5.1.2 Influence of EN and LN states on MJO-related convection and circulation
The next figures show the MJO global anomalous convection and circulation patterns (Figs. 7, 8, 9, 10) during EN and LN states.
When the MJO-related anomalies are favored by oceanic and atmospheric background ENSO-related changes (Fig. 2), their intensity tends to be enhanced. Figures 7 and 8 show that in MJO phases 8, 1, 2, 3 low-level anomalous divergence and reduced convection over the eastern Indian Ocean-Maritime Continent/west Pacific and low-level anomalous convergence and enhanced convection in central Pacific are more intensified on the equatorial belt during EN than during LN. This is consistent with the EN prevalent anomalous subsidence over the former regions and dominant anomalous convection over the latter (Fig. 2 and Figs. 7 and 8). It is more visible in the anomalous OLR field (Fig. 8), and for phases 8 and 1. On the other hand, during MJO phases 5, 6, 7 the low-level anomalous convergence and enhanced convection over the eastern Indian Ocean-Maritime Continent/west Pacific and low-level anomalous divergence and subsidence in central Pacific are more enhanced during LN, because these anomalies are also prevalent in LN (Fig. 2 and Figs. 7 and 8).
The position and size of MJO-related anomalies is also altered by ENSO opposite states. For instance, when the MJO enhanced convection is near/over the Maritime Continent, in phases 4, 5, and 6, it is more intense and extensive during LN than EN (Fig. 8), since ascending motion is enhanced in this region and the warm pool occupies a larger latitudinal extension (Figs. 2e, 2i). On the other hand, the MJO equatorial convection reaches and crosses the Date Line already in phases 5 and 6 during EN, favored by the SST and Walker circulation background conditions in this ENSO state (Fig. 2), while in NT state this only occurs in phases 6 and 7 (Figs. 4 and 8). This is consistent with the eastward shift of MJO activity during EN events reported by Hendon, Zhang and Glick (1999). On the other hand, in LN state the MJO convection is suppressed in equatorial central-east Pacific in phases 7 and 8 (Figs. 4 and 8), but appears in subtropical central-east South Pacific, which is coherent with the background in this ENSO state. As in LN the equatorial SST in the Pacific is colder east of the Date Line and the subsidence is enhanced, the tropical enhanced convection is shifted south of the equator (Fig. 2g).
In another example, when the equatorial MJO enhanced convection crosses the Date Line (phases 6 + 7), and MJO and EN (LN) anomalies have same (opposite) sign east of the Date Line, the MJO-related low-level divergence and subsidence over the equatorial northeast SA is enhanced (weakened) in MJOENphases6+7 (MJOLNphases6+7) and the enhanced convection (or negative OLR) in MJOENphases8 + 1 + 2 (MJOLNphases8 + 1 + 2) is weakened (increased) in this region (Fig. 8) with respect to NT state (Fig. 4). This is consistent with the effects of ENSO on the equatorial northeast SA (or eastern Amazon), shown in the upper panels of Fig. 2. In NT years, without EN or LN influence, the MJO-related subsidence observed in phases 4 + 5 + 6 over northeastern SA weakens in MJONTphases6 + 7 + 8, when the MJO convection crosses the Date Line (Fig. 4). However, during EN years there is EN-related background subsidence over northeastern SA due to the EN-related enhanced convection in central-east equatorial Pacific (Fig. 2d, 2f, 2h), which is further increased in MJO phases 6 + 7 + 8 (Fig. 8, left column). Therefore, during MJOENphases6 + 7 + 8, the subsidence over northeastern SA is stronger and stays longer than during NT or LN state (cf. Figures 4 and 8).
Some of the aspects described in the paragraphs above are also visible in Fig. 9, in which the MJO anomaly composites for the NT years are removed from the composites for EN and LN years. For instance, on the left column (EN-NT) the red ellipses show the enhancement of subsidence over the western Pacific in phases 8 and 1, when the strongest subsidence associated with the MJO coincides with the strongest subsidence associated with EN. On the other hand, on the right column (LN-NT) the blue ellipses show the enhancement of convection over the western Pacific in phases 5 and 6, when the strongest convection associated with the MJO coincides with the strongest convection associated with LN. The black ellipses will be mentioned in the next section.
5.1.3 Influence of EN and LN states on MJO propagation and teleconnections
Propagation speeds of MJO features are approximately indicated by the inclination of the arrows in Figs. 4 and 7, but are also visible in the displacement of the convection and circulation anomalies (Figs. 4, 5, 8, 10 and 11), especially in the low-level streamfunction quadrupoles, associated with the Rossby and Kelvin wave responses to the dipole of mass sink and source (heating and cooling) over the equatorial region (Matsuno 1966; reviews in Zhang 2005; Grimm 2019). The circulation anomalies are presented to facilitate discussion of the teleconnections and precipitation anomalies over SA observed in each ENSO phase. The equatorial propagation characteristics are also summarized in Fig. 12.
Since higher (lower) phase propagation speed is expected for weaker (stronger) convection (Zhang 2005; Pohl and Matthews 2007), the stronger and more extensive anomalous convection over the Maritime Continent in phase 5 during LN is consistent with much slower eastward propagation over this region (Figs. 7, 8). Besides, there is great weakening of the convection when it moves from over the Maritime Continent to the western Pacific, which is consistent with the barrier effect of the Maritime Continent (Zhang and Ling 2017). On the other hand, during EN, the anomalous convection over the Maritime Continent is weaker and does not reduce much when it moves from the Maritime Continent to the western Pacific Ocean, with higher propagation speed (Fig. 7). It crosses the Date Line in phase 6, extending further eastward in the equatorial band than in LN state (Figs. 7, 8), since the equatorial warm pool is extended further eastward during EN than during LN (Figs. 2d, 2e). On the other hand, during LN there is a quicker eastward propagation of anomalous convection on the subtropical central-eastern South Pacific, from phase 6 to 7, probably favored by the warm pool features during LN (Fig. 2e).
After the MJO convection crosses the Maritime Continent and approaches the Date Line (phase 6) (Figs. 4, 7, 8), the propagation speed over the climatologically colder ocean waters shows considerable differences between EN and LN. During EN, although stronger equatorial anomalous convection extends eastward of the Date Line, there is little propagation from phase 6 to 7 (Figs. 7, 8), and this aspect also appears in the circulation anomalies, since there is little propagation of the equatorial streamfunction anomalies over the Pacific between these phases (Figs. 10, 11). On the other hand, during LN there is no enhanced convection on the equatorial central-east Pacific, but subtropical convection is extended eastward from phase 6 to 7 into central-east South Pacific (Fig. 8). The equatorial wind perturbation propagates faster eastward and the low-level convergence and westerly winds are already strong over the eastern Pacific and SA in phase 8, while during EN they reach the maximum in phase 1 (Figs. 7, 10, 12). The same is true for the upper-level easterlies, which are stronger and propagate eastward faster over the central-east Pacific during LN than during EN (Figs. 11, 12), favoring stronger and earlier equatorial convection in the Western Hemisphere, over SA, Atlantic Ocean and Africa in phases 8 through 1 during LN.
Therefore, the tropics-tropics teleconnection between the Pacific and SA during LN state, favoring the enhancement of precipitation in northeast SA and tropical CESA, is already established in MJOLNphase8, but during EN state is only well established in MJOENphase1 (Figs. 7, 8, 10, 11). The differences between EN and LN in the propagation of the equatorial convection anomalies across the Date Line and in the propagation of the zonal wind signal is also clear in the Hovmoller diagrams (Fig. 12). This difference between propagation speeds in the Western Hemisphere is consistent with the reduction (increase) of the MJO phase velocity over warmer (colder) SSTs and associated with stronger (weaker) convection (Zhang 2005). Also the subtropical low-level wind convergence and associated OLR anomalies in central-east South Pacific, which are important for the extratropical teleconnection with SA, are first established in phase 7 (8) during LN (EN) (Figs. 7, 8). This subtropical convection is an important aspect related to the extratropical teleconnection between the central-east Pacific and SA (indicated schematically by the curved arrows in Figs. 5, 10, 11), since the associated upper-level divergence in this region is shown to be very efficient in triggering such teleconnection (Grimm and Silva Dias, 1995; Grimm, 2019). This region is indicated by the ellipses in Figs. 4, 7, 8. The anomalous convection in it is stronger in MJOENphases8+1 and MJOLNphases7+8 (Fig. 8) than in MJONTphases8+1 (Fig. 4). This is emphasized in Fig. 9, where the differences EN-NT and LN-NT are displayed. The black ellipses on the subtropical central-east South Pacific show that in these phases there is enhancement of convection in this region in both EN and LN states, starting earlier in LN than in EN, and that there is enhanced convection over CESA (and SACZ), especially in MJOENphase1 and MJOLNphase8. The Influence Functions displayed in Fig. 11 of Grimm (2019) shows that the upper-level divergence anomalies in the subtropics of the central-east South Pacific lead to a teleconnection pattern that produces a pair anticyclonic-cyclonic circulation anomaly over the extratropics-subtropics of SA that enhances precipitation in CESA in the phases indicated above. The subtropical cyclone favors moisture flux from the Amazon into CESA (and the SACZ), where there is moisture convergence, and divergence of moisture flux from the middle/lower Parana/La Plata Basin (SESA), tending to form a convective dipole between CESA and SESA (Grimm 2019).
The extratropical teleconnection pattern is approximately indicated on the streamfunction patterns of Figs. 5, 10 and 11, in phases 8 and 1, and, with opposite sign in phase 4. It is more visible at 200 hPa, and at 850 hPa it is only indicated on the barotropic actions centers. The MJO phase in which it is more clearly established varies with the ENSO state. While it is still developing in MJOENphase8 (and strongest at MJOENphase1), it is already fully established in MJOLNphase8 (Figs. 10 and 11). In NT years it is more consistently established with the precipitation anomalies over CESA in phase 1 (Fig. 5). The maximum enhancement of the pair extratropical anticyclone-subtropical cyclone over SA, and the strongest negative OLR anomalies in CESA, produced by the cyclonic circulation, happens earlier in the MJO cycle during LN (MJOLNphase8) with respect to EN (MJOENphase1) and NT (MJONTphase1). However, the highest impact on the southern edge of the SACZ in LN occurs in MJOLNphase1, associated with the subtropical cyclonic anomaly a little displaced westward, with respect to MJOENphase1. The same advancement during LN happens to the propagation of velocity potential anomalies (Fig. 7), as well as to anomalies of precipitation and frequency of extreme events, as will be detailed later.
It is interesting to point out that there are some similar effects of EN and LN states, compared to NT state, regarding the MJO-related extratropical teleconnection from subtropical central-east South Pacific to SA. Taking phase 1 as reference (although phase 8 could also be used for the LN state), there are great differences between the MJO circulation anomalies over the continent in NT years and those in EN and LN years (cf. Figure 5 and Figs. 10 and 11). While EN and LN years show a strong anomalous pair extratropical anticyclone-subtropical cyclone over SA (Figs. 10 and 11), in NT years it is weaker and shifted northwestward, as is also the extratropical teleconnection (Fig. 5). This teleconnection fades in MJO phase 2 (Figs. 5, 11), due to the reduction of convection in that subtropical region in South Pacific (Figs. 4, 8), weakening the positive (negative) precipitation anomalies in CESA (SESA) (next section).
Figure 13 zooms mostly the Western Hemisphere and shows OLR anomalies for MJO phases 8 and 1 in EN, NT, as well as for the difference EN-NT, to emphasize that EN increases MJO related convection anomalies in these phases in the subtropical central-east South Pacific, a region in which the upper-level divergence is very efficient in triggering the MJO extratropical teleconnection towards SA (Grimm and Silva Dias, 1995; Grimm, 2019). This figure also shows that this teleconnection pattern (visible in the 200hPa streamfunction anomalies) is much enhanced during EN, especially the cyclonic circulation pattern over subtropical SA associated with increased precipitation over CESA (and SACZ). As mentioned before and is visible in Fig. 9, and Figs. 10 and 11 when compared to Fig. 5, a similar enhancement happens during LN state.
Besides the similarity over SA, the circulation anomalies during MJOENphase1 and MJOLNphase1 are also similar over the South Atlantic Ocean (Fig. 11). One of the notable features is the strong anomalous barotropic anticyclonic circulation near the southeastern coast of SA, which merges with the extratropical anticyclone over southern SA that is part of the teleconnection pattern from the Pacific. This does not appear in MJONTphase1 (Fig. 5), since it is probably partially produced by the enhanced convection in the southern edge of the SACZ during EN and LN MJO phase 1 (Figs. 4, 8).
Thus far, emphasis has been put on the circulation and convection anomalies associated with the most extensive enhanced convection over SA, in phases 8 through 2, especially phases 8 and 1. However, in phases 3 through 6, especially phases 4 and 5, the anomalous convection displays approximately opposite signs compared to that in phases 8 and 1 (Figs. 4, 8). The convection in the equatorial central Pacific and in the subtropical central-east South Pacific, important for the teleconnection towards SA, is suppressed, and the precipitation dipole is reversed on SA, enhancing the positive (negative) OLR anomalies in CESA (SESA) (Figs. 4, 8). Approximately opposite anomaly patterns have been shown in all ENSO states between phases8 + 1 and phases4 + 5, in convection (Figs. 4, 6, 8, 12) and circulation anomalies (Figs. 5, 10, 11), suggesting that the region of subsidence in the subtropical central-east South Pacific can trigger tropics-extratropics teleconnections, suppressing convection in the SACZ. This teleconnection is schematically represented in phase 4 (Figs. 5, 10, 11), when the wave train is best defined, especially in NT state, but in case of EN the subtropical anticyclonic circulation is strongest in phase 5 (Fig. 11).
In phases 8 and 1 the enhanced convection is strongest and shifted east in subtropical central-east South Pacific in EN with respect to LN, and therefore the teleconnection pattern towards SA is also shifted east with respect to LN, and so is the enhanced convection over SA (Figs. 8, 11) and precipitation (next section). On the other hand, enhanced subsidence over central-east subtropical South Pacific is stronger and shifted east in LN with respect to EN in phases 3 and 4, and therefore this teleconnection pattern is stronger and slightly shifted eastward, and so is anomalous convection over SA (Figs. 8, 11). It also starts earlier, in phase 3, while in EN it is more visible in phase 4, as is more clearly seen at 850 hPa (Fig. 10).
The MJO most intense and extensive dry anomalies over SA (which peak in phases 4 and 5) take longer to be established and last longer in EN state (Fig. 8). The tropical teleconnection is slower and in EN it takes longer for the enhanced convection to move to the subtropical central-east South Pacific and reverse the convection dipole over SA, because of the slower propagation between phases 6 and 7 over the Pacific, mentioned before. During these phases in EN state positive OLR anomalies still predominate in subtropical central-east South Pacific, while in LN already negative anomalies predominate (Fig. 8). Therefore, the teleconnection pattern leading to an inverse dipole over SA starts being established in phase 8 in EN, while in this phase it is already fully established in LN. Furthermore, in the equatorial belt the enhanced subsidence during EN over the northern part of CESA also contributes to extend the dry anomalies over SA. Hence, the interannual circulation anomalies extend (shorten) the intraseasonal circulation anomalies over SA in EN (LN) during the MJO phases with dryness in CESA.
Extratropical teleconnection patterns over other regions of the globe also change between EN and LN states, as shown by Moon, Wang and Ha (2011), especially in the Northern Hemisphere, where the winter basic state favors the extratropical teleconnections triggered by tropical convection. For instance, in extratropical North Pacific, North America and North Atlantic there are noticeable differences between EN and LN in phases 7, 8, 1, such as the anomalies over northern Pacific and eastern US (Figs. 10 and 11). Yet during phases 8 and 1 the differences in Southern Hemisphere are not great, especially in the Western Hemisphere, probably because the regions with more influence on the extratropical teleconnection reaching SA are in the subtropics of central-east South Pacific and anomalous MJO convection is not that different between EN and LN in those regions.
Some features pointed out thus far in the equatorial belt are visible and summarized in the MJO Hovmoller diagrams for the three ENSO states. The quicker and more eastward propagation of the OLR anomalies over the Indian Ocean/western Pacific in EN (Fig. 12b) than LN (Fig. 12c) is clearly visible in the lower slope of the OLR diagram for EN in this range of longitudes (between 60°E-120°E). The equatorial convection stays longer a little east of the Date Line during EN (phases 6 and 7), which is coherent with the anomalous Walker circulation with ascending motion over this region, and then extends eastward (and southward) till 120°W (cf. Figures 12b and 8 left). Yet in LN the equatorial propagation of convection is slower till a little west of the Date Line (although it extends southeastward in the subtropics) (cf. Figures 12c and 8 right). Therefore, the equatorial propagation of anomalous convection ends near the Date Line in LN, also coherent with the Walker circulation basic state in this category. The equatorial enhanced convection only reappears over equatorial SA in phases 8-1-2, and is stronger during LN than EN (cf. Figures 12b, 12c and 8). The MJO eastward propagation over colder SSTs in the central-eastern equatorial Pacific is better represented in the upper-level zonal winds (Figs. 12g, 12h, 12i). They have a higher phase velocity since they are associated with free Kelvin waves uncoupled with convection and quicker/stronger for LN than for EN. They propagate faster and are more intense in the region near SA (a little east of 120°W) during LN, when SSTs are colder, there is no deep convection over the eastern Pacific, and the convection over tropical SA is stronger. The anomalous positive SSTs over central-eastern equatorial Pacific in EN decrease the phase velocity of the MJO (Figs. 12e, 12h, between 180°W-160°W, phases 6–7), slowing its convection across the region (Fig. 8), and causing a delay in the inversion of the intraseasonal dipole over SA until MJOENphase7.