Intraseasonal oscillation(ISO)is a dominant pattern of variability in the tropical atmosphere, witha period of approximately 40 days. ISO was firstlydiscovered in the 1970s by Madden and Julian(1971, 1972). Subsequent studies have found that ISO existsin the extratropics as well as in the tropics. The dominant component of tropical ISO, which propagateseastward, is the Madden-Julian oscillation(MJO). Intensive research on ISO(especially MJO)since the1980s has revealed elements of the structure and activity of the MJO(Krishinamurti and Subrahmanyam, 1982; Murakami et al., 1984; Lau and Chan, 1985; Li, 1991; Madden and Julian, 1994; Zhang, 2005). ISO impacts the onset and activity of the Asiansummer monsoon(Mu and Li, 2000; Li et al., 2001;Lin et al., 2005) and summertime rainfall over China( Yang and Li, 2003; He et al., 2006; Zhang et al., 2009). The MJO also influences precipitation overmany other regions, such as East Asia(Jeong et al., 2008), Southwest Asia(Barlow et al., 2005), Australia(Wheeler et al., 2008), and North America(Jones, 2000; Bond and Vecchi, 2003). These nonlocal influences are communicated through the circulation and teleconnections induced by anomalous convection associated with ISO in the tropics. The MJO can alsomodulate tropical cyclone(TC)genesis in northwestern Pacific, northern Indian Ocean, and the areas surrounding Australia. The active phase of the MJOin these regions sets a favorable state for TC genesis(Sobel and Maloney, 2000; Hall et al., 2001). Maloney and Hartmann(2000a, b)showed that the frequencyof TC genesis in the western Pacific increases whenthe active phase of MJO is also located in the westernPacific; however, other studies have suggested that therole of the MJO might not be the most important inmodulating TC genesis(Liebmann et al., 1994). Thedetails of how and why the MJO modulates TC genesis have yet to be clearly revealed.

Although the timescales of tropical ISO differfrom that of ENSO, several studies have indicated significant interactions between these two phenomena. Li(1989)proposed that eastward propagating westerlyanomalies in the tropical western Pacific associatedwith the MJO may play a pivotal role in the onsetof El Niƥno. Many subsequent studies have confirmedthe interactions between ISO in the tropical westernPacific(including the MJO) and ENSO, and have further shown that interannual variability in the intensity of ISO may also influence ENSO variability(Li, 1990; Li and Zhou, 1994; Li and Liao, 1998; Zhang and Gottschalck, 2002).

Numerical studies of ISO using atmosphericgeneral circulation models(AGCMs)have drawnwidespread attention in recent years. T otal errors innumerical forecasts are significantly related to the ability of the forecast model to accurately predict the MJO(Hendon et al., 2000). These errors arise in large partfrom biases in the amplitude(weaker than observed) and eastward propagation(faster than observed)ofthe simulated MJO(Jones et al., 2000). Slingo et al. (1996)ev aluated the ability of 15 AGCMs participating in the Atmospheric Model Intercomparison Project(AMIP)to simulate the MJO. They found that, although most models could capture ISO signals on intraseasonal timescales, none could reproduce the basicfeatures of the observed MJO(such as the eastwardpropagation speed of 5-9 m s^-1or the seasonal cycle of MJO activity). Numerical studies of the MJOhave been conducted by using a wide variety of models(Slingo and Madden, 1991; Maloney and Hartmann, 2001; Sperber, 2004; Kim et al., 2009), but the reasons that models are unable to accurately simulate theMJO remain unknown.

Chinese scientists have conducted a number ofstudies on ISO and the MJO in recent years, with significant progress on several fundamental issues. Someof these studies have focused on the impacts of theMJO on the spatiotemporal distribution of precipitation over China(Wu et al., 2009; Zhang et al., 2009;Bai et al., 2011; Jia and Liang, 2011; Jia et al., 2011;Zhang et al., 2011; Lü et al., 2012; Lin et al., 2013). Others have focused on how and why ISO influenceTC activity in the northwestern Pacific(Zhu et al., 2004; Chen and Huang, 2009; Sun et al., 2009; Pan et al., 2010; Tian et al., 2010a, b; Zhu et al., 2013). Recently, some promising results of ISO have been usedin the extended-range weather forecast, and positiveresults are obtained(Liang and Ding, 2012; Sun et al., 2013). Numerical studies have shown that the ability of a model to simulate the MJO depends not onlyon the ability of the model to simulate the large-scaleatmospheric circulation and climate, but also on themethod used to represent cumulus convection. Numerous studies have revealed the important role of diabatic heating in the lower level of the atmosphere forsimulations of the MJO(Dong and Li, 2007; Jia and Li, 2007a, b; Li et al., 2007; Jia et al., 2008, 2009; Li et al., 2009; Ling et al., 2009; Jia et al., 2010; Yang et al., 2012). Important results have also been obtained regarding the interactions between the MJO and ENSO. Here, we provide a general review and discussion aboutstudies of ISO and MJO conducted by Chinese scientists over the past decade. 2. Mechanisms of ISO/MJO

The mechanisms of ISO/MJO have been studied intensively since these phenomena were first identified; however, no existing hypothesis has been accepted as able to explain all of the observed featuresof ISO. Some early studies suggested that ISO wasrelated to gravity waves in the tropical atmosphere(Chang, 1977), while others suggested that ISO mightbe induced by symmetric and asymmetric instabilitynear the equator(Dunkerton, 1983). However, none ofthese hypotheses could explain the observed structure and eastward propagation of the MJO.

Li(1985)introduced the cumulus convective heating feedback mechanism(i. e., conditional instability ofthe second kind, or CISK), and showed that it couldplay an important role in the initiation and maintenance of MJO events in the tropics. Lau and Peng(1987)developed a mobile wave CISK theory thatcould better explain the slow eastward propagationof the MJO. Wang(1988)developed a frictional waveCISK theory and showed that the speed of the eastward propagating mode induced by deep convectivelatent heating triggered by low-level moisture convergence was similar to the observed propagation speedof the MJO. Li(1993)identified the potential for a dispersion CISK-Rossby wave to generate in the tropicalatmosphere. This type of disturbance can propagateeither westward or eastward when the CISK mechanism is introduced, and may be crucial for initiating and driving 30-60-day oscillations in the extratropicalatmosphere.

Neelin et al. (1987)proposed an ev aporation windfeedback mechanism, but subsequent studies indicatedthat this mechanism alone could not explain the initiation of the MJO. However, the ev aporation windfeedback mechanism could induce unstable waves inthe tropics, and may explain the characteristics of theMJO when combined with the CISK mechanism(Li, 1996). Li et al. (2002)studied MJO mechanismsusing an air-sea coupled model that included boththe CISK and ev aporation wind feedback mechanisms, and found that the CISK mechanism played a morecritical role in the dynamics of the simulated MJO. Air-sea coupling can reduce the frequency of the tropical waves induced by atmospheric disturbances, and could therefore play an important role in MJO dynamics.

Many numerical studies have confirmed the importance of lower tropospheric diabatic heating forsimulating the MJO. We discuss results of these studies in detail later. Latent heating released during theformation of stratus clouds in the lower tropospherecan also influence simulations of the MJO(Cha and Luo, 2011). Future studies of MJO mechanisms shouldpay particular attention to the distribution of diabaticheating in the middle and lower troposphere.

Zhang and Ling(2012)explored the dynamical structure and evolution of the MJO in terms ofpotential vorticity(PV). The structure of the MJOis an equatorial quadrupole of cyclonic and anticyclonic PV that tilts westward and poleward. ThisPV quadrupole is closely related to the swallowtailpattern of positive and negative precipitation anomalies associated with the MJO. Two processes dominatePV generation in the MJO. The first process is linear, and involves only MJO diabatic heating. The secondis nonlinear, and involves diabatic heating and relative vorticity perturbations outside the MJO domain. These results highlight that the MJO has characteristics of both self-sustenance(linearity) and multiplescale interactions(nonlinearity). The swallowtail precipitation pattern and the distribution of PV play important roles in the generation of the MJO.

Ling et al. (2014a)studied MJO events duringthe winter of 2006/2007 and identified large-scale patterns of PV and precipitation that favored MJO initiation. These patterns included a basin-scale(in thezonal direction)positive precipitation anomaly overthe Indian Ocean and a persistent vertical dipole ofPV generation(cyclonic PV generation in the lowertroposphere and anticyclonic PV generation in the upper troposphere). Ling et al. (2013b)identified threelarge-scale patterns that precede MJO initiation overthe Indian Ocean: low-level easterly anomalies thatmove from the western to the eastern Indian Ocean, zonal wavenumber-1 surface pressure anomalies withan equatorial low-pressure surge penetrating eastwardfrom Africa through the Indian Ocean to the Maritime Continent, and eastward-moving negative temperature anomalies in the middle to upper troposphereover the Indian Ocean. All three of these patternsemerge approximately 20 days before convective initiation of the MJO and propagate eastward at speedsclose to the typical propagation speed of the MJO(without any direct connection to MJO convection). These large-scale signals will be helpful for predictingthe onset of MJO events. 3. Numerical simulation of ISO

Numerical simulation of ISO has been a hot topic n studies of ISO over the past several decades. ManyAGCMs fail to reproduce the most salient features ofthe MJO, particularly its slow eastward propagationspeed(Slingo et al., 1996; Kim et al., 2009). Errorsrelated to the simulation of the MJO in numerical forecasts contribute substantially to total errors(Hendon et al., 2000). Typical errors in simulations of the MJOinclude an amplitude that is weaker than observed, an eastward propagation speed that is faster than observed, and an inability of the simulated MJO to propagate through the Maritime Continent(Jones et al., 2000). Improved numerical simulations of the MJOcould substantially improve the accuracy of weatherforecasts.

A number of factors can affect the ability of a numerical model to simulate the MJO. These factors often differ among different models. Some studies haveshown that increasing horizontal resolution can improve simulations of the MJO(Hayashi and Golder, 1986), while other studies have shown little improvement with enhanced resolution(Duffy et al., 2003). Li and Yu(2001)found that introducing air-sea interactions substantially improved the simulated MJO, while other studies found no significant improvement. A variety of studies have suggested that the ability ofa model to simulate the MJO depends critically on theconvective parameterization employed in the model(Slingo et al., 1996; Wang and Schlesinger, 1999; Maloney and Hartmann, 2001). Jia et al. (2009)con-firmed this result by using a variety of different cumulus parameterizations in the SAMIL(Spectral Atmosphere Model of State Key Laboratory of Numerical Modeling for Atmospheric Sciences and Geophysical Fluid Dynamics, Institute of Atmospheric Physics)AGCM. The ability of the model to simulate the MJOchanged substantially when different cumulus parameterization schemes were used. 3. 1 Imp acts of the vertic al distribution of diabatic heating

Li(1983)suggested that the vertical structure oflatent heating due to condensation substantially in-fluences the structure of the atmospheric circulation. Li et al. (2009)studied the impacts of the verticalstructure of diabatic heating on the simulation of theMJO in SAMIL-R42L9. This model includes three optional cumulus parameterization schemes: the Tiedtke(TK)scheme(Tiedtke, 1989), the moisture convection adjustment(MCA)scheme(Manabe and Strickler, 1964), and the Zhang-McF arlane(ZM)scheme(Zhang and McFarlane, 1995). Figure 1 shows thezonal propagation of MJO zonal wind anomalies at850 hPa simulated by SAMIL-R42L9 using the MCA and TK schemes. The model simulates the eastwardpropagation of the MJO well when the MCA schemeis used, but fails to simulate this feature when the TKscheme is used. These two model configurations generate substantially different vertical profiles of diabatic heating. Diabatic heating peaks in the lower troposphere when the MCA scheme is used, but no such apeak exists when the TK scheme is used(figure omitted). These results suggest that the ability of thismodel to simulate the MJO depends critically on thevertical structure of diabatic heating in the model atmosphere.

Fig. 1. Lag regression of MJO b and -filtered(time periods of 30-90 days and zonal wavenumbers 1-5)zonal wind at850 hPa(contour interv al 0. 2 m s-1)averaged over 15°S-15°N on the reference point at 150°E. Data are from SAMILsimulations using(a)the MCA convection scheme and (b)the Tiedtke convection scheme. Dashed contours indicatenegative values and zero contours are omitted. Shadings indicate significance at the 90% confidence level or above. Boldsolid lines indicate eastward propagation at a speed of 5 m s-1. [From Ling et al., 2013a]

Figure options

Besides the control(CT)experiment, two additional numerical experiments were conducted by usingSAMIL-R42L9 with the MCA scheme to further illustrate the impacts of different diabatic heating profileson the simulated MJO. In one experiment, the latentheating profiles were modified at each time step and each grid to be "top heavy"(TH)in the tropical region(20°S-20°N). In the other experiment, the latentheating profiles were modified to be "bottom heavy"(BH), with a peak within 500-600 hPa. The vertically integrated diabatic heating remained unchangedwhen vertical distribution of diabatic heating profileswas modified in these two experiments. The tropical mean profiles of total diabatic heating in the CT and BH experiments are largely similar(although onewas unmodified and the other was modified to increaselatent heating at lower levels and decrease latent heating at upper levels). The tropical mean profile of totaldiabatic heating in the TH experiment is clearly topheavy, with a smaller amplitude.

Figure 2 shows the time evolution of intraseasonaldiabatic heating profiles from these three simulationsalong with their associated zonal and vertical circulations. The diabatic heating anomaly and its associatedcirculation propagate eastward in the CT simulationat a phase speed of approximately 5. 5 m s^-1(periodof about 40 days). The circulation is characterizedby a deep baroclinic mode that extends upward to150 hPa, with mid-level upward motion, low-level convergence, and upper-level divergence in the region ofheating. This manifestation of the MJO is completelyabsent in the TH simulation. The heating and cooling are weaker with a smaller zonal scale than in CT. Furthermore, no eastward propagation is apparent ineither diabatic heating or the circulation. Evidently, the TH simulation does not produce the MJO. The results from the BH simulation are very similar to thosefrom CT. Even though the tropical latent heating pro-files were modified to be bottom heavy in BH, diabaticheating still penetrates upward into the upper troposphere. The eastward propagation speed of the MJO isslower( and closer to the observed propagation speed)in BH than in CT.

Fig. 2. Lag regression of intraseasonal(30-60-day)diabatic heating(color shading; K day-1) and zonal-vertical windvectors averaged over the tropics(15°S-15°N)on vertically integrated heating at 150°Ebased on the CT(left), TH(middle), and BH(right)experiments. [From Li et al., 2009]

Figure options

The results of these numerical experiments support the hypothesis proposed by Li(1983). The modelreproduces the eastward propagation of the MJO whenthe vertical profile of diabatic heating peaks in thelower troposphere. Heating at lower levels inducesstrong upward motion and low-level moisture convergence, and sets favorable conditions for the generation and maintenance of deep convection. Furthermore, lowering the peak of the diabatic heating profileleads to slower eastward propagation in the simulatedMJO. It is di±cult to induce strong upward motion and low-level moisture convergence when the diabaticheating profile peaks in the upper troposphere. Theseconditions impede the development and maintenanceof deep convection, and eliminate the MJO from themodel simulation. 3. 2 Imp acts of latent heating in the boundarylayer

Figure 1 shows that the SAMIL-R42L9 cannot accurately simulate the MJO when the TK scheme isused. This inability to simulate the MJO arises fromthe parameterization itself. Ling et al. (2013a)showedthat the vertical distribution of diabatic heating abovethe planetary boundary layer(PBL)in SAMIL is similar regardless of whether the MCA or TK scheme isused. The main differences in diabatic heating arelocated within the PBL. The TK scheme produces asignificant peak in latent heating in the PBL that doesnot appear when the MCA scheme is used. This distinct peak in latent heating in the PBL also appearsin global reanalysis data and in other AGCMs, such asthe Community Atmosphere Model version 3(CAM3) and the Tropical channel Weather Research and F orecasting(TWRF)model. These models are also unableto simulate the eastward propagation of the MJO. Distributions of latent heating in two recent global reanalysis datasets, the Climate Forecast System Reanalysis(CFSR; Saha et al., 2010) and the Modern Era Retrospective Analysis for Research and Applications(MERRA; Bosilovich et al., 2006), have obvious peaksin the PBL(Ling and Zhang, 2011). Free runs of the atmospheric models used in these two reanalysisdatasets have been unable to reproduce the characteristics of the MJO well(see results based on CFS and GEOS5 presented by Kim et al., 2009). The peakin latent heating in the PBL may be the reason thatSAMIL is unable to simulate the eastward propagation of MJO when the TK scheme is used. Numericalexperiments conducted using a version of SAMIL withincreased vertical resolution(SAMIL-R42L26)help toimprove the description of convection in the verticaldirection.

The Tiedtke scheme is a mass flux convectionscheme that can represent shallow, deep, and mid-levelconvective events(Tiedtke, 1989; Nordeng, 1994). Theimpacts of the latent heating peak in the PBL on thesimulation of the MJO are explored by using anotherset of sensitivity simulations. In one simulation, latent heating generated by shallow convection was setto zero at each time step; this simulation is referredto as the no-shallow-latent-heating(NSLH)run. Latent heating due to shallow convection was doubledrelative to the control simulation in the second sensitivity simulation; this simulation is referred to as thedouble-shallow-latent-heating(DSLH)run. T otal latent heating therefore does not match precipitation inthe NSLH or DSLH runs at grid locations where shallow convection occurs. The effects of this mismatchbetween precipitation and latent heating are relativelyminor because the water vapor source over the ocean(where most shallow convection occurs in this model)does not depend on precipitation. The equiv alent precipitation(EPR)derived from latent heating is usedas the regression index in place of precipitation. EPRis linearly related to vertically integrated latent heating. Figure 3 shows the propagation of b and -pass filtered 850-hPa zonal wind and EPR anomalies averaged over the tropics(15°S-15°N)for each of the threeexperiments. Neither 850-hPa zonal wind anomaliesnor EPR anomalies propagate eastward in the controlor DSLH runs(in fact, these anomalies show signs ofweak westward propogation). By contrast, anomaliesin 850-hPa zonal wind and EPR both propagate eastward from the Indian Ocean to the western PacificOcean in the NSLH run. The propagation speed for these anomalies is similar to(though slightly fasterthan)the observed propagation speed of the MJO. These results show that SAMIL using the TK schemecan simulate the eastward propagation of the MJOwhen latent heating from shallow convection is removed.

Fig. 3. Lag regression of MJO b and -filtered 850-hPazonal wind(U850; contours; interv al 0. 2 m s-1) and EPR(color shading; mm day-1)upon EPR at 90°E(day0)from(a)the control run, (b)the no-shallow-latentheating(NSLH)run, and (c)the double-shallow-latentheating(DSLH)run. All data are averaged over 15°S-15°N. Dashed contours indicate negative values and zerocontours are omitted. Only results passing a significancetest at the 90% confidence level or above are shown forEPR. Significance at the same level is shown by thick contours for U850. The thick blue solid lines indicate eastwardpropagation at a speed of 5 m s-1. [From Ling et al., 2013a]

Figure options

Figure 4 shows the distributions of latent heating and large-scale horizontal moisture convergence inthe control and NSLH runs together with their associated zonal and vertical circulations. Low-level moisture convergence is weak, shallow, and small(on zonalscale)in the control run. By contrast, low-level moisture convergence is stronger and deeper in the NSLH run, with a broader zonal scale. The NSLH run includes a strong center of low-level moisture convergence extending up to 500 hPa located east of thelatent heating center. This center of low-level moisture convergence supplies moisture to and establishesfavorable conditions for the generation of deep convection over the region east of the existing convection. This result is consistent with the role of boundary layer frictional convergence in MJO theory(Wang, 1988), and may explain the eastward propagation ofthe simulated MJO in the NSLH run. The differencesin low-level moisture convergences between these twosensitivity simulations are related to shallow convection simulated by the Tiedtke scheme, which generatesa peak in latent heating in the PBL. This peak tendsto confine moisture to the PBL and limit moisturetransport into the free atmosphere, leading to a drybias in the free atmosphere that effectively eliminatesMJO variability from the simulation.

Fig. 4. Longitude-pressure distributions of tropical(15°S-15°N)mean latent heating(contours; interv al 0. 1 K day-1), moisture convergence(color shading; 10-5g kg-1s-1), and zonal-vertical wind vectors regressed on MJO EPR at 90°Ewith a 0-day time lag based on(a)the control run and (b)the no-shallow-latent-heating(NSLH)run. Dashed contoursindicate negative values and zero contours are omitted. Only results passing a significance test at the 90% confidencelevel or above are plotted for moisture convergence and zonal-vertical wind. Significance at the same level is indicatedby thick contours for latent heating. [From Ling et al., 2013a]

Figure options

Li(1984)showed that vertical cumulus momentum transport could act as a kind of Ekman pumping that influences the formation and maintenance oftyphoons, as well as the development of the ITCZ. Modeling studies show that the mean state of the atmospheric circulation can be improved by introducingconvective momentum transport(Zhang and McFarlane, 1995; Gregory et al., 1997; Inness and Gregory, 1997). In particular, convective momentum transportsuppresses the development of deep convection(Tung and Yanai, 2002) and reduces precipitation( Wu et al., 2007)in the tropics. Relatively few studies have focused on the impacts of vertical cumulus momentumtransport on the simulation of the MJO. Ling et al. (2009)conducted several numerical experiments usingthe CAM2 AGCM with the Tiedtke scheme, and foundthat, although the general circulation in the model isimproved when vertical momentum transport is introduced, the ability of the model to simulate the MJOdeteriorates.

None of the model simulations presented in Section 3. 2 contain vertical cumulus momentum transport, and yet the simulated MJO is substantially different between the control and NSLH simulations. Ling and Li(2014)introduced vertical cumulus momentum transport to these two experiments(referredto as CTCMT and NSCMT, respectively).

The strength of the simulated MJO is weakenedwhen vertical cumulus momentum transport is introduced. In particular, the zonal propagation of EPR and 850-hPa zonal wind anomalies are dramaticallydifferent between the NSCMT and NSLH runs. Theamplitudes of these anomalies are reduced, and theeastward propagating signals over the western Pacificdisappear. The diabatic heating profiles in the controlexperiment do not change significantly when verticalcumulus momentum transport is included; however, the diabatic heating profiles in the NSLH experimentchange dramatically when vertical cumulus momentum transport is included. The reason for this difference is that deep convection is not simulated wellin the control experiment, so the impact of verticalmomentum transport by deep convection is not adequately represented. This difference is also reflected inthe simulations of precipitation. Changes in tropicalprecipitation following the addition of vertical momentum transport are much larger in the NSLH experiment than in the control experiment(figure omitted).

These results show that the ability of a modelto simulate the MJO can be significantly affected bythe inclusion(or exclusion)of vertical cumulus momentum transport, even for models that simulate theMJO well. The influence of vertical cumulus momentum transport on the simulation of the MJO also appears to depend on the ability of the original modelto simulate the MJO(especially its eastward propagation). 4. Relationship between the MJO and ENSO

The MJO and ENSO are two modes of climatevariability that operate on different timescales. Alarge number of studies have examined the relationships between these two modes of climate variability. Li(1989)suggested that a strong East Asian wintermonsoon could help to initiate an El Niƥno event by exciting a strong MJO over the equatorial western Pacific. Many subsequent studies have shown evidenceof interactions between the MJO over the equatorialwestern Pacific and ENSO(Li, 1990; Li and Zhou, 1994; Long and Li, 2002; Zhang and Gottschalck, 2002). The influence of the MJO on ENSO is largelyattributable to interannual variability in the intensityof MJO activity(Li and Liao, 1998; Li et al., 2003, 2008).

Numerous studies have shown strong relationships between the intensity of MJO activity overthe equatorial western Pacific and the occurrence ofENSO. MJO activity over the equatorial western Pacific is generally stronger before the occurrence of ElNiƥno, and then weakens rapidly as El Niƥno develops. A strong MJO is associated with strong westerly windanomalies over the equatorial western Pacific, whichinduce anomalous oceanic Kelvin waves that can trigger El Niƥno. The st and ard deviation of low frequency(30-60 days)kinetic energy is largest(> 0. 9)over theequatorial western Pacific(figure omitted), which indicates that MJO activity over the equatorial westernPacific plays an important role in the interannual variation of low frequency oscillations in the tropical atmosphere. Figure 5 shows the composite evolution of seasurface temperature anomalies(SST A)in the Niƥno 3. 4region and low frequency(30{60 days)kinetic energyanomalies at 850 hPa over the equatorial western Pacific(10°S-10°N, 130°E-180°)based on five strong ElNiƥno events. The SST A in the Niƥno 3. 4 region showsthe life cycle of El Niƥno. The MJO-related kinetic energy anomalies over the equatorial western Pacific arepositive during the spring and summer preceding themature phase of El Niƥno, but decrease significantlywhen El Niƥno reaches its mature phase. These resultssuggest a significant relationship between the interannual variability of MJO activity over the equatorial western Pacific and El Niƥno. A strong positiveanomaly in MJO activity over the equatorial westernPacific favors the development of an El Niƥno event;conversely, the development of an El Niƥno event inhibits MJO activity.

Fig. 5. Composite evolution of SST A(dashed line; ℃)inthe Niƥno 3. 4 region and kinetic energy anomalies associated with the atmospheric intraseasonal oscillation(KEA;solid line; m2s-2)over the equatorial western Pacific overfive strong El Niƥno events. The axis label 0 indicates theyear of El Niƥno onset. [From Li et al., 2008]

Figure options

A number of studies during the early 21st century have focused on characterizing the relationships between ENSO and the MJO. Several studiesabroad show clear relationships between MJO variability and ENSO variability(Roundy and Kriavitz, 2009;Gushchina and Dewitte, 2012), and support the ideathat the intensification of MJO activity over the equatorial western Pacific during boreal spring favors thedevelopment of El Niƥno(Hendon et al., 2007; Marshall et al., 2009). Some studies suggested that the relationships between the MJO and ENSO are due to theresponse of the ocean to MJO activity(Zavala-Garay and Zhang, 2004; Zavala-Garay et al., 2008; Seiki et al., 2009), which is inherently nonlinear( Tang and Yu, 2008). The influence of the MJO on ENSO has alsobeen investigated by using the delayed oscillator model(Richard et al., 2008). Analyses of observ ations and numerical simulations indicate that the influence ofMJO activity on ENSO can be treated as a stochasticforcing(Kapur et al., 2011).

Chinese scientists made several promising advances toward underst and ing the relationships between the MJO and ENSO over the past several years. Liu et al. (2008)used an intermediate air-sea coupledmodel to study the influence of the MJO on ENSO. Their experiments involved forcing the ocean component with MJO zonal wind anomalies of different amplitudes. Several major features of ENSO are successfully simulated in the control run, such as the typical 4-yr period of oscillations in SST A in the Niƥno 3region(5°S-5°N, 90°-150°W), the tendency of ENSOevents to occur predominantly between September and December, and the 2-7-yr interv al between successiveENSO events. The results of these simulations suggest that weak MJO signals can intensify the amplitude of ENSO, while very strong MJO signals decreasethe amplitude of ENSO. The typical period of ENSOvariability in the model is unaffected by the existenceof the MJO forcing. Peng et al. (2011)suggested thatstochastic forcing associated with the MJO affects thepredictability of ENSO. Rong et al. (2011)showedthat high frequency(< 90 days)variability in nearsurface winds impacts the development of SST A associated with ENSO. Most high frequency variability innear-surface wind in the equatorial western Pacific isrelated to MJO activity, so these impacts can be regarded as indicative of the relationship between MJOactivity and ENSO.

El Niƥno events can be classified into three categories: Eastern Pacific(EP)El Niƥno, Central Pacific(CP)El Niƥno, and mixed-type El Niƥno( Yuan, 2009). MJO kinetic energy over the western Pacificmay influence the evolution of these different types ofEl Niƥno in different ways. Figure 6 shows time-latitudecross-sections of MJO kinetic energy at 850 hPa and its anomaly averaged over the western Pacific region(120°-160°E)associated with the three types of ElNiƥno. EP El Niƥno(Fig. 6a)is associated with strongMJO activity both in the Northern Hemisphere fromthe previous winter through the onset of El Niƥno and in the Southern Hemisphere between June and October before the onset of El Niƥno. The strength ofMJO is reduced dramatically after the onset of EPEl Niƥno. This negative anomaly can last through thefollowing winter. CP El Niƥno(Fig. 6b)is associatedwith seasonal north-south migration in MJO anomalies before the onset of El Niƥno. Anomalies in MJOkinetic energy are strong and positive in the SouthernHemisphere during the preceding boreal winter and inthe Northern Hemisphere during the preceding borealspring and summer. The onset of CP El Niƥno leads tothe emergence of negative anomalies in MJO kineticenergy near 10°N, but enhances the positive anomaliesin the Southern Hemisphere. Mixed-type El Niƥno(Fig. 6c)is associated with enhanced MJO activity in boththe Southern and Northern Hemispheres during theperiod May{August preceding the El Niƥno. Anomalies in MJO kinetic energy are positive near 10°S butnegative in the Northern Hemisphere following the onset of El Niƥno. MJO activity is anomalously strongover the equatorial western Pacific before the onset ofeach type of El Niƥno, despite some differences in the detailed distribution.

Fig. 6. Time-latitude cross-sections of MJO kinetic energy at 850 hPa(contours; m2s-2) and its anomaly(colorshading; m2s-2)in the western Pacific(120°-160°E)during(a)EP, (b)CP, and (c)mixed-type El Niƥnos, respectively. The axis label 0 indicates the year of El Niƥno onset(1 indicates the following year).

Figure options

MJO activity can also be described by using outgoing longwave radiation(OLR), which captures thestrong convective activity associated with the MJO. The canonical dipole pattern of OLR anomalies associated with ENSO can be identified over the tropical Pacific Ocean for all three types of El Niƥno(figure omitted), but the distribution of OLR anomalies is somewhat different. The dipole is strongest in amplitude, largest in zonal scale, and longest in duration(nearly 1yr)for EP El Niƥno; and weakest in amplitude, smallestin zonal scale, and shortest in duration(6 months)forCP El Niƥno. The positive and negative OLR anomalies are centered near 160°W and 130°E, respectively, during EP El Niƥno; and near 180° and 110°E, respectively, during CP El Niƥno. The entire dipole is shiftedwestward during CP El Niƥno relative to that during EP El Niƥno. The evolution of these OLR anomalieshas some similarities for the three types of El Niƥno, along with several differences. The characteristics ofthis evolution indicate that MJO activity and El Niƥnodevelopment are closely related, and that MJO activity might be different under different types of El Niƥno. 5. Influences of ISO on weather and climate 5. 1 ISO and TC genesis over the northwesternPacific 5. 1. 1 Imp acts of MJO on TC genesis over the northwestern Pacific

Pan et al. (2010)examined relationships betweenMJO variability and TC activity over the northwesternPacific using the real-time multivariate MJO(RMM)index(Wheeler and Hendon, 2004)generated at theCentre for Australian Weather and Climate Research. The RMM index describes both the strength of theMJO and its spatial distribution(particularly the location of the active phase of MJO convection). Thenumber of TCs generated over the western Pacific during typhoon season(June-October)are counted and classified according to different phases of the MJO between 1979 and 2004 based on the TC data compiledby the Joint Typhoon Warning Center(JTWC), theShanghai Typhoon Institute, and the Japan Meteorological Agency(JMA). MJO variability has substantialimpacts on TC genesis. The number of TCs generatedduring typhoon season under strong MJO events is almost twice the number of TCs generated under weakMJO events. The results are consistent regardless ofthe TC dataset used. This result indicates that TCsare more likely to form during strong MJO events. TCgenesis also depends on the phase of the MJO. Thetotal number of TCs generated during phases 5 and 6(when the active convection center is located overthe western Pacific; hereafter referred to as the Pacificphase), is much greater than the number of TCs generated during phases 2 and 3(when the active convection center is located over the Indian Ocean; hereafterreferred to as the Indian Ocean phase).

A number of factors contribute to the differencesin the number of TCs generated during the Pacific and Indian Ocean phases of the MJO. Composite distributions of the large-scale circulation, sea surfacepressure, ITCZ, vertical motion, and convective heating during the Pacific phase are substantially differentfrom those during the Indian Ocean phase. The IndianOcean phase of the MJO is characterized by positiveanomalies of sea surface pressure, large vertical windshear, negative anomalies in convective heating, and weak vertical motion over the northwestern Pacific. All of these conditions inhibit the genesis and development of TCs over the northwestern Pacific. By contrast, the Pacific phase of the MJO is characterizedby negative anomalies of sea surface pressure, smallvertical wind shear, positive anomalies in convectiveheating, and strong vertical motion over the northwestern Pacific. These conditions all favor TC genesisover the northwestern Pacific. The large-scale dynamical environment over the northwestern Pacific, whichplays an important role in TC genesis, changes significantly as the MJO propagates eastward from thetropical Indian Ocean. 5. 1. 2 Imp acts of ISO on TC genesis over the northwestern Pacific

Tian et al. (2010b)further studied the impactsof ISO on TC genesis in the northwestern Pacific bycomparing the differences between composites of lowfrequency kinetic energy at 850 hPa for years withlarger and smaller numbers of TCs. Positive anomalies in low-frequency kinetic energy are observed overthe northwestern Pacific east of the Philippines and south of 15°N(an area typically dominated by themonsoon trough)during the years with large numbersof TCs. This relationship indicates that strong lowfrequency activity can enhance the monsoon trough and promote TC genesis. By contrast, the years withless TC genesis are characterized by positive anomalies in low-frequency kinetic energy over the IndianPeninsula and the southern part of the South ChinaSea and negative anomalies over the area east of thePhilippines.

The amplitude of ISO can also be quantified interms of the 30-60-day b and pass-filtered zonal windat 850 hPa. ISO activity is considerably different during years with more TC genesis than during years withless TC genesis. In particular, ISO activity in the monsoon trough region over the northwestern Pacific tendsto be strong during years with more TC genesis and weak during years with less TC genesis.

A strong low-frequency cyclonic circulation extends eastward to 160°E over the western Pacific during years with more TC genesis, covering approximately the same area as the monsoon trough. Thissuggests that strong ISO activity intensifies the lowpressure system associated with the monsoon trough and extends it toward the east. F urthermore, the lowfrequency velocity potential at 200 hPa indicates significant divergence over the northwestern Pacific tothe east of the Philippines during years with more TCgenesis. Both of these conditions are favorable for TCgenesis in the northwestern Pacific. 5. 2 ISO and TC tr acks over the northwesternPacific

Forecasts of TC tracks are an important part ofpredicting the activity and impacts of TCs. Previousstudies have shown that the effects of ISO on the largescale circulation(particularly the distribution and location of the monsoon trough and subtropical high)influence TC tracks in the northwestern Pacific(Harr and Elsberry, 1991; Carr and Elsberry, 1995; Hu et al., 2005). These relationships between ISO activity and TC tracks can be used to improve TC forecasts.

TC tracks in the northwestern Pacific can traditionally be classified into three categories: tracks thatmove directly westward, tracks that move northwestward, and tracks that recurve. Tian et al. (2010a)further divided the recurving tracks into three categories based on the direction of movement after therecurve: tracks that recurve to the west of Japan(i. e., those that approach the Korean Peninsula after recurving), tracks that make l and fall over Japan, and tracks that recurve to the east of Japan. Tracks thatmove directly westward are the most common of thesetypes during the typhoon season, followed by recurving TCs that make l and fall over Japan and recurving TCs that pass to the east of Japan. The occurrence frequency of these different types of TC tracksalso varies from month to month. The occurrence frequency of westward-moving typhoons is highest inJuly, while the occurrence frequencies of northwestward-moving TCs and recurving TCs that make l and fall over Japan are highest in August. A full half of thenorthwestward-moving TCs occur in August, while noTCs of this type occur in June or October. Most ofthe TCs that recurve to the east of Japan occur inOctober.

Figure 7 shows composite low-frequency(ISO)horizontal wind fields at 850 hPa corresponding to thefive different types of TC tracks(Tian et al., 2010a). These composites are constructed relative to the timeof TC genesis over the northwestern Pacific. All fivetypes of TC tracks occur under the influence of lowfrequency cyclonic(LFC)anomalies over East Asia and the northwestern Pacific, but the location and pattern of the LFC anomaly are unique for each type. TCs that move directly toward the west(Fig. 7a)areassociated with an anomalous LFC circulation overthe South China Sea(SCS) and the Philippine Sea, with a low-frequency anticyclone(LF AC)anomaly located to its north. TC tracks that move directly toward the northwest(Fig. 7b)are associated with ananomalous LFC circulation over T aiwan Isl and, whichextends eastward to 160°E. This circulation acts to intensify the monsoon trough and extends it toward thewest, and is located southwest of an anomalous LF AC. The belt of the anomalous LFC(i. e., the region of maximum positive vorticity)extends zonally from theSCS to the Philippine Sea for westward-moving TCs, but tilts toward the northwest for northwestwardmoving TCs. TCs tend to move along this belt ofpositive low-frequency vorticity.

Fig. 7. Composites of 30-60-day b and pass-filtered wind vectors(m s-1)at 850 hPa for different types of typhoontracks:(a)typhoons that move directly westward, (b)typhoons that move northwestward, (c)typhoons that recurveto the west of Japan, (d)typhoons that make l and fall over Japan, and (e)typhoons that recurve to the east of Japan. Light shadings indicate significance at the 95% confidence level, while dark shadings indicate significance at the 99%confidence level. [From Tian et al., 2010a]

Figure options

TCs that recurve to the west of Japan(Fig. 7c)are associated with an anomalous LFC circulation overthe East China Sea and Y ellow Sea and a stronganomalous LF AC circulation near 30°N, 130°-160°E. TCs in this regime generally move northward over thebelt of positive low-frequency vorticity west of 130°E and make l and fall over the Korean Peninsula. Recurving TCs that make l and fall over Japan(Fig. 7d)areassociated with an anomalous LFC circulation thatcovers a large area between the Philippines and Japan(10°-35°N, 120°-145°E). TCs under this regime movenorthward over Japan following this belt of positivelow-frequency vorticity. TCs that recurve to the eastof Japan(Fig. 7e)are associated with an anomalousLFC circulation that slants northeastward from southern Japan and an anomalous LF AC circulation centered near 20°N, 150°-166°E. These TCs follow thisbelt of positive low-frequency vorticity northeastwardover the ocean to the east of Japan. The pattern ofthe anomalous LFC circulation at 850 hPa(particularly the belt of positive low-frequency vorticity)playsa key role in determining TC tracks over the northwestern Pacific. This relationship between ISO and TC tracks can be used to improve predictions of TCtracks in the northwestern Pacific.

All five types of TC tracks are consistently associated with low-frequency westerly anomalies near theequator between the Indian Ocean and 150°E. Thisrelationship further confirms the positive relationshipbetween ISO activity in the tropical Pacific and TCgenesis over the northwestern Pacific.

Analysis of composite low frequency wind fieldsat 200 hPa(figure omitted)further supports these results. TC tracks generally follow regions of strong horizontal winds located southwest of the upper tropospheric LF AC. The composite location and horizontaldistribution of the upper tropospheric LF AC and theamplitude and direction of the maximum horizontalwind southwest of this LF AC differ for different typesof TC tracks. These differences are then manifested inthe direction of TC movement. 5. 3 Imp acts of ISO on the Asian summermonsoon 5. 3. 1 ISO and the onset of the Asian summer monsoon

Mu and Li(2000)studied the relationship of theonset of the South China Sea(SCS; 5°-20°N, 105°-120°E)summer monsoon and local ISO activity. Theresults have highlighted the important roles of thetemporal evolution of 850-hPa zonal wind, the lowfrequency(30-60 days)variability in zonal wind, and low-frequency variability in kinetic energy. Increasesin the amplitude of low-frequency westerly winds overthe SCS are primarily attributable to both westwardextension of low-frequency westerly wind variabilityfrom the east and local excitement(figure omitted).

The onset of the SCS summer monsoon is preceded by the development of a symmetric pair of off-equatorial cyclones at 850 hPa over the tropical IndianOcean(Zhou and Chan, 2005). The development ofthis symmetric pair of cyclones is dominated by 30-60-day low-frequency waves. These results suggest thatthe onset of the SCS summer monsoon is fundamentally related to ISO variability over the tropical IndianOcean. The development of the symmetric cycloniccirculations leads the onset of the SCS summer monsoon by approximately 5-10 days. The emergence ofthe low-frequency cyclonic circulations over the tropical Indian Ocean can therefore be considered an indicator of the imminent onset of the SCS(or Asian)summer monsoon. 5. 3. 2 Imp acts of ISO on the East Asian summer monsoon

Variability in the East Asian(EA)summer monsoon dramatically impacts the weather and climate ofAsia, including the occurrence of floods and droughtsin East China. Li et al. (2001)have identified yearsof strong(1981, 1984, 1985, 1986, 1990, 1992, and 1997) and weak(1980, 1983, 1987, 1989, 1991, 1993, and 1998)SCS/EA summer monsoons. The composite circulation anomalies for years with strong summermonsoons are distinctly different from the composite anomalies for years with weak summer monsoons. Y ears with strong summer monsoons are characterizedby stronger westerly winds between 5° and 20°N and stronger easterly winds between 5° and 20°S. The cyclonic circulation located northeast of the SCS is alsostronger during these years. ISO activity at 850 hPa isintense during strong monsoon years, particularly overthe SCS and Philippines. ISO activity is much weaker(by about a factor of two)during years with weak summer monsoons. The most intense ISO activity duringweak years is located over the northwestern Pacific(20°N, 140°E). The formation of a strong cyclonic circulation is one of the most important characteristicsof a strong SCS summer monsoon. ISO activity and low-frequency cyclonic circulation anomalies over theSCS and adjacent regions play an important role in determining the strength of this cyclonic circulation and therefore the strength of the EA summer monsoon.

A powerful anticyclone develops in the upper troposphere(approximately 200 hPa)over the TibetanPlateau during the Asian summer monsoon. This upper tropospheric anticyclone is observed during bothstrong and weak summer monsoons, but it is stronger and displaced toward the northwest during strongsummer monsoons. Moreover, the most pronounceddifference in ISO activity between strong and weaksummer monsoons is located over the Tibetan Plateau(figure omitted). In particular, low-frequency anticyclonic circulation anomalies over the Tibetan Plateauare more intense during strong summer monsoons. This relationship indicates that ISO activity over theTibetan Plateau(particularly, the low-frequency anticyclonic circulation anomalies)plays an importantrole in the onset and maintenance of strong EA summer monsoons. 5. 3. 3 Imp acts of ISO on the Indian summer monsoon

and precipitation over Yunnan Province, ChinaThe Indian(South Asian)summer monsoon is animportant component of the Asian summer monsoonsystem and has significant impacts on summer weather and climate throughout Asia, including China. Theonset and variability of the South Asian summer monsoon are closely related to ISO activity( Murakami et al., 1986; Wang and Xu, 1997; Goswami and Mohan, 2001; Goswami et al., 2003). The evolution of prevailing westerlies in the lower troposphere(one of thedistinguishing features of the Indian summer monsooncirculation)depends fundamentally on the nonlineareffects of ISO(Qi et al., 2009). Nonlinear momentum transport by ISO disturbances contributes approximately 40% of the momentum required to establish westerly winds over the Indian monsoon regionin early June, and also contributes to the weakening of these westerly winds after mid July. Surfacewind convergence and air-sea interactions associatedwith MJO convection in the western equatorial IndianOcean propagate both eastward and northward, during the onset of the Indian summer monsoon(Qi et al., 2008).

Interannual variability in the Indian summermonsoon and interannual variability in local ISO activity are out of phase. Strong ISO activity is associatedwith anticyclonic anomalies in the lower troposphereover the Indian subcontinent, corresponding to a weakIndian summer monsoon(Qi et al., 2009). Conversely, weak ISO activity is associated with cyclonic anomalies that correspond to a strong Indian monsoon. Astrong Indian summer monsoon also tends to suppressconvection in the eastern equatorial Indian Ocean and weaken northward-propagating ISO, thus resulting inweak ISO activity in the Indian monsoon region(Qi et al., 2008). These results indicate substantial interactions between ISO and the Indian summer monsoon.

Lü et al. (2012)have shown that persistent MJOanomalies over the central and eastern tropical IndianOcean can influence summer precipitation in Y unnanProvince, China. A persistent positive phase of theMJO suppresses convective activity over the Bay ofBengal. This situation is associated with drought conditions in Y unnan Province. By contrast, a persistentnegative phase of the MJO enhances convective activity over the Bay of Bengal and increases summerrainfall over most parts of Y unnan Province(exceptnorthwestern Y unnan and the eastern parts of centralY unnan). These results indicate that interannual variations in MJO activity over the tropical Indian Oceanhave a significant impact on summer precipitation overY unnan Province. A positive MJO induces subsidence over the tropical Indian Ocean region(70°-110°E). This may lead to a weaker monsoon vertical circulationover South Asia, decreasing moisture transport fromthe tropical Indian Ocean, and reducing precipitationover Y unnan Province. 5. 4 The MJO and precipitation over China

Impacts of MJO variability on the spatiotemporaldistribution of precipitation over China have been investigated extensively during recent years(Wu et al., 2009; Zhang et al., 2009; Bai et al., 2011; Jia and Liang, 2011; Jia et al., 2011; Zhang et al., 2011; Lü et al., 2012; Lin et al., 2013). The distribution and magnitude of precipitation over different parts of Chinavary significantly under different phases of the MJO.

The MJO is a cluster of strong convective eventspropagating eastward along the equator. In addition to tropical Rossby and Kelvin waves, this clusterof strong convective events induces an extratropicalRossby wave train that emanates from the tropics toward high latitudes. This extratropical Rossby wavetrain induces remote responses in precipitation. Thespatial pattern of these responses is very different under different seasons and different phases of the MJOdue to changes in the atmospheric background state and the location of the anomalous convective heating. Anomalous convective activity associated withthe different phases of the MJO helps to establish favorable(or unfavorable)conditions for rainfall overvarious parts of China(especially East China), and istherefore an important factor in determining the distribution of anomalous precipitation over China. 6. Conclusions

ISO(including the MJO)is an important component of the climate system. ISO( and associatedanomalies)exerts substantial influences on many aspects of weather and climate, so underst and ing thestructures, characteristics, and dynamical mechanismsof ISO is prerequisite to underst and ing the climate system. Studies of ISO therefore represent one of the frontiers of atmospheric sciences. This study has brieflyreviewed and summarized the main achievements ofChinese scientists in this field over the past decade. The main results are as follows.

The dynamical mechanisms of tropical ISO, and particularly the MJO, have been under investigationsince the 1970s. Studies have suggested that feedbacksassociated with heating due to cumulus convection arefundamental to the existence of the MJO. Recent numerical simulations support this idea, and reveal thatthe vertical distribution of diabatic heating in the tropics plays an important role in the ability of a model tosimulate the MJO. The convective initiation of MJOhas drawn considerable attention in recent years, and was a primary focus of the DYNAMO(see Ling et al., 2014b for abbreviation)field campaign(2011-2012). However, further studies are still needed to resolve theconsiderable uncertainties regarding the mechanismsthat underlie the MJO.

Numerical simulations are an important resourcefor underst and ing the mechanisms behind tropical ISO and the MJO; however, AGCMs still fail to simulatemany of the observed features of these phenomena. The choice of convective parameterization appears tobe one of the most important factors in determiningwhether a model is able to simulate a realistic MJO. Itis critical that diabatic heating profiles simulated bythe convection scheme peak in the middle and lowertroposphere. This distribution of diabatic heating facilitates strong upward motion and low-level moistureconvergence, and establishes favorable conditions forthe generation and development of deep convection.

Tropical ISO and the MJO evidently interact withENSO, especially over the equatorial western Pacific. Strong MJO can induce westerly wind bursts over thewestern Pacific and excite oceanic Kelvin waves thatfavor the onset and development of El Niƥno. Composite analysis indicates that MJO activity is typicallystrong during the boreal spring preceding an El Niƥnoevent. Interannual variability in the MJO can alsoaffect the onset of El Niƥno via the response of theair-sea coupled system. The mature phase of El Niƥnosuppresses MJO activity. The relationships betweenMJO activity and different types of El Niƥno are different. F urther studies will be required to better characterize the relationships between ENSO and the MJO.

Tropical ISO and their associated circulation patterns have important influences on TC activity in thewestern Pacific. Approximately twice as many TCsform in the northwestern Pacific during strong MJOevents relative to weak MJO events. Similarly, approximately twice as many TCs form in the northwesternPacific during the Pacific phase of MJO relative to theIndian Ocean phase of the MJO. The MJO also modulates the locations of TC genesis in the northwesternPacific. The locations of TC genesis are generallyconfined to the western Pacific region south of 20°Nduring the Indian Ocean phase of the MJO, but extend northward to 30°N during the Pacific phase ofthe MJO. These two phases of the MJO are characterized by obvious differences in the distributionsof the large-scale circulation, sea surface pressure, the ITCZ, vertical motion, and convective heating. The Indian Ocean phase is associated with positiveanomalies in sea surface pressure, large vertical windshear, negative anomalies in convective heating, and weak vertical motion over the northwestern Pacific. These conditions inhibit the genesis and developmentof TCs over the northwestern Pacific. By contrast, the Pacific phase is associated with negative anomalies in sea surface pressure, small vertical wind shear, positive anomalies in convective heating, and strongvertical motion over the northwestern Pacific. Theseconditions favor TC genesis over the northwestern Pacific.

ISO impacts TCs via the formation of LFC and LF AC circulation anomalies in the lower troposphere. The genesis locations for five typical TC tracks inthe western Pacific all have close relationships withlow-frequency circulation patterns at 850 hPa. Theorientation of the belt of low-frequency cyclonic vorticity at the time of TC genesis is a good predictorof subsequent TC movement. All five TC track typesare also associated with an anomalous LF AC at 200hPa. The airflow along the south and west sides ofthis LF AC is also a good predictor of the TC track. The direction of the airflow along the south and westsides of the LF AC and the spatial extent of the LF ACboth vary among different TC track types.

Tropical ISO activity(including the MJO)has asignificant impact on the onset and variability of theSCS and EA summer monsoons. The onsets of theSCS and Indian summer monsoons are both associatedwith stronger tropical ISO; however, these two monsoon systems have different relationships with ISO. ISO activity over the SCS and its adjacent regionsplays an important role in determining the strengthof the EA summer monsoon, with strong ISO activity indicating a strong monsoon. Conversely, strongISO activity corresponds to a weak Indian summermonsoon, and vice versa. Precipitation over Chinais closely related to tropical ISO activity, particularlythe MJO.

The study of tropical ISO(including the MJO)remains a hot scientific topic with a number of unresolved issues. Ongoing research focuses on severalareas, including the numerical simulation and forecasting of tropical ISO, the influences of tropical ISO onweather and climate(especially the role of ISO in 15-30-day extended forecasts), the interactions betweentropical ISO and ENSO, and the physical mechanismsbehind the convective initiation of the MJO over thewestern equatorial Indian Ocean.

Several other items are worthy of special attention. First, tropical ISO is the most important component of the low-frequency variability in the tropicalatmosphere, but quasi-biweekly(10-20-day)oscillations also contribute. The structures, activity, and mechanisms of ISO and quasi-biweekly oscillationsare significantly different. As a consequence, quasibiweekly oscillation should not be treated as tropicalISO, and low-frequency oscillations should not betreated in a uniform way. Second, the MJO is an important and well-known example of tropical ISO. TheMJO refers specifically to an eastward-propagatingISO near the equator identified in the 1970s. ISOwith characteristics that differ from those of the MJO(especially with respect to zonal propagation)exists inthe subtropics. Therefore, the MJO and tropical ISOshould not be treated as synonymous or interchangeable. Third, the convective initiation of the MJO hasrecently drawn considerable attention. The convectiveMJO signal generally initiates over the western equatorial Indian Ocean, and then propagates eastwardbefore disappearing around the dateline. The MJO activity over the Indian Ocean and the western Pacificcan be well represented by the associated convection;however, Madden and Julian(1971, 1972)showed thatthe MJO has a planetary scale. The convection associated with the Indian Ocean and western PacificMJO is only a part of the MJO structure, and doesnot represent the full characteristics of the MJO.