1. Introduction
During summer 1997, the National Oceanic and Atmospheric Administration (NOAA) ship Ronald H. Brown conducted a research cruise in the tropical eastern Pacific to explore the properties of convective clouds in the intertropical convergence zone (ITCZ). This project was called the Tropical Eastern Pacific Process Study (TEPPS; Yuter and Houze 2000). One of the primary objectives of TEPPS was to understand differences between convection in the eastern and western Pacific tropical basin. From satellite data, the convective clouds over the eastern Pacific seem to be shallower but produce more rain (Spencer 1993). One of the first steps in understanding the convection over the eastern Pacific as compared to that over the western Pacific is to determine the synoptic-scale environment in which the convection occurs. This paper seeks to describe and understand the synoptic environment of TEPPS as a background for more specific studies of the clouds and precipitation observed during the cruise.
The intertropical convergence zone occurs over both the eastern-to-central Atlantic and Pacific oceans as a mean band of persistent cloudiness and precipitation located between approximately 10°S and 10°N, depending on the season. The ITCZ structure and location vary strongly, with the magnitude and location of the band of cloudiness varying on submonthly to interannual timescales. Mitchell and Wallace (1992) found that the annual solar cycle controls the position of the eastern Pacific ITCZ, which is south of the equator from approximately December through April, and north of the equator for the remainder of the year. The annual maximum in precipitation for this region occurs during August and September and is centered at approximately 10°N. At this time, the eastern Pacific equatorial cold tongue in the sea surface temperature (SST) is well-established, while the maximum SST and precipitation is north of the equator. The north–south surface pressure gradient over the equatorial ocean forces large-scale southerly winds across the southern edge of the ITCZ. Strong surface southerly flow is enhanced by the monsoonal land–sea pressure gradient between the cold tongue and the low-pressure region over Central and South America.
The ITCZ structure fluctuates strongly as synoptic-scale conditions vary. This paper is concerned primarily with identifying the synoptic-scale variability in the vicinity of the ship in TEPPS. Observations from sounding stations in the equatorial central and western Pacific during the Northern Hemisphere summer show westward-propagating disturbances, referred to as “easterly waves” or “TD waves” (tropical depression waves), which have 3–6-day periods and wavelengths on the order of 2000–3000 km (Dunn 1940; Riehl 1948; Wallace and Chang 1969; Reed and Recker 1971). Similar disturbances occur over the tropical Atlantic and Caribbean in connection with waves traced to an origin over North Africa (Simpson et al. 1968; Carlson 1969; Wallace and Chang 1969; Burpee 1972; Reed et al. 1977; Riehl 1979, 342–353). Satellite data support the results found from synoptic data by indicating westward-propagating cloud patterns in both the equatorial Pacific (Chang 1970; Nitta et al. 1985; Nitta and Takayabu 1985; Tai and Ogura 1987; Takayabu and Nitta 1993) and Atlantic (Simpson et al. 1968; Frank and Johnson 1969; Nitta et al. 1985; Nitta and Takayabu 1985; Reed et al. 1988a,b; Avila and Clark 1989; Molinari et al. 1997) during boreal summer. The westward-propagating cloud patterns have been described as “inverted Vs” because they appear as open circulations in the visible channel data (Simpson et al. 1968). They are sometimes referred to as TD waves because a subset of these westward-propagating disturbances lead to tropical cyclogenesis in the Caribbean and eastern Pacific (e.g., Reed et al. 1988a,b; Avila and Clark 1989; Raymond et al. 1998; Molinari et al. 1997, 2000) and typhoons in the western Pacific (Riehl 1948; Nitta et al. 1985; Nitta and Takayabu 1985). Ferreira and Schubert (1997) suggested that barotropic instability of the ITCZ might play a role in tropical cyclogenesis. Molinari et al. (2000) suggested that easterly waves in the eastern Pacific may originate over Africa and rejuvenate from barotropic instability as they pass through the Caribbean and eastern Pacific regions.
In addition to easterly (TD) waves, which form as a result of instabilities of the large-scale wind field, the ITCZ region is populated by near-equatorial trapped waves (Kelvin waves, MRG waves, inertio-gravity waves; Holton 1992, 392–398). In addition, tropical storms develop on the poleward side of the ITCZ and generally move out of the region in a northwesterly direction. The objective of this study is to determine the mix of wave and disturbance types that defined the synoptic-scale environment of the ship in TEPPS. The study capitalizes on four special datasets: 1) surface, sounding, and radar data taken on the Ronald H. Brown during TEPPS, 2) visible, water vapor, and infrared channel data from the NOAA Geostationary Operational Environmental Satellite-9 (GOES-9), 3) data from the Woods Hole Oceanographic Institution (WHOI) Improved Meteorological Instrumentation (IMET) buoy near the ship, and 4) the NOAA Tropical Atmosphere–Ocean (TAO) buoy array in the eastern tropical Pacific. We will show that these data point to easterly waves as accounting for most of the synoptic variability during TEPPS but that at one point the easterly waves interacted with a Kelvin wave (Straub and Kiladis 2002), which substantially modified the convective ensemble.
2. Data
The NOAA ship Ronald H. Brown was located as part of TEPPS in the eastern tropical Pacific Ocean at 7.8°N, 125°W from 8–23 August 1997 (Fig. 1). The ship was equipped with a scanning, stabilized C-band Doppler radar; multiple rainfall sensors; surface meteorological sensors to measure air temperature, humidity, wind speed, and direction; and a thermosalinograph to measure 2-m salinities and ocean temperatures. The ship was also equipped with a Vaisala rawinsonde system. Sondes were launched six times per day while on station and measured pressure, air temperature, relative humidity, and GPS-determined wind speed and direction at a raw vertical resolution of about 30 m. An overview of TEPPS and a more detailed description of the dataset can be found in Yuter and Houze (2000).
The radar returns are quantitatively useful within a 100–150 km radius of the ship. At greater distances the volume of the beam is too large to resolve precipitation regions in fine detail or to detect light rain. A radar convective index was defined by Yuter and Houze (2000) as the largest horizontal scale of rain area present within 100 km of the ship and consists of five categories of horizontal scale: 0–2, 2–10, 10–50, and 50–100 km. An index is assigned every hour to classify the lowest elevation-angle scan. These categories serve as a measure of the convective activity in the vicinity of the ship.
NOAA's GOES-9 satellite data provide an overview of the deep convection over the eastern Pacific. These data have 4-km spatial resolution and 3-hourly time resolution. The subset of GOES-9 data used in this study are from July–August 1997 and extend from 75°–150°W, 0°–20°N, and were interpolated to a 1.5° regular horizontal grid for minimal smoothing of the raw data.
The WHOI IMET buoy located at 10°N, 125°W was deployed in April 1997 and recovered in December 1998, providing 18 months of surface meteorological data at hourly time resolution. The NOAA TAO moored buoy array (McPhaden et al. 1998) also provides hourly surface meteorological data at several locations throughout the tropical eastern Pacific including 8°N, 125°W. Records from these buoys included in this study date from 1996–1998. Surface winds from these buoys and from the WHOI IMET buoy collected at roughly 4 m and 3.5 m above mean sea level, respectively, are used to extend the ship surface observations to a broader region within the eastern Pacific ITCZ. Figure 1 shows the IR data horizontal grid spacing and the locations of the IMET and TAO buoys used in this study.
3. Large-scale context for ship data
a. Wind and SST
Equatorial eastern Pacific Florida State University (FSU) wind stress vectors at 2° resolution (Legler and O'Brien 1985) and Reynold's SSTs at 1° resolution (Reynolds 1988) are shown in Fig. 2 for July–September 1997. The wind pattern is in general agreement with that described by Mitchell and Wallace (1992) for the eastern Pacific boreal summer ITCZ region. Convergence is evident over the warmest surface waters marking the ITCZ at approximately 8°N, east of 145°W, and dipping southward west of this longitude. The strength of the easterlies increases to the west as the ITCZ dips to the south. The cold tongue in SST is also evident along the equator from the coast of South America. Figure 3 shows the twenty-fifth percentile GOES-9 IR temperatures for July–August 1997 (September 1997 not available). Twenty-five percent of the IR temperature values at a given point lie below this value. The twenty-fifth percentile temperatures do not smooth the region of high cloud as much as the mean, yet they do not overemphasize the anomalies. The center latitude of the ITCZ is coincident with the surface convergence evident in Fig. 2. The north–south extent of this region ranges from 2.5°–5°N in the far western portion of the domain, and from 2.5°–15°N near the coast of Central and South America. This pattern is not a simple function of SST. Colder cloud tops occur over the 30°C water off the coast of Panama than over the 30°C water west of 2.5°N, 140°W. Evidently, the nearby land mass affects IR temperature near the coast.
The location of the ITCZ slightly south of its long-term climatological position of 10°N and the weak cold tongue are the consequences of the strong El Niño conditions in 1997. The ITCZ tends to move equatorward during these conditions because of the southward shift of the SST maximum, located north of the cold tongue, and the shift of the low-level convergence associated with the SST gradient. Figure 2 indicates that the ship resided in the middle of the warm water in the center of the ITCZ. The ship position was intentionally adjusted southward from the original experimental plan (made before it was clear that it was an El Niño year) in order to compensate for the effect of the El Niño on the location of the ITCZ. It is our contention that the structure of convection was qualitatively normal, and that by having the ship correctly positioned relative to the ITCZ, we sampled convection generally representative of the ITCZ.
b. Variability in the cold cloud cover: Satellite data
In order to understand the temporal and spatial development of the convective events within the ITCZ during July–August 1997, we have reviewed the looped satellite images for evidence of propagating features and/or periodicity throughout the tropical northeast Pacific. Traditionally, this type of analysis has used visible imagery, since the visible images best show the lines of low-level clouds forming inverted Vs defining synoptic-scale wave troughs (e.g., Simpson et al. 1968; Chang 1970). Deep convection often appears along the trough line, the trough being a favorable environment for the development of mesoscale systems. The deep systems tend to initiate along the low-cloud line defining the inverted V. Figure 4 shows the albedo from the GOES-9 visible channel at 0000 UTC for 5–28 August 1997, identifies the inverted V structures seen during this period, and includes the ITCZ portion of the TEPPS cruise, which collected data at 8°N, 125°W from 8–23 August 1997.
Tropical meteorologists have sometimes debated whether quasi-periodic signatures in synoptic-scale tropical wind data are indicative of true waves in the easterlies or simply signatures of tropical cyclones passing nearby at intervals of a few days (Simpson et al. 1968). This question is acute in the TEPPS region where tropical cyclones are frequent during late summer. Figure 5 shows the tracks of all the eastern Pacific tropical cyclones in 1997. Our analysis indicates that there is a close connection between some of these storms and the easterly waves passing over the ship in TEPPS. However, the waves at the ship do not appear to be an aliasing of the cyclones onto the ship data. Rather, the cyclones appear to be initiated by the waves.
We have identified the inverted V patterns in the visible satellite data by inspection of the images at 3-h intervals. Because of space limitation in this article, Fig. 4 contains only a single image for each day. (The reader may view the complete set of satellite images at http://www.atmos.washington.edu/gcg/MG/teppsloops.)
On 5 August, three inverted Vs were evident. The vertex of the inverted V is used to identify the structure. Inverted V3 spawned Hurricane Guillermo, which began as a tropical depression on 30 July and became a hurricane on 1 August (storm number 7 in Fig. 5). The convection associated with Guillermo was connected with what appears to have been a second inverted V, near 12.5°N, 142.5°W, or ∼3000 km from V3 and labeled V2. The feature farthest to the west labeled V1 at 10°N, 170°W, or ∼3000 km from V2, was the least pronounced.
These features were evident until 8 August (except V1, which could not be identified in the 0000 UTC 6 August visible image), moving westward across the basin. About 3000 km remained between V2 and Guillermo; however, as V1 and V2 reached the central Pacific, the distance between them decreased to about 2000 km. Early on this day the Brown arrived on station at 7.8°N, 125°W, southeast of Guillermo, located at approximately 17°N, 135°W. Guillermo had spun off to the northwest, leaving a gap in the V3 cloud stream at 125°W.
By 9 August, V1 was no longer evident and the distance between V2 and V3 had increased to 3500 km. The V3 feature was visible with a gap at the top of the inverted V until 12 or 13 August. In the meantime, V2—at 10°N, 170°W on 9 August—had become more organized. By 11 August, convection developed somewhat east along the trough line suggesting eastward as well as westward propagation of cloud features. The reformed inverted V is labeled V2a and was ∼3000 km from V3 at this time. On 13 August, V2a was at 175°W, and a new inverted V labeled V2b had appeared around 160°W. By 14 August, V2a and V2b were no longer distinct, and one active region is seen at 175°W, labeled V2ab. The formation of V2ab is coincident with the arrival of the Kelvin wave at 180° (Straub and Kiladis 2002), and is likely the result of interference between the inverted V and Kelvin-wave convective envelopes. The V2ab feature was last evident on 16 August at 180°, at which time convective development also continued eastward, between the equator and 5°N, evidently in association with the Kelvin wave.
On 9 August, a fourth region of convection to the east of the gap left by Guillermo was evident near 10°N, 112.5°W, labeled V4. This feature became more pronounced by 12 August, at 10°N, 115°W as tropical depression Hilda (storm number 8 in Fig. 5), which transitioned into a hurricane on 13 August. Like Hurricane Guillermo, Hilda spun off to the northwest, leaving a gap in the cloud stream across the region. The gap from Guillermo is still evident near 142.5°W, ∼3000 km west of Hilda at this time. Finally, on 17 August Hilda moved north of 25°N. However, convection at V4, the structure that spun off Hilda, had developed again on 14 August as the feature moved west, reaching 142.5°W, on 17 August. At this time, the convective envelope from the westward-propagating disturbance intersected with that of the eastward-propagating Kelvin wave. The V4 feature tracked westward until 23 August.
In the far eastern portion of the domain yet another convective region (V5) appeared around 12°N, 110°W on 14 August. By 17 August, this feature became tropical storm Ignacio (storm number 9 in Fig. 5). On 15 August, Hilda, at 130°W, and Ignacio, at 112.5°W, were ∼2000 km apart. By 18 August, Ignacio moved north of 25°W at 120°W. On 19 August, the Kelvin-wave envelope intersected with V5 around 125°W. The V5 feature tracked westward until 23 August.
The movement of the V4 and V5 features is illustrated in more detail by water-vapor channel imagery in Fig. 6. The water vapor imagery shows clearly the inverted Vs of the synoptic-scale waves V4 and V5 moving westward. Looping through water-vapor images shows that the pattern associated with V4 appears similar to a breaking wave, curling over at 15°N, 145°W by 18 August. In addition, the Kelvin-wave convective envelope can be seen to have moved eastward between the equator and 5°N; during the period of Fig. 6, deep convection moved eastward across the trough line of V4.
A new convective feature (V6) was only barely visible on 18 August around 10°N, 112.5°W, becoming more visible on 19 August (Fig. 4). The final evidence of the Kelvin-wave convective envelope is seen on 20 August at around 120°W, where it intersected V6. Unlike other features that had developed in this region up to this time, V6 never spun off a hurricane. It tracked westward across the eastern Pacific until at least 28 August. The feature became part of a string of inverted V structures most evident at 2100 UTC 26 August (the image for 0000 UTC 27 August was not available, so 2100 UTC 26 August is used in its place). Within this string were also V7, appearing for the first time on 25 August, and V8, appearing for the first time at 2100 UTC 26 August. Feature V7 became Hurricane Jimena (storm number 10 in Fig. 5) on 25 August, at 12°N, 130°W (see 2100 UTC 26 August, 10°N, 135°W).
This series of images suggests a pattern throughout August 1997. Areas of convection within the context of inverted V structures, initially appeared in the eastern portion of the eastern Pacific and moved westward. Several of the identified features spun hurricanes off the northern apex of the inverted V, while others never reached that intensity. The westward motion of these features was on the order of 7.6–8.9 m s–1 (7°–8° day–1), with distances between the tips of the inverted V structures being about 2000–3000 km. Shorter wavelengths occurred in the western portion of the domain, where the trades were stronger (Fig. 2). As these easterly waves progressed westward across the domain, four of them (V2ab, V4, V5, and V6) intersected the eastward-propagating Kelvin wave described by Straub and Kiladis (2002).
4. Ship observations
a. Time–height analysis of ship soundings
The ship observations provide detailed information about the thermodynamic and wind structure of the atmosphere from 8–23 August 1997 in the center of the ITCZ. Figure 7 shows the time series of meridional and zonal wind anomalies, relative humidity anomalies, and potential temperature anomalies from 4-hourly sounding data collected at the ship while on station at 7.8°N, 125°W. The data shown in the figure have been fitted to a 50-m vertical resolution and a 6-h time resolution for minimal smoothing of the raw data. Anomalies are calculated with respect to the 8–23 August time period.
Figure 7a shows three periods of enhanced southerlies in the lower troposphere; the first from 0900 UTC 10 August to 0900 UTC 11 August (event 1), the second from 0900 to 1500 UTC 14 August (event 2), and the third from 2100 UTC 18 August to 0900 UTC 19 August (event 3). These time periods had sustained low-level meridional wind anomalies greater than one standard deviation (2.8 m s–1). For this purpose, we defined the low-level meridional wind anomaly with respect to the meridional wind vertically averaged between the surface and 700 mb. A weaker event began on 21 August, but the anomalies did not attain magnitudes greater than 2.8 m s–1.
The mean profiles of meridional and zonal wind during TEPPS, as well as the wind anomaly profiles for the three southerly events, are shown in Figs. 8a and 8b. The mean shear between the upper-level and low-level winds during TEPPS is shown in Fig. 8c, where the low-level winds have been averaged from the surface to 700 mb (υL, uL), and the upper-level winds have been averaged from 600 to 300 mb (υU, uU). The wind anomaly profiles were similar in all cases, except for the zonal wind at low levels for the third southerly event (Fig. 8b, dotted line). In general, wind anomalies were from the southwest at low levels and from the northeast at upper levels. The third event is the exception, with low-level wind anomalies being primarily from the south to southeast. Further examination indicates that the southerly anomalies were coincident with maximum easterly shear between the upper and lower levels in the zonal wind and maximum northerly shear in the meridional wind (Fig. 8c).
If it is assumed that the wind anomalies were propagating westward, then the tilt with height of the TEPPS meridional wind anomaly maxima is westward for the first and second events, and eastward for the third event (Fig. 7a). Referring to the mean zonal wind profile in Fig. 8b, the westward tilt in the meridional wind axis for the first and second events corresponds to an easterly shear. Holton (1971) found that the easterly wave axis in his model tilted westward with height in the lower troposphere for a zonal wind profile with easterly shear from the surface to approximately 300 mb. The tilt in the TEPPS meridional wind axis for the first and second events is consistent with Holton's (1971) model of easterly wave characteristics in the presence of vertical shear in the mean zonal wind. The tilt of the meridional wind axis for the third event is more consistent with Holton's case of westerly shear in the mean zonal wind, and therefore is not consistent with the model results. This particular event was probably complicated by the superposition of the easterly waves with the Kelvin wave (Straub and Kiladis 2002). If the wind anomalies for the third event were propagating eastward, as would be the case for a Kelvin wave, then the tilt of the zonal wind anomalies with height would be westward (Fig. 7b). This result would be consistent with the vertical structure of a Kelvin wave as found by Wheeler et al. (2000). Finally, although the event of 22 August did not have strong surface southerlies, the tilt of the meridional wind axis (Fig. 7a) is westward, as was seen for the first two events, and it coincides with V6 from Fig. 4.
Relative humidity anomalies (Fig. 7c) also had a periodicity of roughly four days, with the positive anomalies above the 0°C contour slightly lagging the lower-tropospheric anomalies and the surface southerlies. The magnitude of the lower-tropospheric anomalies in relative humidity averaged from the surface to 700 mb associated with the 4-day variability was approximately 7%–10%, corresponding to moisture fluctuations of about 1–2 g kg–1. Upper-level anomalies averaged from 600–300 mb, ranged from 10%–30%, corresponding to moisture fluctuations of 0.5–1.0 g kg–1.
Chang et al. (1970) suggest that latent heat released near 300 mb serves to enhance horizontal temperature gradients at this level, creating a source of available potential energy for the wave. While the potential temperature anomalies (Fig. 7d) did not have a 4-day signal as clear as that in the meridional wind or humidity fields, warm anomalies are evident between 200 and 400 mb during the first and third events, and between 350 and 450 mb during the second event. The warm cores are coincident with positive moisture anomalies, consistent with the Chang et al. (1970) proposed energetics of these waves. The warm anomaly aloft is also consistent with the tendency of mesoscale convective systems to produce a maximum of net heating in the upper troposphere (Houze 1982, 1989). The synoptic-scale wave-induced temperature anomalies are estimated by dry dynamics to be <1°C (Holton 1971). The TEPPS potential temperature anomalies shown by the 4-hourly TEPPS soundings (Fig. 7d) were as much as 2°C preceding and during the low-level southerly events. These anomalies likely include the effects of both the synoptic-scale and mesoscale convective heating.
Easterly waves are also often observed to have cold cores in the lower troposphere (Chang et al. 1970). The potential temperature anomalies in Fig. 7d indicate negative anomalies below the warm cores for all three events. The third event differs in that a second positive anomaly is observed near the surface. The cold anomalies could be seen as a negative feedback on the waves, but calculations by Nitta (1970) find that the upper-level creation of available potential energy is enough to overcome any negative feedback from these cold cores.
b. Spectral analysis of ship soundings
The variance as a function of frequency (power spectral density) for the low-level and upper-level meridional winds [P(υL), P(υU)], and low-level and upper-level humidities [P(rL), P(rU)] are shown in Figs. 9a and 9b, respectively. Because our time series is short compared to the time period of variability of interest, the Maximum Entropy Method (MEM) was employed for calculating these spectra, based on the methods of Ghil et al. (2001, hereafter Gh 01). The peaks in the spectra shown here compare well with the Blackman–Tukey correlogram and the Multitaper Method spectra, also useful for short time series spectral analysis (Gh 01). The power spectral density has been multiplied by frequency and plotted as a function of the log of the period so that the areas under all portions of the curve are proportional to the variance. The low-level meridional wind had a strong signal, significant at the 99% confidence level, at the 4-day period (Fig. 9a), as expected from the time series in Fig. 7a. No significant variance is observed in the zonal wind at lower or upper levels during this period (data not shown). The lack of power in the zonal wind for easterly wave-type disturbances is noted by Chang et al. (1970), who also noted the lack of any significant coherence between the zonal wind and other variables at times when easterly waves were present in the data. The low-level moisture had a strong signal at about 4.5 days, while a much smaller peak is observed at upper levels for this variable (Fig. 9b). Both peaks are significant at the 95% confidence level. The smaller signal at upper levels is likely the result of the small amount of total moisture present from 600–300 mb. The potential temperature spectrum revealed no significant peaks at low or upper levels (data not shown).
Figure 9c shows the time-lagged autocovariance of the low-level meridional wind [C(υL, υL)], and cross covariance of the low-level meridional wind with low-level and upper-level moisture [C(υL, rL), C(υL, rU)]. The four-day periodicity in these variables is evidenced by the time difference between the peaks in the autocovariance and cross covariances. The cross covariance between low-level moisture and meridional wind peaks at approximately 4- and 8-day lags, with a smaller peak at zero lag. The coincidence of these peaks with those of the autocovariance of the meridional wind indicates the low-level moisture was nearly in phase with the southerlies. The peak at zero lag is smaller than those at 4- and 8-day lags, possibly because the low-level moisture leads the southerlies by about half a day or so, shifting the maximum peak to slightly negative lags. The cross covariance between the upper-level moisture and the low-level meridional wind peaks at about 0.5-, 4.5-, and 8.5-day lags. This implies that the upper-level moisture lags the southerlies by about a half a day. The cross-covariances confirm the relationships observed in Fig. 7 between the southerlies and the low-level and upper-level humidity anomalies.
c. Clouds and precipitation in the vicinity of the ship
Figure 10 shows the radar convective index (section 2) along with the IR temperature from the grid closest to the ship. The satellite and radar data agree well, with both datasets indicating three periods of sustained convective activity coincident with the southerly events at the ship. The second event was weaker and did not last as long as the first and third. A fourth event was beginning on 21 August.
To indicate the zonal scale and progression of convective systems in the vicinity of the ship, the percent area within 240 km of the ship with reflectivities greater than 20 dBZ was calculated from the radar reflectivity data. Reflectivities from the surveillance scans were interpolated to a 4 × 4 km grid, extending 480 km in both the north–south and east–west directions (ship at the center of the 480-km box). The percentage of 4 × 4 km grids in the north–south direction with reflectivity values greater than the specified threshold was determined for each longitude. The result of this calculation is the percent area coverage of precipitation associated with mesoscale systems and smaller cloud elements as a function of longitude and time, with a 4-km resolution in the east–west direction and roughly 15 min in the time domain. Figure 11 shows the percent area of precipitation in a time–longitude plot (Fig. 11a), along with 6-hourly wind vectors at 300 mb during TEPPS (Fig. 11b). The ship's location is along the center line of the horizontal axis in Fig. 11a. The dates on the vertical axis are at 0000 UTC in both plots. Unlike the convective index, the percent area coverage does not provide information on the scale of contiguous cloud area, but rather provides the percent of cloud echoes greater than 20 dBZ in intensity. The three major convective events seen in Fig. 11a occurred on approximately 10–11, 14–15, and 18–19 August. These events contained small-scale precipitation areas moving to the west, as evidenced by the tilt in the dark areas from east to west. The speed of these convective elements was approximately 8.3 m s–1 (black line), consistent with the speeds estimated from the IR data for the inverted V structures. However, the motion of these small convective rain areas must also have been strongly affected by advection and gust front propagation. The 300-mb wind vectors (Fig. 11b) indicate that upper-level easterlies were present for 14–15 and 18–19 August and were about 7–10 m s–1, similar to the speed of the precipitation areas.
Also evident for 18–20 August is an eastward-moving envelope of convection, within which the individual lines of convection move westward. Development of mesoscale convective systems to the east at this time was shown to be on the trough line of the westward-moving synoptic-scale wave (section 3b). However, the eastward-moving envelope was likely associated with the Kelvin wave present in the region at this time (Straub and Kiladis 2002).
5. Seasonal distribution of the synoptic-scale disturbances over the whole tropical eastern Pacific
a. Buoy observations
Surface winds associated with the synoptic variability in cloud cover have been investigated using data from buoys (TAO and WHOI IMET). Figure 12 shows time series of the meridional wind for each of the TAO buoys between 110°W and 180° at 8°N for July–August 1997. Also shown are the IMET meridional winds at 10°N, 125°W overlaid on the TAO data at 8°N, 125°W. Except for the beginning of the time period, the TAO and IMET datasets are nearly indistinguishable.
In general, the meridional wind varied on subweekly timescales at all locations throughout July and August. Time periods when inverted V structures passed over the buoys (Fig. 4) are identified in Fig. 12. The implied speed for these features is about 3.8 m s–1 in the central Pacific, and about 8.9 m s–1 in the eastern Pacific. The slower phase speed in the central Pacific is coincident with shorter wavelengths in this region based on the visible satellite imagery. The sensitivity of easterly wave characteristics to local conditions is well documented (e.g., Chang 1970; Reed and Recker 1971; Takayabu and Nitta 1993). The differences in wave characteristics seen in the buoy data across the eastern Pacific are therefore not inconsistent with easterly waves studied previously and are highly consistent with the coherent westward-moving features seen in the visible and water-vapor channel satellite imagery (Figs. 4 and 6).
Variability on 3–6-day timescales was present throughout the equatorial eastern Pacific basin in July–August 1997. Similar scales of variability occurred in the 0°, 2°, and 5°N buoy time series, with the 3–6-day variability being most evident off the equator (data not shown).
b. Spectral analysis: Seasonal and areal distribution of disturbances over the eastern Pacific
In order to determine the dominant periods of variability in cloud cover, the satellite IR, rather than visible data, are used, as they provide information over an entire 24-h period. Figure 13a shows the mean IR power spectral density [P(IR)] for grids centered on 125.25°W at latitudes 0.75°, 2.25°, 5.25°, 8.25°, and 9.75°N for August 1997. As with the ship data, the IR power spectra were calculated using the Maximum Entropy Method (Gh 01) and multiplied by frequency for presentation in Fig. 13a. The most energetic peak is observed at 8.25°N at approximately the 5-day period, consistent with visual inspection of the data in Fig. 4. Peaks at this period are significant at the 99% confidence level for the 5°–10°N spectra, with no significant power at this period for the 0° or 2°N spectra. These spectra may be compared to the IMET and TAO buoy meridional wind MEM spectra [P(υIMET), P(υTAO)] for August 1997 along 125°W, and latitudes 10°, 8°, 5°, and 2°N (data at 0° not available; Fig. 13b). Spectral peaks significant at the 99% confidence level were observed at approximately a 5-day period in the 10°N and 8°N spectra. As with the IR spectra, power at this period decreases towards the equator.
The off-equatorial peak in the IR and meridional wind spectra at the 5-day period is most consistent with an easterly (TD) wave. Mixed Rossby–gravity (MRG) waves (wavelength ∼8000 km, phase speed ∼15–20 m s–1; see Liebmann and Hendon 1990; Takayabu and Nitta 1993; Wheeler et al. 2000) have periods similar to easterly waves, making them difficult to distinguish from easterly waves using data at a single location. As MRG waves have their maximum variance in the meridional wind on the equator and are typically symmetric about the equator (Liebmann and Hendon 1990; Takayabu and Nitta 1993), these waves can be distinguished from easterly waves given some information about the north–south distribution of the meridional wind variance. We have calculated the power spectra for the TAO buoy meridional wind south of the equator, at 8°, 5°, and 2°S, for August 1997. These spectra had little power (<0.5 m2 s–2) at 3–6-day periods, consistent with the notion that we are detecting an easterly wave rather than an MRG wave (data not shown).
It is possible that MRG waves were present but decoupled from lower-tropospheric winds due to stable boundary layers over cold waters in the eastern Pacific south of the SST front at about 1.5°N. TAO monthly averaged sea surface temperatures indicate that ocean surface temperatures were about 1.5°C colder at 8°S than at 8°N during August 1997. However, because 1997 was an El Niño year, surface temperatures at 8°S for August 1997 were 27.5°C, 1.5°C warmer than climatology. Nevertheless, a decoupling of the surface layer and the lower-tropospheric winds cannot be discounted.
The time-lagged cross covariance of the IR data from 8.25°N, 125.25°W and the buoy meridional wind at 8°N, 125°W [C(IR, υTAO)] is shown in Fig. 13c, along with the cross covariance of the IR data at the same location with the ship low-level meridional wind [C(IR, υL,Brown)]. The calculations were done such that a covariance peak at positive lag indicates southerlies lead IR maxima (fair weather). The negative covariance at zero lag indicates that the southerlies at the buoy are coincident with a minimum in the IR data (cloudiness). The period of variability for the IR-buoy wind covariance is approximately 5 days, as seen by the difference in time between consecutive covariance peaks. Similarly, the covariance of the IR and ship low-level meridional wind [C(IR, υL,Brown)] indicates the southerlies are coincident with an IR minimum (cloudiness), with a period of variability of about four and a half days (Fig. 13c). This relationship is consistent with the cross covariance of low-level meridional wind with upper-level moisture at the ship [C(υL, rU); Fig. 9c], which indicated that the upper-level moisture lagged the southerlies by less than a day.
The presence of easterly waves in the ITCZ in the eastern Pacific is reflected in the spatial distribution of 3–6-day variability of the satellite IR data. The spatial distribution of the 75th percentile 3–6-day IR spectral energy density is shown in Fig. 14a for July–August 1997 (September data was not available). The 75th percentile values are chosen, rather than the mean or maximum values, as they do not smooth or emphasize anomalies in the distribution. The spectra were multiplied by frequency when constructing this figure. Comparison of this spectral distribution to the 25th percentile IR temperature distribution (Fig. 3) indicates that the convection off the coast of Central America (7.5°N, 85°W) consists of high 3–6-day variability. The Central American coast is also the region of enhanced barotropic instability found by Molinari et al. (2000). Other regions of high 3–6-day activity are seen to be contained within the band of ITCZ convection across the eastern Pacific (5°–15°N, 150°–110°W). Areas of this activity form bands or streaks extending toward the west-northwest from the ITCZ region (for example see 12°N, 113°W to 17°N, 125°W pattern in Fig. 14a). This streakiness probably arises from hurricanes spinning off the inverted Vs of easterly waves toward the west-northwest in this general region (Fig. 4; e.g., Guillermo on 5 August).
The cloud-top pattern in Fig. 14a is closely related to the variability of the surface wind field indicated by buoys. The spatial distribution of the mean 3–6-day meridional wind spectral energy density from the TAO and IMET moorings is shown in Figs. 14b–e for all seasons during 1997. As with the IR data, the spectra were multiplied by frequency when constructing these figures. When a continuous 3-month period was not available from the TAO or IMET buoys, that period was replaced with one from 1998 or 1996. Here, it is assumed that seasonal variability of the synoptic waves is of significantly larger amplitude than any variability imposed on the atmosphere on these timescales by the 1997–1998 El Niño event. Some limited 3–6-day activity is seen in the eastern portion of the buoy domain during January–March (Fig. 14b), a time when the ITCZ is least active in the tropical northeastern Pacific. The least amount of 3–6-day activity occurred during April–June (Fig. 14c), when the ITCZ was still developing in the eastern Pacific. The meridional wind spectral energy peaked in the eastern Pacific in July–September (Fig. 14d). The maximum activity shifted to the central Pacific in October–December (Fig. 14e).
While the buoy observations are on a coarser grid scale than the IR data, it is nonetheless instructive to compare the late summer 3–6-day wind variability (Fig. 14d) with the late summer 3–6-day IR variability (Fig. 14a) over the eastern Pacific. The meridional wind spectral energy over this region was most prominent during the late summer, when the ITCZ cloud pattern was also most active. The spectral energy was maximum at the northern edge of the buoy domain with a rapid decrease southward, a pattern more consistent with easterly waves than with equatorial trapped waves. This result is consistent with the streakiness in the satellite data, noted above, which was evidently related to easterly waves spinning off tropical cyclones. The generally west–east band of maximum meridional wind variability at the northern limit of the buoy array in July–September, tilts southwestward toward the equator in the central Pacific, following the pattern of cold cloud tops seen in Fig. 3, and the 3–6-day activity in cold cloudiness seen in Fig. 14a. The general consistency between the ITCZ and the variance in the meridional wind on 3–6-day timescales is further evidence of easterly wave activity. Easterly waves are more tightly coupled to convection than are equatorial trapped MRG waves (Takayabu and Nitta 1993; Dunkerton and Baldwin 1995).
c. Cross-spectral analysis: Propagation characteristics
To determine the zonal propagation velocity of the 3–6-day disturbances, we employ time-lagged cross covariances. Figure 15a shows the lagged covariance for the IR data at 8.25°N, 125.25°W with all other longitudes at this latitude for August 1997. The eastward-propagating Kelvin wave (Straub and Kiladis 2002) dominated the IR signature at this location, although there is also some hint of westward-propagating disturbances superimposed on the eastward disturbance. The phase speed of the eastward disturbance is about 12.7 m s–1 (white line, Fig. 15a). As seen in Fig. 4, easterly wave convection varies a great deal with location, as the undulating chain of low clouds and moisture defining the inverted Vs meanders north and south. To see the westward disturbances of interest we select a location where the 3–6-day activity is a maximum, and where the Kelvin wave is likely to have a weak signal. The IR data at 11.25°N, 117.75°W meets these criteria, as there is a maximum in the 3–6-day activity at this location, and it is north of 10°N, which appears to be the northern limit of the Kelvin-wave disturbance from Fig. 4. A central location on the domain is also chosen to give the best cross covariances across the basin. Figure 15b shows the time-lagged cross covariance of the IR data at 11.25°N, 117.75°W with all other grids at this latitude. The most prominent feature is coherent for about 30° of longitude and has an apparent westward phase velocity cp of 6.4 m s–1 (white line). Figure 15c shows a similar analysis, but for the 3–6-day band only. In this case the features are seen to have a period of about 5 days and wavelengths on the order of 2200 km.
Figure 16a shows time-lagged covariances for TAO meridional wind data at 8°N, 125°W with all available longitudes, also at 8°N, for August 1997. While these lagged covariances alone cannot be used to determine phase relationships or wavelengths, the information is used to estimate these quantities based on the results of the IR analysis and the 3–6-day variability found in the meridional winds at these locations. For instance, the peak at ∼1.8 days in the 155°W curve (bold dashed line, Fig. 16a) indicates that a southerly event at 155°W lags that at 125°W by this time period, or by 0.45 wave periods for a 4-day disturbance. A possible interpretation would be disturbances with 0.45 wavelengths between these two locations. Such an assumption would imply a wavelength of about 7330 km and a 21.2 m s–1 phase speed. These values are consistent with an MRG wave. A second interpretation is that there are 1.45 wavelengths between these two locations, which implies a wavelength of 2360 km and a phase speed of 6.8 m s–1. These values are more consistent with the IR data and to values in the literature for easterly waves. This type of analysis is also applied to the cross covariances at 170°W and 180° and is summarized in Table 1. The first column in Table 1 notes the data source, followed by the period of the covariance series T determined from the difference in time between the most prominent covariance minima or maxima. Using the period and distance between the two buoy locations and all possible fractional wavelengths between locations N, the implied wavelength λ and phase speed cp are calculated. The fractional wavelength, wavelength of the disturbance, and phase speed of the disturbance are listed in the last three columns of Table 1. Only those fractional wavelengths between buoy locations that gave reasonable results are listed in the table. Analysis of 8°N, 125°W with 8°N, 170°W is inconclusive, as the period of variability appears to be 6 days based on a covariance peak at 6.5 days and minimum at about 3.25 days, but no peak is observed at 0.5 days. Also listed are the results of this type of analysis for the data at 5°N (Fig. 16b), as well as the results from analysis of all other data used in this study for comparison.
The longer wavelengths and faster phase speeds in Table 1 can be compared to MRG wave characteristics, which typically have wavelengths on the order of 8000 km and phase speeds of about 15–20 m s–1 (Liebmann and Hendon 1990; Takayabu and Nitta 1993; Wheeler et al. 2000). The shorter wavelengths and slower phase speeds in Table 1 (1890–2520 km with phase speeds of 4.4–7.3 m s–1) are more consistent with easterly waves. The results for all TAO cross covariances in Table 1 are consistent with easterly waves, while only one (8°N, 125° and 155°W) cross covariance indicates characteristics of an MRG wave. While either wave type is plausible, the persistent easterly wave-type characteristics compared to the intermittent appearance of MRG wave characteristics in Table 1 suggests that the easterly waves are the more likely explanation of the 3–6-day spectral peaks in buoy winds. Moreover, visible and water-vapor channel satellite imagery in Figs. 4 and 6 suggest that the actual wavelengths were 2000–3000 km and that the waves propagated at easterly wave speeds.
6. Conclusions
This study documents the passage of easterly waves in the eastern Pacific ITCZ during August 1997. The disturbances moved westward with wavelengths on the order of 2000–3000 km, periods of 3–6 days, and phase speeds of 4–9 m s–1. The maxima of the surface-to-700 mb meridional wind anomalies were 5–12 m s–1. Maxima of the moisture anomalies at the same levels were ∼1–3 g kg–1, in phase with the low-level southerlies, while upper-level moisture lagged the southerlies by less than a day. The vertical structure of the apparent synoptic-scale wave had a reversal in both the meridional and zonal wind directions between 850 mb and 450 mb during the low-level southerly phase of the disturbance. The meridional and zonal components had a maximum 850–450 mb wind shear during the southerly events. The observed tilts of the wave axis for the given wind shear profiles during the first and second events and the weaker fourth event were westward, consistent with those expected from theoretical studies of easterly waves (Holton 1971). The third event tilted eastward, possibly because at that time the region around the ship was under the influence of both a Kelvin wave (as shown by Straub and Kiladis 2002) and the easterly wave V5. Moisture profiles and a convective radar index indicate a close connection between the southerlies and convective activity over the ship. The coupling between convection and the wave supports the notion that the waves derive their energy from the conversion of available potential energy established by latent heating near 300 mb, to kinetic energy of the wave.
Meridional wind data from the IMET buoy and TAO array show the spatial as well as temporal characteristics of these disturbances in the near-surface winds. The buoy data not only detect these disturbances at the surface across the tropical eastern Pacific basin, but also identify regions and seasons of greater and lesser activity. The regions of heightened activity in the wind field were coincident with those observed in the satellite IR data, providing observational evidence for a link between convection and 3–6-day disturbances of the wind field over a large spatial domain. Furthermore, the disturbances seen in the spectrally analyzed satellite and buoy data exhibited easterly wave characteristics by decreasing in intensity toward the equator in the buoy wind data and having a signature of waves spinning off tropical cyclones in the satellite IR data.
Variability in the lower-tropospheric wind components on synoptic timescales have been noted in numerous studies of tropical dynamics. Disturbances on 3–6-day timescales in the lower troposphere, like those discussed in this study, are generally of two types: the first being equatorial-trapped MRG waves with a maximum signature in the meridional wind on the equator and the second being easterly waves, which tend to follow the axis of the ITCZ in the Northern Hemisphere and sometimes spin off tropical cyclones. The data used for this study suggest that easterly waves, rather than MRG waves, were observed in the eastern Pacific ITCZ during the TEPPS study in August 1997. This conclusion is supported primarily by the off-equatorial maxima in 3–6-day variance observed in both the wind and IR data. In addition, loops of the visible and water-vapor channel imagery show classic inverted V structure and apparent open circulations traveling across the eastern Pacific basin at the latitude of the ITCZ on these timescales, sometimes spinning off hurricanes. Not only do these images suggest off-equatorial phenomena, but also confirm a well-established relationship between easterly waves and hurricanes in both the Atlantic and eastern Pacific.
This study has established the large-scale context for the TEPPS ship observations. Ongoing and future studies of the detailed ship observations in TEPPS can now be placed into a synoptic context consisting of a family of easterly waves passing through the eastern Pacific ITCZ throughout August 1997. In addition to the easterly waves, an equatorial Kelvin wave propagated eastward through the region (Straub and Kiladis 2002). Time-lagged covariance expressed as a function of latitude and longitude (Fig. 15) clearly distinguishes the Kelvin- and easterly wave contributions to the synoptic-scale variability at the ship's location. The Kelvin wave approaching from the west created an interference pattern with the easterly waves, and in future studies the observations of clouds, precipitation, wind, and thermodynamic conditions in the vicinity of the ship in TEPPS must be interpreted in terms of the synoptic-scale interference pattern composed of the Kelvin and easterly waves.
Acknowledgments
The sounding and radar data used in this study were collected aboard the NOAA ship Ronald H. Brown, under the direction of Sandra Yuter. Without the help of Richard Reed we could not have done the satellite data analysis in section 3. We benefited from discussions with Adam Sobel, Robert Tomas, and Matthew Wheeler. Reviews by Katherine Straub, Harry Hendon, and George Kiladis provided constructive criticism and insight, which strengthened the final draft considerably. Candace Gudmundson edited the manuscript and the Edit–Design Center refined the figures. David Ovens provided the satellite loops for the website. The TAO buoy data were provided by the Pacific Marine Environmental Laboratory's Tropical Ocean Atmosphere (TAO) Project. Data from the mooring at 10°N, 12°S was provided by R. Weller and S. Anderson from the Woods Hole Oceanographic Institution, who were funded by the NOAA Office of Global Programs, Contract Number NA96GPO428. This work was supported by JISAO cooperative agreement No. NA67RJ0155.
REFERENCES
Avila, L. A. and G. B. Clark, 1989: Atlantic tropical systems of 1988. Mon. Wea. Rev., 117:2260–2265.
Burpee, R. W., 1972: The origin and structure of easterly waves in the lower troposphere in North Africa. J. Atmos. Sci., 29:77–90.
Carlson, T. N., 1969: Synoptic histories of three African disturbances that developed into Atlantic hurricanes. Mon. Wea. Rev., 97:256–276.
Chang, C-P., 1970: Westward-propagating cloud patterns in the tropical Pacific as seen from time-composite photographs. J. Atmos. Sci., 27:133–138.
Chang, C-P., V. F. Morris, and J. M. Wallace, 1970: A statistical study of easterly waves in the western Pacific: July–December 1964. J. Atmos. Sci., 27:195–201.
Dunkerton, T. J. and M. P. Baldwin, 1995: Observation of 3–6-day meridional wind oscillations over the tropical Pacific, 1973–1992: Horizontal structure and propagation. J. Atmos. Sci., 52:1585–1601.
Dunn, G. E., 1940: Cyclogenesis in the tropical Atlantic. Bull. Amer. Meteor. Soc., 21:215–229.
Ferreira, R. N. and W. H. Schubert, 1997: Barotropic aspects of ITCZ breakdown. J. Atmos. Sci., 54:261–285.
Frank, N. L. and H. M. Johnson, 1969: Vortical cloud systems over the tropical Atlantic during the 1967 hurricane season. Mon. Wea. Rev., 97:124–129.
Ghil, M., Coauthors,. 2001: Advanced spectral methods for climate time series. Rev. Geophys., in press.
Holton, J. R., 1971: A diagnostic model for equatorial wave disturbances: The role of vertical shear of the mean zonal wind. J. Atmos. Sci., 28:55–64.
Holton, J. R., . 1992: An Introduction to Dynamic Meteorology. Academic Press, 511 pp.
Houze, R. A. Jr,, 1982: Cloud clusters and large-scale vertical motions in the tropics. J. Meteor. Soc. Japan, 60:396–410.
Houze, R. A. Jr,, . 1989: Observed structure of mesoscale convective systems and implications for large-scale heating. Quart. J. Roy. Meteor. Soc., 115:424–461.
Legler, D. M. and J. J. O'Brien, 1985: Atlas of Tropical Pacific Wind-Stress Climatology: 1971–1980. Florida State University Press, 187 pp.
Liebmann, B. and H. H. Hendon, 1990: Synoptic-scale disturbances near the equator. J. Atmos. Sci., 47:1463–1479.
McPhaden, M. J., Coauthors,. 1998: The Tropical Ocean–Global Atmosphere observing system: A decade of progress. J. Geophys. Res., 103:14169–14240.
Mitchell, T. P. and J. M. Wallace, 1992: The annual cycle of equatorial convection and sea surface temperature. J. Climate, 5:1140–1156.
Molinari, J., D. Knight, M. Dickinson, D. Vollaro, and S. Skubis, 1997: Potential vorticity, easterly waves, and eastern Pacific tropical cyclogenesis. Mon. Wea. Rev., 125:2699–2708.
Molinari, J., D. Vollaro, S. Skubis, and M. Dickinson, 2000: Origins and mechanisms of eastern Pacific tropical cyclogenesis: A case study. Mon. Wea. Rev., 128:125–139.
Nitta, T., 1970: A study of generation and conversion of eddy available potential energy in the tropics. J. Meteor. Soc. Japan, 48:524–528.
Nitta, T. and Y. Takayabu, 1985: Global analysis of the lower tropospheric disturbances in the tropics during the northern summer of the FGGE year. Part II: Regional characteristics of the disturbances. Pure Appl. Geophys., 123:272–292.
Nitta, T., Y. Nakagomi, Y. Suzuki, N. Hasegawa, and A. Kadokura, 1985: Global analysis of the lower-tropospheric disturbances in the tropics during the northern summer of the FGGE year. Part I: Global features of the disturbances. J. Meteor. Soc. Japan, 63:1–19.
Raymond, D. J., C. Lopez-Carrillo, and L. L. Cavazos, 1998: Case studies of the developing east Pacific easterly waves. Quart. J. Roy. Meteor. Soc., 124:2005–2034.
Reed, R. J. and E. E. Recker, 1971: Structure and properties of synoptic-scale wave disturbances in the equatorial western Pacific. J. Atmos. Sci., 28:1170–1133.
Reed, R. J., D. C. Norquist, and E. E. Recker, 1977: Structures and properties of African wave disturbances as observed during phase III GATE. Mon. Wea. Rev., 105:317–333.
Reed, R. J., A. Hollingsworth, W. A. Heckley, and F. Delsol, 1988a: An evaluation of the performance of the ECMWF operational system in analyzing and forecasting easterly wave disturbances over Africa and the tropical Atlantic. Mon. Wea. Rev., 116:824–865.
Reed, R. J., E. Klinker, and A. Hollingsworth, 1988b: The structure and characteristics of African easterly wave disturbances as determined from the ECMWF operational analysis/forecast system. Meteor. Atmos. Phys., 38:22–33.
Reynolds, R. W., 1988: A real-time global sea surface temperature analysis. J. Climate, 1:75–86.
Riehl, H., 1948: On the formation of typhoons. J. Meteor., 5:247–264.
Riehl, H., . 1979: Climate and Weather in the Tropics. Academic Press, 611 pp.
Simpson, R. H., N. Frank, D. Shideler, and H. M. Johnson, 1968: Atlantic tropical disturbances, 1967. Mon. Wea. Rev., 96:251–259.
Spencer, R. W., 1993: Global oceanic precipitation from the MSU during 1979–91 and comparison to other climatologies. J. Climate, 6:1301–1326.
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.
Tai, K-S. and Y. Ogura, 1987: An observational study of easterly waves over the eastern Pacific in the northern summer using FGGE data. J. Atmos. Sci., 44:339–361.
Takayabu, Y. N. and T. Nitta, 1993: Three-to-five-day period disturbances coupled with convection over the tropical Pacific Ocean. J. Meteor. Soc. Japan, 71:221–246.
Wallace, J. M. and C. P. Chang, 1969: Spectrum analysis of large-scale wave disturbances in the tropical lower troposphere. J. Atmos. Sci., 26:1010–1025.
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.
Yuter, S. E. and R. A. Houze Jr., 2000: The 1997 Pan American Climate Studies Tropical Eastern Pacific Process Study. Part I: ITCZ region. Bull. Amer. Meteor. Soc., 81:451–481.
Wave characteristics of 3–6-day features estimated from the datasets listed in the first column. Here, 200=1T is the dominant period of variability observed in each dataset. Parameter N indicates the number of wavelengths between the two TAO buoy locations used to estimate wavelength λ and phase speed cp for the TAO data. See text for details
Joint Institute for the Study of the Atmosphere and Ocean Contribution Number 823.