Abstract

An international field campaign, Dynamics of the Madden Julian Oscillation (DYNAMO), took place in the Indian Ocean during October 2011–March 2012 to collect observations for the Madden–Julian oscillation (MJO), especially its convective initiation processes. The large-scale atmospheric and oceanic conditions during the campaign are documented here. The ENSO and the Indian Ocean dipole (IOD) states, the monthly mean monsoon circulation and its associated precipitation, humidity, vertical and meridional/zonal overturning cells, and ocean surface currents are discussed. The evolution of MJO events is described using various fields and indices that have been used to subdivide the campaign into three periods. These periods were 1) 17 September–8 December 2011 (period 1), which featured two robust MJO events that circumnavigated the global tropics with a period of less than 45 days; 2) 9 December 2011–31 January 2012, which contained less coherent activity (period 2); and 3) 1 February–12 April 2012, a period that featured the strongest and most slowly propagating MJO event of the campaign (period 3). Activities of convectively coupled atmospheric Kelvin and equatorial Rossby (ER) waves and their interaction with the MJO are discussed. The overview of the atmospheric and oceanic variability during the field campaign raises several scientific issues pertaining to our understanding of the MJO, or lack thereof. Among others, roles of Kelvin and ER waves in MJO convective initiation, convection-circulation decoupling on the MJO scale, applications of MJO filtering methods and indices, and ocean–atmosphere coupling need further research attention.

1. Introduction

The importance of the Madden–Julian oscillation (MJO; Madden and Julian 1971, 1972) to global weather and climate has been increasingly recognized. Through its global influences on a variety of phenomena over a wide range of temporal and spatial scales, including extreme events such as modulation of tropical cyclone activity, the MJO plays a unique role of bridging weather and climate (Zhang 2013). As a major source of global predictability on the subseasonal time scale up to 40 days (e.g., Tung et al. 2011; Waliser 2011), the ability to forecast the MJO would yield tremendous societal benefit (Jones et al. 2011; Waliser 2011). Recent progress has increased useful MJO forecasts to a general range of 15–25 days (Kang and Kim 2010; Rashid et al. 2011; Vitart and Molteni 2010; Matsueda and Endo 2011; Gottschalck et al. 2010; Gottschalck 2011), thanks to advanced physical parameterizations, initial conditions, and forecast calibration. But this upper limit of our current MJO forecast skill is still approximately half of the known MJO predictability (Tung et al. 2011; Waliser 2011). For many models, the MJO forecast skill is the lowest when the current location of the MJO active convection is located over the Indian Ocean (Matsueda and Endo 2011). Meanwhile, the inability of many global climate models to reproduce MJO signals, which has persisted for decades (Slingo et al. 1996; Lin et al. 2006; Kim et al. 2009; Sperber and Kim 2012), limits their applications to the study of climate fluctuations in extreme events that are subject to MJO influences.

To expedite the progress of advancing our understanding of the MJO and improving our forecasting of it, an international field campaign with participation from 16 countries took place in the central equatorial Indian Ocean (IO) during October 2011–March 2012 to collect observations for the study of convective initiation of the MJO (hereafter referred to as MJO initiation). The field campaign was coordinated internationally under the program of the Cooperative Indian Ocean Experiment on Intraseasonal Variability in Year 2011 (CINDY2011). The U.S. participation was organized under Dynamics of the MJO (DYNAMO) with contributions from the Atmospheric Radiation Measurement (ARM) MJO Investigation Experiment (AMIE) and Littoral Air–Sea Processes (LASP). While the majority of the field observations were taken over the central-equatorial IO, data were also collected from an extended sounding network that stretched from East Africa to the western Pacific (WP). AMIE established twin observing sites at Gan Island in the IO (AMIE-Gan) and Manus Island in the WP (AMIE-Manus) to monitor the same MJO events over much of the Indo-Pacific warm pool region during the campaign using a nearly identical suite of instruments. The science rationale and hypotheses, experimental design and operation, data collection, and preliminary results of the field campaign (hereafter referred to as the DYNAMO campaign) are summarized in Yomeyama et al. (2013).

In this article, the atmospheric and oceanic conditions, both their means and variability, during DYNAMO are documented with the purpose to provide the large-scale background within which the detailed instrumental observations were collected. It also brings attention to scientific issues related to the large-scale background that should be considered when interpreting instrumental observations. Satellite data and global data assimilation products are used here. Even though the global reanalyses were constrained by some field observations (e.g., radiosondes), their details over the campaign domain are subject to validation against field observations. Discussions on the background state include the phases of El Niño–Southern Oscillation (ENSO) and the Indian Ocean dipole (IOD; Saji et al. 1999), the Asian monsoon circulation and its associated rainfall, the meridional- and zonal-vertical circulations and their related humidity field, and the oceanic surface current and equatorial thermocline (section 2). Detailed descriptions are given on the evolution of MJO events (section 3) and activities of equatorial Rossby (ER) waves (section 4) and convectively coupled Kelvin waves (section 5). Finally, section 6 discusses the evolution of oceanic signals associated with forcing by subseasonal atmospheric variability. The conclusions are given in section 7 and a list of acronyms used are provided in the  appendix.

2. Background state

a. ENSO and IOD

The start of the DYNAMO campaign in October 2011 took place during reemergence of a moderate cold phase of ENSO, or La Niña, with eastern equatorial Pacific sea surface temperature (SST) anomalies close to −1°C in the Niño-3.4 region (Fig. 1), and near the peak of a moderate positive phase of the IOD that peaked in October of 2011 (Fig. 2). During this month, negative SST anomalies offshore of Sumatra were less than 1°C and positive SST anomalies in the western IO were close to +1°C (Fig. 3b). The warmest sea surface (29°–30°C) of the IO was mostly north of the equator (Fig. 3a). As the negative SST anomalies offshore of Sumatra quickly faded away after October (Fig. 3d), the positive phase of the IOD practically ended (Fig. 2), except for positive SST anomalies scattered in the IO for two more months (Figs. 3d,f). The strength of La Niña continued until February, but quickly decreased thereafter as indicated by lessening negative Niño-3.4 SST anomalies (Fig. 1) and reduced coverage of negative SST anomalies in the equatorial Pacific Ocean in February and March (Figs. 3j,l).

Fig. 1.

Time series of Niño-3.4 SST anomalies for (top) 1993–2012 and (bottom) June 2011–May 2012. Vertical dashed lines mark the beginning and end of the field campaign. Courtesy of NOAA/Climate Prediction Center (CPC) Climate Diagnostic Bulletin.

Fig. 1.

Time series of Niño-3.4 SST anomalies for (top) 1993–2012 and (bottom) June 2011–May 2012. Vertical dashed lines mark the beginning and end of the field campaign. Courtesy of NOAA/Climate Prediction Center (CPC) Climate Diagnostic Bulletin.

Fig. 2.

Time series of an Indian Ocean dipole model index (°C). Vertical dashed lines mark the beginning and end of the field campaign. Courtesy of APEC Climate Center.

Fig. 2.

Time series of an Indian Ocean dipole model index (°C). Vertical dashed lines mark the beginning and end of the field campaign. Courtesy of APEC Climate Center.

Fig. 3.

Monthly means of (left) SST and (right) SST anomalies (base period 1981–2010) for (a),(b) October, (c),(d) November, and (e),(f) December 2011; and (g),(h) January, (i),(j) February, and (k),(l) March 2012. Courtesy of NOAA/CPC Climate Diagnostic Bulletin.

Fig. 3.

Monthly means of (left) SST and (right) SST anomalies (base period 1981–2010) for (a),(b) October, (c),(d) November, and (e),(f) December 2011; and (g),(h) January, (i),(j) February, and (k),(l) March 2012. Courtesy of NOAA/CPC Climate Diagnostic Bulletin.

b. State at the beginning of the field campaign

In October, the heaviest mean precipitation over the IO was near the equator in the eastern sections (~400 mm) and slightly slanted southward across western sections with increased intensity (>450 mm), as shown in Fig. 4a. A gap in rainfall was present along the equatorial East African coast between the heavy rainfall areas over the IO and central-equatorial Africa. Precipitation over the Maritime Continent (MC) and WP was moderate, with the heaviest rainfall over major islands (Borneo, Philippines, and New Guinea). The typical low-level (850 hPa) circulation of the boreal summer monsoon that prevailed in September (not shown) was replaced by weak easterlies over the Indian subcontinent and the Arabian Sea. The DYNANO sounding arrays were surrounded by low-level easterlies to the north and south, westerlies to the west, and weak wind to the east (Fig. 5a).

Fig. 4.

Monthly mean SSM/I precipitation (mm) for (a)–(c) October–December 2011 and (d)–(f) January–March 2012. Courtesy of NOAA/CPC Climate Diagnostic Bulletin.

Fig. 4.

Monthly mean SSM/I precipitation (mm) for (a)–(c) October–December 2011 and (d)–(f) January–March 2012. Courtesy of NOAA/CPC Climate Diagnostic Bulletin.

Fig. 5.

(left) Monthly means and (right) anomalies (base period 1981–2010) of 850-hPa wind vector (m s−1) and RH (shades/colors, %) from the European Centre for Medium-Range Weather Forecasts (ECMWF) reanalysis for (a),(g) October, (b),(h) November, and (c),(i) December 2011; and (d),(j) January, (e),(k) February, and (f),(l) March 2012. White dots mark the DYNAMO sounding array and ARM Manus site.

Fig. 5.

(left) Monthly means and (right) anomalies (base period 1981–2010) of 850-hPa wind vector (m s−1) and RH (shades/colors, %) from the European Centre for Medium-Range Weather Forecasts (ECMWF) reanalysis for (a),(g) October, (b),(h) November, and (c),(i) December 2011; and (d),(j) January, (e),(k) February, and (f),(l) March 2012. White dots mark the DYNAMO sounding array and ARM Manus site.

A wide ascending branch of the local (70°–80°E) vertical-meridional circulation resided over the DYNAMO sounding array (7°N–8°S) with classic deep overturning cells on both sides (Fig. 6a). The ascending motions over the DYNAMO array were anomalously strong in comparison to the October climatology (Fig. 6g). Coinciding with this upward branch, was a moisture pillar with its column split from the lower to midtroposphere (800–300 hPa). As expected, air in the descending branches in the subtropics in both hemispheres was very dry [relative humidity (RH < 10%)], especially in the lower troposphere. The largest negative anomalies in RH were found in the 700–500-hPa layer immediately south of the DYNAMO array (Fig. 6g).

Fig. 6.

(left) Monthly means and (right) anomalies (base period 1981–2010) of vw streamlines and RH (shades/colors, %) from ECMWF reanalysis for (a),(g) October, (b),(h) November, and (c),(i) December 2011; and (d),(j) January, (e),(k) February, and (f),(l) March 2012, averaged over 70°–80°E. Vertical velocities have been scaled by 3000. Vertical white lines mark the latitudinal range of the DYNAMO sounding array.

Fig. 6.

(left) Monthly means and (right) anomalies (base period 1981–2010) of vw streamlines and RH (shades/colors, %) from ECMWF reanalysis for (a),(g) October, (b),(h) November, and (c),(i) December 2011; and (d),(j) January, (e),(k) February, and (f),(l) March 2012, averaged over 70°–80°E. Vertical velocities have been scaled by 3000. Vertical white lines mark the latitudinal range of the DYNAMO sounding array.

In the zonal direction, the Walker circulation was robust, with its upward branch sitting over the eastern IO, MC, and WP (60°–150°E) and its broader downward branch over the central and eastern Pacific (Fig. 7a). Two other ascending branches were almost equally robust, if narrower, over Africa and South America. But their associated downward branches to the east were much weaker. In fact, there was no well-defined downward motion penetrating through the troposphere between the upward branch of the Walker circulation and that over Africa. Consequently anomalous upward vertical motions were found there (Fig. 7g). The strongest anomalous downward motions were located near the date line.

Fig. 7.

(left) Monthly means and (right) anomalies (base period 1981–2010) of uw streamlines and RH (shades, %) from ECMWF reanalysis for (a),(g) October, (b),(h) November, and (c),(i) December 2011; and (d),(j) January, (e),(k) February, and (f),(l) March 2012 averaged over 5°S–5°N. Vertical velocity has been scaled by 3000. Vertical while lines mark the general longitudes of the DYNAMO array and ARM Manus site.

Fig. 7.

(left) Monthly means and (right) anomalies (base period 1981–2010) of uw streamlines and RH (shades, %) from ECMWF reanalysis for (a),(g) October, (b),(h) November, and (c),(i) December 2011; and (d),(j) January, (e),(k) February, and (f),(l) March 2012 averaged over 5°S–5°N. Vertical velocity has been scaled by 3000. Vertical while lines mark the general longitudes of the DYNAMO array and ARM Manus site.

Based on the National Oceanic and Atmospheric Administration (NOAA) Ocean Surface Current Analysis—Real time (OSCAR) product, a seasonal, healthy Wyrtki jet, an energetic westward equatorial surface current (Wyrtki 1973), extended from 55°E all the way to Sumatra and a South Equatorial Current (SEC) that meandered slightly in the central IO (Fig. 8a). The IO thermocline was flat at about 100-m depth (Fig. 9a).

Fig. 8.

Monthly mean near surface current for (a) October, (b) November, and (c) December 2011 and (d) January, (e) February, and (f) March 2012 from NOAA Ocean Surface Current Analysis—Real time.

Fig. 8.

Monthly mean near surface current for (a) October, (b) November, and (c) December 2011 and (d) January, (e) February, and (f) March 2012 from NOAA Ocean Surface Current Analysis—Real time.

Fig. 9.

Monthly mean equatorial ocean temperature (°C) in the Indian Ocean for (a) October, (b) November, and (c) December 2011 and (d) January, (e) February, and (f) March 2012 produced by the NOAA/NCEP’s global ocean data assimilation system, courtesy of NOAA/CPC Climate Diagnostic Bulletin.

Fig. 9.

Monthly mean equatorial ocean temperature (°C) in the Indian Ocean for (a) October, (b) November, and (c) December 2011 and (d) January, (e) February, and (f) March 2012 produced by the NOAA/NCEP’s global ocean data assimilation system, courtesy of NOAA/CPC Climate Diagnostic Bulletin.

c. Large-scale evolution

The subsequent evolution of the large-scale atmospheric and oceanic conditions in the tropical IO and the surrounding regions through DYNAMO generally followed their expected seasonal cycle. Rainfall over the IO continued to migrate southward, along with rainfall over Africa, the MC, and South America (Fig. 4). The IO ITCZ south of the equator, if ambiguous during October–December, was clearly established as a permanent feature after December (Figs. 4d–f). The lack of rain over the northern DYNAMO array in January and March formed a sharp contrast to the earlier months when there was abundant rainfall.

The southward migration of rainfall was accompanied by a steady southward migration of the vertical-meridional circulation over the longitudes of the DYNAMO array (Fig. 6). The ascending branch of the circulation and its associated moisture pillar were almost centered at the equator in November and moved to south of the equator thereafter. The quasi-symmetric overturning cells on both sides persisted until January when the northern cell dominated and the southern cell lost its identity. RH in the boundary layer beneath the descending branch of the meridional circulation became lower in the Northern Hemisphere than in the Southern Hemisphere during November and more so in the later months. The DYNAMO sounding array was always in the ascending branch, in both mean and anomalies, except in January and March when most areas were covered by anomalously strong descending motions (Figs. 6d,j,f,l). Accompanying these descending motions was anomalously dry air that was apparently connected to the upper troposphere of the Northern Hemispheric extratropics (Fig. 6j). There were subtle differences in the structure of the meridional circulations through these months. A shallow circulation distinctly embedded in the classic deep circulation, which has been observed in many other parts of the tropics (Zhang et al. 2008), appeared in the northern cell in November (Fig. 6b) and January (Fig. 6d). The low-level southerly flow of the southern cell was shallow (from the surface to 700 hPa) in October and December, but deep (from the surface to 500 hPa) in November.

The horizontal distribution of low-level moisture was closely related to the gradual establishment of the austral summer monsoon circulation (Fig. 5). In November, strong low-level easterlies on both sides of the equator (10°–20°N/S) flanked weak westerlies near the equator. The northeasterly monsoon flow started in December over the Arabian Sea further developed in January and advected continental dry air toward the equator, continuously decreasing the mean humidity over the northern DYNAMO array. Strong equatorial westerlies were established in December, persisted in January, and weakened in February. As the moisture belt over the IO migrated southward from October through February, high humidity appeared to remain over the MC. There was a tendency for a reversal in the equatorial zonal gradient of RH from the earlier months (more moisture over the IO) to the later months (more moisture over the MC).

In contrast to the vertical-meridional circulation, no strong seasonal shift was observed in the vertical-zonal circulation (Fig. 7). The mean ascending branch of the Walker circulation remained over the MC, with its width expanding and contracting. The DYNAMO array was mainly at the edge of the mean ascending branch, except during January and March, when the ascending motion became narrow and shifted eastward (Figs. 7d,f). Mean descending motions and associated drying over the IO were much weaker than over the Pacific and Atlantic Oceans. The anomalous upward motions appeared to move gradually from the IO in October to the WP in January, then back to the IO and the DYNAMO array in February (Figs. 7g–k). The anomalous dry lower troposphere above 850 hPa over the DYNAMO array in January (Fig. 6j) actually extended westward and covered most of the equatorial IO (Fig. 7j). The central Pacific was always dominated by anomalous downward motion, consistent with ongoing La Niña conditions (Fig. 1).

The monsoon flow shown in Fig. 5 produced substantial effects on the ocean surface currents. The Wyrtki jet persisted through November and became surprisingly strong in December (Fig. 8c) when it occupied nearly the entire equatorial IO, with a maximum amplitude reaching 0.7 m s−1. December is not the typical month for a strong Wyrtki jet. The strong low-level (and presumably surface) equatorial westerlies over the IO (Fig. 5c) might have played an indispensible role in maintaining and strengthening the Wyrtki jet. It is interesting, however, that while the equatorial westerlies persisted into January (Fig. 5d), the Wyrtki jet disappeared (Fig. 8d). More puzzling is that a strong equatorial surface westward current emerged in February (Fig. 8e) against strong westerly wind (Fig. 5e).

The zonally flat equatorial thermocline persisted from October through November but began to tilt in December (Fig. 9c). The shoaling to the west and deepening to the east of the thermocline were consistent with the westerly wind forcing that became strong in December (Fig. 5c). This zonal tilt of the equatorial thermocline persisted through February.

3. MJO

a. Recent historical context

The MJO was active during much of DYNAMO. Not only was a robust MJO evident, but its varying characteristics and interactions with other coherent subseasonal tropical variability make it an interesting period. It is important to put DYNAMO into recent historical context to note the increase in activity as compared to the recent past but also to highlight how potentially bountiful this period, so heavily observed, may be to future research efforts.

Figure 10 shows time series of two MJO activity measures from January 2002 through March 2012: 1) a 91-day running mean of the Wheeler and Hendon (2004) MJO index [hereafter the Real-time Multivariate MJO (RMM) index] amplitude (Fig. 10a) and 2) a 91-day running mean of the absolute value of standardized MJO-filtered OLR anomalies (Fig. 10b). The latter is a localized measure encompassing the DYNAMO array only in the central-equatorial IO (averaged from 10°N–10°S and from 70°–80°E). The OLR anomalies are filtered for eastward propagation with wavenumbers 0–9 and periods of 20–100 days (Table 1), similar to Wheeler and Kiladis (1999) and Kiladis et al. (2005).

Fig. 10.

Time series of (a) the RMM index amplitude and (b) the local Indian Ocean index (OLR version). Plotted is the absolute value of MJO filtered daily OLR anomalies from January 2002 through March 2012. The values in (b) are standardized with respect to the 1979–2012 period. A 91-day running mean is applied to each time series.

Fig. 10.

Time series of (a) the RMM index amplitude and (b) the local Indian Ocean index (OLR version). Plotted is the absolute value of MJO filtered daily OLR anomalies from January 2002 through March 2012. The values in (b) are standardized with respect to the 1979–2012 period. A 91-day running mean is applied to each time series.

Table 1.

Filters used in this study to target MJO, Kelvin, and ER wave signals. No equatorial symmetry constraints are assumed.

Filters used in this study to target MJO, Kelvin, and ER wave signals. No equatorial symmetry constraints are assumed.
Filters used in this study to target MJO, Kelvin, and ER wave signals. No equatorial symmetry constraints are assumed.

The RMM index summarizes a portion of the complexities of the MJO into a simple global scale index for easier interpretation and a time series is shown in Fig. 10a illustrating the limited amount of long-lived, robust MJO activity after May 2005 through August 2011. Only during the periods November 2007–February 2008 and March–May 2009 did the MJO approach the duration and robustness seen during DYNAMO. Figure 10b also shows the overall lack of robust, long-lived MJO activity during the 2005–11 period similar to the global RMM measure, although variability is larger (i.e., OLR only, over a smaller domain). The several months just prior to DYNAMO were one of the least active periods in both time series during 2002–12.

b. RMM index review

The RMM index was utilized by the DYNAMO extended-range forecast team and many others during DYNAMO and we begin our assessment by reviewing its evolution. Figure 11 illustrates the RMM index during DYNAMO and, along with Fig. 12 [time–longitude diagram of velocity potential, section 3c(1)], is used to define three distinct DYNAMO periods (DP1, DP2, and DP3). These periods are (i) DP1, 17 September–8 December 2011; (ii) DP2, 9 December 2011–31 January 2012; and (iii) DP3, 1 February–12 April 2012. Although the index amplitude was small, the MJO strengthened during the second half of September as distinct eastward propagation from the eastern IO to the WP is evident (counterclockwise motion from phase 3 to phase 6 of the blue line).

Fig. 11.

RMM index phase diagrams for (a) DP1 17 Sep–8 Dec 2011, (b) DP2 9 Dec 2011–31 Jan 2012, and (c) DP3 1 Feb–12 Apr 2012. Each point represents a daily value with the specific day of the month just to its left. Different color lines (with annotated text) represent the different months. This point represents the location of the enhanced phase of the MJO. Counterclockwise motion indicates eastward propagation and the farther away from the origin, the stronger the MJO amplitude. Phases of the MJO enhanced phase are labeled and numbered on the diagram. The red arrow highlights the subseasonal event described in the text.

Fig. 11.

RMM index phase diagrams for (a) DP1 17 Sep–8 Dec 2011, (b) DP2 9 Dec 2011–31 Jan 2012, and (c) DP3 1 Feb–12 Apr 2012. Each point represents a daily value with the specific day of the month just to its left. Different color lines (with annotated text) represent the different months. This point represents the location of the enhanced phase of the MJO. Counterclockwise motion indicates eastward propagation and the farther away from the origin, the stronger the MJO amplitude. Phases of the MJO enhanced phase are labeled and numbered on the diagram. The red arrow highlights the subseasonal event described in the text.

Fig. 12.

Time–longitude diagram for 5-day running mean of daily 200-hPa velocity potential anomalies for the period from 1 Sep 2011 through 31 Mar 2012. The data are averaged from 10°S to 10°N with the period mean removed. The base period is 1981–2010. Red dashed (dotted) lines highlight eastward propagation of upper-level divergence (convergence). The dark blue solid and dashed lines represent the less defined subseasonal December event referenced in the text and the black vertical lines denote the DYNAMO array and Manus.

Fig. 12.

Time–longitude diagram for 5-day running mean of daily 200-hPa velocity potential anomalies for the period from 1 Sep 2011 through 31 Mar 2012. The data are averaged from 10°S to 10°N with the period mean removed. The base period is 1981–2010. Red dashed (dotted) lines highlight eastward propagation of upper-level divergence (convergence). The dark blue solid and dashed lines represent the less defined subseasonal December event referenced in the text and the black vertical lines denote the DYNAMO array and Manus.

The MJO became strong during October 2011 and the enhanced phase propagated to the IO and the DYNAMO array by late October (red line in Fig. 11a) and this is termed the October MJO event. The period, as represented by this index, was approximately 45 days. Particularly interesting is the large amplitude when in the Western Hemisphere and western IO phases during mid-October. The amplitude observed in phase 1 was a record for the 1979–2012 period and 87% of the index projection could be attributed to the circulation components of the index (i.e., zonal wind at 850 and 200 hPa; M. Wheeler 2011, personal communication). As the RMM index does not utilize any explicit temporal filtering (a benefit for real-time application), it is susceptible at times to considerable contributions from other modes of subseasonal tropical variability (Roundy et al. 2009). This is evident during October, as the amplitude was reduced during early October across the WP and potentially overly amplified during mid-October. Although the amplitude decreased as the MJO entered the eastern IO, the MJO continued to propagate eastward during November and remained strong (hereafter termed the November event). The period of this event in DP1 to propagate eastward and reenter the IO and once again impact the DYNAMO array was approximately 30 days, considerably shorter than the October event.

Figure 11b illustrates the RMM index for DP2 and shows a considerably less coherent signal than during October and November 2011. There was a tendency for the index to persist in proximity to the MC and the WP and periods when the amplitude was less than one standard deviation (daily points within the displayed circle). Figure 11b also shows, however, a subseasonal MJO-like event that began in mid-to-late December. It is seen here as an increase in amplitude and counterclockwise motion in the phase diagram from approximately 21 December 2011 to 4 January 2012. This event is termed the December event and additional remarks on this case are provided in section 3c and in the conclusion.

Clear renewed eastward propagation in the RMM index is not evident until the development of the strong MJO during DP3 in February and March 2012 (Fig. 11c). The enhanced phase propagated eastward and was centered across the IO at the turn of the month and crossed the WP by early April. The period of this MJO event was closer to 60 days and of longer duration than the MJO events in October and November 2011 (DP1). Interference with other modes of coherent subseasonal tropical variability is evident across the WP during early-to-mid February and again in late March into early April.

c. Velocity potential, OLR, rainfall, 850-hPa winds, moisture, and vertical structure

Section 3b used the RMM index to provide an overview of the MJO during DYNAMO and contribute to defining three periods to aid analysis and discussion here and in subsequent studies. We now focus on important variables to study the MJO in more detail.

1) 200-hPa velocity potential

A common measure to diagnose the MJO is 200-hPa velocity potential and it is often used to assess the full global propagation of the MJO through the tropics to aid in areas where other measures such as OLR and low-level winds often become less distinct. Figure 12 shows the three distinct DYNAMO periods (DP1, DP2, and DP3) outlined in section 3b: 1) alternating periods of enhanced divergence (green shades) and convergence (brown shades) with time from late September 2011 to early December 2011, indicative of coherent eastward propagating MJO activity (labeled DP1); 2) an interval from mid-December 2011 to January 2012 where anomalies were less coherent and at times more persistent in nature (labeled DP2); and 3) a second long-lived period of eastward-propagating MJO activity as shown by alternating anomalies (labeled DP3).

DP1 shows successive MJO activity (Matthews 2008) with two circumnavigating signals (green/brown shades spanning the entire global tropics 2 times) observed over the DYNAMO array. The greatest enhanced divergence was observed over the IO as highlighted by the blue shades while divergence across the equatorial central Pacific was below average in part due to ongoing La Niña conditions during DYNAMO. The November event showed the strongest enhanced divergence across the Eastern Hemisphere to have occurred slightly eastward as compared to the October event, extending across parts of the MC, potentially related to strengthening La Niña conditions.

From mid-December 2011 through January 2012, DP2 showed less coherent, robust MJO activity as anomalies were smaller, often more persistent in certain regions, and eastward propagation was less clear. For several reasons, however, DP2 is just as interesting and important to study. First, DP2 was not absent of activity exhibiting MJO-like dynamical properties as a distinct subseasonal event occurred from mid-to-late December to early January 2012 (highlighted by dark blue solid and dashed lines in Fig. 12). Second, there are varying viewpoints on this subseasonal feature due to a few, sometimes conflicting factors. These include, but are not limited to, (i) short reoccurrence time for enhanced divergence over the IO in mid-to-late January 2012, (ii) less clear global eastward propagation over the period in question, and (iii) questions on the nature of succession from the prior MJO activity in early December. Third, there were several interactions between the MJO and other modes of coherent subseasonal tropical variability, in particular ER waves (section 4).

Last, DP3 arguably saw the strongest MJO activity during DYNAMO. Clear eastward propagation began in late January 2012 near the eastern MC and proceeded to result in the strongest upper-level divergence in the Western Hemisphere during the campaign in February. The enhanced phase of the MJO across the IO occurred during late February into early March and will be termed the February–March event. This renewed activity was slower in propagation speed as compared to 2011 and La Niña conditions played a greater role. Upper-level enhanced convergence was strongest (weakest) near the date line as the suppressed (enhanced) phase of the MJO constructively (destructively) interfered with La Niña. Moreover, upper-level enhanced divergence was strongest (weakest) near the MC as the enhanced (suppressed) phase of the MJO crossed this area.

2) OLR and rainfall

To provide greater detail for periods of anomalous convection and rainfall we now turn to Figs. 13 and 14 where we focus on the Eastern Hemisphere and DYNAMO array, respectively. Since OLR and rainfall each have strengths when monitoring coherent subseasonal tropical variability and also provide different observational perspectives during DYNAMO, both are shown here.

Fig. 13.

Time–longitude diagram of (a) OLR and (b) rainfall averaged 10°S–10°N. Unfiltered anomalies are shaded, and filtered anomalies are contoured with the MJO in blue, ER waves in magenta, and Kelvin waves in red. Contours are drawn every 10 W m−2 and 0.2 mm h−1 for OLR and rainfall, respectively. Solid (dashed) contours correspond to the wet (dry) phases with the zero contour omitted. Only the first contour of the wet phase is shown for Kelvin waves. Vertical solid lines identify the DYNAMO array and Manus with the horizontal dashed lines indicating the DYNAMO periods.

Fig. 13.

Time–longitude diagram of (a) OLR and (b) rainfall averaged 10°S–10°N. Unfiltered anomalies are shaded, and filtered anomalies are contoured with the MJO in blue, ER waves in magenta, and Kelvin waves in red. Contours are drawn every 10 W m−2 and 0.2 mm h−1 for OLR and rainfall, respectively. Solid (dashed) contours correspond to the wet (dry) phases with the zero contour omitted. Only the first contour of the wet phase is shown for Kelvin waves. Vertical solid lines identify the DYNAMO array and Manus with the horizontal dashed lines indicating the DYNAMO periods.

Fig. 14.

Time series of standardized (a) unfiltered, (b) MJO-filtered, (c) ER-filtered, and (d) Kelvin-filtered OLR anomalies (W m−2, black and inverted) and TRMM 3B42 rainfall (mm h−1, blue) averaged 10°S–10°N, 70°–80°E. Faint vertical black lines denote the DYNAMO periods (DP1, DP2, and DP3). Total rainfall (blue) is shown in (a).

Fig. 14.

Time series of standardized (a) unfiltered, (b) MJO-filtered, (c) ER-filtered, and (d) Kelvin-filtered OLR anomalies (W m−2, black and inverted) and TRMM 3B42 rainfall (mm h−1, blue) averaged 10°S–10°N, 70°–80°E. Faint vertical black lines denote the DYNAMO periods (DP1, DP2, and DP3). Total rainfall (blue) is shown in (a).

Starting with DP1, suppressed convection (positive anomalies in Fig. 13a) is observed across the DYNAMO array from late September into early October, early-to-mid-November and last early-to-mid-December. The latter period saw the weakest suppressed convection in part because the MJO showed signs of weakening. Enhanced convection was observed during mid-to-late October and mid-to-late November with the November event resulting in the strongest enhanced convection. Across the DYNAMO array, the October event resulted in the strongest suppressed convection during DP1.

The MJO-related contribution to the data shown in Fig. 13 for the DYNAMO array can be quantified with the localized MJO index introduced in section 3a and this index provides a local scale complement to the RMM index, reviewed in section 3b. Contributions from Kelvin and ER waves are also calculated and these wave projections correspond to the blue, red, and magenta ovals depicted in Fig. 13 for the MJO, Kelvin, and ER waves, respectively. The filters used for these projections, which impose no equatorial symmetry constraint, are listed in Table 1 and are adapted from those used by Kiladis et al. (2009). The methodology is also applied here to rainfall data [the Tropical Rainfall Measuring Mission (TRMM) Multisatellite Precipitation Analysis; TRMM Product 3B42; Huffman et al. 2007]. Figure 14 shows time series of OLR and rainfall daily standardized anomalies during DYNAMO: unfiltered OLR anomalies along with total rainfall (top), MJO associated anomalies (top middle), Kelvin wave associated anomalies (bottom middle), and ER wave associated anomalies (bottom), respectively. Table 2 provides a summary for reference of the dates and magnitudes of extrema based on the filtered rainfall data at the DYNAMO array for the MJO, Kelvin, and ER wave contributions.

Table 2.

Dates and magnitudes of extrema of filtered rainfall at the DYNAMO array. Only events exceeding 1.0σ for the MJO and ER waves or 1.5σ for Kelvin waves are shown.

Dates and magnitudes of extrema of filtered rainfall at the DYNAMO array. Only events exceeding 1.0σ for the MJO and ER waves or 1.5σ for Kelvin waves are shown.
Dates and magnitudes of extrema of filtered rainfall at the DYNAMO array. Only events exceeding 1.0σ for the MJO and ER waves or 1.5σ for Kelvin waves are shown.

The enhanced phase of the MJO peaked across the DYNAMO array for the October and November events during DP1 on 26 October and 26 November, respectively, with suppressed phase peaks occurring on 10 October, 10 November, and 6 December. Although the RMM index suggested a 45-day period for the October event, Fig. 13 suggests that the strongest deep convection over the array was concentrated into a zonally narrow anomaly that propagated eastward at less than 4 m s−1 and lasted less than 1 week centered around 26 October. This period of most concentrated convection did not project onto any widely recognized mode of organized convection even though it occurred embedded within the larger-scale MJO event. This concentrated active event was followed by a final burst of convection associated with a Kelvin wave near 31 October (red oval in Fig. 13).

The enhanced convection and rainfall peak on 26 November was considerably stronger than that on 26 October while the first two suppressed convection peaks were similar in magnitude. The most intense rainfall in the November event was highly concentrated within a relatively brief period between 24 and 28 November. This period of activity was further concentrated within two zonally narrow regions of eastward-propagating convection. Inspection of Fig. 14 also shows that ER wave activity during DP1 was weak, while Kelvin wave activity was strong and contributed to anomalous convection and rainfall at several times (see section 5 for additional details). During much of DP1, Manus in the WP observed suppressed convection (Fig. 13) with only the late September period, prior to DYNAMO, where Manus received substantial enhanced convection associated with the MJO in 2011 (Fig. 13).

The OLR data during DP2 are quite interesting and show development of an area of enhanced convection near and just to the east of the DYNAMO array in mid-to-late December that demonstrates some eastward propagation to the MC. The anomalous convection of this “mini-MJO” projects most strongly onto the MJO rather than on Kelvin wave activity because of its more broad horizontal scale and our choice of filtering boundaries (i.e., 20–100 days for the MJO band; Table 1; Fig. 14), although the associated duration of most intense rainfall over the array was shorter than 5 days. Also, the area of enhanced convection developed considerably farther east as compared to the enhanced convection observed during the October and November 2011 events. Suppressed convection followed thereafter from late December to mid-January with the peak in MJO associated suppressed convection occurring near 1 January 2012. Although the activity projects strongly on the MJO, it is important to note that a strong ER wave was also evident at this time (section 4) and modulated substantially anomalous convection over the IO and MC during this subseasonal event. Over DP2, specifically during early January 2012, Manus experienced increased wetness associated with the MJO.

As previously observed with velocity potential (Fig. 12), there was not a clear signal of eastward propagation into and through the Western Hemisphere during mid-December 2011. The filtering applied to OLR data do not show as clear of an MJO projection across the Western Hemisphere (not shown) toward the end of the November event during DP1 as that earlier in 2011. This evidence suggests that the late December 2011 activity may be a distinct, separate event as compared to the successive activity observed earlier in December 2011 as part of DP1. Additional discussion on this event is provided in the conclusions section. All metrics applied here suggest that the MJO was weak for most of the remainder of January 2012 (Figs. 1214).

The strong MJO activity during DP3 resulted in a substantial period of enhanced convection over the DYNAMO array from mid-February through early March, somewhat longer than the periods in 2011. The enhanced phase of the MJO (with respect to OLR) maximized on 7 March. Kelvin wave activity modulated the anomalous convection several times during this MJO enhanced phase especially during mid-to-late February where near-to-below-average convection was actually observed at times (see section 5 for additional details). Strong suppressed convection entered the DYNAMO array area in mid-March and persisted throughout the remainder of the campaign with a peak in MJO associated suppressed convection near 20 March 2012 (Fig. 14). For Manus, the anomalous convection (first suppressed than enhanced) observed during DP3, was the strongest as any observed during the entire DYNAMO campaign. Enhanced convection occurred in this region for approximately 20–25 days during March 2012.

3) 850-hPa zonal winds

Considerable subseasonal low-level wind variations were observed during DYNAMO. As shown in Fig. 15, easterly wind anomalies were evident across the IO for much of October and November 2011. The strongest easterly winds (maximum anomalies <−6 m s−1) occurred during mid-to-late October and again in mid-to-late November. These two instances correspond to the transition from the suppressed to enhanced phases of the MJO and contributed to gradual moistening across the IO. A period of long-lived and strong westerly wind anomalies averaged across the equator during the enhanced phase of the October event did not materialize. The wind anomalies were of opposite sign about the equator with westerly anomalies across the northern DYNAMO array and easterly anomalies across the southern portion. It was not until the enhanced phase of the November event that strong, broadscale persistent westerly wind anomalies occurred across the IO and the DYNAMO array.

Fig. 15.

Time–longitude diagram for 5-day running mean of daily 850-hPa zonal wind anomalies for the period from 1 Sep 2011 through 31 Mar 2012. The data are averaged from 10°S–10°N with the period mean removed. The base period is 1981–2010. The green rectangle is the strong westerly wind burst described in the text and the black lines denote the DYNAMO array area and Manus.

Fig. 15.

Time–longitude diagram for 5-day running mean of daily 850-hPa zonal wind anomalies for the period from 1 Sep 2011 through 31 Mar 2012. The data are averaged from 10°S–10°N with the period mean removed. The base period is 1981–2010. The green rectangle is the strong westerly wind burst described in the text and the black lines denote the DYNAMO array area and Manus.

A noteworthy event during DYNAMO occurred during DP2 when a prolonged period of large westerly wind anomalies arose in late December into early January. Anomalies maximized near +10 m s−1 and contributed to rough seas for monitoring vessels. These anomalies correspond to the development of enhanced convection across the eastern IO and MC associated with the mini-MJO activity previously described in section 3c(2). Westerly wind anomalies remained in place over the IO throughout much of the remainder of January and continued into early February as convection was more persistent across the MC than in previous months of the campaign.

Weak easterly anomalies returned during February in advance of the enhanced phase of the MJO, but were modulated by wind anomalies associated with an active period of Kelvin wave activity (section 5) and wind departures along the equator were not too far removed from average. The strongest westerly wind anomalies during DP3 occurred in early March 2012, but these were less than those observed in late December 2011.

4) Moisture and vertical structure

One of the primary goals of DYNAMO was to observe the development of low-level moisture over the IO and investigate its role leading up to MJO initiation and maintenance across the Indo-Pacific warm pool region. Figures 16 and 17 review the moisture variability during DYNAMO. Figure 16 illustrates a time–longitude diagram of precipitable water (PW; shading) from the Special Sensor Microwave Imager/Sounder (SSMIS) along with contours that identify MJO-, ER-, and Kelvin-filtered OLR anomalies as in Fig. 13. Land values have been masked in the SSMIS PW data, but some spurious signals from coastlines can still be seen as vertical lines (e.g., near 120°E). Figure 16 is also important in that it explores the moisture variability in the context of interaction between the MJO and Kelvin and ER waves.

Fig. 16.

Time–longitude diagram of precipitable water (mm, shaded), MJO-filtered OLR (black contours), ER-filtered OLR (magenta contours), and KW-filtered (red contours) averaged 10°S–10°N. Contours are drawn every 10 W m−2. Solid (dashed) contours correspond to the wet (dry) phases with the zero contour omitted. Only the first contour of the wet phase is shown for Kelvin waves. Vertical solid black lines identify the DYNAMO array and Manus with horizontal dashed lines denoting DYNAMO periods.

Fig. 16.

Time–longitude diagram of precipitable water (mm, shaded), MJO-filtered OLR (black contours), ER-filtered OLR (magenta contours), and KW-filtered (red contours) averaged 10°S–10°N. Contours are drawn every 10 W m−2. Solid (dashed) contours correspond to the wet (dry) phases with the zero contour omitted. Only the first contour of the wet phase is shown for Kelvin waves. Vertical solid black lines identify the DYNAMO array and Manus with horizontal dashed lines denoting DYNAMO periods.

Fig. 17.

Pressure–time cross section of anomalous uw vectors (plotted as streamlines) and anomalous specific humidity (shading) for (a) September–December 2011 and (b) January–April 2012. The data are averaged from 10°S to 10°N and from 75° to 85°E and anomalies are with respect to a 1981–2010 base period. Green/blue shades indicate above-average moisture while yellow/red shades indicate below-average moisture.

Fig. 17.

Pressure–time cross section of anomalous uw vectors (plotted as streamlines) and anomalous specific humidity (shading) for (a) September–December 2011 and (b) January–April 2012. The data are averaged from 10°S to 10°N and from 75° to 85°E and anomalies are with respect to a 1981–2010 base period. Green/blue shades indicate above-average moisture while yellow/red shades indicate below-average moisture.

Perhaps as expected, Fig. 16 shows that the largest PW occurred near Indonesia (90°–150°E). The IO had its largest values in PW during DP1 and early DP2, when MJO convection was active in this region. This moisture propagated eastward and westward in association with the MJO and ER waves, consistent with a previous climatology (Roundy and Frank 2004). Note, for example, the periods of enhanced moisture (>48 mm) that moved eastward from the western IO with the convective phases of the MJO in October, November, and December.

Figure 17 illustrates a pressure–time cross section (averaged from 10°N–10°S, 75°–85°E) of anomalous uw vectors (plotted in streamline form) and specific humidity (%) for September 2011–April 2012. Figure 17a depicts gradual moistening (i.e., green shades of positive specific humidity anomalies) of the atmospheric column with time from the low levels (900–700 hPa) to just above 400 hPa as the enhanced phase of both the October and November events entered and maximized at the DYNAMO array. Negative moisture anomalies during the suppressed phases of these two MJO events were quite marginal at the surface.

As shown in Fig. 17b, below-average specific humidity and sinking motion are indicated during January in much of the atmosphere. Northeasterly monsoon flow (Fig. 5) dried this region significantly (<42 mm in Fig. 16) during DP2 and this lack of moisture may have played a role in the weakened MJO signal. Moisture anomalies remain highly variable and generally small during the first half of February before positive anomalous specific humidity increased as the enhanced phase of the late February–March event crossed the DYNAMO array. These anomalies, however, are less than the events in 2011 and in fact below-average specific humidity is evident in the middle troposphere in early March 2012. Strong suppressed phases of both a Kelvin and ER wave were coincident over the DYNAMO array at this time and likely contributed to somewhat drier conditions (Fig. 16; Table 2). An extended period of below-average specific humidity and strong sinking air accompanied the MJO suppressed phase the second half of March into April 2012 and were the driest conditions throughout the entire atmosphere during DYNAMO.

4. Atmospheric equatorial Rossby waves

a. ER waves

ER waves are one of the dominant modes of westward-moving convective variability in the tropics. These waves can be comparable in scale to the MJO with periods of 10–40 days and wavelengths up to 15 000 km (Kiladis et al. 2009). The structure and propagation characteristics of convectively coupled ER waves are reminiscent of linear shallow-water solutions (Matsuno 1966). The idealized n = 1 ER waves feature alternating pairs of cyclones and anticyclones that straddle the equator. Cyclonic ER waves can provide a favorable environment for tropical cyclone formation (Molinari et al. 2007), sometimes resulting in “twins” with storms on both sides of the equator (Schreck and Molinari 2009; Gall et al. 2010).

b. ER waves during DYNAMO

Figures 13 and 14 and Table 2 document the ER wave activity during DYNAMO. Figures 13 and 14 provide a broad view of the evolution of these waves and their specific time series over the DYNAMO array, respectively, using both OLR and rainfall. Based on the rainfall time series shown in Fig. 14, Table 2 lists the extrema of ER waves at the DYNAMO array that had amplitude of at least 1.0 standard deviation (SD). The array experienced seven convectively enhanced waves and six convectively suppressed ones. The magenta contours in Fig. 13 show that the ER wave activity was relatively weak during DP1. ER waves became the dominant convective systems during DP2 as the MJO weakened. These waves were also more equatorially symmetric than their earlier counterparts (not shown). As the MJO reintensified in DP3, ER waves continued modulating its convection.

Figure 13 shows the intersection of the convective phases of an ER wave, a Kelvin wave, and the MJO over the DYNAMO array on 27 November. This superposition of modes probably led to the genesis of Tropical Storm Five, which formed in the Arabian Sea near the northern portion of the array (Fig. 18a). However, the identification of the MJO, ER wave, and Kelvin wave is not straightforward in this case. The convection projects most strongly onto the 20–100-day band that we assign to the MJO, but it also retained characteristics of convectively coupled Kelvin waves, similar to the composite Kelvin wave shown in Fig. 6h of Roundy (2008). The convection also projects onto the ER band, and this ER signal can be traced back to the WP around 11 November. However, the enhanced convection is largely confined to the Northern Hemisphere (not shown), suggesting that it might be more akin to a tropical depression-type disturbance (i.e., an easterly wave).

Fig. 18.

Infrared satellite images from Meteosat-7 at (a) 0000 UTC 25 Nov 2011 and (b) 0000 UTC 26 Dec 2011. The gray color scale is proportional to cloud-top temperature (K) with the gray shades depicting warmer cloud tops while the green shades indicate cold cloud-top temperatures and thus deep convection. Specifically, the black colors indicate 330 K, medium gray 290 K, white 245 K, red 225 K, light blue 205 K, dark green 195 K, and white within green 185 K. Courtesy of the DYNAMO Data Catalog.

Fig. 18.

Infrared satellite images from Meteosat-7 at (a) 0000 UTC 25 Nov 2011 and (b) 0000 UTC 26 Dec 2011. The gray color scale is proportional to cloud-top temperature (K) with the gray shades depicting warmer cloud tops while the green shades indicate cold cloud-top temperatures and thus deep convection. Specifically, the black colors indicate 330 K, medium gray 290 K, white 245 K, red 225 K, light blue 205 K, dark green 195 K, and white within green 185 K. Courtesy of the DYNAMO Data Catalog.

DP2 marked a shift in the ER wave regime. As the MJO became less organized, the ER waves amplified and became more equatorially symmetric (not shown). The first of these ER waves reached its convective maximum over the DYNAMO array on 28 December. This wave developed in early December over the WP, but it amplified within the MJO’s convective envelope over the eastern IO. As the MJO and the ER waves constructively interfered, twin Tropical Cyclones, Thane and Benilde, formed over the eastern IO (Fig. 18b). Thane went on to make landfall over eastern India on 30 December (Kruk and Gleason 2012), making it the latest landfalling tropical cyclone ever in that region. ER waves were the dominant mode of convective variability for the remainder of DP2 and the beginning of DP3 and these waves developed near 150°E and took about two weeks to reach the DYNAMO array (Fig. 13).

5. Atmospheric Kelvin waves

a. Convectively coupled Kelvin waves

The Kelvin wave is the leading mode of eastward-moving convection near the equator on time scales between a few days and three weeks (Kiladis et al. 2009). The term “convectively coupled Kelvin wave” refers to observed disturbances that are distinguishable in some respects from their theoretical counterparts (Wheeler et al. 2000; Matsuno 1966). These observed Kelvin waves include the formation of low-level cyclones poleward of the associated moist deep convection and substantial meridional outflow from that convection in the upper troposphere (Roundy 2008). These Kelvin waves tend to be smaller-scale features than the MJO and many of them propagate at more than twice the phase speed of the MJO (Dunkerton and Crum 1995; Wheeler and Kiladis 1999; Straub and Kiladis 2002; Roundy and Frank 2004; Kiladis et al. 2009). When associated with high rainfall rates, these waves gain some structural characteristics of the planetary-scale MJO (Roundy 2012).

b. Convectively coupled Kelvin waves during DYNAMO

Similar to ER waves, Figs. 13 and 14 and Table 2 document the Kelvin wave activity during DYNAMO by equivalent means. One difference, however, is that Kelvin wave extrema in Table 2 are only listed if the amplitude is at least 1.5 standard deviations (SD). By this definition, the array experienced 15 convectively enhanced waves and 11 convectively suppressed ones and the red contours in Fig. 13 show the enhanced convective phases of these moist Kelvin waves during DYNAMO. Three Kelvin wave events exceeded +3.0 SDs and two of these strong enhanced convective Kelvin waves maximized across the DYNAMO array on 24 and 27 November. These “double barrel” Kelvin waves are clearer in rainfall data (Fig. 13b) than OLR (Fig. 13a) where only one Kelvin wave is depicted. The enhanced convective phases of these Kelvin waves occurred within the MJO enhanced convective envelope, but played a large role in focusing enhanced rainfall across the DYNAMO array during the November event.

The third +3.0 SD Kelvin wave event occurred in late February and maximized across the DYNAMO array on 22 February. It was timed with the early portion of the strong enhanced phase of the MJO event of late February through early March. The corresponding strong suppressed phase of this Kelvin wave event (−3.24 SDs; Table 2) crossed the DYNAMO array near 24 February and actually resulted in suppressed convection and rainfall for a period of time within this enhanced MJO phase.

Of the 15 enhanced Kelvin wave events listed in Table 2, 9 occurred collocated with the enhanced phase of the MJO and 7 occurred within the suppressed phase of the MJO. This pattern does not significantly tilt toward previous reports of enhanced numbers of Kelvin waves during the local enhanced convective phase of the MJO. There is also no conclusive evidence that the Kelvin wave amplitudes were any statistically stronger during the local enhanced phase of the MJO as compared to the suppressed phases. Nonetheless, all three of the strongest (+3 SD) Kelvin wave events during DYNAMO occurred during the enhanced phase of the MJO.

6. Subseasonal ocean–atmosphere interactions

Despite reports that subseasonal forecast skill is higher using coupled ocean–atmosphere models as compared to atmosphere-only models (Flatau et al. 1997; Vitart 2004), the importance of ocean–atmosphere coupling for subseasonal variability has yet to be fully quantified. Here we investigate subseasonal variability in SST, OLR, and sea surface height (SSH). The latter is an oceanic variable important to internal ocean dynamics (Rossby and Kelvin waves, horizontal advection) and we explore and its potential role in remotely controlling SST in parallel with mixed layer processes.

The observational datasets we use are microwave weekly SST TMI data (do not require cloud free view) and daily SSH data. Missing SST values are filled by averaging SST from nonmissing neighbor grid points and we then filter observations from 1 January 2010 to 6 October 2012 using a high-pass Lanczos filter with cutoff period at 91 days (13 weeks). Resulting SST data are further analyzed using empirical orthogonal function (EOF) decomposition over the DYNAMO period for revealing horizontal large-scale patterns and possible covariability with the MJO.

Figure 19 compares time–longitude diagrams of subseasonal SST (left panel), OLR (middle panel), and SSH (right panel) averaged between 8°S and the equator (i.e., the southern DYNAMO array latitudes). The longitudes of the array are indicated by thick vertical black lines. Although subseasonal SST variability (left panel) was present during the entire DYNAMO period, stronger, clearer and more coherent subseasonal SST variations started in late January 2012 and coincided with DP3. We note both a stationary SST signal in the IO (horizontal dashed lines) and eastward propagation (dashed arrows) from the eastern IO with an estimated speed of 2 m s−1. The periodicity of SST variations during DP3 was circa 60 days.

Fig. 19.

Time–longitude diagrams of subseasonal variations of (left) SST (°C), (middle) OLR (W m−2), and (right) SSH (cm). All variables are averaged between 8°S and the equator, which were the DYNAMO latitudes. The vertical thick lines show the longitudinal location of the DYNAMO observing system and Manus. Diagonal dashed lines indicate propagating features. Horizontal dashed lines allow for easier comparison of the timing of SST, OLR, and SSH subseasonal variations.

Fig. 19.

Time–longitude diagrams of subseasonal variations of (left) SST (°C), (middle) OLR (W m−2), and (right) SSH (cm). All variables are averaged between 8°S and the equator, which were the DYNAMO latitudes. The vertical thick lines show the longitudinal location of the DYNAMO observing system and Manus. Diagonal dashed lines indicate propagating features. Horizontal dashed lines allow for easier comparison of the timing of SST, OLR, and SSH subseasonal variations.

The MJO events of October and November 2011 (DP1) and of March 2012 (DP3) are clearly depicted on the OLR time–longitude diagram (middle panel). It is interesting to note that during the October MJO event there are no corresponding subseasonal SST fluctuations. For the November MJO event we note a fluctuation in SST in the eastern IO with positive SST leading enhanced convection (negative OLR values) followed by negative SST leading suppressed convection (positive OLR values). During DP3 we observe the same phase-lag relationship between SST and OLR, but in this case we also observe SST fluctuations in the western IO (i.e., eastward propagation of OLR is led by eastward propagation of SST).

Periods of strong subseasonal variability are also evident when examining SSH (Fig. 20, right panel), especially after early December 2011. These subseasonal fluctuations of a periodicity of about 60 days radiated from the western shore of the MC and propagated westward at approximately 1 m s−1 indicative of oceanic equatorial Rossby waves. When considering the covariability of SST and SSH, we note that at the DYNAMO array, maximum cooling (centered in late January 2012) or warming (centered in late February 2012), followed the arrival from the east of, respectively, negative and positive SSH fluctuations. These results corroborate findings of Webber et al. (2010). The mid-March cooling, however, developed prior to the arrival of the next wave from the eastern IO.

Fig. 20.

(top) Principal component 1 (blue) vs principal component 2 (red) derived from EOF analysis of the subseasonal variations of SST. (middle) Comparison of normalized PC1 (blue) with the −RMM1 (MJO active phase over Africa). (bottom) Comparison of the normalized −PC2 (red) with the −RMM2 (MJO active phase over the Indian Ocean).

Fig. 20.

(top) Principal component 1 (blue) vs principal component 2 (red) derived from EOF analysis of the subseasonal variations of SST. (middle) Comparison of normalized PC1 (blue) with the −RMM1 (MJO active phase over Africa). (bottom) Comparison of the normalized −PC2 (red) with the −RMM2 (MJO active phase over the Indian Ocean).

To explore whether the observed SST subseasonal variability was related to the MJO, we performed an EOF analysis of subseasonal SST variations using data from 30°S–30°N, 40°–160°E and during DYNAMO (1 October 2011–31 March 2012). The first two principal components (PCs), explaining 16% and 10% of total subseasonal variance, respectively, are shown Fig. 20 (top panel). Unlike earlier periods of DYNAMO, DP3 is characterized by larger amplitude signals and a clear phase quadrature between PC1 and PC2 suggesting propagating large-scale spatial modes (EOFs). The covariability between the MJO and subseasonal SST modulation is shown by comparing PC1 with −RMM1 (Fig. 20, middle panel), which along with zero RMM2 corresponds to the MJO enhanced phase over Africa. The bottom panel of Fig. 20 shows −PC2 and −RMM2 or the MJO enhanced phase over the IO. The succession of PC1 => −PC2 reflects the propagation suggested by the phase quadrature between the principal components depicted in the top panel of Fig. 20. Again we note a strong dissimilarity between DP3 and the earlier DYNAMO periods. The SST principal components varied more coherently with the RMM indices during DP3 and this suggests some potential for stronger ocean–atmosphere coupling associated with the MJO during DP3 as compared to DP1.

Figure 21 shows from top to bottom the succession of SST EOF1 to −EOF2 to −EOF1 suggested by the phase quadrature of PC1 and PC2, and also the geographical locations of enhanced convective phases of the MJO as defined by the RMM index (left-hand margin). During DP3, when the MJO enhanced phase was over Africa, the entire equatorial IO was warmer than normal with a maximum mode load centered on the southern DYNAMO array latitudes. These above-normal SST fluctuations would likely aid approaching MJO convection to maintain its strength and large-scale organization. When the MJO enhanced convection was over the IO (middle panel), SST anomalies became negative as expected from the heat budget at the surface and steering of the ocean mixed layer. However, there was a tongue of warm water that persists along the southern DYNAMO array latitudes emanating from still warming waters northwest of Australia potentially encouraging MJO convection to propagate eastward. When MJO-enhanced convection reached the MC (bottom panel) the equatorial IO was anomalously cold in its entirety while warm surface waters were present in the WP centered at 20°N and across the South Pacific convergence zone (SPCZ) area.

Fig. 21.

EOFs of subseasonal variations of SST. (from top to bottom) The succession of (EOF1) => (−EOF2) => (−EOF1). The geographical position on the left side of these maps indicates the location of the active phase of the MJO co-occurring with each mode (see text). The amplitude of the SST fluctuations can be obtained by multiplying the EOF with its corresponding principal component.

Fig. 21.

EOFs of subseasonal variations of SST. (from top to bottom) The succession of (EOF1) => (−EOF2) => (−EOF1). The geographical position on the left side of these maps indicates the location of the active phase of the MJO co-occurring with each mode (see text). The amplitude of the SST fluctuations can be obtained by multiplying the EOF with its corresponding principal component.

7. Conclusions

The background state for DYNAMO included the presence of two low-frequency patterns, the IOD during October 2011 and strengthening of La Niña conditions as the campaign progressed. These factors complicated interpretation of observations at times, but they provide additional perspectives for enhancing our understanding of the MJO with and without interaction with low-frequency tropical variability across the Indo-Pacific warm pool region and the role they may play in MJO initiation and maintenance. In addition, comprehensive observations were taken of the ITCZ across the western and central IO during the first half of DYNAMO. This feature frequently appeared during the suppressed phases of the MJO.

The MJO was active during the campaign and showed varying characteristics, which we have used to subdivide the campaign into three periods for discussion here and potential reference in subsequent studies. DP1 began 17 September 2011, just before the start of the official campaign, and lasted until 8 December 2011. It featured two robust MJO events (labeled the October and November events) that circumnavigated the global tropics with a period of less than 45 days. The MJO was less coherent during DP2, which lasted until 31 January 2012. DP2 did feature one “mini-MJO” event in December, the unique details of which are discussed below, but the remainder of DP2 was dominated by ER wave activity. DP3 extended from 1 February 2012 through the end of the campaign and continued until 12 April 2012. The February–March MJO event during DP3 was the strongest of the campaign and also propagated more slowly than the earlier events, which raises interesting questions regarding coupling with the ocean.

As noted above, the December event was particularly interesting and controversial. The event was MJO-like in many ways with respect to dynamics as outlined in section 3. However, it also could be considered a hybrid event as it showed some characteristics of a convectively coupled atmospheric Kelvin wave (i.e., smaller spatial scale and fast propagation speed). The filtering band utilized to extract MJO signals in this study included periods of 20–100 days (Table 1). It is important to note that this range is an arbitrary choice and application of this filter to OLR data placed this event into the MJO category. The cutoff period between the MJO and Kelvin wave filters could have been assigned to the traditional 30 days, which would have classified this event closer to the Kelvin wave band and not the MJO. Assignment of the 20–100-day band for the MJO was a choice made by the authors after viewing several different types of data and in our opinion seeing dynamics consistent with the MJO, even though it lacked some traditional attributes of the MJO. Our comments here are to provide perspective and note issues that complicate such decisions. Moreover, additional techniques and diagnostics may be necessary to best characterize modes of variability if that is deemed appropriate or necessary in future DYNAMO studies.

DYNAMO highlighted or reinforced the need for monitoring several types of MJO indices and data when interpreting the status of the MJO and making forecasts for its evolution. For example, a cornerstone of the campaign was the use of the RMM index for real-time applications to assess the global signal of the MJO. A principal benefit of this index is its simplicity. It can, however, be influenced by both lower- and higher-frequency variability, even with good algorithms in place to minimize these impacts. The MJO monitoring and forecast community carefully interprets the RMM index in the context of additional indices and data to evaluate the details of MJO structure and propagation, but not all users of the RMM index are familiar with such more detailed analysis. Future research studies and real-time users are encouraged to take advantage of this index and use it together with other MJO diagnostics to assess the details of the evolution of individual MJO events.

DYNAMO highlights the need for human forecasters in the MJO monitoring and forecast problem. Interpretation and daily monitoring remains quite important and effective as contributions from the DYNAMO extended-range forecast team added value to official campaign assessments. In several instances, contributions from the team outperformed the model forecasts in their assessment of whether the MJO would remain active, even during times when available forecast tools indicated less coherent or weakening activity. A case in point was the period in mid-October when the enhanced phase of the MJO was crossing the Western Hemisphere. Some model guidance and model-derived forecast tools indicated potential breakdown of the MJO signal as it entered the IO. The MJO remained robust and enhanced convection actually maximized across the DYNAMO array in late October into early November, consistent with the extended-range forecast team assessment thinking.

Kelvin and ER wave activity was substantial during DYNAMO and in many instances augmented convection across the DYNAMO array within the MJO enhanced and suppressed convective phases. Some of these instances were noted in sections 35. Of particular interest are the ER waves that occurred near the initiation of the enhanced convective phases of MJO events. For example, the enhanced ER wave convection that crossed the array on 27 November was associated with enhanced precipitable water (PW) that continued westward, arriving at 40°E around 11 December. The enhanced convective phase of the MJO developed at that time, and the enhanced moisture began moving back to the east with the convection. Similar westward propagation of moisture is observed with each of the MJO events. Future studies should explore whether ER waves play an active role in preconditioning the eastern IO for MJO development. Enhanced convective phases of Kelvin waves also seemed to coincide with onset of the enhanced convective MJO phases, for the October and February–March events. These assertions are qualitative and it is recommended that future research be conducted to better quantify the role of ER and Kelvin waves with respect to MJO initiation.

An important area not discussed in this study is a review of extratropical intrusions into the deep tropics and what role, if any, these interactions may have played in MJO initiation or maintenance during DYNAMO. This is an area that requires considerable attention in forthcoming research, as extratropical–tropical interactions are considered critical to better understanding the MJO initiation problem.

The MJO activity during DYNAMO highlighted some interesting differences with respect to ocean–atmosphere coupling as evidenced by comparing the October–November 2011 and February–March 2012 periods. Analysis of SST variability over the IO and WP indicated subseasonal fluctuations of SST, OLR, and SSH, most notably a well-marked difference in behavior between DP3 and the rest of DYNAMO. Not only did the amplitude of SST fluctuations double during DP3, but ocean temperatures also varied synchronously with the MJO in a way that would most strongly promote eastward propagation of organized convection. These findings could lead one to state that although ocean–atmosphere coupling may not be necessary for the MJO (periods DP1 and DP2), the MJO may become better organized on planetary scales and 40–50-day periods when it is present. The DP3 period may be such a case and ongoing work is further investigating this question in much greater detail.

Acknowledgments

This study was supported by NSF Grant AGS-1062202 and NOAA Climate Program Office (CPO) Earth System Science (ESS) Grants (Zhang, Gottschalck). Schreck received support for this research from NOAA’s Climate Data Record (CDR) Program through the Cooperative Institute for Climate and Satellites (CICS-NC). We are also grateful to Hilawe Semunegus for processing the SSMIS precipitable water data. Real-time analysis and forecast data provided by ECMWF were tremendously helpful during the DYNAMO field campaign. The authors thank Matt Wheeler for dedicated service and very helpful discussions during the DYNAMO campaign. We would like to also thank John Knaff and one anonymous reviewer for careful review of the manuscript and extremely helpful suggestions.

APPENDIX

List of Select Acronyms

ER Equatorial Rossby

IO Indian Ocean

WP Western Pacific

CINDY Cooperative Indian Ocean Experiment on Intraseasonal Variability

AMIE ARM MJO Investigation Experiment

ARM Atmospheric Radiation Measurement

LASP Littoral Air–Sea Processes

MC Maritime Continent

RH Relative humidity

OSCAR Ocean Surface Current Analysis—Real time

SEC South Equatorial Current

RMM Real-time Multivariate MJO

DP1 DYNAMO period 1

DP2 DYNAMO period 2

DP3 DYNAMO period 3

TRMM Tropical Rainfall Measuring Mission

PW Precipitable water

SSMIS Special Sensor Microwave Imager/Sounder

SD Standard deviation

TMI TRMM Microwave Imager

SSH Sea surface height

REFERENCES

REFERENCES
Dunkerton
,
T. J.
, and
F. X.
Crum
,
1995
:
Eastward propagating ~2- to 15-day equatorial convection and its relation to the tropical intraseasonal oscillation
.
J. Geophys. Res.
,
100
(
D12
),
25
781
25
790
.
Flatau
,
M.
,
P. J.
Flatau
,
P.
Phoebus
, and
P.
Niiler
,
1997
:
The feedback between equatorial convection and local radiative and evaporative processes: The implication for intraseasonal oscillations
.
J. Atmos. Sci.
,
54
,
2373
2386
.
Gall
,
J. S.
,
W. M.
Frank
, and
M. C.
Wheeler
,
2010
:
The role of equatorial Rossby waves in tropical cyclogenesis. Part I: Idealized numerical simulations in an initially quiescent background environment
.
Mon. Wea. Rev.
,
138
,
1368
1382
.
Gottschalck
,
J.
,
2011
: A framework for assessing operational model MJO forecasts. Preprints, 23rd Conf. on Climate Variability and Change, Seattle, WA, Amer. Meteor. Soc., 1B.3. [Available online at https://ams.confex.com/ams/91Annual/webprogram/Paper182653.html.]
Gottschalck
,
J.
, and
Coauthors
,
2010
:
A framework for assessing operational Madden-Julian oscillation forecasts: A CLIVAR MJO Working Group Project
.
Bull. Amer. Meteor. Soc.
,
91
,
1247
1258
.
Huffman
,
G. J.
, and
Coauthors
,
2007
:
The TRMM Multisatellite Precipitation Analysis (TMPA): Quasi-global, multiyear, combined-sensor precipitation estimates at fine scales
.
J. Hydrometeor.
,
8
,
38
55
.
Jones
,
C.
,
L.
Carvalho
,
J.
Gottschalck
, and
W.
Higgins
,
2011
:
The Madden–Julian Oscillation and the relative value of deterministic forecasts of extreme precipitation in the contiguous United States
.
J. Climate
,
24
,
2421
2428
.
Kang
,
I.
, and
H.
Kim
,
2010
:
Assessment of MJO predictability for boreal winter with various statistical and dynamical models
.
J. Climate
,
23
,
2368
2378
.
Kiladis
,
G. N.
,
K. H.
Straub
, and
P. T.
Haertel
,
2005
:
Zonal and vertical structure of the Madden–Julian Oscillation
.
J. Atmos. Sci.
,
62
,
2790
2809
.
Kiladis
,
G. N.
,
M. C.
Wheeler
,
P. T.
Haertel
,
K. H.
Straub
, and
P. E.
Roundy
,
2009
:
Convectively coupled equatorial waves
.
Rev. Geophys.
,
47
,
RG2003
,
doi:10.1029/2008RG000266
.
Kim
,
D.
, and
Coauthors
,
2009
:
Application of MJO simulation diagnostics to climate models
.
J. Climate
,
22
,
6413
6436
.
Kruk
,
M. C.
, and
K. L.
Gleason
,
2012
:
[The Tropics] North Indian Ocean
.
Bull. Amer. Meteor. Soc.
,
93
,
S109
S111
.
[in “State of the Climate in 2011.”]
Lin
,
J.-L.
, and
Coauthors
,
2006
:
Tropical intraseasonal variability in 14 IPCC AR4 climate models. Part I: Convective signals
.
J. Climate
,
19
,
2665
2690
.
Madden
,
R.
, and
P.
Julian
,
1971
:
Detection of a 40-50 day oscillation in the zonal wind in the tropical Pacific
.
J. Atmos. Sci.
,
28
,
702
708
.
Madden
,
R.
, and
P.
Julian
,
1972
:
Description of global-scale circulation cells in the tropics with a 40-50 day period
.
J. Atmos. Sci.
,
29
,
1109
1123
.
Matsueda
,
M.
, and
H.
Endo
,
2011
:
Verification of medium-range MJO forecasts with TIGGE
.
Geophys. Res. Lett.
,
38
,
L11801
,
doi:10.1029/2011GL047480
.
Matsuno
,
T.
,
1966
:
Quasi-geostrophic motion in the equatorial area
.
J. Meteor. Soc. Japan
,
44
,
25
43
.
Matthews
,
A.
,
2008
:
Primary and successive events in the Madden–Julian Oscillation
.
Quart. J. Roy. Meteor. Soc.
,
134
,
439
453
,
doi:10.1002/qj.224
.
Molinari
,
J.
,
K.
Lombardo
, and
D.
Vollaro
,
2007
:
Tropical cyclogenesis within an equatorial Rossby wave packet
.
J. Atmos. Sci.
,
64
,
1301
1317
.
Rashid
,
H. A.
,
H. H.
Hendon
,
M. C.
Wheeler
, and
O.
Alves
,
2011
:
Prediction of the Madden-Julian Oscillation with the POAMA dynamical prediction system
.
Climate Dyn.
,
36
,
649
661
.
Roundy
,
P. E.
,
2008
:
Analysis of convectively coupled Kelvin waves in the Indian Ocean MJO
.
J. Atmos. Sci.
,
65
,
1342
1359
.
Roundy
,
P. E.
,
2012
:
The spectrum of convectively coupled Kelvin waves and the Madden–Julian oscillation in regions of low-level easterly and westerly background flow
.
J. Atmos. Sci.
,
69
,
2107
2111
.
Roundy
,
P. E.
, and
W. M.
Frank
,
2004
:
A climatology of waves in the equatorial region
.
J. Atmos. Sci.
,
61
,
2105
2132
.
Roundy
,
P. E.
,
C.
Schreck
, and
M.
Janiga
,
2009
:
Contributions of convectively coupled equatorial Rossby and Kelvin waves to the real-time multivariate MJO indices
.
Mon. Wea. Rev.
,
137
,
469
478
.
Saji
,
N. H.
,
B. N.
Goswami
,
P. N.
Vinayachandran
, and
T.
Yamagata
,
1999
:
A dipole mode in the tropical Indian Ocean
.
Nature
,
401
,
360
363
.
Schreck
,
C. J.
, and
J.
Molinari
,
2009
:
A case study of an outbreak of twin tropical cyclones
.
Mon. Wea. Rev.
,
137
,
863
875
.
Slingo
,
J. M.
, and
Coauthors
,
1996
:
Intraseasonal oscillations in 15 atmospheric general circulation models: Results from an AMIP diagnostic subproject
.
Climate Dyn.
,
12
,
325
357
.
Sperber
,
K. R.
, and
D.
Kim
,
2012
:
Simplified metrics for the identification of the Madden–Julian oscillation in models
.
Atmos. Sci. Lett.
,
13
,
187
193
,
doi:10.1002/asl.378
.
Straub
,
K. H.
, and
G. N.
Kiladis
,
2002
:
Observations of a convectively coupled Kelvin wave in the eastern Pacific ITCZ
.
J. Atmos. Sci.
,
59
,
30
53
.
Tung
,
W. W.
,
J. B.
Gao
,
J.
Hu
, and
L.
Yang
,
2011
:
Recovering chaotic signals in heavy noise environments
.
Phys. Rev. E Stat. Nonlinear Soft Matter Phys.
,
83
,
046210
,
doi:10.1103/PhysRevE.83.038601
.
Vitart
,
F.
,
2004
:
Monthly forecasting at ECMWF
.
Mon. Wea. Rev.
,
132
,
2761
2779
.
Vitart
,
F.
, and
F.
Molteni
,
2010
:
Simulation of the Madden–Julian Oscillation and its teleconnections in the ECMWF forecast system
.
Quart. J. Roy. Meteor. Soc.
,
136
,
842
855
.
Waliser
,
D. E.
,
2011
: Predictability and forecasting. Intraseasonal Variability of the Atmosphere Ocean Climate System, 2nd ed. W. K. M. Lau and D. E. Waliser, Eds., Springer, 433–476.
Webber
,
B. M.
,
A. J.
Matthews
, and
K. J.
Heywood
,
2010
:
A dynamical ocean feedback mechanism for the Madden-Julian Oscillation
.
Quart. J. Roy. Meteor. Soc.
,
136
,
740
754
,
doi:10.1002/qj.604
.
Wheeler
,
M.
, and
G. N.
Kiladis
,
1999
:
Convectively coupled equatorial waves: Analysis of clouds in the wavenumber-frequency domain
.
J. Atmos. Sci.
,
56
,
374
399
.
Wheeler
,
M.
, and
H. H.
Hendon
,
2004
:
An all-season real-time multivariate MJO index: Development of an index for monitoring and prediction
.
Mon. Wea. Rev.
,
132
,
1917
1932
.
Wheeler
,
M.
,
G. N.
Kiladis
, and
P. J.
Webster
,
2000
:
Large-scale dynamical fields associated with convectively coupled equatorial waves
.
J. Atmos. Sci.
,
57
,
613
640
.
Wyrtki
,
K.
,
1973
:
An equatorial jet in the Indian Ocean
.
Science
,
181
,
262
264
.
Yomeyama
,
K.
,
C.
Zhang
, and
C.
Long
,
2013
:
Tracking pulses of the Madden–Julian oscillation
.
Bull. Amer. Meteor. Soc.
, in press.
Zhang
,
C.
,
2013
:
Madden–Julian oscillation: Bridging weather and climate
.
Bull. Amer. Meteor. Soc.
, in press.
Zhang
,
C.
,
D. S.
Nolan
,
C. D.
Thorncroft
, and
H.
Nguyen
,
2008
:
Shallow meridional circulations in the tropical atmosphere
.
J. Climate
,
21
,
3453
3470
.

Footnotes

This article is included in the DYNAMO/CINDY/AMIE/LASP: Processes, Dynamics, and Prediction of MJO Initiation special collection.