Abstract

Kelvin waves in the Pacific Ocean occasionally develop and propagate eastward together with anomalies of deep convection and low-level westerly wind. This pattern suggests coupling between the oceanic waves and atmospheric convection. A simple composite analysis based on observed coupled events from October through April demonstrates that this apparent coupled mode is associated with significant large anomalies in the global flow that extend to high latitudes. These high-latitude anomalies are significantly larger than those that are linearly associated with the El Niño–Southern Oscillation (ENSO), and they evolve on time scales between those of the Madden–Julian oscillation and ENSO, potentially providing an opportunity for enhanced subseasonal predictability in the flow of the global atmosphere.

1. Introduction

Moist deep convection organized on large spatial scales in the tropics exerts profound influence on the flow throughout the global atmosphere (Ferranti et al. 1990; Bladé and Hartmann 1995; Jin and Hoskins 1995). Rossby wave trains originating within the tropics or at higher latitudes might subsequently influence the temporal and spatial evolution of the tropical convection (Matthews and Kiladis 1999; Hoskins and Yang 2000; Matthews et al. 2004; Ray et al. 2009). The Madden–Julian oscillation (MJO; Madden and Julian 1994; Zhang 2005) evolves together with weather patterns around the globe (e.g., Hendon and Salby 1994; Weickmann and Sardeshmukh 1994; Mo and Higgins 1998; Mo 2000; Nogues-Paegle et al. 2000; Matthews et al. 2004). These patterns include dramatic transitions in extratropical weather that tend to be handled poorly by numerical weather prediction models days to weeks in advance (Hendon et al. 2000; Higgins et al. 2000; Jones 2000; Jones and Schemm 2000; Jones et al. 2004a,b). Deep convection associated with the El Niño–Southern Oscillation (ENSO) has also been linked to well-defined seasonal response patterns throughout the global atmosphere (e.g., Wallace and Gutzler 1981; Brankovic et al. 1994; Hoerling and Kumar 2002).

Mean global circulation anomaly patterns associated with the MJO and ENSO have been applied to generate intraseasonal and seasonal weather forecasts. Such forecasts are complicated by active nonlinear interactions between the two modes. Roundy et al. (2010) recently demonstrated that the structure of the global flow anomaly associated with the MJO depends on ENSO, such that when both the MJO and ENSO are simultaneously active, global patterns tend not to evolve in the same way as indicated by the linear sum of patterns associated with the MJO and ENSO considered separately.

Not only does ENSO modify the global flow associated with the MJO, but also the tropical signals of both modes evolve differently when they act in combination. ENSO alters the structure of sea surface temperature (SST) anomalies around the globe, thereby allowing them to modulate variations in atmospheric convection acting on all time scales. Since MJO convection responds to SST anomalies (Hendon et al. 2000; Inness et al. 2003; Fu and Wang 2004; Pohl and Matthews 2007), ENSO must influence the development and propagation of the MJO. The MJO might in turn influence ENSO by modulating fluxes of momentum, radiation, and sensible and latent heat across the air–sea interface (e.g., Hendon et al. 1998; Waliser et al. 2003).

The low-level equatorial westerly phase of the MJO has been associated with development of downwelling oceanic Kelvin waves (Hendon et al. 1998; Zhang and Gottschalck 2002), which systematically force the thermocline downward to the east and advect warm water eastward (Harrison and Schopf 1984; Federov and Melville 2000; Roundy and Kiladis 2006). As the western and central equatorial Pacific warms prior to adjustment toward El Niño, the zonal fetch of surface westerly wind anomalies associated with the MJO extends farther across the Pacific, allowing the MJO to more efficiently amplify these Kelvin waves (e.g., Hendon et al. 1998). Further active convection associated with the MJO also cools the western equatorial Pacific, allowing SST to become more zonally uniform as El Niño matures (Bergman et al. 2001).

Modulation of the MJO by El Niño is not limited to the translation of the geographical extent of its active convection eastward. ENSO modulates the global amplitude and frequency of the MJO (Pohl and Matthews 2007). ENSO also modifies the structure of the wind field associated with the MJO (e.g., Seiki et al. 2009). Roundy and Kravitz (2009) showed that the variation of MJO structure with ENSO and the seasonal cycle is associated with a quasi-systematic interplay between zonal wind anomalies and oceanic Kelvin waves. They demonstrated that Kelvin waves arriving into the east Pacific tend to intersect with local westerly surface wind anomalies there during adjustment toward El Niño, and that the waves tend to intersect with easterly wind anomalies during adjustment away from El Niño. These patterns would amplify the waves and associated SST signals during adjustment toward El Niño and weaken them during adjustment away from El Niño.

On some occasions when the low-level westerly wind phase of the MJO moves over the equatorial west Pacific, the anomalies of convection and winds amplify, reduce in zonal extent, and propagate eastward together with the developing oceanic Kelvin wave at roughly 2 m s−1 (Roundy and Kiladis 2006). This pattern suggests coupling between the wave and atmospheric convection. Although this apparent coupled mode frequently develops in association with the MJO during particular phases of ENSO, the coupling process constitutes an intraseasonal mode that can be distinguished from both the MJO and ENSO (Lau and Shen 1988; Roundy and Kiladis 2006). Such apparently coupled events occur throughout the calendar year. Anomalies of atmospheric convection that evolve along with these waves might associate with a preferred progression of the mid- and high-latitude flow resulting from overturning and Rossby wave dispersion.

Recently, Cordeira and Bosart (2010) analyzed the large-scale patterns that preconditioned the atmosphere for development of the so-called “perfect storms” of the Northern Hemisphere in fall 1991. They noted that convection in the deep tropics associated with the MJO and one of these oceanic Kelvin waves apparently dramatically modified the high-latitude atmospheric flow over the North Pacific, setting the stage for storm development. The purpose of this work is to generalize the associations of the apparent coupled Kelvin mode with the global flow for comparison with global response patterns to ENSO and the MJO. Although we found substantial anomalies associated with these waves during both northern and southern winters, for brevity we focus our analysis in this paper on the Northern Hemisphere cool season.

2. Data

Interpolated outgoing longwave radiation (OLR; Liebmann and Smith 1996) and National Centers for Environmental Prediction (NCEP)–National Center for Atmospheric Research (NCAR) reanalysis (Kalnay et al. 1996) geopotential height, streamfunction, and wind data were obtained from the Earth System Research Laboratory (ESRL). Anomalies were generated by subtracting the local mean and the seasonal cycle (along with its first three harmonics) estimated for the period 1974–2008. Dynamic height data (a proxy for sea level height) were obtained from the Tropical Atmosphere Ocean (TAO) array of buoys moored in the tropical Pacific. The seasonal cycle was subtracted by a least squares fit to its primary and first three harmonics, and the resulting anomalies were filtered for periods of 20–120 days (also by a least squares fit). Missing data were reconstructed by regression relationships to sea level gauge data from the University of Hawaii Sea Level Center, following Roundy and Kiladis (2007). This reconstruction extends the Kelvin wave database back to January 1974.

3. Identification of events

Kelvin waves evolving with atmospheric convection in a manner consistent with the coupled mode were identified by an objective windowed correlation analysis between the Kelvin wave–filtered dynamic height on the equator and unfiltered OLR anomalies averaged from 5°N to 5°S. One time series segment was selected from the OLR anomalies at the OLR grid point closest to a given TAO buoy on the equator. Another time series segment was selected for the same period of time from the dynamic height data at that buoy. The Pearson correlation coefficient was calculated for the two segments and its value placed in a new time series on the date in the middle of the time segment. The process was then repeated for new consecutive pairs of time segments throughout the dataset. An 81-day centered, sliding correlation window was chosen (11 days longer than the average period of the Kelvin wave in the east Pacific). Results are not sensitive to small changes in the length of this window. By definition, the events of interest have positive dynamic height anomalies moving eastward together with negative OLR anomalies, so events were considered for inclusion only when a negative correlation was present, and when the OLR anomaly was negative and locally less than −1 standard deviation (SD). Events were retained only if the correlation pattern was present continuously in time moving eastward across at least three TAO buoys between 140°E and 155°W (the buoys are 10°–15° of longitude apart). The algorithm identified 40 events across the full seasonal cycle during 1974–2006. The set of events analyzed in this project was then limited to the 19 that occurred between October and April (listed in Table 1) to focus on the associated patterns of global flow during the Northern Hemisphere cold season. The set of events were indexed by the dates on which the associated Kelvin-band dynamic height anomalies were maximized at the date line. Monthly mean Niño-3.4 SSTs and SST anomalies are included in Table 1 for reference. The SST on the equator and the date line is also reported when available. The date-line data are useful for reference to Roundy and Kiladis (2006), who argued based on observations and the Lau and Shen (1988) advective mode that coupled events occur when the local SST anomalies in the central Pacific exceed 28°C. One listed event during 1985 indicated SSTs below that level, but the corresponding OLR anomalies indicate that the coupling of the associated wave to convection occurred west of the date line, where SSTs were higher (not shown). Most of these events occur during or just prior to El Niño, during brief warm periods in the central Pacific or during portions of the seasonal cycle when the central Pacific is warmer than average, such as the Northern Hemisphere spring. Although 19 might seem like a small number of events, it is large in comparison with the number of events that are possible, given the background states that apparently support their development (Roundy and Kiladis 2006). During the process of selecting these events, we examined the full climatology of equatorial OLR anomalies for patterns of active convection moving eastward at roughly 2 m s−1 through the western and central Pacific, characterized by zonal scales similar to those observed during the coupled events. We did not find any such convective events that were not associated with downwelling Kelvin waves, suggesting that these selected events are not simply random coincidences of convection and Kelvin waves.

Table 1.

List of date-line crossing dates of the oceanic Kelvin waves analyzed, along with the corresponding Niño-3.4 SST. Temperatures labeled OI are taken from the National Oceanic and Atmospheric Administration (NOAA) optimum interpolation dataset. Temperatures labeled TAO are measured from the Tropical Atmosphere Ocean array of moored buoys. Values labeled N/C indicate that the authors lacked sufficient confidence in available SST data at the date line for the dates noted because of insufficient observations near the date line.

List of date-line crossing dates of the oceanic Kelvin waves analyzed, along with the corresponding Niño-3.4 SST. Temperatures labeled OI are taken from the National Oceanic and Atmospheric Administration (NOAA) optimum interpolation dataset. Temperatures labeled TAO are measured from the Tropical Atmosphere Ocean array of moored buoys. Values labeled N/C indicate that the authors lacked sufficient confidence in available SST data at the date line for the dates noted because of insufficient observations near the date line.
List of date-line crossing dates of the oceanic Kelvin waves analyzed, along with the corresponding Niño-3.4 SST. Temperatures labeled OI are taken from the National Oceanic and Atmospheric Administration (NOAA) optimum interpolation dataset. Temperatures labeled TAO are measured from the Tropical Atmosphere Ocean array of moored buoys. Values labeled N/C indicate that the authors lacked sufficient confidence in available SST data at the date line for the dates noted because of insufficient observations near the date line.

4. Compositing approach

Composite averages of OLR, dynamic height, geopotential height, and wind data were calculated by averaging over the set of dates listed in Table 1, as well as time lags from those dates. To smooth out the impacts of higher-frequency noise, all dates from 5 days before to 5 days after the listed dates were also included in the averages. Composites of the corresponding low-frequency background states were found by calculating the averages again, including all dates from 60 days before to 60 days after the listed dates. Results are insensitive to small changes in the lengths of these time windows (the 11-day window for the first composite is less than 16% of the 70-day leading period of the Kelvin waves, while the 120-day smoothing is a similar fraction of the length of an ENSO cycle). Statistical significance was assessed by means of 1000-member bootstrap tests (e.g., Wilks 2006), assuming that the number of degrees of freedom is equal to the number of Kelvin wave events included (rather than the number of days, following Roundy et al. 2010). Resampling was accomplished by randomly drawing from the set events (with replacement) and finding the mean of the new sample. Repeating the process yielded a distribution of means at each grid point. Tests were then applied to diagnose differences from zero and differences from the composite mean background state (to diagnose whether the signal timed with the Kelvin waves is significantly different from the ENSO signal). Another set of tests was applied to diagnose statistical differences between the composite signal associated with these Kelvin waves and that of a mean MJO signal during the same months and phases of ENSO. However, this attempt proved useless because the signals in the MJO wavenumber–frequency band reaching the date line during those months are dominated by the coupled Kelvin wave events. A similar composite analysis was applied for the same set of events by Cordeira and Bosart (2010), but over a smaller range of time lags and for a base longitude farther west (to emphasize flow features that developed early in the life cycle of the waves). They focused their discussion on the impacts of these waves on the development of the perfect storms of late 1991. We extend the composite across a broader range of the globe and across a broader range of time lags, with the center date of the composite at the date-line crossing time, to help the composite better resolve the associated flow patterns through a greater portion of the life cycle of the composite wave.

5. Results

a. Composite pattern in the tropics

Figure 1a shows the composite OLR anomaly (shading, blue suggests enhanced convection), with positive dynamic height anomalies associated with downwelling Kelvin waves contoured in red. Easterly wind anomalies are contoured in black and westerly anomalies in light gray. Figure 1b shows the same quantities, except that the composite background has been subtracted, leaving only the intraseasonal signal. Comparison of the two panels suggests a background pattern of enhanced convection and low-level westerly winds over the central Pacific and, to a lesser extent, the western Indian basin. Anomalies of active intraseasonal convection and westerly wind (Fig. 1b) begin over the Indian basin near lag −30 days. These anomalies propagate eastward at roughly 5–7 m s−1, consistent with the MJO, but with reduced amplitude, apparently due to suppression of convection over the far eastern Indian basin and the Maritime Continent by background mean El Niño conditions (Roundy et al. 2010). Upon arrival over the west Pacific, these anomalies become associated with a positive dynamic height anomaly. These three anomaly fields move eastward together and amplify prior to crossing the date line, and they continue eastward together until the wind and convective anomalies become insignificant near 150°W. The eastward phase speed marked by the solid black line is 2 m s−1. This pattern represents the apparent coupled mode (Lau and Shen 1988; Roundy and Kiladis 2006), and it is inconsistent over the Pacific with prevailing views of the MJO because of its smaller phase speed and narrower zonal extent. [This statement is based on traditional composites of the MJO such as Hendon and Salby (1994).]

Fig. 1.

Composite OLR anomaly (shading, W m−2), Kelvin wave dynamic height (positive contoured in red, and negative in blue, with a contour interval of 1 cm and the 0 contour omitted), and 850-hPa zonal wind anomaly (westerly in gray, every 2 m s−1 beginning at +1; easterly in black, every 2 m s−1 beginning at −1). The heavy dashed black line references 5 m s−1 for the MJO; the solid black line references 2 m s−1 for the oceanic Kelvin wave. (a) Total composite anomaly; (b) composite anomaly with the low-frequency background subtracted as described in the text.

Fig. 1.

Composite OLR anomaly (shading, W m−2), Kelvin wave dynamic height (positive contoured in red, and negative in blue, with a contour interval of 1 cm and the 0 contour omitted), and 850-hPa zonal wind anomaly (westerly in gray, every 2 m s−1 beginning at +1; easterly in black, every 2 m s−1 beginning at −1). The heavy dashed black line references 5 m s−1 for the MJO; the solid black line references 2 m s−1 for the oceanic Kelvin wave. (a) Total composite anomaly; (b) composite anomaly with the low-frequency background subtracted as described in the text.

Most of the Kelvin wave events that apparently become coupled to atmospheric convection develop initially as the active convective phase of the MJO arrives over the equatorial west Pacific. Anomalies of atmospheric convection and surface winds then amplify and reduce in zonal scale. In spite of differences between the coupled pattern and the MJO, it is interesting to analyze its evolution with tools that have become popular to diagnose and track the MJO. The Wheeler and Hendon (2004) real-time multivariate MJO (RMM) indices are widely applied for tracking MJO-like signals in real time. To diagnose the RMM signal associated with the coupled mode, the composite OLR and 850- and 200-hPa zonal winds were averaged over 15°N to 15°S, normalized by dividing by the respective standard deviations of the unfiltered wind and OLR anomalies, and projected onto the zonal patterns of the first and second EOFs that were previously applied by Wheeler and Hendon (2004) to calculate the RMM indices. The resulting projection values (normalized by dividing by their respective SDs) are plotted in the phase diagram shown in Fig. 2. From lag −40 to +40 days, the composite is associated with more than one complete orbit through the phase space. Although most of the data points reside within the 1-SD circle, all points are significantly different from zero at above the 95% level. The 0 time lag occurs near the start of phase 8, and the residence time in phases 7–8 is substantially longer than average, consistent with the reduced eastward phase speed across the Pacific.

Fig. 2.

Projection of the composite OLR and 200- and 850-hPa zonal winds onto the Wheeler–Hendon real-time multivariate MJO EOFs. Each × represents the values on the indicated time lag.

Fig. 2.

Projection of the composite OLR and 200- and 850-hPa zonal winds onto the Wheeler–Hendon real-time multivariate MJO EOFs. Each × represents the values on the indicated time lag.

b. Composite anomalous background state

Figure 3 shows composite background OLR anomalies (shaded, with blue suggesting enhanced convection) and 300-hPa geopotential height anomalies (contours, positive in red) at the zero-day time lag. Anomalous active convection straddles the date line and suppressed convection extends across the Maritime Continent region. Upper-level ridges form a twin pattern across the equator over the east Pacific, with a wave train extending northward and eastward across the east Pacific, North America, the North Atlantic, and then southeastward across Europe. This result is consistent with an El Niño background state (e.g., Hoerling and Kumar 2002) weighted by the months of the events listed in Table I.

Fig. 3.

Composite background state in OLR (shading, W m−2) and 300-hPa geopotential height (contours, positive in red). The 0 contour is omitted, and the contour interval is 10 m.

Fig. 3.

Composite background state in OLR (shading, W m−2) and 300-hPa geopotential height (contours, positive in red). The 0 contour is omitted, and the contour interval is 10 m.

c. Composite total field

Figures 4a–h show the composite OLR, 850-hPa wind, and 300-hPa geopotential height anomalies for consecutive 5-day lags from day −35 to day +35. Geopotential height anomalies that are significantly different from the 120-day mean background state are enclosed in heavy black contours. At lag −35 days, active convection is located along the equator near the date line and in the Indian basin. Low-level westerly anomalies occur along the equator across much of the Pacific, with low-level easterly anomalies over the Indian basin. A significant upper-level trough anomaly occurs over extreme northeastern Asia, and a significant wave train is apparent from near Greenland southeastward to the Middle East. Similar convective anomalies are evident at lag −25 days, but the high-latitude wave train amplifies and becomes more zonally uniform, with a substantial ridge anomaly over the far North Pacific and trough anomalies over northern and eastern Europe and northwestern Canada. At lag −15 days, positive dynamic height anomalies associated with the developing downwelling Kelvin wave become evident in the west Pacific. At the same time, enhanced convection amplifies over the equatorial Pacific along with low-level westerly wind anomalies. Suppressed convection begins to amplify over the eastern equatorial Indian Ocean basin. The North Pacific ridge anomaly evident before at lag −25 days regresses westward but also extends eastward across the Gulf of Alaska and the far eastern North Pacific. The trough anomaly previously over northern Europe is replaced by a ridge that extends across northern Asia. The trough anomaly previously evident over northwestern Canada expands, and a significant ridge anomaly develops over eastern North America. At lag −5 days, the Kelvin wave approaches the date line, suppressed convection and low-level easterly anomalies expand over the Maritime Continent region including Northern Australia, and enhanced convection and low-level westerly winds amplify over the central Pacific. This pattern of winds and convection is consistent with constructive interference between the MJO and an El Niño background state. A train of significant 300-hPa geopotential height anomalies extends around the northern Pacific Rim, with a significant ridge anomaly over northern Europe. The oceanic Kelvin wave crosses the date line at lag 0 days (Fig. 1). By lag +5 days, the crest of the Kelvin wave is east of the date line and the convective and low-level wind anomalies across the tropics continue to amplify and nudge eastward, with a new anomaly of enhanced convection developing over the western equatorial Indian basin. A trough anomaly that was evident at lag −5 days expands and nudges eastward across the far northeast Pacific and extreme northeast Asia. This trough anomaly also extends eastward into the Gulf of Alaska, replacing the ridge anomaly previously evident there at lag −5 days. A ridge anomaly remains over northern Europe. At lag +15 days, active convection remains collocated with the oceanic Kelvin wave centered near 145°W. The convective signal weakens with increasing time lag, but the pattern in geopotential height intensifies and takes on a structure similar to that of the El Niño background state (Fig. 3), but with substantially greater amplitude. Composite trough anomalies over the eastern North Pacific exceed −90 m. A trough anomaly expands over eastern and northern Europe. A trough anomaly also becomes evident over Texas and the Gulf of Mexico. By lag +25 days, equatorial convection backs to the west of the Kelvin wave dynamic height anomaly, the extratropical wave train across the Pacific and North America deamplifies and reduces in meridional extent, and a trough anomaly expands westward over Europe. By lag +35 days, the more El Niño–like pattern returns with a substantial east Pacific trough and a ridge extending northwestward across the northwestern two-thirds of North America. The trough anomaly remains over the Gulf of Mexico and the southeastern United States.

Fig. 4.

Composite OLR (shading, blue indicates active convection, with a color axis range between −30 and +30 W m−2), 300-hPa geopotential height (contours; red is positive and blue negative, with a contour interval of 10 m and the 0 contour omitted), oceanic Kelvin wave dynamic height (cyan indicates negative and magenta positive, with a contour interval of 1 cm), and 850-hPa wind vectors (m s−1). Heavy black contours enclose regions in which the composite geopotential height anomalies are significantly different from the 120-day moving average background state (represented at lag 0 in Fig. 3). Results are shown for specific pentad time lags noted in the titles (in days from the date-line crossing date of the oceanic wave).

Fig. 4.

Composite OLR (shading, blue indicates active convection, with a color axis range between −30 and +30 W m−2), 300-hPa geopotential height (contours; red is positive and blue negative, with a contour interval of 10 m and the 0 contour omitted), oceanic Kelvin wave dynamic height (cyan indicates negative and magenta positive, with a contour interval of 1 cm), and 850-hPa wind vectors (m s−1). Heavy black contours enclose regions in which the composite geopotential height anomalies are significantly different from the 120-day moving average background state (represented at lag 0 in Fig. 3). Results are shown for specific pentad time lags noted in the titles (in days from the date-line crossing date of the oceanic wave).

To clarify the phase relationships between the oceanic Kelvin wave and high-latitude flow, Fig. 5 shows Hovmöller diagrams of composite OLR and dynamic height anomalies (averaged over 5°N to 5°S) as plotted in Fig. 1, along with contours of 300-hPa geopotential height anomalies averaged from 40° to 50°N (positive anomalies contoured in gray, with negative anomalies in black). Figure 5a represents the composite total anomaly, and Fig. 5b shows only the intraseasonal part. Westward-moving features prevail in geopotential height across Europe and Asia. Eastward propagation seems to dominate across the North Pacific, although the composite suggests some westward-moving features near lag 0 days. Across most of the Pacific and at most time lags, positive high-latitude geopotential height anomalies (black contours) correspond at the same longitude with negative dynamic height anomalies and suppressed equatorial convection. North Pacific trough anomalies remain roughly in phase with the main positive dynamic height anomaly, although the trough anomaly appears stationary within the longitude range of the propagating positive dynamic height anomaly prior to lag zero, after which it shifts abruptly eastward, remaining within the longitude range of the positive dynamic height anomaly. The high- and low-latitude features demonstrate remarkable coherence. Although causality cannot be assured from this analysis, the temporary quasi-stationary motion in the high-latitude feature prior to lag zero and continuous propagation in the low-latitude convective feature suggests that the transition eastward in the high-latitude feature was a response to eastward translation of the heating in the tropical convection. Further, the eastward propagation of oceanic Kelvin waves is a robust process that can occur independent of continuous atmospheric forcing across the basin, whereas no known high-latitude process yields systematic eastward propagation of high-amplitude features near 2 m s−1. The high amplitude of the high-latitude signal ensures that the composite signal does not result from a correlation of local noise with the tropical mode. Although the orientation and characteristics of the high-latitude features can facilitate meridional ventilation of the tropical convection along with Rossby wave dispersion from the tropical features, these patterns together are consistent with the view that the tropical coupled mode drives changes in high-latitude features through redistribution of mass by the tropical convection, yielding potential extended-range predictability in the high-latitude features because the oceanic wave influences the evolution of the convection.

Fig. 5.

Longitude–time representation of composite OLR (shading, W m−2), dynamic height (red contours indicate positive and blue negative, with an interval of 1 cm), and 300-hPa composite geopotential height anomalies averaged from 40° to 50°N (gray is negative and black positive, with a contour interval of 10 m). Heavy dashed and solid straight lines are reproduced from Fig. 1 for reference to the MJO and the oceanic Kelvin wave, respectively. (a) Total composite anomaly; (b) as in (a), but with the low-frequency background subtracted.

Fig. 5.

Longitude–time representation of composite OLR (shading, W m−2), dynamic height (red contours indicate positive and blue negative, with an interval of 1 cm), and 300-hPa composite geopotential height anomalies averaged from 40° to 50°N (gray is negative and black positive, with a contour interval of 10 m). Heavy dashed and solid straight lines are reproduced from Fig. 1 for reference to the MJO and the oceanic Kelvin wave, respectively. (a) Total composite anomaly; (b) as in (a), but with the low-frequency background subtracted.

6. Conclusions

A simple composite analysis shows that significant, high-amplitude patterns of global flow evolve in association with Pacific Ocean Kelvin waves that are apparently coupled to atmospheric convection during October through April. Composite geopotential height and wind anomalies are significantly different from zero across broad regions of the globe, and they exceed the amplitudes of anomalies linked to the background El Niño conditions by at least a factor of 3. Although these Kelvin waves usually develop in association with active convective phases of the MJO over the west Pacific, they evolve differently from the average MJO, with phase speeds near 2 m s−1 with the associated convective anomalies extending over a narrower range of longitudes. In spite of these fundamental differences from the MJO, the events project well onto diagnostics of the MJO signal, including the Wheeler and Hendon (2004) RMM indices. These projections indicate longer than normal residence times in the Pacific and Western Hemisphere phases when the coupled mode is present. Further, the active convection associated with these events extends eastward farther than would be suggested by MJO signals described by these indices (without considering the influence of background El Niño conditions during which these coupled events most frequently occur). The continuous propagation of the Kelvin waves and their associated signals in atmospheric convection and more discrete propagation of the high-latitude anomalies suggests that the high-latitude features progress eastward in response to the evolving forcing by tropical convection. The observed eastward propagation of the tropical features would occur independently of other features, although the coupled mode propagates more slowly than the free first baroclinic mode Kelvin wave (Lau and Shen 1988; Roundy and Kiladis 2006). The Kelvin wave likely modulates the progress of the convection, which then modulates the global flow. We also cannot rule out that some portion of the relationship between the waves and the global flow includes some modulation of the waves by that flow.

The coherence of high-latitude features with the tropical convection and dynamic height anomalies suggests that when such events are identified in real time at any stage of their progression, they could be used to greatly enhance the detail of intraseasonal to seasonal weather forecasts during El Niño events, simply by finding the location of the Kelvin wave. Although the composite features are robust, they are likely to vary with the seasonal background state. However, a sufficient number of events is not available to generate similar composites by month. Further, individual events might vary substantially from the composite in response to high-latitude features that might develop independently from the response to convection associated with the coupled mode. For example, high-latitude blocking would modify the structure of the global response pattern. Nevertheless, these results add substantial detail to the background patterns associated with ENSO.

In addition to the composite based on the 19 events that apparently coupled to convection during extended northern winter, we have also analyzed the association between the global flow and Kelvin waves that were not apparently coupled to atmospheric convection. The signal in these events evolves in a similar manner to the composite generated by Roundy et al. (2010) for the MJO during cold ENSO, although the amplitude of convective and geopotential height anomalies is smaller in the composite based on the Kelvin waves (presumably because the location of the wave does not sufficiently specify the timing and amplitude of the associated west Pacific convection that occurred previous to the date-line crossing times of the waves). The authors’ future work includes setting up an algorithm for identifying and tracking such coupled events in real time and performing a contingency analysis of the associations of surface temperatures around the world with observed coupled events. The identification and tracking algorithm might be applied together with the results of the contingency analysis to enhance prediction of global temperature variations. We are also preparing a more detailed analysis of the weather associated with the coupled mode across the tropics of the Pacific basin, and its association with the coupling process itself. This analysis includes a detailed case study of a high-amplitude event during 1986–87 and a composite analysis.

Acknowledgments

Funding was provided by NSF Grant 0850642 to Paul Roundy. The NOAA Earth System Research Laboratory provided OLR and reanalysis data, and the Australian Bureau of Meteorology provided the first and second EOFs of the Wheeler and Hendon (2004) RMM PCs. Conversations with George Kiladis and the MJO-GTH working group improved the presentation, and we thank Klaus Weickmann and two anonymous reviewers for helpful comments.

REFERENCES

REFERENCES
Bergman
,
J. W.
,
H. H.
Hendon
, and
K. M.
Weickmann
,
2001
:
Intraseasonal air–sea interactions at the onset of El Niño.
J. Climate
,
14
,
1702
1719
.
Bladé
,
I.
, and
D. L.
Hartmann
,
1995
:
The linear and nonlinear extratropical response of the atmosphere to tropical intraseasonal heating.
J. Atmos. Sci.
,
52
,
4448
4471
.
Brankovic
,
C.
,
T.
Palmer
, and
L.
Ferranti
,
1994
:
Predictability of seasonal atmospheric variations.
J. Climate
,
7
,
217
237
.
Cordeira
,
J. M.
, and
L. F.
Bosart
,
2010
:
The antecedent large-scale conditions of the “Perfect Storms” of late October and early November 1991.
Mon. Wea. Rev.
,
138
,
2877
2900
.
Federov
,
A. V.
, and
W. K.
Melville
,
2000
:
Kelvin fronts on the equatorial thermocline.
J. Phys. Oceanogr.
,
30
,
1692
1705
.
Ferranti
,
L.
,
T. N.
Palmer
,
F.
Molteni
, and
K.
Klinker
,
1990
:
Tropical–extratropical interaction associated with the 30–60-day oscillation and its impact on medium and extended range prediction.
J. Atmos. Sci.
,
47
,
2177
2199
.
Fu
,
X.
, and
B.
Wang
,
2004
:
Differences of boreal summer intraseasonal oscillations simulated in an atmosphere–ocean coupled model and an atmosphere-only model.
J. Climate
,
17
,
1263
1271
.
Harrison
,
D. E.
, and
P. S.
Schopf
,
1984
:
Kelvin-wave-induced anomalous advection and the onset of surface warming in El Niño events.
Mon. Wea. Rev.
,
112
,
923
933
.
Hendon
,
H. H.
, and
M. L.
Salby
,
1994
:
The life cycle of the Madden–Julian oscillation.
J. Atmos. Sci.
,
51
,
2225
2237
.
Hendon
,
H. H.
,
B.
Liebman
, and
J. D.
Glick
,
1998
:
Oceanic Kelvin waves and the Madden–Julian oscillation.
J. Atmos. Sci.
,
55
,
80
101
.
Hendon
,
H. H.
,
B.
Liebmann
,
M.
Newman
,
J. D.
Glick
, and
J. E.
Schemm
,
2000
:
Medium-range forecast errors associated with active episodes of the Madden–Julian oscillation.
Mon. Wea. Rev.
,
128
,
69
86
.
Higgins
,
R. W.
,
J-K. E.
Schemm
,
W.
Shi
, and
A.
Leetmaa
,
2000
:
Extreme precipitation events in the western United States related to tropical forcing.
J. Climate
,
13
,
793
820
.
Hoerling
,
M. P.
, and
A.
Kumar
,
2002
:
Atmospheric response patterns associated with tropical forcing.
J. Climate
,
15
,
2184
2203
.
Hoskins
,
B. J.
, and
G-Y.
Yang
,
2000
:
The equatorial response to higher-latitude forcing.
J. Atmos. Sci.
,
57
,
1197
1213
.
Inness
,
P. M.
,
J. M.
Slingo
,
E.
Guilyardi
, and
J.
Cole
,
2003
:
Simulation of the Madden–Julian oscillation in a coupled general circulation model. Part II: The role of the basic state.
J. Climate
,
16
,
365
382
.
Jin
,
F.
, and
B. J.
Hoskins
,
1995
:
The direct response to tropical heating in a baroclinic atmosphere.
J. Atmos. Sci.
,
52
,
307
319
.
Jones
,
C.
,
2000
:
Occurrence of extreme precipitation events in California and relationships with the Madden–Julian oscillation.
J. Climate
,
13
,
3576
3587
.
Jones
,
C.
, and
J-K. E.
Schemm
,
2000
:
The influence of intraseasonal variations on medium- to extended-range weather forecasts over South America.
Mon. Wea. Rev.
,
128
,
486
494
.
Jones
,
C.
,
D. E.
Waliser
,
K. M.
Lau
, and
W.
Stern
,
2004a
:
Global occurrences of extreme precipitation and the Madden–Julian oscillation: Observations and predictability.
J. Climate
,
17
,
4575
4589
.
Jones
,
C.
,
D. E.
Waliser
,
K. M.
Lau
, and
W.
Stern
,
2004b
:
The Madden–Julian oscillation and its impact on Northern Hemisphere weather predictability.
Mon. Wea. Rev.
,
132
,
1462
1471
.
Kalnay
,
E.
, and
Coauthors
,
1996
:
The NCEP/NCAR 40-Year Reanalysis Project.
Bull. Amer. Meteor. Soc.
,
77
,
437
471
.
Lau
,
K-M.
, and
S.
Shen
,
1988
:
On the dynamics of intraseasonal oscillations and ENSO.
J. Atmos. Sci.
,
45
,
1781
1797
.
Liebmann
,
B.
, and
C. A.
Smith
,
1996
:
Description of a complete (interpolated) outgoing longwave radiation dataset.
Bull. Amer. Meteor. Soc.
,
77
,
1275
1277
.
Madden
,
R. A.
, and
P. R.
Julian
,
1994
:
Observations of the 40–50-day tropical oscillation—A review.
Mon. Wea. Rev.
,
122
,
814
837
.
Matthews
,
A. J.
, and
G. N.
Kiladis
,
1999
:
The tropical–extratropical interaction between high-frequency transients and the Madden–Julian oscillation.
Mon. Wea. Rev.
,
127
,
661
677
.
Matthews
,
A. J.
,
B. J.
Hoskins
, and
M.
Masutani
,
2004
:
The global response to tropical heating in the Madden–Julian oscillation during the northern winter.
Quart. J. Roy. Meteor. Soc.
,
130
,
1991
2011
.
Mo
,
K. C.
,
2000
:
Intraseasonal modulation of summer precipitation over North America.
Mon. Wea. Rev.
,
128
,
1490
1505
.
Mo
,
K. C.
, and
R. W.
Higgins
,
1998
:
Tropical influences on California precipitation.
J. Climate
,
11
,
412
430
.
Nogues-Paegle
,
J.
,
L. A.
Byerle
, and
K. C.
Mo
,
2000
:
Intraseasonal modulation of South American summer precipitation.
Mon. Wea. Rev.
,
128
,
837
850
.
Pohl
,
B.
, and
A. J.
Matthews
,
2007
:
Observed changes in the lifetime and amplitude of the Madden–Julian oscillation associated with interannual ENSO sea surface temperature anomalies.
J. Climate
,
20
,
2659
2674
.
Ray
,
P.
,
C.
Zhang
,
J.
Dudhia
, and
S. S.
Chen
,
2009
:
A numerical case study on the initiation of the Madden–Julian oscillation.
J. Atmos. Sci.
,
66
,
310
331
.
Roundy
,
P. E.
, and
G. N.
Kiladis
,
2006
:
Observed relationships between oceanic Kelvin waves and atmospheric forcing.
J. Climate
,
19
,
5253
5272
.
Roundy
,
P. E.
, and
G. N.
Kiladis
,
2007
:
Analysis of a reconstructed oceanic Kelvin wave dynamic height dataset for the period 1974–2006.
J. Climate
,
20
,
4341
4355
.
Roundy
,
P. E.
, and
J. R.
Kravitz
,
2009
:
The association of the evolution of intraseasonal oscillations to ENSO phase.
J. Climate
,
22
,
381
395
.
Roundy
,
P. E.
,
K.
MacRitchie
,
J.
Asuma
, and
T.
Melino
,
2010
:
Modulation of the global atmospheric circulation by combined activity in the Madden–Julian oscillation and the El Niño–Southern Oscillation during boreal winter.
J. Climate
,
23
,
4045
4059
.
Seiki
,
A.
,
Y. N.
Takayabu
,
K.
Yoneyama
,
N.
Sato
, and
M.
Yoshizaki
,
2009
:
The oceanic response to the Madden–Julian oscillation and ENSO.
SOLA
,
5
,
93
96
.
Waliser
,
D. E.
,
K. M.
Lau
,
W.
Stern
, and
C.
Jones
,
2003
:
Potential predictability of the Madden–Julian oscillation.
Bull. Amer. Meteor. Soc.
,
84
,
33
50
.
Wallace
,
J. M.
, and
D. S.
Gutzler
,
1981
:
Teleconnections in the geopotential height field during Northern Hemisphere winter.
Mon. Wea. Rev.
,
109
,
784
812
.
Weickmann
,
K. M.
, and
P. D.
Sardeshmukh
,
1994
:
The atmospheric angular momentum cycle associated with a Madden–Julian oscillation.
J. Atmos. Sci.
,
51
,
3194
3208
.
Wheeler
,
M. C.
, 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
.
Wilks
,
D. S.
,
2006
:
Statistical Methods in the Atmospheric Sciences.
Academic Press, 627 pp
.
Zhang
,
C.
,
2005
:
The Madden–Julian Oscillation.
Rev. Geophys.
,
43
,
RG2003
.
doi:10.1029/2004RG000158
.
Zhang
,
C.
, and
J.
Gottschalck
,
2002
:
SST anomalies of ENSO and the Madden–Julian oscillation in the equatorial Pacific.
J. Climate
,
15
,
2429
2445
.

Footnotes

Corresponding author address: Paul E. Roundy, Department of Atmospheric and Environmental Sciences, DAES-ES351, Albany, NY 12222. Email: roundy@atmos.albany.edu