Observations of the Structures of Deep Convections and ... - J-Stage

Journal of the Meteorological Society of Japan, Vol. 84, No. 1, pp. 115--128, 2006

115

Observations of the Structures of Deep Convections and Their Environment during the Active Phase of an Madden-Julian Oscillation Event over the Equatorial Western Pacific Hisayuki KUBOTA, Ryuichi SHIROOKA, Tomoki USHIYAMA, Jingyang CHEN1, Takashi CHUDA, Kensuke TAKEUCHI2, Kunio YONEYAMA, and Masaki KATSUMATA Institute of Observational Research for Global Change, Japan Agency for Marine-Earth Science and Technology, Yokosuka, Japan (Manuscript received 12 October 2002, in final form 16 October 2005)

Abstract The structures of deep convections and their environment were examined during the latter half of the active phase of the Madden-Julian oscillation (MJO), using fine temporal resolution of research vessel observation data. During the intensive observation period, the MJO associated with deep convections passed through the observation area. A strong westerly wind also appeared during the active phase of the MJO. Vertical structures associated with convective activity were detected using lag correlations of atmospheric parameters. Convective activity included deep convections, with a nocturnal maximum and another peak 12 to 18 hours earlier. Active convections associated with the first peak developed in the afternoon and decayed within a few hours. Those active convections transported water vapor to the lower to middle troposphere, making the environment favorable for the more intensive deep convections that developed during the night. Westerly winds were intensified in the lower layers during the deep convections.

1.

Introduction

Tropical convection plays a significant role in global climate and general circulation through latent heat release and radiative processes. Deep convections in the warm water pool region have broad temporal and spatial scales. The most prominent features are the intraseasonal oscillations, which have a time scale of 30 to 60 days. Madden and Julian (1971, 1972, 1994) first characterized the intraseasonal osCorresponding author: Hisayuki Kubota, Institute of Observational Research for Global Change, Japan Agency for Marine-Earth Science and Technology, 2-15, Natsushima-Cho, Yokosukacity, Kanagama, 237-0061, Japan. E-mail: [email protected] 1 Present affiliation: Fudan University, Shanghai, China. 2 Present affiliation: National Fisheries University, Shimonoseki, Japan. ( 2006, Meteorological Society of Japan

cillation, using rawinsonde and sea-level pressure [Madden-Julian oscillation (MJO)]. Nakazawa (1988) presented a hierarchical structure of intraseasonal oscillations over the western Pacific. Several super-cloud clusters near the equator, with a horizontal scale of several thousand kilometers and an eastward propagation were identified. Within a supercloud cluster, several westward-propagating cloud clusters were found, having a horizontal scale of several hundred kilometers and a timescale of 1 to 2 days. Westward propagating convective activity is strongly connected with the quasi-two-day wave, as found by Takayabu et al. (1996). Chen and Houze (1997) explained these westward propagating two-day disturbances, using a combination of diurnal varying cloud-radiation interaction and inertia-gravity waves. Diurnal variations of convection were observed for cases with heavy rainfall, and for

116

Journal of the Meteorological Society of Japan

strong convection in which the maximum appeared in the early morning hours (Chen et al. 1996; Sui et al. 1997; Kubota and Nitta 2001). Lau et al. (1989) related the eastwardpropagating super-cloud clusters and westward-propagating cloud clusters, to equatorially trapped Kelvin and Rossby wave modes, respectively. The Kelvin and Rossby modes combine to produce a zonally asymmetric circulation. Over the western Pacific, strong nearsurface westerly winds are observed in association with paired Northern and Southern hemisphere cyclones, that form within a few days and several degrees of longitude of each other, on opposite sides of the equator (Keen 1982). These short-lived episodes of strong westerly winds are called ‘‘westerly wind bursts.’’ Westerly wind bursts are associated with deep convections and occur more frequently in the active phase of the MJO (Nitta and Motoki 1987). These also have an impact of triggering El Nin˜o events (McPhaden 1999). In this study, the latter half of the active phase of the MJO was captured during the intensive observation period (IOP) of 27 November to 11 December in 2000, with the research vessel (R / V) Mirai, and we observed strong westerly wind in lower layers. With low-level westerly winds, the convective systems moved eastward (Wu and LeMone 1999). Halverson et al. (1999) analyzed eastward-propagating convective systems, and showed that the stratiform rainfall fraction tended to spread during the nighttime. They suggested that moisture around the middle troposphere plays some role for increasing the stratiform rainfall area. Kubota and Nitta (2001) and Kubota et al. (2004) suggested that moisture transport from the lower layers was important for the development of deep convections during the nighttime. Our purpose is to study the structures of deep convection and their environment, by examining vertical moisture transport associated with deep convections during the active phase of the MJO. The Frontier Observational Research System for Global Change implemented observational project PALAU (Pacific Area Long-term Atmospheric observation for the Understanding of climate change) 2000 from November to December 2000. The R / V Mirai made a fine temporal resolution of stationary observations during the active phase of the

Vol. 84, No. 1

MJO accompanied by westerly wind bursts. In this paper, satellite data and in situ observations are used to clarify the relation of largescale disturbances, deep convections and their environment. Section 2 describes observations and satellite data. Section 3 consists of four subsections. First, the broad-scale features are described by analyzing satellite data. Second, observational data from the R / V Mirai are used to examine the vertical structure of deep convection and westerly wind bursts. Third, the atmospheric distributions in the life cycle of convective activity are investigated, using lag correlations between the fractional rainfall area and atmospheric parameters. Fourth, the life cycle and its environment are examined and discussed, with a focus on the convective activity that developed around 4 to 6 December over the observation area. Moisture transport associated with deep convections and the twoday oscillations of convective activities, are discussed in a later section. 2.

Meteorological instruments and observations

The R / V Mirai performed stationary observations from 27 November to 11 December 2000 at 2 N, 138 E. This period is called an Intensive Observation Period (IOP). Radiosondes launched every three hours during this period, were used to observe the vertical structure of the troposphere. Continuous radar observations detected the intensity of rainfall every 10 minutes. Hourly infrared equivalent blackbody temperature (TBB) data derived from the Japan geostationary meteorological satellite (GMS), were used to determine the behavior of convective activity during the IOP. The spatial resolution of GMS data achieved by Kochi university is 0.05 degrees. The microwave scatterometer QuikSCAT data were also used to obtain the surface wind over the ocean during the IOP. QuikSCAT scans about twice a day, and its spatial resolution is 0.25 degrees. The horizontal scale of rainfall clouds was classified, using a threshold of 15 dBZ reflectivity within the 200-km range hourly PPI (planposition indicator) at an elevation of 0.7 degree. NCEP/NCAR reanalysis 6 hourly 850 hPa height wind data, were used to compensate the spatial wind field during the IOP with 2.5 degrees resolution.

February 2006

H. KUBOTA et al.

117

Fig. 2. As in Fig. 1, except for the surface wind vectors from QuikSCAT data averaged around 2 N to 2 S. Fig. 1. Time-longitude diagram of convective activity averaged around GMS data for 5 N to 5 S from 16 November to 15 December 2000. The solid line indicates the intensive observation period over 138 E. The contour intervals are 20 K.

3.

Results

3.1

Overview during the intensive observation period During the IOP, an MJO associated with westerly wind bursts and deep convection passed through the observation area. Figure 1 depicts a time-longitude diagram of convective activity along the equator from the GMS data. An area of strong convective activity propagated eastward from the Indian Ocean in the middle of November, reaching the observation area on 25 November. The active phase of the MJO continued over the observation area until 5 December. The center of the large-scale convection was observed to the east of the stationary observation region. During the IOP, indicated with a solid line, two strong signals of convection passed through the observation area, one on 28–29 November and another on 4–5 December. During this active phase of the MJO, an area with a large value of precipitable water also propagated eastward associated with the convective signals (not shown). Figure 2 presents a time-longitude diagram of surface winds from the QuikSCAT data.

Fig. 3. One-day-average of GMS TBB data and surface wind vectors from QuikSCAT data on 28 November 2000. TBB data are represented by shades, and their intervals are 20 K. White circles indicate 200-km radar range centered at 2 N, 138 E.

Westerly winds exceeding 10 m/s were observed from 26 November in the observation region before the intensive observation period. A tropical cyclone became Typhoon 0022 on 28 November at 8.7 N, 131.0 E. Westerly wind extended to 155 E where convergence occurred. This convergence was to the east of the center of the large-scale convection (Fig. 1). Figure 3 depicts the one-day-average spatial feature of

118

Journal of the Meteorological Society of Japan

Vol. 84, No. 1

the typhoon and strong westerly winds on 28 November. Clouds are represented by GMS TBB data and surface winds by the QuikSCAT data. Counterclockwise circulation appears around the typhoon near 8.7 N, 131.0 E. Strong westerly winds are observed along the equator reaching 152 E in zonal, and spread away from the equator from 6 N to 9 S in meridional including the observation area. The structures of the typhoon and westerly wind bursts correspond to a Matsuno-Gill pattern (Matsuno 1966; Gill 1980). The typhoon moved westward and its circulation caused strong westerly winds in the observation area from 26 to 29 November. Another tropical depression moved westward along 10 N, and caused strong westerly winds in the observation area from 2 to 5 December. These results demonstrate that convection became active from 25 November to 5 December in the observation area, and that strong westerly winds (westerly wind bursts) also appeared during this period. 3.2

Deep convection and westerly wind bursts observed by a research vessel The R / V Mirai observed the active phase of the MJO during the intensive observation period. Figure 4a illustrates the time-series of the fractional rainfall area, using a threshold of Mirai radar reflective intensity of 15 dBZ within the 200 km range PPI. When the active phase of the MJO passed through, convection became active and the fractional rainfall area was relatively large until 5 December. The rainfall echo fraction was used as an index of convective activity over the observation region. Convection was especially strong around 28 to 29 November and around 4 to 5 December, consistent with Fig. 1. Mainly three peaks of fractional rainfall area appeared at 2000 UTC on 28 November, at 1600 UTC on 30 November, and at 1900 UTC on 4 December. Each peak has another small peak about twelve to 18 hours before the main peak of the fractional rainfall area. After the first peak in rainfall clouds decayed or moved away, a second region of rainfall clouds developed or passed through the observation area the following night (this is discussed in more detail in Section 3.4). After the active phase of the MJO has passed over the observation region, rainfall clouds almost disappeared on 6 to 7 December. The rainfall

Fig. 4. A time series of the fractional rainfall area in Mirai radar using the threshold of reflective intensity of 15 dBZ within 200 km range PPI (a). Numbers of horizontal scales of rainfall clouds less than 100 km 2 (b), 100 to 1000 km 2 and greater than 1000 km 2 (c) from 26 November to 12 December 2000. The units are % (a). The horizontal scales are less than 100 km 2 (dotted line) (b), 100 to 1000 km 2 (dashed line), and greater than 1000 km 2 (thick solid line) (c).

echo fraction gradually increased again after 8 December. The horizontal scale of rainfall clouds was classified into three levels using radar data: less than 100 km 2 , 100 to 1000 km 2 , and over 1000 km 2 . Figures 4b and c present a time series of observed numbers of the three types of rainfall clouds. Rainfall clouds with areas of less than 100 km 2 appeared in 86% of the total clusters numbers during the intensive observa-

February 2006

H. KUBOTA et al.

Fig. 5. As in Fig. 4, except for (a) the vertical profile of relative humidity and (b) zonal wind from 27 November to 11 December 2000. The contour intervals are 5% (only layers with relative humidity exceeding 70% are drawn) (a) and 5 m/s (b). Missing data are left blank. Data are missing on 0300Z on 5 December and 1200Z on 6 December under relative humidity (a).

tion period. This frequency of small-scale rainfall clouds exceeds the 67% in GATE observations over the Atlantic (Houze and Cheng 1977). The frequency of 100 to 1000 km 2 was 12%. The frequency of over 1000 km 2 was 2%. During the active phase of the MJO until 5 December, the smallest-scale rainfall clouds had two peaks, around 27 to 29 November and 3 to 5 December. Rainfall clouds were not present from 0700 to 1000 UTC on 6 December. After 8 December, the number of rainfall clouds increased again, from the smallest scale rainfall clouds. Figure 5 shows a time-series of vertical profiles of relative humidity and zonal winds ob-

119

tained from radiosonde observations. During the first half of the IOP, a moist layer associated with the active phase of the MJO extended to 11 km. Moist layers persisted for about three days in the upper troposphere around 5 to 10 km, after strong convections were observed on 28 November and 4 December. Moistening at high levels indicates high-level clouds, i.e., anvil spreading. For the 28 November case, upper level moist layers originating from deep convection, were observed from 28 to 30 November. After the deep convection developed on 4 December, the anvil spread at upper levels, and lower-to-middle levels above 1 km became dry; convection was then suppressed. After 8 December, low levels below 2 km were moistened, and then convection was regenerated (see Fig. 4). The evolution of low-level relative humidity after 5 December indicates that moistening around 1 to 2 km may be one of the key layers for subsequent development of convection. Westerly wind bursts were observed in the observation region on 26 November, before the IOP (Fig. 2). Westerly winds exceeding 10 m/s were observed below 6 km, especially around 1 to 4 km, during the active phase of the MJO (Fig. 5b). The cores of the strong westerly winds appeared around 27 to 29 November, 1 to 3 December and 4 to 5 December. The maximum wind speed was 23.1 m/s near 1.6 km altitude at 2100 UTC on 28 November. The cores of strong winds appear to be associated with deep convection. Deep convection was suppressed until 7 December, and westerly winds in the lower layers were weakened. After 8 December, convection gradually activated again, and easterly winds were strengthened at upper levels above 8 km; vertical wind shear did not decrease, however. 3.3

Vertical atmospheric structures associated with deep convections Lag correlations between the fractional rainfall area and atmospheric parameters were calculated. The focus was on atmospheric structures associated with deep convections during the active phase of the MJO, so the analysis period was selected as from 1200 UTC on 26 November to 0000 UTC on 6 December. The focus was not on individual clouds observed within the 200-km radius radar range, but on the ag-

120

Journal of the Meteorological Society of Japan

gregate of deep convections with scales of up to 400 km. It was difficult to observe the life cycle of convective activity by radar, because the convection propagated eastward with the environmental flow. Furthermore, convective activity had different lengths of life cycle. The aim of this subsection is to describe a composite life cycle of convective activity during the active phase of the MJO. The radar-echo area is used as an index of convective activity. The rainfall echo fraction increases when convections pass through the radar coverage. We defined the peak of echo fraction as a peak of convective activity. We focus on the lag correlation of atmospheric parameters against the radar-echo area, which shows the vertical atmospheric structures during the life cycle of the convective activities. Figure 6a illustrates the lag correlations between the water vapor mixing ratio and the radar-echo area. At the peak of echo fractions, the highest positive correlation was found around 4 km. This high positive correlation started 18 hours prior to the peak, at a height of around 2 to 4 km, and spread upward with time. Kubota and Nitta (2001) suggested that the low-level moisture increase prior to deep convection, plays an important role in the development of nocturnal convection. Another high positive correlation appeared at 5 to 7 km, nine hours prior to the peak. The lowest negative correlation peak was found below 1 km at lower layers three hours prior to the peak, which indicates that the dry air descended due to the downdraft. After the peak, the region of positive correlation gradually ascended, reaching 6 km at 18 hours after the peak. Another positive correlation appeared above 8 km after the peak. The water vapor mixing ratio increases at low levels prior to the peak of echo fractions and at high levels after the peak, were similar to what was observed associated with convective activities of two-day oscillations, as observed during the TOGA COARE (Takayabu et al. 1996). Radiosondes were launched every six hours during TOGA COARE, whereas we launched every three hours during the intensive observation period of this study, and thus finer temporal structures of moisture distribution were observed. Moisture increases about 18 hours before the peak of echo fraction. The peak of echo fraction

Vol. 84, No. 1

appears about 3 hours later than convective echo fraction peak (discussed in subsection 3.4), due to anvil stratiform rain spread wide in the radar range. Therefore our results demonstrated a high correlation of moisture, spread upward before the peak of convection. Figure 6b depicts the lag correlations of temperature against the echo fractions. Negative correlations existed in lower layers around 1 km centered about three hours prior to the peak, consistent with a downdraft and drying (Fig. 6a). A positive correlation (warming) appeared in the middle troposphere around 3 to 7 km until 12 hours prior to the peak, and a negative correlation (cooling) appeared at 4 km after the peak. Figure 6c shows the lag correlation of zonal wind against the echo fractions. During the active phase of the MJO, westerly winds prevailed below 6 km, and a strong westerly wind of greater than 10 m/s was observed below 4 km (Fig. 5b). A high positive correlation appeared around 1 km from 12 hours prior to the peak to 6 hours after the peak. The westerly wind was simultaneous accelerated with convections that were activated. When two major peaks of rainfall clouds (2000 UTC on 28 November, and 1900 UTC on 4 December) appeared, westerly wind was also accelerated (Fig. 5b). However, for the case at 1600 UTC on 30 November, the relation was not apparent. In Fig. 6c, about twelve hours after the peak, a high positive correlation appeared around 7 to 9 km. Easterly wind was usually dominant at these levels during the intensive observation period, but it switched briefly to westerly winds after the major peaks (Fig. 5b). The TBB data indicate cloud top heights, if clouds are present below the satellite, and are used as an index of convective activity. We constructed histograms, with 0:05  0:05 grid mean TBB data to investigate the vertical structure of convective activity over the 200  200-km grid, corresponding to the radar echo range (3:6  3:6 ) centered at 2 N, 138 E. The histogram which consists of smaller-grid numbers is binned to every 5 K, from 180 K to 300 K in the vertical (see subsection 2.1 in Kubota and Nitta 2001). If only high-level clouds cover an area, all TBB values are low. In reality, clear regions and several levels of clouds may be present in one area. We can use TBB

February 2006

H. KUBOTA et al.

121

Fig. 6. Lag correlation of (a) a water vapor profile, (b) a temperature profile, (c) a zonal wind profile, (d) TBB histogram data and (e) the numbers of rainfall clouds at three horizontal scales less than 100 km 2 (broken line), 100 to 1000 km 2 (dashed line), and greater than 1000 km 2 (thick solid line), compared to the convective activity. The abscissa indicates lag hours; the contour intervals are 0.1.

histogram data as an index of vertical cloud distribution, by their fraction of each level. Note, however, that if an area is covered by high-level clouds, GMS cannot detect lower clouds. Figure 6d shows the lag correlation of the TBB histogram data against the echo fractions. High positive correlations centered on

210 to 230 K, are formed from three hours before the peak time to seven hours after the peak. The high correlations after the peak in convective activity occur because the satellite detects the anvil spread, after the deep convection (Houze et al. 1981). In contrast, the negative correlations about three hours after the

122

Journal of the Meteorological Society of Japan

peak, are found at around 270 to 290 K, which correspond to the level of low-level clouds. This could reflect either a decrease in low-level clouds, or their obscuration by higher-level clouds. After the peak, the peak temperature increased indicative of the decay of the anvil shields associated with the convection. The top height of high relative humidity descended after it reached up to 11 km during 28 to 30 November and 5 to 6 December. It means that cloud-top heights decreased (Fig. 5a). This result also supports the decay of anvil clouds; during 28 to 30 November and 5 to 6 December (Fig. 5a). Figure 6e shows the lag correlations, between the numbers of rainfall clouds classified into three horizontal scales and the echo fractions. Rainfall clouds larger than 1000 km 2 had the same phase as the peak of echo fractions, and the correlation coefficient exceeded 0.8 at hour 0. Rainfall clouds with scales of 100–1000 km 2 , had the highest positive correlations 15 to 18 hours prior to the peak; the correlations subsequently decreased, while lag time became small. It corresponds to what we have seen in Fig. 4c, that a time series of the numbers of 100 to 1000 km 2 scale rainfall clouds reached maximum, about 18 hours prior to the peak of those for more than 1000 km 2 of 2000 UTC 28 November and 1900 UTC 4 December. The rainfall area also revealed a small peak, 12 to 18 hours prior to the larger peak in those cases (Fig. 4a). Thus, we can assume that the small peak of rainfall area observed prior to the larger peak, mainly consist of rainfall clouds with a scale of 100 to 1000 km 2 . Rainfall clouds smaller than 100 km 2 have a positive correlation of around 0.5 until the peak of echo fractions, after which time the correlations decrease. 3.4

Case study of deep convection on 4 to 6 December 2000 Deep convections were observed during the active phase of the MJO, and echo fraction peaks were observed especially at 2000 UTC 28 November, 1600 UTC 30 November and 1900 UTC 4 December (Fig. 4). These three peaks all occurred after midnight. The diurnal variability of convection over the warm pool in the tropical western Pacific is well known (Gray and Jacobson 1977; Chen and Houze

Vol. 84, No. 1

Fig. 7. A time series of the echo fractions of all the rainfall clouds (solid line) and the convective clouds (dashed line) for the study of 0300 LT on 4 December to 0300 LT on 6 December, 2000. Convective cloud fractions are multiplied by 5. The units are %.

1997; Sui et al. 1997; Kubota and Nitta 2001). Furthermore, quasi two-day oscillation of largescale convections was observed during the TOGA COARE, and most of their peaks appeared at nighttime (Takayabu et al. 1996). Case studies of deep convection during westerly wind bursts in TOGA COARE, also reveal nocturnal maximums (Halverson et al. 1999). We described the composite vertical structures of the life cycle of convective activity in subsection 3.3. Our observation also found convective activities that include deep convections, with two peaks (nighttime and 12 to 18 hours earlier) in subsection 3.2. In this subsection we focus on convective activity observed at the case from 4 to 6 December, and describe the relation of two peaks convection to the vertical structure of the environment. Figure 7 depicts a time series of the echo fractions of all rainfall areas and convective rain areas, from 0300 LT (local time) on 4 December to 0300 LT on 6 December. The local time here is 9 hours ahead of UTC. The rainfall area was divided into a convective area and stratiform area, after Steiner et al. (1995). The rainfall area as well as the convective rain fraction, increased from 0800 LT on 4 December to 1100 LT on 4 December. After 1100 LT, the convective rain fraction sub-

February 2006

H. KUBOTA et al.

stantially decreased, and the total rainfall area fraction decreased from 1600 LT. This was the first sequence of the convective system. The convective rain fraction started to increase again from 1800 LT, i.e., another convective cloud started to develop. All rainfall area fractions then increased from 2000 LT until 0400 LT on 5 December. Convective cloud fractions attained a maximum value three hours before the maximum of the total rainfall area fraction. Figure 8 presents rainfall cloud maps observed on 4 to 6 December, derived from the 300 km range Mirai radar PPI. We used a 200 km radius radar range for analyzing lag correlations. We used 300 km radius data in Fig. 8, to describe a broad picture for viewing the movement of rainfall clouds. An east-west oriented band with a scale of 100 km, first emerged from the southwest edge of the radar (not shown). This band weakened before 1300 LT on 4 December. Several rainfall clouds with east-west scales of tens of kilometers, developed on the west to northwest side of the radar range around 1300 LT (Fig. 8a). These rainfall areas moved northeast, and coalesced into a stratiform rain area, with an east-west scale of about 200 km by 1500 LT (Fig. 8b). These rainfall clouds corresponded to the first peak fraction at 1600 LT. The intensive area decreased and separated into groups, while moving eastward at 1700 LT (Fig. 8c). These groups of rainfall clouds weakened, and other rainfall clouds appeared southwest of the radar range at 1900 LT (Fig. 8d). They coalesced into cloud systems, while moving eastward on the southern region of the radar range at 2100 LT (Fig. 8e). They moved eastward out of the 200 km radius radar range, and another group of convective systems emerged from the west side of the radar at 2300 LT (Fig. 8f ). Convective systems moved eastward, and appeared on the north and south sides of the radar range. They coalesced in the east-west direction, and developed up to the 200-km scale at 0100 LT (Fig. 8g). The fraction of convective clouds reached maximum. The clouds coalesced further, and were manifested as a broad stratiform canopy at 0300 LT (Fig. 8h). The rainfall cloud cover reached 36% of the whole radar range at 0400 LT (not shown). Figure 9 illustrates a time series of the water vapor mixing ratio anomalies, and zonal wind for 4 to 6 December. Water vapor increased

123

about 1 to 3 km from 0900 LT on 4 December, and reached a maximum of about 2 to 4 km at 1500 LT, which corresponded to the first peak in echo fractions. After 1800 LT, moistening layers spread upward to 6 km until the second peak of echo fractions. After the second peak at 0400 LT on 5 December, the humidity began to decrease from lower layers. Drying occurred predominately below 2 km from 1200 LT. Water-vapor variations associated with the life cycle of convective activity, corresponded to the lag correlation of convective activity, during the active phase of the MJO. The developing process of convective activity has two peaks of deep convections. However, the water vapor increased from the first peak to the second peak of echo fractions. This suggests that water vapor was supplied into the atmosphere in association with the development and decay of the first peak of active convections. This moistening could create a favorable environment for the subsequent development of deeper convections. Short periods of rainfall were observed above the research vessel, around 1200 LT on 4 December and 0600 LT on 5 December. These convections affected the low-level atmosphere, and increased moisture below 1 km. Figure 9b shows the time series of zonal wind for 4 to 6 December. Strong westerly winds exceeding 15 m/s were observed from 0.5 to 2 km, from 1500 LT on 4 December to 1800 LT on 5 December. During this period, two peaks of total rainfall echo fractions appeared, one at 1600 LT on 4 December, and one at 0400 LT on 5 December. During the second peak of convective activity, the echo fraction of the convective region increased until 0100 LT 5 December (Fig. 7). After that stratiform region still increased until 0400 LT. Westerly winds were accelerated at about 1 to 2 km while stratiform region spread over the radar range. 4.

Discussions

4.1 Developing process of deep convections An MJO associated with deep convections, passed through the observation area during the intensive observation period, from 27 November to 5 December 2000 at 2 N, 138 E over the equatorial western Pacific. The life cycle of convective activity observed in the MJO over the radar range included two peaks of active convections. The first developed in the after-

124

Journal of the Meteorological Society of Japan

Vol. 84, No. 1

Fig. 8. Descriptions of rainfall clouds observed on 4 to 5 December derived from 300-km range Mirai radar PPI at 1300 LT on 4 December (a), 1500 LT (b), 1700 LT (c), 1900 LT (d), 2100 LT (e), 2300 LT (f ), 0100 LT on 5 December (g) and 0300 LT (h). Thin dotted circles are drawn from every 100 km from the research vessel located in the center of the circle.

February 2006

H. KUBOTA et al.

Fig. 9. As in Fig. 7, except for (a) the water vapor anomaly and (b) the zonal wind profiles. The contour intervals are (a) 0.5 g/kg and (b) 5 m/s, except for 2.5 m/s from 15 to 20 m/s.

noon and decayed within a few hours. In the second, rainfall clouds developed and reached the peak in the following night, which spread the rainfall area further than the preceding peak. The life-cycle of convective activity was around 24 hours. Halverson et al. (1999) investigated convective systems observed by radar, during the westerly wind bursts episode of TOGA COARE. They examined nine convective systems for 11 February 1993. The horizontal scale of the convective systems was relatively small before midnight, and the maximum fraction of the ra-

125

dar range was observed after midnight. They suggested that convective systems spread laterally, and include broad stratiform clouds after midnight. However, data limitations precluded discussion of the convective system environment. In our study, a small rainfall echo area increased prior to the larger-scale rainfall echo area during IOP (Fig. 6e). Relatively high cloud-top temperatures were highly correlated before the peak of the rainfall echo (Fig. 6d). These results indicate that a relatively small horizontal scale and lower top height clouds, were observed prior to the maximum deep convections. On 4 to 6 December, convective activities had two peaks, one at 1100 to 1500 LT, and a larger one the following night. The second peak of rainfall echo area was twice that of the former peak. Water vapor was transported upward from the lower to the middle troposphere from the first peak in the rainfall clouds. Moistening layers spread until the second peak of rainfall clouds occurs. Therefore, the first peak of convections acted as a developing stage for water vapor expansion, within the life-cycle of convective activity. Low-level moistening was also discussed on the preconditioning stage of the MJO life cycle (Kemball-Cook and Weare 2001; Kikuchi and Takayabu 2004). KemballCook and Weare (2001) showed that moistening was started from low-level about two weeks before the onset of MJO active convections. Discussions of temporal and spatial scale is larger compared to our study. However, the process of low-level moistening created a favorable environment for deep convections in both time scales of studies. Our study showed this process promoted subsequent deep convection at night, during the active phase of MJO. A fine temporal resolution of three-hour interval upper air observation enabled us to detect them. 4.2 Horizontal structures of disturbances Two-day oscillations appeared in lag correlations, between the fractional rainfall area and TBB, atmospheric parameters as discussed in Takayabu et al. (1996). However, time series of echo fraction of rainfall area represented that two-day oscillation was not continuously appeared, compared to Takayabu et al. (1996) (Fig. 4). Figure 10 depicts lag correlations,

126

Journal of the Meteorological Society of Japan

Vol. 84, No. 1

Fig. 10. Lag correlations of TBB horizontal distribution compared to the fractional rainfall area. Lag hours are described from 12 hours to 18 hours on every 6 hours in (a) to (f ). Black circles indicate 200-km radar range centered at 2 N, 138 E. Contour intervals are 0.1.

between the fractional rainfall area and spatial distribution of TBB. Negative correlation means that TBB values decreased and highlevel clouds increased, when the echo fraction

increased. Strong negative correlations less than 0.5, appeared around the observation area from 6 hours prior to 6 hours after the peak. The horizontal scale of correlated convec-

February 2006

H. KUBOTA et al.

5.

Fig. 11. Correlation of spatial distribution of zonal wind at 850 hPa compared to fractional rainfall area. Black circles indicate 200-km radar range centered at 2 N, 138 E. Contour intervals are 0.1.

tive activity was about 500 km, which was about half of the two-day oscillation of largescale disturbances discussed in Takayabu et al. (1996). The movement of convective activity was weakly eastward, and was opposite from Takayabu et al. (1996). The environment of the horizontal distribution is discussed using NCEP/NCAR reanalysis data. There was a high positive correlation at 850 hPa (1.5 km) of zonal wind in the observation area during the peak of echo fractions (Fig. 6c). Figure 11 depicts the correlation between the fractional rainfall area and spatial distribution of zonal wind at 850 hPa at lag 0. A large value of the positively correlated region, expands over the northwestern observation area toward northeastward up to 10 N. The positive correlation area was paired with a strong negative correlation area, that was located northwest of the positive region. The circulation of a tropical cyclone will reflect the positive and negative pattern. The westerly wind region spread broadly during the latter half of the active phase of the MJO in this study. The case study of rainfall clouds on 4 to 6 December 2000 propagated northeastward (Fig. 8). The intensified westerly wind propelled cloud systems eastward.

127

Summary and conclusions

R / V Mirai performed a stationary observation from 28 November to 11 December 2000 at 2 N, 138 E over the equatorial western Pacific. During this intensive observation period, an MJO associated with convective activity passed through the observation area. Two strong convective systems passed through the observation area on 28 to 29 November and on 4 to 5 December. Strong westerly winds (westerly wind bursts) also occurred during the active phase of the MJO. A typhoon moved westward, and its circulation caused strong westerly winds in the observation area. The cores of the strong westerly winds appeared at 1 to 4 km height. We successfully observed deep convections and their environment, with a fine temporal resolution. Lag correlations between the fractional rainfall area and atmospheric parameters were constructed to detect the atmospheric structures associated with the life cycle of convective activity. Water vapor increased at low levels prior to the peak of echo fractions and at high levels after the peak. Westerly winds had a high positive correlation with the fractional rainfall area, at about 1 km from twelve hours before the peak to six hours after the peak. The life cycle of convective activities included two peaks of active convections. The first active convections developed in the afternoon and decayed within a few hours, supplying water vapor in the low to middle troposphere. This made a favorable environment for subsequent development of deeper convections during the night. Westerly winds were accelerated at about 1 to 2 km, while stratiform region spread over the radar range. Acknowledgments The authors wish to thank Captain Takaaki Hashimoto, his crew, and the technicians of the research vessel Mirai. We also thank Dr. Tokio Kikuchi for use of the geostationary meteorological satellite. QuikSCAT data was obtained from the Remote Sensing System. NCEP/NCAR reanalysis data sets were provided from the anonymous ftp site of the National Oceanic and Atmospheric Administration/Climate Diagnostic Center. The authors thank anonymous reviewers for providing positive comments for improvement of this paper.

128

Journal of the Meteorological Society of Japan

The GFD-DENNOU library was used for drawing of figures. References Chen, S.S. and R.A. Houze, Jr., 1997: Diurnal variation and life-cycle of deep convective systems over the tropical Pacific warm pool. Quart. J. Roy. Meteor. Soc., 123, 357–388. ———, ———, and B.E. Mapes, 1996: Multiscale variability of deep convection in relation to large-scale circulation in TOGA-COARE. J. Atmos. Sci., 53, 1380–1409. Gill, A.E., 1980: Some simple solutions for heatinduced tropical circulation, Quart. J. Roy. Meteor. Soc., 106, 447–462. Gray, W.M. and R.W. Jacobson, Jr., 1977: Diurnal variation of deep cumulus convection. Mon. Wea. Rev., 105, 1171–1188. Halverson, J.B., B.S. Ferrier, T.M. Rickenbach, J. Simpson, and W.-K. Tao, 1999: An ensemble of convective systems on 11 February 1993 during TOGA COARE: Morphology, rainfall characteristics, and anvil cloud interactions. Mon. Wea. Rev., 127, 1208–1228. Houze, R.A., Jr. and C.-P. Cheng, 1977: Radar characteristics of tropical convection observed during GATE: Mean properties and trends over the summer season. Mon. Wea. Rev., 105, 964–980. ———, S.G. Geotis, F.D. Marks, Jr., and A.K. West, 1981: Winter Monsoon convection in the vicinity of north Borneo, part I: Structure and time variation of the clouds and precipitation. Mon. Wea. Rev., 109, 1595–1614. Keen, R.A., 1982: The role of cross-equatorial cyclone pairs in the Southern Oscillation. Mon. Wea. Rev., 110, 1405–1416. Kemball-Cook, S.R. and B.C. Weare, 2001: The onset of convection in the Madden-Julian oscillation. J. Climate, 14, 780–793. Kikuchi, K. and Y.N. Takayabu, 2004: The development of organized convection associated with the MJO during TOGA COARE IOP: Trimodal characteristics. Geophys. Res. Lett., 31, 10.1029/ 2004GL019601. Kubota, H. and T. Nitta, 2001: Diurnal variations of tropical convection observed during the TOGACOARE. J. Meteor. Soc. Japan, 79, 815–830. ———, A. Numaguti, and S. Emori, 2004: Numerical

Vol. 84, No. 1

experiments examining the mechanism of diurnal variation of tropical convection. J. Meteor. Soc. Japan, 82, 1245–1260. Lau, K.-M., L. Peng, C.-H. Sui, and T. Nakazawa, 1989: Dynamics of super cloud clusters, westerly wind bursts, 30–60 day oscillations and ENSO: An unified view. J. Meteor. Soc. Japan, 67, 205–219. Lin, X. and R.H. Johnson, 1996: Kinematic and thermodynamic characteristics of the flow over the western pacific warm pool during TOGA COARE. J. Atmos. Sci., 53, 695–715. Madden, R.A. and P.R. Julian, 1971: Detection of a 40–50 day oscillation in the zonal wind in the tropical Pacific. J. Atmos. Sci., 28, 702–708. ——— and ———, 1972: Description of global-scale circulation cells in the tropics with a 40–50 day period. J. Atmos. Sci., 29, 1109–1123. ——— and ———, 1994: Observations of the 40–50 day tropical oscillation—A review. Mon. Wea. Rev., 122, 814–837. Matsuno, T., 1966: Quasi-geostrophic motions in the equatorial area. J. Meteor. Soc. Japan, 44, 25– 63. McPhaden, M.J., 1999: Genesis and evolution of the 1997–98 El Nin˜o. Science, 283, 950–954. Nakazawa, T., 1988: Tropical super clusters within intraseasonal variations over the western Pacific. J. Meteor. Soc. Japan, 66, 823–839. Nitta, T. and T. Motoki, 1987: Abrupt enhancement of convective activity and low-level westerly burst during the onset phase of the 1988–87 El Nin˜o. J. Meteor. Soc. Japan, 65, 497–506. Steiner, M., R.A. Houze, Jr., and S.E. Yuter, 1995: Climatological characterization of threedimensional storm structure from operational radar and rain gauge data. J. Appl. Meteor., 34, 1978–2007. Sui, C.-H., K.-M. Lau, Y.N. Takayabu, and D.A. Short, 1997: Diurnal variations in tropical oceanic cumulus convection during TOGA COARE. J. Atmos. Sci., 54, 639–655. Takayabu, Y.N., K.M. Lau, and C.H. Sui, 1996: Observation of quasi two-day wave during TOGA COARE. Mon. Wea. Rev., 124, 1892–1913. Wu, X. and M.A. LeMone, 1999: Fine structure of cloud patterns within the intraseasonal oscillation during TOGA COARE. Mon. Wea. Rev., 127, 2503–2513.