (Figure 3E). In the following spring, the positive rainfall anomaly patterns are quite similar to that in the last winter with more precipitation in the mountainous region and a wider range of rainfall around the Aral Sea (Figure 3F). We also investigate the partial correlations between seasonal precipitation anomalies in Central Asia and the normalized DJF CT/WP ENSO indices in the same three seasons. And the results from the partial correlation are compatible with the above composite analyses (Figures not shown). Only that the area where the partial correlation coefficient between the precipitation anomaly and the CT/WP ENSO indices can pass the 90% significance test that is broader than that of the composite analysis. This is related to the size of the samples for which significance tests were carried out.

In summary, when comparing the precipitation anomalies in Central Asia associated with the CT and WP El Niño, three conspicuous characteristics need to be emphasized. First, overall, El Niño are associated with more precipitations over Central Asia in SON(0), DJF(0) and MAM(1) which is consistent with the findings of Chen et al. (2018), although the wet condition patterns are not consistent spatially and temporally. Second, the impacts of WP El Niño on Central Asia precipitation are stronger, more extensive, and longer lasting than that of CT El Niño. Third, the precipitation anomaly spatial patterns associated with the two types of El Niño is different from the El Niño developing autumn to the decaying spring. To be specific, the precipitation associated with CT El Niño is mostly concentrated in the plains and hilly areas of Central Asia and is more dispersed in space. Whereas the precipitation associated with WP El Niño is mostly concentrated in the mountainous areas of Central Asia. In its peaking winter and decaying spring, WP El Niño also brings a small amount of precipitation to Xinjiang and the plains around the Aral Sea.

The different impacts of CT and WP El Niño on precipitation in Central Asia in time and space suggest that precipitation in Central Asia in different seasons may be controlled by different atmospheric circulations when the two types of ENSO occur. In this section, we explore the possible physical mechanisms for the above features of precipitation in Asia, based on composite analyses of the geopotential height (Figures 4, 5) and wind field (Figures 6, 7) of the lower and middle troposphere, the vertical integral of eastward and northward water vapor flux and the vertical integral of divergence of moisture flux (Figure 8) using the NCEP/NCAR Reanalysis 1 data. Previous studies suggested that ENSO has strong influences on the south-westerly water vapor flux coming from the Arabian Sea and tropical Africa and northeastern water vapor flux from Russia and thus can cause precipitation anomalies over Central Asia (Mariotti, 2007; Hu et al., 2017; Chen et al., 2018). In the following content, we focus on the difference between these two sources of water vapor flux in the developing, peaking, decaying season of two types of El Niño.

In the autumn before the CT El Niño peaks, anomalous high pressure dominates most of the low latitudes south of 30°N, and north of this anomalous high pressure there are two weaker low pressure centers located in the eastern Mediterranean and over the Caspian Sea, with a deeper trough near the Mediterranean Sea (Figures 4A, 5A), such that the atmospheric circulation pattern enhances westerly winds (Figures 6A, 7A) and facilitates the transport of water vapor from the Mediterranean and Arabian Sea to the Central Asian region (Figure 8A); while in Central Asia a weak high-pressure anomaly exists in northeastern Central Asia, which cooperates with the low-pressure center over the Caspian Sea (Figures 4A, 5A), resulting in an easterly wind anomaly (Figures 6A, 7A) in central Asia and a convergence of water vapor fluxes in south-central Central Asia (Figure 8A), thus causing precipitation in these regions (Figure 3A).

In the winter when CT El Niño matures, the strength of anomalous high pressure at low latitudes increases significantly, the Mediterranean Sea and North Africa are controlled by anomalous high pressure, a weak low pressure center exists in the Arabian Peninsula, and a significant anomalous high pressure exists over the Tianshan Mountains (Figures 4B, 5B), thus enhancing the northward transport of water vapor from the Indian Ocean (Figures 6B, 7B, 8B); at the same time, the disappearance of anomalous low pressure over the Mediterranean Sea (Figures 4B, 5B) weakens the southwest water vapor flux in Central Asia (Figure 8B), resulting in more scattered precipitation in the northern part of the region (Figure 3B).

In the spring when CT El Niño decays, anomalous high pressure dominates most of the middle and low latitudes south of 40°N, and a narrow anomalous low pressure exists at middle and high latitudes with two centers, one in western Europe and one near Siberia (Figures 4C, 5C), and the above atmospheric circulation pattern also leads to stronger westerly winds (Figures 6C, 7C), thus strengthening the western water vapor path in Central Asia (Figure 8C); Meanwhile, there is anomalous high pressure at high latitudes, whose center is located near the Ural Mountains, and together with the anomalous low pressure at middle and low latitudes (Figures 4C, 5C), it causes significant northeasterly wind anomalies in northern Central Asia (Figures 6C, 7C), which makes water vapor from high latitudes transport to northern Central Asia and strengthens the northern water vapor path in Central Asia (Figure 8C).

For WP El Niño, in the autumn prior to its maturation, most of the low and middle latitudes south of 45°N are controlled by significant anomalous high pressure, and anomalous high pressure exists at high latitudes in western Europe, while a widespread low pressure anomaly exists north of 45°N in northeastern Europe and northern Asia (Figures 4D, 5D), such a circulation pattern enhances northwesterly winds in northern Central Asia (Figures 6D, 7D), allowing water vapor transport from high latitudes to Central Asia (Figure 8D). Unlike CT El Niño, a high-pressure anomaly center exists in eastern Central Asia, and a significant southwesterly wind anomaly exists at the low troposphere (Figures 4D, 5D), enhancing the southwesterly water vapor path in Central Asia, and water vapor fluxes converge in the Tianshan Mountains and the Pamir Plateau (Figure 8D), causing positive precipitation anomalies in these regions (Figure 3D).

During the peaking season of WP El Niño, anomalous low pressure in the higher latitudes of Eurasia further intensified and pushed southward, and anomalous high pressure in the middle and low latitudes also strengthened (Figures 4E, 5E), thus enhancing the southwestern water vapor flux and northern water vapor flux in Central Asia (Figures 6E, 7E, 8E) and bringing stronger precipitation to the aforementioned areas (Figure 3E).

During its decaying season, a significant high pressure anomaly is maintained over most of the low latitudes south of 40°N, and what was originally an anomalous low pressure at high Eurasian latitudes becomes a weak anomalous high pressure, while a low pressure center develops over central Europe at mid-latitudes, and a narrow low pressure trough exists in western Central Asia (Figures 4F, 5F), such a circulation situation enhances water vapor transport from the Mediterranean, which is particularly significant in the troposphere (Figures 6F, 7F). This circulation enhances water vapor transport from the Mediterranean, especially in the middle troposphere, resulting in increased southwest water vapor fluxes in Central Asia (Figure 8F). Unlike CT El Niño, the anomalous low pressure near Siberia is weak at this time, and there is no significant northeasterly wind level in the middle and lower troposphere, so the water vapor flux in the northern part of Central Asia fails at this time (Figure 8F).

There are bound to be differences in the Walker circulation excited by the different SSTA patterns of the two types of El Niño illustrated in Figure 1, and such differences may have different effects on precipitation in more countries and regions around the world through the transmission of atmospheric bridges. To further analyze how the two types of El Niño affect seasonal precipitation in Central Asia through changes in atmospheric circulation, we first examine the Walker circulation anomaly (Figure 9) excited by the two types of El Niño, and the local meridional circulation anomaly between 45°E-75°E (Figure 10). Figure 9 shows the partial correlations between the anomalous Walker circulation averaged in 5°S-5°N and normalized DJF CT ENSO index (left panel) and WP ENSO index (right panel) from El Niño developing autumn (SON0) to the decaying spring (MAM1). Figure 10 shows the partial correlations between the anomalous Hadley circulation averaged in 45°E-75°E and normalized DJF CT ENSO index (left panel) and WP ENSO index (right panel). In both figures, shadings indicate correlations above the 95 and 90% confidence levels.

For the CT El Niño, two anomalous cells form over the equatorial Indian-Pacific Ocean in DJF when it peaks (Figure 9B), a strong one over the Pacific and a weak one over the Indian Ocean. The rising branch over the Indian Ocean is weaker in autumn (Figure 9A), strengthens in winter, and disappears in spring when CT El Niño decays (Figure 9C). The anomalous Walker circulation caused by WP El Niño is significantly different from CT El Niño. The most dominant upwelling branch is located in the equatorial central Pacific at 150°E-180°, with two sinking branches on each side of it (Figures 9D,E). In the autumn before the peak of WP El Niño, there is no significant rising motion over the Indian Ocean (Figure 9D), and in the winter when WP El Niño matures, there is a significant rising motion near 90E (Figure 9E), and the extent of the rising branch over the Indian Ocean continues to expand as WP El Niño decays (Figure 9F). To sum up, there are significant differences in the evolution of the upwelling branch over the Indian Ocean during the development of the two types of El Niño.

To examine the possible impact of the Indian Ocean upwelling branch on precipitation in Central Asia during the two El Niño development phases, we next examine the anomalous Hadley circulation averaged over 45°E-75°E (Figure 10). Note that since our results in Section Possible Teleconnection indicate that the two types of El Niño have no significant effect on precipitation in Xinjiang, China, during their lifecycle, here we narrow the meridional range of the region of interest to 45°E-75°E to exclude the blocking effect of the Tibetan Plateau. It can be seen that around 30°N-40°N over Central Asia, there is a clear rising motion in the autumn before the peak of CT El Niño, extending to the top of the troposphere (Figure 10A); during the peaking winter of El Niño, this rising branch moves northward to around 50°N and weakens significantly (Figure 10B); in the spring when El Niño decays, there is no clear rising motion over Central Asia (Figure 10C). For WP El Niño, upward motion is present over Central Asia (30°N-55°N) in the autumn before its peak, in the winter when it matures, and in the spring when it decays (Figures 10D–F). This upward branch is strongest in the winter of El Niño’s maturation and reaches its widest extent and extends to the top of the troposphere (Figure 10E). In spring, this upwelling branch is weaker but still present, which is a significant difference from CT El Niño (Figure 10F). That is, the influence of the Walker circulation anomaly over the Indian Ocean on precipitation anomalies in Central Asia differs during the life cycle of the two types of El Niño, especially in winter and spring. We further examine the 200 hPa velocity potential (VP200) composite over the two types of El Niño (Figure 11). A clear, persistent pattern of upper-level convergence over the Indian Ocean is present throughout the WP El Niño peaking winter and decaying spring. Whereas the positive VP200 anomalies are weak in the winter of CT El Niño peaks and disappear in the following spring.