• Allan, R. P., B. J. Soden, V. O. John, W. Ingram, and P. Good, 2010: Current changes in tropical precipitation. Environ. Res. Lett., 5, 025205, doi:10.1088/1748-9326/5/2/025205.

    • Search Google Scholar
    • Export Citation
  • Ashfaq, M., L. C. Bowling, K. Cherkauer, J. S. Pal, and N. S. Diffenbaugh, 2010: Influence of climate model biases and daily scale temperature and precipitation events on hydrological impacts assessment: A case study of the United States. J. Geophys. Res., 115, D14116, doi:10.1029/2009JD012965.

    • Search Google Scholar
    • Export Citation
  • Bourlioux, A., and A. Majda, 2002: Elementary models with probability distribution function intermittency for passive scalars with a mean gradient. Phys. Fluids, 14, 881897.

    • Search Google Scholar
    • Export Citation
  • Draxler, R. R., and G. D. Hess, 1998: An overview of the Hysplit4 modeling system for trajectories, dispersion, and deposition. Aust. Meteor. Mag., 47, 295308.

    • Search Google Scholar
    • Export Citation
  • Duffy, P. B., B. Govindasamy, J. Milovich, K. Taylor, M. Wehner, A. Lamont, and S. Thompson, 2003: High resolution simulations of global climate, Part 1: Present climate. Climate Dyn., 21, 290371.

    • Search Google Scholar
    • Export Citation
  • Gershgorin, B., and A. Majda, 2011: Filtering a statistically exactly solvable test model for turbulent tracers from partial observations. J. Comput. Phys., 230, 16021638.

    • Search Google Scholar
    • Export Citation
  • Held, I. M., and B. J. Soden, 2000: Water vapor feedback and global warming. Annu. Rev. Energy Environ., 25, 441475.

  • Hilburn, K. A., and F. J. Wentz, 2008: Intercalibrated passive microwave rain products from the Unified Microwave Ocean Retrieval Algorithm (UMORA). J. Appl. Meteor. Climatol., 47, 778794.

    • Search Google Scholar
    • Export Citation
  • Holloway, C. E., and J. D. Neelin, 2009: Moisture vertical structure, column water vapor, and tropical deep convection. J. Atmos. Sci., 66, 16651683.

    • Search Google Scholar
    • Export Citation
  • Holloway, C. E., and J. D. Neelin, 2010: Temporal relations of column water vapor and tropical precipitation. J. Atmos. Sci., 67, 10911104.

    • Search Google Scholar
    • Export Citation
  • Hu, Y., and R. T. Pierrehumbert, 2001: The advection–diffusion problem for stratospheric flow. Part I: Concentration probability function. J. Atmos. Sci., 58, 14931510.

    • Search Google Scholar
    • Export Citation
  • Hu, Y., and R. T. Pierrehumbert, 2002: The advection–diffusion problem for stratospheric flow. Part II: Concentration probability function of tracer gradients. J. Atmos. Sci., 59, 28302845.

    • Search Google Scholar
    • Export Citation
  • Kalnay, E., and Coauthors, 1996: The NCEP/NCAR 40-Year Reanalysis Project. Bull. Amer. Meteor. Soc., 77, 437471.

  • Kollias, P., and B. Albrecht, 2010: Vertical velocity statistics in fair weather cumuli at the ARM TWP Nauru Climate Research Facility. J. Climate, 23, 65906604.

    • Search Google Scholar
    • Export Citation
  • Lintner, B. R., and J. D. Neelin, 2008: Eastern margin variability of the South Pacific convergence zone. Geophys. Res. Lett., 35, L16701, doi:10.1029/2008GL034298.

    • Search Google Scholar
    • Export Citation
  • Lintner, B. R., and J. D. Neelin, 2010: South America–Atlantic sector convective margins and their relationship to low-level inflow. J. Climate, 23, 26712685.

    • Search Google Scholar
    • Export Citation
  • Majda, A. J., and B. Gershgorin, 2010: Quantifying uncertainty in climate change science through empirical information theory. Proc. Natl. Acad. Sci. USA, 107, 14 95814 963.

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

    • Search Google Scholar
    • Export Citation
  • Mather, J. H., T. P. Ackerman, W. E. Clements, F. J. Barnes, M. D. Ivey, L. D. Hatfield, and R. M. Reynolds, 1998: An atmospheric radiation and cloud station in the tropical western Pacific. Bull. Amer. Meteor. Soc., 79, 627642.

    • Search Google Scholar
    • Export Citation
  • Muller C. J., L. E. Back, P. A. O’Gorman, and K. A. Emanuel, 2009: A model for the relationship between tropical precipitation and column water vapor. Geophys. Res. Lett., 36, L16804, doi:10.1029/2009GL039667.

    • Search Google Scholar
    • Export Citation
  • Neelin, J. D., B. R. Lintner, B. Tian, Q. Li, L. Zhang, P. K. Patra, M. T. Chahine, and S. N. Stechmann, 2010: Long tails in deep columns of natural and anthropogenic tropospheric tracers. Geophys. Res. Lett., 37, L05804, doi:10.1029/2009GL041726.

    • Search Google Scholar
    • Export Citation
  • Peters, O., and J. D. Neelin, 2006: Critical phenomena in atmospheric precipitation. Nat. Phys., 2, 393396.

  • Porch, W. M., S. C. Olsen, P. Chylek, M. K. Dubey, B. G. Henderson, and W. Clodius, 2006: Satellite and surface observations of Nauru Island clouds: Differences between El Niño and La Niña periods. Geophys. Res. Lett., 33, L13804, doi:10.1029/2006GL026339.

    • Search Google Scholar
    • Export Citation
  • Sherwood, S. C., 2010: Direct versus indirect effects of tropospheric humidity changes on the hydrologic cycle. Environ. Res. Lett., 5, 025206, doi:10.1088/1748-9326/5/2/025206.

    • Search Google Scholar
    • Export Citation
  • Solomon, S., D. Qin, M. Manning, M. Marquis, K. Averyt, M. M. B. Tignor, H. L. Miller Jr., and Z. Chen , Eds., 2007: Climate Change 2007: The Physical Science Basis. Cambridge University Press, 996 pp.

    • Search Google Scholar
    • Export Citation
  • Stechmann, S. N., and J. D. Neelin, 2011: A stochastic model for the transition to strong convection. J. Atmos. Sci., in press.

  • Stokes, G. M., and S. E. Schwartz, 1994: The Atmospheric Radiation Measurement (ARM) program: Programmatic design of the cloud and radiation test bed. Bull. Amer. Meteor. Soc., 75, 12021221.

    • Search Google Scholar
    • Export Citation
  • Walker, M. D., and N. S. Diffenbaugh, 2009: Evaluation of high-resolution simulations of daily scale temperature and precipitation over the United States. Climate Dyn., 33, 11311147.

    • Search Google Scholar
    • Export Citation
  • Wehner, M., R. L. Smith, G. Bala, and P. Duffy, 2010: The effect of horizontal resolution on simulation of very extreme US precipitation events in a global atmosphere model. Climate Dyn., 34, 241247.

    • Search Google Scholar
    • Export Citation
  • Zhang, C., B. Mapes, and B. J. Soden, 2003: Bimodality in tropical water vapor. Quart. J. Roy. Meteor. Soc., 129, 28472866.

  • View in gallery

    Time series of 5-min-averaged ARM passive radiometer cwv at Nauru.

  • View in gallery

    The pdfs for selected spectral bands of the FFT Nauru radiometer cwv anomalies. Here, anomalies are defined by first removing a climatology computed over the entire period from the raw data and then forming 30-day running means of the deseasonalized data and subtracting these from the deseasonalized data. Results shown correspond to all submothly variability (gray) as well as periods <1 day (subdaily; black), periods between 1 and 5 days (fast synoptic; red), and periods between 5 and 30 days (slow synoptic; blue). Also included are Gaussian best fits to the cores of the pdfs, estimated from fit through the maximum and half-maximum points of each.

  • View in gallery

    Anomalous cwv pdfs and their relationship to precipitation. (left) Total Nauru radiometer cwv anomaly pdf (black) and pdfs sorted into precipitating (blue) or nonprecipitating (red) categories using the 5-min Nauru optical gauge scanner measurements of precipitation. Gaussian best fits to the cores are also shown. (right) As in the left panel, but sorting using running 6-h averages of precipitation.

  • View in gallery

    (left) The pdf of Nauru’s radiosonde interpolated pressure level specific humidity (q, g kg−1, along the x axis) anomalies, defined with respect to 5-day running means (after subtracting climatology). The line contours are Gaussian core best fits for each pressure level. (right) Nauru radiosonde cwv anomalies bin averaged for the pressure level specific humidity anomalies in the left panel. The dashed lines represent the positive and negative 2σ levels at each pressure level.

  • View in gallery

    Total Nauru radiosonde specific humidity profiles for extreme cwv departures. The blue–red line denotes the mean q vertical profile when cwv anomalies (defined with respect to a 5-day running mean) exceed ±2σ. The black line is the mean over all available radiosonde profiles. Light blue–red shading denotes the q envelope for all q anomalies smaller than 2σ (defined with respect to a 5-day running mean).

  • View in gallery

    Composites of (left 2 columns) TMI satellite cwv and (right 2 columns) precipitation rate anomalies on positive excursions of daily mean radiometer cwv anomalies (from a 5-day running mean) at Nauru (centered in maps). (top to bottom) Time lags are from −2 to +2 days. Note that if events overlap, only the larger one is used. Dashed contour is 52-mm total cwv for each composite, with higher values to the left of the contour.

  • View in gallery

    As in Fig. 6, but for negative excursions of daily mean radiometer cwv anomalies.

  • View in gallery

    Nauru HYSPLIT once-daily Lagrangian back trajectories composited on (a) all positive–negative cwv anomalies and (b) anomalies exceeding ±2σ cwv (blue–red) for arrival altitudes of 50 m (circles), 2000 m (squares), and 4000 m (triangles). Also shown are means over all available back trajectories (black). The symbol locations correspond to 24-h intervals along the trajectories. (c),(d) As in (a),(b), but for departures of back trajectories from 5-day running means.

  • View in gallery

    Comparison of NCEP–NCAR reanalysis 4 times daily (circles) and radiometer (squares) cwv anomalies at Nauru. Radiometer cwv is first aggregated to 6-h averages, and then anomalies are formed with respect to 5-day running means. The pdfs shown are for all conditions (black) as well as sorted into vertical averages of reanalysis anomalous vertical velocity (blue, anomalous ascent; red, anomalous descent). Gaussian core best fits are included.

  • View in gallery

    Summary schematic for vertical processes hypothesized to contribute to the formation of long-tailed pdfs.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 213 115 28
PDF Downloads 165 84 21

Column Water Vapor Statistics and Their Relationship to Deep Convection, Vertical and Horizontal Circulation, and Moisture Structure at Nauru

View More View Less
  • 1 Department of Environmental Sciences, Rutgers, The State University of New Jersey, New Brunswick, New Jersey
  • | 2 Department of Meteorology, University of Reading, Reading, United Kingdom
  • | 3 Department of Atmospheric and Oceanic Sciences and Institute of Geophysics and Planetary Physics, University of California, Los Angeles, Los Angeles, California
© Get Permissions
Full access

Abstract

Relationships among relatively high-frequency probability distribution functions (pdfs) of anomalous column water vapor (cwv), precipitating deep convection, and the vertical and horizontal structures of circulation and tropospheric moisture are investigated for the Atmospheric Radiation Measurement (ARM) climate observing facility at Nauru in the western equatorial Pacific. At the highest frequencies (subdaily) analyzed, the cwv pdf exhibits a Gaussian core with pronounced longer-than-Gaussian, approximately exponential tails, with the relatively lower-frequency submonthly pdfs becoming more Gaussian distributed across the entire range of cwv variability. The genesis and morphology of the longer-than-Gaussian tails are examined within the context of several hypothetical mechanisms outlined in prior work. For example, pdf conditioning on ARM optical gauge precipitation measurements reveals an association of the positive-side tail with precipitating deep convective conditions; thus, despite the condensation and fallout of cwv during rainfall events, it is argued that updraft vertical motions associated with deep convection locally compensate the loss by increasing cwv. Using vertical moisture profiles from ARM radiosonde measurements, vertical structures of specific humidity anomalies associated with tail-regime cwv excursions are computed, with the negative cwv profile significantly departing from the mean profile in the lower free troposphere. Such behavior is consistent with local restorative surface evaporative forcing and turbulent mixing in the atmospheric boundary layer and drying of the column from above during descent conditions. Analysis of cwv variability with respect to the horizontal moisture structure, using gridded measurements from the Tropical Rainfall Measuring Mission (TRMM) Microwave Imager (TMI) and trajectories from the Hybrid Single-Particle Lagrangian Integrated Trajectory (HYSPLIT) model driven by NCEP–NCAR reanalysis meteorology underscores how the horizontal and vertical components modulate Nauru cwv: in particular, high cwv conditions at Nauru are often associated with weakened low-level inflow from the dry regions to the east of Nauru and stronger along-trajectory ascent. Finally, comparison of the ARM-based pdfs to those estimated from the reanalysis illustrates how pdf-based diagnostics may be useful tools for model intercomparison and validation.

Corresponding author address: Benjamin R. Lintner, Dept. of Environmental Sciences, Rutgers, The State University of New Jersey, 14 College Farm Rd., New Brunswick, NJ 08901-8551. E-mail: lintner@envsci.rutgers.edu

Abstract

Relationships among relatively high-frequency probability distribution functions (pdfs) of anomalous column water vapor (cwv), precipitating deep convection, and the vertical and horizontal structures of circulation and tropospheric moisture are investigated for the Atmospheric Radiation Measurement (ARM) climate observing facility at Nauru in the western equatorial Pacific. At the highest frequencies (subdaily) analyzed, the cwv pdf exhibits a Gaussian core with pronounced longer-than-Gaussian, approximately exponential tails, with the relatively lower-frequency submonthly pdfs becoming more Gaussian distributed across the entire range of cwv variability. The genesis and morphology of the longer-than-Gaussian tails are examined within the context of several hypothetical mechanisms outlined in prior work. For example, pdf conditioning on ARM optical gauge precipitation measurements reveals an association of the positive-side tail with precipitating deep convective conditions; thus, despite the condensation and fallout of cwv during rainfall events, it is argued that updraft vertical motions associated with deep convection locally compensate the loss by increasing cwv. Using vertical moisture profiles from ARM radiosonde measurements, vertical structures of specific humidity anomalies associated with tail-regime cwv excursions are computed, with the negative cwv profile significantly departing from the mean profile in the lower free troposphere. Such behavior is consistent with local restorative surface evaporative forcing and turbulent mixing in the atmospheric boundary layer and drying of the column from above during descent conditions. Analysis of cwv variability with respect to the horizontal moisture structure, using gridded measurements from the Tropical Rainfall Measuring Mission (TRMM) Microwave Imager (TMI) and trajectories from the Hybrid Single-Particle Lagrangian Integrated Trajectory (HYSPLIT) model driven by NCEP–NCAR reanalysis meteorology underscores how the horizontal and vertical components modulate Nauru cwv: in particular, high cwv conditions at Nauru are often associated with weakened low-level inflow from the dry regions to the east of Nauru and stronger along-trajectory ascent. Finally, comparison of the ARM-based pdfs to those estimated from the reanalysis illustrates how pdf-based diagnostics may be useful tools for model intercomparison and validation.

Corresponding author address: Benjamin R. Lintner, Dept. of Environmental Sciences, Rutgers, The State University of New Jersey, 14 College Farm Rd., New Brunswick, NJ 08901-8551. E-mail: lintner@envsci.rutgers.edu

1. Introduction

The spatial and temporal distributions of tropospheric moisture have important implications for the global climate system and its variability. As Earth’s principal greenhouse gas, water vapor strongly affects the cloud optical properties and their distribution and atmospheric column and surface radiative budgets (Held and Soden 2000). Phase changes of tropospheric water provide important sources of heating (condensation of the vapor phase) and cooling (evaporation or sublimation of condensed phases) (Sherwood 2010). In turn, horizontal and vertical transport processes and mixing, evaporation, and condensation impart a nontrivial space–time structure to tropospheric water vapor (Zhang et al. 2003). Interactions between water vapor and atmospheric dynamic and thermodynamic processes lead to nonlinear feedbacks operating across many space and time scales.

Given the climatic importance of tropospheric water vapor, it is not surprising that the predictive skill of weather and climate system models depends critically on it. On the one hand, model development and refinement require detailed mechanistic knowledge of the observed sources of tropospheric water variability across a wide range of spatial and temporal scales, so that these sources can be realistically incorporated into models. On the other hand, it is also necessary to validate the performance of numerical models against observations through the application of suitable diagnostic measures. For water vapor, the assessment of model performance is especially challenging because of the many orders of magnitude, in both space and time, over which water vapor variability requires validation.

The quantification of the underlying statistical distributions of climate variables has important implications for the interpretation and application of such data. For example, the shape of the probability distribution function (pdf) of a climate variable affects how the frequency of extremes is expected to change under conditions for which the entire pdf is shifted, a common framework for interpreting global warming impacts on extremes (Solomon et al. 2007; Walker and Diffenbaugh 2009). If a process is statistically Gaussian, the pdf is characterized by two moments (mean and variance); thus, a change in extreme occurrence with respect to a fixed threshold, induced by a shift in the pdf mean, is readily quantified in terms of an error function. In contrast, for non-Gaussian pdfs, quantification of an extreme value change will depend on the detailed structure of the tails. For water vapor, relationships to deep convection and precipitation processes impart even greater significance when elucidating pdf structure, as extreme precipitation events are often associated with extreme tropospheric moisture excursions, with the extremes strongly impacting the overall strength and even the sign of the hydrologic cycle variability (Ashfaq et al. 2010; Allan et al. 2010). Moreover, since approaches for data assimilation and filtering, as well as inverse modeling, are sensitive to assumptions about the underlying data statistics, different methodologies may be required when pdfs exhibit long tails (Gershgorin and Majda 2011).

Here, we examine pdfs of column water vapor (cwv) anomalies for Nauru (0.5°S, 166.9°E) and its immediate vicinity in the western equatorial Pacific. This site-specific focus is motivated by two factors. First, there exists a considerable amount of validated high-frequency moisture data for Nauru from the Department of Energy’s Atmospheric Radiation Measurement (ARM; Stokes and Schwartz 1994) climate observing station (Mather et al. 1998) maintained there. Prior analyses by two of us (Holloway and Neelin 2009, 2010, hereafter HN09 and HN10, respectively) documented the observed vertical and temporal variability characteristics of water vapor at Nauru and thus provide a context for interpreting the results presented here. Second, Nauru is situated in a complex convective environment in the western Pacific: in particular, it lies to the east of the western Pacific warm pool, to the south of the Pacific intertropical convergence zone (ITCZ), and to the north of the South Pacific convergence zone (SPCZ). Nauru’s location thus renders it ideal for studying how the principal large-scale Pacific convection zones interact with each other as well as with adjacent nonconvecting areas. It is anticipated that the moisture variability at Nauru should be qualitatively representative of processes operating elsewhere over warm tropical ocean regions.

Building on prior work addressing pdf structures in tropospheric water vapor and other “passive” tracers (Neelin et al. 2010, hereafter N10), our objective is to relate features of Nauru cwv pdfs to precipitating deep convection and vertical and horizontal characteristics of tropospheric circulation and moisture in the vicinity of Nauru. In doing so, we will review several hypotheses to explain the characteristics of the pdfs, such as the shapes of the distributions and possible asymmetries, and discuss the plausibility of these hypotheses to the Nauru pdfs.

2. Anomalous Nauru radiometer cwv pdfs

a. Overview of the radiometer cwv time series and its variability

Figure 1 illustrates the time series of ARM passive radiometer-derived cwv at Nauru for the period 20 November 1998–16 August 2006. The native temporal resolution of the radiometer instrument is 1 min, although the data shown in Fig. 1 have been aggregated to 5-min means to reduce noise. The cwv time series is characterized by a time mean of 50 mm over the whole record but with substantial variability between 20 and 70 mm. The power spectral density of the total radiometer cwv estimated from a fast Fourier transform (FFT) of the raw cwv (not shown) manifests a red noise structure.

Fig. 1.
Fig. 1.

Time series of 5-min-averaged ARM passive radiometer cwv at Nauru.

Citation: Journal of Climate 24, 20; 10.1175/JCLI-D-10-05015.1

Two points are worth noting. First, the systematically lower cwv values during the period 1999–2001 appear to be related to large-scale conditions in the equatorial Pacific, as persistent La Niña conditions were present from summer 1998 through late winter of 2001; during the remainder of the cwv record, normal or El Niño–like conditions prevailed. During La Niña events, relatively cool sea surface temperatures in the central–eastern equatorial Pacific are associated with infrequent deep convective activity and large-scale descent. These conditions have been related to lower cwv values at Nauru (Porch et al. 2006). For our purposes here, we do not explicitly consider the very low frequency (intraseasonal and longer) behavior dictated by the large-scale background Pacific mean state, although that does not mean that such variability does not impact the high-frequency statistics. For example, in section 3, we consider how anomalous cwv relates to precipitating deep convection; the latter is more likely during the warmer surface conditions after mid-2001.

Second, the total cwv data manifest only a weak seasonal cycle. Examination of the 7-yr climatology (1999–2005) hints at a weak climatological maximum in NH spring and minimum in late NH summer, with a peak-to-trough amplitude of roughly 10 mm (not shown). One intriguing feature of the data is the asymmetry of maximum and minimum excursions with respect to the interval climatology over the period: during the course of the year, the maxima are more tightly clustered than the minima; that is, the range above climatology over which the maxima vary is smaller (by roughly 50%) than the range below climatology over which the minima vary. Such asymmetric behavior is consistent with prior work on cwv relationships to precipitating deep convection in which a critical (temperature dependent) cwv threshold is apparent; HN09 estimated the value of the threshold at Nauru to be ~66 mm. Although cwv values above critical do in fact occur, pdfs of total cwv tend to be peaked just below critical. On the low side, the absolute limit on cwv is 0, although physical constraints maintain cwv at a nonzero lower bound: for a well-mixed atmospheric boundary layer (ABL) underlying a dry free troposphere (FT), this limit is ~qbΔpb/g, where Δpb is the pressure depth of the ABL. We return to this point below.

b. Spectral band pdfs

Given the red noise nature of the Nauru cwv variability, we analyze the variability within spectral bands by reconstructing bandpass-filtered time series through application of an FFT. Because we wish to emphasize here the pdf behavior at submonthly time scales, the 7-yr climatology (1999–2006) as well as a running 30-day box-car mean are presubtracted from the data. Also, because the radiometer does not function properly when moist (the “wet window problem”; see HN10), cwv values during periods of intense rainfall are missing, resulting in some gaps in the time series. These gaps have been filled here with a nearest-neighbors linear interpolation scheme. While some finescale pdf features are sensitive to whether data in the gaps are retained or neglected, the results discussed below are largely robust to this procedure.

Figure 2 illustrates pdfs for selected high-frequency spectral intervals. For comparison, Gaussian “core” fits to the half-maximum points for each spectral band are included. The pdf of all submonthly scale cwv anomalies (gray) is observed to be effectively Gaussian over much of the anomaly range. The shape of this pdf is very sensitive to the low-frequency variability present in the data (i.e., the width of the running mean), as this truncates the power spectrum on the low-frequency side. For the principal time scales of interest here, subdaily (<1 day; black) to a few days (1–5 days; red), the pdfs are seen to deviate substantially from the Gaussian core fit with the emergence of longer-than-Gaussian tails. These tails, which are more pronounced in the subdaily data, are effectively linear in the y coordinate over at least two decades, suggesting approximately exponential behavior.

Fig. 2.
Fig. 2.

The pdfs for selected spectral bands of the FFT Nauru radiometer cwv anomalies. Here, anomalies are defined by first removing a climatology computed over the entire period from the raw data and then forming 30-day running means of the deseasonalized data and subtracting these from the deseasonalized data. Results shown correspond to all submothly variability (gray) as well as periods <1 day (subdaily; black), periods between 1 and 5 days (fast synoptic; red), and periods between 5 and 30 days (slow synoptic; blue). Also included are Gaussian best fits to the cores of the pdfs, estimated from fit through the maximum and half-maximum points of each.

Citation: Journal of Climate 24, 20; 10.1175/JCLI-D-10-05015.1

What accounts for the presence of the long tails in these cwv pdfs? A class of models based on advection–diffusion of a passive tracer (for a constituent with concentration χ) has been shown to produce long-tailed pdfs (Bourlioux and Majda 2002; Majda and Gershgorin 2010) even under conditions for which the underlying statistics of the flow field (v) are Gaussian because of the nonlinear advection term of the form v · χ. Insights from these models have been widely applied to passive stratospheric tracers to infer properties of stratospheric flow and its variability that would be otherwise difficult to capture with available observations (Hu and Pierrehumbert 2001, 2002). As in N10, we note the plausibility of the simple prototype as a potential source of long-tailed behavior of Nauru’s cwv pdfs—to the extent that tropospheric water vapor behaves as a passive tracer—although we speculate that other processes are involved [see, for example, Stechmann and Neelin (2011) for a stochastic model with explicit sinks].

We also point out the apparent asymmetries between the positive and negative tails evident in Fig. 2. Although sample size limits confidence at the extremes, for the subdaily data, the slope of the positive side tail is less steep than the negative side tail. In contrast, for the fast synoptic time scale of a few days, the negative side tail is fatter than the positive tail. It is reasonable to ask, what might account for the presence of asymmetry as well as its apparent change between subdaily and fast synoptic time scales? N10 in fact noted the asymmetry in the anomalous cwv pdf for the entire tropics, with the negative tail falling off more rapidly than the positive tail, although in this analysis, anomalies were formed with respect to 30-day running means.

N10 hypothesized several potential sources for cwv pdf asymmetry. First, the presence of precipitating deep convection may act to limit cwv through condensation and rain out: this process may be expected to truncate the positive side, and is in fact consistent with the behavior of maximum and minimum excursions described in section 2a. On the other hand, deep convecting regions are also characterized by ascending motions that tend to be strong and narrow compared to weaker compensatory descent over a broader area. Because of water vapor’s strongly maintained vertical gradient, the asymmetry of strong upward versus weak downward motion might be expected to produce asymmetries in the positive and negative tail patterns of behavior; that is, the cwv pdf may simply inherit asymmetries of the vertical motion field. In the absence of deep convection, or in the presence of convective downdrafts, descent through the relatively dry FT above the ABL may act to dry out the FT portion of the column, limiting the negative-side tail. Finally, lateral advective transport associated with horizontal gradients can modify or replace local air masses, and asymmetries may arise from spatially nonuniform gradients.

In the sections below, we interpret the Nauru cwv pdf tails and asymmetries in light of the aforementioned hypotheses. We begin by evaluating the potential role of convective processes (section 3), followed by the effects of vertical and horizontal gradients (sections 4 and 5). Section 6 presents a comparison of the Nauru site results to cwv estimated from the National Centers for Environmental Prediction–National Center for Atmospheric Research (NCEP–NCAR) reanalysis as a way of highlighting the utility of pdfs as potential diagnostics of model performance.

3. Pdf relationships to deep convection and precipitation

The Nauru ARM radiometer optical gauge instrument provides measurements of precipitation rates concurrent with cwv. Here, we use the precipitation data to estimate conditional pdfs under precipitating and nonprecipitating conditions. It is important to note that the gap filling has its strongest impact on the results presented here, as the cwv rejected because of the wet-window problem are expected to be on the high side of the distribution. However, as the gaps are typically short (under 12 h) and the temporal persistence of water vapor values for strong convecting events is of order a few hours to a day (see HN10), we argue that it is reasonable to retain the linearly interpolated values as an approximation to the true behavior.

Figure 3 depicts the total (black), precipitating (blue), and nonprecipitating (red) pdfs for two cases: the left panel uses the precipitation in each 5-min time interval while the right panel uses precipitation smoothed over 6 h. The cwv anomalies here are defined in terms of the instantaneous values with climatology and a 5-day running mean removed. For the 5-min precipitation values, there are relatively few occurrences in the precipitating pdf, since the likelihood of precipitation during any 5-min interval is small. Despite the small population of points, long-tailed behavior is nevertheless evident, particularly on the positive side of the distribution. Within the context of the hypotheses outlined in section 2b, this pattern of behavior suggests that while precipitation may be a sink for cwv, the associated dynamical processes are locally sufficient to overcome the loss that would otherwise lower the positive side tail. Only at the most extreme positive values is there some indication of a truncated tail.

Fig. 3.
Fig. 3.

Anomalous cwv pdfs and their relationship to precipitation. (left) Total Nauru radiometer cwv anomaly pdf (black) and pdfs sorted into precipitating (blue) or nonprecipitating (red) categories using the 5-min Nauru optical gauge scanner measurements of precipitation. Gaussian best fits to the cores are also shown. (right) As in the left panel, but sorting using running 6-h averages of precipitation.

Citation: Journal of Climate 24, 20; 10.1175/JCLI-D-10-05015.1

Use of a 6-h moving-average of precipitation (Fig. 3, right) indicates that the likelihood of precipitation during a given window is now much closer to the likelihood of no precipitation, as each of the precipitating and nonprecipitating pdfs account for roughly half of the total. Over much of the total pdfs linear range, precipitating and nonprecipitating conditions are associated with the positive and negative tails, respectively. The positive tail of the nonprecipitating pdf is observed to cut off more sharply than the negative tail of the precipitating pdf. Of course, the precipitating pdf contains instances of very weak precipitation; restricting the pdf to successively larger minimum rainfall values tends to reduce the negative side tail (not shown).

We suggest that strong updraft vertical motion in precipitating deep convection and the associated low-level moisture convergence are responsible for the relationship between precipitating conditions and the positive pdf tail; unfortunately, direct observations of vertical velocity needed to verify this connection are not available at Nauru. Kollias and Albrecht (2010) have recently estimated in-cloud vertical motion and mass fluxes for fair-weather cumuli conditions at Nauru, though these are not generally applicable to deep convecting conditions. In section 6, we consider the analysis in terms of the vertical velocity fields from the first NCEP–NCAR reanalysis dataset (Kalnay et al. 1996) to condition the pdfs, although the results are found to be inconclusive, for reasons to be discussed. For now, we turn to relating the vertical structure of specific humidity anomalies to total cwv as a check on consistency.

4. Vertical structure

The presence of a strongly maintained vertical gradient in tropospheric moisture suggests that fluctuations in vertical motion may produce longer-than-Gaussian tails in the sense suggested by simple advective-diffusion prototypes (Bourlioux and Majda 2002; Gershgorin and Majda 2011). While the radiometer does not provide vertically resolved information, an ARM radiosonde measurement program was carried out at Nauru. The available data (here, we consider the period of 1 April 2001–16 August 2006) typically consist of two soundings per day, at 0000 and 1200 UTC, but with sporadic soundings also occurring at 0230 and 1430 UTC. Specific humidity (q; g kg−1) is interpolated to 5-mb increments from 1000 to 125 mb; cwv (in units of mm) is also estimated as the pressure integral over the available vertical levels. HN09 and HN10 compared the patterns of behavior of radiometer and radiosonde cwv, sampling the radiometer data at times coincident with the soundings, and noted generally strong temporal correlations.

Spectral analysis of the sonde-derived cwv (not shown) yields qualitatively similar behavior to the radiometer cwv; in particular, longer-than-Gaussian tails emerge at relatively high-frequency (daily) time scales, although the small sample size (~1100 measurements) is considerably smaller than for the radiometer. Still, we suggest that analysis of the vertical profiles should provide some understanding of the processes that give rise to long tails and their asymmetry that would be obtained in the limit of greater sampling available with either a lengthier or finer resolved dataset.

The pdfs of sonde pressure-level q anomalies (defined with respect to a 5-day running mean) are contoured in Fig. 4 (left). The shape of the distribution indicates the largest spread in anomalies in the LFT, consistent with the standard deviation analysis of HN09. Here, we draw attention to the asymmetry of the pressure-level q pdfs with height: in the LFT, the pdfs are skewed negative, while those in the ABL are skewed slightly positive. The right panel of Fig. 4 depicts sonde cwv anomalies bin averaged onto specific humidity at each pressure level. These results demonstrate that the largest excursions in sonde-derived cwv occur for the q excursions in the LFT. Further, that there is a vertical offset between where the largest positive and negative cwv excursions occur (i.e., the largest positive cwv anomalies are reflected in q anomalies occurring above the q anomalies associated with the largest negative cwv anomalies) is consistent with a local vertical redistribution of water vapor. To state this in another way, if horizontal redistribution were primarily responsible for generating the largest cwv excursions, we might expect the peak positive and negative cwv values to map onto q anomalies at the same pressure level.

Fig. 4.
Fig. 4.

(left) The pdf of Nauru’s radiosonde interpolated pressure level specific humidity (q, g kg−1, along the x axis) anomalies, defined with respect to 5-day running means (after subtracting climatology). The line contours are Gaussian core best fits for each pressure level. (right) Nauru radiosonde cwv anomalies bin averaged for the pressure level specific humidity anomalies in the left panel. The dashed lines represent the positive and negative 2σ levels at each pressure level.

Citation: Journal of Climate 24, 20; 10.1175/JCLI-D-10-05015.1

Figure 5 illustrates averages of the daily sonde profiles indexed on anomalous daily sonde cwv, using those sonde profiles for which the magnitude of the cwv anomalies exceed 2σ, that is, profiles in the tails of the distribution. We note that positive (negative) excursions of daily sonde cwv are associated with moistening (drying) throughout much of the depth of the troposphere, although vertical amplitude differences are again evident, with the largest deviations from the mean profile (black line) in the LFT. In general, the −2σ profile in the lower FT deviates more from the mean profile than does the +2σ profile. Figure 5 also depicts values of specific humidity anomalies at each pressure level (shading). Comparing these profiles to the cwv composites points to the importance of coherence in the vertical for generating large cwv excursions; as the anomalies that are largest at a specific pressure level do not necessarily translate to the largest cwv anomalies. Interestingly, the +2σ cwv specific humidity profile largely matches the +2σ specific humidity anomaly profile over 900–700 mb. One interpretation is that positive excursions of cwv are limited by the size of positive specific humidity excursions in the LFT because saturation is reached. The behavior on the negative side suggests that negative cwv excursions are not so limited, although the circulation conditions for larger negative excursions would need to be sufficiently persistent and coherent.

Fig. 5.
Fig. 5.

Total Nauru radiosonde specific humidity profiles for extreme cwv departures. The blue–red line denotes the mean q vertical profile when cwv anomalies (defined with respect to a 5-day running mean) exceed ±2σ. The black line is the mean over all available radiosonde profiles. Light blue–red shading denotes the q envelope for all q anomalies smaller than 2σ (defined with respect to a 5-day running mean).

Citation: Journal of Climate 24, 20; 10.1175/JCLI-D-10-05015.1

5. Horizontal structure

Thus far, our analysis has emphasized processes primarily impacting the vertical distribution of moisture, such as deep convection and advection by vertical flow. However, as noted above, Nauru is located in a complex convective environment: for example, during the late NH spring, Nauru is situated in the vertex between the Pacific ITCZ and SPCZ with easterly mean trade wind flow toward the island from the relatively dry area in between these large-scale convecting features. It is thus reasonable to consider what role horizontal moisture gradients and the circulations interacting with them may play for Nauru cwv variability. To address contributions from horizontal moisture structure, we make use of the gridded cwv estimates in the vicinity of Nauru from the Tropical Rainfall Measuring Mission (TRMM) Microwave Imager (TMI) at 0.25° × 0.25° resolution from the Hilburn and Wentz (2008) algorithm. We also analyze back-trajectory output from the Hybrid Single-Particle Lagrangian Integrated Trajectory (HYSPLIT; Draxler and Hess 1998) model.

a. Lead–lag spatial composite analysis

To get an idea of the horizontal and temporal variabilities associated with the cwv variability at Nauru, we first consider composite maps of anomalous cwv and precipitation from TMI at lags of −2 to +2 days centered on positive radiometer cwv anomalies at Nauru (Fig. 6). At zero lag, there is an area of >1 mm cwv surrounding Nauru spanning approximately 5° latitude and 10° longitude, which is comparable to the scale of mesoscale convective system composites in Mapes et al. (2009). This anomaly structure appears to propagate from east to west fairly coherently at approximately 6 km s−1 across the 5-day lead–lag window, suggesting anomalous cwv conditions associated with either easterly waves or mean wind advection. In fact, the intermittency of easterly waves may contribute to the genesis of long-tailed pdfs at Nauru (see Gershgorin and Majda 2011), although further analysis is required to support whether the signature in the horizontal moisture field is indeed reflecting such waves.

Fig. 6.
Fig. 6.

Composites of (left 2 columns) TMI satellite cwv and (right 2 columns) precipitation rate anomalies on positive excursions of daily mean radiometer cwv anomalies (from a 5-day running mean) at Nauru (centered in maps). (top to bottom) Time lags are from −2 to +2 days. Note that if events overlap, only the larger one is used. Dashed contour is 52-mm total cwv for each composite, with higher values to the left of the contour.

Citation: Journal of Climate 24, 20; 10.1175/JCLI-D-10-05015.1

The precipitation anomalies in Fig. 6, while less coherent, tend to follow the cwv anomalies, although the positive cwv anomalies in regions below ~52 mm of total cwv (e.g., to the east of the dashed contour along the equator in Fig. 6). are less likely to correspond to positive precipitation anomalies; this follows from studies indicating a strong relationship between absolute cwv and precipitation, with much larger amounts of precipitation above a certain cwv threshold (Peters and Neelin 2006; Muller et al. 2009), as well as studies of tropical convective margins (Lintner and Neelin 2008, 2010).

Negative anomalies (Fig. 7) exhibit similar behavior to the positive anomalies. Note that the removal of the 5-day running mean emphasizes the differences between the center of the window and the edges, likely resulting in the opposite signs of the anomalies over Nauru between lag 0 and lag ±2 for both Figs. 6 and 7. However, composites on significant positive events without the running mean removed (not shown) exhibited very similar westward-propagating coherent features and did not show significant opposite-signed anomalies at the edge of the time window. We further note that we performed similar analyses emphasizing the pdf tails by including in the composites only those data corresponding to cwv excursions with magnitudes exceeding 2σ. While qualitatively similar results were obtained (not shown), these involved considerably smaller sample sizes, resulting in much lower signal-to-noise ratios.

Fig. 7.
Fig. 7.

As in Fig. 6, but for negative excursions of daily mean radiometer cwv anomalies.

Citation: Journal of Climate 24, 20; 10.1175/JCLI-D-10-05015.1

b. HYSPLIT back-trajectory analysis

As noted above, during some seasons, Nauru is located in a region of horizontal moisture gradients between the dry eastern equatorial Pacific and the moist western Pacific warm pool (WPWP) as well as between the ITCZ or SPCZ and equator. For simple flow configurations, it may be possible to interpret moisture variability associated with fluctuations of (low level) horizontal winds in the presence of mean moisture gradients by considering endpoint moisture values on the dry and moist sides of the moisture gradient to be set by large-scale controls. On the one hand, for dry, nondeep convecting conditions, moisture balance occurs primarily between evaporation into the ABL and large-scale subsidence above the ABL. On the other hand, for moist, strongly deep convecting conditions, the convective threshold (and saturation) may strongly constrain moisture values.

A potential complication for characterizing the spatial footprint of the variability, particularly at high frequencies, is the variable nature of the flow, that is, how much the daily to synoptic time-scale flow varies from expectations based on, say, monthly mean winds. We thus present a simple analysis in terms of HYSPLIT using NCEP–NCAR reanalysis meteorological input fields. For our purposes, we examine once-daily trajectories obtained over the 120 h (5 days) prior to their arrival at Nauru. Because of our interest in vertical structure, the trajectories are integrated from three low-altitude points above Nauru (50, 2000, and 4000 m).

Five-day back trajectories in the longitude–height plane, composited on daily mean radiometer cwv anomalies with respect to a 5-day running mean, are shown in Fig. 8a (Fig. 8b) for all positive (negative) cwv anomalies and extremes exceeding ±2σ. For reference, the mean trajectory over all days is also shown at each of the three vertical target levels. In a mean sense, the trajectories reflect origination to the east of Nauru, with parcels at the two lowermost levels experiencing slight descent until roughly 170°E, or 24 h upstream of Nauru, followed by ascent over the remainder of the trajectory. The parcels arriving at 4000 m experience ascending conditions along the entire trajectory. Projection of these data onto a 2D plane belies variations in the meridional direction such as upper-level outflow from either the ITCZ or SPCZ into the dry, descent zone along the equator.

Fig. 8.
Fig. 8.

Nauru HYSPLIT once-daily Lagrangian back trajectories composited on (a) all positive–negative cwv anomalies and (b) anomalies exceeding ±2σ cwv (blue–red) for arrival altitudes of 50 m (circles), 2000 m (squares), and 4000 m (triangles). Also shown are means over all available back trajectories (black). The symbol locations correspond to 24-h intervals along the trajectories. (c),(d) As in (a),(b), but for departures of back trajectories from 5-day running means.

Citation: Journal of Climate 24, 20; 10.1175/JCLI-D-10-05015.1

Figure 8a demonstrates the interplay of vertical and horizontal characteristics associated with Nauru cwv anomalies. For anomalously positive cwv values at Nauru, the trajectories at the 2000- and 4000-m levels are clearly lower in altitude (blue) than for negative cwv conditions. (The trajectories at 50 m, within the ABL, show vertical separation in the same sense that is not apparent on the scale of this plot.) Because of the strong vertical moisture gradient, the lower-originating trajectories might be expected to reflect enhanced moisture content at the beginning of the trajectory, although this depends on many factors over the history of the trajectory. At the same time, the shape of the trajectory implies stronger ascent upstream of Nauru, which implies stronger moisture convergence along the trajectory.

Comparing the composites’ longitudinal structures shows that the dry-phase trajectories extend farther to the east than do the moist-phase ones by 1°–2° longitude. In general, the more easterly originating trajectories may be expected to carry less moisture, as the air mass to the east of the equator is relatively dry. Again, however, the full 3D nature of the trajectories is critical, as air originating from within the ITCZ–SPCZ to the north–south of the equator would be expected to be moister than air that flowed in along the trades from the dry descent region in the southeast tropical Pacific. For extreme Nauru cwv excursions (Fig. 8b), the vertical and horizontal characteristics of the positive and negative phase trajectories are not cleanly separated; as in the lead–lag analysis, the small number of trajectories involved and the noise inherent to the trajectories have a substantial impact. Since we are interested in the relatively high-frequency variability of Nauru cwv, we repeat the compositing but using departures of the trajectories from their running 5-day means; results of this compositing are shown in Figs. 8c and 8d. It is now clear that for high-frequency moist excursions at Nauru, the associated trajectories are relatively lower and farther to the west than for high-frequency dry excursions.

6. Comparison to NCEP–NCAR reanalysis

N10 demonstrated that the pdf of reanalysis-estimated cwv for daily excursions from 30-day running means computed over the entire 20°S–20°N domain exhibits longer-than-Gaussian tails, with the positive and negative tails strongly associated with ascending and descending conditions, respectively (cf. their Fig. 1). It is not obvious a priori that the single-site, higher-frequency comparison of the reanalysis to Nauru observations should hold.

To carry out our comparison, we make use of the 6-h mean, 4 times daily reanalysis output. The radiometer cwv anomalies are aggregated to 6-h averages corresponding to the output intervals in the reanalysis, and for both the reanalysis and radiometer, anomalies are defined relative to a running 5-day mean (i.e., twenty 6-h intervals). Results of this comparison are highlighted in Fig. 9. In contrast to the observations (squares), the reanalysis pdf (circles) is Gaussian distributed over the whole range of cwv anomalies; the range of reanalysis anomalies is, in turn, considerably smaller than for the radiometer anomalies at these time scales. We have also conditioned the cwv pdfs onto the reanalysis vertical velocity at 500 mb. While the reanalysis pdf exhibits the expected pairing of tails and vertical velocity, the radiometer pdf at these time scales is not sensitive to the reanalysis vertical velocity field.

Fig. 9.
Fig. 9.

Comparison of NCEP–NCAR reanalysis 4 times daily (circles) and radiometer (squares) cwv anomalies at Nauru. Radiometer cwv is first aggregated to 6-h averages, and then anomalies are formed with respect to 5-day running means. The pdfs shown are for all conditions (black) as well as sorted into vertical averages of reanalysis anomalous vertical velocity (blue, anomalous ascent; red, anomalous descent). Gaussian core best fits are included.

Citation: Journal of Climate 24, 20; 10.1175/JCLI-D-10-05015.1

Our interpretation of these results is that the reanalysis fails to capture the behavior at the time (and space) scales inherent to the radiometer cwv values. This is perhaps not too surprising, especially for the vertical velocity pairing, which requires the time behavior in the reanalysis to match the observations. However, the lack of Gaussian tails in the reanalysis cwv indicates that the bulk statistics are also different. On the other hand, N10 documented the emergence of long tails in the NCEP–NCAR reanalysis on monthly time scales for pdfs summing over the entire tropics. This suggests that the NCEP–NCAR reanalysis captures mechanisms capable of generating long-tail pdfs but at larger scales than at the single observation site considered here. Together, our single-site, high-frequency analysis and the broader spatial scale, remote sensing based approach in N10 illustrate how consideration of pdfs may be a useful diagnostics of multiscale cwv evaluation in observations and comparisons of models (e.g., they indicate a model performing reasonably at large scales but failing at smaller scales).

7. Summary and discussion

In this paper, we have constructed pdfs of observed anomalous tropospheric column water vapor conditions at Nauru in the western equatorial Pacific and discussed how these relate to deep convection and the properties of atmospheric flow. By considering bandpass-filtered time series of ARM passive radiometer derived cwv, we demonstrated how the tails of the pdfs at time scales below a few days deviate from Gaussian.

The genesis of these longer-than-Gaussian tails was considered in light of several mechanistic hypotheses. We first conditioned the pdfs on observed precipitation occurrence to demonstrate the correspondence of the positive tail with precipitating conditions; that is, even though condensation of cwv during precipitation is a sink for column moisture, updrafts associated with deep convection locally can nonetheless drive up cwv values. Using ARM radiosonde measurements, we further illustrated how column-integrated moisture relates to the vertical structure of specific humidity anomalies and their pdfs. In particular, we documented large anomalous specific humidity departures in the LFT, with negative pdf skewness, compared to the ABL, with smaller excursions and a (slightly) positive skewness. Such behavior is consistent with vertical motion across a gradient maintained by strong restorative forcing in the ABL.

Apart from vertical processes associated with deep convection and advection across a maintained gradient, horizontal variations in moisture and flow in the vicinity of Nauru also have some impact. Lead–lag compositing of the spatially extensive TMI satellite measurements onto Nauru’s cwv identified coherent, upstream moisture footprints in the region to the east of Nauru over 1–2 days, consistent with quasi-steady mean flow advection of anomalous moisture conditions toward Nauru. Additionally, we analyzed back trajectories simulated by HYSPLIT forced by the NCEP–NCAR reanalysis to illustrate how the interplay of horizontal and vertical flow conditions affects Nauru’s cwv: specifically, we demonstrated how composited high cwv conditions at Nauru are characterized by back trajectories corresponding to both relatively stronger ascent and reduced easterly flow.

It is worth briefly mentioning how the results discussed here fit within the convective margins framework developed in prior work (see Lintner and Neelin 2008, 2010). In the underlying theory of convective margins, it is the vertical motion field (acting through moisture convergence) doing much of the work of enhancing column moisture along an inflow trajectory. For an assumed initial moisture value below what is needed for convection, the ratio of lateral (moisture weighted) inflow wind speed to the convergent component of the wind field determines a characteristic along-trajectory length scale for the moisture to increase. The presence of varying low-level winds (or varying initial moisture loading) has the net effect of shifting the convecting transition point.

The picture that emerges for high-frequency cwv variability and the generation of long-tailed pdfs at Nauru is summarized schematically in Fig. 10. Here, we have chosen to illustrate the local vertical processes; however, in the sense described in the prior paragraph, it is understood that nonlocal lateral transport also impacts cwv values at Nauru largely by influencing where the vertical processes ultimately occur. Moreover, the results in section 5 make it apparent that the vertical and horizontal circulation conditions are entwined.

Fig. 10.
Fig. 10.

Summary schematic for vertical processes hypothesized to contribute to the formation of long-tailed pdfs.

Citation: Journal of Climate 24, 20; 10.1175/JCLI-D-10-05015.1

An important question that arises from our analysis concerns the spatial scale at which the longer-than-Gaussian tails emerge in the pdfs. We note our comparison of the observed pdf to the reanalysis, for which the latter exhibited a smaller overall range of cwv variability and no discernible deviation from Gaussian over this range. Our interpretation of the discrepancy is that mesoscale circulations impart the non-Gaussian behavior to the very high-frequency pdfs, and the reanalysis does not adequately capture these circulations. Within the context of model simulation, it is well known that model resolution strongly affects a model’s capacity for simulating precipitation extremes (Duffy et al. 2003; Wehner et al. 2010). Given the physical connections between extreme moisture and precipitation excursions, cwv bulk statistics may be a powerful diagnostic, particularly in the sense that extreme precipitation episodes are more intermittent in both space and time. For this reason, we advocate the application of pdf-based approaches to high-resolution model output, as in the forthcoming fifth phase of the Coupled Model Intercomparison Project (CMIP5).

Acknowledgments

We thank P. Kollias, M. Miller, V. Ghate, and M. Niznik for useful comments; K. Hales-Garcia for TMI data processing; and J. Meyerson for graphical assistance. ARM data were used with the cooperation of the U.S. Department of Energy as part of the Atmospheric Radiation Measurement Program Climate Research Facility, and NCEP–NCAR reanalysis data were downloaded from the NOAA/OAR/ESRL Physical Sciences Division Web site (http://www.esrl.noaa.gov/psd/). This work was funded by National Science Foundation Grants ATM-0645200 and AGS-1102838 and National Oceanic and Atmospheric Administration Grants NA08OAR4310882 and NA11OAR4310099, and New Jersey Agricultural Experiment Station Hatch Grant NJ07102. CEH was supported by Natural Environment Research Council Grant NE/E00525X/1.

REFERENCES

  • Allan, R. P., B. J. Soden, V. O. John, W. Ingram, and P. Good, 2010: Current changes in tropical precipitation. Environ. Res. Lett., 5, 025205, doi:10.1088/1748-9326/5/2/025205.

    • Search Google Scholar
    • Export Citation
  • Ashfaq, M., L. C. Bowling, K. Cherkauer, J. S. Pal, and N. S. Diffenbaugh, 2010: Influence of climate model biases and daily scale temperature and precipitation events on hydrological impacts assessment: A case study of the United States. J. Geophys. Res., 115, D14116, doi:10.1029/2009JD012965.

    • Search Google Scholar
    • Export Citation
  • Bourlioux, A., and A. Majda, 2002: Elementary models with probability distribution function intermittency for passive scalars with a mean gradient. Phys. Fluids, 14, 881897.

    • Search Google Scholar
    • Export Citation
  • Draxler, R. R., and G. D. Hess, 1998: An overview of the Hysplit4 modeling system for trajectories, dispersion, and deposition. Aust. Meteor. Mag., 47, 295308.

    • Search Google Scholar
    • Export Citation
  • Duffy, P. B., B. Govindasamy, J. Milovich, K. Taylor, M. Wehner, A. Lamont, and S. Thompson, 2003: High resolution simulations of global climate, Part 1: Present climate. Climate Dyn., 21, 290371.

    • Search Google Scholar
    • Export Citation
  • Gershgorin, B., and A. Majda, 2011: Filtering a statistically exactly solvable test model for turbulent tracers from partial observations. J. Comput. Phys., 230, 16021638.

    • Search Google Scholar
    • Export Citation
  • Held, I. M., and B. J. Soden, 2000: Water vapor feedback and global warming. Annu. Rev. Energy Environ., 25, 441475.

  • Hilburn, K. A., and F. J. Wentz, 2008: Intercalibrated passive microwave rain products from the Unified Microwave Ocean Retrieval Algorithm (UMORA). J. Appl. Meteor. Climatol., 47, 778794.

    • Search Google Scholar
    • Export Citation
  • Holloway, C. E., and J. D. Neelin, 2009: Moisture vertical structure, column water vapor, and tropical deep convection. J. Atmos. Sci., 66, 16651683.

    • Search Google Scholar
    • Export Citation
  • Holloway, C. E., and J. D. Neelin, 2010: Temporal relations of column water vapor and tropical precipitation. J. Atmos. Sci., 67, 10911104.

    • Search Google Scholar
    • Export Citation
  • Hu, Y., and R. T. Pierrehumbert, 2001: The advection–diffusion problem for stratospheric flow. Part I: Concentration probability function. J. Atmos. Sci., 58, 14931510.

    • Search Google Scholar
    • Export Citation
  • Hu, Y., and R. T. Pierrehumbert, 2002: The advection–diffusion problem for stratospheric flow. Part II: Concentration probability function of tracer gradients. J. Atmos. Sci., 59, 28302845.

    • Search Google Scholar
    • Export Citation
  • Kalnay, E., and Coauthors, 1996: The NCEP/NCAR 40-Year Reanalysis Project. Bull. Amer. Meteor. Soc., 77, 437471.

  • Kollias, P., and B. Albrecht, 2010: Vertical velocity statistics in fair weather cumuli at the ARM TWP Nauru Climate Research Facility. J. Climate, 23, 65906604.

    • Search Google Scholar
    • Export Citation
  • Lintner, B. R., and J. D. Neelin, 2008: Eastern margin variability of the South Pacific convergence zone. Geophys. Res. Lett., 35, L16701, doi:10.1029/2008GL034298.

    • Search Google Scholar
    • Export Citation
  • Lintner, B. R., and J. D. Neelin, 2010: South America–Atlantic sector convective margins and their relationship to low-level inflow. J. Climate, 23, 26712685.

    • Search Google Scholar
    • Export Citation
  • Majda, A. J., and B. Gershgorin, 2010: Quantifying uncertainty in climate change science through empirical information theory. Proc. Natl. Acad. Sci. USA, 107, 14 95814 963.

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

    • Search Google Scholar
    • Export Citation
  • Mather, J. H., T. P. Ackerman, W. E. Clements, F. J. Barnes, M. D. Ivey, L. D. Hatfield, and R. M. Reynolds, 1998: An atmospheric radiation and cloud station in the tropical western Pacific. Bull. Amer. Meteor. Soc., 79, 627642.

    • Search Google Scholar
    • Export Citation
  • Muller C. J., L. E. Back, P. A. O’Gorman, and K. A. Emanuel, 2009: A model for the relationship between tropical precipitation and column water vapor. Geophys. Res. Lett., 36, L16804, doi:10.1029/2009GL039667.

    • Search Google Scholar
    • Export Citation
  • Neelin, J. D., B. R. Lintner, B. Tian, Q. Li, L. Zhang, P. K. Patra, M. T. Chahine, and S. N. Stechmann, 2010: Long tails in deep columns of natural and anthropogenic tropospheric tracers. Geophys. Res. Lett., 37, L05804, doi:10.1029/2009GL041726.

    • Search Google Scholar
    • Export Citation
  • Peters, O., and J. D. Neelin, 2006: Critical phenomena in atmospheric precipitation. Nat. Phys., 2, 393396.

  • Porch, W. M., S. C. Olsen, P. Chylek, M. K. Dubey, B. G. Henderson, and W. Clodius, 2006: Satellite and surface observations of Nauru Island clouds: Differences between El Niño and La Niña periods. Geophys. Res. Lett., 33, L13804, doi:10.1029/2006GL026339.

    • Search Google Scholar
    • Export Citation
  • Sherwood, S. C., 2010: Direct versus indirect effects of tropospheric humidity changes on the hydrologic cycle. Environ. Res. Lett., 5, 025206, doi:10.1088/1748-9326/5/2/025206.

    • Search Google Scholar
    • Export Citation
  • Solomon, S., D. Qin, M. Manning, M. Marquis, K. Averyt, M. M. B. Tignor, H. L. Miller Jr., and Z. Chen , Eds., 2007: Climate Change 2007: The Physical Science Basis. Cambridge University Press, 996 pp.

    • Search Google Scholar
    • Export Citation
  • Stechmann, S. N., and J. D. Neelin, 2011: A stochastic model for the transition to strong convection. J. Atmos. Sci., in press.

  • Stokes, G. M., and S. E. Schwartz, 1994: The Atmospheric Radiation Measurement (ARM) program: Programmatic design of the cloud and radiation test bed. Bull. Amer. Meteor. Soc., 75, 12021221.

    • Search Google Scholar
    • Export Citation
  • Walker, M. D., and N. S. Diffenbaugh, 2009: Evaluation of high-resolution simulations of daily scale temperature and precipitation over the United States. Climate Dyn., 33, 11311147.

    • Search Google Scholar
    • Export Citation
  • Wehner, M., R. L. Smith, G. Bala, and P. Duffy, 2010: The effect of horizontal resolution on simulation of very extreme US precipitation events in a global atmosphere model. Climate Dyn., 34, 241247.

    • Search Google Scholar
    • Export Citation
  • Zhang, C., B. Mapes, and B. J. Soden, 2003: Bimodality in tropical water vapor. Quart. J. Roy. Meteor. Soc., 129, 28472866.

Save