• Bretherton, C. S., , J. R. McCaa, , and H. Grenier, 2004a: A new parameterization for shallow cumulus convection and its application to marine subtropical cloud-topped boundary layers. Part I: Description and 1D results. Mon. Wea. Rev., 132, 864882.

    • Search Google Scholar
    • Export Citation
  • Bretherton, C. S., , M. E. Peters, , and L. E. Back, 2004b: Relationships between water vapor path and precipitation over the tropical oceans. J. Climate, 17, 15171528.

    • Search Google Scholar
    • Export Citation
  • Chaboureau, J.-P., , F. Guichard, , J.-L. Redelsperger, , and J.-P. Lafore, 2004: The role of stability and moisture in the diurnal cycle of convection over land. Quart. J. Roy. Meteor. Soc., 130, 31053117.

    • Search Google Scholar
    • Export Citation
  • Chelton, D. B., , and M. H. Freilich, 2005: Scatterometer-based assessment of 10-m wind analyses from the operational ECMWF and NCEP numerical weather prediction models. Mon. Wea. Rev., 133, 409429.

    • Search Google Scholar
    • Export Citation
  • Derbyshire, S. H., , I. Beau, , P. Bechtold, , J.-Y. Grandpeix, , J.-M. Piriou, , J.-L. Redelsperger, , and P. M. M. Soares, 2004: Sensitivity of moist convection to environmental humidity. Quart. J. Roy. Meteor. Soc., 130, 30553079.

    • Search Google Scholar
    • Export Citation
  • Dunbar, R. S., 2006: Level 2B data software interface specification (SIS-2): QuikSCAT era. Jet Propulsion Laboratory Doc. D-16079-Rev C, 68 pp. [Available online at ftp://podaac.jpl.nasa.gov/OceanWinds/quikscat/L2B/v2/docs/L2B_SIS_200609.pdf.]

  • Durre, I., , R. S. Vose, , and D. B. Wuertz, 2006: Overview of the Integrated Global Radiosonde Archive. J. Climate, 19, 5368.

  • Elsaesser, G. S., , C. D. Kummerow, , T. S. L'Ecuyer, , Y. N. Takayabu, , and S. Shige, 2010: Observed self-similarity of precipitation regimes over the tropical oceans. J. Climate, 23, 26862698.

    • Search Google Scholar
    • Export Citation
  • Emanuel, K. A., 1994: Atmospheric Convection. Oxford University Press, 580 pp.

  • Fletcher, J. K., , and C. Bretherton, 2010: Evaluating boundary layer–based mass flux closures using cloud-resolving model simulations of deep convection. J. Atmos. Sci., 67, 22122225.

    • Search Google Scholar
    • Export Citation
  • Freilich, M. H., , D. G. Long, , and M. W. Spencer, 1994: SeaWinds: A scanning scatterometer for ADEOS II-Science overview. Proc. Int. Geoscience and Remote Sensing Symp., Pasadena, CA, IEEE, 960–963.

  • Hohenegger, C., , and C. S. Bretherton, 2011: Simulating deep convection with a shallow convection scheme. Atmos. Chem. Phys. Discuss., 11, 83858430.

    • Search Google Scholar
    • Export Citation
  • Huddleston, J. N., , and B. W. Stiles, 2000: Multidimensional Histogram (MUDH) Rain Flag product description, version 3.0. Jet Propulsion Laboratory Tech. Rep., 8 pp. [Available online at ftp://podaac.jpl.nasa.gov/OceanWinds/quikscat/L2B/v2/docs/MUDH_Description_V3.pdf.]

  • Iguchi, T., , T. Kozu, , R. Meneghini, , J. Awaka, , and K. Okamoto, 2000: Rain-profiling algorithm for the TRMM Precipitation Radar. J. Appl. Meteor., 39, 20382052.

    • Search Google Scholar
    • Export Citation
  • Iguchi, T., , T. Kozu, , J. Kwiatkowski, , R. Meneghini, , J. Awaka, , and K. Okamoto, 2009: Uncertainties in the rain profiling algorithm for the TRMM Precipitation Radar. J. Meteor. Soc. Japan, 87A, 130.

    • 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.

  • Joyce, R. J., , J. E. Janowiak, , P. A. Arkin, , and P. Xie, 2004: CMORPH: A method that produces global precipitation estimates from passive microwave and infrared data at high spatial and temporal resolution. J. Hydrometeor., 5, 487503.

    • Search Google Scholar
    • Export Citation
  • Khairoutdinov, M., , and D. Randall, 2006: High-resolution simulation of shallow-to-deep convection transition over land. J. Atmos. Sci., 63, 34213436.

    • Search Google Scholar
    • Export Citation
  • Khairoutdinov, M., , S. K. Krueger, , C.-H. Moeng, , P. A. Bogenschutz, , and D. A. Randall, 2009: Large-eddy simulation of maritime deep tropical convection. J. Adv. Model. Earth Syst., 1, 15, doi:10.3894/JAMES.2009.1.15.

    • Search Google Scholar
    • Export Citation
  • Kingsmill, D. E., 1995: Convection initiation associated with a sea-breeze front, a gust front, and their collision. Mon. Wea. Rev., 123, 29132933.

    • Search Google Scholar
    • Export Citation
  • Kuang, Z., 2010: Linear response functions of a cumulus ensemble to temperature and moisture perturbations and implications for the dynamics of convectively coupled waves. J. Atmos. Sci., 67, 941962.

    • Search Google Scholar
    • Export Citation
  • Kuang, Z., , and C. S. Bretherton, 2006: A mass-flux scheme view of a high-resolution simulation of a transition from shallow to deep cumulus convection. J. Atmos. Sci., 63, 18951909.

    • Search Google Scholar
    • Export Citation
  • Kummerow, C. D., , W. Barnes, , T. Kozu, , J. Shiue, , and J. Simpson, 1998: The Tropical Rainfall Measuring Mission (TRMM) sensor package. J. Atmos. Oceanic Technol., 15, 809817.

    • Search Google Scholar
    • Export Citation
  • Lima, M. A., , and J. W. Wilson, 2008: Convective storm initiation in a moist tropical environment. Mon. Wea. Rev., 136, 18471864.

  • Liu, W. T., 2002: Progress in scatterometer applications. J. Oceanogr., 58, 121136.

  • Mapes, B. E., 2000: Convective inhibition, subgrid-scale triggering energy, and stratiform instability in a toy tropical wave model. J. Atmos. Sci., 57, 15151535.

    • Search Google Scholar
    • Export Citation
  • Mapes, B. E., , and R. Neale, 2011: Parameterizing convective organization to escape the entrainment dilemma. J. Adv. Model. Earth Syst., 3, M06004, doi:10.1029/2011MS000042.

    • Search Google Scholar
    • Export Citation
  • Masunaga, H., 2012: A satellite study of the atmospheric forcing and response to moist convection over tropical and subtropical oceans. J. Atmos. Sci., 69, 150167.

    • Search Google Scholar
    • Export Citation
  • Moeng, C.-H., , M. A. LeMone, , M. F. Khairoutdinov, , S. K. Krueger, , P. A. Bogenschutz, , and D. A. Randall, 2009: The tropical marine boundary layer under a deep convection system: A large-eddy simulation study. J. Adv. Model. Earth Syst., 1, 16, doi:10.3894/JAMES.2009.1.16.

    • Search Google Scholar
    • Export Citation
  • Neale, R. B., , J. H. Richter, , and M. Jochum, 2008: The impact of convection on ENSO: From a delayed oscillator to a series of events. J. Climate, 21, 59045924.

    • Search Google Scholar
    • Export Citation
  • Neale, R. B., and Coauthors, 2010: Description of the NCAR Community Atmosphere Model (CAM 5.0). NCAR Tech. Note NCAR/TN-486+STR, 283 pp. [Available online at http://www.cesm.ucar.edu/models/cesm1.0/cam/docs/description/cam5_desc.pdf.]

  • Neelin, J. D., , O. Peters, , J. W.-B. Lin, , K. Hales, , and C. E. Holloway, 2008: Rethinking convective quasi-equilibrium: Observational constraints for stochastic convective schemes in climate models. Philos. Trans. Roy. Soc. London, A366, 25812604.

    • Search Google Scholar
    • Export Citation
  • Nesbitt, S. W., , and E. J. Zipser, 2003: The diurnal cycle of rainfall and convective intensity according to three years of TRMM measurements. J. Climate, 16, 14561475.

    • Search Google Scholar
    • Export Citation
  • Pereira, L. G., , and S. A. Rutledge, 2006: Diurnal cycle of shallow and deep convection for a tropical land and an ocean environment and its relationship to synoptic wind regimes. Mon. Wea. Rev., 134, 26882701.

    • Search Google Scholar
    • Export Citation
  • Posselt, D. J., , S. C. van den Heever, , and G. L. Stephens, 2008: Trimodal cloudiness and tropical stable layers in simulations of radiative convective equilibrium. Geophys. Res. Lett., 35, L08802, doi:10.1029/2007GL033029.

    • Search Google Scholar
    • Export Citation
  • Raymond, D. J., , S. L. Sessions, , and Z. Fuchs, 2007: A theory for the spinup of tropical depressions. Quart. J. Roy. Meteor. Soc., 133, 17431754.

    • Search Google Scholar
    • Export Citation
  • Rio, C., , F. Hourdin, , J.-Y. Grandpeix, , and J.-P. Lafore, 2009: Shifting the diurnal cycle of parameterized deep convection over land. Geophys. Res. Lett., 36, L07809, doi:10.1029/2008GL036779.

    • Search Google Scholar
    • Export Citation
  • Romps, D. M., , and Z. Kuang, 2011: A transilient matrix for moist convection. J. Atmos. Sci., 68, 20092025.

  • Ross, A. N., , A. M. Tompkins, , and D. J. Parker, 2004: Simple models of the role of surface fluxes in convective cold pool evolution. J. Atmos. Sci., 61, 15821595.

    • Search Google Scholar
    • Export Citation
  • Sahany, S., , J. D. Neelin, , K. Hales, , and R. B. Neale, 2012: Temperature–moisture dependence of the deep convective transition as a constraint on entrainment in climate models. J. Atmos. Sci., 69, 13401358.

    • Search Google Scholar
    • Export Citation
  • Sherwood, S. C., 1999: Convective precursors and predictability in the tropical western Pacific. Mon. Wea. Rev., 127, 29772991.

  • Shige, S., , Y. N. Takayabu, , W.-K. Tao, , and D. E. Johnson, 2004: Spectral retrieval of latent heating profiles from TRMM PR data. Part I: Development of a model-based algorithm. J. Appl. Meteor., 43, 10951113.

    • Search Google Scholar
    • Export Citation
  • Shige, S., , Y. N. Takayabu, , W.-K. Tao, , and D. E. Johnson, 2007: Spectral retrieval of latent heating profiles from TRMM PR data. Part II: Algorithm improvement and heating estimates over tropical ocean regions. J. Appl. Meteor. Climatol., 46, 10981124.

    • Search Google Scholar
    • Export Citation
  • Snodgrass, E. R., , L. Di Girolamo, , and R. M. Rauber, 2009: Precipitation characteristics of trade wind clouds during RICO derived from radar, satellite, and aircraft measurements. J. Appl. Meteor. Climatol., 48, 464483.

    • Search Google Scholar
    • Export Citation
  • Stiles, B. W., 1999: Special wind vector data product: Direction Interval Retrieval with Thresholded Nudging (DIRTH), version 1.1. Jet Propulsion Laboratory Tech. Rep., 10 pp. [Available online at ftp://podaac.jpl.nasa.gov/OceanWinds/quikscat/L2B/v2/docs/Dirth-Product-Description.pdf.]

  • Tompkins, A. M., 2001: Organization of tropical convection in low vertical wind shears: The role of cold pools. J. Atmos. Sci., 58, 16501672.

    • Search Google Scholar
    • Export Citation
  • Wu, C.-M., , B. Stevens, , and A. Arakawa, 2009: What controls the transition from shallow to deep convection? J. Atmos. Sci., 66, 17931806.

    • Search Google Scholar
    • Export Citation
  • Zelinka, M. D., , and D. L. Hartmann, 2009: Response of humidity and clouds to tropical deep convection. J. Climate, 22, 23892404.

  • Zhang, G. J., , and N. A. McFarlane, 1995: Sensitivity of climate simulations to the parameterization of cumulus convection in the Canadian Climate Centre general circulation model. Atmos.–Ocean, 33, 407446.

    • Search Google Scholar
    • Export Citation
  • Zhang, Y., , and S. A. Klein, 2010: Mechanisms affecting the transition from shallow to deep convection over land: Inferences from observations of the diurnal cycle collected at the ARM Southern Great Plains site. J. Atmos. Sci., 67, 29432959.

    • Search Google Scholar
    • Export Citation
  • Zuidema, P., and Coauthors, 2012: On trade wind cumulus cold pools. J. Atmos. Sci., 69, 258280.

  • View in gallery

    Radiosonde near-surface (1000 hPa) temperature (Ta) anomalies as a function of CPKE. The black solid line denotes the mean anomaly and vertical bars outline the extent of the 95% statistical significance interval.

  • View in gallery

    Sample snapshots (4 Jan 2004) of collocated QuikSCAT orbital and CMORPH products. All valid QuikSCAT pixels (maximum of about 400) falling within the spatial extent outlined by the black box are used in the computation of gridbox CPKE. Arrows in the surface wind plot depict the surface wind velocity in various regions. The maximum length of the velocity vector corresponds to a surface wind of 20 m s−1.

  • View in gallery

    (a) Composite temporal evolution of CMORPH rainfall as a function of local rainfall maxima. Vertical lines at ±12 h are shown for visual reference only. (b) Histogram of relative frequency of occurrence (%) for all local maxima in rainfall.

  • View in gallery

    As in Fig. 3a, but for frequency of occurrence of TRMM PR (a) shallow, (b) unorganized deep, and (c) organized deep convection (DC) clusters, (d) frequency of occurrence of stratiform areal coverage greater than 70%, (e) frequency of occurrence that a PR convective ETH greater than 9 km is observed within the CMORPH grid box, and (f) frequency of occurrence that a PR convective ETH greater than 10 km is observed within the CMORPH grid box. Unorganized (organized) DC clusters are referred to as MID-LEV (DEEP) clusters in Elsaesser et al. (2010).

  • View in gallery

    As in Fig. 3a, but for radiosonde-observed saturation fraction, CAPE, and ECAPE.

  • View in gallery

    As in Fig. 3a, but for rainfall, radiosonde-observed CIN, QuikSCAT CPKE, and the ratio of CIN:CPKE.

  • View in gallery

    Composite evolution of select environmental parameters for high-amplitude (strong) and low-amplitude (weak) rainfall cycles. (a)–(f) The gray shading highlights the extended time period during which CPKE exceeds CIN; (g),(h) the vertical lines outline the same period. The axis for CAPE (ECAPE) extends along the left (right)-hand side of (e),(f).

  • View in gallery

    (a) CIN, CPKE, and the ratio of CIN:CPKE as a function of rainfall. (b) The negative buoyancy percentage contributions to CIN as a function of altitude (color shaded). Shown for reference are the traditionally defined LFC (solid white line) and LCLs (dashed white line). The integrated percentage contribution to CIN from the parcel-launch level to the LFC is shown in the bottom part of (b).

  • View in gallery

    Locations of radiosonde stations (solid blue circles).

  • View in gallery

    Relative frequency of occurrence of all hour-0 rainfall maxima as a function of radiosonde site.

  • View in gallery

    Radiosonde, QuikSCAT, and TRMM observation sample sizes as a function of hour lag and hour-0 rainfall maxima. The natural log of the count is shown since sample sizes increase by several orders of magnitude as the hour-0 rainfall maxima approaches 0 mm h−1. The minimum (maximum) radiosonde, QuikSCAT, and TRMM sample counts are 45 (23 106), 85 (39 346), and 23 (11 950), respectively.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 26 26 2
PDF Downloads 21 21 2

A Multisensor Observational Depiction of the Transition from Light to Heavy Rainfall on Subdaily Time Scales

View More View Less
  • 1 Department of Atmospheric Science, Colorado State University, Fort Collins, Colorado
© Get Permissions
Full access

Abstract

Utilizing data from the Quick Scatterometer (QuikSCAT), a new observational parameter related to mesoscale cold pool activity [termed cold pool kinetic energy (CPKE)] is developed and investigated. CPKE and the Climate Prediction Center (CPC) morphing technique (CMORPH) rainfall product (both scaled to 2.25°) are geolocated to 25 tropical island radiosonde sites. CPKE and radiosonde-derived nondilute CAPE, entraining CAPE (ECAPE), saturation fraction, and a new measure of convective inhibition (that takes into account stable layers above the LFC) are investigated with respect to rainfall time tendencies. Over the life cycle of rainfall, the composite temporal evolutions of CPKE and convective inhibition are quantitatively similar, but slightly out of phase. The maximum in CPKE precedes the maximum in convective inhibition by 3–6 h, thus allowing for an oscillation in the ratio of convective inhibition to CPKE relative to maximum rainfall. This ratio falls below unity at the time rainfall begins increasing and averages to near unity over the entire life cycle. These results imply a lagged, coupled relationship between CPKE and convective inhibition during rainfall. The rapid increase in rainfall occurs when saturation fraction and ECAPE exceed approximately 70% and 280 J kg−1, respectively, consistent with previously noted thresholds for deep convection transition. However, since similar thermodynamic conditions are found before the increase in rainfall, observations support a hypothesis that the onset time for transition from light to heavy rainfall occurs when triggering energy (as captured in CPKE) approaches and exceeds convective inhibition. The observed onset and time scale for CAPE depletion by convection is nearly equivalent to the initial temporal appearance and time duration (6–12 h) that CPKE exceeds convective inhibition.

Corresponding author address: Gregory Elsaesser, Department of Atmospheric Science, Colorado State University, 1371 Campus Delivery, Fort Collins, CO 80523-1371. E-mail: elsaesser@atmos.colostate.edu

Abstract

Utilizing data from the Quick Scatterometer (QuikSCAT), a new observational parameter related to mesoscale cold pool activity [termed cold pool kinetic energy (CPKE)] is developed and investigated. CPKE and the Climate Prediction Center (CPC) morphing technique (CMORPH) rainfall product (both scaled to 2.25°) are geolocated to 25 tropical island radiosonde sites. CPKE and radiosonde-derived nondilute CAPE, entraining CAPE (ECAPE), saturation fraction, and a new measure of convective inhibition (that takes into account stable layers above the LFC) are investigated with respect to rainfall time tendencies. Over the life cycle of rainfall, the composite temporal evolutions of CPKE and convective inhibition are quantitatively similar, but slightly out of phase. The maximum in CPKE precedes the maximum in convective inhibition by 3–6 h, thus allowing for an oscillation in the ratio of convective inhibition to CPKE relative to maximum rainfall. This ratio falls below unity at the time rainfall begins increasing and averages to near unity over the entire life cycle. These results imply a lagged, coupled relationship between CPKE and convective inhibition during rainfall. The rapid increase in rainfall occurs when saturation fraction and ECAPE exceed approximately 70% and 280 J kg−1, respectively, consistent with previously noted thresholds for deep convection transition. However, since similar thermodynamic conditions are found before the increase in rainfall, observations support a hypothesis that the onset time for transition from light to heavy rainfall occurs when triggering energy (as captured in CPKE) approaches and exceeds convective inhibition. The observed onset and time scale for CAPE depletion by convection is nearly equivalent to the initial temporal appearance and time duration (6–12 h) that CPKE exceeds convective inhibition.

Corresponding author address: Gregory Elsaesser, Department of Atmospheric Science, Colorado State University, 1371 Campus Delivery, Fort Collins, CO 80523-1371. E-mail: elsaesser@atmos.colostate.edu

1. Introduction

The general problem of isolating and understanding the mechanisms that influence the transition from shallow to deep precipitating convection on varying time scales and in different phenomena remains. While large-scale [i.e., current spatial scales of present-day general circulation model (GCM) grid boxes] parameters related to buoyancy, water vapor, and shear are clearly important in regulating shallow and deep convective states and their transitions, recent investigations have been focused on smaller-scale (i.e., unresolved in a GCM) temporally and spatially fluctuating environmental parameters as determinants in the evolution of convection. Because of the scales involved, this latter line of inquiry has been pursued from either a large-eddy simulation (LES) or cloud-resolving model (CRM) perspective (e.g., Derbyshire et al. 2004; Khairoutdinov and Randall 2006; Kuang and Bretherton 2006; Wu et al. 2009), field campaign perspective (e.g., Kingsmill 1995; Pereira and Rutledge 2006; Lima and Wilson 2008), or through consideration of regional continental locations having high-density networks of observation sites (e.g., Zhang and Klein 2010).

The degree of inhomogeneity in the background field is influenced by many local mechanisms, all of which affect the subsequent evolution of convection, are difficult to observe because of the spatial scales involved (particularly from a satellite perspective), and are missing or unresolved in a GCM. These include cold pools and the role they play in deep convective cloud development through thermodynamic effects (Tompkins 2001) or forced lifting of surface parcels above their inhibition barrier (Mapes 2000; Lima and Wilson 2008; Khairoutdinov and Randall 2006; Khairoutdinov et al. 2009; Hohenegger and Bretherton 2011), temperature and water vapor variations at small scales and the effect they may have on local buoyancy, ensuing cloud development and organization (Chaboureau et al. 2004; Fletcher and Bretherton 2010; Zhang and Klein 2010; Mapes and Neale 2011), and processes allowing for shallow clouds to become buoyant relative to the mean environment promoting the rapid development of deep cloud (Wu et al. 2009).

Of the processes mentioned, consider that some studies (Tompkins 2001; Ross et al. 2004; Moeng et al. 2009; Zuidema et al. 2012) have shown that cold pool activity can significantly affect horizontal wind velocity at a 10–30-km scale (i.e., a mesoscale impact), thus leading to significant inhomogeneity at the scale of a GCM grid box. In light of these spatial scales, the opportunity exists to exploit higher-resolution satellite products in pursuit of a signal related to cold pool activity and evolving larger-scale state of convection. Fortunately, the QuikSCAT orbital and viewing characteristics are such that wind velocity can be retrieved at a scale of 12–25 km, roughly equivalent to the scales discussed above. Therefore, a key element of this study is the focus on mesoscale fluctuations in surface wind fields estimated by QuikSCAT over the tropical oceans, with overarching goals of 1) assessing the relationship between the surface wind variability and cold pool activity, and 2) shedding light on existing ideas related to cold pools and their influence (as a dynamic lifting mechanism) on the transition from shallow to deep convection. To this end, a new observational parameter (with energy units) is developed in this paper that captures the variability in surface wind fields hypothesized to be largely associated with cold pools and/or organized mesoscale downdrafts. Observational products for rainfall and surface temperature anomalies are also used to support the cold pool interpretation.

It should be noted that while most prior modeling studies have discussed transition in the sense of progression from shallow, nonprecipitating clouds to rainfall, transition defined here specifically focuses on progression from light (likely warm-rain dominated; cloud tops below freezing level) to heavier rainfall (shown to often be associated with organized aggregates of deep convective clouds). While substantial attention is paid to cold pool activity as it relates to the discussed transition, temporal composites of nondilute and entraining convective available potential energy, column-integrated negative buoyancy, and relative humidity, centered on local rainfall maxima, are also constructed and discussed relative to rainfall tendencies. In documenting the evolution from light to heavy rainfall, high temporal resolution is desired through the sole use of observational platforms. This is accomplished through the use of multisensor satellite products and radiosonde data, merged in such a manner that 3-hourly resolution can be achieved. Based on the results, for a conditionally unstable atmosphere preconditioned with sufficient moisture, the thresholds for which are established based on prior studies, a causal relation between integrated negative buoyancy, cold pool activity, and the resultant onset and demise of heavy rainfall is proposed and discussed.

2. Observational datasets

All data and derived parameters used in this study span the time frame extending from December 2002 to November 2009. The Climate Prediction Center (CPC) morphing technique (CMORPH) product (Joyce et al. 2004) serves as the basis for rainfall observations. This technique relies on rainfall estimates derived exclusively from a number of orbiting passive microwave sensors. These estimates are propagated in time/space through use of geostationary infrared (IR) data such that global estimates of rainfall rates are produced at an approximate spatial and temporal resolution of 10 km and 30 min, respectively. These estimates, averaged to a 3-hourly, ¼° rainfall product, are used in this analysis.

The Tropical Rainfall Measuring Mission (TRMM; Kummerow et al. 1998) Precipitation Radar (PR) 2A25 product (Iguchi et al. 2000, 2009) is used to investigate convection characteristics associated with the surface rainfall values provided by the CMORPH product (where available). Radar convective echo-top height (ETH) distributions and convective/stratiform rainfall fractions from the 2A25 product are specifically used for assessing the vertical extent and organization of precipitating convection.

Twice-daily surface wind velocity information, available over all oceanic regions, is derived from the SeaWinds scatterometer on the Quick Scatterometer (QuikSCAT) satellite. Details on scatterometry, applications, and instrument specifications can be found in Freilich et al. (1994), Liu (2002), and Chelton and Freilich (2005). Through the use of a geophysical model relating wind stress, ocean emissivity, and backscattered radiation at multiple azimuth angles, surface wind velocity can be retrieved. Two wind vector solutions are provided: a baseline solution produced using a maximum likelihood estimator and ambiguity removal algorithm (Dunbar 2006), and an additional wind vector solution produced through use of the Direction Interval Retrieval with Thresholded Nudging (DIRTH) algorithm (Stiles 1999). Orbital-level 2B (L2B) wind velocity retrievals are available at both the 12.5- and 25-km spatial scale and are calibrated to a 10-m reference height. The baseline solution L2B wind vector retrievals at the 12.5-km resolution serve as the basis for defining cold pool activity.

Rain probability from the scatterometer is determined using the multidimensional histogram (MUDH) rain flag algorithm (Huddleston and Stiles 2000). Because wind vector retrievals may be degraded when rainfall is present, the MUDH rain impact flag (bit 13 of the L2B product wvc_quality_flag) is used to remove pixels potentially contaminated by rainfall. Pixel-level QuikSCAT wind estimates are used if and only if the rain impact flag is set to 0 (if an impact is detected, the flag is set to 1). Additionally, if the MUDH rain impact flag is not usable for any cell, the retrieval is immediately discarded.

Buoyancy and relative humidity characterizing the environment are derived from radiosonde-observed (raob) water vapor and temperature profiles. These profiles are available from the Integrated Global Radiosonde Archive (IGRA; Durre et al. 2006), and are produced by the National Climatic Data Center (NCDC). For any given reporting time, a raob is used if 1) the lowest-level reporting altitude is less than or equal to 150 m, 2) data are available for at least six standard pressure levels (1000, 925, 850, 700, 500, 400, 300, 250 hPa) and 3) data are available to a pressure altitude of at least 200 hPa. There are 25 raob stations used in this study (geographic locations are shown in Fig. A1). Since analyses here involve spatial colocation of multiple datasets, one of which is available for nearby oceanic pixels only (QuikSCAT), only stations located on islands are considered.

3. Atmospheric variable definitions and analysis method

Rainfall data are averaged to 2.25° square boxes and geographically centered on the raob launch sites. Given that this spatial scale is representative of the current size of grid boxes in conventional GCMs or host GCMs in multiscale modeling frameworks (MMFs), it is a pertinent scale for studying the time evolution of precipitation for convective parameterization purposes.

A cluster classification algorithm for assessing the nature of convection, developed in Elsaesser et al. (2010) and applied to TRMM PR 2A25 data, is used to characterize the raining cloud populations and degree of convective organization associated with grid boxes centered on the raob sites. Latent heating profiles from the spectral latent heating product (Shige et al. 2004, 2007) were used to further assist in characterizing the convection (details also found in Elsaesser et al. 2010). The clustering algorithm classifies the convection scene as one of the following: 1) shallow/congestus convection (little stratiform rainfall, convective ETHs < 5 km), 2) unorganized deep convection (substantial numbers of congestus clouds, smaller numbers of deep clouds extending above the freezing level and peak latent heating from 3 to 8 km), 3) organized deep convection (top-heavy latent heating, significant stratiform/anvil cloud, and substantial numbers of convective ETHs extending beyond 9 km), and 4) remnant anvil or nonconvective stratiform rainfall. The 2.25° gridbox distribution of convective ETHs, average convective rainfall, and the ratio of convective to stratiform rainfall are all used in the clustering algorithm.

Cold pool activity is summarized in a parameter termed cold pool kinetic energy (CPKE), defined as
e1
In Eq. (1), ρSFC is set to 1.2 kg m−3, primes denote departures from the spatially averaged 2.25° × 2.25° zonal (u) and meridional (υ) wind components, and the overbar indicates an average over the entire grid box (centered on each raob site). If the number of quality-controlled, valid samples within a grid box falls below 100 (or 25% of the spatial area), the grid box is not considered in the final analysis. Since frontal activity is typically nonexistent over the deep tropical oceans, boundaries over open ocean that would influence the variance in surface wind are likely related to cold pool outflow or mesoscale downdrafts from nearby regions of convection. Evidence supporting this interpretation is provided immediately below.

For each raob site, seasonal and diurnal cycles are removed from the temperature time series by calculating 30-day-average lowest-altitude temperatures (referred to as surface temperatures) valid for each raob reporting hour (typically 0000 and 1200 UTC) and subtracting these values from the actual reported surface temperatures. Figure 1 shows that average raob surface temperature anomalies decrease as CPKE increases, a trend that one would expect for a region filled with cold pools of increasing size and/or strength. Average surface temperature anomalies are on the order of −0.5 to −1.0 K for the largest CPKE values in Fig. 1 (statistically significant at the 95% level using the two-tailed Student's t test). Additionally, significant scatter in the anomalies is also present. This is to be expected given that 1) the raob temperatures are point measurements, while CPKE is characteristic of the centered grid box, and that 2) significant surface temperature inhomogeneity typically exists in a GCM-size grid box filled with cold pools of varying sizes and strengths (Tompkins 2001; Moeng et al. 2009).

Fig. 1.
Fig. 1.

Radiosonde near-surface (1000 hPa) temperature (Ta) anomalies as a function of CPKE. The black solid line denotes the mean anomaly and vertical bars outline the extent of the 95% statistical significance interval.

Citation: Journal of the Atmospheric Sciences 70, 7; 10.1175/JAS-D-12-0210.1

To visualize the variations of CPKE and the spatial scales of wind perturbations alongside surface rainfall rates, a snapshot of QuikSCAT surface wind components, CPKE, and CMORPH rainfall is shown in Fig. 2. Signatures of cold pool outflow are manifested as arcs (e.g., meridional wind panel at 1°N, 142°E), lines (e.g., zonal wind panel at 2°S, 146°E), and smaller circular surface wind enhancements (bottom right of all QuikSCAT panels). This outflow is presumably the result of convection south of the equator (Fig. 2, bottom). CPKE is elevated in regions where outflow boundaries exist or collide. These are likely regions of strong local convergence. As rainfall approaches zero or the distance from substantial rainfall increases, CPKE tends toward minimal magnitudes of less than 3 J kg−1. Figures 1 and 2 together lend observational support for the interpretation that increasing CPKE is associated with increasing cold pool activity.

Fig. 2.
Fig. 2.

Sample snapshots (4 Jan 2004) of collocated QuikSCAT orbital and CMORPH products. All valid QuikSCAT pixels (maximum of about 400) falling within the spatial extent outlined by the black box are used in the computation of gridbox CPKE. Arrows in the surface wind plot depict the surface wind velocity in various regions. The maximum length of the velocity vector corresponds to a surface wind of 20 m s−1.

Citation: Journal of the Atmospheric Sciences 70, 7; 10.1175/JAS-D-12-0210.1

Raob column-integrated relative humidity, or saturation fraction, is defined as the column-integrated mixing ratio (TPW) divided by the column-integrated saturation mixing ratio. Nondilute convective available potential energy (CAPE) is calculated using the method developed by Emanuel (1994). First, for all raob sites and times, temperature and water vapor profiles are interpolated to uniform pressure levels (1-hPa resolution). One-third of these lowest levels initially serve as launch levels for pseudoadiabatic parcel lifting. Of these levels, the launching level that yields the largest CAPE serves as the final chosen origin level. In this study, 89% (95%) of parcels have origin pressure levels exceeding 1000 (960) hPa. Thus, a majority of parcel-launch levels are within 100 m of the surface. These altitudes are consistent with the results described in the modeling study of Romps and Kuang (2011). In that study, it was found that most of the air convecting into the free troposphere originated at altitudes of less than 100 m.

Entraining CAPE (ECAPE) is computed using buoyancy profiles from an offline version of an entraining plume model used in the latest versions of the National Center for Atmospheric Research (NCAR) Community Atmosphere Model (CAM; Neale et al. 2010). In this calculation, the parcel is launched from the level of maximum moist static energy (MSE) and mixes with the environment with an assumed constant fractional entrainment rate of 1 km−1. Typically, the level of maximum MSE coincides with the parcel-launch level used in the CAPE calculation. An exception to this would occur in cases where the maximum MSE occurs above the lowest one-third levels searched during the CAPE calculation procedure. In the latest versions of CAM, a modified version of Zhang–McFarlane deep convection scheme (Zhang and McFarlane 1995; Neale et al. 2008) uses this particular computation of ECAPE for the deep convection cloud-base mass flux closure. The use of this version of ECAPE has led to improvements in the representation of convection and has increased convection sensitivity to tropospheric water vapor (Neale et al. 2008).

Column-integrated negative buoyancy represents a potential barrier to convection and is defined as the vertical integral of all raob-profile negative virtual temperature differences (between the environment and the parcel) above the parcel origin level. Note that this differs from the traditional definition of convective inhibition, which typically is defined as the integral of virtual temperature differences from the parcel origin level to the level of free convection (LFC). Temperature inversions and layers of increased stability have been found above the LFC, including near the freezing level and have been hypothesized to be related to the multimodal distribution of tropical convective clouds found in many studies (e.g., Johnson et al. 1996; Posselt et al. 2008; Elsaesser et al. 2010). A recent modeling study by Kuang (2010) suggests that stable layers located anywhere in the lower troposphere can have as strong an effect on convection as the stable layer associated with traditional convective inhibition. Therefore, the reasoning behind this particular computation of integrated negative buoyancy (often referred to as CIN) stems from the desire to not neglect any component of lower-tropospheric negative buoyancy (even above the LFC) that may have an effect on the gridbox convection. Additional discussion of CIN relative to the traditional calculation of convective inhibition is provided in a later section.

For all raob sites, a seeker function isolates and stores local temporal maxima in each gridbox rainfall time series. A second scan of all time series is performed to remove multiple local maxima that are within ±12 h of the largest-amplitude local rainfall maxima. This is done to minimize data overlap and redundant sampling that would arise in the construction of the temporal composites about local maxima, and to remove smaller-amplitude peaks that may be associated with the same rainfall life cycle that can typically last from several hours to a period approaching that of the diurnal cycle. The relative frequency of occurrence of varying-amplitude rainfall maxima as a function of raob site is shown in the appendix (Fig. A2).

The CMORPH rainfall product provides data at a fixed 3-h sampling (center times of 0150, 0450, 0750, 1050, 1350, 1650, 1950, and 2150 UTC). Raob launches take place at 0000 and 1200 UTC. The orbital configuration of QuikSCAT leads to equatorial crossings at about 0600 and 1800 local time (LT), which implies that UTC varies as a function of longitude/latitude. TRMM samples tropical regions at varying UTCs. If a rainfall maximum occurs at 0150 UTC, raob data are available at hour lags −1350, −0150, +1050, +2250, and so forth. If the same rainfall maximum magnitude occurs at 1050 UTC, then raob data are available at hour lags −1050, +0150, +1350, etc. Since a rainfall maximum can occur at any of the above CMORPH center times, a high-temporal-resolution composite of environmental parameters relative to maximum rainfall can be developed. The same argument can be applied to both QuikSCAT and TRMM. By storing the hourly time lag information for each product, convenient two-dimensional portrayals of rainfall time tendencies and environmental parameter histories (vertical extent of convection, saturation fraction, CAPE, ECAPE, CIN, and CPKE) versus the local rainfall maximum can be constructed. The number of radiosonde launches, QuikSCAT, and TRMM samples associated with this compositing strategy are shown in the appendix (Fig. A3).

4. Temporal evolution

a. Rainfall and convection characteristics

The evolution of rainfall relative to gridbox hour-0 rainfall maxima is depicted in Fig. 3a. To facilitate interpretation of this image, consider a rainfall evolution associated with an hour-0 maximum of 1.8 mm h−1. The composite of the time-lagged (±2 days) rainfall can then be found by viewing the horizontal transect anchored to the 1.8 mm h−1 rainfall rate shown on the ordinate. In this case, rainfall quickly intensifies 6 h prior to the maximum and exhibits a rapid decrease over the following 6-h period. For all hour-0 maxima up to 3.2 mm h−1, the composite time series over multiday periods are displayed. Regarding time scales for rainfall, the diurnal cycle in rainfall is clearly visible, particularly for rainfall histories that correspond to lower-amplitude hour-0 rainfall rates (see the 0.5–2.4 mm h−1 transects, for instance), where clear minima in rainfall composites exist 12 h prior to and after local maxima. An envelope of elevated rainfall extending over an approximate 2-day period (±24 h) is clearly visible for high-amplitude hour-0 rainfall evolutions, with the most substantial increase in rainfall beginning, on average, 12 h prior to the maximum amplitude. These envelopes of higher-amplitude rainfall cycles are possibly associated with 2-day waves and attendant mesoscale convective systems (MCSs).

Fig. 3.
Fig. 3.

(a) Composite temporal evolution of CMORPH rainfall as a function of local rainfall maxima. Vertical lines at ±12 h are shown for visual reference only. (b) Histogram of relative frequency of occurrence (%) for all local maxima in rainfall.

Citation: Journal of the Atmospheric Sciences 70, 7; 10.1175/JAS-D-12-0210.1

Despite the fact that launch sites are located mostly over islands such that a 2.25° rainfall average consists of predominantly oceanic pixels, there will be some influence on the timing of rainfall by land. A histogram of all local times for hour-0 rainfall maxima is shown in Fig. 3b. Local peaks in rainfall can occur at any time but are most likely to occur during the early-morning hours (0600–0800 LST), consistent with the diurnal cycle statistics characteristic of oceanic convection presented in Nesbitt and Zipser (2003). Thus, the ensemble average of all composite rainfall histories is likely representative of the overocean tropical average diurnal cycle found in a number of satellite studies.

Detailed characteristics of the convection associated with the rainfall composites are shown in Fig. 4. For hour-0 rainfall maxima greater than 2 mm h−1, the shallow convection and unorganized deep convection clusters are typically observed prior to −6 h. At hour 0, unorganized and organized deep convection clusters are always observed. Maximum convective ETHs exceed 9 (10) km nearly 85% (75%) of the time, while the areal fraction of stratiform rainfall exceeds 0.7 more than 80% of the time. Furthermore, the maximum in stratiform coverage lags the maximum convective ETH occurrences by about 3 h. Shallow convection clusters occur less than 10% of the time and are never observed for the highest-amplitude rainfall composite. The compositing strategy employed provides an Eulerian perspective of the evolution from light to heavy rainfall, but the TRMM composites suggest that for the highest-amplitude hour-0 rainfall composites, a generalization can be made that the transition from light to heavy rainfall can be thought of as a transition from shallow to deep convection [shallow clouds, to congestus, to deep convection (convective ETHs exceeding 9–10 km), to large stratiform/anvil coverage]. For the lower-amplitude rainfall composites, convection cluster occurrences are more mixed.

Fig. 4.
Fig. 4.

As in Fig. 3a, but for frequency of occurrence of TRMM PR (a) shallow, (b) unorganized deep, and (c) organized deep convection (DC) clusters, (d) frequency of occurrence of stratiform areal coverage greater than 70%, (e) frequency of occurrence that a PR convective ETH greater than 9 km is observed within the CMORPH grid box, and (f) frequency of occurrence that a PR convective ETH greater than 10 km is observed within the CMORPH grid box. Unorganized (organized) DC clusters are referred to as MID-LEV (DEEP) clusters in Elsaesser et al. (2010).

Citation: Journal of the Atmospheric Sciences 70, 7; 10.1175/JAS-D-12-0210.1

b. Saturation fraction, nondilute CAPE, and entraining CAPE

The composite evolution of rainfall (reproduced from Fig. 3a), saturation fraction, CAPE, and ECAPE is illustrated in Fig. 5. On average, the nonraining atmosphere (bottommost transect along the abscissa in all panels) is characterized by lower saturation fractions and decreased CAPE; little variability is evident over the 4-day period, too. At the 2.25° spatial scale, the transition to heavier rainfall is defined to begin at the first appearance of a positive rainfall time tendency. In Fig. 5a, this time occurs along a line extending from near −12 h for a rainfall composite with peak amplitude near 3.0 mm h−1 to approximately −4 h for a composite with peak amplitude approaching 0.5 mm h−1. For 3 h prior to and including these transition hours, saturation fractions, CAPE, and ECAPE are similar for a given composite, but vary from one rainfall composite to another (ranging from 67% to 74%, 2750 to 2900 J kg−1, and 210 to 280 J kg−1, respectively).

Fig. 5.
Fig. 5.

As in Fig. 3a, but for radiosonde-observed saturation fraction, CAPE, and ECAPE.

Citation: Journal of the Atmospheric Sciences 70, 7; 10.1175/JAS-D-12-0210.1

At this point, it is useful to cast these results within the context of prior analyses noting thresholds beyond which convection is notably deeper. A number of studies (e.g., Bretherton et al. 2004b; Raymond et al. 2007) have noted that virtually all deep precipitating convection occurs at saturation fractions exceeding about 70%, a magnitude consistent with that found in this study if one were to average over all rainfall composites in Fig. 5b. While these prior analyses have depicted this nonlinear relation from the perspective of a scatter-graph view of saturation fraction and rainfall, the relationship can be understood from the perspective of the short time-scale rainfall composites as presented in this study, where the hourly time-scale transition to heavier rainfall occurs when the atmosphere is sufficiently primed with moisture such that rising plumes maintain their buoyancy in spite of dry-air entrainment. Following the transition time, saturation fraction exhibits a general increase prior to the rainfall maximum for all composites, with a tendency for maxima in saturation fraction and rainfall to be collocated in time. The integrated water vapor and rainfall time tendencies are consistent with those found in Zelinka and Hartmann (2009).

In addition to moisture, heavy rainfall in the tropics requires sufficient conditional instability (CAPE) to be present in the atmosphere. Interestingly, regardless of the amplitude of the rainfall maximum, average CAPE preceding the heaviest rainfall in all composites appear to be quite similar (Fig. 5c), thereby countering attempts to relate CAPE at any prior time to the rainfall maximum and suggestive of an often-present reservoir of potential energy for convection. More generally, rainfall maximum aside, a visual comparison of the dissimilar rainfall and CAPE patterns highlight the difficulty one encounters in relating CAPE to rainfall over these subdaily time scales. While CAPE is considered necessary for convection, its presence does not guarantee upcoming rainfall nor dictate its intensity. This result is not completely unexpected, since CAPE sufficient for convection has been found to exist 90% of the time in another observational study (Sherwood 1999) of the tropics, despite deep convection occurring at a frequency far less than this. Depletion of CAPE begins once the appropriate triggering mechanisms are in place for convective development. CAPE reaches a minimum 3–6 h after local rainfall maxima but is never completely consumed and, instead, approaches a magnitude that is comparable to that of nonraining scenes. The significant decreases in CAPE (40% reduction in composite average) in the presence of substantial rainfall is consistent with the evolution depicted in Masunaga (2012), and suggests that convection is not in equilibrium with the environment from a CAPE perspective on these time scales.

Realizing that deep convection requires both CAPE and moisture has motivated the development a single parameter (ECAPE) that takes into account both factors. ECAPE, formulated as the column integral of a parcel's buoyancy weighted by entrainment, exhibits substantial differences in shallow versus deep convection (Neelin et al. 2008; Sahany et al. 2012). Data provided in Neelin et al. (2008) suggest that an ECAPE of approximately 300 J kg−1 delineates shallow from deep convection (the mode of convection being based on whether the level of neutral buoyancy is above or below 450 hPa). This threshold value is consistent with that found during the defined transition period of this analysis (Fig. 5d). Compared to saturation fraction and CAPE, composite ECAPE increases and peaks 3–6 h before local rainfall maxima. As discussed in section 3, recent convective parameterizations have related ECAPE to convective mass flux. The ECAPE maximum before rainfall is the result of an optimal combination of buoyancy (on average, peak cold pool cooling and drying of the surface has not yet occurred, as suggested by the still-elevated CAPE values) and moisture (composite saturation fraction is elevated, and increasing).

Saturation fraction, CAPE, and ECAPE are among a number of commonly studied parameters related to the intensity of convection. Thus far, the results have suggested that the transitions from light to heavier rainfall are collocated with various thermodynamic parameter thresholds that are consistent with prior noted threshold values serving to delineate shallow from deep convection. For each rainfall composite, why convection begins to “turn on” and increase at the particular thresholds noted remains an open question. Given a reservoir of CAPE and moisture, discussion of additional parameters that should be considered when developing a conceptual picture of the hourly time-scale development of heavier rainfall is provided below.

c. CIN, CPKE, and the transition from light to heavy rainfall

The composites of CIN and CPKE relative to evolving rainfall are shown in Fig. 6. On average, CIN reaches a minimum 3 h after rainfall begins increasing, and then increases over the entire remaining period of enhanced rainfall (Fig. 6b), eventually peaking 3–6 h after maximum rainfall. The increase in CIN is likely due to the combined effects of lower troposphere cooling by cold pools associated with evaporating rainfall and drying due to convective/mesoscale downdrafts (Tompkins 2001). Steadily increasing inhibition during rainfall was also attributed to lower-tropospheric subsidence warming surrounding deep convection in the modeling study of Khairoutdinov et al. (2009). In Fig. 6b, the large values of CIN found both in nonraining regions and immediately after the time of heaviest rainfall highlight the nonuniqueness that exists between rainfall and CIN, therefore confounding the attempt to relate near-instantaneous rainfall rate and CIN. From an evolution point of view, over these time scales, similar trends in CIN (and additionally, CAPE) with respect to a rainfall maximum can be gleaned from images presented in the modeling studies of Chaboureau et al. (2004) and Khairoutdinov et al. (2009).

Fig. 6.
Fig. 6.

As in Fig. 3a, but for rainfall, radiosonde-observed CIN, QuikSCAT CPKE, and the ratio of CIN:CPKE.

Citation: Journal of the Atmospheric Sciences 70, 7; 10.1175/JAS-D-12-0210.1

Because rainfall begins increasing before the minimum in CIN, the investigation of the transition toward heavier rainfall, and why it occurs at the particular thresholds discussed, proceeds along an avenue involving cold pool activity, as quantified by CPKE. CPKE (Fig. 6c) is highly correlated with rainfall, and peak values are nearly coincident with the hour-0 rainfall axis. For the strongest amplitude rainfall composites, CPKE is slightly asymmetric about local maxima. Guided by modeling studies, and strictly based on observations here, a two-component hypothesis for the influence of cold pools on transition from light to heavy rainfall is proposed: 1) cold pools act to increase the kinetic energy in the boundary layer such that near-surface parcels can acquire enough vertical kinetic energy to overcome the barrier (i.e., CIN), resulting in increasing cloud-base mass flux/positive rainfall time tendency; and 2) cold pool activity increases CIN by cooling and drying the boundary layer in the vicinity of convection, thereby acting as a limiting factor on further increases in rainfall. In light of these contrasting effects, efforts to understand the observed transition to heavier rainfall are pursued through consideration of both CIN and CPKE.

CIN and CPKE have an analog in recent modeling studies (e.g., Bretherton et al. 2004a; Kuang and Bretherton 2006) of transition defined according to the progression of convection from shallow, nonraining clouds to deeper cumulus. Results from those studies suggest the transition to deeper convection can be understood from the perspective of boundary layer mean vertical turbulent kinetic energy (TKE) overwhelming the layers of negative buoyancy. A vertical velocity scale can be empirically derived from boundary layer TKE and compared to a minimum vertical velocity that is necessary for a parcel to overcome any layer of inhibition that may be capping the boundary layer. When the vertical kinetic energy is sufficient so that this minimum threshold is met, the development of raining clouds commences. This idea is quantified by consideration of the ratio of integrated inhibition to TKE, and as it approaches unity (or below), the time tendency for cloud growth becomes positive. The idea extends back to Mapes (2000), where the ratio of convective inhibition to “triggering energy” was deemed important for deep convection occurrence. In that study, triggering energy was thought to encompass subgrid-scale fluctuations in equivalent potential temperature, vertical kinetic energy, and gravity wave effects on local variations in convective inhibition. Incorporating the same concept, Rio et al. (2009), Fletcher and Bretherton (2010), and Hohenegger and Bretherton (2011) have now extended the idea to study the transition to deeper, precipitating convection over larger rainfall ranges.

From an observational perspective, all smaller-scale fluctuations that would contribute to triggering energy cannot be determined. In this analysis, with transition defined as the progression from light to heavy rainfall, it is hypothesized that precipitation processes and cold pool effects are even more important than during the transition from nonraining to rainfall onset, and therefore CPKE can be thought of as a significant triggering mechanism for parcels to overcome CIN as defined in this analysis. The two parameters are evaluated as a ratio with the intention of casting this observational study within the context of prior modeling results that also considered a ratio of similar parameters.

In Fig. 6, as gridbox-mean CPKE approaches the magnitude of CIN, the ratio (Fig. 6d) of CIN:CPKE tends toward unity and the rainfall time tendency becomes positive, thus starting the transition from light to heavier rainfall. The time during which this occurs is temporally located with the saturation and ECAPE thresholds discussed earlier. In this view, reservoirs of moisture and CAPE (as identified by these thresholds) are seen as necessary factors serving to prime the environment for deep convection occurrence; however, CPKE approaching the magnitude of CIN is viewed as the trigger for determining the time at which convection intensifies. Note also that the period during which CPKE overwhelms CIN precedes the period of increased rainfall. This is an important result, particularly relating to the causal argument that CIN:CPKE serves to increase rainfall. Physically, CIN:CPKE would be related to cloud-base mass flux, a parameter sought after in convective parameterization studies. CIN:CPKE and grid-mean cloud-base mass flux would be expected to be in phase, since local ascent associated with small-scale convergence boundaries or cold pool edges quickly influences average cloud-base mass flux. The maximum in rainfall would be expected to occur after the peak in cloud-base mass flux or CIN:CPKE because of the longer time scales associated with precipitation processes.

An examination of Figs. 6a and 6c shows that observed CPKE magnitudes of 3–5 J kg−1 are associated with grid-mean rainfall rates of less than 1 mm h−1. This implies that cold pools have either propagated into the grid box after being generated by nearby deeper convection or that shallower precipitating convection itself can produce enough CPKE to start the progression. The initial increase in CPKE by shallow convection is plausible, particularly when one considers that mesoscale organization of shallow raining cumulus convection (<5-km cloud-top heights) in association with cold pool outflows has been recently observed during the Rain in Cumulus over the Ocean (RICO) field campaign (Snodgrass et al. 2009; Zuidema et al. 2012). In the transition, then, shallow raining cumulus may not serve the sole role of preconditioning the atmosphere for deep convection from a moistening perspective but may also serve the important role of increasing CPKE.

Figures 5 and 6 can be summarized into low- and high-amplitude rainfall composites (Figs. 7a,b), so defined by averaging over all rainfall evolutions associated with local rainfall maxima extending from the median to upper rainfall quartile and all evolutions associated with maxima exceeding the upper quartile in rainfall (referred to as weak and strong rainfall cycles, respectively). The median local maximum rainfall rate is 0.3 mm h−1, with 95% of the total accumulated rainfall exceeding this value. A concise view of the composite evolutions of CIN, CPKE, and CAPE for strong and weak rainfall cycles is shown in Fig. 7. The rapid drop in composite CAPE (Figs. 7e,f) takes place during the time interval that CIN ≤ CPKE, thus defining a time scale for CAPE depletion. The most rapid decrease in CAPE, likely associated with a peak in convective mass flux, occurs when CIN:CPKE reaches a minimum. Notice that the ratio of CIN:CPKE approaches unity before the minimum in CIN (Figs. 7c,d) and is low throughout, and therefore, rainfall increases for several hours before a minimum in CIN is observed. Despite CIN experiencing a rapid increase (atmosphere is stabilizing), rainfall continues to increase since CPKE still exceeds CIN. Because CIN and CPKE are out of phase (Figs. 7c,d), with CPKE peaking during maximum rainfall and CIN afterward, convection shuts down before CAPE is entirely consumed. The time period during which CPKE exceeds CIN is also temporally collocated with ECAPE values that exceed the prior discussed deep convection ECAPE threshold, with the minimum in CIN:CPKE (inferred from Fig. 7c) occurring near the maximum in ECAPE (Fig. 7e). ECAPE, which is highly influenced by midtropospheric moisture, is expected to be collocated with positive water vapor anomalies. This is verified in Figs. 7g and 7h, which depict maxima in ECAPE temporally collocated near the maximum in 600–850-hPa composite water vapor anomalies.

Fig. 7.
Fig. 7.

Composite evolution of select environmental parameters for high-amplitude (strong) and low-amplitude (weak) rainfall cycles. (a)–(f) The gray shading highlights the extended time period during which CPKE exceeds CIN; (g),(h) the vertical lines outline the same period. The axis for CAPE (ECAPE) extends along the left (right)-hand side of (e),(f).

Citation: Journal of the Atmospheric Sciences 70, 7; 10.1175/JAS-D-12-0210.1

d. Discussion of the relationship between CIN and CPKE

Figures 7c and 7d portray an interesting relationship between CIN and CPKE. The amplitudes in CPKE and shape of the evolution curves are similar to that found for CIN, although lagged by 3–6 h. This is suggestive of a coupling between the two parameters in the presence of rainfall. Relative to maximum rainfall found in Fig. 7a, if one were to isolate the identical rainfall rate on either side of the rainfall peak and evaluate the ratio of CIN:CPKE, the ratio would much less than unity if rainfall is increasing with time, and slightly greater than unity if the rainfall time tendency is negative. Based on these results, for any given rainfall rate, the averages of the CIN:CPKE ratios, ignoring whether the rainfall time tendency is positive or negative, are expected to lie close to unity. This is supported in Fig. 8a, which shows a near constant ratio of CIN:CPKE with respect to rainfall. The ratio greatly exceeds unity for nonraining scenes—a result supporting the hypothesis that heavier rainfall and increasing convection does not occur until CPKE approaches and exceeds the magnitude of CIN. An examination of Fig. 8b shows that in contrast to raining scenes, there is a large contribution to CIN from stable atmospheric levels above the traditionally defined LFC (near the 750–825-hPa levels) for nonraining grid boxes. Thus, it can be generalized that CIN, as defined in this study, is largely influenced by the presence of stable layers that may be found at altitudes above the LFC.

Fig. 8.
Fig. 8.

(a) CIN, CPKE, and the ratio of CIN:CPKE as a function of rainfall. (b) The negative buoyancy percentage contributions to CIN as a function of altitude (color shaded). Shown for reference are the traditionally defined LFC (solid white line) and LCLs (dashed white line). The integrated percentage contribution to CIN from the parcel-launch level to the LFC is shown in the bottom part of (b).

Citation: Journal of the Atmospheric Sciences 70, 7; 10.1175/JAS-D-12-0210.1

There are two possible physical interpretations of the coupling between CIN and CPKE. The first involves the local properties associated with any given cold pool, such that when convection is sufficient for rainfall and surface cold pools to develop, stronger cold pools are more dynamic in terms of creating gustiness in the surface wind field and localized regions of boundary layer convergence along their leading edges—an effect that would be reflected as an increase in CPKE. After a number of hours, as the cold pools spread out and have a cooling effect on the average lower-tropospheric temperature, CIN begins increasing. Thus, the horizontal wind velocity variance (CPKE) and the negative temperature perturbation (related to CIN) are time-lagged, characteristic properties of the same mesoscale cold pool. This interpretation requires that in the presence of rainfall, CIN be largely driven by the potential temperature state of the lowest portion of the atmosphere below an altitude that characterizes the vertical extent of the surface cold pool. This is supported in Fig. 8b, where 80%–90% of the magnitude of CIN is the result of integrated negative buoyancies found below the 925 hPa—a level close to the height of the LFC.

An alternate interpretation of the coupling relates to mesoscale downdrafts associated with large raining systems. Organized downdrafts associated with well-developed raining systems would be expected to lead to gustiness over the ocean surface (increase in CPKE). Subsidence warming of the environment surrounding convection, in association with organized mesoscale circulations, could also play a greater role in increasing CIN. In this scenario, CIN would still be influenced by the thermodynamic state of the lower portions of the atmosphere affected by subsidence warming, but may also be affected by an enhancement of the stable layers found above boundary layer. Therefore, as gridbox-mean rainfall increases, a contribution to CIN by negative buoyancy anomalies above the atmospheric boundary layer would be anticipated. The data shown in Fig. 8b may support this. Note that as the mean rainfall rate increases, negative buoyancy at levels above the LFC (above 925 hPa) are found to contribute more substantially to CIN.

Negative potential temperature perturbations associated with cold pools over a warm sea surface and increased gustiness associated with outflow can lead to greater surface fluxes in near vicinity to cold pools. This would act to decrease CIN. While this discussion has focused on the possible mechanisms that can lead to increases in CIN, this aspect of CIN modification (via boundary layer recovery) would need to be further addressed to more completely understand the relationship between CPKE and CIN.

5. Conclusions

Through the exclusive use of observations, a composite analysis of the subdaily evolution of rainfall and environmental parameters as a function of hour-0 rainfall maximum was performed. Parameters such as saturation fraction, CAPE, and ECAPE have been discussed and interpreted with respect to prior studies. On average, ECAPE maximizes 3–6 h before peak rainfall, saturation fraction maximizes near the rainfall maximum, and CAPE minimizes 3–6 h after peak rainfall. Substantial local consumption of CAPE (~40%) is observed during more intense rainfall cycles, suggesting that on average, periods of heavy rainfall are not in equilibrium with and occur in the absence of large-scale CAPE forcing.

Recent modeling studies have conceptualized the transition from shallow, nonprecipitating convection to rainfall as occurring when the boundary layer possess sufficient TKE for parcels to acquire a minimum vertical momentum needed to overcome the stable layers present in the environment. The present study considers a similar argument but geared toward a special case of transition defined as the onset of heavy rainfall relative to an environment dominated by shallow, lightly raining convection. This version of transition is proposed to be causally related to two new observational parameters: CIN, defined differently from the traditional measure of convection inhibition in that stable layers above the LFC are taken into account in determining the final inhibition measure, and CPKE, an energy-unit measure of surface wind fluctuations associated with oceanic mesoscale cold pool and/or organized downdraft activity. CPKE now takes the place of TKE in the side-by-side comparison with CIN as it is now considered the significant lifting mechanism for situations involving the transition from shallow, raining convection to deeper, heavily raining convection.

Through consideration of these two parameters, it is shown that prior established thermodynamic thresholds (in saturation fraction and ECAPE) associated with the onset of deep, precipitating convection temporally occur when CPKE first approaches the magnitude of CIN. However, instead of the reservoirs of moisture and ECAPE (as measured by these thermodynamic thresholds) serving to dictate the time for development of heavier rainfall, it is argued that the triggering mechanism of CPKE exceeding CIN determines when rainfall increases. This implies that CIN should not be evaluated in isolation as it relates to rainfall development, but instead should be quantified relative to CPKE since the ratio of the two quantities is more directly related to rainfall. Given sufficient CPKE relative to CIN in a thermodynamically favorable (in terms of moisture and CAPE) atmosphere, a continued increase in rainfall and depletion of CAPE occurs. CIN itself is a function of cold pool activity. Therefore, as rainfall progresses, cold pool activity may have the effect of increasing CIN to such an extent that it eventually overwhelms CPKE, thus leading to a quick decrease in rainfall and no further consumption of CAPE. Thus, the period of CPKE ≥ CIN outlines the time scale for CAPE depletion.

Because the transition is discussed here in terms of cold pool activity, cold pools either need to be generated by nearby deeper convection (and sufficiently long lived to propagate away to new locations) or generated by shallow raining clouds. The association with shallower raining clouds is supported by recent data collected during the RICO campaign, as discussed. This, in conjunction with the results of this analysis, implies that shallow raining clouds may serve as agents for increasing CPKE in addition to their role of moistening the lower atmosphere. Further examination of the relationship between shallow convection and cold pool activity, along with other inhomogeneous ocean surface features that may affect the magnitude of CPKE, will need to be performed.

CPKE, or “triggering energy” in a broad sense, should be representative of the entire boundary layer, and ideally should be converted to a vertical kinetic energy since vertical motions impinging on a stable lower atmosphere and contributing to cloud-base mass flux are the actual processes that should be investigated as they pertain to the transition process. However, this is where the observational approach is limited, given that retrievals are surface based and attempts to convert to vertical kinetic energy values representative of the boundary layer would require additional assumptions and empiricism beyond current observation capabilities. An additional observations-driven investigation into CPKE (and the ratio of CIN:CPKE), its influence by the large-scale environment, and its relation to cloud-base mass flux will be performed at the time when the vertical component of kinetic energy, the magnitude of which should be strongly related to CPKE, can be derived. The Eulerian analysis performed here, utilizing either a case-study approach or composite technique, is reproducible using output from a CRM simulation or LES. In such a study, an analogous parameter to CPKE could be computed by averaging wind vectors over all model grid columns to achieve the spatial field of view that a satellite observes. Model investigations such as these, along with a Lagrangian analysis of CPKE (in conjunction with CAPE and/or CIN) along the track of propagating precipitation systems, are currently planned. To the extent that CPKE is tied to larger-scale thermodynamics (inhibition, for instance) or dynamics, the overall results may provide guidance for better representation of the effects of small-scale features such as cold pools in GCMs, and thus may be useful in future development of convective parameterizations.

Acknowledgments

This work has been supported by the National Science Foundation Science and Technology Center for Multiscale Modeling of Atmospheric Processes (CMMAP), managed by Colorado State University under Cooperative Agreement ATM-0425247. The comments and suggestions of three anonymous reviewers have substantially improved this manuscript.

APPENDIX

Sample Sizes for All Data Products

Figure A1 shows the radiosonde sites serving as the centered geographic coordinates for all collocated observational datasets. Referencing the radiosonde site numbers in Fig. A1, Fig. A2 shows the contribution by each radiosonde site location to the composite results provided in the text. For the environment and rainfall evolutions associated with high-amplitude hour-0 rainfall maxima, equatorial sites in the central Indian Ocean (sites 4 and 5), western Pacific (sites 12 and 13), South Pacific convergence zone (sites 15 and 16), and western Caribbean (site 22) contribute substantially to the composites, while sites in the Atlantic scarcely contribute.

Fig. A1.
Fig. A1.

Locations of radiosonde stations (solid blue circles).

Citation: Journal of the Atmospheric Sciences 70, 7; 10.1175/JAS-D-12-0210.1

Fig. A2.
Fig. A2.

Relative frequency of occurrence of all hour-0 rainfall maxima as a function of radiosonde site.

Citation: Journal of the Atmospheric Sciences 70, 7; 10.1175/JAS-D-12-0210.1

The data sample counts for radiosonde launches and 2.25° square boxes centered on radiosonde locations containing QuikSCAT and TRMM data are shown as a function of hour lag (1-h resolution clearly depicted) and hour-0 rainfall maximum in Fig. A3.

Fig. A3.
Fig. A3.

Radiosonde, QuikSCAT, and TRMM observation sample sizes as a function of hour lag and hour-0 rainfall maxima. The natural log of the count is shown since sample sizes increase by several orders of magnitude as the hour-0 rainfall maxima approaches 0 mm h−1. The minimum (maximum) radiosonde, QuikSCAT, and TRMM sample counts are 45 (23 106), 85 (39 346), and 23 (11 950), respectively.

Citation: Journal of the Atmospheric Sciences 70, 7; 10.1175/JAS-D-12-0210.1

REFERENCES

  • Bretherton, C. S., , J. R. McCaa, , and H. Grenier, 2004a: A new parameterization for shallow cumulus convection and its application to marine subtropical cloud-topped boundary layers. Part I: Description and 1D results. Mon. Wea. Rev., 132, 864882.

    • Search Google Scholar
    • Export Citation
  • Bretherton, C. S., , M. E. Peters, , and L. E. Back, 2004b: Relationships between water vapor path and precipitation over the tropical oceans. J. Climate, 17, 15171528.

    • Search Google Scholar
    • Export Citation
  • Chaboureau, J.-P., , F. Guichard, , J.-L. Redelsperger, , and J.-P. Lafore, 2004: The role of stability and moisture in the diurnal cycle of convection over land. Quart. J. Roy. Meteor. Soc., 130, 31053117.

    • Search Google Scholar
    • Export Citation
  • Chelton, D. B., , and M. H. Freilich, 2005: Scatterometer-based assessment of 10-m wind analyses from the operational ECMWF and NCEP numerical weather prediction models. Mon. Wea. Rev., 133, 409429.

    • Search Google Scholar
    • Export Citation
  • Derbyshire, S. H., , I. Beau, , P. Bechtold, , J.-Y. Grandpeix, , J.-M. Piriou, , J.-L. Redelsperger, , and P. M. M. Soares, 2004: Sensitivity of moist convection to environmental humidity. Quart. J. Roy. Meteor. Soc., 130, 30553079.

    • Search Google Scholar
    • Export Citation
  • Dunbar, R. S., 2006: Level 2B data software interface specification (SIS-2): QuikSCAT era. Jet Propulsion Laboratory Doc. D-16079-Rev C, 68 pp. [Available online at ftp://podaac.jpl.nasa.gov/OceanWinds/quikscat/L2B/v2/docs/L2B_SIS_200609.pdf.]

  • Durre, I., , R. S. Vose, , and D. B. Wuertz, 2006: Overview of the Integrated Global Radiosonde Archive. J. Climate, 19, 5368.

  • Elsaesser, G. S., , C. D. Kummerow, , T. S. L'Ecuyer, , Y. N. Takayabu, , and S. Shige, 2010: Observed self-similarity of precipitation regimes over the tropical oceans. J. Climate, 23, 26862698.

    • Search Google Scholar
    • Export Citation
  • Emanuel, K. A., 1994: Atmospheric Convection. Oxford University Press, 580 pp.

  • Fletcher, J. K., , and C. Bretherton, 2010: Evaluating boundary layer–based mass flux closures using cloud-resolving model simulations of deep convection. J. Atmos. Sci., 67, 22122225.

    • Search Google Scholar
    • Export Citation
  • Freilich, M. H., , D. G. Long, , and M. W. Spencer, 1994: SeaWinds: A scanning scatterometer for ADEOS II-Science overview. Proc. Int. Geoscience and Remote Sensing Symp., Pasadena, CA, IEEE, 960–963.

  • Hohenegger, C., , and C. S. Bretherton, 2011: Simulating deep convection with a shallow convection scheme. Atmos. Chem. Phys. Discuss., 11, 83858430.

    • Search Google Scholar
    • Export Citation
  • Huddleston, J. N., , and B. W. Stiles, 2000: Multidimensional Histogram (MUDH) Rain Flag product description, version 3.0. Jet Propulsion Laboratory Tech. Rep., 8 pp. [Available online at ftp://podaac.jpl.nasa.gov/OceanWinds/quikscat/L2B/v2/docs/MUDH_Description_V3.pdf.]

  • Iguchi, T., , T. Kozu, , R. Meneghini, , J. Awaka, , and K. Okamoto, 2000: Rain-profiling algorithm for the TRMM Precipitation Radar. J. Appl. Meteor., 39, 20382052.

    • Search Google Scholar
    • Export Citation
  • Iguchi, T., , T. Kozu, , J. Kwiatkowski, , R. Meneghini, , J. Awaka, , and K. Okamoto, 2009: Uncertainties in the rain profiling algorithm for the TRMM Precipitation Radar. J. Meteor. Soc. Japan, 87A, 130.

    • 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.

  • Joyce, R. J., , J. E. Janowiak, , P. A. Arkin, , and P. Xie, 2004: CMORPH: A method that produces global precipitation estimates from passive microwave and infrared data at high spatial and temporal resolution. J. Hydrometeor., 5, 487503.

    • Search Google Scholar
    • Export Citation
  • Khairoutdinov, M., , and D. Randall, 2006: High-resolution simulation of shallow-to-deep convection transition over land. J. Atmos. Sci., 63, 34213436.

    • Search Google Scholar
    • Export Citation
  • Khairoutdinov, M., , S. K. Krueger, , C.-H. Moeng, , P. A. Bogenschutz, , and D. A. Randall, 2009: Large-eddy simulation of maritime deep tropical convection. J. Adv. Model. Earth Syst., 1, 15, doi:10.3894/JAMES.2009.1.15.

    • Search Google Scholar
    • Export Citation
  • Kingsmill, D. E., 1995: Convection initiation associated with a sea-breeze front, a gust front, and their collision. Mon. Wea. Rev., 123, 29132933.

    • Search Google Scholar
    • Export Citation
  • Kuang, Z., 2010: Linear response functions of a cumulus ensemble to temperature and moisture perturbations and implications for the dynamics of convectively coupled waves. J. Atmos. Sci., 67, 941962.

    • Search Google Scholar
    • Export Citation
  • Kuang, Z., , and C. S. Bretherton, 2006: A mass-flux scheme view of a high-resolution simulation of a transition from shallow to deep cumulus convection. J. Atmos. Sci., 63, 18951909.

    • Search Google Scholar
    • Export Citation
  • Kummerow, C. D., , W. Barnes, , T. Kozu, , J. Shiue, , and J. Simpson, 1998: The Tropical Rainfall Measuring Mission (TRMM) sensor package. J. Atmos. Oceanic Technol., 15, 809817.

    • Search Google Scholar
    • Export Citation
  • Lima, M. A., , and J. W. Wilson, 2008: Convective storm initiation in a moist tropical environment. Mon. Wea. Rev., 136, 18471864.

  • Liu, W. T., 2002: Progress in scatterometer applications. J. Oceanogr., 58, 121136.

  • Mapes, B. E., 2000: Convective inhibition, subgrid-scale triggering energy, and stratiform instability in a toy tropical wave model. J. Atmos. Sci., 57, 15151535.

    • Search Google Scholar
    • Export Citation
  • Mapes, B. E., , and R. Neale, 2011: Parameterizing convective organization to escape the entrainment dilemma. J. Adv. Model. Earth Syst., 3, M06004, doi:10.1029/2011MS000042.

    • Search Google Scholar
    • Export Citation
  • Masunaga, H., 2012: A satellite study of the atmospheric forcing and response to moist convection over tropical and subtropical oceans. J. Atmos. Sci., 69, 150167.

    • Search Google Scholar
    • Export Citation
  • Moeng, C.-H., , M. A. LeMone, , M. F. Khairoutdinov, , S. K. Krueger, , P. A. Bogenschutz, , and D. A. Randall, 2009: The tropical marine boundary layer under a deep convection system: A large-eddy simulation study. J. Adv. Model. Earth Syst., 1, 16, doi:10.3894/JAMES.2009.1.16.

    • Search Google Scholar
    • Export Citation
  • Neale, R. B., , J. H. Richter, , and M. Jochum, 2008: The impact of convection on ENSO: From a delayed oscillator to a series of events. J. Climate, 21, 59045924.

    • Search Google Scholar
    • Export Citation
  • Neale, R. B., and Coauthors, 2010: Description of the NCAR Community Atmosphere Model (CAM 5.0). NCAR Tech. Note NCAR/TN-486+STR, 283 pp. [Available online at http://www.cesm.ucar.edu/models/cesm1.0/cam/docs/description/cam5_desc.pdf.]

  • Neelin, J. D., , O. Peters, , J. W.-B. Lin, , K. Hales, , and C. E. Holloway, 2008: Rethinking convective quasi-equilibrium: Observational constraints for stochastic convective schemes in climate models. Philos. Trans. Roy. Soc. London, A366, 25812604.

    • Search Google Scholar
    • Export Citation
  • Nesbitt, S. W., , and E. J. Zipser, 2003: The diurnal cycle of rainfall and convective intensity according to three years of TRMM measurements. J. Climate, 16, 14561475.

    • Search Google Scholar
    • Export Citation
  • Pereira, L. G., , and S. A. Rutledge, 2006: Diurnal cycle of shallow and deep convection for a tropical land and an ocean environment and its relationship to synoptic wind regimes. Mon. Wea. Rev., 134, 26882701.

    • Search Google Scholar
    • Export Citation
  • Posselt, D. J., , S. C. van den Heever, , and G. L. Stephens, 2008: Trimodal cloudiness and tropical stable layers in simulations of radiative convective equilibrium. Geophys. Res. Lett., 35, L08802, doi:10.1029/2007GL033029.

    • Search Google Scholar
    • Export Citation
  • Raymond, D. J., , S. L. Sessions, , and Z. Fuchs, 2007: A theory for the spinup of tropical depressions. Quart. J. Roy. Meteor. Soc., 133, 17431754.

    • Search Google Scholar
    • Export Citation
  • Rio, C., , F. Hourdin, , J.-Y. Grandpeix, , and J.-P. Lafore, 2009: Shifting the diurnal cycle of parameterized deep convection over land. Geophys. Res. Lett., 36, L07809, doi:10.1029/2008GL036779.

    • Search Google Scholar
    • Export Citation
  • Romps, D. M., , and Z. Kuang, 2011: A transilient matrix for moist convection. J. Atmos. Sci., 68, 20092025.

  • Ross, A. N., , A. M. Tompkins, , and D. J. Parker, 2004: Simple models of the role of surface fluxes in convective cold pool evolution. J. Atmos. Sci., 61, 15821595.

    • Search Google Scholar
    • Export Citation
  • Sahany, S., , J. D. Neelin, , K. Hales, , and R. B. Neale, 2012: Temperature–moisture dependence of the deep convective transition as a constraint on entrainment in climate models. J. Atmos. Sci., 69, 13401358.

    • Search Google Scholar
    • Export Citation
  • Sherwood, S. C., 1999: Convective precursors and predictability in the tropical western Pacific. Mon. Wea. Rev., 127, 29772991.

  • Shige, S., , Y. N. Takayabu, , W.-K. Tao, , and D. E. Johnson, 2004: Spectral retrieval of latent heating profiles from TRMM PR data. Part I: Development of a model-based algorithm. J. Appl. Meteor., 43, 10951113.

    • Search Google Scholar
    • Export Citation
  • Shige, S., , Y. N. Takayabu, , W.-K. Tao, , and D. E. Johnson, 2007: Spectral retrieval of latent heating profiles from TRMM PR data. Part II: Algorithm improvement and heating estimates over tropical ocean regions. J. Appl. Meteor. Climatol., 46, 10981124.

    • Search Google Scholar
    • Export Citation
  • Snodgrass, E. R., , L. Di Girolamo, , and R. M. Rauber, 2009: Precipitation characteristics of trade wind clouds during RICO derived from radar, satellite, and aircraft measurements. J. Appl. Meteor. Climatol., 48, 464483.

    • Search Google Scholar
    • Export Citation
  • Stiles, B. W., 1999: Special wind vector data product: Direction Interval Retrieval with Thresholded Nudging (DIRTH), version 1.1. Jet Propulsion Laboratory Tech. Rep., 10 pp. [Available online at ftp://podaac.jpl.nasa.gov/OceanWinds/quikscat/L2B/v2/docs/Dirth-Product-Description.pdf.]

  • Tompkins, A. M., 2001: Organization of tropical convection in low vertical wind shears: The role of cold pools. J. Atmos. Sci., 58, 16501672.

    • Search Google Scholar
    • Export Citation
  • Wu, C.-M., , B. Stevens, , and A. Arakawa, 2009: What controls the transition from shallow to deep convection? J. Atmos. Sci., 66, 17931806.

    • Search Google Scholar
    • Export Citation
  • Zelinka, M. D., , and D. L. Hartmann, 2009: Response of humidity and clouds to tropical deep convection. J. Climate, 22, 23892404.

  • Zhang, G. J., , and N. A. McFarlane, 1995: Sensitivity of climate simulations to the parameterization of cumulus convection in the Canadian Climate Centre general circulation model. Atmos.–Ocean, 33, 407446.

    • Search Google Scholar
    • Export Citation
  • Zhang, Y., , and S. A. Klein, 2010: Mechanisms affecting the transition from shallow to deep convection over land: Inferences from observations of the diurnal cycle collected at the ARM Southern Great Plains site. J. Atmos. Sci., 67, 29432959.

    • Search Google Scholar
    • Export Citation
  • Zuidema, P., and Coauthors, 2012: On trade wind cumulus cold pools. J. Atmos. Sci., 69, 258280.

Save