• Anderberg, M. R., 1973: Cluster Analysis for Applications.Academic Press, 359 pp.

  • Anderson, S. P., , R. A. Weller, , and R. B. Lukas, 1996: Surface buoyancy forcing and the mixed layer of the western Pacific warm pool: Observations and 1D model results. J. Climate, 9, 30563085.

    • Search Google Scholar
    • Export Citation
  • Bauer, P., , P. Lopez, , D. Salmond, , A. Benedetti, , S. Saarinen, , and M. Bonazzola, 2006: Implementation of 1D+4D-Var assimilation of precipitation-affected microwave radiances at ECMWF. II: 4D-Var. Quart. J. Roy. Meteor. Soc., 132, 23072332.

    • Search Google Scholar
    • Export Citation
  • Chelton, D. B., , and F. J. Wentz, 2005: Global microwave satellite observations of sea surface temperature for numerical weather prediction and climate research. Bull. Amer. Meteor. Soc., 86, 10971115.

    • Search Google Scholar
    • Export Citation
  • de Boyer Montégut, C., , G. Madec, , A. S. Fischer, , A. Lazar, , and D. Iudicone, 2004: Mixed layer depth over the global ocean: An examination of profile data and a profile-based climatology. J. Geophys. Res., 109, C12003, doi:10.1029/2004JC002378.

    • Search Google Scholar
    • Export Citation
  • Dee, D. P., and Coauthors, 2011: The ERA-Interim reanalysis: Configuration and performance of the data assimilation system. Quart. J. Roy. Meteor. Soc., 137, 553597, doi:10.1002/qj.828.

    • Search Google Scholar
    • Export Citation
  • Dias, J., , S. Tulich, , and G. Kiladis, 2012: An object-based approach to assessing tropical convection organization. J. Atmos. Sci., 69, 24882504.

    • Search Google Scholar
    • Export Citation
  • Elsaesser, G. S., , and C. D. Kummerow, 2008: Toward a fully parametric retrieval of the nonraining parameters over the global oceans. J. Appl. Meteor. Climatol., 47, 15991618.

    • Search Google Scholar
    • Export Citation
  • 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
  • Freitag, H. P., , M. E. McCarty, , C. Nosse, , R. Lukas, , M. J. McPhaden, , and M. F. Cronin, 1999: COARE Seacat data: Calibrations and quality control procedures. NOAA Tech. Memo. ERL PMEL-115, 89 pp. [Available online at http://www.pmel.noaa.gov/pubs/PDF/frei2034/frei2034.pdf.]

  • Godfrey, J. S., , and E. J. Lindstrom, 1989: The heat budget of the equatorial western Pacific surface mixed layer. J. Geophys. Res., 94 (C6), 80078017.

    • Search Google Scholar
    • Export Citation
  • Gosnell, R., , C. Fairall, , and P. J. Webster, 1995: The sensible heat of rainfall in the tropical ocean. J. Geophys. Res., 100 (C9), 18 43718 442.

    • Search Google Scholar
    • Export Citation
  • Jakob, C., , and G. Tselioudis, 2003: Objective identification of cloud regimes in the tropical western Pacific. Geophys. Res. Lett., 30, 2082, doi:10.1029/2003GL018367.

    • Search Google Scholar
    • Export Citation
  • Jakob, C., , and C. Schumacher, 2008: Precipitation and latent heating characteristic of the major tropical western Pacific cloud regimes. J. Climate, 21, 43484364.

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

    • Search Google Scholar
    • Export Citation
  • 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
  • Katsaros, K. B., , and K. J. K. Buettner, 1969: Influence of rainfall on temperature and salinity of the ocean surface. J. Appl. Meteor., 8, 1518.

    • Search Google Scholar
    • Export Citation
  • Kawai, Y., , and A. Wada, 2007: Diurnal sea surface temperature variation and its impact on the atmosphere and ocean: A review. J. Oceanogr., 63, 721744.

    • Search Google Scholar
    • Export Citation
  • Kikuchi, K., , and B. Wang, 2008: Diurnal precipitation regimes in the global tropics. J. Climate, 21, 26802696.

  • 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
  • Kummerow, C. D., , S. Ringerud, , J. Crook, , D. Randel, , and W. Berg, 2011: An observationally generated a priori database for microwave rainfall retrievals. J. Atmos. Oceanic Technol., 28, 113130.

    • Search Google Scholar
    • Export Citation
  • Lee, D., , L. Oreopoulos, , G. J. Huffman, , W. B. Rossow, , and I.-S. Kang, 2013: The precipitation characteristics of ISCCP tropical weather states. J. Climate, 26, 772788.

    • Search Google Scholar
    • Export Citation
  • Li, Y., and R. E. Carbone, 2012: Excitation of rainfall over the tropical western Pacific. J. Atmos. Sci., 69, 29832994.

  • Machado, L. A. T., , W. B. Rossow, , R. L. Guedes, , and A. W. Walker, 1998: Life cycle variations of mesoscale convective systems over the Americas. Mon. Wea. Rev., 126, 16301654.

    • Search Google Scholar
    • Export Citation
  • Mapes, B., , S. Tulich, , J. Lin, , and P. Zuidema, 2006: The mesoscale convection life cycle: Building block or prototype for large-scale tropical waves? Dyn. Atmos. Oceans, 42, 329.

    • Search Google Scholar
    • Export Citation
  • Mapes, B., , R. Milliff, , and J. Morzel, 2009: Composite life cycle of maritime tropical mesoscale convective systems in scatterometer and microwave satellite observations. J. Atmos. Sci., 66, 199208.

    • 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
  • Masunaga, H., , and C. D. Kummerow, 2006: Observations of tropical precipitating clouds ranging from shallow to deep convective systems. Geophys. Res. Lett., 33, L16805, doi:10.1029/2006GL026547.

    • Search Google Scholar
    • Export Citation
  • McPhaden, M. J., and Coauthors, 2010: The global tropical moored buoy array. Proceedings of OceanObs '09: Sustained Ocean Observations and Information for Society, J. Hall, D. E. Harrison, and D. Stammer, Eds., Vol. 2, ESA Publ. WPP-306. [Available online at http://www.pmel.noaa.gov/tao/proj_over/pubs/McPhaden_Oceanobs09.pdf.]

  • Mitovski, T., , I. Folkins, , K. von Salzen, , and M. Sigmond, 2010: Temperature, relative humidity, and divergence response to high rainfall events in the tropics: Observations and models. J. Climate, 23, 36133625.

    • Search Google Scholar
    • Export Citation
  • Mohr, K. I., , J. S. Famiglietti, , and E. J. Zipser, 1999: The contribution to tropical rainfall with respect to convective system type, size, and intensity estimated from the ice scattering signature. J. Appl. Meteor., 38, 596606.

    • Search Google Scholar
    • Export Citation
  • Reynolds, R. W., , T. M. Smith, , C. Liu, , D. B. Chelton, , K. S. Casey, , and M. G. Schlax, 2007: Daily high-resolution-blended analyses for sea surface temperature. J. Climate, 20, 54735496.

    • Search Google Scholar
    • Export Citation
  • Richards, K. J., , M. E. Inall, , and N. C. Wells, 1995: The diurnal mixed layer and the upper ocean heat budget in the western equatorial Pacific. J. Geophys. Res., 100 (C4), 68656879.

    • Search Google Scholar
    • Export Citation
  • Rossow, W. B., , G. Tselioudis, , A. Polak, , and C. Jakob, 2005: Tropical climate described as a distribution of weather states indicated by distinct mesoscale cloud property mixtures. Geophys. Res. Lett., 32, L21812, doi:10.1029/2005GL024584.

    • Search Google Scholar
    • Export Citation
  • Rossow, W. B., , A. Mekonnen, , C. Pearl, , and W. Goncalves, 2013: Tropical precipitation extremes. J. Climate, 26, 14571466.

  • Sapiano, M. R. P., , and P. A. Arkin, 2009: An intercomparison and validation of high-resolution satellite precipitation estimates with 3-hourly gauge data. J. Hydrometeor., 10, 149166.

    • Search Google Scholar
    • Export Citation
  • Skok, G., , J. Tribbia, , J. Rakovec, , and B. Brown, 2009: Object-based analysis of satellite-derived precipitation systems over the low- and midlatitude Pacific Ocean. Mon. Wea. Rev., 137, 31963218.

    • Search Google Scholar
    • Export Citation
  • Stackhouse, P. W., Jr., , S. K. Gupta, , S. J. Cox, , J. C. Mikovitz, , T. Zhang, , and L. M. Hinkelman, 2011: 24.5-year SRB dataset released. GEWEX News, No. 21 (1), International GEWEX Project Office, Silver Spring, MD, 10–12. [Available online at http://www.gewex.org/images/Feb2011.pdf.]

  • Stephens, G. L., and Coauthors, 2010: Dreary state of precipitation in global models. J. Geophys. Res., 115, D24211, doi:10.1029/2010JD014532.

    • Search Google Scholar
    • Export Citation
  • Tan, J., , C. Jakob, , and T. P. Lane, 2013: On the identification of the large-scale properties of tropical convection using cloud regimes. J. Climate, 26, 66186632.

    • Search Google Scholar
    • Export Citation
  • Tulich, S., , and G. Kiladis, 2012: Squall lines and convectively coupled gravity waves in the tropics: Why do most cloud systems propagate westward? J. Atmos. Sci., 69, 29953012.

    • Search Google Scholar
    • Export Citation
  • Wheeler, M., , and G. N. Kiladis, 1999: Convectively coupled equatorial waves: Analysis of clouds and temperature in the wavenumber–frequency domain. J. Atmos. Sci., 56, 374399.

    • Search Google Scholar
    • Export Citation
  • Williams, M., , and R. A. Houze Jr., 1987: Satellite-observed characteristics of winter monsoon cloud clusters. Mon. Wea. Rev., 115, 505519.

    • Search Google Scholar
    • Export Citation
  • Yuan, J., , and R. A. Houze Jr., 2010: Global variability of mesoscale convective system anvil structure from A-Train satellite data. J. Climate, 23, 58645888.

    • Search Google Scholar
    • Export Citation
  • View in gallery

    Example of a fast tracked system. Rainfall rate from CMORPH is contoured, with the threshold rate for tracking marked with a dashed red contour; the O signifies the position of a moored buoy from which a time series of SST was taken at time step t = 4.

  • View in gallery

    Histograms of (top) duration, (middle) zonal distance traveled, and (bottom) zonal speed for tracked systems, separated into the ocean basins defined in Table 2.

  • View in gallery

    Contoured frequency by altitude diagrams (CFADs) of mean reflectivity profiles from TRMM PR, shown for the TMI-derived (a),(b) midlevel and (c),(d) deep convective precipitation regimes. These are separated into PR pixels marked as stratiform and convective.

  • View in gallery

    Profiles of mean vertical wind shear magnitude at system passage for fast and slow systems of all basins.

  • View in gallery

    ISCCP weather state relative frequencies of occurrence for (top) fast and (bottom) slow systems.

  • View in gallery

    System-relative evolution of (top) TPW, (middle) water vapor convergence, and (bottom) rain-rate evolution by basin, shown for fast (dashed) and slow (solid) systems.

  • View in gallery

    System-relative evolution of (top) LHF and (bottom) SST. The LHF from SeaFlux is shown for fast (dashed) and slow (solid) systems and separated by basin. The SST evolution is representative only of cases collocated with buoys, with all system speeds considered for each basin.

  • View in gallery

    System-relative evolution of SST from SeaFlux (solid) and buoy (dashed) data, with zero defined at a lag of −72 h.

  • View in gallery

    (top) System-relative net radiative flux into the surface and (bottom) downward shortwave radiation at the surface, shown for fast (dashed) and slow (solid) systems and separated by basin.

  • View in gallery

    Surface radiative flux, buoy-based SST anomaly, and the first time derivative of SST coevolution for all tracked systems matched with buoys.

  • View in gallery

    Percent difference between fast and slow rain-rate PDFs [(PDFslow − PDFfast)/PDFslow] following the passage of tracked systems in all basins. Warm (cold) colors denote contours of 20% increments where environments affected by slow (fast) systems are more likely to reach the given rain-rate bin, with values between −20% and +20% left uncontoured. Overplotted is the SST anomaly from SeaFlux for all (green), fast (purple), and slow (blue) systems, with zero defined at a lag of −72 h.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 78 78 11
PDF Downloads 34 34 3

A Lagrangian Analysis of Deep Convective Systems and Their Local Environmental Effects

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

Abstract

Life cycles of deep convective raining systems are documented through use of a Lagrangian tracking algorithm applied to high-resolution Climate Prediction Center morphing technique (CMORPH) rainfall data, permitting collocation with related environmental ancillary fields and the International Satellite Cloud Climatology Project (ISCCP) cloud states (). System life cycles are described in terms of propagation speed, duration, and dominant cloud structures. Tracked systems are usually associated with the ISCCP weather state 1 (WS1) deep convection cloud state and an independent, microwave-based deep convective precipitation regime developed here. The distribution and characteristics of tracked systems are found to be similar between ocean basins in terms of system speed and duration, with westward-propagating systems predominant in every basin.

The effects that these systems have on environmental parameters are assessed, stratified according to their average propagation speed and by ocean basin. Regardless of system speed the net effect on the environment is similar, with the largest difference being how quickly changes occur, with net surface radiation decreasing about 150 W m−2 and total precipitable water perturbed by 5–7 kg m−2; sea surface temperature (SST) drops 0.2°–0.3°C over 24 h, with system speed affecting how long SSTs remain depressed. The observed drop in SST is partly caused by the presence of widespread, optically thick clouds that greatly decrease the net surface radiative flux. Quick changes in SSTs caused by tracked systems are captured by buoys but not represented well in gridded SST products, as these regions remain largely under the precipitating cloud cover associated with these systems.

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

Abstract

Life cycles of deep convective raining systems are documented through use of a Lagrangian tracking algorithm applied to high-resolution Climate Prediction Center morphing technique (CMORPH) rainfall data, permitting collocation with related environmental ancillary fields and the International Satellite Cloud Climatology Project (ISCCP) cloud states (). System life cycles are described in terms of propagation speed, duration, and dominant cloud structures. Tracked systems are usually associated with the ISCCP weather state 1 (WS1) deep convection cloud state and an independent, microwave-based deep convective precipitation regime developed here. The distribution and characteristics of tracked systems are found to be similar between ocean basins in terms of system speed and duration, with westward-propagating systems predominant in every basin.

The effects that these systems have on environmental parameters are assessed, stratified according to their average propagation speed and by ocean basin. Regardless of system speed the net effect on the environment is similar, with the largest difference being how quickly changes occur, with net surface radiation decreasing about 150 W m−2 and total precipitable water perturbed by 5–7 kg m−2; sea surface temperature (SST) drops 0.2°–0.3°C over 24 h, with system speed affecting how long SSTs remain depressed. The observed drop in SST is partly caused by the presence of widespread, optically thick clouds that greatly decrease the net surface radiative flux. Quick changes in SSTs caused by tracked systems are captured by buoys but not represented well in gridded SST products, as these regions remain largely under the precipitating cloud cover associated with these systems.

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

1. Introduction

Clouds, precipitation, and their effects on the local environment are inextricably linked, yet are often treated separately. New paradigms that seek to exploit the relationship between cloud and precipitation states offer some hope to link the two disciplines. Studies that have endeavored to categorize “weather states” have approached the issue from the perspective of either classifying similar cloud states (e.g., Jakob and Tselioudis 2003; Rossow et al. 2005) or precipitation states (Masunaga and Kummerow 2006; Elsaesser et al. 2010). These frameworks can provide the so-called building blocks of the convective life cycle (Mapes et al. 2006) whose constituents are expected to have differing roles in both the water and energy cycles. More recent efforts have combined these types of analyses, examining cloud states and precipitation together (Jakob and Schumacher 2008; Lee et al. 2013; Rossow et al. 2013; Tan et al. 2013) to determine which cloud regimes are the biggest drivers of precipitation in the tropics. Deep convective cloud regimes have been shown to account for the largest fraction of total rainfall in spite of their low frequencies of occurrence (Lee et al. 2013), with active mesoscale convective systems accounting for 56% of total precipitation in the tropics by one estimate (Yuan and Houze 2010).

In the case of the tropical ocean, the mean climate can differ substantially between ocean basins, but these mean states comprise highly similar elements (or weather states) occurring at varying frequencies; this is true whether one examines scenes of precipitation (Elsaesser et al. 2010) or clouds (Rossow et al. 2005). Such findings pave the way for further analysis, allowing for a broader, more global picture of tropical precipitation to emerge from distinct precipitation regimes. This also provides a possible framework through which models may be analyzed to see how accurately these states, and thus the hydrological cycle, are being represented.

Despite the importance of deep convective raining systems to the tropical water and energy budgets, their environmental effects on local scales are not well documented. In a composite analysis, Mapes et al. (2009) found local increases in precipitable water centered on the emergence of cold cloud tops, with surface wind convergence before and divergence after. Local effects may be manifest as an average discharge of convective available potential energy out of the atmospheric column of around 400 J kg−1 (Masunaga 2012) or be as extreme as a 200 W m−2 flux of sensible heat out of the ocean due to rainfall alone (Gosnell et al. 1995). Importantly, if deep convective rainfall events affect the ocean and atmospheric boundary layer, or have a net effect on water vapor in the atmospheric column, they could be part of a feedback on subsequent precipitation in that area.

This study uses Climate Prediction Center (CPC) morphing technique (CMORPH; Joyce et al. 2004) data, a high-resolution blended rainfall product, to track deep convective systems. Other studies have used different metrics for tracking tropical systems: Wheeler and Kiladis (1999) use outgoing longwave radiation as a proxy for deep convection; Machado et al. (1998) use cloud-top temperatures from geostationary infrared sensors to track specific systems; Dias et al. (2012) take brightness temperatures from a high-resolution gridded product and assess deep convection based upon a range of thresholds. The “object based” method of Dias et al. (2012) requires systems to exist between 15°N and 15°S and be contiguous in the latitude–longitude–time domain. Since these criteria are similar to those used in this study, Dias et al. (2012) provide a valuable subject of comparison, despite their focus on equatorial wave types and not precipitation. A more similar method is found in Skok et al. (2009), another study that uses rain rates directly for the purpose of system tracking.

Precipitation events affect the local environment on various spatial and temporal scales, dependent on aspects of the precipitating system and the state of the environment. These effects are manifested specifically in variables that comprise the water and energy cycles, with the environmental response and any associated feedbacks, such as modulations of subsequent precipitation intensity, being closely related to the state of the raining system. A guiding assumption of this study is that the most extreme raining systems will present the clearest signal on constituents of the water and energy cycles such as water vapor, latent heat fluxes, and sea surface temperature (SST); deep convective raining systems and their associated cloud fields produce a large fraction of total tropical precipitation despite being an infrequent weather state (e.g., Elsaesser et al. 2010; Lee et al. 2013) and are thus the focus here.

As a check on the tracking method used, an objective method of system classification based solely on passive microwave data is developed here. This objective method of scene classification also serves to justify the rain-rate threshold chosen to define which systems are tracked. While similar to the method of scene classification presented in Elsaesser et al. (2010), the method used here is more generalizable for other purposes because of its sole reliance on passive microwave data and the extensive spatiotemporal coverage of microwave sensors. Comparisons of these microwave-only precipitation states with International Satellite Cloud Climatology Project (ISCCP) cloud states are provided in section 3.

This study combines a tracking algorithm, mesoscale cloud state, and various ancillary data sources to assess the impacts of deep convective raining systems in the tropical ocean on elements of the water and energy cycles. The remainder of this paper is organized as follows. Data sources are described in the next section. Section 3 details the methods employed for tracking the systems studied, provides an overview of the microwave-based method of scene classification, and describes how the time series of mean environmental evolution were constructed. In section 4 the mean effects of deep convective raining systems on the environment are analyzed. Section 5 examines the relationship between radiative fluxes and SST variability. A discussion of all results follows in section 6.

2. Data sources

A high-resolution gridded rainfall product is used to track precipitation systems in the tropical ocean. CMORPH uses rainfall estimates from available orbiting microwave radiometers, combining them with the help of infrared satellite imagery to project the microwave signals into areas that lack direct microwave measurements (Joyce et al. 2004). CMORPH provides rain-rate estimates for 3-h periods at a ¼° resolution spanning 60°N–60°S. The CMORPH dataset has been found to faithfully resolve the diurnal cycle of precipitation and, when compared to similar high-resolution blended precipitation products, has the highest correlation with validation data and displays a near-zero mean bias (Sapiano and Arkin 2009). CMORPH rain-rate estimates give the average rain rate of a grid box over a 3-h period, say 0600–0900 UTC, and thus other datasets used are interpolated to coincide with the midpoint of these periods, such as 0730 UTC in this example.

Cloud regime data come from ISCCP. Specifically, the cloud data used are exactly those described in Rossow et al. (2005). Cloud weather state data span the tropics from 15°N to 15°S at a 2.5° grid resolution at 3-hourly intervals, but are only available during daylight hours because of a reliance on visible satellite channels. A K-means clustering algorithm (Anderberg 1973) is used to find consistent groups of cloud properties based upon histograms of cloud-top pressure and optical depth measurements; ISCCP data from 2003 and 2004 are used in this study. Frequencies and descriptions of the six ISCCP weather state (WS) classifications are found in Table 1, with more detailed information found in Rossow et al. (2005). The first three weather states, WS1–WS3, are the “convectively active” tropical cloud regimes, with WS1 being especially relevant here as the state signifying vigorous, large-scale deep convection.

Table 1.

ISCCP weather state descriptions (from Rossow et al. 2005).

Table 1.

State variables are culled from a few sources that provide gridded data at a 3- or 6-hourly resolution. Surface latent heat flux (LHF) and SST data are taken from the SeaFlux product (Clayson et al. 2013, manuscript submitted to Int. J. Climatol.). SST is calculated via a diurnal parameterization applied to version 2 of the Reynolds Advanced Very High Resolution Radiometer (AVHRR)-only daily SST product (Reynolds et al. 2007); LHF is calculated by bulk formula, with the component pieces of near-surface humidity difference and winds coming from satellite estimates. Respectively, the SST and LHF datasets exhibit biases of 0.11°C and −2.1 W m−2, with standard errors of 0.72°C and 38 W m−2, when compared with in situ data (Clayson et al. 2013, manuscript submitted to Int. J. Climatol.). SeaFlux data are used at 0.25°, 3-hourly resolution for both LHF and SST. Water vapor and wind vectors at various pressure levels are taken from the European Centre for Medium-Range Weather Forecasts (ECMWF) Interim Re-Analysis (ERA-Interim; Dee et al. 2011). Column-integrated water vapor convergence is calculated from ERA-Interim data, as are total precipitable water (TPW) and vertical wind shear. Wind shear is defined as the magnitude of zonal and meridional combined shear. ERA-Interim data are on a 0.75° global grid with a 6-hourly time resolution at 37 pressure levels from 1000 to 1 hPa. Longwave and shortwave radiation data come from the Surface Radiation Budget (SRB) 3.1 and 3.0 release products, respectively (Stackhouse et al. 2011). Shortwave and longwave fluxes have biases of −3.0 and −0.7 W m−2, with respective RMS errors of 19.3 and 10.4 W m−2, when compared to in situ measurements equatorward of 60°. SRB data are available on a 1° global grid every 3 h. All datasets mentioned are used to provide high-resolution data on the state of the atmosphere for years 2003–06.

Given the variety of gridded datasets employed, interpretation of the results is affected by the varied resolutions and associated errors inherent in each dataset. This is complicated further by the fact that the points of greatest interest are in areas of intense precipitation; since SST, TPW, and surface wind speed are difficult to accurately retrieve by satellite in raining conditions (e.g., Elsaesser and Kummerow 2008), the interpretation of the evolution of these parameters in the results is necessarily problematic because the collocated grid points contain rainfall by definition. For instance, LHF estimates from SeaFlux use passive microwave-derived surface winds and humidity, which need to be interpolated in raining conditions. Model data, like the reanalysis data used here, have their own biases, and indeed assimilate many of the aforementioned parameters that are difficult to retrieve in raining conditions; ERA-Interim uses a special implementation of their data assimilation scheme for rain-affected satellite observations, described in Bauer et al. (2006). Furthermore, ERA-Interim generally underestimates rainfall in the tropical ocean, with modeled precipitation more frequent and less intense than shown in observations (Dee et al. 2011); the intensity of the heaviest rainfall is underestimated by a factor of 3 or more (Mitovski et al. 2010), a finding consistent with analysis presented here.

In situ measurements of SST are taken from the Global Tropical Moored Buoy Array (GTMBA; McPhaden et al. 2010). The GTMBA is a network of moored buoys in the Pacific, Atlantic, and Indian Oceans that provides measurements of SSTs either hourly or every 10 min at an absolute accuracy of ±0.003°C (Freitag et al. 1999). SST measurements from these buoys are mostly given at a depth of 1 m, and should not be confused with ocean skin temperature, which exhibits greater diurnal variation (Kawai and Wada 2007). Oceanic mixed layer depth information comes from a dataset described in de Boyer Montégut et al. (2004), in which mixed layer depth (MLD) is defined as the depth at which ΔT = 0.2°C or Δσθ = 0.03 kg m−3 when compared to a near-surface value at 10-m depth, with data given on a 2° grid at monthly resolution. MLD data will be used to facilitate scale analysis of what drives the observed SST variability. The buoy SST data serve as a highly accurate check on a satellite-derived gridded dataset for which small variations can be of key importance for both weather and climate.

The input data for the objective classification of tracked systems come from the Tropical Rainfall Measuring Mission (TRMM) Microwave Imager (TMI; Kummerow et al. 1998), using the latest version of the Goddard profiling algorithm (GPROF) microwave retrieval (Kummerow et al. 2011), GPROF 2010 version 2 (C. Kummerow 2014, unpublished manuscript). The GPROF retrieval is used to provide rainwater path (RWP), surface rain rates, and convective fraction of rainfall at the pixel level. The TRMM satellite orbits Earth approximately 16 times daily, covering latitudes from 39°N to 39°S. While the method used can be implemented for any microwave imager, TMI is used here because of its ability to nearly perfectly collocate the clustering algorithm output with the TRMM precipitation radar (PR) to obtain vertical structure of the scene classifications; PR also flies on the TRMM satellite (Kummerow et al. 1998), and the PR 2A25 product is used to provide reflectivities at a 250-m vertical resolution.

3. Methods

a. Tracking algorithm

An algorithm was developed that tracks groups of contiguous CMORPH grid points that exceed a given rain-rate threshold. The region chosen for analysis is the tropical ocean between 15°N and 15°S. By contiguous it is meant that the grid points exceeding the threshold are adjacent along at least one lateral edge in the same time step or one time step removed. An example of a tracked system is found in Fig. 1. The algorithm groups the contiguous grid points and tracks them as one entity, storing the latitude and longitude of the system’s center at each time step so as to allow collocation with other datasets. The center of each system is defined as the mean longitude and latitude of all pixels exceeding the rain-rate threshold at each 3-h time interval (i.e., the center of the area marked by the dashed line in Fig. 1). The merging and breaking apart of cloud systems, as detailed in Williams and Houze (1987), is accounted for as long as the spatially separated elements are contiguous with raining pixels one time step away [for a visual example of this, see Fig. 6 in Skok et al. (2009)]. Additionally, a group must contain a minimum of 40 total (¼°) grid points across all time steps to be considered. This essentially removes very small areas with high rain rates and very short-lived systems from the analysis; the results are not particularly sensitive to changes in this additional constraint.

Fig. 1.
Fig. 1.

Example of a fast tracked system. Rainfall rate from CMORPH is contoured, with the threshold rate for tracking marked with a dashed red contour; the O signifies the position of a moored buoy from which a time series of SST was taken at time step t = 4.

Citation: Journal of Climate 27, 5; 10.1175/JCLI-D-13-00285.1

Choosing an appropriate rain-rate threshold for tracking is important for keeping distinct systems separate. A low threshold causes there to be too many grid points that overlap in space–time and thus systems that are artificially large and propagate very long and far, whereas a high threshold will cause some systems to be missed and less cohesive systems to be incorrectly labeled as separate entities. The threshold used, separating grid points that are eligible to form tracked groups from noneligible ones, is a rain rate of 7 mm h−1. Determination of the rain-rate threshold for tracking was made after performing a sensitivity analysis, examining thresholds between 5 and 10 mm h−1, with 7 mm h−1 found to maximize the number of eligible groups. Visible inspection was also important in determining the threshold. Because the tracking method is wholly based upon rain rates, there is no distinction made between mesoscale convective systems (MCSs), tropical cyclones, etc., with all being treated the same way and hence the nomenclature of “system” instead of MCS. Thus, if an object is contiguous in the three-dimensional longitude–latitude–time domain and meets the stated criteria, it enters the analysis.

The distribution of tracked system characteristics is found in Fig. 2, with the ocean basins shown in different colors. The ocean basins are defined by longitude (Table 2). Some key results from this analysis are that westward-propagating systems are more common and faster moving [a topic explored in Tulich and Kiladis (2012)], that systems seldom have lifetimes in excess of 24 h, and that the ocean basins possess strikingly similar distributions of behavior. The Atlantic basin is a slight outlier in that it contains the largest ratio of west-movers to east-movers, and has west-movers that often travel farther because of a tendency for systems in the Atlantic to move quickly westward.

Fig. 2.
Fig. 2.

Histograms of (top) duration, (middle) zonal distance traveled, and (bottom) zonal speed for tracked systems, separated into the ocean basins defined in Table 2.

Citation: Journal of Climate 27, 5; 10.1175/JCLI-D-13-00285.1

Table 2.

Number of tracked systems, with means and standard deviations of latitude (positive northward), propagation speed, and number of ¼° pixels per time step. Shown for fast and slow systems in each ocean basin, 2003–06.

Table 2.

A particular goal of this study is to determine a raining system’s impact on the environment as a function of its speed of propagation. This is motivated by the hypothesis that a longer residence time in a particular locale will translate into a larger environmental impact simply due to a longer period of integration. Therefore, the analysis is split into fast- and slow-moving systems. Fast systems are defined to have a mean propagation speed greater than 6 m s−1 while slow systems propagate at less than 2 m s−1; these limits were chosen for their simplicity and the similar number of tracked objects that fall within those bounds, representing approximately the fastest fifth and slowest fifth of all tracked systems.

b. Classification

As a check on the tracking method, a microwave-based clustering method is used to determine the similarity of tracked objects between basins and provide information regarding vertical structure and precipitation characteristics. The motivation behind the clustering method used in this study is Elsaesser et al. (2010), in which a K-means clustering method applied to precipitating scenes in the tropical ocean finds three distinct, consistent regimes. These three regimes echo the trimodal distribution of tropical cloud structures in Johnson et al. (1999).

Following this method, pixel-level data from TMI are divided into scenes that are roughly 100-km squares, approximating the area of a 1° × 1° grid box in the tropics [the same scene size used in Masunaga (2012)]. The TMI swath width allows for six adjacent scenes across the swath, yielding about 800 000 eligible scenes per month. A scene qualifies only if all pixels are 100% ocean and at least one pixel has a surface rain rate greater than 0.5 mm h−1, a threshold similar to the minimum rate detectable by PR.

Five variables are used as input for the clustering algorithm. Each of these variables represents the state of the entire scene, not just the precipitating pixels. The mean convective rain rate and fraction of total rainfall that is convective are two of the clustering parameters. These two variables provide information on the scene’s overall rain rate as well as the partitioning between stratiform and convective rainfall, an attribute that is highly related to distinct cloud states (Jakob and Schumacher 2008). The other input variables come from a normalized histogram of RWP. Split up into three bins, the histogram of RWP provides a sense of the distribution of cloud depths in the scene, since RWP scales with optical depth.

The clustering method yields a stable set of clusters whose mean properties vary little from month to month, with regimes remarkably consistent in structure and mean properties between ocean basins. An average month has a distribution of 82% shallow, 17% midlevel, and 1% deep convective scenes. The names given to the cluster types refer to the extent of vertical development characteristic of each type, ascertained from RWP distributions and PR reflectivity profiles. The distribution of reflectivity profiles for the midlevel and deep convective precipitation regimes, separated into convective and stratiform pixels, is given in Fig. 3. These scenes have mean rain rates of 0.18, 1.6, and 7.0 mm h−1, accounting for 35%, 49%, and 16% of total oceanic rainfall from 30°N to 30°S, respectively. The two nonshallow regimes, 18% of the total, account for nearly two-thirds of all rainfall, consistent with numerous other studies of precipitation classification that have found that MCSs provide a majority of total rainfall (e.g., Mohr et al. 1999; Lee et al. 2013). It is observed that a certain degree of decay and reorganization is commonplace for longer-lived systems, with tracked systems sometimes alternating between the deep convective and midlevel regimes, whereas shorter-lived systems are more likely to follow a simpler life cycle of shallow or congestus clouds building to deep convective clouds and then dissipating.

Fig. 3.
Fig. 3.

Contoured frequency by altitude diagrams (CFADs) of mean reflectivity profiles from TRMM PR, shown for the TMI-derived (a),(b) midlevel and (c),(d) deep convective precipitation regimes. These are separated into PR pixels marked as stratiform and convective.

Citation: Journal of Climate 27, 5; 10.1175/JCLI-D-13-00285.1

Comparison of the microwave-based precipitation regimes with the ISCCP tropical weather states (Rossow et al. 2005) shows that the TMI-derived deep convective regime matches with WS1 84% of the time, with the remainder being mostly the other convectively active weather states (WS2 and WS3). Consistency between the precipitation regimes defined here and the cloud regimes from Rossow et al. (2005) demonstrates an encouraging continuity between the two methods of scene classification despite the wholly different resolutions and methods employed.

A majority of the tracked systems exhibit the characteristics of a deep cloud structure, a large fractional area of stratiform pixels, and a large convective rain rate. Because the structure and characteristics of tracked objects are found to be consistent independent of location, the following analysis of environmental impacts of deep convective raining systems can reasonably be generalized to all heavy precipitation in the tropical ocean.

c. Capturing local time evolution

Depicting the time evolution of various fields before and after the passage of a tracked system is motivated by an assertion from Mapes et al. (2009): “a mean MCS life cycle can be robustly defined and depicted when hundreds of cases or more are averaged.” With over 21 000 total tracked systems, including over 4300 fast cases and 3600 slow cases over four years of matching ancillary data with tracked systems, the mean life cycle and time evolution of the fields examined should be well defined.

To capture the time evolution of a field, a time series is extracted from the grid point closest to the centroid of a tracked object at every time step. Thousands of time series are thus extracted and averaged together to provide the mean time evolution of a given ancillary field, centered on the passage of a tracked object. Unlike the methodologies of some similar studies (Mapes et al. 2009; Mitovski et al. 2010), it should be stressed that there is no compositing necessary because all the datasets have continuous data fields. However, this method does introduce some peculiarities. For instance, a given time series can be extracted multiple times if the object centroid remains in the same grid box for consecutive time steps, a common occurrence for datasets on coarser grids. This has the effect of slightly smearing the resultant time series and is largely a function of the dataset’s spatial resolution. Also, longer-lived objects will contribute more data points to the average; however, fast- and slow-moving systems have a similar average lifetime in each basin and thus one subset is not biased over the other. The use of so many data points should ameliorate all but systematic errors in each ancillary dataset, and indeed the mean time series exhibit features such as the diurnal nature of heavy precipitation in the tropics quite smoothly.

4. Environmental effects

Given the precipitating systems identified by the tracking algorithm, it is then possible to assess their local environmental impacts via collocation with various ancillary gridded datasets by the method described in section 3c. The calculation of these time series provides a picture of the mean environmental evolution of many different fields, centered on the passage of a deep convective system.

The distribution of propagation characteristics for tracked systems (Fig. 2) is borne out of many complex factors, but the fact that propagation speed is an important determinant of a system’s local impact is more intuitive. Slowly moving systems deposit a great amount of rain in one locale and the associated cloud field severely decreases the solar radiation that reaches the surface, two causes of a drop in SST, whereas both of these effects are shorter lived for faster systems. In the following discussion, analysis of the environmental evolution of various parameters is separated into locations affected by slow- and fast-moving systems.

Before looking at environmental effects, analysis of wind shear at the time of system passage shows a clear difference between fast and slow propagating systems (Fig. 4). Fast systems generally exhibit stronger vertical wind shear in the midtroposphere from 900 to 500 mb. This general pattern holds in all basins except the central Pacific, where shear profiles are nearly identical for fast and slow systems; the Atlantic basin holds the biggest difference between fast and slow system shear profiles, with fast systems exhibiting much stronger shear from 850 to 700 mb (not shown). While significant differences in the shear profiles do exist between basins, the mean shear profiles hold a key contrasting physical feature of fast and slow systems. This discrepancy in shear suggests a prescriptive method for examining the behavior of deep convective raining systems in models.

Fig. 4.
Fig. 4.

Profiles of mean vertical wind shear magnitude at system passage for fast and slow systems of all basins.

Citation: Journal of Climate 27, 5; 10.1175/JCLI-D-13-00285.1

The evolution of clouds before and after the passage of a tracked system generally follows a cycle of shallower clouds building to deeper clouds and then dissipating. More specifically, smaller-scale convection and scattered cumulus clouds grow to large-scale deep convective clouds before dissipating into smaller-scale deep convection and thick cirrus, then the distribution returns back to the mean. A large majority of tracked systems are collocated with the deep convective cloud regime (WS1) in the ISCCP dataset. Figure 5 shows how WS1 dominates nearest the system’s passage with a frequency of occurrence near 90%, while the periods before and after the passage are largely WS3 and WS5. This is similar for fast and slow systems, with the main difference being the persistence of WS1. The residence time of WS1 is much longer than that of the rainfall itself, corroborating a finding in Lee et al. (2013) that the presence of WS1 does not necessarily imply precipitation, as WS1 is nonraining in about half of all occurrences. This particular analysis is necessarily a type of composite resulting from ISCCP weather states being available only during daytime (Rossow et al. 2005).

Fig. 5.
Fig. 5.

ISCCP weather state relative frequencies of occurrence for (top) fast and (bottom) slow systems.

Citation: Journal of Climate 27, 5; 10.1175/JCLI-D-13-00285.1

The evolution of water vapor and water vapor convergence with respect to the passage of the convective system is derived from ERA-Interim. Water budget analysis from reanalyses can be problematic on short time scales. It is known that models and reanalyses do not resolve very intense precipitation well in the tropics (Mitovski et al. 2010; Stephens et al. 2010; Rossow et al. 2013). As such, closure of the water budget was not expected when using both observational and reanalysis datasets, especially on the scale of a single grid box with intense rainfall. The analysis shows that there is not enough water vapor fluxed into the area analyzed to support such heavy precipitation, and so the water vapor convergence and possibly TPW results are not consistent with the other elements of the water budget.

As seen in Fig. 6, fast and slow systems both exhibit an increase in local TPW and general convergence of columnar water vapor. Again some differences between basins are apparent, but the convergence of water vapor and increase in TPW are generally more abrupt for faster systems. Each ocean basin shows large-scale water vapor convergence in the mean evolution that steadily increases until the system’s passage, after which convergence falls sharply. This large-scale flux of water vapor into areas of deep convection seems to be a key source of moisture for the heavy precipitation, but is not of sufficient magnitude to close the water budget with the rainfall estimate from CMORPH. The mean evolution of TPW, for almost every basin and speed, exhibits a nearly symmetric increase of 4–6 kg m−2. The passage of a deep convective system provides neither a moistening nor a drying net effect on the total atmospheric column at long lag times; slight net moistening is more pronounced at shorter lag times. Parsing total column changes by atmospheric levels, there is a modest net effect of drying at lower levels and moistening at upper levels because of deep convection.

Fig. 6.
Fig. 6.

System-relative evolution of (top) TPW, (middle) water vapor convergence, and (bottom) rain-rate evolution by basin, shown for fast (dashed) and slow (solid) systems.

Citation: Journal of Climate 27, 5; 10.1175/JCLI-D-13-00285.1

Evaporation is another component of the water budget, a quantity directly linked to LHF at the surface. As seen in Fig. 7, changes in LHF due to tracked systems show a moderate dependence on ocean basin, partially because the basins have different mean winds and SSTs. Both fast and slow systems exhibit a peak in LHF 6–9 h after system passage, with this peak being more pronounced for fast systems. In the mean evolution, none of the basins shows variations in LHF of greater than 20%, showing changes to be small in magnitude at the time scales considered here. The peaks in LHF after system passage occur at a time of high wind stress with slightly lower near-surface humidity than at the time of system passage. Ultimately, LHF variability is small because of the largely offsetting effects of higher wind stress and lower near-surface humidity difference, and thus LHF is not a key element of variation for the water or energy budgets.

Fig. 7.
Fig. 7.

System-relative evolution of (top) LHF and (bottom) SST. The LHF from SeaFlux is shown for fast (dashed) and slow (solid) systems and separated by basin. The SST evolution is representative only of cases collocated with buoys, with all system speeds considered for each basin.

Citation: Journal of Climate 27, 5; 10.1175/JCLI-D-13-00285.1

5. Radiation and SST variability

The tracked systems cause a drop in SST for all system speeds in every ocean basin, a maximum decrease of 0.2°–0.3°C in the mean evolution (Fig. 7). Many factors may contribute to the interbasin variability of both the drop in SST and its recovery, such as variations in surface wind stress and other meteorological conditions, the residence time of deep cloud cover, ocean MLD, and other oceanic conditions. Much of this is beyond the scope of this study, in that finescale variations in the upper ocean can be of key importance: “the response of the upper ocean to atmospheric forcing is very sensitive to the vertical structure of both the temperature and salinity” (Richards et al. 1995). In addition, the rate of precipitation can change the local MLD and thus affect the rate of change of SST (Anderson et al. 1996). Analysis of 10-m winds from ERA-Interim showed that the mean time evolution of wind stress is quite similar between basins (not shown), but clearly many other factors are at play. For example, despite the central Pacific’s annual mean MLD of around 60 m, deeper than the tropical mean of 44 m, the mean change in SST is very similar to that of other basins with generally shallower MLDs (de Boyer Montégut et al. 2004).

Observation of small-scale changes in SST as a result of raining systems can be problematic given the inability of satellite-based sensors, infrared or passive microwave, to measure SST directly in raining conditions (Chelton and Wentz 2005). Gridded SST datasets must use interpolation to fill in precipitating areas, and thus the values for SST given by a product such as SeaFlux around the time of system passage are not retrieved values by definition. To circumvent this issue, the tracked systems are matched with buoys from the GTMBA (McPhaden et al. 2010) that are within one degree of latitude and longitude of the system center, providing accurate and high temporal resolution measurements of SST. The difference between these two SST measurements is shown in Fig. 8, while Fig. 7 gives the SST evolution from cases matched with buoy data and separated by basin. It should be noted that a large majority of the tracked systems matched with buoy measurements are from the equatorial Pacific because of its great density of moored buoys, with Figs. 7 and 8 (and also 10) being representative only of these matched cases.

Fig. 8.
Fig. 8.

System-relative evolution of SST from SeaFlux (solid) and buoy (dashed) data, with zero defined at a lag of −72 h.

Citation: Journal of Climate 27, 5; 10.1175/JCLI-D-13-00285.1

The buoy-based measurements show that SST starts dropping sharply around 15 h before system passage, drops by approximately 0.2°C in the mean in less than 24 h, and starts recovering fairly quickly thereafter. This is evidence for the cloud shield associated with the tracked deep convection affecting the surface radiation budget before precipitation arrives, consistent with analyses of radiative fluxes (Fig. 9); this is true regardless of system size or speed, as slow and fast systems both see similar drops in SST prior to system passage despite fast systems typically being larger (Table 2). Depressed SSTs persist through +72-h lag in the mean evolution, a result shown by both datasets that holds for all system speeds. The buoy measurements from the GTMBA indicate a more abrupt decrease and subsequent recovery of SST than the SeaFlux dataset because of rain contamination issues mentioned previously. The buoys show a fundamentally different evolution of SST on these short time scales when compared to the gridded product, and thus the buoy data alone are used in the following scale analysis.

Fig. 9.
Fig. 9.

(top) System-relative net radiative flux into the surface and (bottom) downward shortwave radiation at the surface, shown for fast (dashed) and slow (solid) systems and separated by basin.

Citation: Journal of Climate 27, 5; 10.1175/JCLI-D-13-00285.1

If, for the sake of argument, one considers that oceanic variability is negligible, the observed changes in SST can be thought of as being caused by a combination of radiative forcing at the surface, evaporation, and the formation of cooler freshwater pools from accumulated rainfall. The presence of widespread, optically thick cloud cover causes a drop in solar radiation reaching the ocean surface, manifested by an approximate 150 W m−2 decrease of the net radiative flux into the ocean surface from the mean value to that of system passage (Fig. 9). This decrease is fairly uniform for all basins and system speeds, with faster systems displaying a sharper decrease and subsequent recovery of net radiative flux around system passage. Variability in the net radiative flux is almost entirely caused by changes in downwelling shortwave radiation that reaches the surface, as shown by the top and bottom panels of Fig. 9 being nearly identical. The weak diurnal signature of heavy tropical precipitation (Kikuchi and Wang 2008), which occurs preferentially in the early morning hours, is clear (e.g., Fig. 9).

A simple scale analysis of the causes of SST variability, again assuming static oceanic conditions, shows that radiative fluxes are the main driver of the observed change in SSTs. Changes in LHF (Fig. 7) are a relatively insignificant factor in forcing SST as they are on the order of 10 W m−2 for fast systems and 5 W m−2 for slow systems; despite the uncertainty of the LHF estimates, this is still an order of magnitude less than the radiative variability. Sensible heat flux due to the precipitation itself is difficult to quantify, as it is dependent on the near-surface wet-bulb temperature and attributes like the drop size distribution (Katsaros and Buettner 1969). Over long time scales the sensible heat flux due to rainfall is typically small; it can, however, be a significant term in the energy budget on short time scales, with instances of fluxes greater than 200 W m−2 observed coincident with extreme rain rates (Gosnell et al. 1995). As a very rough approximation given observations from Gosnell et al. (1995), the sensible heat flux out of the ocean due to precipitation is estimated to linearly scale as 6 W m−2 for every 1 mm h−1 of rain rate. Sensible heat flux is thus a nonnegligible element of the energy budget in areas of heavy precipitation and, given the very rough estimate used here, averages about 20 W m−2 for the 24-h period around system passage, peaking at over 50 W m−2 at system passage in the mean evolution.

Figure 10 gives a straightforward picture of how the net surface radiative flux drives SST on short time scales, with radiation leading SST anomaly by about 6 h as they covary closely; the net radiative flux into the surface and the first time derivative of SST vary almost perfectly in tandem. To assess whether the changes in radiative fluxes are the key driver of SST variability, an equivalent MLD can be calculated by integrating the departure of the net surface radiative flux Fsfc from the time mean [Eq. (1)]. This calculation assumes an isothermal mixed layer of constant density ρw and specific heat Cw, and is simply used for comparison with the climatological MLD to provide a sense of scale:
e1
Fig. 10.
Fig. 10.

Surface radiative flux, buoy-based SST anomaly, and the first time derivative of SST coevolution for all tracked systems matched with buoys.

Citation: Journal of Climate 27, 5; 10.1175/JCLI-D-13-00285.1

Integrating changes in the net surface radiative flux from −18 h to +6 h lag, the 24-h period of greatest SST decrease (ΔT), yields a calculated MLD of 13.2 m for radiation alone and 15.0 m if including a term for sensible heat as well. This MLD of 15 m is significantly shallower than the annual mean MLD in the tropics of approximately 44 m (de Boyer Montégut et al. 2004). Some of this could be explained by the presence of a “barrier layer” of water stratified by salinity that can sometimes exist below the near surface, effectively insulating the near-surface layer from greater depths (Godfrey and Lindstrom 1989). In a study of the western Pacific warm pool, Anderson et al. (1996) define near-surface layer depths based upon density HD, potential temperature HT, and salinity differences HS; mean values of these depths were found to be 14.8, 20.3, and 29.3 m, respectively, with the barrier layer defined as HTHD. The top near-surface layer (HD) of 14.8 m is nearly equal to the roughly estimated MLD of 15.0 m from the analysis presented here, and indeed HD is intended to demarcate the layer of strongest mixing (Anderson et al. 1996). Despite the danger in generalizing from the western Pacific to the whole tropics, the calculated MLD of 15.0 m seems reasonable if the top layer has strong mixing and a barrier layer is temporarily inhibiting deeper mixing in the ocean column.

If the time period of integration is extended to allow for greater mixing to be more likely, then the scale analysis begins to match the climatological MLD of approximately 44 m (de Boyer Montégut et al. 2004). For instance, integrating the change in surface radiative flux from −18 to +36 h in Fig. 10 yields a calculated MLD of 45.8 m when including the sensible heat term, a value close to the mean climatological MLD for all basins. Radiation appears to be the main driver of the observed SST decrease in these cases; importantly, this result also shows that the initial recovery of SST is not primarily driven by radiation, as cooler water from the surface mixes with warmer water from deeper in the mixed layer. This is clear given the initial recovery of SST seen in Fig. 10 at around +12 h lag while the net radiative flux is still anomalously low, with SST tendency changing drastically while radiation is still depressed. Characteristics of the near-surface ocean, including the depth of the mixed layer, are important for both the magnitude of the total drop in SST and its subsequent recovery.

6. Discussion

A method to identify, track, and characterize deep convective raining systems in the tropical ocean according to precipitation regime and cloud state has been presented. Tracked systems were stratified by propagation speed, and their relationships with radiation, SST, and moisture variables were assessed. It was found that there is a great degree of similarity among deep convective systems of all ocean basins in the tropics. Despite small regional differences, it is remarkable how consistent the spread of tracked systems’ behavior remains from basin to basin (Fig. 2).

Few studies have examined deep convective rainfall’s propagation characteristics from precipitation data in such a broad geographic sense, with the only real comparisons being Skok et al.’s (2009) study, which looked only at the Pacific Ocean, and Rossow et al.’s (2013) study, which used brightness temperatures and collocated rainfall data to track systems and judge their intensity. A cursory analysis of the propagation characteristics found in the Pacific in Skok et al. (2009) appears to be in line with the findings seen in Fig. 2. In spite of using the same latitudinal domain, comparisons with Dias et al. (2012) are more difficult, as phase speed and propagation speed are not directly comparable; however, direct comparison of system duration is possible (see their Fig. 5b). The ratio of west- to east-moving systems is consistent between the two studies as is the finding that westward propagating systems have faster speeds on average. Their “continuous cloud regions” have a peak of 18–24-h duration, whereas the criteria used here signals a peak at 12–15-h duration and a mean duration of just over 15 h (Fig. 2). This is likely a result of the finer grid used here, identifying more small-scale systems, as well as the focus on heavy precipitation instead of cold cloud tops only—deep clouds with cold cloud tops are more persistent than heavy precipitation (Lee et al. 2013).

One study that has examined environmental evolution from an Eulerian, composited perspective is that of Masunaga (2012). Masunaga (2012) found that the passage of a “moist convective” system in the tropics leaves a negative anomaly of convective available potential energy in its wake, while atmospheric water vapor goes from a positive anomaly of almost 4 kg m−2 at the system’s passage back down to a TPW value at or slightly below an equilibrium value (his Fig. 6c). A composite method is used and the tropics are defined differently in Masunaga (2012), but findings in this study agree with the general shape of positive anomaly in TPW and roughly agree with the magnitude of TPW increase resulting from a system’s passage, shown in Fig. 6.

The presence of optically thick cloud fields associated with deep convective systems is the primary cause of the observed decrease in mean SST of 0.2°–0.3°C. Latent heat flux and sensible heat flux due to rainfall each contribute to the observed SST variability, but a scale analysis performed here suggests that the interplay of net radiative flux at the surface and depth of the mixed layer are the key determinants of SST decrease and recovery. Because of interpolation due to rain contamination, gridded products do not accurately represent the finescale SST variability caused by a passing system, as shown by in situ measurements from moored buoys. This may have implications for how air–sea interactions are modeled on short time scales, as small variations in SST and SST gradients can have sizable impacts on convection (Li and Carbone 2012).

As a quick examination of the suspected feedback on subsequent convective activity caused by the observed drop in SSTs, rain-rate data from CMORPH were binned to create probability density functions (PDFs) of rain rates after the passage of tracked systems. Figure 11 displays this analysis by way of a PDF percent difference, where the slow PDF is subtracted by the fast PDF and the difference is divided by the slow PDF, showing which rain-rate bins are more likely to be reached by environments affected by fast- or slow-moving systems at a given lag time. Many factors other than SST affect such a simple analysis and a fair amount of noisiness exists at higher rain-rate bins due to the smaller number of cases. Heavy rain rates are more likely in regions affected by fast-moving systems at a lag of 4–6 days, a result that was expected given that SSTs following slow systems remain nearly 0.1°C lower for many days, as seen in the overlaid mean evolutions of SeaFlux SST in Fig. 11. This feedback seems to have a greater effect at very high rain rates and on deep convection, implicitly, since the PDFs are nearly equal at all lower rain rates once a couple days removed from the tracked system’s passage.

Fig. 11.
Fig. 11.

Percent difference between fast and slow rain-rate PDFs [(PDFslow − PDFfast)/PDFslow] following the passage of tracked systems in all basins. Warm (cold) colors denote contours of 20% increments where environments affected by slow (fast) systems are more likely to reach the given rain-rate bin, with values between −20% and +20% left uncontoured. Overplotted is the SST anomaly from SeaFlux for all (green), fast (purple), and slow (blue) systems, with zero defined at a lag of −72 h.

Citation: Journal of Climate 27, 5; 10.1175/JCLI-D-13-00285.1

Despite the fact that a deep convective system typically moves through a region in under 24 h, SSTs can remain depressed for an extended (i.e., multiday) period following the passage of a tracked system. This implies a thermodynamic modification to the boundary layer that can have an impact on the subsequent evolution of rainfall and the redevelopment of convection in the near vicinity.

Acknowledgments

The authors acknowledge the helpful comments from the anonymous reviewers and Dave Thompson at CSU. The work was supported through NASA Grants NNX08AT04A and NNX10AG75G. Thanks as well to Dave Randall and Wes Berg from the GPROF team.

REFERENCES

  • Anderberg, M. R., 1973: Cluster Analysis for Applications.Academic Press, 359 pp.

  • Anderson, S. P., , R. A. Weller, , and R. B. Lukas, 1996: Surface buoyancy forcing and the mixed layer of the western Pacific warm pool: Observations and 1D model results. J. Climate, 9, 30563085.

    • Search Google Scholar
    • Export Citation
  • Bauer, P., , P. Lopez, , D. Salmond, , A. Benedetti, , S. Saarinen, , and M. Bonazzola, 2006: Implementation of 1D+4D-Var assimilation of precipitation-affected microwave radiances at ECMWF. II: 4D-Var. Quart. J. Roy. Meteor. Soc., 132, 23072332.

    • Search Google Scholar
    • Export Citation
  • Chelton, D. B., , and F. J. Wentz, 2005: Global microwave satellite observations of sea surface temperature for numerical weather prediction and climate research. Bull. Amer. Meteor. Soc., 86, 10971115.

    • Search Google Scholar
    • Export Citation
  • de Boyer Montégut, C., , G. Madec, , A. S. Fischer, , A. Lazar, , and D. Iudicone, 2004: Mixed layer depth over the global ocean: An examination of profile data and a profile-based climatology. J. Geophys. Res., 109, C12003, doi:10.1029/2004JC002378.

    • Search Google Scholar
    • Export Citation
  • Dee, D. P., and Coauthors, 2011: The ERA-Interim reanalysis: Configuration and performance of the data assimilation system. Quart. J. Roy. Meteor. Soc., 137, 553597, doi:10.1002/qj.828.

    • Search Google Scholar
    • Export Citation
  • Dias, J., , S. Tulich, , and G. Kiladis, 2012: An object-based approach to assessing tropical convection organization. J. Atmos. Sci., 69, 24882504.

    • Search Google Scholar
    • Export Citation
  • Elsaesser, G. S., , and C. D. Kummerow, 2008: Toward a fully parametric retrieval of the nonraining parameters over the global oceans. J. Appl. Meteor. Climatol., 47, 15991618.

    • Search Google Scholar
    • Export Citation
  • 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
  • Freitag, H. P., , M. E. McCarty, , C. Nosse, , R. Lukas, , M. J. McPhaden, , and M. F. Cronin, 1999: COARE Seacat data: Calibrations and quality control procedures. NOAA Tech. Memo. ERL PMEL-115, 89 pp. [Available online at http://www.pmel.noaa.gov/pubs/PDF/frei2034/frei2034.pdf.]

  • Godfrey, J. S., , and E. J. Lindstrom, 1989: The heat budget of the equatorial western Pacific surface mixed layer. J. Geophys. Res., 94 (C6), 80078017.

    • Search Google Scholar
    • Export Citation
  • Gosnell, R., , C. Fairall, , and P. J. Webster, 1995: The sensible heat of rainfall in the tropical ocean. J. Geophys. Res., 100 (C9), 18 43718 442.

    • Search Google Scholar
    • Export Citation
  • Jakob, C., , and G. Tselioudis, 2003: Objective identification of cloud regimes in the tropical western Pacific. Geophys. Res. Lett., 30, 2082, doi:10.1029/2003GL018367.

    • Search Google Scholar
    • Export Citation
  • Jakob, C., , and C. Schumacher, 2008: Precipitation and latent heating characteristic of the major tropical western Pacific cloud regimes. J. Climate, 21, 43484364.

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

    • Search Google Scholar
    • Export Citation
  • 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
  • Katsaros, K. B., , and K. J. K. Buettner, 1969: Influence of rainfall on temperature and salinity of the ocean surface. J. Appl. Meteor., 8, 1518.

    • Search Google Scholar
    • Export Citation
  • Kawai, Y., , and A. Wada, 2007: Diurnal sea surface temperature variation and its impact on the atmosphere and ocean: A review. J. Oceanogr., 63, 721744.

    • Search Google Scholar
    • Export Citation
  • Kikuchi, K., , and B. Wang, 2008: Diurnal precipitation regimes in the global tropics. J. Climate, 21, 26802696.

  • 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
  • Kummerow, C. D., , S. Ringerud, , J. Crook, , D. Randel, , and W. Berg, 2011: An observationally generated a priori database for microwave rainfall retrievals. J. Atmos. Oceanic Technol., 28, 113130.

    • Search Google Scholar
    • Export Citation
  • Lee, D., , L. Oreopoulos, , G. J. Huffman, , W. B. Rossow, , and I.-S. Kang, 2013: The precipitation characteristics of ISCCP tropical weather states. J. Climate, 26, 772788.

    • Search Google Scholar
    • Export Citation
  • Li, Y., and R. E. Carbone, 2012: Excitation of rainfall over the tropical western Pacific. J. Atmos. Sci., 69, 29832994.

  • Machado, L. A. T., , W. B. Rossow, , R. L. Guedes, , and A. W. Walker, 1998: Life cycle variations of mesoscale convective systems over the Americas. Mon. Wea. Rev., 126, 16301654.

    • Search Google Scholar
    • Export Citation
  • Mapes, B., , S. Tulich, , J. Lin, , and P. Zuidema, 2006: The mesoscale convection life cycle: Building block or prototype for large-scale tropical waves? Dyn. Atmos. Oceans, 42, 329.

    • Search Google Scholar
    • Export Citation
  • Mapes, B., , R. Milliff, , and J. Morzel, 2009: Composite life cycle of maritime tropical mesoscale convective systems in scatterometer and microwave satellite observations. J. Atmos. Sci., 66, 199208.

    • 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
  • Masunaga, H., , and C. D. Kummerow, 2006: Observations of tropical precipitating clouds ranging from shallow to deep convective systems. Geophys. Res. Lett., 33, L16805, doi:10.1029/2006GL026547.

    • Search Google Scholar
    • Export Citation
  • McPhaden, M. J., and Coauthors, 2010: The global tropical moored buoy array. Proceedings of OceanObs '09: Sustained Ocean Observations and Information for Society, J. Hall, D. E. Harrison, and D. Stammer, Eds., Vol. 2, ESA Publ. WPP-306. [Available online at http://www.pmel.noaa.gov/tao/proj_over/pubs/McPhaden_Oceanobs09.pdf.]

  • Mitovski, T., , I. Folkins, , K. von Salzen, , and M. Sigmond, 2010: Temperature, relative humidity, and divergence response to high rainfall events in the tropics: Observations and models. J. Climate, 23, 36133625.

    • Search Google Scholar
    • Export Citation
  • Mohr, K. I., , J. S. Famiglietti, , and E. J. Zipser, 1999: The contribution to tropical rainfall with respect to convective system type, size, and intensity estimated from the ice scattering signature. J. Appl. Meteor., 38, 596606.

    • Search Google Scholar
    • Export Citation
  • Reynolds, R. W., , T. M. Smith, , C. Liu, , D. B. Chelton, , K. S. Casey, , and M. G. Schlax, 2007: Daily high-resolution-blended analyses for sea surface temperature. J. Climate, 20, 54735496.

    • Search Google Scholar
    • Export Citation
  • Richards, K. J., , M. E. Inall, , and N. C. Wells, 1995: The diurnal mixed layer and the upper ocean heat budget in the western equatorial Pacific. J. Geophys. Res., 100 (C4), 68656879.

    • Search Google Scholar
    • Export Citation
  • Rossow, W. B., , G. Tselioudis, , A. Polak, , and C. Jakob, 2005: Tropical climate described as a distribution of weather states indicated by distinct mesoscale cloud property mixtures. Geophys. Res. Lett., 32, L21812, doi:10.1029/2005GL024584.

    • Search Google Scholar
    • Export Citation
  • Rossow, W. B., , A. Mekonnen, , C. Pearl, , and W. Goncalves, 2013: Tropical precipitation extremes. J. Climate, 26, 14571466.

  • Sapiano, M. R. P., , and P. A. Arkin, 2009: An intercomparison and validation of high-resolution satellite precipitation estimates with 3-hourly gauge data. J. Hydrometeor., 10, 149166.

    • Search Google Scholar
    • Export Citation
  • Skok, G., , J. Tribbia, , J. Rakovec, , and B. Brown, 2009: Object-based analysis of satellite-derived precipitation systems over the low- and midlatitude Pacific Ocean. Mon. Wea. Rev., 137, 31963218.

    • Search Google Scholar
    • Export Citation
  • Stackhouse, P. W., Jr., , S. K. Gupta, , S. J. Cox, , J. C. Mikovitz, , T. Zhang, , and L. M. Hinkelman, 2011: 24.5-year SRB dataset released. GEWEX News, No. 21 (1), International GEWEX Project Office, Silver Spring, MD, 10–12. [Available online at http://www.gewex.org/images/Feb2011.pdf.]

  • Stephens, G. L., and Coauthors, 2010: Dreary state of precipitation in global models. J. Geophys. Res., 115, D24211, doi:10.1029/2010JD014532.

    • Search Google Scholar
    • Export Citation
  • Tan, J., , C. Jakob, , and T. P. Lane, 2013: On the identification of the large-scale properties of tropical convection using cloud regimes. J. Climate, 26, 66186632.

    • Search Google Scholar
    • Export Citation
  • Tulich, S., , and G. Kiladis, 2012: Squall lines and convectively coupled gravity waves in the tropics: Why do most cloud systems propagate westward? J. Atmos. Sci., 69, 29953012.

    • Search Google Scholar
    • Export Citation
  • Wheeler, M., , and G. N. Kiladis, 1999: Convectively coupled equatorial waves: Analysis of clouds and temperature in the wavenumber–frequency domain. J. Atmos. Sci., 56, 374399.

    • Search Google Scholar
    • Export Citation
  • Williams, M., , and R. A. Houze Jr., 1987: Satellite-observed characteristics of winter monsoon cloud clusters. Mon. Wea. Rev., 115, 505519.

    • Search Google Scholar
    • Export Citation
  • Yuan, J., , and R. A. Houze Jr., 2010: Global variability of mesoscale convective system anvil structure from A-Train satellite data. J. Climate, 23, 58645888.

    • Search Google Scholar
    • Export Citation
Save