• Bretherton, C. S., , and P. K. Smolarkiewics, 1989: Gravity waves, compensating subsidence and detrainment around cumulus clouds. J. Atmos. Sci., 46 , 740759.

    • Search Google Scholar
    • Export Citation
  • Hogan, R. J., , C. Jakob, , and A. J. Illingworth, 2001: Comparison of ECMWF winter-season cloud fraction with radar-derived values. J. Appl. Meteor., 40 , 513525.

    • Search Google Scholar
    • Export Citation
  • Horie, H., , T. Iguchi, , H. Hanado, , H. Kuroiwa, , H. Okamoto, , and H. Kumagai, 2000: Development of a 95-GHz airborne cloud profiling radar (SPIDER): Technical aspects. IEICE Trans. Commun., E83-B , 20102020.

    • Search Google Scholar
    • Export Citation
  • Johnson, R. H., , P. E. Ciesielski, , and K. A. Hart, 1996: Tropical inversions near the 0°C level. J. Atmos. Sci., 53 , 18381855.

  • Johnson, R. H., , T. M. Rickenbach, , S. A. Rutledge, , P. E. Ciesielski, , and W. H. Schubert, 1999: Trimodal characteristics of tropical convection. J. Climate, 12 , 23972418.

    • Search Google Scholar
    • Export Citation
  • Lin, J., , B. Mapes, , M. Zhang, , and M. Newman, 2004: Stratiform precipitation, vertical heating profiles, and the Madden–Julian oscillation. J. Atmos. Sci., 61 , 296309.

    • Search Google Scholar
    • Export Citation
  • Mapes, B. E., 2001: Water’s two height scales: The moist adiabat and the radiative troposphere. Quart. J. Roy. Meteor. Soc., 127 , 23532366.

    • Search Google Scholar
    • Export Citation
  • Mapes, B. E., , and R. A. Houze Jr., 1995: Diabatic divergence profiles in western Pacific mesoscale convective systems. J. Atmos. Sci., 52 , 18071828.

    • Search Google Scholar
    • Export Citation
  • Mapes, B. E., , and P. Zuidema, 1996: Radiative–dynamical consequences of dry tongues in the tropical troposphere. J. Atmos. Sci., 53 , 620638.

    • Search Google Scholar
    • Export Citation
  • Raymond, D. J., , and A. M. Blyth, 1992: Extension of the stochastic mixing model to cumulonimbus clouds. J. Atmos. Sci., 49 , 19681983.

    • Search Google Scholar
    • Export Citation
  • Sassen, K., 1977: Lidar observations of High Plains thunderstorm precipitation. J. Atmos. Sci., 34 , 14441457.

  • Sassen, K., , and B. S. Cho, 1992: Subvisual-thin cirrus lidar dataset for satellite verification and climatological research. J. Appl. Meteor., 31 , 12751285.

    • Search Google Scholar
    • Export Citation
  • Sassen, K., , and T. Chen, 1995: The lidar dark band: An oddity of the radar bright band analogy. Geophys. Res. Lett., 22 , 35053508.

  • Sassen, K., , J. R. Campbell, , J. Zhu, , P. Kollias, , M. Shupe, , and C. Williams, 2005: Lidar and triple-wavelength Doppler radar measurements of the melting layer: A revised model for dark- and brightband phenomena. J. Appl. Meteor., 44 , 301. 312.

    • Search Google Scholar
    • Export Citation
  • Steiner, M., , R. A. Houze Jr., , and S. E. Yuter, 1995: Climatological characterization of three-dimensional storm structure from operational radar and rain gauge data. J. Appl. Meteor., 34 , 19782007.

    • Search Google Scholar
    • Export Citation
  • Stewart, R. E., , J. D. Marwitz, , J. C. Pace, , and R. E. Carbone, 1984: Characteristics through the melting layer of stratiform clouds. J. Atmos. Sci., 41 , 32273237.

    • Search Google Scholar
    • Export Citation
  • Sugimoto, N., , I. Matsui, , and A. Shimizu, 2000: Climatological characteristics of cloud distribution and planetary boundary layer structure in Jakarta, Indonesia revealed by lidar observation. Geophys. Res. Lett., 27 , 29092912.

    • Search Google Scholar
    • Export Citation
  • Sugimoto, N., , I. Matsui, , Z. Liu, , A. Shimizu, , and K. Asai, 2001: Latitudinal distribution of aerosols and clouds in the western Pacific observed with a lidar on board the research vessel Mirai. Geophys. Res. Lett., 28 , 41874190.

    • Search Google Scholar
    • Export Citation
  • Szyrmer, W., , and I. Zawadzki, 1999: Modeling of the melting layer. Part I: Dynamics and microphysics. J. Atmos. Sci., 56 , 35733592.

  • Yasunaga, K., , H. Kida, , and T. Satomura, 2003: The 600–750 hPa relative humidity minimum observed during PEM-Tropics B. Geophys. Res. Lett., 30 .2282, doi:10.1029/2003GL018739.

    • Search Google Scholar
    • Export Citation
  • Zuidema, P., 1998: The 600–800-mb minimum in tropical cloudiness observed during TOGA COARE. J. Atmos. Sci., 55 , 22202228.

  • View in gallery

    Time–longitude cross section of infrared equivalent blackbody temperature. Data are from the Geostationary Meteorological Satellite of the Japan Meteorological Agency. Dashed line indicates the longitude of R/V Mirai.

  • View in gallery

    Time–height cross sections of (a) zonal and (b) meridional wind components derived from radiosonde. Observations were conducted every 3 h from 0000 UTC 9 November until 0000 UTC 9 December. Solid and dashed contours indicate positive and negative values, respectively. Periods of no data are left blank.

  • View in gallery

    Cumulative frequency of stability (dT/dZ) for values greater than −5 (solid lines), −4.5 (dashed lines), and −4 K km−1 (dotted lines) during the (a) IOP, (b) inactive period, and (c) active period. Thin dotted line in each panel indicates the 0°C level.

  • View in gallery

    Evolution of the coverage of C-band radar echo for reflectivity factors >15 dBZ at 2 km height over a 200 km × 200 km area during the IOP (unit: km2).

  • View in gallery

    Time–height cross section of (a) 532-nm lidar backscattering coefficient, (b) 1064-nm lidar backscattering coefficient, (c) radar reflectivity factor, and (d) ultimate cloud base (Zb) distributions. Crosses in the upper three figures indicate the levels where the backscattering coefficient and reflectivity first exceeded certain thresholds during the vertical scan from the height of 2 km (see text for details). In (d), the area with high relative humidity (>80%) is shaded by grayscale.

  • View in gallery

    TSI images taken at (a) 0611 UTC and (b) 2211 UTC 10 November 2001.

  • View in gallery

    Frequency distribution of the base heights of thin cloudy layers (thick solid lines) during the (a) IOP, (b) inactive period, and (c) active period. Error bars represent the ranges within one standard deviation calculated from 30 samples. Cumulative frequency of stability (dT/dZ) for values greater than −4.5 K km−1 (see Fig. 3) and the 0°C level are overplotted in each panel with thin solid and dotted lines, respectively.

  • View in gallery

    Time–height cross section of (a) 532-nm lidar backscattering coefficient, (b) 1064-nm lidar backscattering coefficient, (c) radar reflectivity factor, and (d) ultimate cloud base (Zb) distributions on 24 November 2001. In (d), the area with high relative humidity (>80%) is shaded by grayscale.

  • View in gallery

    As in Fig. 8 except on 4 December 2001.

  • View in gallery

    TSI images taken at (a) 0041 UTC 24 November and (b) 0711 UTC 4 December 2001.

  • View in gallery

    Variations in the composite relative humidity of eight precipitation events during the active period. Time = 0 corresponds to the time when the coverage of convective type radar echoes at 2 km was at a maximum. Dashed line indicates the 0°C level.

  • View in gallery

    As in Fig. 11 except for the composite temperature lapse rate.

  • View in gallery

    Frequency distributions of Zl (solid lines) and Zr (dashed lines) during the (a) IOP, (b) inactive period, and (c) active period. The frequency of Zr is doubled in order to easily compare with the frequency of Zl. Thin dotted line in each panel indicates the 0°C level.

  • View in gallery

    Variations in (a) Zl, and (b) Zb of the eight precipitation events during the active period. Each color represents one event. Time = 0 corresponds to the time when the coverage of convective-type radar echoes at 2 km was at a maximum. Dashed line indicates the 0°C level.

  • View in gallery

    Vertical profile of the lidar backscattering coefficient (solid line) and radiosonde-derived relative humidity (dashed line) at 1200 UTC 29 November 2001.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 25 25 11
PDF Downloads 15 15 6

Melting Layer Cloud Observed during R/V Mirai Cruise MR01-K05

View More View Less
  • * Institute of Observational Research for Global Change, JAMSTEC, Yokosuka, Kanagawa, Japan
  • + Center for Atmospheric and Oceanic Studies, Graduate School of Science, Tohoku University, Sendai, Japan
  • # National Institute for Environmental Studies, Tsukuba, Ibaraki, Japan
  • @ National Institute of Information and Communications Technology, Tokyo, Japan
© Get Permissions
Full access

Abstract

In this study, cloud profiling radar and lidar were used to determine the frequency distribution of the base heights of cloudy layers with little (or no) falling condensate particles. The data were obtained from stationary observations conducted from Research Vessel Mirai over the tropical western Pacific (around 1.85°N, 138°E) from 9 November to 9 December 2001. The observed cloudy layers had base heights predominantly in the range of 4.5–6.5 km. Almost all cloudy layers with a base in the range of 4.5–6.5 km had thickness thinner than 500 m, and the frequency peak of the base heights of measured cloudy layers is considered to represent the common occurrence of midlevel thin clouds.

Midlevel thin clouds were frequently observed even during the active phase of the Madden–Julian oscillation (MJO). Composite analysis of radiosonde-derived relative humidity and temperature lapse rate indicates that the midlevel thin cloud in the MJO active period is generated via melting within the stratiform cloud, rather than by detrainment of surface-based convection.

Corresponding author address: Kazuaki Yasunaga, Institute of Observational Research for Global Change, JAMSTEC, Yokosuka Headquarters, 2-15 Natsushima-Cho, Yokosuka-city, Kanagawa, 237-0061, Japan. Email: yasunaga@jamstec.go.jp

Abstract

In this study, cloud profiling radar and lidar were used to determine the frequency distribution of the base heights of cloudy layers with little (or no) falling condensate particles. The data were obtained from stationary observations conducted from Research Vessel Mirai over the tropical western Pacific (around 1.85°N, 138°E) from 9 November to 9 December 2001. The observed cloudy layers had base heights predominantly in the range of 4.5–6.5 km. Almost all cloudy layers with a base in the range of 4.5–6.5 km had thickness thinner than 500 m, and the frequency peak of the base heights of measured cloudy layers is considered to represent the common occurrence of midlevel thin clouds.

Midlevel thin clouds were frequently observed even during the active phase of the Madden–Julian oscillation (MJO). Composite analysis of radiosonde-derived relative humidity and temperature lapse rate indicates that the midlevel thin cloud in the MJO active period is generated via melting within the stratiform cloud, rather than by detrainment of surface-based convection.

Corresponding author address: Kazuaki Yasunaga, Institute of Observational Research for Global Change, JAMSTEC, Yokosuka Headquarters, 2-15 Natsushima-Cho, Yokosuka-city, Kanagawa, 237-0061, Japan. Email: yasunaga@jamstec.go.jp

1. Introduction

The vertical distribution of radiatively active tracers, aerosols, and clouds influences the earth’s climate via the radiative properties of the atmosphere. Of these factors, clouds have the largest radiation-forcing properties. At low latitudes, the shortwave flux incident at the top of the atmosphere and longwave radiation flux emitted from the earth surface are large. The vertical distribution of tropical clouds is therefore an important component for understanding the energy balance of the earth.

Over regions of warm tropical sea surface temperatures (SSTs), the atmosphere is conditionally unstable, and many isolated or highly organized convections occur. Johnson et al. (1996) described prominent stable layers at heights of 2, 5, and 15–16 km, observed during the Tropical Ocean Global Atmosphere Coupled Ocean–Atmosphere Response Experiment (TOGA COARE). Johnson et al. (1999) also found that maxima in the vertical distributions of radar-echo (cloud) tops exist in the vicinity of these three stable layer heights.

It is a well-known fact that environmental static stability influences the vertical profiles of detrainment from cumulus convection as well as the top heights of cumulus convection. On the basis of numerical and analytical models, Bretherton and Smolarkiewicz (1989) proposed that mesoscale circulation accompanying deep convection is associated with a gravity wave response and that mesoscale circulation enhances the detrainment of cloud air at levels where the cloud buoyancy decreases. If we apply the Bretherton and Smolarkiewicz (1989) model to the mean temperatures recorded during the TOGA COARE, detrainment from convective clouds is predicted to occur at approximately 500 hPa. Zuidema (1998) applied the buoyancy-sorting cloud model of Raymond and Blyth (1992) to the Kapingamarangi soundings gathered during the TOGA COARE and concluded that detrainment from cloud ensembles is enhanced at around the 400-hPa and 800-hPa levels and inhibited at around the 600-hPa level. Mapes (2001) applied a one-dimensional single-column model to the TOGA COARE mean temperature profiles and concluded that diabatic divergence (Mapes and Houze 1995) occurs at around the 200-, 600-, and 900-hPa levels within heated regions in order to maintain existing temperature profiles in the column.

The findings of the three studies summarized above raise the possibility that detrainment from convective clouds over warm tropical oceans is enhanced at specific pressure levels, although the three studies provide contrasting estimates of the exact detrainment levels. Yasunaga et al. (2003) performed a statistical analysis of the NASA Pacific Exploratory Mission-Tropics B (PEM-Tropics B) observational data and found that tropical cumulus convection directly transports boundary layer air to the 350–600-hPa and 750–800-hPa levels. These pressure levels coincide with the heights where stable layers were frequently observed during the PEM-Tropics B. It is therefore reasonable to expect that clouds detrained from cumulus convection (detrainment shelves of Mapes and Zuidema 1996) would be frequently observed at around the specific layers of 2, 5, and 15–16 km as well as cloud tops.

Previous studies have used radiosonde-derived relative humidity profiles to indirectly infer vertical cloud distributions (e.g., Mapes and Zuidema, 1996; Zuidema, 1998). Mapes and Zuidema (1996) discovered a high occurrence of cloudy layers around the melting level and deduced that the cloudy layer corresponds to detrainment shelves enhanced by the 0°C stable layer (Fig. 18 in Mapes and Zuidema 1996). The peak distribution of cloudy layers at the melting level is not recognizable, however, when the relative humidity is calculated with respect to ice above the 0°C level. It is completely arbitrary as to whether the relative humidity is calculated with respect to water or with respect to ice at ∼0°C. Although these previous studies provide strong indirect evidence for melting-level cloudiness, direct evidence is still lacking.

We consider that detrainment shelves are thin cloudy layers composed of small condensate particles of negligible terminal velocity (<0.1 mm in radius). First, in the present study, the frequency distribution of base height of clouds without falling condensate particles (>0.1 mm in radius) is examined, making use of 95-GHz cloud profiling radar (wavelength of 3.16 mm) and lidar with a dual wavelength of 532 and 1064 nm. Second, thickness of the cloud is estimated with lidar backscattering coefficient and radiosonde-derived relative humidity.

Lidar (or ceilometer) utilizes wavelengths around the region of the visible spectrum, and the size of condensate particles corresponds to the Mie or geometrical optics region. The number density of cloud droplets (<0.1 mm in radius) is generally much larger than that of rain droplets (>0.1 mm in radius); lidar is therefore more sensitive to cloud droplets than rain droplets. However, in the case of an optically thick cloud, lidar cannot observe structure above the cloud due to strong attenuation. At the very least, lidar can determine the base height of cloudy layers by means of the threshold method in which an object is defined as a cloud when a rising edge in the range-corrected signal exceeds a certain threshold (Sassen and Cho 1992; Sugimoto et al. 2000). However, lidar alone is unable to distinguish a cloudy layer with little or no falling condensate particles from a cloudy layer with falling condensate particles.

Radar is more sensitive than lidar in terms of detecting falling condensate particles. The size of cloud droplets corresponds to the Rayleigh region, where the backscattering cross-section area of a sphere is proportional to the sixth power of diameter. Therefore, a thin cloudy layer without falling condensate particles is commonly transmitted with little reflectivity, even if the radar utilizes a wavelength of just several millimeters. Taking into account the characteristics of lidar and radar, it is possible to detect the base height of cloudy layers that contain little (or no) falling condensate particles by excluding the cases with a large value of radar reflectivity from those cases with a strong lidar backscatter signal.

In section 2, we present a brief description of data and meteorological conditions. Details of analytical procedures are described in section 3, and the frequency distribution of base height of clouds without falling condensate particles is presented in section 4. In these sections, it is also demonstrated that thickness of the cloud is thin enough, like detrainment shelves, that the cloud-base height can be regarded as the height of the cloudy layer. In section 5, we compare the frequency distributions of the thin cloudy layer with those obtained by previous works, and suggest the possibility that the peak frequency of the thin cloudy layer in the midlevel atmosphere is associated with cloud microphysics (melting process) within stratiform clouds.

2. Data and meteorological conditions during MR01-K05 Leg 3–4 cruise

The data utilized in the present study were obtained by stationary observation over the tropical western Pacific (around 1.85°N, 138°E) from 9 November to 9 December 2001 by the research vessel (R/V) Mirai of the Japan Marine Science and Technology Center (JAMSTEC). During the intensive observation period (IOP), which is divided into two periods (active and inactive) as mentioned later in the present section, the principal mission objectives were C-band weather Doppler radar observations, atmospheric sounding by radiosonde, surface meteorological measurement, conductivity–temperature–depth profiler castings to 500 m, and current measurement by acoustic Doppler current profiler. Additional tasks included turbulent flux measurement, solar radiation measurement, cloud profiling radar (SPIDER), lidar, and greenhouse gas measurement.

The cloud profiling radar (SPIDER) has a frequency of 95 GHz (wavelength 3.16 mm). The vertical resolution of the radar data is 82.5 m, with the maximum altitude of the observation being 20 km, with a time resolution of about 0.1 s. The radar reflectivity factor is commonly indicated in the logarithmic form (dBZe), and the minimum detectable radar reflectivity factor averaged over 1 min was −43 dBZe at 5 km, which corresponds to the ice water content of about 10−4 g m−3 (ice water path of 10−3 g m−2 for the cloud thickness of 1 km), using the Z–IWC relationship based on Hogan et al. (2001). A full description of the cloud profiling radar SPIDER is provided by Horie et al. (2000).

The lidar utilized in the present study is a two-wavelength (532 nm and 1064 nm) Mie-scattering lidar with a depolarization measurement capability of 532 nm. The original data have a vertical resolution of 6 m and a temporal resolution of 10 s. The maximum altitude of observations is 20 km. Technical specifications of the lidar are provided by Sugimoto et al. (2001), while additional details concerning the shipborne cloud profiling radar and lidar system are provided by Okamoto et al. (2005, manuscript submitted to J. Geophys. Res.).

Figure 1 shows the infrared equivalent blackbody temperature (TBB) measured during the IOP by the Geostationary Meteorological Satellite of the Japan Meteorological Agency. Westward-propagating and eastward-propagating convective areas reach the site in the first and latter half of the IOP, respectively. According to Madden–Julian oscillation (MJO) monitoring by the National Oceanic and Atmospheric Administration/Climate Diagnostics Center (http://www.cdc.noaa.gov/), eastward-propagating convective areas in the late IOP are identified as MJO signals. In the present study, we use satellite TBB data to simply define the intraseasonal convectively active period from the first entry of the eastward-moving large-scale low TBB (<250 K) area into the observational site; other periods are defined as inactive (9–28 November: inactive period; 29 November–9 December: active period). Therefore, it should be noted that westward-propagating convective areas in the first half of the IOP are included in the inactive period.

Figure 2 presents time–height cross sections of zonal and meridional winds derived from radiosonde data. Westerly winds are dominant in the lower troposphere during the entire observational period (Fig. 2a). There are two peaks of strong westerly wind during the middle of November and early December, each time accompanied by cloud activities. The meridional wind component shows the periodic replacement of positive and negative values between the 100-hPa and 300-hPa levels (Fig. 2b), which would indicate that equatorial trapped waves prevailed during this period. The periodic variations of the meridional wind are also seen in the midlevel and would be affected by the upper-level meridional wind variations.

The frequency distribution of temperature lapse rates exceeding the thresholds of −4, −4.5, and −5 K km−1 are shown in Fig. 3. The results presented in Fig. 3 were calculated for a depth of 500 m. During the IOP, stable layers were frequently observed at the heights of 2 and 5.5 km, while weakened stability layers were detected at a height of around 4.5 km (Fig. 3a). The frequency distribution of the stable layer during the IOP is similar to that recorded during TOGA COARE (e.g., Johnson et al. 1996).

The frequency peak of the stable layers at the 2-km height was more prominent in the inactive period than in the active period (Figs. 3b,c). The frequency peak of the stable layer at midlevels is more pronounced in the active period, although the height of the peak is lower in the active period (5 km) than in the inactive period (6 km). While a layer of weakened stability is observed around 4.5-km height during both periods, the layer is much shallower in the active period.

Figure 4 shows variation in the coverage of C-band radar echoes with reflectivity factors >15 dBZ at 2 km height over a 200 × 200 km area. Radar echo was classified as either convective or stratiform type, following the method by Steiner et al. (1995). It should be noted that vertical scale is large in the lowest panel in Fig. 4.

The characteristics of precipitation differ between the active and inactive period. During the inactive period (9–28 November) precipitation occurs almost daily, and convective type echoes are dominant except on 9–10 and 23 November. In the (MJO) active period (29 November–9 December) coverage of radar echoes is much larger than in the inactive period, and stratiform type echoes are dominant, which is consistent with the results of Lin et al. (2004).

3. Data analysis and procedures

Radar and lidar data were interpolated to produce consistent vertical and temporal resolution. The combined data used in the present analysis has a vertical resolution of 82.5 m and a time resolution of 1 min. The base height of cloudy layers with little or no falling condensate particles was determined via the radar reflectivity factor and lidar backscattering coefficient according to the following three steps.

  1. The provisional base level of the cloudy layer (Zl) is defined as the level at which the lidar backscattering coefficient first exceeds a certain threshold (Lc) during a vertical scan from the height of 2 km. The lidar has two wavelengths, and the lower level is selected as Zl. The starting height of the vertical scan, 2 km, is selected in order to avoid aerosols and clouds within the boundary layer.
  2. The provisional base level of the cloudy layer (Zr) is defined as the level at which the radar reflectivity factor first exceeds a certain threshold (Rc) during a vertical scan from a height of 2 km.
  3. The ultimate base level of the cloudy layer (Zb) is defined by Zl for the case where the height of Zl is less than that of Zr, or where Zl is defined and Zr cannot be determined; Zb is considered to be indefinite in the case of the height of Zr being equal to or lower than that of Zl, or when Zl cannot be determined.

Radar is more sensitive than lidar in terms of falling condensate particles (>0.1 mm in radius), and is able to detect such large particles at lower levels than lidar, provided an adequate Rc value is chosen. Therefore, when the height of Zr is equal to or lower than that of Zl, we can conclude that falling condensate particles exist. This forms the basis of defining Zb as indefinite in the third step outlined above. Conversely, lidar is more sensitive to small particles of negligible terminal velocity (<0.1 mm in radius) than radar. Accordingly, when the height of Zl is lower than that of Zr, or when Zr is indefinite, Zl can be considered to represent the base height of a cloudy layer with little or no falling condensate particles.

Figure 5 shows observations made on 10 November 2001. In this case, Lc and Rc are −5.25 [= log10 β, where β (m−1 sr−1) is the backscattering coefficient] and −35 dBZe, respectively. During 0100–0200 UTC, the lidar backscattering coefficient is large around the height of 5 km, while high radar reflectivity is recorded at lower levels. These observations indicate that falling condensate particles exist and that Zb is defined as indefinite (see the lowest panel in Fig. 5). In contrast, during 0400–0500 UTC and 2000–2200 UTC, lidar detects thin cloudy layers at midlevels (5–7 km), while the cloud profiling radar detects only weak (or no) signals. Although attenuation effects prevent lidar from observing structure above the optically thick cloud, the high relative humidity areas indicate the reality of the cloud thickness observed with lidar. Images from the total sky imager (TSI), a full-color digital imaging and software system designed to automatically monitor cloud conditions, support the existence of thin cloudy layers (Figs. 6a and 6b). While the TSI cannot identify the height of the cloud, neither lidar nor cloud profiling radar showed strong signals above the height of 8 km. Therefore, the thin cloudy layers in the Figs. 6a and 6b would be located at the midlevels with large backscattering coefficients of lidar (Fig. 5). This example clearly demonstrates that the thickness of the cloud with little falling condensate particles detected in the present analysis procedure is thin.

4. Results

Figure 7 shows the frequency distribution of the measured base heights of cloudy layers; Lc varies from −4.25 to −5.25 [= log10 β, where β (m−1 sr−1) is the backscattering coefficient] at 0.25 intervals, while Rc varies from −10 to −35 dBZe at 5-dBZe intervals. Values averaged over 30 (5 × 6) samples are displayed in Fig. 7. The error bars indicate the ranges within one standard deviation as calculated from the 30 samples.

A peak exists between the heights of 4.5 and 6.5 km during the IOP (Fig. 7a). During the inactive period, when coverage of the convective type echoes were as large as that of the stratiform type echoes (Fig. 4), two peaks are found around the heights of 6 and 7 km (Fig. 7b). The peaks coincide with the height of the frequent occurrence of the stable layer. In the active period, when coverage of the stratiform type echoes was much greater than convective type echoes (Fig. 4), there are two peaks around the heights of 5 and 6 km (Fig. 7c). The levels of both peaks also agree with the height of the common occurrence of the stable layer. Although the characteristics of precipitation are different in the inactive and active periods, the peak between the heights of 4.5 km and 6.5 km is apparent in both periods. The midlevel peak apparent in Fig. 7a, therefore, does not reflect a specific event. The frequent occurrence of the base of cloudy layers around 4.5–6.5-km height is presumably a general phenomenon over warm tropical oceans. The heights of the peaks in Fig. 7 do not vary when the starting height of the vertical scan is changed from 0 to 2 km.

Figures 8 and 9 show examples in which the base of cloudy layers was observed between the heights of 4.5 and 6.5 km. In Fig. 8, the base of the cloudy layers occurs at midlevels during 0300–0400, 1000–1400, and 1900–2100 UTC. The thickness of observed cloudy layers is relatively thin. Figure 9 shows the occurrence of midlevel thin cloudy layers for almost the entire day. TSI images and radiosonde-derived relative humidity support the existence and thin thickness of the cloudy layers detected with the present analysis procedure (Figs. 8d, 9d, 10a and 10b).

When clouds are identified as objects with a lidar backscattering coefficient exceeding a certain threshold, almost all of the clouds whose base heights (Zb) are detected between 4.5 and 6.5 km height had a thickness smaller than 500 m (Table 1). Clouds are also identified with relative humidity derived from radiosonde, which was launched every 3 h, because attenuation effects prevent lidar from observing structure above optically thick cloud. In the relative humidity case, 60%–70% of the clouds have thickness less than 500 m (Table 2). Thick cloudy layers, therefore, are unlikely to exist above the cloud bases determined with the analytical method described in the present study (Zb). As a result, the frequency peak of the base heights of midlevel cloudy layers shown in Fig. 7 can be regarded as representing the frequent occurrence of midlevel thin clouds. This interpretation is also consistent with the vertical cloud distribution inferred from radiosonde-derived relative humidity data (Mapes and Zuidema 1996).

5. Discussion

The peak of the frequency distribution of thin cloudy layers (Fig. 7) is located near the stable layer often observed (Fig. 3). Accordingly, part of the thin cloud probably represents detrainment shelves from the convective cloud promoted by the stable layer. The upper peak in the inactive period (Fig. 7b) agrees with the detrainment level estimated by Zuidema (1998; 400 hPa). The lower peak in the inactive period (Fig. 7b) and the upper peak in the active period (Fig. 7c) are consistent with the Bretherton and Smolarkiewicz (1989) model (500 hPa). The lower peak in the active period (Fig. 7c) coincides with the detrainment level estimated by Mapes (2001; 600 hPa). The peak levels of the frequency distribution of thin cloudy layers observed in the present study, therefore, seem to support all results obtained in the previous works.

On the other hand, the midlevel thin cloudy layer was frequently observed even during the MJO active phase (Fig. 7c), when coverage of the stratiform type echoes was much greater than convective type echoes (Fig. 4). Moreover, the lower peak in the active period is located just at the 0°C level. We, therefore, discuss the possibility that the lower peak in the active period is associated with cloud microphysics (melting process) within the stratiform cloud rather than detrainment from the convective cloud.

Condensation can occur within the layer that contains melting ice particles, as melting is generally independent of relative humidity and diabatic cooling due to melting enhances relative humidity. Szyrmer and Zawadzki (1999) used a numerical model to demonstrate that the nonuniformity of snow content causes horizontal variability in various atmospheric properties within the melting layer, which in turn leads to the generation of convective cells. In addition, Stewart et al. (1984) took measurements from an aircraft flown through the melting layer of stratiform clouds over the California Valley and documented an isothermal layer at 0°C and a large value of cloud liquid water content slightly below the 0°C layer. Accordingly, it is unlikely that the melting process within the stratiform cloud is able to account for the common occurrence of cloudy layers in the midlevel, especially at the 0°C level.

Figures 11 and 12 show variations in radiosonde-derived relative humidity and stability (dT/dZ) profiles, respectively. The data shown in Figs. 11 and 12 represent the composite of eight precipitation events during the active period. Time = 0 corresponds to the time when the coverage of convective type echoes of the C-band radar with the reflectivity factors larger than the value of 15 dBZ at 2 km within an area of 40 000 km2 (200 km × 200 km) reached a maximum. Therefore, time = 0 can be considered the peak time of the convective activity.

A high value of relative humidity occurs around the 600-hPa level between 2 and 12 h after the convection peak (Fig. 11). During the active period, coverage of the stratiform type echoes reached a maximum 3–6 h after the convection peak. The occurrence of the relative humidity peak at ∼600 hPa corresponds to the time when stratiform clouds were active. Moreover, enhanced- and weakened-stability layers are simultaneously found above and below the level of the high relative humidity (Fig. 12). The enhanced- and weakened-stability layers of high relative humidity at around 600-hPa level is consistent with the isothermal layer at 0°C documented by Stewart et al. (1984) within stratiform cloud, and is easily explained by the melting process. Johnson et al. (1996) also proposed that melting is responsible for the generation of a stable layer of high relative humidity. It is, therefore, reasonable to conclude that the high relative humidity around the 600-hPa level results from melting processes within the stratiform cloud.

Figure 13 shows the frequency distribution of the base heights of cloudy layers determined solely from lidar observations (Zl). It should be noted that the results shown in Fig. 13 include cases of cloud that contain falling condensate particles. The difference between Zb and Zl in Figs. 7 and 13 indicates that about half of the clouds around the melting level detected by lidar are accompanied with falling condensate particles. A prominent peak is apparent in the melting level during the IOP (Fig. 13a). The peak also appears in the active period (Figs. 13c) when stratiform precipitation was dominant (Fig. 4), but the peak is not observed during the inactive period when convective precipitation was dominant (Fig. 13b).

Figures 14a and 14b show variations in the base heights of Zl and Zb of the eight precipitation events during the active period, respectively. Time = 0 is also the peak time of convective activity, as is described in Figs. 11 and 12. While the coverage of the stratiform type echoes reached a maximum 3–6 h after the convection peak, Zl and Zb were frequently observed near the melting level at 7 and 9 h after the convection peak, respectively. The delay in the occurrence of the Zl and Zb near the melting level from the stratiform precipitation peak would result from the attenuation by rainfall. Although lidar cannot determine the structure above the optically thick cloud, the coincidence between the occurrence of Zl and Zb (Figs. 13 and 14) and the high relative humidity distribution around the melting layer (Fig. 11) indicates that condensation is promoted only at around 600 hPa. Namely, the peak evident in Figs. 13c indicates the existence of a “melting layer cloud” that is generated via melting processes within the stratiform cloud.

A lidar brightband phenomenon, however, occurs at visible frequencies when condensate particles fall through the melting level (Sassen 1977). According to Sassen and Chen (1995), the increase in particle size encountered by the laser pulse as it ascends through the melting region initially causes a rapid increase in backscattering followed by marked attenuation, which produces the returned energy peak known as the lidar brightband analog. This effect can result in an overestimate of the frequency of the cloudy layer around the melting level if estimates are based solely on lidar observations (Fig. 13). Despite this potential uncertainty, it is commonly observed that the peak of the backscattering coefficient in the melting level is in good agreement with high values of relative humidity (e.g., Fig. 15). Moreover, Sassen (1977) observed the lidar bright band at a temperature greater than 3.0°C in light continuous precipitation. The 3.0°C level corresponds to the height of 4410 m in the present case, and the brightband level is about 500 m lower than the peak level in Fig. 13c.

The frequency distribution of the base heights determined by cloud profiling radar (Zr) also show a minor midlevel peak (Fig. 13c), even though the peak height is slightly lower (500 m) than that by lidar. Recently, Sassen et al. (2005) reported that no clear examples of the bright band could be seen at the wavelength of 3.2 mm because of the dominance of non-Rayleigh scattering effects. Therefore, the peak of Zr would not reflect the bright band. The reason why the peak is less pronounced and located at lower level than the peak by lidar would be caused by the high sensibility of the cloud profiling radar to falling condensate particles.

From the discussion in the present section, it is unlikely that all of the high values of the lidar backscattering coefficient around the melting layer, which form the apparent peak in Fig. 13c, result from the lidar brightband phenomenon; we do not consider the peak to be entirely artificial.

Mapes and Houze (1995) documented diabatic convergence and divergence (reverberations) centered on the melting level. It is therefore possible that the melting layer cloud can be extended to a wider area due to the divergence, which would provide favorable conditions for the frequent observation of midlevel thin clouds.

6. Summary

In the present study, the frequency distribution of the base heights of cloudy layers was determined using 95-GHz cloud radar and dual wavelength lidar (532 and 1064 nm). The data were obtained from stationary observations conducted over the tropical western Pacific (approximately 1.85°N, 138°E) from 9 November to 9 December 2001 by the Research Vessel Mirai. The analytical method described in this study enabled the detection of cloudy layers composed of small falling condensate particles of negligible terminal velocity. This is achieved by the exclusion of those cases with large values of radar reflectivity from those cases with a strong lidar backscatter signal.

The frequency distribution of the base heights of cloudy layers has a peak between the heights of 4.5 and 6.5 km. Almost all cloudy layers with a base in the range 4.5–6.5 km have thickness thinner than 500 m. The frequency peak of the base heights is therefore considered to represent the common occurrence of midlevel thin clouds.

Midlevel thin clouds were frequently observed even during the MJO active phase when the coverage of stratiform type echoes was much greater than that of convective type echoes. If clouds with falling condensate particles are included in the analysis, the frequency distribution of cloud bases has a more prominent peak within the melting layer. Relative humidity was enhanced around the 600-hPa level at the time when stratiform clouds became active. Melting within the stratiform cloud is considered responsible for the enhancement of relative humidity around the 600-hPa level. The coincidence of levels that contain high relative humidity and a peak in cloud base frequency indicates that clouds are generated (and extended) around the 0°C level via melting processes within stratiform cloud (melting layer cloud).

The present study does not negate the possibility that most of the thin cloud detected in the midlevel are associated with detrainment shelves from convective clouds. We, however, consider that cloud microphysics within the stratiform cloud plays a central role in the generation of thin cloud around the 0°C level during the (MJO) active period, as coverage of the stratiform cloud is much greater than that of the convective cloud. To properly reproduce the MJO in a numerical model, it might be important to adequately consider the effects of the melting process.

Acknowledgments

We express our sincere thanks to Captain M. Akamine and his crew for their skillful operation of the R/V Mirai. Technical staffs of Global Ocean Development, Inc., are also acknowledged for their continuous support for meteorological observations. We greatly appreciate helpful comments of the anonymous reviewer. The authors also thank Mr. Nishi for helpful discussions. All of the figures were drawn using GrADS (http://www.iges.org/grads/) and GFD-DENNOU Library (http://www.gfd-dennou.org/index.html.en).

REFERENCES

  • Bretherton, C. S., , and P. K. Smolarkiewics, 1989: Gravity waves, compensating subsidence and detrainment around cumulus clouds. J. Atmos. Sci., 46 , 740759.

    • Search Google Scholar
    • Export Citation
  • Hogan, R. J., , C. Jakob, , and A. J. Illingworth, 2001: Comparison of ECMWF winter-season cloud fraction with radar-derived values. J. Appl. Meteor., 40 , 513525.

    • Search Google Scholar
    • Export Citation
  • Horie, H., , T. Iguchi, , H. Hanado, , H. Kuroiwa, , H. Okamoto, , and H. Kumagai, 2000: Development of a 95-GHz airborne cloud profiling radar (SPIDER): Technical aspects. IEICE Trans. Commun., E83-B , 20102020.

    • Search Google Scholar
    • Export Citation
  • Johnson, R. H., , P. E. Ciesielski, , and K. A. Hart, 1996: Tropical inversions near the 0°C level. J. Atmos. Sci., 53 , 18381855.

  • Johnson, R. H., , T. M. Rickenbach, , S. A. Rutledge, , P. E. Ciesielski, , and W. H. Schubert, 1999: Trimodal characteristics of tropical convection. J. Climate, 12 , 23972418.

    • Search Google Scholar
    • Export Citation
  • Lin, J., , B. Mapes, , M. Zhang, , and M. Newman, 2004: Stratiform precipitation, vertical heating profiles, and the Madden–Julian oscillation. J. Atmos. Sci., 61 , 296309.

    • Search Google Scholar
    • Export Citation
  • Mapes, B. E., 2001: Water’s two height scales: The moist adiabat and the radiative troposphere. Quart. J. Roy. Meteor. Soc., 127 , 23532366.

    • Search Google Scholar
    • Export Citation
  • Mapes, B. E., , and R. A. Houze Jr., 1995: Diabatic divergence profiles in western Pacific mesoscale convective systems. J. Atmos. Sci., 52 , 18071828.

    • Search Google Scholar
    • Export Citation
  • Mapes, B. E., , and P. Zuidema, 1996: Radiative–dynamical consequences of dry tongues in the tropical troposphere. J. Atmos. Sci., 53 , 620638.

    • Search Google Scholar
    • Export Citation
  • Raymond, D. J., , and A. M. Blyth, 1992: Extension of the stochastic mixing model to cumulonimbus clouds. J. Atmos. Sci., 49 , 19681983.

    • Search Google Scholar
    • Export Citation
  • Sassen, K., 1977: Lidar observations of High Plains thunderstorm precipitation. J. Atmos. Sci., 34 , 14441457.

  • Sassen, K., , and B. S. Cho, 1992: Subvisual-thin cirrus lidar dataset for satellite verification and climatological research. J. Appl. Meteor., 31 , 12751285.

    • Search Google Scholar
    • Export Citation
  • Sassen, K., , and T. Chen, 1995: The lidar dark band: An oddity of the radar bright band analogy. Geophys. Res. Lett., 22 , 35053508.

  • Sassen, K., , J. R. Campbell, , J. Zhu, , P. Kollias, , M. Shupe, , and C. Williams, 2005: Lidar and triple-wavelength Doppler radar measurements of the melting layer: A revised model for dark- and brightband phenomena. J. Appl. Meteor., 44 , 301. 312.

    • Search Google Scholar
    • Export Citation
  • Steiner, M., , R. A. Houze Jr., , and S. E. Yuter, 1995: Climatological characterization of three-dimensional storm structure from operational radar and rain gauge data. J. Appl. Meteor., 34 , 19782007.

    • Search Google Scholar
    • Export Citation
  • Stewart, R. E., , J. D. Marwitz, , J. C. Pace, , and R. E. Carbone, 1984: Characteristics through the melting layer of stratiform clouds. J. Atmos. Sci., 41 , 32273237.

    • Search Google Scholar
    • Export Citation
  • Sugimoto, N., , I. Matsui, , and A. Shimizu, 2000: Climatological characteristics of cloud distribution and planetary boundary layer structure in Jakarta, Indonesia revealed by lidar observation. Geophys. Res. Lett., 27 , 29092912.

    • Search Google Scholar
    • Export Citation
  • Sugimoto, N., , I. Matsui, , Z. Liu, , A. Shimizu, , and K. Asai, 2001: Latitudinal distribution of aerosols and clouds in the western Pacific observed with a lidar on board the research vessel Mirai. Geophys. Res. Lett., 28 , 41874190.

    • Search Google Scholar
    • Export Citation
  • Szyrmer, W., , and I. Zawadzki, 1999: Modeling of the melting layer. Part I: Dynamics and microphysics. J. Atmos. Sci., 56 , 35733592.

  • Yasunaga, K., , H. Kida, , and T. Satomura, 2003: The 600–750 hPa relative humidity minimum observed during PEM-Tropics B. Geophys. Res. Lett., 30 .2282, doi:10.1029/2003GL018739.

    • Search Google Scholar
    • Export Citation
  • Zuidema, P., 1998: The 600–800-mb minimum in tropical cloudiness observed during TOGA COARE. J. Atmos. Sci., 55 , 22202228.

Fig. 1.
Fig. 1.

Time–longitude cross section of infrared equivalent blackbody temperature. Data are from the Geostationary Meteorological Satellite of the Japan Meteorological Agency. Dashed line indicates the longitude of R/V Mirai.

Citation: Journal of the Atmospheric Sciences 63, 11; 10.1175/JAS3779.1

Fig. 2.
Fig. 2.

Time–height cross sections of (a) zonal and (b) meridional wind components derived from radiosonde. Observations were conducted every 3 h from 0000 UTC 9 November until 0000 UTC 9 December. Solid and dashed contours indicate positive and negative values, respectively. Periods of no data are left blank.

Citation: Journal of the Atmospheric Sciences 63, 11; 10.1175/JAS3779.1

Fig. 3.
Fig. 3.

Cumulative frequency of stability (dT/dZ) for values greater than −5 (solid lines), −4.5 (dashed lines), and −4 K km−1 (dotted lines) during the (a) IOP, (b) inactive period, and (c) active period. Thin dotted line in each panel indicates the 0°C level.

Citation: Journal of the Atmospheric Sciences 63, 11; 10.1175/JAS3779.1

Fig. 4.
Fig. 4.

Evolution of the coverage of C-band radar echo for reflectivity factors >15 dBZ at 2 km height over a 200 km × 200 km area during the IOP (unit: km2).

Citation: Journal of the Atmospheric Sciences 63, 11; 10.1175/JAS3779.1

Fig. 5.
Fig. 5.

Time–height cross section of (a) 532-nm lidar backscattering coefficient, (b) 1064-nm lidar backscattering coefficient, (c) radar reflectivity factor, and (d) ultimate cloud base (Zb) distributions. Crosses in the upper three figures indicate the levels where the backscattering coefficient and reflectivity first exceeded certain thresholds during the vertical scan from the height of 2 km (see text for details). In (d), the area with high relative humidity (>80%) is shaded by grayscale.

Citation: Journal of the Atmospheric Sciences 63, 11; 10.1175/JAS3779.1

Fig. 6.
Fig. 6.

TSI images taken at (a) 0611 UTC and (b) 2211 UTC 10 November 2001.

Citation: Journal of the Atmospheric Sciences 63, 11; 10.1175/JAS3779.1

Fig. 7.
Fig. 7.

Frequency distribution of the base heights of thin cloudy layers (thick solid lines) during the (a) IOP, (b) inactive period, and (c) active period. Error bars represent the ranges within one standard deviation calculated from 30 samples. Cumulative frequency of stability (dT/dZ) for values greater than −4.5 K km−1 (see Fig. 3) and the 0°C level are overplotted in each panel with thin solid and dotted lines, respectively.

Citation: Journal of the Atmospheric Sciences 63, 11; 10.1175/JAS3779.1

Fig. 8.
Fig. 8.

Time–height cross section of (a) 532-nm lidar backscattering coefficient, (b) 1064-nm lidar backscattering coefficient, (c) radar reflectivity factor, and (d) ultimate cloud base (Zb) distributions on 24 November 2001. In (d), the area with high relative humidity (>80%) is shaded by grayscale.

Citation: Journal of the Atmospheric Sciences 63, 11; 10.1175/JAS3779.1

Fig. 9.
Fig. 9.

As in Fig. 8 except on 4 December 2001.

Citation: Journal of the Atmospheric Sciences 63, 11; 10.1175/JAS3779.1

Fig. 10.
Fig. 10.

TSI images taken at (a) 0041 UTC 24 November and (b) 0711 UTC 4 December 2001.

Citation: Journal of the Atmospheric Sciences 63, 11; 10.1175/JAS3779.1

Fig. 11.
Fig. 11.

Variations in the composite relative humidity of eight precipitation events during the active period. Time = 0 corresponds to the time when the coverage of convective type radar echoes at 2 km was at a maximum. Dashed line indicates the 0°C level.

Citation: Journal of the Atmospheric Sciences 63, 11; 10.1175/JAS3779.1

Fig. 12.
Fig. 12.

As in Fig. 11 except for the composite temperature lapse rate.

Citation: Journal of the Atmospheric Sciences 63, 11; 10.1175/JAS3779.1

Fig. 13.
Fig. 13.

Frequency distributions of Zl (solid lines) and Zr (dashed lines) during the (a) IOP, (b) inactive period, and (c) active period. The frequency of Zr is doubled in order to easily compare with the frequency of Zl. Thin dotted line in each panel indicates the 0°C level.

Citation: Journal of the Atmospheric Sciences 63, 11; 10.1175/JAS3779.1

Fig. 14.
Fig. 14.

Variations in (a) Zl, and (b) Zb of the eight precipitation events during the active period. Each color represents one event. Time = 0 corresponds to the time when the coverage of convective-type radar echoes at 2 km was at a maximum. Dashed line indicates the 0°C level.

Citation: Journal of the Atmospheric Sciences 63, 11; 10.1175/JAS3779.1

Fig. 15.
Fig. 15.

Vertical profile of the lidar backscattering coefficient (solid line) and radiosonde-derived relative humidity (dashed line) at 1200 UTC 29 November 2001.

Citation: Journal of the Atmospheric Sciences 63, 11; 10.1175/JAS3779.1

Table 1.

Occurrence frequency (%) of the thickness of the cloud whose base (Zb) is located between 4.5- and 6.5-km height. Clouds are identified as objects with a lidar backscattering coefficient exceeding −5.25 [= log10 β, where β (m−1 sr−1)].

Table 1.
Table 2.

Occurrence frequency (%) of the thickness of the cloud whose base is located between 4.5- and 6.5-km height. Clouds are identified as areas with the relative humidity exceeding 95% or 85%.

Table 2.
Save