A Diagnostic Study on Interactions Between Atmospheric ... - J-Stage

Journal of the Meteorological Society of Japan, Vol. 84A, pp. 151--169, 2006

151

A Diagnostic Study on Interactions Between Atmospheric Thermodynamic Structure and Cumulus Convection over the Tropical Western Pacific Ocean and over the Indochina Peninsula

Yukari N. TAKAYABU, Jun’ichi YOKOMORI Center for Climate System Research, the University of Tokyo, Kashiwa, Japan

and Kunio YONEYAMA Japan Agency for Marine-Earth Science and Technology, Yokosuka, Japan (Manuscript received 30 October 2005, in final form 22 March 2006)

Abstract Interactions between the convective activity and the atmospheric thermodynamic structures are analyzed utilizing upper-air rawinsonde observations obtained by R / V Mirai, R / V Kaiyo, R / V Natsushima, of the Japan Agency for Marine-Earth Science and Technology (JAMSTEC) over the western tropical Pacific Ocean, and those over three GEWEX Asian Monsoon Experiment (GAME) stations: Chiang Mai, Non Khai, and Ubon Ratchathani. Special emphases are placed on understanding the correlation between convection and the atmospheric thermodynamic structures in relation to the recent findings of trimodal cloud levels over the warm ocean (e.g., Johnson et al. 1999) and to the cloud diagnostics proposed by Raymond and Blyth (1992). We first examine the relationships between a convection index and thermodynamic structure indices. A large correlation is found between the convective activity and lower-tropospheric (600–800 hPa) humidity, while there is no significant correlation between the convective activity and Convective Available Potential Energy (CAPE) or Convective Inhibition (CIN). Next, we apply a cloud diagnostic model introduced by Raymond and Blyth (1992) (referred to as RB92) to the observed profiles. As a result, it is shown that there are fundamentally 3 peaks of detrainment levels, which are lower-troposphere (near 900 hPa), mid-troposphere (near 450 hPa), and uppertroposphere (near 150 hPa), over ocean as well as over land. In the soundings over ocean, when the lower-troposphere (600–800 hPa) is dry, there is a tendency for simultaneous existence of stable layers both in the lower-troposphere and in the mid-troposphere. Such atmospheric thermodynamic structure is diagnosed as favorable for strengthened detrainments in the low- and mid-troposphere and weakened in the upper-troposphere. Finally, meridional winds are composited to the north and to the south of the maximum convective activity in the Inter Tropical Convergence Zone (ITCZ) region, respectively, over the tropical western Pacific Ocean. It is confirmed with upper-air soundings that there is a significant meridional divergence

Corresponding author: Dr. Yukari N. Takayabu, Center for Climate Systems Research, the University of Tokyo, 5-1-5 Kashiwano-ha, Kashiwa, Chiba, 277-8568, Japan. E-mail: [email protected] ( 2006, Meteorological Society of Japan

152

Journal of the Meteorological Society of Japan

Vol. 84A

near the melting layer level in the mid-troposhere around 500–600 hPa and a significant meridional convergece near 350–400 hPa, in addition to the lower-tropospheric convergence and the upper-tropospheric divergence of the local Hadley Circulation. These additional circulations in the mid-troposphere are consistent with detrainment profiles diagnosed for observed atmospheric profiles utilizing RB92 cloud model. After all, it is strongly suggested that the cloud microphysics, such as melting and freezing, play significant roles in determining the large-scale circulation.

1.

Introduction

The tropical cumulus convection is known to affect the atmospheric circulation through various processes. In considering the Coupling Processes in the Equatorial Atmosphere (CPEA), deep cumulus convection is an essential process as emphasized in Fukao (2006). Therefore, we would like to clarify characteristics of the tropical cumulus convection in details. The most important role of the tropical deep cumulus convection is understood as transporting excessive energy gained at the surface to the upper troposphere. For a several decades, a ‘hot tower’ hypothesis introduced by Riehl and Malkus (1958) was widely accepted as an approximate image of tropical cumulus convection. However, there have been objections placed to the noentrainment hypothesis of a hot tower cumulus, for there are scarcely no such strong updraft in the cumulus convection found in numerous numbers of aircraft observations over ocean (Zipser and LeMone 1980; Zipser 2003). Also, based on observational studies by Betts (1982), it is suggested that tropical atmosphere should be considered as moist neutral to the reversible vertical motion, instead of assuming the conditional instability (e.g., Emanuel 1986; Xu and Emanuel 1989). In the usual general circulation model at present, it is necessary to parameterize the effects of cumulus convection, which is really a subgrid-scale phenomenon, with largescale field variables. It is also well known that cumulus parameterization largely affects the performance of the climate models (Houghton et al. 2001, IPCC third assessment report). Therefore, it is indispensable to adequately understand the relationship between the atmospheric thermodynamic structure and the cumulus convection. Recent cumulus parameterizations are mostly based on the concept of the adjustment, such as Manabe and Strickler (1964) and Arakawa and Schubert (1974). How-

ever, it is still unclear how entrainments and detrainments occur in the real cumulus convection; how they relate to the atmospheric structure, and how they affect the hierarchical structure and step-wise developments of organized convective systems such as Madden-Julian Oscillation (Madden and Julian 1971, 1972; Nakazawa 1988; Kikuchi and Takayabu 2004). In order to examine the entrainment effects for the cumulus convection, Blyth et al. (1988) analyzed the cumulus convection over the land utilizing the aircraft observation and upper-air observation data during the Cooperative Convective Precipitation Experiments (CCOPE) and the High Plains Cooperative Program (HILPLEX). As a result, they showed an adequacy of a thermal type cumulus model, in which each parcel experiences an entrainment, loses its buoyancy, and detrains. Based on these results, Raymond and Blyth (1992), hereafter referred to as RB92, introduced a cloud model to diagnose the vertical distribution of detrainments, which reasonably reproduced the observed detrainments from cumulus convection. On the other hand, interesting new features of tropical cumulus convection have been revealed by recent studies. With data obtained from Tropical Ocean Global Atmosphere Coupled Ocean-Atmosphere Response Experiments (TOGA COARE), Johnson et al. (1999) and its related studies (Johnson et al. 1996; Zuidema 1998) indicated that there are ubiquitous cumulus congestus over the tropical western Pacific warm pool. They presented a new schematic image of tropical oceanic cumulus convection as a trimodal structure with cumulus congestus with a melting-level height, in addition to trade inversion-height cumuli and very tall cumulus convection which reaches the tropopause. Kikuchi and Takayabu (2004) utilized TOGA-COARE soundings together with infrared TBB histograms obtained by Japanese Geostationary Meteorological Satellite, and dis-

July 2006

Y.N. TAKAYABU, J. YOKOMORI and K. YONEYAMA

played that there is a stepwise development in convection associated with the Madden-Julian Oscillation (Madden and Julian 1971, 1972). They showed that after the suppressed stage, when the cumulus stops at the trade inversion level, there is a development stage where frequent convection stops at the mid-troposphere and moistens the lower half of the troposhere, then finally mature stage where convection reaches the tropopause level appears. Redelsperger et al. (2002) applied the TOGA COARE data to a cloud resolving model and showed that atmospheric moisture profile and stability suppress the development of cumulus convection. These studies indicate that the midtropospheric thermodynamic stability and incidental convective detrainments are important factors to determine the characteristics of cumulus convection. The relationship between cumulus convection and environmental conditions is complicated through various factors. Fu et al. (1994) analyzed the effects of sea surface temperature (SST) to the surface divergence field and resulting deep convection, utilizing upper sounding and SST data. They concluded that high SST and surface convergence are both important for the realization of deep convection through affecting the atmospheric vertical profiles. They also indicated that there are conditions that deep convection is suppressed by a dry and stable atmospheric boundary layer associated with a surface divergence; even the convective available potential energy (CAPE) attains a large value. More recently, Sherwood (1999) statistically showed that accumulation of lower tropospheric moisture is essential for the initiation of convection over the tropical ocean. Raymond et al. (2003), on the other hand, examined the entropy budget in the atmospheric boundary layer with observations obtained in Eastern Pacific Investigation of Climate processes program (EPIC) 2001. They indicated that two thirds of the infrared reflectivity can be explained with the atmospheric stability index and the surface moisture flux. As for the effect of cumulus convective activity to the large-scale circulation fields, vice versa, Mapes (2001) pointed it out that a midlevel convergence appears in the meridional wind field of the reanalysis data both in ERA40 and in NCEP, which is an indication of

153

impacts of melting-level atmospheric structure found with TOGA-COARE studies (Johnson et al. 1996, 1999; Zuidema 1998). Zhang and McGauley (2004) recently discovered a meridional return flow at around 2–4 km just above the trade inversion level over the eastern Pacific Ocean, utilizing the rawinsonde data, dropsonde data, and wind profiler data. Their finding suggests that the atmospheric stability structure significantly affects the large-scale circulation. Primary objectives of this study are to investigate the interaction between the convective activity and the atmospheric thermodynamic structures, by utilizing the rawinsonde upperair observations and a diagnostic cloud model. Special emphases are placed on understanding the correlation between convection and the atmospheric thermodynamic structures in relation to the recent findings of trimodal cloud levels over the warm ocean (e.g., Johnson et al. 1999) and to the cloud diagnostics proposed by RB92. We first examine the characteristics of observed atmospheric profiles and their statistical relationship to the convective activity. We then analyze the effect of stratification utilizing a diagnostic cloud model. Finally, we examine the effect of atmospheric profile and convection onto the large-scale circulation over the tropical western Pacific Ocean. 2.

Data and methodology

For the analysis, upper-air rawinsonde observations obtained by R / V Mirai, R / V Kaiyo, R / V Natsushima, of JAMSTEC over the western tropical Pacific Ocean during the period from February 1993 to December 2002 are utilized for the oceanic data (Table 1). Details of the observational cruises and radiosonde data processing can be found in Yoneyama (2003). Those at three GAME-tropics stations, Chiang Mai, Non Khai, and Ubon Ratchathani during the wet seasons from May to September in 1997 and in 1998 are utilized for the continental data (Table 2). Upper air observations are done either three hourly or six hourly, depending on the observation period. In order to avoid statistical bias on local time, we selected six hourly data from all available data. Note that because we perform the correlation analysis in this study, we only utilize observational months in which at least

154


Vol. 84A

Table 1. Numbers of upper air soundings used in this study over ocean in each month. year

1993

1994

1995

1996

1997

1998

1999

2000

2001

2002

Numbers 36 (Feb) 21 (Dec) 25 (Jan) 6 (Jan) 13 (Jan) 44 (Mar) 54 (Jun) 8 (Feb) 79 (Nov) 33 (Nov) 57 (Jul) 65 (Feb) 70 (Feb) 32 (Jun) 17 (Jul) 13 (Mar) 27 (Dec) 47 (Dec) over 34 (Jun) 58 (Jul) 70 (Aug) ocean 12 (Nov) 5 (Aug) (month) 38 (Dec)

Table 2. Numbers of upper air soundings used in this study, at GAMEtropics stations in each month. year

1997

Ubon Ratchathani

1998 46 (May), 54 (Jun), 58 (Aug), 56 (Sep)

Non Khai

23 (Aug), 22 (Sep)

Chiang Mai

15 (Aug)

five consecutive days of six hourly data are obtained. Total launch numbers utilized in this study are 865 over ocean and 484 over land. Their geographical distribution is indicated in Fig. 1. Most oceanic data are obtained in the equatorial western Pacific between 10 N and 10 S. Figure 2 depicts the launch time distribution of the utilized data in local time. Although there is a slight shift in sampling times between over ocean and over land, we do not further concern about this shift here. Tables 1 and 2 show the periods of data utilized in this study, over ocean and over land, respectively. We must be aware that continental data utilized here are basically collected during the summer monsoon wet season over the Indochina Peninsula at three stations; Chiang Mai, Non Khai, and Ubon Ratchathani, so that they are not representing the tropical continental conditions. Yet, we think it is still interesting to compare wetseason continental conditions with the conditions over the warm water pool, since in both situations plentiful supply of moisture from the wet surface are expected. All soundings are interpolated into 5 hPa intervals with a Spline interpolation method. Three hourly infrared equivalent black body temperature (IR-TBB) data observed from Japanese Geostationary Meteorological Satellite (GMS) and gridded into 1 deg 1 deg longitude-latitude grids for the corresponding period are also utilized to represent the convective activity.

43 (May), 52 (Jun), 63 (Aug), 52 (Sep)

In order to examine statistical relationship between cumulus convection and atmospheric thermodynamic structure, we utilized IR-TBB for the convection index and five indices for the

Fig. 1. Geographic distribution of upper air data used in this study. Numbers found in boxes indicate the launch numbers of rawinsonde observations utilized in this study, over the western tropical Pacific Ocean by R / V Mirai, R / V Kaiyo, and R / V Natsushima of JAMSTEC during the period of Feb 1993–Dec 2002. Total number is 865. Over Indochina Peninsula, rawinsonde observations at Chiang Mai, Non Khai, and Ubon Ratchathani for Aug 1997– Sept 1998 of GAME-tropics were utilized. Their total number is 484.

July 2006


Fig. 2. Launch time distribution of utilized rawinsonde data numbers in local time, over ocean (solid line) and over land (dashed line).

thermodynamic structure. The latters are summarized in Table 3. In order to understand the relationships between convective detrainments and thermodynamic structures, we examined the effects of atmospheric stratifications with a diagnostic cloud model by RB92, which will be described later in details. 3.

Statistical relationships between convection and atmospheric profiles

3.1 Characteristics of atmospheric profiles First, we investigated the relationships between IR-TBB and thermodynamic structure indices. In order to calculate some stratification indices, we first need to determine the cloud base height, where a parcel starts to condense. In order to do so, we compared the diagnosed

lifting condensation levels (LCLs) for parcels from different atmospheric layers near the surface to cloud base heights observed over the ocean by ceilometer data (not shown). As for the cloud base heights from the ceilometer, we selected only those values less than 1200 m, since we only wanted the low-level cloud base of the cumulus convection. As a result, LCLs of any parcels from 1000 to 950 hPa coincide well with the cloud base observed with the ceilometer. Hereafter we define a cloud-base parcel as a parcel with average temperature and humidity in the 1000–980 hPa layer. In calculating CAPE and the convective inhibition (CIN), we slightly modified a definition of the level of free convection (LFC) as a level where buoyancy integration from the cloud base upward attains the minimum value, ignoring the occasional appearance of very thin layer with negative buoyancy. Figure 3 shows mean temperature lapse rate profiles averaged for all oceanic data and all continental data utilized in this study. Aside from a strong surface inversion over the land, three stable layers are found both in oceanic and in continental profiles. They are trade inversion around 2–3 km, a weaker stable layer near the melting level around 6 km, and near the tropopause. The temperature profile over ocean is very similar to those obtained in the Coupled Ocean-Atmosphere Response Experiment (COARE), the Tropical Eastern Pacific Process Study (TEPPS, Yuter and Houze 1999),

Table 3. List of indices for stratification Abbreviation

Indices for stratification

155

Definition

CAPE

Convective available potential energy

Amount of work done by buoyancy to a cloudbase parcel lifted from LFC (level of free convection) to LNB (level of neutral buoyancy)

CIN

Convective inhibition

Amount of work need to be put to a cloud base parcel in order to lift it from the surface to LFC, against the negative buoyancy.

RHlt

Lower tropospheric mean relative humidity

Mean relative humidity averaged in the layer of 600–800 hPa.

TIS

Trade wind inversion level stability

A minimum observed lapse rate for all 100 hPa layers centered with 5 hPa intervals between 900 and 600 hPa

FLS

Freezing level stability

Same as TIS but for 600–350 hPa

156


Vol. 84A

Fig. 3. Temperature lapse rate ðdT/dzÞ. Profiles averaged for all data in analysis over ocean (solid line) and over land data (dashed line). Error bars indicate the 95% significance intervals. Note that the lapse rate in ordinate increases downward, so that the stability increases upward.

and the GARP (Global Atmospheric Research Program) Atlantic Tropical Experiment (GATE) as shown in Mapes (2001) and confirmed to be a very robust feature over the tropical ocean. The level of trade inversion over land, on the other hand, is higher than that over ocean, so that three-peak structure of atmospheric stratification is slightly more obscure compared to that over ocean. Here we define two stability indices. One is the trade inversion stability (TIS) defined as a minimum lapse rate for all 100 hPa layers centered with 5 hPa intervals between 900 and 600 hPa. The other is the freezing level stability (FLS) defined as same as TIS but for between 600–350 hPa. The three stable levels at TIS, at FLS and at the tropopause, needless to say, correspond to the trimodal structure of the cumulus convection presented by Johnson et al. (1999). In individual observations, very strong stable layer is occasionally found over ocean but scarcely over land (not shown). Histograms of

TIS and FLS in 0.5 K km1 bins are shown in Fig. 4, for those over ocean and those over land. Maximum counts of TIS are observed in the [5.0 to 4.5 K km1 ] bin, while those of FLS are found in the slightly weaker-stability bin of [5.5 to 5.0 K km1 ] both over ocean and over land. It is noticed that there is no TIS stronger than 2 K km1 over land, while there are some stronger TIS over ocean. Similarly, there is no FLS more stable than 3.5 K km1 over land, while there are some stronger FLS over ocean. Next, we examined another stratification index, RHlt, or the 600–800 hPa (lowertropospheric) mean relative humidity. Figure 5 compares frequency distributions of RHlt over ocean and over land. It is found that dry lower troposphere conditions are more frequently found over ocean than over land. For example, frequency of RHlt < 40% are 14% (125/865) over ocean while it is only 3.5% (17/484) over land. It may be due to the period of observation over land, which is during the monsoon wet

July 2006


Fig. 4. Frequency Distributions of temperature lapse rates at FLS and TIS over ocean (solid lines) and over land (dashed lines), respectively. Lines with crosses are for TIS level lapse rates, and those with asterisks or open squares are for FLS levels. Note that the lapse rate in abscissa increases leftward, so that the stability increases rightward.

Fig. 5. Frequency distributions of RHlt (600–800 hPa) over ocean (solid line) and over land (dashed line).

season. Still, it is interesting to find more soundings with very dry lower troposphere over the warm ocean where plentiful moisture supply from the surface is expected. 3.2 Comparisons of diurnal variations Figure 6 shows diurnal variations of TBB in the 1 deg 1 deg grid including the location of

157

Fig. 6. Diurnal variations of GMS IRTBB for the 1 deg 1 deg longitudelatitude grid including the upper-air sounding point and for the corresponding time. Solid curves indicate those over the sea, while dashed curves represent those for over continent. Error bars indicate 95% significance intervals with the Student-t test.

corresponding upper-air observation, which is employed as a cumulus convection index. We have to be careful about the discrete sampling timings shown in Fig. 2. Over land, the 19 LT spot is found to be the most convectively active from the TBB data. Over ocean, there is no significant diurnal variation in TBB. Note that TBB values include the surface temperature information when it is clear, so that they do not purely represent the cloud top temperatures, but are contaminated with the surface temperature. Over ocean, where the sea surface temperature is fairly homogeneous, the contamination do not affect much. However over land, it could largely affect the information. Therefore, in future studies, we would like to remove the surface effects from the analysis. Diurnal variations of stratification indices are depicted in Fig. 7. Diurnal variations over land are significant for CAPE (Fig. 7a) and marginal for CIN (Fig. 7b) with a maximum in the evening in both indices, which is favorable for convection. As for the oceanic case, both CAPE and CIN variations are marginal with a common maximum in the early morning. It is also notable, that a near surface parcel always faces a higher hurdle of CIN over land than

158


Vol. 84A

Fig. 7. Same as Fig. 6 but for stratification indices. (a) CAPE, (b) CIN, (c) TIS, (d) FLS, and (e) RHlt (relative humidity averaged for 600–800 hPa. Error bars indicate 95% significance intervals with the Student-t test.

over ocean before reaching the level of free convection (LFC). Both TIS (Fig. 7c) and FLS (Fig. 7d) work in a manner to discourage deep convection. There is a significant diurnal variation of TIS over land with a maximum at around the local noon. On the other hand, there is almost no diurnal variation of TIS over the sea, but note that it is always as large as the maximum value over the land. FLS diurnal variations are

not significant, but the maximum is observed near the local noon over land, while in the early morning over ocean. FLS is also always larger over ocean than over land. The relative humidity in the lower troposphere (RHlt) is significantly varying over land and marginally over ocean. Over land, a significant minimum is found around noon and RHlt is larger from evening to morning, favorable

July 2006


159

Table 4. Correlation coefficients between the index of cumulus convection and indices of atmospheric stratification. Bold Italic fonts indicate 95% significant values.

Indices for stratification

Index for convection (TBB) OCEAN

LAND

CAPE

0.031

0.020

CIN

0.056

0.049

RHlt

C0.54

C0.45

TIS

0.21

0.16

FLS

0.053

C0.12

for cumulus convection. Over ocean, a marginal maximum is seen in the early morning. Again it is noticed that lower-tropospheric relative humidity is mostly smaller over ocean than over land. 3.3

Correlations among convective indices and stratification indices Finally in this section, we would like to present the correlations between the TBB and stratification indices. The correlation coefficients are calculated utilizing the all available data over ocean and over land, separately, and shown in Table 4. Clearly, there is no correlation between TBB and CAPE, or between TBB and CIN. Although in the cumulus parameterization in numerical models we often refer to variables akin to CAPE for adjustment, in the real world it often happens that CAPE is already adjusted when the convection is observed. Examples for smaller CAPE profiles are found in Fig. 8 (a: @202 J kg1 ) and (b: @85 J kg1 ), and for larger CAPE profile in Fig. 8 (c:@2480 J kg1 ). For two smaller CAPE cases, completely different profiles are found. Figure 8 (a) shows a dry case, where convection is severely suppressed, while (b) is an adjusted case, where total troposphere is completely wet. Figure 9 shows vertical frequency distribution of water vapor mixing ratio, for smaller CAPE (700 J kg1 ) cases. It is noticed that there are two branches of distribution in smaller cases (Fig. 9a), one is around very

Fig. 8. Examples of small-CAPE profiles (a,b) and large-CAPE profiles (c). Three profiles in each panel indicate potential temperature, equivalent potential temperature, and saturated equivalent potential temperature, respectively from the left. Horizontal lines indicate levels of LCL, LFC, and LNB upward.

small mixing ratio (@2 J kg1 ) through the troposphere, and the other branch is in moister area. Comparing its moister branch in the larger CAPE distribution, it is confirmed that this branch represents the adjusted conditions because the lower troposphere is dryer and the upper troposphere is moister than larger CAPE cases. Note that among smaller CAPE cases, adjusted cases are much more frequent than severely dry cases. It is consistent with Sobel

160


Vol. 84A

Fig. 9. Vertical frequency distribution of water vapor mixing ratio for smaller CAPE (700 J/kg) profiles. Contours and shadings indicate the numbers of frequency in the corresponding bins of mixing ratio and pressure.

et al. (2004)’s results showing a negative correlation between CAPE and precipitation. On the other hand, both over ocean and over land, there are large correlations between TBB and lower-troposphere (600–800 hPa)-mean humidity (RHlt). Figure 10 shows lag correlations of the water vapor mixing ratio at each levels versus TBB. First of all, it is noticeable that vertically coherent large correlation is observed at around day 0, which indicates that water vapor is vertically transported by the cumulus convection throughout the troposphere. From six days to four days before the reference time, there is a slight increase of the mixing ratio in the boundary layer under @800 hPa. From five days to two days before the reference time, there is a significant increase of the mixing ratio below the melting level of @450 hPa. Significant moistening seen with absolute correlations larger than 0.3 (green shades) in the lower troposphere, around 800–700 hPa (found

from 2.3 to 1.9 days), slightly leads those in the upper troposphere around 400–200 hPa (found from @1.6 days). These are consistent with the result of Sherwood (1999), showing the lower-tropospheric humidity is a precursor to the deep convection. Moreover, it is interesting to find stepwise increase of humidity preceding the mature phase of convection, while after the convection peak a very gradual and smooth downward decrease of humidity profiles is found. This is a common feature found in convectively organized systems in different scales (e.g., Takayabu et al. 1996; Kikuchi and Takayabu 2004). Looking back the Table 4, secondary significant correlations between TBB and TIS are found both over ocean and over land, in a sense that TIS discourage the convection. As for FLS, marginally significant negative correlation is found only over land. However, its value is very small, and the signature is in the manner that a stable layer enhances

July 2006


161

Fig. 10. Lag correlations of the water vapor mixing ratio at each levels versus TBB over ocean. Negative lag means that the correlated mixing ratio leads TBB. Contours indicate the lag correlation values, with color shades for those values over 95% significance levels.

the convection, which opposes the physical sense. Therefore, we do not discuss about this correlation further. In the following section, we will consider how these atmospheric profiles affect the convection, with a special emphasis on the lower tropospheric relative humidity. 4.

Effects of lower-tropospheric relative humidity and diagnosed detrainments

4.1 Atmospheric profiles for RHlt classes In the previous section, we have shown that convective activity (TBB) is highly correlated with the lower tropospheric relative humidity. So, in this section, we first examine actually what vertical profiles are found with the analyzed data. We classified all sounding data we utilized into 10% bins of RHlt (600–800 hPa mean RH) and averaged their profiles over ocean and over land, respectively. Average pro-

files for lower RHlt classes are shown in Fig. 11. Figures 11 (a) and (b) show the profiles for potential temperature, equivalent and saturated equivalent potential temperature. Over ocean, as already pointed out in the previous section, extremely dry RHlt case is more often found than over land. When we observe the saturated equivalent potential temperature over ocean, it is notable that freezing-level stable layers tend to appear in concert with the trade wind inversions: smaller the RHlt becomes, it is found that TIS (@900– 800 hPa) and FLS (@550–500 hPa) becomes more significant. Figures 11 (c) and (d) depict relative humidity profiles in same RHlt bins over ocean and over land. Over ocean (Fig. 11c), there are clear local maxima at around 500–550 hPa levels, with steeper peak for lower RH. These levels correspond with the mid-level stable layer

162


Vol. 84A

Fig. 11. Atmospheric Profiles of potential temperature, equivalent potential temperature, and saturated equivalent potential temperature (upper panels), and relative humidity (lower panels) averaged in 10% lower-tropospheric humidity bins from 10 to 40% over ocean (a,c) and from 30 to 40% over land (b,d). As seen in Fig. 5, there is little samples for 10–30% RHlt over land. Profiles labeled ‘all’ indicate the average profiles calculated from the all soundings. Error bars in (c) and (d) indicate 95% significance intervals with the Student-t test.

(FLS) found in Fig. 11a. With error bars indicating 95% significance intervals, we can confirm these mid-level maxima in RH are robust features for lower RHlt. Over land (Fig. 11d), on the other hand, there is no such robust maximum found for 30–40% RHlt curve, corresponding to the smooth saturated equivalent potential curve in Fig. 11(b). Figure 12 shows average profiles for higher RHlt classes. Comparing RH profiles over ocean and those over land, we can notice clear separation below and above around 550 hPa over ocean, but rather monotonic shift through the entire troposphere is found over land. After all, it is notable that as frequently

found over ocean, when the lower-troposphere (600–800 hPa) is dry, there is a tendency for simultaneous existence of stable layers both in the lower-troposphere and in the midtroposphere. 4.2

Application of the RB92 diagnostic model to the observed profiles In order to diagnose the effects of atmospheric profiles to detrainments, we applied a stochastic-mixing model of RB92 to the averaged atmospheric profiles for each RHlt classes. Figure 13 is a schematic explanation of RB92 model. First of all, a cloud base (LCL) and a cloud top (LNB) are determined from the given

July 2006


Fig. 12.

163

Same as Fig. 11 but for moister RH bins from 70–100%.

Fig. 13. Schematics for a StochasticMixing Model by Raymond and Blyth (1992).

profile. Then this model virtually forces the cumulus convection from the LCL to the LNB. Same amount of the cloud base air is forced to move to each I’th layer, each of 200 layers in this study, maintaining the entropy and mixed with the environmental air. The mixing occurs in the manner that same amount of nine-type parcels with 10–90% mixture of environmental air are produced. Then each parcel moves upward or downward to its own neutral buoyancy level and detrains there. More details of the model is found in RB92. As characteristics of RB92 model, this model represents the temporarily and spatially inhomogeneous mixing, and the effect of losing the buoyancy through entrainments. The diagnosis depends on the atmospheric stability. Simple precipitation and freezing processes are included, and it is assumed that one parcel experiences only one mixing event. It is shown that these assump-

164


Vol. 84A

Fig. 14. Detrainments diagnosed with RB92 model from the average profiles over ocean (a) and over land (b). Abscissa is the detrainment normalized with the total amount of cloud-base air parcel and the ordinate is pressure. ‘Mode 1,2,3’ and ‘strong’ in the figure show the conversion parameter differences. Please see the details in the text.

tions are based on observations of cumulus convections over the land and expresses the observed detrainments well (Raymond and Blyth 1992). We applied this RB92 model to the mean atmospheric structures over ocean and over land, respectively, and diagnosed the convective detrainments. Simple conversion processes from cloud water to rain and cloud water to ice are included as follows: Dqp ¼ Rp qL DT;

and

DqI ¼ RI qL DT;

where, qp represents the precipitating water mixing ratio, qI is the ice water mixing ratio, and qL is the cloud water mixing ratio, T is temperature and D stands for the deviation from the starting point of the parcel. Rp and RI are conversion constants. We performed a brief sensitivity study against their conversion parameters as shown in Fig. 14. Conversion constants for mode 1 are set as Rp ¼ RI ¼ 0:02, for mode 2 as Rp ¼ RI ¼ 0:05, for mode 3 as Rp ¼ RI ¼ 0:1, and for mode strong as Rp ¼ RI ¼ 0:2. As a result, it was shown that fundamentally three peaks of detrainment levels, lower-troposphere (near 900 hPa), mid-troposphere (near 450 hPa), and upper-troposphere (near 150 hPa), were diagnosed irrespective of the conversion parameters over ocean as well as over land.

Next we applied the RB92 diagnosis to the mean profiles for different classes of RHlt (Fig. 15). In the previous subsection, we showed that when the lower troposphere (600–800 hPa) is dry, there is a tendency to have two stable layers simultaneously at the low and at the middle levels. Figure 15 (a) indicates that these low RHlt profiles certainly results in larger detrainment at the middle level (520–420 hPa), compared to those for the average profile (black line). Correspondingly, the detrainment at the upper level becomes smaller. On the other hand, for high-RHlt (Fig. 15b) soundings, detrainments are mostly decreased in the lower (900–700 hPa) and in the middle (7000– 500 hPa) levels, and also in the upper levels below 200 hPa-altitude, compared to those for the average sounding. The upper-most detrainments are significantly increased for the high RHlt cases. 5.

Effects of atmospheric structure and convection on the large-scale circulation

Finally, in order to investigate the effect of the convective activity to the large-scale circulation, in turn, we composited meridional winds in the north side and in the south side of ITCZ (Inter Tropical Convergence Zone) region. First, TBB minimum was determined within the

July 2006


165

Fig. 16. TBB composite referring to the latitudes of convection centers.

Fig. 15. Normalized detrainments for atmospheric profiles classified with lowlevel RH over ocean, for lower bins of RHlt (a) and for higher bins (b). Lines labeled ‘all’ indicate those for average profiles calculated from the all soundings.

41 deg 41 deg area with a center at each location of upper air observation over ocean. Then, whole TBB maps are latitudinally shifted in the manner that the latitude of TBB minimum points comes to the zero relative latitude, and composited. In Fig. 16, composite TBB field is shown, with relative longitude zero corresponds to the longitude of the observation sites, and relative latitude zero corresponds to the latitude of TBB minimum. It is noticed that TBB field exhibits a quite zonal distribution with enhanced convection within 5 degrees and suppressed convection in 10–20 degrees in both sides from the TBB minimum. Average TBB

Fig. 17. Composite meridional winds to the south (left) and to the north (right) of the convection center. Data from soundings with maximum distance of 20 degrees in latitude are composited. Error bars are for the 95% significance intervals.

minimum is found @11 degrees to the west of the upper air observation sites. Next, all observed meridional winds within the 20 degrees in latitude of the convection center were composited to the south and to the north of the reference separately. Figure 17 shows the composite meridional winds. Error bars are for the 95% significance intervals. From Fig. 17, we can see that there is a significant meridional divergence near the

166


melting layer level in the mid-troposhere, and also a significant meridional convergence near 350–400 hPa, besides the lower-tropospheric convergence and the upper-tropospheric divergence associated with the Hadley Circulation over the tropical western Pacific region. This additional circulation in the mid-troposphere suggests the significant role of cloud microphysics plays in determining the large-scale circulation: mid-tropospheric detrainment resulting from the melting-level stability, and midlevel convergence may be associated with the reboost of convection with the latent heat of freezing (Zipser 2003), as diagnosed by the RB92 model. Needless to say, this is another manifestation of the dominance of trimodal cumulus convection over the warm water pool suggested by Johnson et al. (1999). Although these features are previously reported to exist in the 15-year averages of ECMWF and NCEP objective reanalysis data (Mapes 2001), objective analysis data are largely influenced by the utilized numerical model and it is difficult to affirm this large-scale mid-level meridional convergence really exists. We could confirm these effects of trimodal cumulus convection over the western Pacific Ocean directly utilizing ship-born upper-air soundings. 6.

Summary

In order to investigate the interaction between the convective activity and the atmospheric thermodynamic structures, rawinsonde upper-air soundings obtained over ocean and over land, and GMS IR data are analyzed. Upper-air data are obtained by R / V Mirai, R / V Kaiyo, and R / V Natsushima of JAMSTEC over the western tropical Pacific Ocean primarily in boreal summer and in boreal winter seasons, and also at three stations of GAMEtropics; Chiang Mai, Non Khai, and Ubon Ratchathani, primarily during the wet-seasons of the Indochina monsoon. First, statistical relationships between the indices of cumulus convection and atmospheric stratification are examined, by utilizing the upper air sonde data. Then, characteristics of atmospheric stratification over ocean and over land are compared. A cloud diagnostic model by Raymond and Blyth (1992) is applied to the upper-air data to diagnose the effect of atmospheric profiles to the cumulus convection. Finally, we examined the ef-

Vol. 84A

fect of cumulus convection onto the large-scale meridional circulation, in turn, over the tropical western Pacific Ocean. Primary results are listed below. 1. Convective activity (TBB) is highly correlated with the low-level relative humidity (RHlt: 600–800 hPa), while there are no significant correlations between convective activity and CAPE or CIN. 2. Very dry conditions of the lower troposphere (600–800 hPa) are often found over the tropical warm pool, while it is scarcely found in the three station data over the wet-season Indochina Peninsula. 3. It is found that low-RHlt soundings over ocean tend to have stable layers at the trade inversion level as well as at the freezing level, simultaneously. There was no such significant tendency observed with the wetseason GAME-tropics soundings. 4. By applying RB92 model to observed upper air soundings, it was shown that fundamentally three peaks of detrainment levels, lower-troposphere (near 900 hPa), midtroposphere (near 450 hPa), and uppertroposphere (near 150 hPa), were diagnosed over ocean as well as over land. 5. Taller convection is diagnosed with largerRHlt profiles, which is consistent with the correlation statistics. 6. It is confirmed with upper-air soundings that there is a significant meridional divergence near the melting layer level in the mid-troposhere around 500–600 hPa and a significant meridional convergece near 350–400 hPa, in addition to the lowertropospheric convergence and the uppertropospheric divergence of the local Hadley Circulation. Combining above six results together, we can summarize the contributions of this study as follows. Upper air sounding data are classified in the manner examining the RHlt effects on the convection suggested by previous studies (e.g., Sherwood et al. 1999; Sobel et al. 2004) and these characteristics are described. A diagnostic cloud model by RB92 is applied to these classified sounding data, and effects of the atmospheric structure to the cumulus convective detrainment profiles in the real world are diagnosed. We also confirmed the additional large-

July 2006


scale meridional divergence (500–600 hPa) and convergence (350–400 hPa) to the local Hadley circulation, previously suggested with objective analysis data. It is emphasized that these results are consistent with diagnosed detrainment profiles with actual atmospheric soundings, and shown as another manifestation of the trimodal cumulus convection presented by Johnson et al. (1999). 7.

Concluding discussions

In this study, we aimed to diagnose the convection-atmosphere interaction utilizing observed sounding data. We compared oceanic cases with cases for continental wet-seasons over the Indochina Peninsula. It was shown that the lower-tropospheric (600–800 hPa) mean relative humidity highly correlates with the convective index, consistent with Sherwood (1999). We put special emphasis on understanding the correlation between the convective index and the atmospheric thermodynamic structure indices, in relation to the recent findings of trimodal cloud levels over the warm ocean (e.g., Johnson et al. 1999) utilizing the diagnostic models proposed by Raymond and Blyth (1992). Over the ocean, soundings with very dry lower-troposphere were often found. In such cases, there is a clear tendency that both the freezing level stable layer and the trade wind inversion appear simultaneously. These three conditions, dry lower troposphere, stronger FLS and TIS, all work to enhance detrainments at TIS or FLS levels and discourage deeper convection. Simultaneous occurrences of these three conditions suggest the more frequent low-level dry condition over ocean is associated with long-term subsidence, which is also suggested in Mapes (2001). Otherwise, abundant moisture supply from the warm water surface would never allow such dry condition in the lower troposphere. On the other hand, a lack of soundings with such dry lowertroposphere accompanied by larger FLS and TIS suggests that wet season conditions over land do not provide such continuous subsidence. It is interesting that despite the warm sea surface provides plentiful moisture to the atmosphere, signals of longer-term subsidence are observed than over wet-season continent. It is probably because the oceanic atmosphere lacks the mixed layer development during the

167

day which forces the showery afternoon turn over of the atmosphere over land. This is also consistent with the suggestions that precipitation systems are more organized over ocean than over land (e.g., Takayabu 2002; Schumacher and Houze 2003). We here recall that freezing level stability is attributed to the melting of stratiform rain, which originates from the organized convective systems over ocean. Differences in existence of longer-term subsidence may be essential to the differences in levels of organization of convective systems over ocean and over land. Further studies are needed to clarify these points. In the last part, we analyzed the effect of cumulus convection to the large-scale environment, in turn. This time, we focused on how the large-scale meridional circulation is affected by cumulus convection, referring to works by Mapes (2001) suggesting the additional mid-level meridional circulation to the deep local Hadley cell over the western Pacific Ocean with two reanalysis data, and by Zhang and McGauley (2004) that reported a existence of a shallow meridional return flow from the ITCZ over the eastern Pacific Ocean. As a result, we could confirm the existence of a significant divergent component around the freezing level (500–600 hPa), and a significant convergent component just above it around 350– 400 hPa with upper-air soundings, in addition to the deep meridional cell of the local Hadley circulation. These mid-level additional flows over the warm western Pacific Ocean are contrasting to trade-wind inversion-level return flow over the eastern Pacific. The level of the secondary outflow from the ITCZ almost coincides with the level of detrainment diagnosed from the RB92 model with observed atmospheric profiles. It is considered that the midlevel detrainment associated with the weak stable layer causes the mid-level meridional outflow. Observed meridional inflow above the melting level, on the other hand, is considered to be attributable to a reboost of the convection due to the latent heating of freezing, which is suggested by Williams and Renno (1993) and Zipser (2003). After all over ocean, larger correlation with convection of lower tropospheric humidity than with CAPE or CIN, detrainment effects of midlevel stable layer besides the trade-wind inver-

168


sion layer, and large-scale mid-level meridional divergent and convergent flows in addition to the deep Hadley cell are analyzed. These three points are individually presented by previous studies. We would like to emphasize that we not only statistically confirmed them with longterm R / V sounding data, but also showed their close mutual relationships by applying a diagnostic cloud model to the observed soundings. We showed that in the real oceanic data over the warm water pool, lower-tropospheric relative humidity (RHlt) actually varies in concert with the trade-inversion level and the mid-level atmospheric stability, because a coexistence of these three conditions is the manifestation of atmospheric subsidence. Through RB92 model, we showed that these larger trade-inversion level and melting-level stability and lower RHlt all work in the same sense with the cumulus convection, and realizes the dominant trimodal cumulus convection observed in TOGA-COARE (Johnson et al. 1999). It is also interesting to find that the same condition is not necessarily observed with the wet-season continental data. Extended analyses are expected to further clarify the relationship among atmospheric thermodynamic conditions, cumulus convective systems, and large-scale circulations suggested here. Analyzing the cloud resolving model in comparison with the real data may also shed light on further understandings of such phenomena. Acknowledgements The authors would like to acknowledge Mr. Kiyotoshi Takahashi at MRI for providing GMS TBB-IR data. We would like to express our sincere gratitude to cruise of JAMSTEC R / V ‘Mirai’, ‘Natsushima’ and ‘Kaiyo’, and members of GAME-tropics for all their efforts to obtain the valuable observation data. On the process of revising the manuscript, authors are indebted to constructive comments by two anonymous reviewers, as well as sincere help by Dr. Shuichi Mori as an editor. This work is partially supported by Grant-in-Aid for Scientific Research on Priority Area-764 of the Ministry of Education, Culture, Sports, Science and Technology of Japan, and also by Grant-in-Aid for Scientific Research by Japan Society for the Promotion of Science.

Vol. 84A

References Arakawa, A., and W. Schubert, 1974: Interaction of a cumulus cloud ensemble with the large-scale environment. J. Atmos. Sci., 31, 674–701. Betts, A. K., 1982: Saturation point analysis of moist convective overturning. J. Atmos. Sci., 39, 1484–1505. Blyth, A. M., W. A. Cooper, and J. B. Jensen, 1988: A Study of the Source of Entrained Air in Montana Cumuli. J. Atmos. Sci., 45, 3944–3964. Emanuel, K. A., 1986: An air-sea interaction theory for tropical cyclones. Part I: Steady-state maintenance, J. Atmos. Sci. 43, 585–604. Fu, R., A. D. Del Genio, and W. B. Rossow, 1994: Influence of Ocean Surface Conditions on Atmospheric Vertical Thermodynamic Structure and Deep Convection. J. Climate, 7, 1092– 1108. Fukao, S., 2006: Coupling processes in the equatorial atmosphere (CPEA): A project overview. J. Meteor. Soc. Japan, this issue. Houghton, T., Y. Ding, D. J. Griggs, M. Noguer, P. J. van der Linden, and D. Xiaosu (Eds.), 2001: The Scientific Basis. Contribution of Working Group I to the Third Assessment Report of the Intergovernmental Panel on Climate Change (IPCC). Cambridge University Press, U.K. pp. 944. Johnson, R. H., P. E. Ciesielski, and K. A. Hart, 1996: Tropical Inversions near the 0 C Level. J. Atmos. Sci., 53, 1838–1855. ———, T. M. Rickenbach, S. A. Rutledge, P. E. Ciesielski, and W. H. Schubert, 1999: Trimodal Characteristics of Tropical Convection. J. Climate, 12, 2397–2418. Kikuchi, K., and Y. N. Takayabu, 2004: The development of organized convection associated with the MJO during TOGA COARE IOP: Trimodal characteristics. Geoph. Res. Lett., 31, L10101. Kingsmill, D. E., and R. A. Houze, Jr., 1999: Thermodynamic characteristics of air flowing into and out of precipitating convection over the west Pacific warm pool. Q. J. R. Meteorol. Soc., 125, 1209–1229. 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. Manabe, S., and R. F. Strickler, 1964: Thermal equilibrium of the atmosphere with a convective adjustment, J. Atmos. Sci., 21, 361–385. Mapes, B. E., 2001: Water’s two height scales: The moist adiabat and the radiative troposphere., Quart. J. Roy. Meteor. Soc., 127, 2353–2366.

July 2006


———, P. E. Ciesielski, and R. H. Johnson, 2003: Sampling Errors in Rawinsonde-Array Budgets. J. Atmos. Sci., 60, 2697–2714. Nakazawa, T., 1988: Tropical super clusters within intraseasonal variations over the western Pacific. J. Meteor. Soc. Japan, 66, 823–839. Raymond, D. J., and A. M. Blyth, 1992: Extension of the Stochastic Mixing Model to Cumulonimbus Clouds. J. Atmos. Sci., 49, 1968–1983. ———, G. B. Raga, C. S. Bretherton, J. Molinari, C. Lopez-Carrillo, and Z. Fuchs, 2003: Convective Forcing in the Intertropical Convergence Zone of the Eastern Pacific. J. Atmos. Sci., 60, 2064–2082. Redelsperger, J.-L., D. B. Parsons, and F. Guichard, 2002: Recovery Processes and Factors Limiting Cloud-Top Height following the Arrival of a Dry Intrusion Observed during TOGA COARE. J. Atmos. Sci., 59, 2438–2457. Renno, N. O., and A. P. Ingersoll, 1996: Natural Convection as a Heat Engine: A Theory for CAPE. J. Atmos. Sci., 53, 572–585. Riehl, H., and J. S. Malkus, 1958: On the heat balance of the equatorial trough zone. Geophysica, 6, 503–538. ———, and J. Simpson, 1979: The heat balance of the equatorial trough zone, revisited. Beitr. Phys. Atmos., 52, 287–305. Sherwood, S. C., 1999: Convective precursors and predictability in the tropical western Pacific. Mon. Wea. Rev., 127, 2977–2991. Sobel, A. H., S. E. Yuter, C. S. Bretherton, and G. N. Kiladis, 2004: Large-Scale Meteorology and Deep Convection during TRMM KWAJEX. Mon. Wea. Rev., 132, 422–444.

169

Takayabu, Y. N., K.-M. Lau, and C.-H. Sui, 1996: Observation of a quasi-2-day wave during TOGA COARE. Mon. Wea. Rev., 124, 1892– 1913. ———, 2002: Spectral representation of rain profiles and diurnal variations observed with TRMM PR data over the equatorial area. Geophys. Res. Lett., 29, 10.1029/2001GL014113. Williams, E., and N. Renno, 1993: An analysis of the conditional instability of the tropical atmosphere. Mon. Wea. Rev., 121, 21–36. Xu, K.-M., and K. A. Emanuel, 1989: Is the Tropical Atmosphere Conditionally Unstable?. Mon. Wea. Rev., 117, 1471–1479. Yoneyama, K., 2003: Moisture Variability over the Tropical Western Pacific Ocean. J. Meteor. Soc. Japan. 81, 317–337. Yuter, S. E., and Houze, Jr., R. A., 1999: The 1997 Pan American Climate Studies Tropical Eastern Pacific Process Study. Part I: ITCZ region. Bull. Amer. Meteor. Soc. 81, 451–482. Zhang, C., and M. McGauley, 2004: Shallow Meridional Circulation in the Tropical Eastern Pacific. J. Climate, 17, 133–139. Zipser, E. J., and M. A. LeMone, 1980: Cumulonimbus vertical velocity events in GATE. Part II: Synthesis and model core structure. J. Atmos. Sci., 37, 2458–2469. ———, 2003: Some Views On ‘‘Hot Towers’’ after 50 Years of Tropical Field Programs and Two Years of TRMM Data. Meteor. Monographs, 29, 49–58. Zuidema, P., 1998: The 600–800-mb Minimum in Tropical Cloudiness Observed during TOGA COARE. J. Atmos. Sci., 55, 2220–2228.