• Barnes, G. M., , and M. Garstang, 1982: Subcloud layer energetics of precipitating convection. Mon. Wea. Rev., 110 , 102117.

  • Bretherton, C. S., , and M. C. Wyant, 1997: Moisture transport, lower-troposheric stability, and decoupling of cloud-topped boundary layers. J. Atmos. Sci., 54 , 148167.

    • Search Google Scholar
    • Export Citation
  • Brost, R. A., , J. C. Wyngaard, , and D. H. Lenschow, 1982: Marine stratocumulus layers. Part II: Turbulence budgets. J. Atmos. Sci., 39 , 818836.

    • Search Google Scholar
    • Export Citation
  • Comstock, K. K., , C. S. Bretherton, , and S. E. Yuter, 2005: Mesoscale variability and drizzle in southeast Pacific stratocumulus. J. Atmos. Sci., 62 , 37923807.

    • Search Google Scholar
    • Export Citation
  • Duynkerke, P. G., and Coauthors, 2004: Observations and numerical simulations of the diurnal cycle of the EUROCS stratocumulus case. Quart. J. Roy. Meteor. Soc., 130 , 32693296.

    • Search Google Scholar
    • Export Citation
  • Fairall, C. W., , E. F. Bradley, , D. P. Rogers, , J. B. Edson, , and G. S. Young, 1996: Bulk parametrization of air–sea fluxes for Tropical Ocean Global Atmosphere Coupled Ocean Atmosphere Response Experiment. J. Geophys. Res., 101 , 37473764.

    • Search Google Scholar
    • Export Citation
  • Faloona, I., and Coauthors, 2005: Observations of entrainment in eastern Pacific marine stratocumulus using three conserved scalars. J. Atmos. Sci., 62 , 32683285.

    • Search Google Scholar
    • Export Citation
  • Ferek, R. J., and Coauthors, 2000: Drizzle suppression in ship tracks. J. Atmos. Sci., 57 , 27072728.

  • Fu, Q., , and K. N. Liou, 1993: Parameterization of the radiative properties of cirrus clouds. J. Atmos. Sci., 50 , 20082025.

  • Han, Q., , R. Welch, , J. Chou, , W. Rossow, , and A. White, 1995: Validation of satellite retrievals of cloud microphysics and liquid water path using observations from FIRE. J. Atmos. Sci., 52 , 41834195.

    • Search Google Scholar
    • Export Citation
  • Jensen, J. B., , S. Lee, , P. B. Krummel, , J. Katzfey, , and D. Gogoasa, 2000: Precipitation in marine cumulus and stratocumulus. Part I: Thermodynamic and dynamic observations of closed cell circulations and cumulus bands. Atmos. Res., 54 , 117155.

    • Search Google Scholar
    • Export Citation
  • Paluch, I. R., , and D. H. Lenschow, 1991: Stratiform cloud formation in the marine boundary layer. J. Atmos. Sci., 48 , 21412158.

  • Perez, J. C., , F. Herrera, , F. Rosa, , A. Gonzales, , M. Wetzel, , R. Borys, , and D. Lowenthal, 2000: Retrieval of marine stratus cloud droplet size from NOAA AVHRR night imagery. J. Remote Sens. Environ., 73 , 3145.

    • Search Google Scholar
    • Export Citation
  • Petters, M. D., , J. R. Snider, , B. Stevens, , G. Vali, , I. Faloona, , and L. Ryssell, 2005: Accumulation mode aerosol, pockets of open cells, and particle nucleation in the remote subtropical pacific marine boundary layer. J. Geophys. Res., in press.

    • Search Google Scholar
    • Export Citation
  • Sharon, T. M., , B. A. Albrecht, , H. Jonsson, , P. Minnis, , M. M. Khaiyer, , T. M. VanReken, , J. Seinfeld, , and R. Flagan, 2006: Aerosol and cloud microphysical characteristics of rifts and gradients in maritime stratocumulus clouds. J. Atmos. Sci., in press.

    • Search Google Scholar
    • Export Citation
  • Stevens, B., 2000: Cloud-transitions and decoupling in shear-free stratocumulus topped boundary layers. Geophys. Res. Lett., 27 , 25572560.

    • Search Google Scholar
    • Export Citation
  • Stevens, B., , W. R. Cotton, , G. Feingold, , and C-H. Moeng, 1998: Large-eddy simulations of strongly precipitating, shallow, stratocumulus-topped boundary layers. J. Atmos. Sci., 55 , 36163638.

    • Search Google Scholar
    • Export Citation
  • Stevens, B., and Coauthors, 2003a: Dynamics and chemistry of marine stratocumulus—DYCOMS-II. Bull. Amer. Meteor. Soc., 84 , 579593.

  • Stevens, B., and Coauthors, 2003b: On entrainment rates in nocturnal marine stratocumulus. Quart. J. Roy. Meteor. Soc., 129 , 34693493.

    • Search Google Scholar
    • Export Citation
  • Stevens, B., and Coauthors, 2005a: Evaluation of large-eddy simulations via observations of nocturnal marine stratocumulus. Mon. Wea. Rev., 133 , 14431462.

    • Search Google Scholar
    • Export Citation
  • Stevens, B., , G. Vali, , K. Comstock, , R. Wood, , M. C. Zanten, , P. Austin, , C. S. Bretherton, , and D. H. Lenschow, 2005b: Pockets of open cells (POCs) and drizzle in marine stratocumulus. Bull. Amer. Meteor. Soc., 86 , 5157.

    • Search Google Scholar
    • Export Citation
  • Vali, G., , R. D. Kelly, , J. French, , S. Haimov, , D. Leon, , R. E. McIntosh, , and A. Pazmany, 1998: Finescale structure and microphysics of coastal stratus. J. Atmos. Sci., 55 , 35403564.

    • Search Google Scholar
    • Export Citation
  • vanZanten, M. C., , B. Stevens, , G. Vali, , and D. H. Lenschow, 2005: Observations of drizzle in nocturnal marine stratocumulus. J. Atmos. Sci., 62 , 88106.

    • Search Google Scholar
    • Export Citation
  • Wang, S., , and Q. Wang, 1994: Roles of drizzle in a one-dimensional third-order turbulence closure model of nocturnal stratus-topped marine boundary layer. J. Atmos. Sci., 51 , 15591576.

    • Search Google Scholar
    • Export Citation
  • View in gallery

    Early morning (1430 UTC, 0730 PDT) visible satellite imagery of RF02. The shaded region is the GOES-10 channel-1 reflectance on a linear scale ranging from 0 to 0.6. The general region of aircraft sampling is shown with the aid of three flight tracks (SC2, CT2, and SF from south to north, respectively) whose estimated airmass relative position is shown superimposed on the cloud field.

  • View in gallery

    Mean thermodynamic state as observed during RF02. (left to right) Liquid water potential temperature Θl, total water specific humidity qt, and liquid water specific humidity ql. Dots represent averages over 30-m intervals based on all flight data; black (gray) lines denote inner (outer) quartiles. The LCL of the mean PBL air and the inversion height are specified on the vertical axis, where the average leg heights are also indicated. The LCL is calculated from mean thermodynamical properties and the inversion height is located where the sharpest increase in the mean profile of Θl occurs. For Θl and qt the numbers on the horizontal axis represent the mean PBL value, the value at the inversion height, and the maximum value above the PBL, while for ql maximum liquid water (based on the 30-m interval mean values) and adiabatic liquid water content at the cloud top are noted. The mean SST (292.0 K) for the whole flight area is also indicated.

  • View in gallery

    Airmass-relative position of the two CT flight tracks at the midpoint of CT1. (left) Starting points of the legs are denoted by a closed circle. (right) Radar-based drizzle rate, R, at cloud base (light gray: drizzle rate of 0.1 mm day−1 or higher, dark gray: 1 mm day−1 or higher, black: 5 mm day−1 or higher), cloud droplet number, N, and downward LW radiative flux, FLW, as a function of location. All data points are plotted as averages over 1-km lengths. The data from CT2 are plotted in reversed order; thus note that the leftmost data values of the CT1 and CT2 legs are actually an hour apart in time. The vertical scale is denoted by the maximum value, leg mean, and minimum value of the variable.

  • View in gallery

    Difference in brightness temperature, ΔTB, of the 11- and 4-μm channels for the (left) 0930 and (middle) 1315 UTC satellite images. (right) The 1315 UTC image with all values of ΔTB greater than 1.5 K is masked, thus isolating pixels that are candidates for POC-like regions. The dashed lines indicate the trajectory of the mean PBL flow for a 225-min period. The ellipse shows the expected position the air mass at the northernmost point on the left dashed line at 0930 UTC should be achieved at 1315 UTC.

  • View in gallery

    Low-pass filtered ΔTB, the difference in brightness temperature of the 4- minus 11-μm channel as measured by the GOES-10 satellite; also superimposed on this figure are 1-km averages of the radar-based drizzle rates, R, at 70 m above sea level. For the SF leg in situ drizzle data are used because radar data are not available. POC regions (i.e., regions with ΔTB less than 1.5 K) are denoted with solid black lines, non-POC regions with dashed lines, and R with thin gray lines.

  • View in gallery

    Potential temperature Θ, specific humidity qυ, and vertical velocity w (with the leg mean w subtracted) along the flight path for two subcloud legs (SC2 and SF). All data points represent 500-m averages. Solid lines denote POC per the ΔTB criterion (i.e., ΔTB less than 1.5 K) with gray segments denoting the precipitating part of the POC. The thin dashed line shows the values outside the POC. The horizontal lines represent the leg mean values, and interior tick marks on the right vertical axis indicate the average over the non-POC area (thin black line), the POC (thick black line), and the drizzle areas within the POC (thick gray line). The vertical scale can be inferred by the marking of the maximum, leg mean, and minimum value of the variable.

  • View in gallery

    Vertical velocity variance () as a function of height. The variance in the POC area is given by an open circle, while the variance in the closed cell region is denoted by a closed circle plotted 10 m higher for clarity. The uncertainty specified is the standard deviation of the mean. The open gray circles without specified uncertainty are variance values for drizzling areas only. Information on how () was estimated can be found in the appendix Cloud depth is indicated by the gray bar.

  • View in gallery

    (left) Total-water specific humidity flux budget and (right) net flux Fqt as function of height. The net flux consists of the sum of the turbulent flux and the drizzle flux. The turbulent flux is given by circles, and the in situ drizzle flux FR is denoted by diamonds. Open symbols denote values in the POC region, while values in the non-POC region are given by closed symbols. Cloud depth is indicated by the gray bar. The uncertainty specified is the standard deviation of the mean.

  • View in gallery

    (left) Liquid water potential temperature flux budget and (right) net flux FΘl as a function of height. Symbol representation similar to Fig. 8 with the total LW radiative flux FLW denoted by open triangles. The net flux is the sum of the turbulent, radiative, and drizzle flux, yet the drizzle flux is not repeated here but can be found in Fig. 8. The uncertainty specified is the standard deviation of the mean.

  • View in gallery

    Conceptual rendering of POC and neighboring non-POC or stratiform region. Also shown is a schematic of the horizontal and vertical variations in Θe and inferred mesoscale circulations. Note that the net upward motion in the POC is left out, as explained in the text.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 9 9 3
PDF Downloads 11 11 5

Observations of the Structure of Heavily Precipitating Marine Stratocumulus

View More View Less
  • 1 Institute of Marine and Atmospheric Research Utrecht, Utrecht University, Utrecht, Netherlands, and Department of Atmospheric and Oceanic Sciences, University of California, Los Angeles, Los Angeles, California
  • 2 Department of Atmospheric and Oceanic Sciences, University of California, Los Angeles, Los Angeles, California
© Get Permissions
Full access

Abstract

The second research flight of the Second Dynamics and Chemistry of Marine Stratocumulus (DYCOMS-II) field study is analyzed. This case attracted attention because it combined the presence of high drizzle rates with the occurrence of clearings in the cloud cover, which previous work has suggested could be due to a drizzle-induced change in cloud structure. Recent work has named the configuration of these open-cell-like features pocket of open cells (POC). A division of the data, based on the difference of brightness temperature of the 11- and 4-μm channels of the Geostationary Operational Environmental Satellite-10 (GOES-10), is used to condition averages over POC and non-POC sections of the data. Based on this division, significant precipitation is observed almost entirely within the POC. Overall, the observed PBL was markedly energetic and well mixed, commensurate with observations of similarly forced nonprecipitating boundary layers. Regions of elevated equivalent potential temperature, Θe, are encountered in association with the POC and are particularly pronounced below cloud in regions of active precipitation. In the same areas, evaporative cooling and moistening of the subcloud layer air and a marked reduction of the vertical velocity variance below cloud is noted. The POC, and in particular the drizzling areas in the POC, have a mean upward velocity, while the non-POC is associated with descending air. In addition to exhibiting more mesoscale variability the POC differs microphysically from the non-POC: in the POC liquid water amount is higher, cloud droplet number concentration lower, and the effective radii larger. Factors maintaining regions of elevated Θe are discussed, and targets for future modeling or observational studies are identified.

Corresponding author address: Dr. M. C. vanZanten, KNMI, P.O. Box 201, 3730 AE De Bilt, Netherlands. Email: Zanten@knmi.nl

Abstract

The second research flight of the Second Dynamics and Chemistry of Marine Stratocumulus (DYCOMS-II) field study is analyzed. This case attracted attention because it combined the presence of high drizzle rates with the occurrence of clearings in the cloud cover, which previous work has suggested could be due to a drizzle-induced change in cloud structure. Recent work has named the configuration of these open-cell-like features pocket of open cells (POC). A division of the data, based on the difference of brightness temperature of the 11- and 4-μm channels of the Geostationary Operational Environmental Satellite-10 (GOES-10), is used to condition averages over POC and non-POC sections of the data. Based on this division, significant precipitation is observed almost entirely within the POC. Overall, the observed PBL was markedly energetic and well mixed, commensurate with observations of similarly forced nonprecipitating boundary layers. Regions of elevated equivalent potential temperature, Θe, are encountered in association with the POC and are particularly pronounced below cloud in regions of active precipitation. In the same areas, evaporative cooling and moistening of the subcloud layer air and a marked reduction of the vertical velocity variance below cloud is noted. The POC, and in particular the drizzling areas in the POC, have a mean upward velocity, while the non-POC is associated with descending air. In addition to exhibiting more mesoscale variability the POC differs microphysically from the non-POC: in the POC liquid water amount is higher, cloud droplet number concentration lower, and the effective radii larger. Factors maintaining regions of elevated Θe are discussed, and targets for future modeling or observational studies are identified.

Corresponding author address: Dr. M. C. vanZanten, KNMI, P.O. Box 201, 3730 AE De Bilt, Netherlands. Email: Zanten@knmi.nl

1. Introduction

The question of how precipitation affects the turbulent dynamics and cloud morphology of the stratocumulus-topped boundary layer is central to our understanding of this climatologically important flow regime. Indeed, many hypothesized effects of atmospheric aerosol perturbations on the radiative properties of the atmosphere as a whole hinge on it. The observational basis for addressing this question is, however, quite limited. From the smattering of case studies that exist, clear trends or patterns in how the structure of the boundary layer changes under the influence of drizzle have yet to emerge; thus the target, or object, of any theoretical development of the subject is itself outstanding. In this context, a number of missions, flown as part of the Second Dynamics and Chemistry of Marine Stratocumulus (DYCOMS-II; Stevens et al. 2003a) field study, offer the possibility of expanding our understanding of the basic phenomenology of precipitating boundary layers. The second research flight (RF02) is particularly intriguing in this respect, as it sampled areas of heavy drizzle coinciding with more open cellular-like cloud structures, which Stevens et al. (2005b) called pockets of open cells (POCs); similar cloud structures have been named cloud rifts by Sharon et al. (2006). A tilde-shaped POC is evident near 31°N, 122°W in the early morning satellite imagery taken during RF02 (Fig. 1). The airmass-relative flight tracks1 suggest that this particular POC was bisected during the northernmost surface leg (SF) (see Table 1 for the flight leg nomenclature) in Fig. 1 and that the northwestern sector of the CT2 and SC2 flight legs may have sampled its southeast edge. These data, and more preliminary findings presented by Stevens et al. (2005b), suggest that RF02 provides an excellent opportunity to investigate the interaction among drizzle, cloud dynamics, and differences between regions of open and closed cellular convection.

Drizzle is often associated with a marked differentiation of subcloud and cloud-layer state variables (e.g., Brost et al. 1982; Stevens et al. 1998); however, we found little evidence of this during RF02. Both liquid water potential temperature (Θl) and total water specific humidity (qt) were surprisingly well mixed (Fig. 2), with PBL mean profiles nearly indistinguishable from those for RF01 (Stevens et al. 2003b); a flight that had been flown the previous day in the same area but was representative of a homogeneous nondrizzling stratocumulus-topped PBL. Overall the mean state of RF02 was slightly cooler (0.3 K) and moister (0.4 g kg−1) than RF01, resulting in a 120 m lower LCL and consequently a deeper cloud (cloud top height of RF02 was only 50 m lower than the cloud top height of RF01). From the perspective of the mean profiles within the PBL, perhaps the biggest clue that precipitation played more of a role during RF02 is the subadiabaticity of its cloud layer, with ql tending to be relatively flat over the upper half of the cloud. Other, more subtle, signals are also present. These include larger fluctuations of qt within the PBL and larger spatial fluctuations in the height of cloud top, the latter being responsible for a less well delineated cloud-top height in the mean profile. Both characteristics were missing from RF01, but are evident in a previous analysis of precipitating stratocumulus (Paluch and Lenschow 1991). The ql profile also reveals greater variations in cloud base height during RF02, with the presence of finite values of ql in the hectometer below the LCL, and even nonzero ql values at much lower levels associated with the presence of showers. The boundary forcings for RF01 and RF02 appear rather similar. The most notable difference is evidence that a humid layer similar to that measured well above the cloud layer during RF01 appeared to intersect the cloud layer as sampled during RF02. The presence of a humid layer leads to smaller jumps in both qt and Θl compared to RF01. Stevens et al. (2005b) hypothesized that the intersection of this moist layer with the PBL might have played a role in either the formation and/or the maintenance of the drizzle, and hence the POC. The mean wind in the PBL was 7.3 m s−1 out of the north-northwest; so during the seven hours of the flight the air mass was advected over roughly 185 km. Above the PBL the geostrophic wind backed with height, indicative of cold advection.

Despite its rather subtle influence on the mean state, heavy drizzle (defined as rates higher than 1 mm day−1) is present over large areas and extended periods of time. This is shown for the CT legs in Fig. 3, where the horizontal bar (or ticker-tape-like feature at the base of each time series in the right-hand panels) indicates drizzle rates in four different intensity intervals. The precipitation rate, R, is derived at two heights (mean cloud base and close to the sea surface, respectively) from radar data, following the procedures described by vanZanten et al. (2005) and summarized in the appendix. Moderate values of R (higher than 0.1 mm day−1) were present more often than not. Locally regions of 10–20 km are evident where cloud base drizzle rates are greater than 5 mm day−1. Precipitation at these rates is equivalent to a latent heat flux of 145 W m−2 and is thus expected to play a significant role in both heat and water budgets of the layer. Also shown in the right-hand panels of Fig. 3 is the time series of cloud-drop number concentration and downwelling longwave radiation for both CT legs, the latter being an indicator of cloud thickness as cloud clearings—or at least thinning—can be associated with large (80 W m−2) reductions in downwelling longwave radiation. Contrasting the regions marked a, b, and c in the figure we see that region b evinces persistent drizzle with only hints of clearing, showers and clearings are evident in region c, and clearings with no drizzle are evident in region a. Each region raises its own questions: The absence of cloud holes in region b is engaging because drizzle rates that high are, without replenishment, able to deplete RF02’s flight mean liquid water path on time scales of 30 min or less. Is the absence of drizzle in region a and the tendency toward less pronounced clearings in the latter CT2 evidence of cloud filling in this region? Is the POC, whose edge is apparently sampled along region c (cf. Fig. 1), maintained by drizzle? If so what role does lower aerosol, and hence droplet concentration, play in fueling drizzle? Are the cellular features so persistent in this region characteristic of the changes predicted by past (e.g., Stevens et al. 1998) modeling studies? And to what extent are such features passively advected by the mean flow, or is their spatial evolution determined by dynamical interactions with the ambient environment (e.g., Paluch and Lenschow 1991; Jensen et al. 2000)? Figure 3 and the questions it raises help encapsulate why RF02 is so fascinating and motivates our further study.

Although many of these questions will prove difficult to answer definitively based on the data collected during RF02 alone, these data can be used to explore the generality of past ideas, test previously stated hypotheses, and characterize what we believe to be essential features of RF02, thereby providing a target for future modeling studies. As such this analysis can be an important component in ongoing attempts to first characterize, and then understand, how precipitation modifies stratiform cloud sheets and how susceptible this climatologically critical cloud type is to external perturbations.

2. Methodology

The basis of our attempt to use RF02 data to understand the interaction between drizzle and cloud dynamics is to compare regions of the flow where precipitation is light, or not evident, with regions where it is more pronounced. Toward this end one would like to partition the data based on its propensity to precipitate. While the radar reflectivity (as a proxy for drizzle) would seem like a natural indicator, radar data is not available for the surface leg and is heavily attenuated on the RL legs. More importantly, when using a simple precipitation based indicator, it is difficult to recognize that locally nonprecipitating clearings might be elements of broader regions of precipitation, and thus from a dynamical perspective should be grouped with their neighboring regions of intense precipitation. To avoid these difficulties we chose instead to use an indicator based on the infrared channels of the Geostationary Operational Environmental Satellite-10 (GOES-10). Previous studies have shown that ΔTB, the difference in brightness temperature of the 4-μm minus 11-μm channels can be a good indicator of precipitation. The ΔTB acts as a proxy for the effective radius (cf. Fig. 1 of Perez et al. 2000), which itself is a proxy for drizzle (Han et al. 1995; Ferek et al. 2000). Here ΔTB values less than 2 K have also been found to be correlated with areas with heavy drizzle during the specific conditions encountered during DYCOMS-II (vanZanten et al. 2005). The coarse-graining inherent in the 4-km GOES-10 IR channels provides a natural way to partition the flow at larger scales, allowing us to look at broader POC-like regions and compare them to more stratiform, less precipitating regions of the flow.

To associate a value of ΔTB with each aircraft measurement we calculate the airmass-relative flight tracks at either 0930 or 1315 UTC (whichever is closer) and then select the corresponding pixel value from the GOES-10 imagery at these times (Fig. 4), thus generating a time series of ΔTB along the flight track. Each leg of the time series is then spectrally filtered by retaining only the first four harmonics (roughly corresponding to scales greater than 20 km). Areas where the filtered ΔTB is less than 1.5 K are flagged as belonging to the POC; otherwise the data are classified as non-POC. Our choices of exactly how many, and which, satellite snapshots to use in creating the original ΔTB time series, how to filter the segments, and what value to choose as a threshold are all open to question, of which the latter might be the most critical.

Figure 5 shows not only that depressions in ΔTB are associated with precipitation but also that the magnitude of the ΔTB depression correlates well with the magnitude of drizzle, with both becoming more pronounced with time. As hinted at above, this temporal evolution is probably an artifact of horizontal displacements of the flight track with respect to the air being sampled. Although the flight tracks attempted to remain in the same air mass, the necessity of staying out of restricted airspace to the southeast required systematic backtracking upwind through the course of our time on station; the effects of which are evident in Fig. 1. Serendipitously it led to increasingly better sampling of the POC through the course of the flight—something we became aware of only in the post flight analysis. It is worth emphasizing that the ΔTB threshold appears to be a useful, but certainly not perfect, indicator for both POCs and drizzle during RF02. It is by no means clear that it would work as well for other situations. Indeed, even for RF02 its merits are not without question: in some cases a low ΔTB value could simply result from an especially thin cloud layer, which is neither associated with greater inhomogeneity nor (heavy) drizzle. To the extent that such circumstances bias our subsequent analysis, and interpretations, we attempt to account for them explicitly.

In addition to providing a basis for identifying the POC, this analysis also suggests that on a time scale of hours the POC appears to simply advect passively with the flow. To make this point, estimates of 225-min trajectories following the mean PBL wind are indicated by the dashed lines in Fig. 4. The ellipse along the leftmost line in the middle panel represents the expected position at 1315 UTC of an air mass originating at the start of this same line 225 min earlier at 0930 UTC. The size of the ellipse characterizes our uncertainty in the mean wind and, hence, particle trajectories. Given this uncertainty, the alignment of flow features with these lines suggests that the null hypothesis (i.e., that the POC as a whole is passively advected by the mean flow) is quite viable. Whether or not the cell walls retain this advective rigidity, or more dynamically evolve within the envelope of the POC, is not possible to discern from this analysis.

3. Analysis

a. Mean structure

Within the POC we find evidence of repetitive spatial fluctuations in thermodynamic state parameters. Examples from the SF and SC legs are shown in Fig. 6, where a solid line is used to denote the portions of the time series where the ΔTB indicates a POC. Over the POC, fluctuations in qυ and Θ are negatively correlated on scales of 10–20 km, commensurate with the scale of the cellular patterning in the satellite images (e.g., Fig. 1). These features, which are also similar in both scale and amplitude to those observed by Paluch and Lenschow (1991) for a case of precipitating stratocumulus observed during the First International Satellite Cloud Climatology Project (ISCCP) Regional Experiment (FIRE) and in the analysis of drizzle over the southeast Pacific by Comstock et al. (2005), are most striking in the first half of the SF leg where the flight track most directly bisected the POC (e.g., Fig. 1). They are also evident in the SC time series where the aircraft flew through a region of significant precipitation at the edge of the POC. Anticorrelated fluctuations in Θ and qυ are not evident in the region of low ΔTB sampled toward the end of SF; however, this portion of the time series appears to have been near the eastern edge of the POC and is not associated with strong precipitation, thus suggesting that such fluctuations are related to the presence of precipitation.

Indeed, Paluch and Lenschow (1991) argued that these fluctuations were simply the signature of precipitation evaporating in the subcloud layer. As they pointed out, this argument works insofar as the fluctuations are not apparent in Θe, which is conserved2 under precipitation. However, from the perspective of the Θe budget, qυ fluctuations in Fig. 6 are not compensated by fluctuations in Θ. To make this more quantitative we composited over regions of showers during the SF leg (with the composite constructed of windows, 71 seconds long, centered around the maximum amount of drizzle water). On average the maximum fluctuation in Θ is 0.5 K, while the maximum fluctuation in qυ is 0.6 g kg−1. If Θe were to be conserved for this magnitude of moisture fluctuation, one would expect depressions in Θ of roughly δ Θ = Lυδ qυ/cp ≈ 1.5 K. Given the measured fluctuations of Θ, this implies a nearly 1-K enhancement of Θe in regions of precipitation. Although such effects can always be attributed to instrumental biases, obvious biases such as probe wetting effects (e.g., discussed in the appendix) would, if anything, damp such fluctuations by measuring artificially cold temperatures. Further, no signs of wetting of the cross-flow Lyman-α were found (while the stub Lyman-α does show signs of wetting at the same location). The idea that precipitation from shallow clouds actually coincides with elevated values of Θe is, upon closer examination, also consisted with the FIRE data, as well as measurements made during periods of light precipitation (relative to that expected in deep convection) during the Global Atmospheric Research Program (GARP) Atlantic Tropical Experiment (GATE) (Barnes and Garstang 1982). In the case of the FIRE data, these fluctuations were relatively modest and, to the extent they existed, they were attributed to baroclinic circulations, which in a dry atmosphere act to damp horizontal fluctuations in Θυ. Paluch and Lenschow (1991) envisioned such circulations arising in response to the evaporation of drizzle and, in so doing, preferentially damping Θ fluctuations, thus helping to generate positive fluctuations in Θe. The ability of the mechanism to contribute to the observed structure of the RF02 data is discussed below.

Even after averaging over clear and precipitating regions during SF and SC2 the POC tends to be cooler and moister than the background flow, this being most pronounced for SC2 where the horizontal extent of drizzle is larger than close to the surface. The observed differences might seem small in absolute value, but it is important to keep in mind that, for example, a change of 0.1 K in the averaged temperature leads to a change of roughly 10 m in the LCL. The colder and moister profile in the POC as a whole leads to a roughly 50 m lower LCL compared to the closed cell region. These differences become more pronounced when we conditionally sample the POC region over drizzling areas only; whereby the drizzling areas are defined as the location where drizzle drop concentrations (as measured by the 260X) exceed 5 L−1. To the extent that effects of spatial and temporal aliasing can be ignored there is also an indication that Θl increases and qt decreases slightly with height, albeit to a smaller extent than is evident in the precipitating LES of Stevens et al. (1998, hereafter SCFM98).

The time series of vertical velocity, w, is also presented in Fig. 6. On the scale of the POC, as a whole, upward motion is evident, becoming more pronounced when looking at the drizzling areas only. Although the absolute vertical velocities measured from the aircraft are typically not reliable on these scales, we are interested in the relative differences between the two (POC and non-POC) regions, which should be more reliably measured. The reproducibility of the signal at multiple levels (e.g., Table 2) further supports the idea that the differences in the mean vertical motion in and outside of the POC is real.

Ascending air in the POC is in line with a change to a more cumulus-like cloud dynamics. In the snapshots in SCFM98, the occurrence of cumulus is clearly visible in the heavy drizzle case and their values obtained for the skewness of w are commensurate with cumulus convection. However, this dynamical signature of cumulus did not emerge from our analysis of the RF02 data. Furthermore, the cloud droplets measured by the SPP-100 below cloud are most likely due to splash artifacts by drizzle drops, although the occurrence of freshly nucleated cloud droplets cannot be ruled out.

Microphysically a distinction is also evident between the POC and the non-POC region (see Table 3). The amount of ql in the POC is higher than in the closed cell region, consistent with a lower LCL as estimated from the mean thermodynamic state. SCFM98 found a smaller liquid water path in the heavy drizzling case, albeit with more variance. We did not encounter this for RF02, although the similar amount of ql variance for the POC and nonPOC region could very well be due to the occurrence of cloud holes in both regions. In the POC the number of droplets is less but the effective radius of these droplets is larger, both favorable conditions for drizzle. Because in regions of drizzle droplet sizes are probably underestimated,3 the data in the table may actually understate this effect. Those results are consistent with a study by Sharon et al. (2006) who compared the aerosol and cloud microphysical properties of a cloud rift with the properties of the surrounding solid stratocumulus. A more in-depth study of the microphysical structure and aerosol characteristics during RF02, and their possible role in maintaining POCs, can be found in a companion study by Petters et al. (2005).

b. Energetics

Past modeling studies and simulations have suggested that precipitation suppresses the turbulent kinetic energy (TKE) production by buoyancy in the PBL. Overall the magnitude of the vertical velocity variance (as a proxy for TKE) during RF02 (Fig. 7) is commensurate with values observed during RF01. That said, a partitioning of the data into POC and non-POC is suggestive of a decrease in similar to what has been predicted by modeling studies (Wang and Wang 1994; SCFM98). On the basis of their simulations the latter explicitly note that in regions of precipitation “The reduction in is dramatic, particularly in the subcloud layer and should be observable.” Such a reduction is observable during RF02 although it is most visible in the values of when averaged over drizzling areas only. Analysis of w spectra for the SC legs indicates that this reduction in variance mainly takes place on scales between a few hundred meters and a few kilometers (see also Fig. 10 in Petters et al. 2005). Larger vertical velocity variances in the POC during the CT (at 674 m) legs might be due simply to the layer being deeper in these regions, resulting in the measurements being located farther from local cloud top than for the closed cell region. This interpretation is supported by measurements during the SP leg for which in the drizzling area is also larger. The values for the POC as a whole, or just its precipitating area, are similar at CT because measurements sampled very little area with no precipitation within the POC.

We also analyzed fluxes of moisture and heat, results of which are presented in Figs. 8 and 9, respectively. The turbulent fluxes in these figures are calculated similarly to Faloona et al. (2005) and described in detail in the appendix The drizzle flux FR is computed from in situ data using the method described in vanZanten et al. (2005) (summarized in the appendix). Attempts to factor the effects of precipitation on the heat and moisture budgets are complicated by a clear trend in the precipitation rate. Table 1 and Fig. 5 show that the earlier legs more marginally sampled the POC and hence regions of pronounced precipitation. The differences are most evident for the CB legs, which were flown at the beginning of our time on station. Thus to avoid this bias we omit estimates of CB fluxes within the POC. Even so, the effect of this aliasing is still evident in the profiles of FR, which fail to display the common picture of highest precipitation rate at cloud base with decreasing values toward cloud top and the sea surface. For the CT legs we could use the radar data to calculate the drizzle flux in the POC at cloud base and the surface, with values of 103 and 28 W m−2, respectively. This illustrates that the instantaneous drizzle profile conforms to our expectations.

The radiative forcing across cloud top and across the PBL was determined by combining a radiative transfer model and measured longwave (LW) radiative fluxes (see the appendix). This leads to a more detailed LW flux profile, especially around cloud top, than what would be obtainable from measurements alone, while at the same time computed values can be compared to measured values at heights where those values are available. The LW radiative flux divergence across cloud top is estimated to be 70 ± 5 W m−2, which is offset by a cloud base warming of 20 W m−2 making the total LW radiative flux divergence across the PBL about 50 ± 5 W m−2. This value is similar in magnitude to the drizzle flux divergence across the PBL in the POC.

Overall this analysis demonstrates that the drizzle flux is an important term in the budget of both Θl and qt in the POC. Not surprisingly, drizzle moistens the subcloud layer and dries the cloud layer (although the temporal and spatial biasing in the precipitation fluxes hides the drying effect in Fig. 8), but the resultant turbulent circulations in the POC organize to distribute the net precipitation drying over the entire layer. In the non-POC region a net moistening of the PBL is visible. The heat budget proves more difficult to interpret. The estimated net flux divergence across the PBL in the POC suggests a slight warming, while it is ambiguous whether the nonPOC region displays a net flux divergence. Overall, these budgets provide little evidence of diabatic processes (turbulent fluxes, radiation, or drizzle) driving an enhancement of Θe within the POC, as the apparent tendency toward a net drying of the POC is not clearly offset by commensurate warming, as would be necessary to increase, or even maintain, values of Θe within the POC.

The surface latent and sensible heat fluxes (LHF and SHF, respectively) were estimated using the Tropical Ocean Global Atmosphere Coupled Ocean–Atmosphere Response Experiment (TOGA COARE) bulk flux algorithm (Fairall et al. 1996, version 2.5b). This yielded a SHF flux of 16 ± 5 W m−2 and a LHF of 93 ± 12 W m−2. For this we assumed the mean PBL values as specified in the Fig. 2 to be valid at a height of 50 m. In absolute terms these values compare reasonably well with direct, eddy- correlation-based estimates along the SF leg, for which we estimate a SHF of 9 ± 4 W m−2 and a LHF of 83 ± 6 W m−2. If the POC and non-POC regions have the same surface exchange coefficient and wind speed, surface fluxes would not be expected to drive differences in Θe. That said, reductions in Θυ at the surface or larger wind speeds will lead to enhanced exchange between the surface and overlying atmosphere, both of which may contribute to some enhancement of Θe within the POC region.

4. Discussion

a. Conceptual model

One of the more robust features to emerge from this analysis is that the POC is clearly associated with enhanced regions of precipitation, which are in turn associated with regions of elevated subcloud Θe. The maintenance of such features seems conducive to the longevity of the precipitating regions but is difficult to explain on the basis of the flux data or the boundary conditions alone. Because the POC appears to advect with the mean flow, it seems difficult to attribute enhancement of subcloud Θe to differential surface structure (i.e., warmer SSTs over the POC). Similarly, our analysis of the net turbulent and diabatic fluxes suggests that these would, if anything, act to reduce Θe in the region of the POC. We also find such features difficult to attribute to simple instrumental artifacts.

Figure 10 was drafted to help organize our thoughts with respect to the processes affecting Θe on the scale of both the POC itself and its component cells. To the extent to which elevations of Θe on the scale of the POC, such as are visible for the surface leg, are more generally evident we argue that baroclinic circulations of the type envisioned by Paluch and Lenschow (1991) are not significant contributers to the POC-wide increase in Θe, if only because the POC appears as an area of net convergence and, hence, net upward motion. While from Fig. 6 it is apparent that most of the motion appears to be manifest near the precipitating cells, the data are insufficient to more clearly determine the spatial structure of the circulation. On the scale of the POC, reduced radiative cooling associated with more broken clouds, a larger (meso) scale circulation implied by the vertical velocity measurements, and perhaps entrainment of air richer in Θe may all contribute to a POC-wide increase in the mean PBL Θe. A larger, mesoscale, circulation is consistent with, and reinforces, higher values of Θe in the POC region, as the latter would drive regions of deeper, precipitating, convection within the POC. Such circulations could also help maintain a source of moisture to the POC.

On the scale of the cells within a POC, the type of circulations envisioned by Paluch and Lenschow (1991), as well as enhanced surface fluxes generated by further destabilization of the surface layer, may play more of a role. A baroclinic circulation in the subcloud layer (indicated by the small arrows in the subcloud layer in Fig. 10) would tend to homogenize temperature fluctuations more effectively than moisture fluctuations, but in the absence of diabatic effects it is unable to explain the enhancement in Θe near the precipitating region or on the scale of the POC as a whole. Radiative fluxes, which could be expected to export Θe from the clear regions but enhance Θe near the cloud base in saturated regions, may also play a homogenizing role. These types of radiative effects would be more pronounced in the upper subcloud layer, consistent with greater homogenization of Θ between precipitating and nonprecipitating regions of the POC at that level (cf. Fig. 6). The spreading of precipitation-induced cold pools associated with such a circulation may also support convection at the boundaries of the precipitating regions. Because the drizzle rate profile from the CT legs shows most of the heating to be confined to the upper part of the cloud layer, the counterpart to this circulation would be in the reverse sense above cloud base, which would tend to drive convergence toward the middle of the PBL. Such circulations would seem to imply greater differentiation between the cloud and subcloud layer than we observed. Since the POC was sampled serendipitously, the data were not suitable for shedding light on a possible mesoscale circulation. Neither an analysis of local divergence and convergence of the horizontal wind speed nor an analysis of the difference between pressure altitude and radar altimeter led to conclusions constrained by the data. So, on these scales many mysteries remain: in particular, an evaluation of the efficacy of various processes is hampered by a lack of knowledge as to the time scales over which the Θe perturbations develop. If they develop slowly and are long lived, they may impart some rigidity to the internal structure of the POC. Clearly these and other questions are attractive targets for future modeling and observational studies.

b. LES

Before evaluating such ideas with new modeling studies, it is worthwhile to take stock of the observations of RF02 in light of previous studies. In this respect the most obvious target is the heavy drizzling case analyzed by SCFM98 (hereafter HDLES), which has been commented on throughout. Even for this case the comparison is not trivial. The precipitating region of the PBL during RF02 was sampled irregularly in space and time, which impairs our ability to recreate budgets and vertical profiles of critical quantities. The limited horizontal domain of the HDLES (3.2 km) may have inhibited the formation of clearings and thus it is not clear exactly how to compare its circulations with those exhibiting a fundamental modulation on scales of 10–20 km, although we note that, if we restrict our comparisons to only the precipitating regions of the POC, the simulations begin to look more like the data.

With these caveats in mind it still seems fair to say that the central prediction of SCFM98, that “persistent, well-mixed, shallow, radiatively driven stratocumulus-topped PBLs, in which drizzle is heavy and downdrafts are negatively buoyant through a deep layer, do not exist in nature” is not supported by our analysis of the RF02 data. One of the more striking findings of RF02 is the degree to which the cloud and subcloud layer are undifferentiated despite the presence of heavy drizzle. Further, an analysis of SC leg data suggest that negatively buoyant downdrafts do exist on the scale of the POC.

Nonetheless, several features of the HDLES are evident in our data, especially on the scale of the precipitating regions within a POC. On these scales, downdrafts appear less vigorous in the subcloud layer and more marked reductions in are evident. Marked cooling and moistening of the subcloud layer due to the evaporation of drizzle are apparent; however, if this leads to differentiation between the cloud and subcloud layers, it was not sampled by our soundings. Similarly, at least at a qualitative level, differences in the turbulent fluxes between HDLES and its nondrizzling counterpart are similar to the differences between regions of precipitation within a POC and nonprecipitating regions away from the POC. All of which suggests that the SCFM98 hypothesis might fare somewhat better on more local scales.

The differences in circulation between RF02 and the HDLES might be due to two artifacts resulting from the configuration of the LES study: First, due to its small horizontal domain, statistics of the HDLES are determined over one single cell, which might have biased the value of the vertical velocity skewness toward more extreme values. Second, the HDLES can be expected to entrain more efficiently than in nature (particularly at the vertical grid spacings used for the HDLES; cf. Duynkerke et al. 2004; Stevens et al. 2005a). As shown in an LES study of RF01 (Stevens et al. 2005a), this makes the LES more susceptible to decoupling than would otherwise be, hence explaining the tendency of the HDLES to produce a more differentiated cloud/subcloud structure than observed during RF02, as well as its tendency to produce more skewed, or cumulus-like circulations, all of which have come to be associated with the decoupling transition (Bretherton and Wyant 1997; Stevens 2000; Stevens et al. 2005a).

Both to evaluate our hypothesis for the maintenance of large Θe and to reexamine how precipitation interacts with dynamic circulations it seems worthwhile to revisit the simulating framework of SCFM98. By using simpler parameterizations of precipitation, larger horizontal domains, and by taking care to limit the amount of dissipation at cloud top (e.g., by suitably choosing the subgrid model and vertical grid spacings; Stevens et al. 2005a) much could be learned from further simulations. Based on our analysis we would look to the LES to (i) reproduce the large values throughout the boundary layer, (ii) better maintain the lack of cloud/subcloud differentiation in state variables, (iii) spontaneously generate larger scales with features (elevated Θe) commensurate with the observations, and (iv) maintain the longevity of the precipitating cells. To the extent LES is capable of reproducing these features it may then be used to fine tune questions and hypotheses in ways which make them suitable to testing by future measurements. Toward this end we have idealized the basic state and boundary forcings observed during RF02 for use as the basis of the ninth intercomparison of the Global Energy and Water Cycle Experiment (GEWEX) cloud system studies boundary layer cloud working group. This intercomparison study will be similar to that performed by (Stevens et al. 2005a) and stands to tell us much about the ability of LES to teach us more about the interaction of precipitation and cloud dynamics in stratocumulus.

5. Concluding remarks

Analyzing the second flight of the DYCOMS-II field study proved to be worthwhile for studying the influence of drizzle on the mean structure and energetics of the PBL—this despite the fact that the imperfect sampling of the POC impaired our ability to study budgets and vertical profiles of critical profiles in detail. The effect of drizzle is most clearly noticeable in the subcloud layer where evaporation of rainwater leads to a moistening and cooling of the air and causes a marked reduction of the vertical velocity variance. However, despite high rain rates a stabilization of the thermodynamical profiles across the cloud base was absent. The high vertical velocity variances encountered are consistent with the observed well mixedness of the PBL. As a whole, the PBL in the POC was slightly cooler and moister than in the non-POC, which led to a lowering of the LCL. The flux budgets of total water specific humidity suggest opposite trends for the POC and the non-POC, with the PBL in the POC drying out. The temperature flux budgets are harder to interpret, although they seem to suggest a slight warming for the POC region.

A striking feature of RF02 is the elevated Θe in the showers in the POC. The enhanced values are remarkable since they seem to be clearly coupled to the presence of precipitation, even though Θe is approximately invariant to flux divergences of liquid water specific humidity. We speculate as to the efficacy of a variety of diabatic processes versus induced mesoscale circulations as a means for maintaining regions of elevated Θe and the moisture necessary to support the apparent longevity of the precipitating cells. On the scale of the POC as a whole, such circulations are consistent with the mean upward vertical velocity observed in the POC. On the scale of individual cells, upward velocities are even more pronounced, suggesting that cell-scale circulations may play a role as well.

The observed PBL allows us to consider and reject the SCFM98 hypothesis that persistent, well-mixed, shallow, stratocumulus-topped PBL in which LW radiative forcing and the maximum drizzle flux are of the same magnitude do not exist in nature. This makes RF02 an interesting candidate for future studies. If LES is capable of reproducing characteristic features of the RF02, it may be useful in testing our ideas as to the role of various processes, ranging from local diabatic effects to larger-scale circulations.

Acknowledgments

Simona Bordoni provided an analysis of the satellite-retrieved SSTs used in this study. Ongoing conversations on these topics with Bruce Albrecht, Donald Lenschow, Markus Petters, and Gabor Vali greatly enriched our analysis. Gabor Vali and Herm Gerber are thanked for their contribution in the investigations of the occurrence of cloud droplets below clouds. Andrew Ackerman and Verica Savic-Jovcic are appreciated for their constructive suggestions on (earlier) drafts of this manuscript. Rob Wood is thanked for catching an error in the last minute. Discussions with Adam Sobel on the origins of the Θe fluctuations are also gratefully acknowledged. Of course none of this would have been possible if it were not for the competence and professionalism of the NCAR and University of Wyoming staff scientists and technicians who helped maintain and deploy the facilities during DYCOMS-II. In particular both Jörgen Jensen and Alan Schanot are thanked for their contributions to our analysis of the state-variable data especially in relation to possible effects of wetting on instrumental behavior. Support by the National Science Foundation, both through Grant ATM 0097053, and the attention and support of the program officer in charge at that time (Roddy Rogers) is also gratefully acknowledged. For the first author this work is part of the research programme of the Stichting voor Fundamenteel Onderzoek der Materie (FOM), which is financially supported by the Nederlandse Organisatie voor Wetenschappelijk Onderzoek (NWO).

REFERENCES

  • Barnes, G. M., , and M. Garstang, 1982: Subcloud layer energetics of precipitating convection. Mon. Wea. Rev., 110 , 102117.

  • Bretherton, C. S., , and M. C. Wyant, 1997: Moisture transport, lower-troposheric stability, and decoupling of cloud-topped boundary layers. J. Atmos. Sci., 54 , 148167.

    • Search Google Scholar
    • Export Citation
  • Brost, R. A., , J. C. Wyngaard, , and D. H. Lenschow, 1982: Marine stratocumulus layers. Part II: Turbulence budgets. J. Atmos. Sci., 39 , 818836.

    • Search Google Scholar
    • Export Citation
  • Comstock, K. K., , C. S. Bretherton, , and S. E. Yuter, 2005: Mesoscale variability and drizzle in southeast Pacific stratocumulus. J. Atmos. Sci., 62 , 37923807.

    • Search Google Scholar
    • Export Citation
  • Duynkerke, P. G., and Coauthors, 2004: Observations and numerical simulations of the diurnal cycle of the EUROCS stratocumulus case. Quart. J. Roy. Meteor. Soc., 130 , 32693296.

    • Search Google Scholar
    • Export Citation
  • Fairall, C. W., , E. F. Bradley, , D. P. Rogers, , J. B. Edson, , and G. S. Young, 1996: Bulk parametrization of air–sea fluxes for Tropical Ocean Global Atmosphere Coupled Ocean Atmosphere Response Experiment. J. Geophys. Res., 101 , 37473764.

    • Search Google Scholar
    • Export Citation
  • Faloona, I., and Coauthors, 2005: Observations of entrainment in eastern Pacific marine stratocumulus using three conserved scalars. J. Atmos. Sci., 62 , 32683285.

    • Search Google Scholar
    • Export Citation
  • Ferek, R. J., and Coauthors, 2000: Drizzle suppression in ship tracks. J. Atmos. Sci., 57 , 27072728.

  • Fu, Q., , and K. N. Liou, 1993: Parameterization of the radiative properties of cirrus clouds. J. Atmos. Sci., 50 , 20082025.

  • Han, Q., , R. Welch, , J. Chou, , W. Rossow, , and A. White, 1995: Validation of satellite retrievals of cloud microphysics and liquid water path using observations from FIRE. J. Atmos. Sci., 52 , 41834195.

    • Search Google Scholar
    • Export Citation
  • Jensen, J. B., , S. Lee, , P. B. Krummel, , J. Katzfey, , and D. Gogoasa, 2000: Precipitation in marine cumulus and stratocumulus. Part I: Thermodynamic and dynamic observations of closed cell circulations and cumulus bands. Atmos. Res., 54 , 117155.

    • Search Google Scholar
    • Export Citation
  • Paluch, I. R., , and D. H. Lenschow, 1991: Stratiform cloud formation in the marine boundary layer. J. Atmos. Sci., 48 , 21412158.

  • Perez, J. C., , F. Herrera, , F. Rosa, , A. Gonzales, , M. Wetzel, , R. Borys, , and D. Lowenthal, 2000: Retrieval of marine stratus cloud droplet size from NOAA AVHRR night imagery. J. Remote Sens. Environ., 73 , 3145.

    • Search Google Scholar
    • Export Citation
  • Petters, M. D., , J. R. Snider, , B. Stevens, , G. Vali, , I. Faloona, , and L. Ryssell, 2005: Accumulation mode aerosol, pockets of open cells, and particle nucleation in the remote subtropical pacific marine boundary layer. J. Geophys. Res., in press.

    • Search Google Scholar
    • Export Citation
  • Sharon, T. M., , B. A. Albrecht, , H. Jonsson, , P. Minnis, , M. M. Khaiyer, , T. M. VanReken, , J. Seinfeld, , and R. Flagan, 2006: Aerosol and cloud microphysical characteristics of rifts and gradients in maritime stratocumulus clouds. J. Atmos. Sci., in press.

    • Search Google Scholar
    • Export Citation
  • Stevens, B., 2000: Cloud-transitions and decoupling in shear-free stratocumulus topped boundary layers. Geophys. Res. Lett., 27 , 25572560.

    • Search Google Scholar
    • Export Citation
  • Stevens, B., , W. R. Cotton, , G. Feingold, , and C-H. Moeng, 1998: Large-eddy simulations of strongly precipitating, shallow, stratocumulus-topped boundary layers. J. Atmos. Sci., 55 , 36163638.

    • Search Google Scholar
    • Export Citation
  • Stevens, B., and Coauthors, 2003a: Dynamics and chemistry of marine stratocumulus—DYCOMS-II. Bull. Amer. Meteor. Soc., 84 , 579593.

  • Stevens, B., and Coauthors, 2003b: On entrainment rates in nocturnal marine stratocumulus. Quart. J. Roy. Meteor. Soc., 129 , 34693493.

    • Search Google Scholar
    • Export Citation
  • Stevens, B., and Coauthors, 2005a: Evaluation of large-eddy simulations via observations of nocturnal marine stratocumulus. Mon. Wea. Rev., 133 , 14431462.

    • Search Google Scholar
    • Export Citation
  • Stevens, B., , G. Vali, , K. Comstock, , R. Wood, , M. C. Zanten, , P. Austin, , C. S. Bretherton, , and D. H. Lenschow, 2005b: Pockets of open cells (POCs) and drizzle in marine stratocumulus. Bull. Amer. Meteor. Soc., 86 , 5157.

    • Search Google Scholar
    • Export Citation
  • Vali, G., , R. D. Kelly, , J. French, , S. Haimov, , D. Leon, , R. E. McIntosh, , and A. Pazmany, 1998: Finescale structure and microphysics of coastal stratus. J. Atmos. Sci., 55 , 35403564.

    • Search Google Scholar
    • Export Citation
  • vanZanten, M. C., , B. Stevens, , G. Vali, , and D. H. Lenschow, 2005: Observations of drizzle in nocturnal marine stratocumulus. J. Atmos. Sci., 62 , 88106.

    • Search Google Scholar
    • Export Citation
  • Wang, S., , and Q. Wang, 1994: Roles of drizzle in a one-dimensional third-order turbulence closure model of nocturnal stratus-topped marine boundary layer. J. Atmos. Sci., 51 , 15591576.

    • Search Google Scholar
    • Export Citation

APPENDIX

Data

For cloud-liquid-water specific humidity the King probe was used. The amount of liquid water contained by drizzle drops was estimated by centered Riemann sums of the PMS 260X data. A combination of the two represents total amount of liquid water specific humidity.

Although leg mean values of both Rosemount temperature probes (model 102EAL, stated accuracy 0.5 K) differed only slightly (maximum difference 0.1 K), careful inspection of the data revealed a ql dependence of the difference between the two, an indication of wetting of both sensors. An objective way to determine whether the probes suffered from wetting is lacking for DYCOMS-II since the OPHIR-III radiometric thermometer did not perform satisfactorily. However, we applied a wetting correction of 0.7 K (g kg, liquid water)−1, which improved both the temperature profiles (as compared to calculated moist adiabatic temperature profiles) as the match between qυ, as measured by the cross-flow Lyman-α, and calculated saturated specific humidity values during cloud legs.

Specific humidity values were obtained with the cross-flow Lyman-α sensor. An oxygen correction was applied to the raw Lyman-α output and absolute humidity (ρ) values were obtained by fitting a third-order empirical function (A. Schanot 2004, personal communication). The Lyman-α absolute humidity data was tied to the General Eastern hygrometer absolute humidity (ρGE) by correcting the raw Lyman-α output with a linear fit to the instantaneous correction for all eight boundary layer legs. The hygrometer performed well compared to saturation humidity as calculated from temperature measurements. To obtain a similar good comparison with respect to the Lyman-α sensor the offset between ρ and ρGE was removed for the boundary layer legs. This was achieved by replacing the flight-averaged offset value in the linear fit with leg-averaged offset values.

The SPP-100 measured the cloud droplet size distribution between 2 and 47 μm divided into 40 size intervals. We combined the SPP-100 data into 19 unequal-sized bins in order to minimize sizing ambiguities. For RF02 the data is slightly questionable because it was determined that the instrument was overestimating droplet sizes by approximately two bin sizes. During CB1 the instrument failed and no data is available for this leg. To determine the sizes of drizzle drops the one-dimensional Particle Measuring Systems 260X was used. The probe functioned well during the flight. More detailed information about the microphysical instruments and data checks applied can be found in vanZanten et al. (2005).

Details on the computation of the drizzle rate can be found in vanZanten et al. (2005) but here we give an overview of the most important steps. The drizzle rate has been estimated based on both in situ and remotely sensed data by the Wyoming cloud radar (Vali et al. 1998). Truncated lognormal functions have been fitted to the cloud droplet size distribution as measured by the SPP-100 and to the drizzle drop size distribution as measured by the 260X. Based on these fits, the drizzle rate and reflectivity values have been calculated analytically over a diameter range up to 1 mm for each leg. Next, a drizzle rate–reflectivity relationship was derived specifically for RF02 at two heights: close to the sea surface and at the mean cloud base height. These relationships were used to convert radar reflectivity data at those two levels into a rain rate. The radar measured close to the surface during all legs except the surface leg, so an almost continuous estimate of the surface drizzle rate has been obtained.

Two Heimann radiometers were available on the aircraft for estimating the surface temperature. The instruments precisely tracked each other within the stated accuracy of 0.5 K. Superimposing the flight area on SSTs as derived from the Tropical Rainfall Measurement Mission (TRMM) Microwave Imager (TMI) indicates that the range of SST covered during RF02 lies between 17.5° and 19°C with an average value of 18.3°C. This range compares well to the 1°C span observed during the SF leg by the Heimann radiometers. The SF leg has a mean SST of 19.6°C, displaying a bias toward warmer values because of its location in the southeast corner of the target area. There is a consistent bias between the Heimann radiometer–derived SST and the satellite-derived estimate of 0.5°C. The satellite-derived mean SST is corrected for this bias and this value is given in Fig. 2.

Turbulent fluxes and variances were computed by using 2.5-min. segments over which the variables were demeaned and detrended, identical to the method used by Faloona et al. (2005). Leg-mean values have been computed from 23 (or 46, for two legs at the same height) half overlapping segments while the error bars represent a standard deviation of the mean based on 12 (or 24, respectively) independent segments. With respect to the conditionally sampled fluxes of the POC and non-POC area, each segment has been tagged “POC” or “non-POC” according to whether it was for 60% or more in the POC or non-POC area. Starting times of the legs have been modified slightly in a few cases such that every segment could be labeled. We also estimated the fluxes by first dividing the leg into POC and non-POC areas. We demeaned and detrended each area separately and removed fluctuations on a scale of 15 km and larger, obtaining a flux by summing and averaging over all POC (non-POC) data points. Both methods led to comparable flux values, but values of the first method are mainly used in the article since the method supplies a straightforward way of calculating uncertainties.

The radiative forcing across cloud top and across the PBL is determined by combining a radiative transfer model (Fu and Liou 1993) and measured longwave radiative (LW) fluxes. Similarly to RF01 (Stevens et al. 2003b) offsets of 17 and 14 W m−2 are added to the downward and upward LW flux values. The radiative transfer code is forced by the observed state of the atmosphere. For the PBL and the lowest part of the free troposphere the mean state as presented in Fig. 2 is taken (albeit piecewise linearized) while for the overlying free troposphere a constant lapse rate was used for the potential temperature and a constant value for the total water specific humidity (both based on the San Diego, California, sounding). Cloud liquid water was defined as half the adiabatic value. The control run had a similar acceptable agreement between model output and data as is shown for RF01 in Stevens et al. (2003b).

Fig. 1.
Fig. 1.

Early morning (1430 UTC, 0730 PDT) visible satellite imagery of RF02. The shaded region is the GOES-10 channel-1 reflectance on a linear scale ranging from 0 to 0.6. The general region of aircraft sampling is shown with the aid of three flight tracks (SC2, CT2, and SF from south to north, respectively) whose estimated airmass relative position is shown superimposed on the cloud field.

Citation: Journal of the Atmospheric Sciences 62, 12; 10.1175/JAS3611.1

Fig. 2.
Fig. 2.

Mean thermodynamic state as observed during RF02. (left to right) Liquid water potential temperature Θl, total water specific humidity qt, and liquid water specific humidity ql. Dots represent averages over 30-m intervals based on all flight data; black (gray) lines denote inner (outer) quartiles. The LCL of the mean PBL air and the inversion height are specified on the vertical axis, where the average leg heights are also indicated. The LCL is calculated from mean thermodynamical properties and the inversion height is located where the sharpest increase in the mean profile of Θl occurs. For Θl and qt the numbers on the horizontal axis represent the mean PBL value, the value at the inversion height, and the maximum value above the PBL, while for ql maximum liquid water (based on the 30-m interval mean values) and adiabatic liquid water content at the cloud top are noted. The mean SST (292.0 K) for the whole flight area is also indicated.

Citation: Journal of the Atmospheric Sciences 62, 12; 10.1175/JAS3611.1

Fig. 3.
Fig. 3.

Airmass-relative position of the two CT flight tracks at the midpoint of CT1. (left) Starting points of the legs are denoted by a closed circle. (right) Radar-based drizzle rate, R, at cloud base (light gray: drizzle rate of 0.1 mm day−1 or higher, dark gray: 1 mm day−1 or higher, black: 5 mm day−1 or higher), cloud droplet number, N, and downward LW radiative flux, FLW, as a function of location. All data points are plotted as averages over 1-km lengths. The data from CT2 are plotted in reversed order; thus note that the leftmost data values of the CT1 and CT2 legs are actually an hour apart in time. The vertical scale is denoted by the maximum value, leg mean, and minimum value of the variable.

Citation: Journal of the Atmospheric Sciences 62, 12; 10.1175/JAS3611.1

Fig. 4.
Fig. 4.

Difference in brightness temperature, ΔTB, of the 11- and 4-μm channels for the (left) 0930 and (middle) 1315 UTC satellite images. (right) The 1315 UTC image with all values of ΔTB greater than 1.5 K is masked, thus isolating pixels that are candidates for POC-like regions. The dashed lines indicate the trajectory of the mean PBL flow for a 225-min period. The ellipse shows the expected position the air mass at the northernmost point on the left dashed line at 0930 UTC should be achieved at 1315 UTC.

Citation: Journal of the Atmospheric Sciences 62, 12; 10.1175/JAS3611.1

Fig. 5.
Fig. 5.

Low-pass filtered ΔTB, the difference in brightness temperature of the 4- minus 11-μm channel as measured by the GOES-10 satellite; also superimposed on this figure are 1-km averages of the radar-based drizzle rates, R, at 70 m above sea level. For the SF leg in situ drizzle data are used because radar data are not available. POC regions (i.e., regions with ΔTB less than 1.5 K) are denoted with solid black lines, non-POC regions with dashed lines, and R with thin gray lines.

Citation: Journal of the Atmospheric Sciences 62, 12; 10.1175/JAS3611.1

Fig. 6.
Fig. 6.

Potential temperature Θ, specific humidity qυ, and vertical velocity w (with the leg mean w subtracted) along the flight path for two subcloud legs (SC2 and SF). All data points represent 500-m averages. Solid lines denote POC per the ΔTB criterion (i.e., ΔTB less than 1.5 K) with gray segments denoting the precipitating part of the POC. The thin dashed line shows the values outside the POC. The horizontal lines represent the leg mean values, and interior tick marks on the right vertical axis indicate the average over the non-POC area (thin black line), the POC (thick black line), and the drizzle areas within the POC (thick gray line). The vertical scale can be inferred by the marking of the maximum, leg mean, and minimum value of the variable.

Citation: Journal of the Atmospheric Sciences 62, 12; 10.1175/JAS3611.1

Fig. 7.
Fig. 7.

Vertical velocity variance () as a function of height. The variance in the POC area is given by an open circle, while the variance in the closed cell region is denoted by a closed circle plotted 10 m higher for clarity. The uncertainty specified is the standard deviation of the mean. The open gray circles without specified uncertainty are variance values for drizzling areas only. Information on how () was estimated can be found in the appendix Cloud depth is indicated by the gray bar.

Citation: Journal of the Atmospheric Sciences 62, 12; 10.1175/JAS3611.1

Fig. 8.
Fig. 8.

(left) Total-water specific humidity flux budget and (right) net flux Fqt as function of height. The net flux consists of the sum of the turbulent flux and the drizzle flux. The turbulent flux is given by circles, and the in situ drizzle flux FR is denoted by diamonds. Open symbols denote values in the POC region, while values in the non-POC region are given by closed symbols. Cloud depth is indicated by the gray bar. The uncertainty specified is the standard deviation of the mean.

Citation: Journal of the Atmospheric Sciences 62, 12; 10.1175/JAS3611.1

Fig. 9.
Fig. 9.

(left) Liquid water potential temperature flux budget and (right) net flux FΘl as a function of height. Symbol representation similar to Fig. 8 with the total LW radiative flux FLW denoted by open triangles. The net flux is the sum of the turbulent, radiative, and drizzle flux, yet the drizzle flux is not repeated here but can be found in Fig. 8. The uncertainty specified is the standard deviation of the mean.

Citation: Journal of the Atmospheric Sciences 62, 12; 10.1175/JAS3611.1

Fig. 10.
Fig. 10.

Conceptual rendering of POC and neighboring non-POC or stratiform region. Also shown is a schematic of the horizontal and vertical variations in Θe and inferred mesoscale circulations. Note that the net upward motion in the POC is left out, as explained in the text.

Citation: Journal of the Atmospheric Sciences 62, 12; 10.1175/JAS3611.1

Table 1.

Overview of RF02 flight legs. RL: remote sensing leg, CB: cloud base leg, SC: subcloud leg, CT: cloud-top leg, SP: special leg, and SF: surface leg; leg names numbered when two legs of the same type have been flown. All legs were circles, 30 min in duration, flown in a clockwise (CW) or counterclockwise (CCW) direction. Time is in UTC, which is LT plus 7 h. The radar-based drizzle rate at the surface, R, was estimated by following the procedures described by vanZanten et al. (2005), which are summarized in the appendix (note that for SF radar data are not available so in situ measurements are used instead). Averages are given over the whole leg, R, the POC section, RPOC, and the non-POC section, RnonPOC, respectively. Definitions of the POC and non-POC region can be found in section 2. BDL denotes below detection limit. The fraction of time spent in the POC region is given by APOC.

Table 1.
Table 2.

Vertical velocity w averaged over the POC and nonPOC areas separately after removing the leg mean. The uncertainty specified is the standard deviation of the mean. The fraction of time spent in the composite POC region is given by APOC. Note that the composite legs are arranged in chronological order.

Table 2.
Table 3.

Liquid water ql, cloud droplet number N, and effective radius re averaged over the POC and non-POC areas separately. We have calculated the averages over cloud areas (N larger than 20 cm−3) only due to a less than total cloud cover during the CB legs; uncertainty specified as std dev.

Table 3.
1

In principle, the flight position is only known relative to the geoid; however by advecting the earth-relative position backward or forward in time using flight-averaged PBL winds the airmass-relative position at the time of the satellite image can be estimated. This technique was used extensively by Stevens et al. (2003b) and throughout this manuscript as well.

2

Strictly speaking, this is only true if one neglects compositional effects on the specific heats, and for evaporation into saturated air (i.e., a reversible process); however, on short time scales these effects are expected to be too small to explain the magnitude of the observed signal.

3

The effective radius is based on the PVM-100 probe with a range up to 70 μm. With respect to RF02 the PVM-100 is known to have underestimated ql, especially with respect to CT legs, which suggests that re is underestimated in general and particularly for the POC region (e.g., cf. Fig. 8 in vanZanten et al. 2005).

Save