Abstract

This study examines the modulation of tropical cyclone (TC) activity by the Madden–Julian oscillation (MJO) in the Fiji, Samoa, and Tonga regions (FST region), using Joint Typhoon Warning Center best-track cyclone data and the MJO index developed by Wheeler and Hendon. Results suggest strong MJO–TC relationships in the FST region. The TC genesis patterns are significantly altered over the FST region with approximately 5 times more cyclones forming in the active phase than in the inactive phase of the MJO. This modulation is further strengthened during El Niño periods. The large-scale environmental conditions (i.e., low-level relative vorticity, upper-level divergence, and vertical wind shear) associated with TC genesis show a distinct patterns of variability for the active and inactive MJO phases. The MJO also has a significant effect on hurricane category and combined gale and storm category cyclones in the FST region. The occurrences of both these cyclone categories are increased in the active phase of the MJO, which is associated with enhanced convective activity. The TCs in the other MJO phases where convective activity is relatively low, however, show a consistent pattern of increase in hurricane category cyclones and a concomitant decrease in gale and storm category cyclones. Finally, TC tracks in different MJO phases are also objectively described using a cluster analysis technique. Patterns seen in the clustered track regimes are well explained here in terms of 700–500-hPa mean steering flow.

1. Introduction

It is well known that tropical cyclone (TC) activity in most ocean basins is strongly influenced by various modes of natural climate variability (see Camargo et al. 2010 for a review). In the tropical Pacific, the El Niño–Southern Oscillation (ENSO; e.g., Trenberth 1997) and the Madden–Julian oscillation (MJO; Madden and Julian 1971, 1994) are the two major modes of natural climate variability affecting TC activity on seasonal and intraseasonal time scales, respectively. A review of literature reveals that the influence of ENSO on TC activity has garnered more attention in several TC basins and subbasins than that of the influence of the MJO. Recently, for example, Chand and Walsh (2009, hereafter CW09) examined the effects of ENSO on TC genesis positions and tracks in the Fiji region. The current investigation builds on the work of CW09, except that the emphasis now is on the MJO.

Gray (1979) studied each of the world’s TC basins and found that TC formations are not a series of evenly distributed events, but rather have a tendency to cluster in time with periods of approximately 2–3 weeks of active formation followed by a similar period of quiescence. Subsequent studies, for example, in the western North Pacific basin (Liebmann et al. 1994; Sobel and Maloney 2000), Australian region (Hall et al. 2001), South Indian Ocean (Bessafi and Wheeler 2006; Ho et al. 2006), and elsewhere (e.g., Maloney and Hartmann 2000a,b, 2001; Barrett and Leslie 2009), have related such intraseasonal clusterings to the MJO.

The MJO is a leading mode of intraseasonal atmospheric variability in the tropical Pacific and can be summarized as an eastward-propagating disturbance with a period of about 30–90 days (see the comprehensive review of Zhang 2005). The passage of an active (inactive) phase of the MJO is associated with enhanced (suppressed) tropical disturbances over a region. Liebmann et al. (1994), for example, observed that temporal and spatial clusterings of TC formations in the Indian and the western Pacific Oceans are a result of subsequent enhancement and suppression of climatologically favorable environmental conditions by the MJO. Maloney and Hartmann (2000b) found a significant modulation of eastern North Pacific TC numbers and intensity and have also attributed the modulation to changes in low-level relative vorticity and vertical wind shear caused by MJO influences. Hall et al. (2001) determined that the MJO strongly modulates TC genesis in the Australian region (i.e., between the longitudes 80° and 170°E), with strengthening of the MJO–TC relationship during El Niño periods. They also found that changes in TC genesis patterns associated with the MJO are strongly related to 850-hPa cyclonic relative vorticity anomalies. Similarly, Bessafi and Wheeler (2006) found a substantial MJO modulation of TC genesis over the South Indian Ocean and attributed such modulation to the changes in large-scale environmental conditions such as low-level cyclonic relative vorticity, vertical wind shear, and deep convection. More recently, Camargo et al. (2009) explored the MJO modulation of TC genesis in all TC basins using an empirical genesis potential index developed by Emanuel and Nolan (2004). They found that the modulation of TC genesis frequency by the MJO, as implied by the genesis potential index, is primarily caused by changes in midlevel relative humidity and low-level absolute vorticity with only minor contributions from vertical wind shear and potential intensity.

Results from the above studies have greatly advanced our knowledge of the modulation of TC activity by the MJO for various TC basins across the world. However, a similar quantitative study, like that of Hall et al. (2001) for the Australian region or Bessafi and Wheeler (2006) for the South Indian Ocean, is lacking for Fiji, Samoa, and Tonga (FST) regions (Fig. 1). The primary objective of this study is therefore to document and understand the modulation of TC genesis, tracks, and intensity by the MJO in the FST region. The MJO modulation of TC activity in different phases of ENSO will also be investigated. The other objective is to determine the influence of MJO on large-scale environmental conditions necessary for TC genesis in the region. Some earlier studies in the South Pacific basin that spanned the FST region in full (e.g., von Storch and Smallegange 1991) or in part (e.g., Liebmann et al. 1994) were restricted to relatively few years of data, which made it possible that any signal could have been contaminated by, for example, ENSO-related phenomena (Hall et al. 2001). Our study, however, differs in that it makes use of the data since routine satellite observations became available to examine quantitatively the extent to which MJO modulates not only TC genesis patterns but also TC tracks and intensity over the FST region. It also seeks to establish the relationship of these observed changes in TC genesis patterns, tracks, and intensity to the large-scale environmental conditions.

The present investigation therefore has obvious benefits for the FST region. In addition to improved understanding of MJO–TC relationships, the results will be helpful for potential future development of intraseasonal TC prediction schemes (such as Leroy and Wheeler 2008). We begin with a description of datasets used in the study and an outline of methodology adopted for data analysis (section 2). Section 3 describes the results of the investigation. Finally, a discussion and summary are given in section 4.

2. Data and methodology

a. TC data and definitions

This study is restricted to the austral summer season (i.e., only November–April) for the period beginning November 1970 and ending April 2006. The TC data used here is that archived by Joint Typhoon Warning Center (JTWC 2009) at 6-h intervals. The time period chosen for study is consistent with the era after which routine satellite observations became available. To ensure maximum accuracy, observations within the JTWC database were cross referenced with information from the Fiji Meteorological Services. Consistent with CW09, we have included all TCs forming in the FST region (defined as the area between 5°–25°S and 170°E–170°W), as well as those crossing this defined boundary during some part of their lifetime. Altogether, 115 TCs are considered in the analysis. The TC formation is defined here as the first track point in the JTWC database and is used interchangeably with genesis and development throughout the paper.

b. MJO index and TC binning procedure

It is not the scope of the present investigation to derive the basic states of the MJO. In fact, several recent studies (e.g., Hall et al. 2001; Wheeler and Hendon 2004; Bessafi and Wheeler 2006 and others) have already done so. In our study, we used the real-time multivariate MJO (RMM) index of Wheeler and Hendon (2004) to investigate the extent to which MJO modulates TC activity in the FST region. This RMM index is based on the first two empirical orthogonal functions (EOFs) of the combined fields of near-equatorially averaged 850-hPa zonal wind, 200-hPa zonal wind, and satellite-observed outgoing longwave radiation (OLR) data. The time series coefficients of each of these two EOFs are called RMM1 and RMM2. Using the two-dimensional phase space defined by RMM1 and RMM2, Wheeler and Hendon (2004) obtained eight phases of the MJO and assigned each available day into one of these eight phases. An additional phase, called a weak phase, is also included for cases where magnitude of time coefficients were below unity (i.e., [RMM12 + RMM22]1/2 < 1). The originally defined RMM index is available only from 1974 when satellite OLR data first became available. To be consistent with the period of TC data considered in this study, the RMM index extending back to 1970 was obtained from Leroy and Wheeler (2008). The index for this extended period is based only on 850- and 200-hPa zonal wind fields as OLR data was unavailable.

The TC data were then binned into one of these eight MJO phases (plus a weak phase) whereby each cyclone from the JTWC database was assigned a phase that was operating during the time of its genesis. Since genesis can have different definitions, differences in the date arising because of the choice of genesis definition may result in a TC being assigned a different MJO phase (Hall et al. 2001). Therefore, in order to assess the magnitude of this problem, we have also performed the TC binning procedure using an alternate definition where genesis was taken to have occurred at a point when a TC has first attained an estimated 17.0 m s−1 wind speed. The differences between the results obtained using the two definitions were minor and so we have adopted to the cyclone binning procedure using the formal definition of genesis mentioned in section 2a above.

The geographical distribution of TC genesis positions in each MJO phase was objectively clustered using the kernel density estimation (KDE; Bowman and Azzalini 1997) approach. This approach was successfully used in Ramsay et al. (2008) for the Australian region and in CW09 for the Fiji region TC studies. KDE is a nonparametric technique whereby a density function or kernel can be used to construct a smooth probability density estimate of the observed data. The constructed density estimate can then be used as a statistical representation of a larger population distribution. In the present investigation, the KDEs are constructed to encapsulate the pattern of spatial variations in TC genesis positions for different MJO phases. Unlike a simple scatterplot, the constructed KDE contours give an objective measure of the density estimates of TC genesis positions. Here, the KDEs are displayed with contours that enclosed proportions of genesis positions corresponding to the median and to quartiles (i.e., the contour labeled 75 contained 75% of the total observations and similarly for contours labeled 50 and 25).

A statistical test, following Hall et al. (2001), was performed at 90% and 95% significance levels to examine the connection between the MJO and the distribution of TC numbers in each MJO phase (the null hypothesis being that TCs are uniformly distributed across all phases of the MJO, i.e., no modulation). The relevant test statistic, as computed separately for each MJO phase, is given by the following equation:

 
formula

where and pe are observed and expected daily genesis rates (DGR; defined as the ratio of TC number to the number of MJO days N in each phase), respectively, and z is the test statistic. The critical values of z at 90% and 95% levels, using two-tailed test, are ±1.64 and ±1.96, respectively. Questions, however, may arise on the validity of z test due to relatively few TCs forming in the FST region (i.e., <30 per phase). Therefore, an additional nonparametric bootstrap method (e.g., Camargo et al. 2007b) was performed to test the significance of TC modulations in each MJO phase. The results obtained by the two tests were very similar so we only show the statistics associated with z tests.

The effect of ENSO on the MJO–TC relationship was also determined. The TC data were first stratified into different phases of ENSO, namely El Niño (the warm phase), La Niña (the cold phase), and the neutral phase. Years corresponding to these phases of ENSO were obtained from the Climate Prediction Center (CPC 2009) Web site. Cyclones associated with each ENSO phase were then binned into the respective MJO phases using the cyclone binning procedure discussed above. The z test was also repeated to identify statistically MJO phases associated with enhanced or suppressed TC activity in respective ENSO conditions, except that the value in each MJO phase is now evaluated against the pe value for the associated ENSO condition (i.e., the modulation of TCs in the El Niño phase, for example, used pe only for the El Niño phase).

Finally, we have determined the influence of the MJO on different intensity categories of TCs in the FST region, namely, on the hurricane category and on the combined gale and storm category cyclones. This classification scheme (Table 1), which is based on the maximum 10-min sustained wind speed, is same as one proposed by Revell (1981) and adopted by Thompson et al. (1992), Holland (1984), and Sinclair (2002) for TC studies in the southwestern tropical Pacific region. There are, however, some reservations raised regarding the homogeneity of the intensity estimates that have evolved over time since the first satellite observation (e.g., Landsea et al. 2006; Harper et al. 2008). These relate to advances in satellite technology and improvements in the application of Dvorak analysis techniques. Nevertheless, the data from 1985 onward has been suggested to be the most reliable for intensity studies due to routine satellite observations (Harper et al. 2008). Accordingly, we have only used the past 20 years of TC data (i.e., from 1985 onward) for this part of the study. All TCs are first grouped into one of the two cyclone categories (i.e., a hurricane category or a gale plus storm category) and then they were allocated to the respective MJO phases using the aforementioned TC binning procedure. The statistical test for enhancement or suppression of TCs in each MJO phase of the two categories also uses the z statistic as previously discussed.

c. TC track clustering

The TC tracks associated with different phases of the MJO were examined here for related behaviors in their track types using the cyclone track clustering model developed by Gaffney (2004; available online at http://www.datalab.uci.edu/resources/CCT). Only a brief summary of the clustering technique is presented here. A complete description of its implementation over the FST region is given in CW09. Details on model specification and clustering methodology can be found in Gaffney (2004) and Camargo et al. (2007a).

Earlier studies on TC track clustering (e.g., Elsner and Liu 2003; Elsner 2003) relied on techniques like K means, where tracks of varying lengths could not be accommodated. This poses serious limitation when studying TC tracks (Camargo et al. 2007a). The technique used here, however, accommodates tracks of varying lengths and can objectively cluster TC tracks of different trajectory shapes and locations. This curve clustering method is based on a polynomial regression mixture model, which is an extension of the standard finite mixture model (e.g., Everitt and Hand 1981), to allow for the representation of appropriate regression equations from which TC tracks might have been generated. In this case, we used a second-order polynomial regression model (as opposed to other order polynomials and splines) as it provided the best trade-off for the observed TC tracks over the FST region in terms of goodness-of-fit and ease of interpretation.

Following CW09, we selected a set of three clusters for our study such that each TC track is assumed to be generated by one of these clusters. Applying cluster analysis separated all tracks into one of these three clusters and generated cluster-specific mean regression curves using a second-order polynomial regression function. Two distinct types of TC motions, namely, “straight” and “recurver” types were clearly identified from the mean regression curves. The typical tracks in clusters 1 and 2 were characteristic of straight-moving cyclones while those in cluster 3 corresponded to recurving cyclones. To quantify the relationship between TC tracks and the MJO, TC tracks in each of the three clusters were binned into different MJO phases. The number of TC tracks in all MJO phases and their respective clusters were subjectively evaluated and only those clusters that had a relatively higher number of TCs were retained for regression analysis using second-order polynomial regression.

d. Atmospheric data and compositing technique

The relationship between TC characteristics and large-scale environmental conditions are well established in the scientific literature. Gray (1968, 1975, 1979) identified several preexisting dynamical and thermodynamical large-scale environmental conditions necessary for TC formations. Subsequent studies (e.g., McBride and Zehr 1981) showed that while the thermodynamical conditions are usually satisfied for considerable portions of the tropical oceans for long periods of time, dynamical conditions can change significantly at much shorter spatial and temporal scales. The three most important dynamical parameters that are widely used in association with TC genesis studies in several basins are (i) low-level cyclonic relative vorticity, (ii) vertical wind shear, and (iii) upper-level divergence. A recent study by CW09 has also shown the importance of these parameters for cyclone genesis in the FST region for the different phases of ENSO. While these dynamical parameters are arguably more important to investigate short-term TC variability, Camargo et al. (2009) have recently shown the ability of some thermodynamical parameters (e.g., midlevel relative humidity and potential intensity) to modulate the MJO–TC relationships as well.

Consistent with CW09, our focus in this study is on low-level relative vorticity, upper-level divergence, and vertical wind shear as these dynamical parameters are known to strongly affect TC genesis patterns in the FST region (see CW09 for details). The composites of these parameters are constructed using the daily National Centers for Environmental Prediction–National Center for Atmospheric Research (NCEP–NCAR) reanalysis products (Kalnay et al. 1996). For ease of interpretation, only composites associated with phases 2 and 3 and 7 and 8 are constructed as these phases show significant suppression and enhancement of TC activity in the FST region (as shown later in section 3), respectively. The 850-hPa cyclonic relative vorticity and the 200-hPa divergence fields were determined using a centered-finite differencing scheme while the environmental vertical wind shear (EVWS) was determined using the following equation:

 
formula

where u and υ are zonal and meridional components of wind vectors, respectively. For completeness, the wind flow patterns at 850- and 200-hPa levels were also composited.

To investigate factors affecting TC tracks, we have examined the mean environmental flow patterns between the 700- and 500-hP levels for all TC clusters. While the other levels such as 500 hPa, 700 hPa, and a deep-mean layer between 850 and 200 hPa can also possibly be considered as steering flows of TC motion (e.g., Chan and Gray 1982; Aberson 2003), CW09 showed that mean 700–500-hPa level is the best representation of TC tracks in the FST region. Thus, the mean daily wind fields between 700- and 500-hPa levels were composited over the lifetime of each TC in their respective clusters.

Often it is argued that large-scale patterns evident in composites are a result of TCs themselves. To address this issue, composites were constructed for cases with and without TCs. Since the flow patterns investigated here are composites of wind vectors, the question of whether composited results are truly representative of individual cases can also arise. To ascertain the reliability of flow patterns associated with each composite, it was necessary to examine the steadiness of flow associated with each composite. This was done following Chan and Kwok (1999), who defined steadiness as the ratio of the magnitude of the mean wind vector to the average speed of the wind at the same grid point. A ratio close to unity (greater than 75% in our case) indicates that the flow in the composite is representative of most individual cases that were used to construct the composite.

3. Results

a. Climatological characteristics

To better understand the MJO–TC relationships in the FST region, it is important to first explore the climatological distribution of TCs and the large-scale environmental conditions in which they occur. Altogether 122 TCs from the 1970/71 to 2005/06 periods are observed in the FST region for all seasons and 115 TCs when considering only austral summer seasons. On rare occasions and possibly tied to ENSO effects (e.g., Ramage and Hori 1981), TC formations in the FST region can be observed as early as October and as late as June (Fig. 2a). The major season, however, starts from November and continues through April, accounting for about 94% of the total formations. The peak activity is in January, with January and February alone accounting for approximately 50% of the total formations.

The geographical distributions of TCs during November through April show a poleward progression of the mean seasonal genesis locations in tandem with the associated zero shear lines (Fig. 2b). The mean genesis locations, for example, during November and December seasons are equatorward of 12°S and are around 14°S during the peak seasons (i.e., January and February). The formations during March and April are, on average, poleward of 15°S. The changes in these mean seasonal genesis positions are consistent with the associated shifts in the zero shear lines, with the exception of the April season, where the mean genesis location is quite far south of the associated zero shear line. While most of the formations in the FST region occur between 170°E and the date line, those forming east of the date line (west of 170°E) are generally attributed to El Niño (La Niña) conditions (see CW09 for details on interannual distribution of TCs in the FST region).

The overall distribution of genesis positions shows a consistent relationship with the climatological conditions of dynamical large-scale environments (Fig. 3). Cyclonic 850-hPa relative vorticity (Fig. 3a) exists over almost the entire region with large negative values (i.e., <−4 × 10−6 s−1) noticed west of the date line where mean seasonal formations are more common. The mean divergence aloft (Fig. 3b) also favors TC formations in the FST region. Similarly, the low values of mean EVWS (i.e., <8 m s−1) are well collocated with the mean TC genesis positions in the FST region, particularly the climatological zero zonal shear line that lies between 170°E and the date line.

To summarize thus far, TC genesis positions tend to be concentrated in an area of climatological low-level cyclonic relative vorticity, divergent atmosphere aloft, and a weak vertical wind shear. It is therefore hypothesized that perturbations in these climatological environmental conditions due to MJO influences significantly contribute to the intraseasonal variability of TC activity in the FST region. Once favorable conditions are established, a number of mechanisms have been suggested to cause genesis. Maloney and Hartmann (2001), for example, suggested that growing eddy kinetic energy in the mean westerly flow provided seed disturbances for cyclone development. Rossby wave accumulation through convergence of low-level zonal flow is also suggested to amplify tropical disturbances into “tropical depression” type storms (Sobel and Maloney 2000). Determining these mechanisms is, however, not within the scope of the present investigation. Examination of large-scale environmental conditions associated with different MJO phases is arguably sufficient here to understand the MJO–TC relationships in the FST region.

b. The MJO–TC relationships

This section is divided into four parts. The first part looks at the relationship between the MJO and TC genesis while the second part examines the strength of this relationship in different ENSO conditions. In the third part, the influences of the MJO on different intensity category cyclones are examined. Finally, we identify clusters in TC track regimes associated with different phases of the MJO.

1) MJO and TC genesis

The TC genesis locations, together with the associated DGRs, for different MJO phases as defined by Wheeler and Hendon (2004) are shown in Fig. 4. The corresponding statistics are also shown in Table 2. Altogether there are 6525 MJO days observed in the FST region during the November through April seasons for 1970/71 to 2005/06 periods. Of these days, only 115 are associated with TC formations. Assuming a null hypothesis of no effect of the MJO on TC genesis (i.e., no modulation) in the FST region, the expected DGR would be 1.76% (i.e., pe = 1.76%). Examination of each MJO phase, however, reveals that there are cases where DGRs are statistically different from 1.76% (i.e., TC formations are modulated by the MJO). Of particular interest are phases 2 and 3 and phases 7 and 8 where the z statistics reveal suppression and enhancement of TC genesis, respectively.

The TC formations in phases 2 and 3 are four each with 523 and 586 MJO days, respectively. Their associated z values indicate statistical suppression of TC numbers at 90% and 95% significance levels. The TC formations during a weak phase of the MJO are also statistically suppressed at the 90% significance level. On the other hand, the TC formations for phases 7 and 8 are 26 and 18 with 609 and 554 MJO days, respectively. Consequently, the TC formations in phases 7 and 8 are statistically enhanced at the 95% significance level. These statistics on TC genesis modulation by the MJO yield an enhancement to suppression ratio of, on average, about 5 to 1 in the FST region. This indicates that TCs are 5 times more likely to form when the MJO is in its active phase over the FST region (i.e., phases 7 and 8) than when it is in its inactive phase (i.e., phases 2 and 3). The causality of this observed modulation will be established in relation to the large-scale environmental conditions later in the section. The DGRs in the other MJO phases are not significantly different from the expected DGR of 1.76%. Therefore, TC formations in these phases are not statistically modulated by the MJO.

A close examination of the KDE contours from phases 6–8 followed by phase 1 reveals an eastward propagation of TC genesis maxima. This is consistent with the concomitant propagation of the mean convective activity for phases 6–8 (and phase 1) over the FST region (see Fig. 8 of Wheeler and Hendon 2004). On average, the genesis maxima in phase 6 is around 170°E while those in phase 7 are around 170°E as well as near the date line extending eastward. Similarly, the distribution of TC genesis positions in phase 8 is mostly around the date line while it shifts farther eastward near 170°W in phase 1. The rates of cyclone genesis in the other MJO phases (including the weak phase) are relatively low and the spatial distributions of TCs are also not very well defined as they are widely spread over the entire domain, particularly between 170°E and the date line.

2) Effect of ENSO on the MJO–TC relationship

The TCs in the FST region are strongly affected by the ENSO phenomenon (Chand and Walsh 2009). In El Niño years, for example, more cyclones are observed to be forming near the date line as opposed to La Niña years where cyclones usually form near 170°E; there are also fewer cyclones forming in the FST region during La Niña years than during El Niño years. Moreover, the ENSO phenomenon is also known to alter MJO characteristics, for example, by enhancing MJO-related convective activity east of the date line during El Niño years (Hendon et al. 1999). It is therefore anticipated that the MJO–TC relationship is likely to be strengthened in El Niño years relative to that in the La Niña or neutral years. To test this hypothesis, DGRs in El Niño, La Niña, and neutral phases are examined separately.

There is indeed evidence of different amount of modulations of TC genesis by the MJO in El Niño, La Niña, and neutral years (Fig. 5). During El Niño years, DGRs in the MJO phases 6–8 are statistically greater (at the 95% significance level) than the expected El Niño value of pe = 2.36% (Fig. 5a), indicating that TC formations in these MJO phases are statistically enhanced. Two characteristic features that distinguish modulation of TC genesis during El Niño years from that in the all-year case of Fig. 4j are (i) no statistical suppression of TC genesis occurs during El Niño years (except for the weak phase) and (ii) the DGR in phase 6 is double the associated all-year case (Table 3). The DGR for phases 7 and 8 during El Niño years also shows an increase by a factor of about 1.6 from the respective all-year cases, though this increase is not statistically significant. Consistent with the results of Hall et al. (2001) for the Australian region study, our results also support the hypothesis that the MJO–TC relationship in the FST region is substantially strengthened when only El Niño conditions are included. In La Niña years, TC genesis is also significantly strengthened (at the 95% significance level) but only in the MJO phase 1 (Fig. 5b) when formation rate is also significantly greater than the associated all-year case by a factor of about 1.9. In neutral conditions, TC genesis is statistically enhanced in the MJO phase 7 at the 95% significance level and this behavior remains unchanged from the associated all-year case. Moreover, the total rates of cyclone genesis during El Niño (La Niña) years are significantly greater (less) than the climatological genesis rate of 1.76%. This is consistent with CW09 who showed a greater potential of TC formation in the FST region during El Niño years than the La Niña years.

In summary, the modulation of cyclone genesis in the FST region by the MJO is substantially different in different ENSO conditions. In El Niño years, for example, the rate of cyclone genesis is enhanced in the MJO phases 6–8, while in La Niña years the rate is enhanced in phase 1. In neutral conditions, the genesis rate is statistically enhanced in the MJO phase 7. Of particular interest here are the rates of cyclone genesis in the MJO phases 1 and 6, which are statistically different from their respective all-year cases and appear to be associated with La Niña and El Niño conditions, respectively.

3) MJO and TC intensity

Here we show the modulation of cyclones by the MJO in two separate intensity categories: (i) hurricane intensity and (ii) combined gale and storm intensity categories as defined earlier in section 2b. Figure 6 shows the DGRs of these two categories in different MJO phases. Two interesting features emanate from this figure. First is a clear indication of an increase in the DGRs of hurricane category cyclones with a concomitant decrease in the DGRs of gale and storm category cyclones as we move from phase 1 to 6. Second is the increase in DGRs for phases 7 and 8 in both the categories. The differences in these two features are possibly linked to the different degree of the MJO and ENSO influences in each of these phases. The enhanced convective activity associated with the MJO in, for example, phases 7 and 8 seems more likely to generate potential seed disturbances throughout the FST region than that in the other phases, resulting in increased TC genesis. As these cyclones are spread over the lower latitudes as well as over the higher latitudes, their potential to develop into hurricane category or gale and storm category is equally high. In phase 8, however, there is slightly more skewness toward the hurricane category than toward the gale and storm category cyclones, partially due to the fact that relatively more cyclones in phase 8 are formed equatorward than poleward (as per Fig. 4h). This enhances the probability of cyclones reaching hurricane intensity (e.g., Thompson et al. 1992; Camargo and Sobel 2005).

A similar argument can also be used to explain the concomitant increase and decrease in the DGRs of the hurricane and gale plus storm category cyclones from phases 1 to 6. Considering the all-year case, the DGRs in these phases are not statistically enhanced by the MJO (as per Fig. 4j) and they do not differ substantially from each other. This implies that the total genesis rate in phases 1–6 is approximately constant. As such, the MJO provides changes to the basic state that then allow a greater proportion of cyclones to develop into hurricanes in phases 4–6. To a lesser extent, the observed changes in the cyclone intensity from phases 1 to 6 can also be linked to the ENSO phenomenon. As determined earlier, DGRs in phase 1 (phase 6) are relatively more common in La Niña (El Niño) conditions. The cyclones forming in phase 1 are more poleward than those in phase 6 (see Figs. 4a and 4f), consistent with formation in La Niña and El Niño conditions, respectively, as determined by CW09. Cyclones in El Niño conditions are arguably more likely to reach hurricane intensity categories than those in the La Niña conditions because of more equatorward displacement of their formation regions (e.g., Thompson et al. 1992; Sinclair 2002; Camargo and Sobel 2005). As a result, cyclones forming in phase 1 reach only gale and storm category while the majority of those forming in phase 6 reach hurricane intensity. Warm SST anomalies are also found equatorward of about 20°S during El Niño (Sinclair 2002), further explaining the likely increased occurrence of hurricane category cyclones.

4) MJO and TC tracks

As mentioned earlier in section 2c, TC tracks in the FST region can be conveniently described in three separate clusters determined by CW09 with distinct patterns in different ENSO conditions. When binned into the associated clusters for different MJO phases (Fig. 7a), distinct TC track regimes are evident, predominantly in phase 1 and phases 6–8. In phase 1, for example, cyclone tracks are mainly associated with cluster 1 as those in clusters 2 and 3 are very few or none at all. Phase 1 cyclones that form near 170°E usually pass over Fiji into the Tonga region (Fig. 7b) while those forming in the maximal genesis region (as per Fig. 4a) have, however, very erratic tracks that could not be separated by cluster analysis. The TCs forming in phase 6 are also associated with cluster 1. Most of these cyclones form around 170°E (as per Fig. 4f) and track east-southeastward over the northern part of the Fiji islands into the Tonga region (Fig. 7c). This flow pattern is also a characteristic of cluster 1 during El Niño conditions (see Fig. 6a of CW09). The TCs in phase 7 are associated with clusters 1 and 2 whereby cluster 1 encompasses cyclones mainly forming west of the date line whereas cluster 2 encompasses cyclones forming east of the date line. The mean regression curve associated with cluster 1 passes over the southern parts of the Fiji islands while that associated with cluster 2 passes near the Samoa region (Fig. 7d). Relative to the other phases, phase 7 includes the maximum number of cyclones in both clusters 1 and 2. This is different from phase 8 where larger numbers of cyclones are associated with the “recurving” cluster 3 (Fig. 8e). Cluster 2 cyclones in phase 8, particularly those that form east of the date line, take a straight path over Samoa.

c. Large-scale environmental conditions

This section is divided into two parts. The first part looks at the influences of 850-hPa relative vorticity, 200-hPa divergence, and vertical wind shear on TC genesis in the FST region during enhanced (i.e., phases 7 and 8) and suppressed (i.e., phases 2 and 3) phases of the MJO. The second part looks at the mean 700–500-hPa steering flow patterns for each of the clusters associated with different phases of the MJO.

1) Association with TC genesis positions

(i) Relative vorticity, divergence, and wind shear

The large-scale environmental fields associated with TC genesis in the FST region show distinct spatial patterns for the enhanced and suppressed phases of the MJO (Fig. 8). Composites of 850-hPa wind vectors in the suppressed phase, for example, are predominantly easterly throughout the region, with small variations around 4°–12°S and west of the date line (Fig. 8a1). This shows only a weak cyclonic relative vorticity straddling the FST region, particularly west of the date line (grayscale in Fig. 8a1). On the other hand, the associated wind vectors in the enhanced phase are predominantly westerly in the region equatorward of 15°S extending east-southeastward up to about 170°W (Fig. 8a2). This westerly flow converges with predominantly northeasterly flows east of 170°W resulting in the net poleward flow. The steady easterly flow is also evident west of the date line and poleward of 18°S. This easterly flow, together with broad aforementioned westerly flow, gives rise to relatively large values of cyclonic relative vorticity that strongly favors TC genesis, particularly in the region around 10°–15°S, 170°E and 10°–15°S, 180° where the density of TC genesis is the highest (see Figs. 4g,h).

Similarly, the 200-hPa wind vectors and the associated divergence also show characteristically different spatial patterns for the suppressed and enhanced phases of the MJO (Figs. 8b1 and 8b2). The 200-hPa winds poleward of 16°S are consistently westerly in both the phases. In the enhanced phase, however, winds equatorward of 8°S become predominantly easterly as opposed to the suppressed phase where the flow patterns remain weak. This facilitates increased divergence in the entire FST region during the enhanced MJO phase (Fig. 8b2). The divergence in the suppressed phase is relatively weak (Fig. 8b1).

Another distinguishing large-scale environmental condition separating observed TC genesis patterns in the suppressed and enhanced MJO phases is the vertical wind shear (Figs. 8c1 and 8c2). A high wind shear environment is known to be one of the major reasons for the failure of development and weakening of TCs (e.g., Gray 1968; McBride and Zehr 1981). In this study, we examine the zonal wind shear of vertical winds as well as the EVWS. A characteristic difference between the wind shear composites of suppressed and enhanced phases of the MJO is the location of zero shear line. In the suppressed phase, the zero shear line and the associated low EVWS are located in the northwestern corner of the region (Fig. 8c1). In contrast, the zero shear line and the associated low EVWS for the enhanced phase are located farther poleward to about 12°S extending eastward in the region where TCs are frequently spawned (Fig. 8c2).

As stated earlier in section 2d, it may be argued that differences in the above large-scale patterns between suppressed and enhanced MJO phases could be the result of more TC genesis in the latter. To assess the magnitude of this problem, we have reexamined the composites after excluding all TC cases. The results (not shown) are substantially similar. Consistent with other investigations (e.g., Bessafi and Wheeler 2006; Camargo et al. 2009), this indicates that the extent to which TCs can influence the composite fields in the FST region is limited.

(ii) Genesis potential parameter

Here we examine the collective effect of 850-hPa relative vorticity, 200-hPa divergence, and vertical wind shear on the suppressed and enhanced phases of the MJO. These dynamical large-scale environmental fields are combined into a single genesis potential parameter (GPP) using the product concept of Gray (1975) such that

 
formula

where ζr is the relative vorticity (s−1), D is the divergence (s−1), and Vshear is the vertical wind shear of zonal winds between the 850- and 200-hPa levels (m s−1). The coefficients of ζr and D appearing in this equation are arbitrarily chosen to give the parameter value of order unity. The additive term and the power factor associated with Vshear limits the parameter value approaching infinity. Further work is, however, necessary to determine a constant multiplier that can give this parameter an appropriate magnitude and dimension so that it can reliably replicate annual cycle of TC genesis rate in the FST region (e.g., Emanuel and Nolan 2004).

Regardless, this basic parameter is able to reproduce observed large-scale patterns of TC genesis positions in the FST region. An example is shown for the cases of suppressed and enhanced phases of the MJO (Fig. 9). In the suppressed phase, there is relatively small potential for TC genesis in the FST region, particularly around 12°S, 172°E (Fig. 9a). In contrast, the potential for genesis remarkably increases in the enhanced phase of the MJO with the negative maxima in the Coral Sea region extending eastward into the Fiji and Samoa regions where formation is relatively more common.

(iii) Effect of individual fields on the GPP

In this section, we examine the separate contributions of relative vorticity, divergence, and wind shear fields to the GPP. For ease of interpretation, we limit our investigation only to the suppressed and enhanced phases of the MJO. A modified genesis potential parameter (denoted GPP modified) is constructed such that one field is allowed to vary while the others are set to their respective climatologies (e.g., Camargo et al. 2009). Assuming linearity in all modified composites (i.e., the composite of GPP is a sum of the three modified parameters), the degree of contribution by each field to the GPP can be determined qualitatively.

Comparing the composites of each modified genesis parameter with the GPP composite (Fig. 10), some distinctions on the degree of contributions by each field can be made. West of the date line, for example, all factors appear to be substantially contributing to the GPP. East of the date line, however, the dominating factor appears to be the vertical wind shear followed by approximately equal contributions from divergence and vorticity. Similar results (not shown) are obtained for the suppressed phase. These results indicate that all large-scale environmental fields (i.e., relative vorticity, divergence, and wind shear) considered in this study are substantially important for the MJO modulation of TC genesis in the FST region. However, we make no claim that these fields are optimal representation of the GPP for the FST region. There may be other fields that can additionally be used in association with relative vorticity, divergence, and wind shear to optimize the GPP (see, e.g., Emanuel and Nolan 2004). Regardless, the parameter (as mentioned above) is sufficient to elucidate differences in TC genesis structures between enhanced and suppressed phases of the MJO for the FST region.

2) Association with TC motions

The track patterns identified earlier in section 3b using cluster analysis for different phases of the MJO are explained here in terms of 700–500-hPa mean steering flow composited over both the lifetime of cyclones in respective clusters (Fig. 11) as well as over their first positions (not shown). The wind fields for first-position and the lifetime composites are very similar, thus indicating the role of the former in possible TC track forecasts. Only dominant clusters in phase 1 and phases 6–8 are considered as the numbers of TCs in these phases are relatively large and the flow patterns are well defined (see Fig. 7).

The initial flow pattern in phase 1 associated with TCs in cluster 1 is eastward and later becomes northwesterly (Fig. 11a). This steers most TCs forming in the Coral Sea region southeastward over Fiji and Tonga. The cyclones forming between 170°E and the date line in the MJO phase 6 are also associated with cluster 1. Here, the flow is also initially eastward and later becomes northwesterly, with a slight equatorward shift in the mean regime compared to phase 1 (Fig. 11b). This steers TCs over to the northern part of the Fiji islands and the Tonga region. The TCs forming in phase 7, which is the phase where the MJO has the strongest signal in the FST region (e.g., Wheeler and Hendon 2004), show two characteristic flow patterns associated with clusters 1 and 2. Those in cluster 1 are, on average, generated around 170°E and track southeastward by steady northwesterly steering winds into the southern part of the Fiji region (Fig. 11c). Those TCs in cluster 2 are usually generated around the date line and steered eastward into the Samoa region by mean westerly midtopospheric flows (Fig. 11d). Similarly, TCs in phase 8, which are mostly formed around the date line, have two distinct flow regimes associated with clusters 2 and 3. The TCs associated with cluster 2 usually track eastward over the Samoa region in the mean westerly steering winds (Fig. 11e) while those TCs associated with cluster 3 turn into the southern part of the Fiji islands associated with veering winds around 15°S, 170°E (Fig. 11f).

4. Discussion and summary

This paper presents a detailed study on the influences of the MJO on TC activity over the FST region. Investigations particularly focused on addressing variations in TC genesis rates, tracks, and intensity during different phases of the MJO (as defined by Wheeler and Hendon 2004) and on encapsulating the role of large-scale environmental conditions associated with such variations. Results indicate strong modulation of TC genesis, tracks, and intensity in the FST region by the MJO.

The daily genesis rates of TCs are strongly affected in the FST region. Statistics reveal that the genesis rate is suppressed in the MJO phases 2 and 3 (inactive MJO phases) at the 90% and 95% significance levels, respectively. Similarly, the genesis rate is enhanced in phases 7 and 8 (active MJO phases) at the 95% significance level yielding an enhancement to suppression ratio of about 5:1. This is consistent with the associated enhanced and suppressed convective activity over the FST region as determined by Wheeler and Hendon (2004). Because the FST region lies well within the South Pacific convergence zone (SPCZ; see Vincent 1994 for a review) where modulation of convective activity by the MJO is strong (e.g., Matthews et al. 1996; Zhang and Dong 2004), the observed TC enhancement to suppression ratio is also strong. The large-scale environmental conditions (i.e., low-level cyclonic relative vorticity, upper-level divergence, and weak vertical wind shear) substantially favor an enhanced TC genesis rate in the active phase of the MJO. These favorable large-scale environmental conditions are, however, lacking in the inactive phase of the MJO. The combined product of these environmental conditions into a single genesis parameter also has a substantial potential to discern TC genesis patterns observed in the enhanced and suppressed MJO phases. Moreover, a close examination of the TC genesis positions from phases 6 to 8 (and phase 1) reveals an eastward propagation of TC genesis maxima. This is consistent with the concomitant propagation of the mean convective activity from phases 6 to 8 followed by phase 1 over the FST region (see Fig. 8 of Wheeler and Hendon 2004). The modulation of the MJO–TC relationship is further strengthened during El Niño conditions, consistent with the results of Hall et al. (2001) for the Australian region.

Furthermore, the MJO also has a significant effect on the ratio of the hurricanes versus weaker cyclones in the FST region. The increase in the hurricane category and concomitant decrease in the gale and storm category cyclones for phases 1–6 are attributed to different degree of influences by ENSO. The TCs in the MJO phase 1, for example, are relatively more common in La Niña conditions as opposed to phase 6 where they are more common in El Niño conditions. As mentioned earlier, cyclones in El Niño conditions are arguably more likely to reach hurricane intensities than those in La Niña conditions because of more equatorward and eastward displacement of their formation regions (e.g., Thompson et al. 1992; Sinclair 2002; Camargo and Sobel 2005). As a result, cyclones forming in phase 1 reach only gale and storm intensity while the majority of those forming in phase 6 reach hurricane intensity. Warm SST anomalies are also found equatorward of about 20°S during El Niño (Sinclair 2002), further explaining the increased occurrence of hurricane category cyclones east of the date line. The increase in both hurricane category and gale plus storm category cyclones in phases 7 and 8 are a result of strong convective activity possibly seeding more potential disturbances throughout the FST region than that in the other phases, resulting in increased TC genesis. As these cyclones are spread over the lower latitudes as well as over the higher latitudes, their potential to develop into hurricane category or gale plus storm category is equally high.

The TC tracks in different phases of the MJO, particularly in phase 1 and phases 6–8, are also conveniently described by three clusters using objective cluster analysis technique. In phases 1, 6, and 7, for example, most TCs forming in the Coral Sea region and around 170°E are steered southeastward to over Fiji and Tonga by steady 700–500-hPa mean flow. In phase 7, most of the cyclones are also generated around and east of the date line. These cyclones are associated with cluster 2 and they usually track eastward in the mean 700–500-hPa flow over the Samoa region. Similarly, TCs forming east of the date line in phase 8 are associated with cluster 2 whereby they are steered eastward into the Samoa region by mean 700–500-hPa winds. Some TCs that are generated west of the date line in phase 8 are associated with a recurving cluster 3. These cyclones initially move southwestward and then recurve around 15°S, 170°E into the southern parts of the Fiji region by the recurving 700–500-hPa mean flow regimes.

To summarize, TC genesis, tracks, and intensity in the FST region are substantially modulated by the MJO, particularly during El Niño conditions. These modulations are caused by large-scale atmospheric variability such as low-level relative vorticity, upper-level divergence, and vertical wind shear. The modulation of TCs here is identifiable to an extent that the results may be incorporated into future work on the intraseasonal TC prediction schemes for the FST region.

Acknowledgments

The best-track tropical cyclone data were obtained from the Joint Typhoon Warning Center Web site at http://www.usno.navy.mil/NOOC/nmfc-ph/RSS/jtwc/best_tracks/. The NCEP–NCAR reanalysis products were also obtained online from the NOAA Web site (http://www.cdc.noaa.gov). The authors are thankful to Dr. Matthew Wheeler for providing the MJO indices for the period extending back to 1969 (obtained online from the Web site at http://www.bom.gov.au/bmrc/clfor/cfstaff/matw/maproom/RMM/RMMestimates.69toRealtime.txt). We also thank Professor David Karoly for his advice. We appreciate constructive comments by Professor Neville Nicholls and the other two anonymous reviewers. Finally, I acknowledge the Australian government–sponsored Endeavour Postgraduate Award under which I am doing my doctorate degree at the University of Melbourne.

REFERENCES

REFERENCES
Aberson
,
S. D.
,
2003
:
Targeted observations to improve operational tropical cyclone track forecast guidance.
Mon. Wea. Rev.
,
131
,
1613
1628
.
Barrett
,
B. S.
, and
L. M.
Leslie
,
2009
:
Links between tropical cyclone activity and Madden–Julian oscillation phase in the North Atlantic and northeast Pacific basins.
Mon. Wea. Rev.
,
137
,
727
744
.
Bessafi
,
M.
, and
M. C.
Wheeler
,
2006
:
Modulation of south Indian Ocean tropical cyclones by the Madden–Julian oscillation and convectively coupled equatorial waves.
Mon. Wea. Rev.
,
134
,
638
656
.
Bowman
,
A. W.
, and
A.
Azzalini
,
1997
:
Applied Smoothing Techniques for Data Analysis: The Kernel Approach with S-Plus Illustrations.
Oxford University Press, 193 pp
.
Camargo
,
S. J.
, and
A. H.
Sobel
,
2005
:
Western North Pacific tropical cyclone intensity and ENSO.
J. Climate
,
18
,
2996
3006
.
Camargo
,
S. J.
,
A. W.
Robertson
,
S. J.
Gaffney
,
P.
Smyth
, and
M.
Ghil
,
2007a
:
Cluster analysis of typhoon tracks. Part I: General properties.
J. Climate
,
20
,
3635
3653
.
Camargo
,
S. J.
,
A. W.
Robertson
,
S. J.
Gaffney
,
P.
Smyth
, and
M.
Ghil
,
2007b
:
Cluster analysis of typhoon tracks. Part II: Large-scale circulation and ENSO.
J. Climate
,
20
,
3654
3676
.
Camargo
,
S. J.
,
M. C.
Wheeler
, and
A. H.
Sobel
,
2009
:
Diagnosis of the MJO modulation of tropical cyclogenesis using an empirical index.
J. Atmos. Sci.
,
66
,
3061
3074
.
Camargo
,
S. J.
,
A. H.
Sobel
,
A. G.
Barnston
, and
P. J.
Klotzbach
,
2010
:
The influence of natural climate variability, and seasonal forecasts of tropical cyclone activity.
Global Perspectives on Tropical Cyclones, 2nd ed., J. C. L. Chan, Ed., World Scientific, in press
.
Chan
,
J. C. L.
, and
W. M.
Gray
,
1982
:
Tropical cyclone movement and surrounding flow relationships.
Mon. Wea. Rev.
,
110
,
1354
1374
.
Chan
,
J. C. L.
, and
R. H. F.
Kwok
,
1999
:
Tropical cyclone genesis in a global NWP model.
Mon. Wea. Rev.
,
127
,
611
624
.
Chand
,
S. S.
, and
K. J. E.
Walsh
,
2009
:
Tropical cyclone activity in the Fiji region: spatial patterns and relationship to large-scale circulation.
J. Climate
,
22
,
3877
3893
.
CPC
,
cited
.
2009
:
Cold and warm episodes by season: Changes to the oceanic Niño index.
.
Elsner
,
J. B.
,
2003
:
Tracking hurricanes.
Bull. Amer. Meteor. Soc.
,
84
,
353
356
.
Elsner
,
J. B.
, and
K. B.
Liu
,
2003
:
Examining the ENSO-typhoon hypothesis.
Climate Res.
,
25
,
43
54
.
Emanuel
,
K. A.
, and
D. S.
Nolan
,
2004
:
Tropical cyclone activity and global climate.
Preprints, 26th Conf. on Hurricanes and Tropical Meteorology, Miami, FL, Amer. Meteor. Soc., 240–241
.
Everitt
,
B. S.
, and
D. J.
Hand
,
1981
:
Finite Mixture Distributions.
Chapman and Hall, 143 pp
.
Gaffney
,
S. J.
,
2004
:
Probabilistic curve-aligned clustering and prediction with regression mixture models. Ph.D. thesis, University of California, Irvine, 281 pp. [Available online at http://www.ics.uci.edu/~sgaffney/papers/sgaffney_thesis.pdf]
.
Gray
,
W. M.
,
1968
:
Global view of the origin of tropical disturbances and storms.
Mon. Wea. Rev.
,
96
,
669
700
.
Gray
,
W. M.
,
1975
:
Tropical cyclone genesis.
Department of Atmospheric Sciences Paper 234, Colorado State University, Fort Collins, CO, 121 pp. [Available from Department of Atmospheric Sciences, Colorado State University, Ft. Collins, CO 80523]
.
Gray
,
W. M.
,
1979
:
Hurricanes: Their formation, structure, and likely role in the tropical circulation.
Meteorology over Tropical Oceans, D. B. Shaw, Ed., Royal Meteorological Society, 155–218
.
Hall
,
J. D.
,
A. J.
Matthews
, and
D. J.
Karoly
,
2001
:
The modulation of tropical cyclone activity in the Australian region by the Madden–Julian oscillation.
Mon. Wea. Rev.
,
129
,
2970
2982
.
Harper
,
B. A.
,
S. A.
Stroud
,
M.
McCormack
, and
S.
West
,
2008
:
A review of historical tropical cyclone intensity in northwestern Australia and implications for climate change trend analysis.
Aust. Meteor. Mag.
,
57
,
121
141
.
Hendon
,
H. H.
,
C.
Zhang
, and
J. D.
Glick
,
1999
:
Interannual variation of the Madden–Julian oscillation during austral summer.
J. Climate
,
12
,
2538
2550
.
Ho
,
C-H.
,
J-H.
Kim
,
J-H.
Jeong
,
H-S.
Kim
, and
D.
Chen
,
2006
:
Variation of tropical cyclone activity in the South Indian Ocean: El Niño–Southern Oscillation and Madden–Julian Oscillation effects.
J. Geophys. Res.
,
111
,
D22101
.
doi:10.1029/2006JD007289
.
Holland
,
G. J.
,
1984
:
On the climatology and structure of tropical cyclones in the Australian/southwest Pacific region: I. Data and tropical storms.
Aust. Meteor. Mag.
,
32
,
1
15
.
JTWC
,
cited
.
2009
:
JTWC tropical cyclone best track data site.
.
Kalnay
,
E.
, and
Coauthors
,
1996
:
The NCEP/NCAR 40-Year Reanalysis Project.
Bull. Amer. Meteor. Soc.
,
77
,
437
471
.
Landsea
,
C. W.
,
B. A.
Harper
,
K.
Hoarau
, and
J. A.
Knaff
,
2006
:
Can we detect trends in extreme tropical cyclones?
Science
,
313
,
452
454
.
Leroy
,
A.
, and
M. C.
Wheeler
,
2008
:
Statistical prediction of weekly tropical cyclone activity in the Southern Hemisphere.
Mon. Wea. Rev.
,
136
,
3637
3654
.
Liebmann
,
B.
,
H. H.
Hendon
, and
J. D.
Glick
,
1994
:
The relationship between tropical cyclones of the western Pacific and Indian Oceans and the Madden–Julian Oscillation.
J. Meteor. Soc. Japan
,
72
,
401
411
.
Madden
,
R. A.
, and
P. R.
Julian
,
1971
:
Detection of a 40–50-day oscillation in the zonal wind in the tropical Pacific.
J. Atmos. Sci.
,
28
,
702
708
.
Madden
,
R. A.
, and
P. R.
Julian
,
1994
:
Observations of the 40–50-day tropical oscillation— A review.
Mon. Wea. Rev.
,
122
,
814
837
.
Maloney
,
E. D.
, and
D. L.
Hartmann
,
2000a
:
Modulation of hurricane activity in the Gulf of Mexico by the Madden–Julian oscillation.
Science
,
287
,
2002
2004
.
Maloney
,
E. D.
, and
D. L.
Hartmann
,
2000b
:
Modulation of eastern North Pacific hurricanes by the Madden–Julian oscillation.
J. Climate
,
13
,
1451
1460
.
Maloney
,
E. D.
, and
D. L.
Hartmann
,
2001
:
The Madden–Julian oscillation, barotropic dynamics, and North Pacific tropical cyclone formation. Part I: Observations.
J. Atmos. Sci.
,
58
,
2545
2558
.
Matthews
,
A. J.
,
B. J.
Hoskins
, and
J. M.
Slingo
,
1996
:
Development of convection along the SPCZ within a Madden–Julian Oscillation.
Quart. J. Roy. Meteor. Soc.
,
122
,
669
688
.
McBride
,
J. L.
, and
R.
Zehr
,
1981
:
Observational analysis of tropical cyclone formation. Part II: Comparison of nondeveloping versus developing systems.
J. Atmos. Sci.
,
38
,
1132
1151
.
Ramage
,
C. S.
, and
A. M.
Hori
,
1981
:
Meteorological aspects of El Niño.
Mon. Wea. Rev.
,
109
,
1827
1835
.
Ramsay
,
H. A.
,
L. M.
Leslie
,
P. J.
Lamb
,
M. B.
Richman
, and
M.
Leplastrier
,
2008
:
Interannual variability of tropical cyclones in the Australian region: Role of large-scale environment.
J. Climate
,
21
,
1083
1103
.
Revell
,
C. G.
,
1981
:
Tropical cyclones in the southwest Pacific, Nov. 1969 to April 1979.
New Zealand Meteorological Service Misc. Pub. 170, 53 pp
.
Sinclair
,
M. R.
,
2002
:
Extratropical transition of southwest Pacific tropical cyclones. Part I: Climatology and mean structure changes.
Mon. Wea. Rev.
,
130
,
590
609
.
Sobel
,
A. H.
, and
E. D.
Maloney
,
2000
:
Effect of ENSO and MJO on the western North Pacific tropical cyclones.
Geophys. Res. Lett.
,
27
,
1739
1742
.
Thompson
,
C. S.
,
S.
Ready
, and
X.
Zheng
,
1992
:
Tropical cyclones in the southwest Pacific: November 1979 to May 1989.
New Zealand Meteorological Service, 35 pp
.
Trenberth
,
K. E.
,
1997
:
The definition of El Niño.
Bull. Amer. Meteor. Soc.
,
78
,
2771
2777
.
Vincent
,
D. G.
,
1994
:
The South Pacific convergence zone (SPCZ): A review.
Mon. Wea. Rev.
,
122
,
1949
1970
.
von Storch
,
H.
, and
A.
Smallegange
,
1991
:
The phase of the 30- to 60-day oscillation and the genesis of tropical cyclones in the western Pacific.
Max-Planck-Institut fur Meteorologie Rep. 66, 22 pp
.
Wheeler
,
M.
, and
H. H.
Hendon
,
2004
:
An all-season real-time multivariate MJO index: Development of an index for monitoring and prediction.
Mon. Wea. Rev.
,
132
,
1917
1932
.
Zhang
,
C.
,
2005
:
Madden–Julian Oscillation.
Rev. Geophys.
,
43
,
RG2003
.
doi:10.1029/2004RG000158
.
Zhang
,
C.
, and
M.
Dong
,
2004
:
Seasonality of the Madden–Julian oscillation.
J. Climate
,
17
,
3169
3180
.

Footnotes

Corresponding author address: Savin S. Chand, School of Earth Sciences, University of Melbourne, Melbourne, VIC 3010, Australia. Email: schand@pgrad.unimelb.edu.au