The east Pacific warm pool exhibits basic-state variability associated with the Madden–Julian oscillation (MJO) and Caribbean low-level jet (CLLJ), which affects the development of easterly waves (EWs). This study compares and contrasts composite changes in the background environment, eddy kinetic energy (EKE) budgets, and EW tracks during MJO and CLLJ events. While previous studies have shown that the MJO influences jet activity in the east Pacific, the influence of the MJO and CLLJ on the east Pacific and EWs is not synonymous. The CLLJ is a stronger modulator of the ITCZ than the MJO, while the MJO has a more expansive influence on the northeastern portion of the basin. Anomalous low-level westerly MJO and CLLJ periods are associated with favorable conditions for EW development paralleling the Central American coast, contrary to previous findings about the relationship of the CLLJ to EWs. Easterly MJO and CLLJ periods support enhanced ITCZ EW development, although the CLLJ is a greater modulator of EW tracks in this region, which is likely associated with stronger moisture and convection variations and their subsequent influence on the EKE budget. ITCZ EW growth during easterly MJO periods is more reliant on barotropic conversion than during strong CLLJ periods, when eddy available potential energy (EAPE)-to-EKE conversion associated with ITCZ convection is more important. Thus, the influence of these phenomena on east Pacific EWs should be considered distinct.
Easterly waves (EWs) are synoptic-scale disturbances that are important components of the weather and climate of the tropics. EWs are the primary precursor disturbances for tropical cyclones in the Atlantic and Pacific (Avila and Pasch 1992; Landsea 1993; Avila and Guiney 2000; Avila et al. 2003; Pasch et al. 2009; Russell et al. 2017) and are prominent sources of westward-propagating tropical variability in wavenumber–frequency space (Wheeler and Kiladis 1999; Kiladis et al. 2006). Northern Hemisphere EW activity maximizes in boreal summer and fall (Roundy and Frank 2004; Serra et al. 2008). While the origins of east Pacific EWs are still controversial, they are thought to originate either from reinvigorated African easterly waves (AEWs) that have propagated over the Central American isthmus (e.g., Frank 1970; Shapiro 1986; Avila and Pasch 1992; Rappaport and Mayfield 1992; Zehnder et al. 1999; Serra et al. 2008, 2010) or from local disturbances in favorable background conditions (Nitta and Takayabu 1985; Tai and Ogura 1987; Maloney and Hartmann 2001; Serra et al. 2010; Toma and Webster 2010a, b; Rydbeck and Maloney 2014; Rydbeck et al. 2017). The east Pacific warm pool, where these EWs can reintensify or locally form, has been shown to exhibit variability on multiple time scales as a result of both local and remote processes. Two important phenomena that affect the east Pacific warm pool are the Madden–Julian oscillation (MJO) and Caribbean low-level jet (CLLJ). In this study, we investigate the respective influence of the MJO and CLLJ on the east Pacific background state, and their distinct impacts on EW eddy kinetic energy (EKE) budgets and tracks.
The MJO (Madden and Julian 1994; Zhang 2005) modulates conditions in the east Pacific warm pool on intraseasonal time scales (Maloney and Esbensen 2003; Serra et al. 2014). Previous work has found that boreal summer east Pacific precipitation, low-level convergence, low-level zonal wind, low-level relative vorticity, and vertical wind shear vary with MJO phase, although Rydbeck et al. (2013) suggest that the east Pacific can exhibit similar intraseasonal variability in these quantities independent of the MJO (e.g., Maloney and Hartmann 2000; Maloney and Esbensen 2003, 2007; Aiyyer and Molinari 2008). Maloney and Hartmann (2000) found that during anomalous MJO low-level westerly periods, a significant enhancement of east Pacific tropical cyclogenesis occurs as a result of strong positive anomalies in low-level relative vorticity and near-zero vertical shear of the zonal wind in the genesis region. More energetic EWs during MJO westerly periods may also contribute to this modulation of cyclogenesis (Maloney and Hartmann 2001). Anomalous MJO low-level easterly periods are associated with suppressed convection as well as anomalous low-level divergence and negative low-level relative vorticity anomalies (e.g., Maloney and Hartmann 2000, 2001). By accounting for the phase of the MJO, Slade and Maloney (2013) demonstrated increased skill at predicting east Pacific tropical cyclogenesis at 2–3-week lead times in a statistical model.
East Pacific EW activity is notably modulated by the MJO. Maloney and Hartmann (2001) found that low-level EKE is enhanced during westerly MJO phases relative to easterly phases, and that enhanced barotropic conversion helps to explain the more vigorous EWs and hence modulation of east Pacific tropical cyclones by the MJO. Using case studies from August to September 1998, Aiyyer and Molinari (2008) showed that westerly, convective phases of the MJO are associated with enhanced barotropic energy conversions along the Central American coast, while easterly, nonconvective phases of the MJO have enhanced barotropic conversion along the east Pacific intertropical convergence zone (ITCZ). Similarly, Crosbie and Serra (2014) found that low-level EKE and barotropic conversion increase along the Central American coast during MJO westerly periods in regional WRF Model simulations, and they documented shifts in moisture anomalies as a function of MJO phase. Rydbeck and Maloney (2014) highlighted that barotropic conversion and the conversion from eddy available potential energy (EAPE) to EKE are the leading terms in the EKE budget during neutral and westerly intraseasonal events and that EAPE-to-EKE conversion, while still important to the energy budget, is reduced during easterly intraseasonal events. Rydbeck and Maloney (2014) also found that strong midlevel barotropic conversion occurs during westerly intraseasonal events along typical EW tracks, further indicating that westerly intraseasonal periods provide more favorable energy conditions for EW growth than easterly periods.
The CLLJ has also been shown to modify atmospheric variables that are important to EW growth and tropical cyclogenesis in the Intra-Americas Sea. Wang and Lee (2007) found that variations in the CLLJ were connected to larger-scale circulations over the Atlantic, specifically the North Atlantic subtropical high (NASH), and that pressure gradients that develop over the Caribbean Sea because of the strength and location of the NASH play an important role in enhancing or weakening the jet. Wang (2007) found that a strong easterly CLLJ in boreal summer is positively correlated with increased rainfall along the east Pacific ITCZ, but is negatively correlated with rainfall along the west coast of Central America, similar to the results of Cook and Vizy (2010). Further, the CLLJ and the mean summer easterly flow in the Caribbean have been cited as providing a favorable background state for east Pacific EWs. Using an EW tracking algorithm, Serra et al. (2010) found that strong easterly CLLJ periods are associated with a greater frequency of east Pacific EWs. Molinari et al. (1997) and Molinari and Vollaro (2000) found that the reversal of the meridional PV gradient associated with the CLLJ and its extension, the Papagayo jet, provide conditions favorable for EWs to intensify in the east Pacific. Serra et al. (2010) noted that barotropic conversion caused by the CLLJ supports EW intensification in this region, while Serra et al. (2008) and Serra et al. (2010) also found that EAPE-to-EKE conversion is important to east Pacific EWs. Zehnder (1991) used a barotropic shallow water model to show that mean easterly flow over an idealized Sierra Madre Range led to the initiation of westward-propagating Rossby waves.
The aforementioned CLLJ results, which suggest the importance of a strong easterly CLLJ to EW growth, appear to conflict with the findings presented above about the MJO; namely, that low-level westerly anomalies associated with the MJO provide more favorable background environments and enhance EW energy conversions relative to easterly periods. The MJO also produces variations in the CLLJ (Maloney and Esbensen 2007). However, the analysis below will show that the effects of the MJO and CLLJ on EWs and the east Pacific background state are not synonymous. This study will investigate low-level zonal wind anomaly periods of the MJO and CLLJ to help distinguish which anomalous low-level wind direction is favorable for east Pacific EW development, and where and why this enhancement occurs. An EW energetics analysis will show that westerly periods of the MJO and CLLJ have similar energy budget term structures and profiles that favor EW intensification in the northeastern portion of the basin, though the influence of MJO has a more expansive reach westward into the basin. However, these phenomena produce different effects during easterly periods along the east Pacific ITCZ that are consistent with strong CLLJ period EWs relying more on convection for their energetics than in easterly MJO periods.
This paper is structured as follows. Section 2 provides an overview of the data and methods used to composite periods based on MJO and CLLJ phases and of the EW tracking algorithm that is employed. Section 3 discusses the mean state of the east Pacific and how this background state is modulated by the MJO and CLLJ. Section 4 describes the vertically averaged EKE budget during MJO and CLLJ composite events, while section 5 looks at the vertical structure of the energy budget terms and regressed fields along the ITCZ. Section 6 discusses how EW tracks vary based on MJO and CLLJ composite phases. Section 7 provides a discussion of the results and conclusions.
2. Data and methods
This study uses data from the European Centre for Medium-Range Weather Forecasts interim reanalysis (ERA-Interim; Dee et al. 2011) over the years 1990–2010, and the months of May–October when EWs are most active. The ERA-Interim data employed are 6-hourly, cover 17 pressure levels from 1000 to 200 hPa in 50-hPa increments, and have a grid spacing of 1.5°. Further, the NOAA/National Climatic Data Center outgoing longwave radiation (OLR) daily dataset (Lee 2014) from 1990 to 2010 is used as a proxy for deep convection. In section 3, both ERA-Interim and OLR data are interpolated to have a 1° grid spacing and 6-h time step for the composite analysis, similar to what is done in Rydbeck and Maloney (2014).
To investigate behavior during MJO and CLLJ events, compositing is used. MJO phases are defined using the real-time multivariate MJO (RMM; http://www.bom.gov.au/climate/mjo/) index based on Wheeler and Hendon (2004). The daily RMM index uses combined EOF analysis on equatorial-averaged 850- and 200-hPa zonal wind and OLR to define eight MJO phases. Only days with RMM amplitude greater than one standard deviation are considered in our analysis. To distinguish different sign wind events, MJO phase 2 will represent westerly events, while MJO phase 6 will represent easterly events. For the CLLJ and its extension to the Papagayo jet, we define an index based on the ERA-Interim 925-hPa zonal wind over the domain 9°–13°N, 86°–89°W for the months of May–October, similar in method to Wang (2007) and Serra et al. (2010). The CLLJ index is defined by first removing the 6-hourly climatological mean, then taking the 5-day running average of zonal wind averaged over this box. We define strong (weak) jet events as one standard deviation below (above) zero, where strong CLLJ events are more easterly and weak CLLJ events are more westerly compared to the mean easterly CLLJ, respectively. Only CLLJ events in which the MJO RMM index is within plus and minus one standard deviation are considered to limit the impact of the MJO on the CLLJ index in the analysis, since Maloney and Esbensen (2007) showed that jet events often accompany strong MJO events. In fact, roughly 33% of strong MJO westerly and easterly periods identified by the RMM index had significant jet activity that was not included in the CLLJ composite periods, and a sensitivity analysis investigating MJO periods with these jet periods removed yielded qualitatively similar results to the full MJO composite periods. By designing the CLLJ analysis this way, CLLJ periods are defined to have little to no influence by the MJO, which is a method that is unique compared to other studies involving the CLLJ in the literature. In total, 1472 MJO phase 2, 1028 MJO phase 6, 837 weak CLLJ, and 918 strong CLLJ 6-hourly observations occurred over the May–October, 1990–2010 period.
Before computing the EKE budget for EWs, the effects of tropical cyclone (TC) anomalies were removed at all pressure levels using the National Hurricane Center best-track data (Landsea and Franklin 2013) for TC locations over the time domain. After the cyclone center was identified, the weighting function and process described in Aiyyer et al. (2012) and Rydbeck and Maloney (2014) was used to remove fields around the cyclone center. The weighting function is given by
where r = r(x, y) is the distance of a grid point from the TC center and R = 500 km is the length scale of the weighting function. The weighting function is applied to a 6° × 6° box centered on the location of a TC at all time steps and over all pressure levels of data. By removing TCs, the EKE budget will more accurately display the energy conversions for EWs.
In section 6, we investigate how May–October EW and TC tracks vary during phases of the MJO and CLLJ. To do this, we utilize the newly created NOAA/National Centers for Environmental Information African easterly wave climatology dataset at 600 hPa (Belanger et al. 2014), which provides track information about individual easterly wave events and TCs. The EW track data are used to illustrate the general spatial characteristics of EW activity and to determine whether EW activity is favored in certain regions of the east Pacific during particular MJO and CLLJ periods, as opposed to investigating individual EWs that may be modulated by these phenomena. The 6-hourly track data used are derived from ERA-Interim data and an EW tracking algorithm described in Belanger et al. (2016) that uses curvature vorticity anomalies, among other variables, to identify and track EWs and TCs. To ensure that only typical, more robust EWs are included, we mandate that EWs must persist for more than 2 days and travel at least 15° in Pythagorean distance, at least 12° to the west in longitude, and no more than 4° to the south in latitude relative to their genesis location. EW track information is presented using track density, which is calculated by binning EW track observations every 2° and scaled to produce units of the number of waves per May–October season, similar to work done by Serra et al. (2010) and others. Finally, a centered 8° × 8° two-dimensional Gaussian weighting box with σ = 1.4° is applied over the track density plots for smoothing.
3. MJO and CLLJ influence on the east Pacific background state
To indicate where east Pacific EWs and TCs tend to occur, Fig. 1 shows average EW track density over May–October of 1990–2010, along with line contours of mean boreal summer OLR. Track density is in units of EW (4°2)−1 yr−1 and represents the number of EW and TC observations that occur in a given grid box per May–October season. The mean track density shown has similar features to results shown by Thorncroft and Hodges (2001), Serra et al. (2010), and Belanger et al. (2016). In the eastern Atlantic, track density values are mostly around 7–8 EW (4°2)−1 yr−1 near the exit of the African easterly jet, and slightly lower values continue westward across the North Atlantic hurricane main development region. Further, a relative minimum in track density occurs in the western Caribbean Sea as waves weaken after traversing the Atlantic. In the east Pacific, EWs and eventually TCs travel in a more southeast–northwest-oriented path that is parallel to the Central American coast, with values over 8.5 EW (4°2)−1 yr−1. This area of high track density will henceforth be referred to as the main EW path. The transition from low track density in the western Atlantic to high track density in the east Pacific is consistent with previous studies on EW tracks (Thorncroft and Hodges 2001; Serra et al. 2010; Belanger et al. 2016) and supports the notion that while some waves may propagate from the Atlantic into the Pacific, the local conditions and energetics in the east Pacific warm pool may be sufficient enough for EW generation (e.g., Maloney and Hartmann 2001; Serra et al. 2010; Rydbeck and Maloney 2014). Track density is also fairly high along the east Pacific ITCZ, with the location of the ITCZ being associated with the lower mean OLR values in the basin that extend from 7° to 10°N and from 80° to 140°W. This ITCZ EW activity is potentially related to preexisting AEWs and Panama Bight convective disturbances having more of a zonal track in the basin, or from EWs that form locally in the ITCZ as a result of processes like ITCZ breakdown (Ferreira and Schubert 1997) or inertial instability in the ITCZ that results from cross-equatorial pressure gradients (Toma and Webster 2010a).
Having discussed locations of east Pacific EW activity, it is important to investigate how the MJO and CLLJ affect the environmental conditions of this basin. Figure 2 shows the monthly frequency of occurrence for MJO and independent CLLJ low-level westerly and easterly events. MJO phase 2 events in our sample have a modest preference for the months of June–September, with August having a frequency of around 0.2. MJO phase 6 events in our sample have a modest preference for the months of May and October, with May having a frequency of about 0.25. Significant jet events independent from the MJO appear to favor the later months of boreal summer and into fall. Interestingly, although Chelton et al. (2000) notes that jet activity is particularly strong in boreal winter, the bottom panels of Fig. 2 indicate that substantial jet activity also occurs in summer months. For weak-jet periods, the months of August–October all have frequencies over 0.2, while May and July values are below 0.1. For strong-jet periods, October has the highest frequency of around 0.25, while June is the second highest month of occurrence. Although Wang (2007) notes that the CLLJ has a boreal summer maximum in July, our index finds that July is not the most prominent month with strong CLLJ activity. This difference may be due to the emphasis on synoptic time-scale activity and the removal of high-amplitude MJO events in our CLLJ index.
To analyze the influence of the MJO and CLLJ on the east Pacific warm pool large-scale environment, composite anomalies are computed relative to the 6-hourly climatological mean for all fields. Figures 3 and 4 show anomalous total column water vapor and 850-hPa relative vorticity during westerly and easterly periods of the MJO and CLLJ. For the MJO and CLLJ, westerly low-level wind anomaly periods are associated with enhanced moisture and positive relative vorticity anomalies along the main EW path shown in Fig. 1, while negative moisture and vorticity anomalies occur along the ITCZ. Easterly low-level wind anomaly periods for both phenomena are associated with negative moisture and relative vorticity anomalies along the main EW path, and have positive moisture and relative vorticity anomalies along the ITCZ. Although general similarities exist between these MJO and CLLJ composites, there are also some notable differences in how they modulate the east Pacific background state. For example, a greater modulation of moisture occurs along the ITCZ for the CLLJ relative to the MJO, particularly during easterly periods. Strong CLLJ periods have moisture anomalies that are around 1 kg m−2 above those found in MJO phase 6. In particular, CLLJ moisture anomalies are much higher along the ITCZ east of 105°W relative to MJO easterly periods. The increase in ITCZ moisture anomalies during strong CLLJ periods may provide a more favorable environment for ITCZ EWs relative to MJO phase 6. Further, it appears that the MJO has a more spatially extensive influence along the coast from Nicaragua to the Baja California Peninsula relative to the CLLJ, while the absolute value of the anomalies during both CLLJ periods are higher than in MJO periods. MJO moisture and vorticity anomalies have more of a northward and westward extent over the main EW path than those for the CLLJ, with CLLJ moisture anomalies being more concentrated in the immediate vicinity of the Central American coast. The strong influence along the entirety of the main EW path by the MJO provides additional evidence of its role as a major modulator of east Pacific EWs and TCs.
Figures 5 and 6 show mean May–October OLR and anomalous composite OLR to highlight the MJO’s more expansive influence along the main EW path and the stronger modulation of ITCZ convection by the CLLJ. The time-mean contours indicate that during boreal summer, deep convection is prominent in the Panama Bight region, with values below 200 W m−2. This OLR minimum then extends westward between 7° and 10°N, again highlighting the mean position of the east Pacific ITCZ. The composite OLR anomalies indicate that strong (weak) CLLJ periods are associated with suppression (enhancement) of convection near the Central American coast, but these anomalies are more localized relative to the MJO periods, similar to what was seen for the moisture anomalies in Figs. 3 and 4. Negative (positive) OLR anomalies occur more extensively along the main EW path during MJO phase 2 (phase 6). CLLJ periods are associated with a greater modulation of convection along and to the south of the ITCZ with a sign opposite to that near the Central American coast. During strong CLLJ periods, negative OLR anomalies at ITCZ latitudes begin around 90°W, while MJO phase 6 anomalies only start to become negative around 110°W, underscoring that the influence of these easterly periods on the east Pacific is not synonymous. This result is consistent with Figs. 3 and 4, which show higher moisture anomalies occurring along the ITCZ during strong CLLJ periods relative to MJO phase 6. Westerly period composite OLR anomalies also emphasize the greater influence the CLLJ has on ITCZ convection. Weak CLLJ period positive OLR anomalies also start around 90°W while positive MJO anomalies only begin west of 105°W. Though differences in westerly MJO and CLLJ period anomalies are notable, differences in the modulation of moisture and convection are most strongly apparent for easterly periods along the ITCZ, and the differences between easterly MJO and CLLJ periods will be explored further in subsequent sections. Overall, these findings indicate that while the MJO has a more extensive spatial influence along the main EW path, the modulation of moisture and deep convection along and to the south of the ITCZ is stronger and more extensive from the CLLJ. The enhancement of ITCZ moisture and convection during strong CLLJ periods may be associated with stronger moisture convergence because of the CLLJ, as mentioned by Wang (2007). However, understanding the differing responses of the east Pacific to the MJO and CLLJ is a topic for future research.
4. Eddy kinetic energy budgets
To support the notion that both MJO and CLLJ westerly low-level wind anomaly periods are more favorable for EW growth in the northeastern portion of the east Pacific basin, and to show that EWs during easterly periods of the MJO and CLLJ are influenced differently along the ITCZ, an EKE budget (vertically averaged from 1000 to 200 hPa) is computed. As in Rydbeck and Maloney (2014), EKE is defined as
where a bar represents the 11-day running mean, and a prime represents a deviation from the 11-day running mean. The EKE budget describes the components of the EKE tendency equation, which is given by
where v is the three-dimensional wind vector, is the two-dimensional wind vector, is the geopotential, R is the gas constant for dry air, p is the pressure, ω is the vertical pressure velocity, T is the temperature, and D is the budget residual. The first and second terms on the right-hand side of the EKE tendency equation, and , account for the advection of EKE by the time-mean and perturbation flow, respectively. The third term, , accounts for barotropic conversion, or the conversion of mean-state kinetic energy to EKE. The fourth term, , accounts for the generation of EKE by geopotential flux convergence. The fifth term,, accounts for the conversion from EAPE to EKE. Serra et al. (2008) documented that a maximum in EAPE-to-EKE conversion occurs between 400 and 200 hPa in the east Pacific, suggesting that EAPE is generated by convection in this region. EAPE is converted to EKE where ascent occurs in areas of warm temperature anomalies and descent occurs in areas of cold anomalies. The final term D is the budget residual and accounts for any uncalculated sources or sinks of EKE, such as friction caused by turbulent mixing or smaller-scale processes, and also contains any budget residuals caused by analysis increments or errors that arise from imprecise calculations.
Figures 7 and 8 show the contrast between EKE during westerly and easterly low-level wind anomaly periods for the MJO and CLLJ, respectively, vertically averaged from 1000 to 200 hPa. A strong meridional gradient in EKE exists because of increasing synoptic-scale wind perturbations with latitude, likely associated with midlatitude systems. MJO phase 2 is characterized by a local maximum of EKE in the east Pacific around 12°N, 105°W, with values reaching 11.5 m2 s−2, and a local minimum in the Panama Bight with values as low as 6.5 m2 s−2. For MJO phase 2, the area where the local maximum occurs is enhanced by around 2 m2 s−2 relative to phase 6 and is a statistically significant difference in EKE at the 90% confidence level using a bootstrapping resampling method by generating over 2000 random samples with replacement. Further, MJO phase 6 EKE values are also low in the Panama Bight, and lower values extend westward to the south of the ITCZ. Given the relatively weak modulation of precursor disturbances in the Bight of Panama region, the expansive modulation of the east Pacific background state shown in Figs. 3 and 5 by the MJO is most likely responsible for stronger EWs occurring along the main EW path during MJO phase 2 relative to phase 6.
Figure 8 shows that weak CLLJ periods resemble MJO phase 2 in that an extension of higher EKE values occurs southward; however, the local maximum is observed farther east than for MJO westerly periods, near the Central American coast and the Gulf of Tehuantepec. Values in this area are on the order of 11 m2 s−2, which are enhanced relative to the mean-state EKE (not shown). However, during strong CLLJ periods, the total EKE in the east Pacific is reduced both relative to weak CLLJ periods and the easterly MJO phase 6. This result highlights that MJO and CLLJ westerly periods are associated with stronger EWs relative to their respective easterly periods and seemingly goes against the notion that a strong CLLJ is associated with strong EW development. In particular, near the Central American coast weak CLLJ period EKE enhancements up to 2 m2 s−2 occur relative to strong CLLJ periods, which are statistically significant at the 90% level after bootstrapping. However, this enhancement of EKE for weak CLLJ periods is not as extensive along the main EW path, or as large in magnitude relative to the MJO periods shown in Fig. 7. It is also noted that at the latitude of the ITCZ, the reduction in EKE for strong CLLJ periods relative to weak CLLJ periods is relatively modest and not significant, consistent with a possible role for convection in maintaining EW activity in these regions as argued in the context of Figs. 4 and 6 above. This feature is not seen for the MJO in Fig. 7. In addition, the EKE during weak CLLJ periods is significantly enhanced to the west of 120°W in the equatorial region south of the ITCZ. This signal is also not present in the MJO composites. The mean-state equatorial 200-hPa zonal wind west of 120°W during weak CLLJ periods is characterized by total westerly winds that are also westerly in an anomalous sense relative to strong CLLJ periods (not shown). Thus, a potential explanation for this increase in equatorial EKE could be cross-equatorial wave propagation through westerly ducts, as described in Webster and Holton (1982). However, further testing of this hypothesis is left for future work.
Interestingly, EKE values in the Panama Bight show only modest differences between strong and weak CLLJ periods, similar to the MJO composite results shown in Fig. 7. During MJO phase 2, EKE in the Panama Bight is actually lowest when EKE in the east Pacific warm pool is highest. One implication of these results is that variability in the strength of east Pacific EW activity may be more dependent on variations in the large-scale downstream environment than the initial strength of diurnal convective disturbances that can grow upscale into EWs (e.g., Rydbeck et al. 2017).
The leading terms of the EKE budget (barotropic conversion and EAPE-to-EKE conversion) for the anomalous westerly and easterly composites of the MJO and CLLJ are provided in Figs. 9 and 10, respectively. Barotropic conversion in MJO phase 2 has a tendency to orient parallel to the Central American coast and has a local maximum of 3 × 10−5 m2 s−3 that is roughly in the same location as the EKE maximum seen in Fig. 7. In comparison to the weak CLLJ periods in Fig. 10, the barotropic conversion signature during MJO phase 2 extends more along the main EW path, although the values are roughly on the same order of magnitude around 2 × 10−5 m2 s−3. Consistent with the results in Figs. 4, 6, and 8, the area of higher EW barotropic conversion is shifted closer to the Central American coast for weak CLLJ periods relative to MJO phase 2. MJO phase 6 barotropic conversion has more of a zonal orientation that is shifted southward toward the ITCZ and has lower values along the main EW path compared to MJO phase 2, except for the exit region of the Papagayo jet. A southward shift in barotropic conversion toward the ITCZ also occurs during strong CLLJ periods and an enhancement of this term also extends westward from the Papagayo jet, which lines up well with the enhanced relative vorticity anomalies in this region for easterly periods in Figs. 3 and 4. These results indicate that topographic effects caused by an enhancement of the basic-state easterly flow through the Central American mountain gaps may play a role in providing a more favorable environment to initiate or reinvigorate EWs downstream during easterly periods, similar to what was found in Zehnder (1991), Molinari et al. (1997), and Molinari and Vollaro (2000).
For the EAPE-to-EKE conversion term, MJO and CLLJ westerly periods are more comparable to each other than for the easterly period composites. For example, an enhancement of this term, which is consistent with convective invigoration, extends from the Central American coast into the main EW path during both low-level westerly periods. EAPE-to-EKE conversion values over 4 × 10−5 m2 s−3 occur in this region during westerly periods. In addition, both periods have a relative minimum in EAPE-to-EKE conversion over the Costa Rica Dome region, where relatively cool sea surface temperatures help disfavor convection climatologically (Xie et al. 2005). Both westerly period EAPE-to-EKE conversion composites are consistent with the anomalies shown in Figs. 3–6, namely, that enhanced convection and moisture along the main EW path and near the Central American coast can help produce EAPE, which then is converted to EKE. Both easterly period composites have higher EAPE-to-EKE conversion along 10°N and the ITCZ. The magnitude of the EAPE-to-EKE conversion term values for strong CLLJ periods are stronger than for the MJO. Strong CLLJ periods also have a broader area of enhanced EAPE-to-EKE conversion along the ITCZ, which lines up with the negative OLR anomalies seen in Fig. 6. The more widespread and stronger EAPE-to-EKE values along the ITCZ during strong CLLJ periods than for easterly MJO periods suggests that this term plays more of a role in EW development relative to MJO phase 6, given their similar structures and magnitudes in barotropic conversion. Interestingly, all composite phases of the MJO and CLLJ have high EAPE-to-EKE conversion in the Panama Bight region, suggesting convective invigoration there. This finding indicates that regardless of the modulation to the east Pacific background state by the MJO or CLLJ, deep convective disturbances in the Panama Bight will continue to initiate and provide seed disturbances that are favorable for downstream EW development.
To provide a sense of the relative importance of the leading EKE budget terms in the composite periods, Figs. 11 and 12 provide the ratio of vertically averaged barotropic conversion to EAPE-to-EKE conversion in areas where the sum of the absolute values of both of the terms in Figs. 9 and 10 is at least 1.5 × 10−5 m2 s−3. Ratio values greater (less) than 1 indicate higher (lower) barotropic conversion values relative to EAPE-to-EKE conversion. Overall, westerly MJO and weak CLLJ periods appear to have many similarities, while their respective easterly periods have different responses. For MJO phase 2 and weak CLLJ periods, most of the main EW path is associated with a greater proportion of EKE generation stemming from EAPE-to-EKE conversion. Given the increased moisture and negative OLR anomalies east of 105°W in Figs. 3–6, we would expect convective processes to play an important role in EWs along the main track during both westerly periods. The ratios observed for weak CLLJ periods are lower than those during MJO phase 2 and are the lowest overall, indicating that EW growth during weak CLLJ periods is most dependent on convective generation of all the composite periods. However, when comparing both easterly periods, the ratios for EW energetics show greater differences along the ITCZ. In phase 6, the ITCZ is associated with higher barotropic conversion to EAPE-to-EKE conversion ratios than during MJO phase 2. Hence, it appears that convection is relatively less important for EW growth during easterly MJO periods, which agrees with the positive OLR anomalies found around the ITCZ in Fig. 5. This result is also consistent with Rydbeck and Maloney (2015), who found that weak convective coupling occurs in EWs during easterly intraseasonal events, and may suggest that EWs during easterly MJO periods act more like dry dynamical waves, growing as a result of interactions with the mean flow. Conversely, the strong CLLJ period ratios seem to indicate more of a contribution from convective invigoration to EW growth along the ITCZ relative to MJO phase 6. Along the ITCZ, statistically significant differences in the ratios occur between strong CLLJ periods and MJO phase 6 at the 90% confidence level using bootstrapping (not shown). When calculating the average of the ratio fields meeting the magnitude threshold for plotting in Figs. 11 and 12 over a box encompassing the mean ITCZ (6°–13°N, 88.5°–120°W), the average ratio is 1.27 for strong-jet periods to 1.52 for MJO phase 6. If the strength threshold used for plotting is relaxed, strong CLLJ periods have an average ratio in the ITCZ box of 1.68, while for MJO phase 6 the ratio is 5.11. Thus, strong CLLJ periods have more of a reliance on convection for EKE growth, reflected in the conversion from EAPE to EKE, than during MJO phase 6 periods. This greater dependence on convection is consistent with the stronger modulation in moisture along the ITCZ during strong CLLJ periods shown in Fig. 4, and from the more extensive and stronger negative OLR anomalies along and south of the ITCZ for strong CLLJ periods versus the positive OLR anomalies east of 110°W in the ITCZ associated with MJO phase 6 (Figs. 5 and 6).
5. ITCZ vertical structure
Given the differences in the east Pacific basic-state modulation and vertically averaged EKE budget terms near the ITCZ for the MJO and CLLJ, this section discusses the vertical structure of EKE budget terms along the ITCZ to diagnose their contributions to the growth of EWs during the MJO and CLLJ events. To do this, vertical cross sections along the ITCZ at 10.5°N from 85.5° to 135°W are computed.
Figure 13 shows the composite vertical cross sections of EKE from east (right) to west (left) along the ITCZ. Overall, a distinct vertical gradient in EKE exists associated with the stronger wind perturbations at upper levels. MJO phase 2 has a secondary maximum (with values over 11.5 m2 s−2) of EKE between 600 and 400 hPa near 10.5°N, 102°W and higher values that extend down through the column to the surface. Additionally, weak CLLJ periods also have an extension of higher EKE at midlevels, though the values are not as strong. The midlevel extensions are consistent with regions of increased EKE during westerly periods in the main EW track shown in Figs. 7 and 8. Although westerly periods for both the MJO and CLLJ exhibit similar behavior, easterly periods have more notable differences in their EKE structure relative to one another. For MJO phase 6, very little structure exists at midlevels. However, strong CLLJ periods have an enhancement of midlevel EKE relative to MJO phase 6 along the ITCZ on the order of 1 m2 s−2. The enhancement of midlevel EKE during strong CLLJ periods may be consistent with the notion that convective invigoration helps support EWs more during strong CLLJ periods relative to MJO phase 6, as suggested in Figs. 9–12.
Having shown the changes to the EKE vertical structure, we will now discuss composite changes in barotropic conversion and EAPE-to-EKE conversion in a vertical cross section along the ITCZ (Figs. 14 and 15). The structure of the energy generation fields along the ITCZ expands upon previous work done by Rydbeck and Maloney (2014) for a cross section along the main EW path, and shows the importance of lower-level barotropic conversion and upper-level EAPE-to-EKE conversion for ITCZ EWs east of 120°W. This analysis also expands upon the findings of Serra et al. (2010), who showed in meridional cross sections along 95°W that strong EAPE-to-EKE conversion occurs at upper levels and enhanced barotropic conversion occurs at lower levels near 10°N.
Consistent with previous results in this study, MJO and CLLJ westerly periods have similar vertical cross sections of their leading budget terms along the ITCZ, while easterly periods have notable differences. For example, Fig. 14 shows that MJO phase 2 is associated with strong midlevel barotropic conversion near 102°W, with values on the order of 4 × 10−5 m2 s−3, and that high values at low levels exist throughout the cross section. Similarly, barotropic conversion during weak CLLJ periods shown in Fig. 15 is enhanced near 400 hPa and has its highest values below 900 hPa. Both westerly MJO and weak CLLJ periods have strong EAPE-to-EKE conversion at upper levels to the east of 115°W, with values over 1.4 × 10−4 m2 s−3. Thus, even though lower- and midlevel barotropic conversion is important to developing EWs, the values of EAPE-to-EKE conversion at upper levels are large enough to keep the ratios shown in Figs. 11 and 12 along the ITCZ during both westerly composites below one to the east of 115°W. So, it appears that these areas may favor the generation of EKE by convective invigoration relative to barotropic conversion.
However, easterly MJO and strong CLLJ composite cross sections of the leading budget terms show differences important to the energetics of ITCZ EWs. While strong CLLJ periods have a similar, though weaker, midlevel extension of barotropic conversion up to 400 hPa, which is consistent with the results of Serra et al. (2010), MJO phase 6 periods have only weak midlevel barotropic conversion. Strong CLLJ periods have stronger and somewhat deeper EAPE-to-EKE conversion at upper levels relative to MJO phase 6 periods, and this feature is located above the enhanced midlevel barotropic conversion feature, similar to what occurs during the westerly composites.
A one-tailed difference-of-means Student’s t test at the 90% confidence level was conducted to examine the statistical significance of this EAPE-to-EKE conversion enhancement during strong CLLJ periods relative to MJO phase 6 in Fig. 14. The sample sizes of the composite periods used in the significance test are the number of individual events spanning at least two consecutive days over the composites—45 strong CLLJ and 52 MJO phase 6 events were used, respectively. Indeed, the enhanced EAPE-to-EKE conversion feature for strong CLLJ periods is significantly different at the 90% confidence level than during MJO phase 6 (Fig. 14), shown by the hatching in the bottom-right panel of Fig. 15. This more prominent area of EAPE-to-EKE conversion during strong CLLJ periods supports the results of Figs. 11 and 12, namely, that strong CLLJ period EWs along the ITCZ rely more on convective processes relative to MJO phase 6. This result is also consistent with the greater modulation of ITCZ moisture and convection during strong CLLJ periods relative to MJO phase 6 seen in Figs. 4 and 6. Thus, while cross sections of the westerly MJO and CLLJ composite budget terms are comparable in structure and magnitude, the easterly MJO and CLLJ composites are not synonymous and support the idea that strong CLLJ period waves depend more on convective invigoration than MJO phase 6 waves along the ITCZ.
To further explain why strong CLLJ period EWs are invigorated more by convection relative to MJO phase 6 EWs, linear regressions of bandpass-filtered apparent heat source Q1, ω, and temperature onto bandpass-filtered 700-hPa vorticity were calculated for easterly MJO and CLLJ periods. A positive covariance between anomalies of Q1 and temperature leads to EAPE generation, which subsequently can be converted to EKE through a negative covariance of ω and temperature as shown in Figs. 14 and 15. The method behind this analysis is described here. The expression for Q1 is calculated as in Yanai and Johnson (1993) and is given by:
where s = cpT + gz is the dry static energy, υ is the horizontal wind velocity, p is the pressure, and ω is the vertical pressure velocity. In this expression, the bars represent an average over a grid cell. For the linear regression, 700-hPa relative vorticity filtered from 2.5 to 12 days in frequency space and to westward wavenumbers 6–30 is used as the reference time series at a base point of 10.5°N, 111°W for investigating ITCZ EWs during the respective composite periods. Anomalous Q1, ω, and temperature on EW time scales are computed at all pressure levels by the same bandpass-filtering approach as for the 700-hPa relative vorticity. These anomalous fields are linearly regressed on to the filtered-vorticity reference time series during easterly and westerly periods of the CLLJ and MJO to yield regression coefficients. These regression coefficients are then scaled by one standard deviation of the respective composite-period-filtered vorticity time series (to represent the passage of a typical EW) for plotting.
Figure 16 shows ITCZ vertical cross sections (averaged between 10.5° and 12°N) of regressed anomalous temperature and Q1 (top row) and temperature and ω (bottom row) for MJO phase 6 and strong CLLJ composite periods. The temperature profiles for EWs in both periods, and particularly in strong CLLJ periods, agree with Serra et al. (2008), who found that east Pacific EWs have an eastward tilt of warm temperature anomalies with height. Further, EWs in both periods have anomalous apparent heating that maximizes in the middle to upper troposphere, suggesting that both deep convection and stratiform structures are present in these waves, which is consistent with the observational results of Petersen et al. (2003). Strong CLLJ period EWs appear to have stronger apparent heating and temperature anomalies and smaller wavelengths compared to MJO phase 6 waves, and appear to more efficiently produce EAPE because of a stronger covariability of temperature and Q1. For example, out in front of the wave there is anomalous cooling collocated with negative temperature anomalies that are more vertically coherent relative to MJO phase 6. Similarly, anomalous heating and positive temperature anomalies are more collocated behind the wave axis in strong CLLJ period waves. The analysis of regressed ω and temperature reveals a consistent story in which a better collocation of opposite-signed and stronger ω and temperature anomalies occurs for strong CLLJ period waves relative to MJO phase 6 waves, particularly in front of the wave axis. The enhanced negative covariance between ω and temperature indicates that strong CLLJ period waves seem to more efficiently convert the generated EAPE to EKE, supporting the results of Figs. 14 and 15. Therefore, strong CLLJ period EWs rely more on convective invigoration for their energetics relative to MJO phase 6. A potential explanation for the more coherent dynamical structures along the ITCZ that lead to more efficient EAPE-to-EKE conversion associated with convection during strong CLLJ periods could be the stronger moisture gradients along and to the north of the ITCZ relative to MJO phase 6, inferred from Figs. 3 and 4. The enhanced moisture gradients during strong CLLJ periods may cause stronger moisture advection anomalies in ITCZ EWs relative to MJO phase 6, leading to a stronger locking of convection to the waves. An examination of the moisture budget during these composite periods will be a topic for future work.
6. Easterly wave tracking
To provide a measure of the number and location of EWs to supplement the strength metric of EKE applied in sections 4 and 5, this section uses the recently released NOAA/National Centers for Environmental Information African easterly wave climatology dataset at 600 hPa to look at the influence of the MJO and CLLJ on EW and TC tracks. As a reminder, the May–October 1990–2010 track density is given by Fig. 1. As previously stated, the units for track density in Fig. 1 are EW (4°2)−1 yr−1 and are created from binning all track observations at a resolution of 2° × 2°.
To discern the impact of the MJO and CLLJ on EW tracks, the difference in composite EW track density is shown in Fig. 17, where the difference is calculated as the westerly period minus the easterly period. Track density in these plots represents the difference in the total number of track observations per grid box per year over each respective composite period [EW (4°2)−1 yr−1]. To do this, the number of waves per grid box over each composite period is calculated, similar to Fig. 1. Next, this value is divided by the number of days that goes into the composite period, and then is scaled by the length of the active season, May–October (184 days), to represent a year. Along the main EW path in MJO phase 2, there are areas with up to 10 additional track observations per grid box per year compared to phase 6. More tracks occur along 6°–8°N during MJO phase 6 relative to phase 2. This result is likely related to the southward shift of the vertically averaged barotropic and EAPE-to-EKE conversion terms toward the ITCZ shown in Fig. 9.
For the CLLJ composites, a strong modulation of tracks between wind periods occurs along the ITCZ. Strong CLLJ periods have higher EW observations along and north of the ITCZ relative to weak CLLJ periods, consistent with the OLR anomalies shown in Fig. 6, but along the main EW path not as clear of a distinction exists between wind phases as was observed with the MJO. An analysis of MJO and CLLJ composite periods against the climatology (not shown) reveals that, indeed, MJO phase 2 (phase 6) is associated with higher (lower) track density relative to the climatology along the main EW path, while weak (strong) CLLJ period track density is close to (slightly above) climatology in this region. For the easterly composites, strong CLLJ periods and MJO phase 6 are associated with higher ITCZ track density relative to climatology, though strong CLLJ periods have broader and generally stronger areas of higher track counts along the ITCZ relative to MJO phase 6. Thus, the MJO modulates the number of tracks along the main EW path to a much greater degree than the CLLJ, while both the CLLJ and MJO favor more EWs occurring along the ITCZ during easterly events, although the modulation by the CLLJ appears to be stronger. The stronger invigoration by convection that occurs during strong CLLJ periods relative to easterly MJO periods may support the higher ITCZ EW track density.
7. Discussion and conclusions
The background state of the east Pacific is modulated on a variety of time scales by events such as the MJO and CLLJ. These phenomena are associated with changes in the low-level wind structure, low-level vorticity, moisture, and convection in the basin, which are important to EW development and TC genesis. Previous studies have found that low-level westerly events associated with the MJO or intraseasonal events were linked to enhanced EW activity (Maloney and Hartmann 2001), tropical cyclogenesis (Maloney and Hartmann 2000; Aiyyer and Molinari 2008), favorable EW energetics (Crosbie and Serra 2014; Rydbeck and Maloney 2014), and stronger convective coupling in EWs (Rydbeck and Maloney 2015). Contrastingly, other studies indicate that strong easterly flow related to the CLLJ is associated with a greater frequency of EWs (Serra et al. 2010), east Pacific EW intensification (Molinari et al. 1997; Molinari and Vollaro 2000), and a possible source of EWs resulting from interactions of the flow with Central American topography (e.g., Zehnder 1991). This study presents a composite analysis and an EKE budget for MJO and CLLJ low-level wind phases to discern whether westerly or easterly low-level wind anomalies associated with these events provide more favorable conditions for east Pacific EW development. Our results show that MJO and CLLJ low-level wind periods provide differing anomalous responses in the east Pacific background state. The CLLJ is a stronger modulator of moisture and convection along the ITCZ, with strong CLLJ periods having enhanced moisture and reduced OLR anomalies relative to MJO phase 6. Strong CLLJ periods are also associated with enhanced ITCZ EW track density relative to MJO phase 6. MJO and CLLJ westerly periods are associated with enhanced convection, moisture, and low-level vorticity in the northeastern portion of the basin, with MJO anomalies extending farther westward across the main EW path.
EKE budgets for the composite periods reveal that westerly MJO and weak CLLJ periods are associated with significantly higher vertically averaged EKE along the main EW track as compared to their respective easterly phases. Along the ITCZ, however, the difference in total EKE between weak and strong CLLJ periods is not statistically significant, which differs from the corresponding MJO result, potentially indicating the importance of convection in supporting strong CLLJ period EW activity. Easterly periods for both phenomena are associated with enhanced barotropic and EAPE-to-EKE conversion along the ITCZ. However, easterly MJO and strong CLLJ period ITCZ EWs have different dependencies on their sources of EKE, with strong CLLJ period EWs relying more on EAPE-to-EKE conversion because of convective generation for their energetics. This result is consistent with the nonsignificant decrease in strong CLLJ period ITCZ EKE relative to weak CLLJ periods and the enhancement in ITCZ EW track density relative to MJO phase 6. Further, the greater reliance on EAPE-to-EKE conversion by strong CLLJ period waves is likely related to the relatively enhanced moisture and convection anomalies in this period seen in Figs. 4 and 6. Westerly MJO and weak CLLJ periods are associated with enhanced barotropic and EAPE-to-EKE conversion along the main EW path relative to easterly MJO and strong CLLJ periods (with the MJO having a more expansive westward reach along the main EW path). This result seems to contrast with previous CLLJ studies by suggesting that weak CLLJ periods, as opposed to strong CLLJ periods, may be associated with more favorable conditions for strong EW development in the northeastern portion of the basin. Regardless of the composite low-level wind period, a maximum in EAPE-to-EKE conversion occurs in the Panama Bight, which suggests that favorable energy conversions in this region are consistently occurring to support local disturbances that can develop and propagate into the east Pacific warm pool and seed downstream EW growth.
Vertical cross sections along the ITCZ reveal that easterly MJO and CLLJ periods have differing vertical structures of EKE budget terms, which also suggest the greater relative importance of convective processes to strong CLLJ period EWs. Strong CLLJ periods have enhanced midlevel barotropic conversion and stronger upper-level EAPE-to-EKE conversion from 114° to 99°W relative to MJO phase 6. Linear regressions of Q1, ω, and temperature anomalies onto filtered low-level vorticity also suggest that strong CLLJ period EWs appear to be able to more efficiently generate EAPE and then convert it to EKE relative to MJO phase 6 period waves. This result is consistent with the enhanced convective activity and column moisture anomalies found along the ITCZ in Figs. 4 and 6 for strong CLLJ periods.
EW track data are employed to show how the MJO and CLLJ alter the count and spatial characteristics of EWs. Easterly MJO and strong CLLJ periods are characterized by increased track density along the ITCZ. However, the CLLJ appears to be a stronger modulator of EW tracks in this area, likely associated with the greater modulation of moisture and convection anomalies along the ITCZ relative to the MJO. Along the main EW path, the CLLJ does not modulate EW track density to the same degree as the MJO.
The CLLJ index developed for our analysis is uniquely defined such that significant strong and weak CLLJ periods have limited influence by the MJO, since Maloney and Esbensen (2007) found that the MJO can also cause variations in the CLLJ. This study highlights the importance of convective invigoration to strong CLLJ period ITCZ EWs relative to MJO phase 6 waves, but more work must be done to understand why this occurs. We hypothesize that the greater modulation of moisture and convection along the ITCZ during strong CLLJ periods leads to enhanced ITCZ moisture gradients, which causes stronger moisture advection and convective coupling to the waves. We plan to conduct a detailed moisture budget analysis for MJO and CLLJ periods to address this hypothesis. In addition, we plan to use regional modeling in subsequent analyses to support mechanistic understanding of how and why MJO and CLLJ events have distinct influences on the east Pacific warm pool, EKE budget, and EW tracks and to further explore the role convection has in the development and maintenance of EWs in this region. These experiments may include alternately constraining the basic state and Bight of Panama convective disturbances to test their respective influences on east Pacific EW formation. Future work may also explore how the MJO and CLLJ affect east Pacific EWs in a future climate and whether the relationships observed in this study still hold. Overall, these findings suggest that subsequent studies involving the east Pacific background state and EWs must consider the distinct influences the MJO and CLLJ have on the region.
We thank Anantha Aiyyer and two anonymous reviewers for their helpful comments on this manuscript. This work was supported by the Climate and Large-Scale Dynamics Program of the National Science Foundation under Grants AGS-1347738 and AGS-1735978. The statements, findings, conclusions, and recommendations do not necessarily reflect the views of NSF.