• Adler, R. F., and Coauthors, 2003: The version-2 Global Precipitation Climatology Project (GPCP) monthly precipitation analysis (1979–present). J. Hydrometeor., 4 , 11471167.

    • Search Google Scholar
    • Export Citation
  • Boehm, M. T., , and J. Verlinde, 2000: Stratospheric influence on upper tropospheric tropical cirrus. Geophys. Res. Lett., 27 , 32093212.

    • Search Google Scholar
    • Export Citation
  • Buck, A. L., 1981: New equations for computing vapor pressure and enhancement factor. J. Appl. Meteor., 20 , 15271532.

  • Clark, H. L., 2005: Longitudinal variability of water vapor and cirrus in the tropical tropopause layer. J. Geophys. Res., 110 , D07107. doi:10.1029/2004JD004943.

    • Search Google Scholar
    • Export Citation
  • Dima, I. M., , and J. M. Wallace, 2007: Structure of the annual-mean equatorial planetary waves in the ERA-40 reanalyses. J. Atmos. Sci., 64 , 28622880.

    • Search Google Scholar
    • Export Citation
  • Dinh, T. P., , D. R. Durran, , and T. P. Ackerman, 2010: Maintenance of tropical tropopause layer cirrus. J. Geophys. Res., 115 , D02104. doi:10.1029/2009JD012735.

    • Search Google Scholar
    • Export Citation
  • Durran, D. R., , T. Dinh, , M. Ammerman, , and T. Ackerman, 2009: The mesoscale dynamics of thin tropical tropopause cirrus. J. Atmos. Sci., 66 , 28592873.

    • Search Google Scholar
    • Export Citation
  • Fu, Q., , Y. Hu, , and Q. Yang, 2007: Identifying the top of the tropical tropopause layer from vertical mass flux analysis and CALIPSO lidar cloud observations. Geophys. Res. Lett., 34 , L14813. doi:10.1029/2007GL030099.

    • Search Google Scholar
    • Export Citation
  • Fueglistaler, S., , A. E. Dessler, , T. J. Dunkerton, , I. Folkins, , Q. Fu, , and P. W. Mote, 2009: Tropical tropopause layer. Rev. Geophys., 47 , RG1004. doi:10.1029/2008RG000267.

    • Search Google Scholar
    • Export Citation
  • Fujiwara, M., and Coauthors, 2009: Cirrus observations in the tropical tropopause layer over the western Pacific. J. Geophys. Res., 114 , D09304. doi:10.1029/2008JD011040.

    • Search Google Scholar
    • Export Citation
  • Gill, A. E., 1980: Some simple solutions for heat-induced tropical circulation. Quart. J. Roy. Meteor. Soc., 106 , 447462.

  • Guichard, F., , D. Parsons, , and E. Miller, 2000: Thermodynamic and radiative impact of the correction of sounding humidity bias in the tropics. J. Climate, 13 , 36113624.

    • Search Google Scholar
    • Export Citation
  • Highwood, E. J., , and B. J. Hoskins, 1998: The tropical tropopause. Quart. J. Roy. Meteor. Soc., 124 , 15791604.

  • Holton, J. R., 1979: An Introduction to Dynamic Meteorology. Academic Press, 391 pp.

  • Holton, J. R., , and R. S. Lindzen, 1968: A note on “Kelvin” waves in the atmosphere. Mon. Wea. Rev., 96 , 385386.

  • Holton, J. R., , P. H. Haynes, , M. E. McIntyre, , A. R. Douglass, , R. B. Rood, , and L. Pfister, 1995: Stratosphere–troposphere exchange. Rev. Geophys., 33 , 403439.

    • Search Google Scholar
    • Export Citation
  • Immler, F., , K. Krüger, , M. Fujiwara, , G. Verver, , M. Rex, , and O. Schrems, 2008: Correlation between equatorial Kelvin waves and the occurrence of extremely thin ice clouds at the tropical tropopause. Atmos. Chem. Phys., 8 , 40194026.

    • Search Google Scholar
    • Export Citation
  • Jensen, E. J., , O. B. Toon, , H. B. Selkirk, , J. D. Spinhirne, , and M. R. Schoeberl, 1996: On the formation and persistence of subvisible cirrus clouds near the tropical tropopause. J. Geophys. Res., 101 , 2136121375.

    • Search Google Scholar
    • Export Citation
  • Jensen, E. J., , L. Pfister, , T-P. Bui, , P. Lawson, , and D. Baumgardner, 2010: Ice nucleation and cloud microphysical properties in tropical tropopause layer cirrus. Atmos. Chem. Phys., 10 , 13691384.

    • Search Google Scholar
    • Export Citation
  • Leith, C. E., 1973: The standard error of time-averaged estimates of climatic means. J. Appl. Meteor., 12 , 10661069.

  • Massie, S., , A. Gettelman, , W. Randel, , and D. Baumgardner, 2002: Distribution of tropical cirrus in relation to convection. J. Geophys. Res., 107 , 4591. doi:10.1029/2001JD001293.

    • 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
  • Matsuno, T., 1966: Quasi-geostrophic motions in the equatorial area. J. Meteor. Soc. Japan, 44 , 2542.

  • McFarquhar, G. M., , A. J. Heymsfield, , J. Spinhirne, , and B. Hart, 2000: Thin and subvisual tropopause tropical cirrus: Observations and radiative impacts. J. Atmos. Sci., 57 , 18411853.

    • Search Google Scholar
    • Export Citation
  • Robinson, G. D., 1980: The transport of minor atmospheric constituents between troposphere and stratosphere. Quart. J. Roy. Meteor. Soc., 106 , 227253.

    • Search Google Scholar
    • Export Citation
  • Roundy, P. E., 2008: Analysis of convectively coupled Kelvin waves in the Indian Ocean MJO. J. Atmos. Sci., 65 , 13421359.

  • Spang, R., , G. Eidmann, , M. Riese, , D. Offermann, , P. Preusse, , L. Pfister, , and P-H. Wang, 2002: CRISTA observations of cirrus clouds around the tropopause. J. Geophys. Res., 107 , 8174. doi:10.1029/2001JD000698.

    • Search Google Scholar
    • Export Citation
  • Virts, K. S., 2009: Cirrus in the tropical tropopause transition layer: Formation mechanisms and influence of local and planetary-scale environment. M.S. thesis, Department of Atmospheric Sciences, University of Washington, 99 pp.

  • Virts, K. S., , and J. M. Wallace, 2010: Annual, interannual, and intraseasonal variability of tropical tropopause transition layer cirrus. J. Atmos. Sci., 67 , 30973112.

    • Search Google Scholar
    • Export Citation
  • Wallace, J. M., , and V. E. Kousky, 1968: Observational evidence of Kelvin waves in the tropical stratosphere. J. Atmos. Sci., 25 , 900907.

    • Search Google Scholar
    • Export Citation
  • Wang, P-H., , P. Minnis, , M. P. McCormick, , G. S. Kent, , and K. M. Skeens, 1996: A 6-year climatology of cloud occurrence frequency from Stratospheric Aerosol and Gas Experiment II observations (1985–1990). J. Geophys. Res., 101 , 2940729430.

    • Search Google Scholar
    • Export Citation
  • Webster, P. J., 1972: Response of the tropical atmosphere to local, steady forcing. Mon. Wea. Rev., 100 , 518541.

  • Wheeler, M., , G. N. Kiladis, , and P. J. Webster, 2000: Large-scale dynamical fields associated with convectively coupled equatorial waves. J. Atmos. Sci., 57 , 613640.

    • Search Google Scholar
    • Export Citation
  • Winker, D. M., , and C. R. Trepte, 1998: Laminar cirrus observed near the tropical tropopause by LITE. Geophys. Res. Lett., 25 , 33513354.

    • Search Google Scholar
    • Export Citation
  • Winker, D. M., , W. H. Hunt, , and M. J. McGill, 2007: Initial performance assessment of CALIOP. Geophys. Res. Lett., 34 , L19803. doi:10.1029/2007GL030135.

    • Search Google Scholar
    • Export Citation
  • Yang, Q., , Q. Fu, , J. Austin, , A. Gettelman, , F. Li, , and H. Vömel, 2008: Observationally derived and general circulation model simulated tropical stratospheric upward mass fluxes. J. Geophys. Res., 113 , D00B07. doi:10.1029/2008JD009945.

    • Search Google Scholar
    • Export Citation
  • Yin, X., , A. Gruber, , and P. Arkin, 2004: Comparison of the GPCP and CMAP merged gauge–satellite monthly precipitation products for the period 1979–2001. J. Hydrometeor., 5 , 12071222.

    • Search Google Scholar
    • Export Citation
  • View in gallery

    (a) Seven-day running mean cold point temperature at Manus (gray; inverted scale at left) and CALIPSO cloud fraction with base above 15 km within a 10° latitude × 10° longitude region centered on Manus (black; scale at right) from June 2006 to June 2009. (b) As in (a), but an 80-day high-pass Lanczos filter has been applied to both variables.

  • View in gallery

    Cross sections of (a) zonal-mean, annual-mean cloud fraction [contour interval (CI) = 0.05], (b) ERA-Interim relative humidity (CI = 10%) and (c) vertical velocity (CI = 1 mm s−1; zero contour in red). Cloud fractions calculated from CALIPSO profiles from June 2006 to June 2009. Latitude resolution is 5°.

  • View in gallery

    (a) Mean 100-hPa temperature (contours at … −80, −79, −76, −73, … °C; outermost dark contour is −79°C), (b) cloud fraction with base above 15 km (CI = 0.05), (c) cloud fraction in the 11–12-km layer (without reference to cloud-base altitude; CI = 0.05), and (d) GPCP precipitation (CI = 1.5 mm day−1). Latitude resolution is 5°, with values every 2.5°; longitude resolution is 10°.

  • View in gallery

    Correlations between filtered ERA 100-hPa temperatures within black reference boxes and filtered CALIPSO TTL cirrus index throughout the tropics (CI = 0.1). The strongest correlation is indicated in lower right corner of each panel.

  • View in gallery

    Correlations between filtered CALIPSO TTL cirrus index within black reference boxes and filtered 5° latitude × 5° longitude ERA 100-hPa temperatures (colors and contours) and winds (vectors) throughout the tropics (CI for temperature–cloud fraction correlations = 0.1). The strongest correlation is indicated in lower right corner of each panel.

  • View in gallery

    As in Fig. 4, but with filtered variables within the Maritime Continent reference box as indicated in the upper left corner of each panel.

  • View in gallery

    Correlations between filtered ERA 100-hPa temperatures within 10° latitude × 10° longitude reference boxes (centered on white stars) and filtered CALIPSO height-dependent tropical cloud index from 5°S to 5°N (CI = 0.1). The strongest correlation is indicated in lower right corner of each panel.

  • View in gallery

    As in Fig. 7, but for filtered ERA 300-hPa ω in Maritime Continent reference box.

  • View in gallery

    Correlations between filtered CALIPSO TTL cirrus index within 10° latitude × 10° longitude reference box (black vertical lines) and filtered ERA temperatures (colors and contours) from 2.5°S to 2.5°N. ERA longitudinal resolution is 5°; CI = 0.1.

  • View in gallery

    (left) Correlation and (right) regression coefficients between CALIPSO-derived cloud fraction and ARM radiosonde temperatures at Manus as functions of height. Correlations indicating 95% significance level are plotted as dotted lines in the left panel.

  • View in gallery

    (left) Composite “clear” (red) and “cloudy” (blue) temperature profiles at Manus as a function of height, constructed from CALIPSO cloud profiles and radiosonde temperatures, along with Monte Carlo–derived 2-standard deviation confidence intervals (red and blue dashed; see text for methodology). (right) As at left, but composite “warm” (red) and “cold” (blue) cloud fraction profiles based on radiosonde cold point temperatures from Manus.

  • View in gallery

    (left) Scatterplot of mean CALIPSO TTL cirrus index within regularly spaced bins of ERA 100-hPa temperatures (°C, both variables averaged over 10° latitude × 10° longitude regions over 7-day periods). (right) Scatterplot of mean cloud fraction within regularly spaced bins of ARM temperatures [°C, both variables averaged over 15–16 km (triangles), 16–17 km (dots), and 17–18 km (asterisks), and cloud fraction averaged over 5° latitude × 5° longitude region centered on Manus]. Error bars indicate 10th and 90th percentiles of cloud fraction for each interval and are plotted for the middle layer only in the right panel.

  • View in gallery

    (left) Threshold temperatures below which mean cloud occurrence greater than 10% (pluses), 30% (dots), 50% (asterisks), and 70% (circles) are observed (cloud fractions are calculated over a 5° latitude × 5° longitude region centered on Manus). Mean temperature (T; dash–dotted) and frost point temperature (Tf; dashed) from Manus and mean frost point temperature, measured by frost point hygrometers, from Biak (Tfb; gray dashed; see text for details) are also plotted. (right) As at left, but best linear fit to each cloud fraction threshold is plotted.

  • View in gallery

    Time–height section of radiosonde-derived temperature anomalies from Manus (colors and contours; CI = 2°C) and CALIPSO-derived cloud fraction within successive 200-m layers, calculated for a 5° latitude × 5° longitude region centered on Manus (vertical black dots, scaled by cloud fraction observed). Cold point tropopause height plotted as thick black line.

  • View in gallery

    (top) Filtered CALIPSO TTL cirrus index and ERA 100-hPa winds throughout tropics regressed onto PC time series of the first EOF of filtered TTL cirrus index (EOF calculated using only points with mean cloud fraction ≥0.1). The variance explained is indicated in the lower right corner. (bottom) Cross section of filtered CALIPSO height-dependent tropical cloud index (from 5°S to 5°N) and ERA winds (plotted every 15° and at selected vertical levels) regressed onto PC time series from the top panel. The vertical component of arrows represents the heating rate (see text); wind vector scaling is arbitrary, and the vertical scale of the gray arrows is smaller than that for the black arrows.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 115 115 12
PDF Downloads 123 123 26

Tropical Tropopause Transition Layer Cirrus as Represented by CALIPSO Lidar Observations

View More View Less
  • 1 Department of Atmospheric Sciences, University of Washington, Seattle, Washington
© Get Permissions
Full access

Abstract

The spatial and temporal variability of cirrus cloud fraction within the tropical tropopause transition layer (TTL) is investigated based on three years of data from the Cloud–Aerosol Lidar and Infrared Pathfinder Satellite Observations (CALIPSO) mission, analyzed in conjunction with fields from the European Centre for Medium-Range Weather Forecasts Re-Analysis (ERA)-Interim and temperature profiles from radiosondes launched at Manus Island, Papua New Guinea (2°S, 147°E). TTL cirrus is found to be mainly confined to the rising branch of the Hadley cell within ∼15° of the equator, with maximum cloud fraction between 14 and 15 km. The time-varying spatial pattern of cloud fraction within this belt does not resemble the pattern of cloud fraction in the layer below, as would be expected if the TTL cirrus were formed by the spreading of the anvils of convective clouds. On the contrary, within the stably stratified layer above ∼13 km, cirrus cloud fraction and temperature both appear to be modulated by the planetary-scale vertical velocity field. The time-varying spatial patterns are reminiscent of the vertical-propagating Kelvin wave response to an equatorial heat source, with the coldest, cloudiest air in the TTL centered approximately 30° of longitude to the east of the strongest heating.

Corresponding author address: Katrina Virts, Department of Atmospheric Sciences, 408 ATG Bldg., Box 351640, Seattle, WA 98195–1640. Email: kvirts@uw.edu

Abstract

The spatial and temporal variability of cirrus cloud fraction within the tropical tropopause transition layer (TTL) is investigated based on three years of data from the Cloud–Aerosol Lidar and Infrared Pathfinder Satellite Observations (CALIPSO) mission, analyzed in conjunction with fields from the European Centre for Medium-Range Weather Forecasts Re-Analysis (ERA)-Interim and temperature profiles from radiosondes launched at Manus Island, Papua New Guinea (2°S, 147°E). TTL cirrus is found to be mainly confined to the rising branch of the Hadley cell within ∼15° of the equator, with maximum cloud fraction between 14 and 15 km. The time-varying spatial pattern of cloud fraction within this belt does not resemble the pattern of cloud fraction in the layer below, as would be expected if the TTL cirrus were formed by the spreading of the anvils of convective clouds. On the contrary, within the stably stratified layer above ∼13 km, cirrus cloud fraction and temperature both appear to be modulated by the planetary-scale vertical velocity field. The time-varying spatial patterns are reminiscent of the vertical-propagating Kelvin wave response to an equatorial heat source, with the coldest, cloudiest air in the TTL centered approximately 30° of longitude to the east of the strongest heating.

Corresponding author address: Katrina Virts, Department of Atmospheric Sciences, 408 ATG Bldg., Box 351640, Seattle, WA 98195–1640. Email: kvirts@uw.edu

1. Introduction

The notion that extensive cirrus layers should exist within the tropical tropopause transition layer (TTL)—that is, the layer extending from ∼14 km, above the tops of deep convective clouds, up to ∼18.5 km, above the cold point—in regions of large-scale ascent is consistent with and was actually anticipated on the basis of dynamical considerations (Robinson 1980; Holton et al. 1995; Fueglistaler et al. 2009). Optically thin cirrus layers have been detected using ground-based lidars (e.g., Mather et al. 1998), lidars aboard experimental aircraft (McFarquhar et al. 2000) and a space shuttle (Winker and Trepte 1998), and satellite radiometers (Wang et al. 1996). TTL cirrus clouds have been detected throughout the tropics but are especially prevalent above the western Pacific warm pool (Wang et al. 1996; Fu et al. 2007). They occur most frequently at levels below 16 km, although some clouds have been observed as high as 18 km (Fu et al. 2007). The Cloud–Aerosol Lidar and Infrared Pathfinder Satellite Observations (CALIPSO) satellite, launched in 2006, carries a two-wavelength polarization lidar, the first space-based lidar optimized for cloud and aerosol layer detection. This instrument is capable of detecting subvisible cirrus layers with optical depths of 0.01 or less (Winker et al. 2007).

Two contrasting formation mechanisms for TTL cirrus have been advanced in the literature:

  • Detrainment: Outflow from the anvil region of deep convective clouds has high ice water content. Larger ice crystals precipitate out within a few hours (Jensen et al. 1996), but the remaining smaller crystals can form an optically thin cirrus cloud layer that may persist long after the underlying cumulonimbus has dissipated.
  • In situ formation: Stratiform ascent in equatorial planetary-scale waves can adiabatically cool air to its frost point, below which ice crystals can form, giving rise to optically thin clouds that can persist for several days if the fall rates of the ice crystals are slow enough (Jensen et al. 1996; Jensen et al. 2010).

In situ cirrus formation in the manner described above can occur in response to, for example, upwelling forced by the Hadley circulation, the Brewer–Dobson circulation, or planetary-scale ascent in diabatically forced equatorial planetary waves, as discussed by Matsuno (1966), Webster (1972), and Gill (1980). These waves are forced primarily by the release of latent heat in the updrafts of deep convective clouds in the midtroposphere, but the diabatically induced, planetary-scale circulation may extend into the stably stratified layer above the convection. Several previous studies have shown evidence of planetary-wave signatures in tropical cirrus. Examining radiosonde and lidar data from the Atmospheric Radiation Measurement Program (ARM) site at Nauru (see section 2), Boehm and Verlinde (2000) found that TTL cirrus are detectable only during the cold phases of Kelvin waves propagating downward from the lower stratosphere into the TTL. The waves observed by Boehm and Verlinde had 5–10-day periods; because of limited data availability, the authors were unable to investigate fluctuations with time scales longer than that. Similar results were obtained by Immler et al. (2008) based on lidar observations over Suriname. In addition, the formation and dissipation of TTL cirrus detected above the warm pool by a ship-based lidar during boreal winter field campaigns were observed to be controlled by the passage of convectively induced perturbations in the TTL, with Kelvin wave packets exerting the stronger influence (Fujiwara et al. 2009).

Global, vertically resolved views of TTL cirrus obtained from instrumentation carried aboard the CALIPSO satellite offer an unprecedented opportunity to investigate the factors that control the distribution of cirrus within the TTL. In this paper, we describe the climatological-mean distribution of TTL cirrus in the annual-mean fields and how it varies on time scales shorter than ∼80 days. In the companion paper (Virts and Wallace 2010), we document the variations in TTL cirrus that occur in association with the annual cycle, the Madden–Julian oscillation, and ENSO. The remainder of this paper is organized as follows: section 2 describes the data sources and analysis techniques. The climatological-mean, annual-mean distribution of TTL cirrus is described in section 3. The observed three-dimensional pattern of cirrus cloud fraction is found to be consistent with the notion that most TTL cirrus forms in situ. The planetary-scale environment and circulations observed in association with TTL cirrus are documented in section 4. Section 5 provides more quantitative information on the statistical relationship between TTL cirrus cloud fraction and temperature, making use of high-resolution radiosonde data at a single station in the equatorial western Pacific. Discussion and concluding remarks are presented in section 6.

2. Data and analysis techniques

a. Data sources

The Cloud–Aerosol Lidar with Orthogonal Polarization (CALIOP) aboard the polar-orbiting CALIPSO satellite has been acquiring vertical profiles of clouds and aerosols around the world since June 2006. CALIPSO data have high spatial resolution—5 km in the horizontal (along-track) and 60 m in the vertical for the altitudes of interest (Winker et al. 2007). Data from the first ∼36 months of CALIPSO’s operations (June 2006–June 2009) are analyzed in this study. Documentation of the algorithms employed to identify clouds and cloud-base and cloud-top heights can be found online (at http://eosweb.larc.nasa.gov/PRODOCS/calipso/table_calipso.html). Following Fu et al. (2007), opaque cloud layers are assigned a cloud base at the earth’s surface, in effect assuming that such layers are deep convective clouds with low-altitude bases.

After this paper and the companion paper were accepted for publication, the Atmospheric Science Data Center released a new version of the CALIPSO data. The primary effect of the changes in the new version was to substantially reduce low cloud fraction. We have updated our cross sections that significantly include low cloud fraction—Fig. 2 in this paper and Figs. 6, 10, and 15 in the companion paper—to reflect the latest CALIPSO version.

Atmospheric conditions are represented by variables from the European Centre for Medium-Range Weather Forecasts (ECMWF) Re-Analysis (ERA)-Interim. ERA-Interim analyses are available 4 times daily; their horizontal resolution is 1.5°, spanning the entire globe, and in the vertical analyses are available at 37 pressure levels from sea level through the stratosphere. The primary focus will be on latitudes equatorward of 30° (hereafter, the “tropics”). Analogous and more extensive analyses than those presented here, based on the initial analyses of the Global Forecast System (GFS) model and the first two years of CALIPSO data, are presented in Virts (2009).

Gridded (1° horizontal resolution) daily precipitation rates from the Global Precipitation Climatology Project (GPCP) are also used in this study. This dataset was created by combining precipitation estimates from microwave imagers, outgoing longwave radiation (OLR) sensors, and vertical sounders aboard an assortment of satellites with ground-based rain gauge data (Adler et al. 2003; Yin et al. 2004). At the time of analysis, GPCP and ERA-Interim data were available through April 2009.

Temperature profiles from radiosonde observations at Manus Island, Papua New Guinea (2°S, 147°E), are analyzed in section 5. This observing site is maintained by the Department of Energy’s ARM (Mather et al. 1998). Data are gathered every 2 s during ascent, yielding soundings with ∼10-m vertical resolution; soundings are twice daily. As a quality control, at each vertical level, temperatures farther than five standard deviations from the mean at that level (in all, less than 0.2% of the total data) are discarded.

b. Analysis techniques

CALIPSO’s sampling in the along-path direction (5 km) is more than ample for comparison with the more coarsely gridded ERA and GPCP variables, but its irregular sampling in time is a serious limitation. Any given location along CALIPSO’s orbital path is sampled only once every 16 days, which is not frequent enough to detect any relationship with convection or to provide a representative time series of cloudiness there. To circumvent this limitation, a CALIPSO TTL cirrus index is defined as follows: the length of the CALIPSO record is divided into 156 consecutive 7-day (“weekly”) segments. Dividing the tropics into 10° latitude × 10° longitude squares, the “TTL cirrus index” (also referred to in this paper as “cloud fraction”) for each week is defined as the fraction of CALIPSO profiles acquired within each 10° × 10° grid square in which a cloud deck with a base above 15 km is identified. Similarly, a CALIPSO height-dependent cloud index is defined within the same 10° × 10° regions and 7-day periods as the TTL cirrus index but is calculated as cloud fraction within successive 200-m layers. On average, the satellite passes over some portion of a 10° × 10° region in the equatorial belt 5 times per week. ERA and GPCP variables have been averaged over each of the 7-day sampling periods to allow for comparison with the cirrus index; horizontal averaging has also been applied to these variables when indicated.

The TTL cirrus index for a 10° latitude × 10° longitude region centered above the ARM site in Manus is plotted in black in Fig. 1a. In this figure only, cloud fraction is calculated as 7-day running means rather than nonoverlapping weekly values. For comparison, cold point temperatures from the Manus radiosonde observations, also represented as 7-day running means, are shown in gray in Fig. 1. (For each radiosonde launch, the level with the lowest reported temperature is defined as the cold point tropopause, provided that its altitude is at least 15 km and that the temperature profile extends at least 1 km above it.) A pronounced annual cycle is evident in both the cold point temperature and cloud fraction time series, with the lowest (highest) temperatures and largest (smallest) cloud fractions observed during the boreal winter (summer) months. Some interannual variability in both cold point temperature and cloud fraction is also apparent: the winter of 2007/08 (2006/07) is coldest (warmest) and cloudiest (clearest) of the three. The seasonal and interannual variability of TTL cirrus are explored in the companion paper.

The present study focuses on the relationship between TTL cirrus and the surrounding environment on the intraseasonal time scale. To highlight this variability, an 80-day high-pass Lanczos filter was applied to each dataset when indicated. This filter was applied to the time series in Fig. 1a to produce the filtered time series shown in Fig. 1b. The filtered temperature and cloud fraction time series are significantly correlated with one another (r ≈ −0.61). Furthermore, both filtered time series exhibit a tendency for quasi-periodic behavior, with cycles of ∼40 days in length. A spectral peak in this frequency range is suggestive of a possible association with the Madden–Julian oscillation, as explored in the companion paper.

The Student’s t test is used to determine the statistical significance of correlation coefficients. The effective number of degrees of freedom of the datasets used in section 4 (after filtering and temporal and spatial averaging, as described earlier in this section), calculated using the formula of Leith (1973), is at least 143, which means that correlations greater than 0.17 (0.22) in absolute value are significant at the 95% (99%) level when a two-sided distribution is assumed. Most relationships of interest in this study are marked by correlation coefficients much higher than these threshold values. In section 5, neither filtering nor time averaging is applied to the data, yielding a larger sample size but a greater interdependence of individual data points. The 95% level is used to evaluate the significance of those results.

3. Annual-mean fields

Figure 2 shows meridional cross sections of annual-mean cloud fraction, as defined by CALIPSO measurements processed as indicated in section 2, and both relative humidity and vertical velocity (derived from ω, the vertical velocity in pressure coordinates; mm s−1), as defined by the ERA-Interim analyses. The highest cloud fractions and relative humidities and the strongest ascent are observed at ∼7°N, along the latitude of the oceanic intertropical convergence zones (ITCZs). At latitudes extending from 15°S to 20°N, cloud fraction and relative humidity exhibit a distinctive maximum within the TTL, with values in excess of 20% and 50%, respectively. This is the region of primary interest in this study.

The annual-mean GPCP precipitation climatology, a measure of the rate of ascent and latent heat release in deep convective clouds, shown in Fig. 3d, is dominated by the Pacific and Atlantic ITCZs, the South Pacific convergence zone (SPCZ), and the continental rain belts. These features are mirrored in the CALIPSO data for cloud fraction in the 11–12-km layer, which is situated just below the convective cloud tops of the convective clouds (Fig. 3c), but the TTL cirrus cloud fraction, represented here as the frequency of occurrence of clouds with bases above 15 km (Fig. 3b), exhibits a smoother, simpler distribution with greater equatorial symmetry and much less emphasis on the oceanic convergence zones. The primary maximum in TTL cirrus over the western Pacific warm pool is displaced eastward of its counterpart in the GPCP rainfall field, indicative of an eastward tilt of the cloud pattern with height.

If TTL cirrus were formed primarily by the spreading of anvils from deep convective clouds, then one might expect the TTL and convective cloud distributions to look broadly similar. It is conceivable that advection by the horizontal wind field could distort the shapes of anvils as they spread, but it is not obvious why it should simplify the pattern of TTL cirrus and make it more symmetric about the equator than the distribution of convective clouds from which it was supposedly formed. Indeed, it seems more likely that TTL cirrus is generated in situ, by the lifting of air in the ascending branch of the zonally symmetric Hadley cell augmented by ascent in the equatorially trapped planetary waves. Because the air in the TTL is stably stratified, planetary-scale ascent also induces cooling. The similarity between the distribution of TTL cirrus and the distribution of 100-hPa temperature, with enhanced cloudiness in the colder regions (Fig. 3, top two panels), provides supporting evidence of in situ formation. With its broader meridional extent, the 100-hPa pattern (Fig. 3a) provides a more complete description of the planetary-wave pattern. The nose of cold, cloudy air extending eastward along the equator from the Maritime Continent toward the date line is the Kelvin-wave signature, and the off-equatorial lobes extending westward into the western Indian Ocean, more prominent in 100-hPa temperature than in cloud fraction, constitute the Rossby-wave signature (Gill 1980).

Throughout the remainder of this paper and the companion paper, we will be describing the structure and evolution of variations about the climatological-mean basic state depicted in Figs. 2 and 3. We will show that the variations about the mean state are also characterized by geometrically simple, equatorially symmetric patterns with analogous Kelvin and Rossby wave signatures and a strong inverse relationship between TTL cirrus cloud fraction and 100-hPa temperature in the equatorial belt. We will argue that these features constitute further evidence of in situ TTL cirrus formation in regions of ascent associated with convectively forced equatorial planetary waves. Hence, the evidence presented in these papers consistently suggests that tropical convection induces TTL cirrus not so much directly, through the spreading of the anvils of convective clouds, as indirectly, by forcing patches of planetary-scale ascent within the stably stratified layer above the cloud tops.

4. Temporal variations in TTL cirrus: A planetary-scale perspective

In this section, we investigate the relationship between TTL cirrus and its planetary-scale environment. We begin by showing the spatial patterns obtained by correlating time series of filtered cloud fraction with bases above 15 km in 10° × 10° boxes throughout the tropics with selected reference time series of atmospheric variables at specified locations. Figure 4 shows correlations between 100-hPa temperatures from the ERA-Interim analyses averaged over five reference boxes (outlined in black) and the gridded cloud index throughout the tropics. The reference boxes have been chosen to represent the three areas of maximum TTL cirrus occurrence—Africa, the Maritime Continent and western Pacific, and South America (see Fig. 3). In each panel, 100-hPa temperature is significantly negatively correlated with TTL cirrus in the region surrounding the reference box, indicative of ascent within the TTL throughout a broad, planetary-scale region. Above the Maritime Continent and the western Pacific (Figs. 4b–d), the cirrus fields in Fig. 4 assume the shape of the Kelvin–Rossby planetary wave signature described in section 3, with negative cirrus correlations extending along the equator to the east of the reference boxes and off the equator to the west of the boxes. Positive (though weaker) TTL cirrus anomalies are discernible to the east and/or west of the reference boxes in all panels of Fig. 4. The reference box in Fig. 4c is centered close to the Manus observation site. An analogous plot with radiosonde-derived cold point temperature at Manus as the reference variable yields a similar cloud pattern (Virts 2009).

Additional information regarding the planetary-scale circulations associated with TTL cirrus occurrence can be obtained by reversing the analysis protocol of Fig. 4 so that cloud fraction is the reference variable. Figure 5 shows correlations between the filtered CALIPSO cirrus index within the same five reference boxes and filtered ERA 100-hPa temperatures (colors and contours) and winds (arrows). In each panel of Fig. 5, TTL cirrus is significantly negatively correlated with 100-hPa temperatures over regions about as large as the regions of significant cloud correlations in Fig. 4. Above the Maritime Continent, correlations exceed 0.7 in absolute value. The temperature fields exhibit a high degree of equatorial symmetry throughout the tropics, and the off-equatorial features tend to be more pronounced than in Fig. 4.

Planetary-wave temperature signatures are apparent in the vicinity of the reference boxes for the Maritime Continent and western Pacific regions (Figs. 5b–d). The wind correlations indicate divergence in the vicinity of the reference boxes—in each panel, anomalous westerlies (easterlies) are observed along the equator to the east (west) of the reference box. Planetary-wave temperature signatures are not observed in the temperature signatures above Africa and South America (Figs. 5a,e). There is a suggestion of a wave train in all the panels of Fig. 5. The patterns in Figs. 5a, 5b, and 5e are suggestive of an out-of-phase relationship between 100-hPa temperature over the Maritime Continent and equatorial Africa and South America.

One-point correlation plots analogous to those in Fig. 4, but with filtered 100-hPa ω and relative humidity in the Maritime Continent reference box, are shown in the first and second panels of Fig. 6, respectively. TTL cirrus occurrence is significantly correlated with vertical velocity and relative humidity at the 100-hPa level, but the correlations are not quite as strong and do not extend over quite as broad a region as those with 100-hPa temperature in Fig. 4b. The same is true of the correlation maps for other reference grid boxes (not shown). We speculate that cirrus formation takes place in air that has a history of ascent and that temperature perturbations may be a better indicator of this recent history than vertical velocity itself; in addition, the ERA-Interim 100-hPa temperature field may be more reliable than the corresponding relative humidity and vertical velocity fields.

Correlation fields between filtered ERA 300-hPa ω and relative humidity in the Maritime Continent reference box and filtered TTL cirrus fraction throughout the tropics are shown in the third and fourth panels of Fig. 6, respectively. The 300-hPa level lies near 10.5-km altitude, just above the level of maximum ascent in the more convective regions of the tropics (Dima and Wallace 2007). An analogous one-point correlation map, with filtered GPCP precipitation as the reference variable, is shown in the fifth panel of Fig. 6. In the third panel, TTL cirrus seems to be correlated with midtropospheric ascent, but the cirrus anomalies are centered ∼30° of longitude to the east of the reference box; in the vicinity of the box itself, the correlation is weak, and the sign varies from one reference box to another (not shown). Likewise, in neither the fourth nor fifth panels is the indicator of deep convection in the reference box significantly correlated with local TTL cirrus cloud fraction. The same is true for other reference boxes; pointwise correlations in excess of 0.2 between precipitation and cloud fraction are observed only in isolated patches (Virts 2009). This confirms our result from section 3 that TTL cirrus cloud formation is not dominated by the shearing of the anvils of deep convective clouds.

The relationship between TTL cirrus and both planetary wave and convective activity is further explored by considering a vertical cross section of the cloud field. Figure 7 shows cross sections of the correlations between filtered 100-hPa temperature within the same five reference boxes (white stars are centered on the 100-hPa level and indicate the center of each box) and the filtered, height-dependent cloud index throughout the equatorial belt. As in the foregoing results, cloud fraction in the upper troposphere in Fig. 7 is negatively correlated with 100-hPa temperature. Below the ∼14-km level, correlations between 100-hPa temperature and cloud fraction within the reference boxes are weak and of inconsistent sign. Above the Maritime Continent and western Pacific (Figs. 7b–d), the cloud signature within the TTL tilts eastward with height. A weaker, but still statistically significant, cloud feature extending down to the earth’s surface ∼30° of longitude to the west of the reference boxes indicates the location of the associated deep convection. No nearby convective signature or tilt with altitude of the primary TTL cloud signature is apparent in the African and South American panels of Fig. 7. Secondary cloud features can also be observed in several panels.

Figure 8 shows an analogous plot for the Maritime Continent reference box, with 300-hPa ω in place of 100-hPa temperature as the reference variable. A vertically aligned cloud feature can be observed in the reference box, with peak correlation of −0.5. Below the 10–12-km level, the feature in the cloud field is nearly vertically aligned; above that level, it exhibits an eastward tilt with height. The tilting bands of strong correlations in Figs. 7 and 8 should not be interpreted as indicating a continuous cloud layer emanating from the top of the convective region and being advected eastward over an extent of several thousand kilometers; in fact, the most extensive cirrus layers observed to date extended over only ∼1000 km (Winker et al. 2007). It seems more likely that this band represents the region of planetary-scale ascent and low temperatures within which the development of cirrus cloud decks is favored.

Figure 9 shows a longitude–pressure cross section of correlations between the filtered temperature field throughout the equatorial belt and the filtered TTL cirrus index for the Maritime Continent reference box (outlined in black). A change in the sign of the temperature anomalies in the waves is evident near the 150-hPa level within and around the reference box. The strongest negative correlations are found at the 100-hPa level, while the strongest positive correlations lie between the 300- and 200-hPa levels. An eastward tilt with height is evident above the ∼250-hPa level, similar to that identified in cross sections of idealized equatorially trapped Kelvin waves in Holton and Lindzen (1968), Wallace and Kousky (1968), and Holton (1979). The cloud fields in Fig. 7 exhibit a similar eastward tilt with height. Below the ∼250-hPa level in Fig. 9, correlations are weaker, and the temperature field tilts westward with height. The shape of the region of anomalous warmth is reminiscent of the boomerang-shaped fields identified in association with convectively coupled waves by Wheeler et al. (2000) and Roundy (2008). Negative temperature anomalies between the 400- and 200-hPa levels over Africa and the central Pacific, and the overlying positive anomalies in the TTL, mark regions of subsidence in the wave train.

The results shown in these figures indicate that TTL cirrus is significantly correlated with Gill (1980)-like planetary wave perturbations within the TTL. The maximum amplitude of the perturbations in the geopotential height and zonal wind fields occurs at the level of the sign reversal in the temperature correlations in Fig. 9, just above the 150-hPa level.

5. Temporal variations in TTL cirrus: A local perspective

We now investigate the statistical relationship between temperature and cirrus cloud fraction in the TTL, making use of the high vertical resolution of the ARM temperature profiles. In comparing these variables, we do not mean to imply that cloud fraction is determined by the temperature; rather, as we stated in section 4, we view both variables as indicators of the air parcel’s recent history of vertical velocity. To compare the ARM and CALIPSO data, CALIPSO profiles acquired within a 5° latitude × 5° longitude box centered on Manus are used. [Analogous analyses based on the Nauru observing site or on wintertime data only yield similar results (Virts 2009).] Manus was overpassed more than 475 times between June 2006 and June 2009. For each CALIPSO pass, a vertical profile of cloud fraction is constructed for consecutive 200-m layers. The closest radiosonde launch in time is designated as indicative of the environmental conditions, provided that the launch took place within 24 h of the time of the CALIPSO overpass. Temperatures are taken as averages over the same 200-m layers.

The left panel of Fig. 10 shows correlations between cloud fraction and temperature at Manus as a function of altitude. The 95% significance level is plotted as dashed lines. In the ∼6–13-km layer, correlations are predominantly positive, though weak at any individual level. The prevalence of correlations of the same sign within this deep layer indicates that this result is more statistically significant than indicated by the Student’s t test applied at individual vertical levels. Taking cloud occurrence as a proxy for rising motion, the circulations within this layer can be viewed as thermally direct. Above the 15-km level, correlations in Fig. 10 between cloud fraction and ambient temperature become significantly negative, with values below −0.6 between 16 and 17 km. These negative correlations are an indication of thermally indirect circulations within the TTL, in agreement with the observations of Dima and Wallace (2007).

To ascertain the magnitude of the temperature perturbations associated with cloud occurrence in the upper troposphere and TTL, we first regress temperature onto cloud fraction at each vertical level to obtain a measure of ∂T/∂CF, where T is temperature and CF is cloud fraction. Regression coefficients from Manus are plotted as a function of height in the right panel of Fig. 10. Between 6 and 13 km, regression coefficients are positive (i.e., clouds tend to occur during warmer conditions) and range up to 0.9°C, indicating temperature differences of <1°C between clear and cloudy skies. Within the upper part of the TTL, the regression coefficients are strongly negative, indicating that much lower temperatures are observed when clouds are present. At 17 km, regression coefficients are near −9.4°C (or a temperature drop of 0.94°C per 10% increase in cloud fraction).

To further investigate the temperature anomalies associated with tropical clouds, the mean cloud fraction at each vertical level is determined, and soundings with cloud fractions less than the mean are designated as “clear” and those greater than the mean as “cloudy” at that level. This analysis yields an estimate of the mean temperatures observed in association with relatively clear and cloudy conditions at each altitude, irrespective of whether clouds are observed at other altitudes. The resulting temperature profiles are shown in the left panel of Fig. 11. To assess the statistical significance of these results, a Monte Carlo test is applied at each vertical level. Radiosonde-derived temperatures at each level are randomly sorted into two subsets of the same size as the cloudy and clear subsets, and the mean temperature of each subset is calculated. The mean subset temperatures derived from 1000 such Monte Carlo simulations are nearly normally distributed. At each vertical level, the temperatures representing two standard deviations of these distributions are plotted in the red and blue dashed lines in Fig. 11.

Below the 14-km level, the temperature anomalies associated with greater than average cloudiness are generally positive but are no greater than 0.5°C in magnitude and are thus not discernible in Fig. 11. Within the stably stratified TTL, temperature differences between the cloudy and clear profiles are much larger. For example, at the 16.5-km level above Manus, near the 100-hPa level, the mean temperature for cloudy soundings is ∼4.1°C lower than the mean temperature for clear soundings (the standard deviation of the temperature distribution at the 16.5-km level above Manus is 3.16°C; at this level, ∼33% of CALIPSO passes were classified as cloudy). At this level, the mean temperatures for both the cloudy and clear soundings are significantly different than the annual-mean temperature at Manus (plotted in black)—in fact, each is separated from the climatology by 11 or more standard deviations as observed in the Monte Carlo simulations. Even stronger cold anomalies are observed between the 17- and 18-km levels in Fig. 11. Anomalies of this magnitude are typical of cloudy conditions above the Maritime Continent and warm pool regions; above Africa and South America, anomalies of ∼2°–4°C are observed (Virts 2009).

The inverse analysis of that in the left panel of Fig. 11 is presented in the right panel of that figure, in which vertical profiles of cloud fraction are plotted for CALIPSO passes associated with cold point temperatures below and above the mean cold point temperature at Manus (profiles are designated as “cold” and “warm,” respectively). Below (above) normal cold point temperatures are associated with significantly high (low) cloud fractions between the 13- and 18-km levels. The largest cloud fraction associated with the mean cold profile is 0.61 at the 15.4-km level. The difference between the mean cold and warm cloud fractions is largest at 16.5 km, where cloud occurrence is ∼7.5 times more likely when the cold point temperature is below normal.

Scatterplots of cloud fraction and ambient temperature are shown in Fig. 12. The left panel shows the relationship between the TTL cirrus index in 10° × 10° grid boxes and ERA-Interim 100-hPa temperature. In preparing this plot, TTL cirrus fraction was paired with collocated ERA temperatures and then averaged over 1°C temperature bins; all data equatorward of 30° latitude were included. The tendency for cloudier conditions in the TTL when temperatures are lower is evident in Fig. 12: the lowest (<−83°C) temperatures are associated with cloud fractions in the range of 0.35–0.45. These relationships are largely insensitive to the size of the grid box used or to whether the data are time averaged (not shown).

An analogous plot, but based on data from the ARM radiosondes, is shown in the right panel of Fig. 12. For this plot, temperature and cloud fraction were averaged over 1-km layers prior to binning. Scatterplots are shown for the 15–16-, 16–17-, and 17–18-km layers. In each layer, larger cloud fractions are associated with lower temperatures, and mean cloud fractions greater than 0.7 are observed at the low end of the temperature range. The results in Fig. 12 are not in agreement with those of Clark (2005), who observed decreasing cloud fraction [identified by the Cryogenic Limb Array Etalon Spectrometer (CLAES)] at the 100-hPa level when ambient temperatures dropped below −84°C. We observe the existence of clouds at temperatures as low as −90°C. The difference between our results and Clark’s may be due to their inclusion of clouds with lower bases (CLAES detects cirrus occurrence within a 2.5-km layer centered on 100 hPa) and to differences in the abilities of the CLAES and CALIPSO sensors to detect optically thin cirrus.

Also of interest in Fig. 12 is that, as altitude increases, cirrus are observed in association with lower and lower temperatures—for the 15–16-km layer, mean cloud fractions ≥50% are observed with temperatures below −77°C, but mean cloud fractions reach 50% within the 16–17-km layer only when temperatures drop below −84°C and in the 17–18-km layer when they drop below −88°C. The temperature thresholds for various cloud fractions are more clearly discernible in the left panel of Fig. 13, in which the highest temperatures associated with mean cloud occurrence greater than 10%, 30%, 50%, and 70%, as observed near Manus, are plotted as a function of height. A straight line is fitted to the temperature profile for each threshold value and plotted in the right panel of Fig. 13. Also shown are observed mean temperature and frost point temperature profiles for Manus [the latter was calculated from dewpoint temperature using http://www.humidity-calculator.com, which uses the formulations of Buck (1981) and his subsequent updates]. Below the ∼16-km level, the fitted threshold lines closely parallel the climatological temperature and frost point profiles. Using the temperature thresholds and mean frost point temperature to derive relative humidity with respect to ice RHi, we find that cloud fractions above 0.5 are observed with RHi values greater than ∼50% and that an increase in cloudiness of 10% corresponds to an increase in relative humidity of ∼10%. From ∼16.5 to 19 km in Fig. 13, climatological-mean temperature and frost point increase with height; as noted above, cirrus decks are observed only in regions of ascent in which temperatures are well below the climatological mean.

One potential issue with the Manus frost point temperature profile is the difficulty of accurately measuring moisture content at such low temperatures using standard radiosonde instruments (Guichard et al. 2000). As a rough calibration of the Manus profiles, a mean frost point temperature profile from Biak, Indonesia (1°S, 136°E), is plotted in gray in both panels of Fig. 13. The Biak profile is based on measurements by frost point hygrometers during eight radiosonde ascents between 8 and 16 January 2006 (Yang et al. 2008). A mean frost point profile was calculated for Manus using only the closest radiosondes in time to those at Biak (not shown). In the TTL, Manus’ frost point temperatures were, on average, 1°C lower than Biak’s, while the mean temperatures at the two sites were virtually identical. This suggests that the Manus frost point temperatures may be biased low. If 1°C is added to the Manus mean frost point profile in Fig. 13, RHi for the 50% threshold line becomes ∼65% for altitudes below 16 km.

The 30-day time–height section from Manus shown in Fig. 14 illustrates the relationship between cloud occurrence in the troposphere and wave activity in the lower stratosphere. The temperature anomalies (colors and contours; calculated with respect to the mean at each level during the 30-day period) reveal a number of downward-propagating wavelike fluctuations with varying descent rates, which are largely confined to altitudes above the cold point tropopause. Cloud layers identified by CALIPSO passes near Manus are plotted as vertical columns of black dots scaled to represent cloud fraction at each level. There is a clear tendency for cloudier conditions in the TTL when negative temperature anomalies are present at the tropopause (21 March–1 April; 10–18 April) than when positive anomalies are present there (1–9 April).

Results of previous studies suggest that cirrus layers are not only optically thin but often geometrically thin as well. McFarquhar et al. (2000) observed mean geometric thicknesses near 0.47 km for cirrus near the tropopause during the Central Equatorial Pacific Experiment (CEPEX), and Winker and Trepte (1998) reported layer thicknesses between 0.25 km and 1 km for TTL cirrus identified by a space shuttle–borne lidar. It is clear from Fig. 14 that some cirrus observed near Manus were geometrically thin—five of the CALIPSO passes identified cirrus layers less than 1 km deep. These observations could be interpreted as suggesting that TTL cirrus are associated with vertically propagating waves with short vertical wavelengths; however, inspection of Fig. 14 reveals that the cold layers in which TTL cirrus tends to be enhanced are not as thin as the cloud layers themselves. Smaller-scale processes that could give rise to thin layers of cirrus are the subject of ongoing research. Results of a recent modeling study suggest that once a thin TTL cirrus cloud layer has formed, radiative heating within the cloud induces a cloud-scale circulation with inflow at the bottom of the cirrus cloud and outflow near the top. If the inflowing air is sufficiently moist, this cloud-induced circulation provides a mechanism for the cloud to become self-sustaining (Durran et al. 2009; Dinh et al. 2010).

6. Summary and conclusions

In this study, data from the CALIPSO satellite have been used to investigate the mechanisms involved in the formation of TTL cirrus and the characteristics of the local and planetary-scale environment that influence its frequency of occurrence. As a means of summarizing these results, the first empirical orthogonal function (EOF) of the filtered CALIPSO TTL cirrus index, which explains ∼15% of the variance in the TTL cirrus field, is shown in the top panel of Fig. 15, with the filtered ERA 100-hPa wind field regressed onto the associated principal component (PC) time series. The bottom panel of Fig. 15 shows an equatorial cross section in which the filtered CALIPSO height-dependent cloud index and ERA winds are regressed onto the same PC time series. Vertical velocities have been multiplied by the horizontally averaged static stability (Γd − Γ, where Γ is the lapse rate) and thus are indicative of the diabatic heating rate with sign reversed (Dima and Wallace 2007). Based on the results presented in this paper, we can draw the following conclusions:

  • The leading mode of variability in the TTL cirrus field on time scales less than 80 days includes a cirrus signature centered above the Maritime Continent that strongly resembles a convectively-induced Kelvin–Rossby wave couplet (top panel of Fig. 15; Gill 1980). Cirrus extending along the equator to the east of the center of this feature marks the Kelvin wave, while the Rossby wave is associated with the bands of enhanced cirrus flanking the equator at around 10° latitude to the west. The Rossby-wave signature is somewhat clearer in the temperature fields in Fig. 5 and in the first EOF of a TTL cirrus field calculated on a 5° latitude × 10° longitude grid (not shown; see section 6 of the companion paper for further discussion of Kelvin and Rossby features). The 100-hPa wind field associated with EOF1 (Fig. 15, top) exhibits divergence out of the center of the cirrus feature over the Maritime Continent, with westerlies associated with the Kelvin-wave signature to the east of it and easterlies associated with the Rossby-wave signature to the west.
  • The cloud and wind fields in the equatorial cross section associated with EOF1 (Fig. 15, bottom) are strongly suggestive of a Kelvin wave signature in the TTL above the Maritime Continent. The cloud feature, which corresponds to the region of cold anomalies (see Figs. 7 and 9), tilts eastward with height, and there is a transition from ascent and westerlies to the east of the axis of maximum cloud fraction to descent and easterly winds to the west of it. The coincidence of ascent and westerly winds, in quadrature with the temperature and cloudiness anomalies, is in agreement with the theoretically derived structure of equatorially trapped Kelvin waves described by Holton and Lindzen (1968), Wallace and Kousky (1968), and Holton (1979).
  • TTL cirrus tends to be concentrated within planetary-scale regions of anomalously low temperature (Figs. 3 and 5); temperatures near the 100-hPa level above the warm pool region are, on average, 4°–6°C lower when cirrus are present (Fig. 11; Virts 2009). Comparison with the ARM data reveals that cirrus occurrence within the TTL and temperature are negatively correlated even at the lowest observed temperatures in radiosonde data at Manus. We find that TTL cirrus is more strongly correlated with temperature than any other variable considered in this study. We interpret this result as indicating that low temperature and high cloud fraction reflect a recent history of planetary-scale ascent.
  • The annual-mean TTL cirrus signature does not resemble the mean precipitation field (Fig. 3), but TTL cirrus over the Maritime Continent and western Pacific is significantly correlated with convective activity approximately 30° of longitude (∼3000 km) to the west. This longitudinal displacement is evident in the third and fifth panels of Fig. 6, which shows the TTL cirrus signature centered ∼30° east of the precipitation and 300-hPa ascent in the reference box, and in Fig. 7, which demonstrates that the convective cloud signature is displaced westward of the perturbation in the TTL. Our results suggest that convection exerts an indirect influence on TTL cirrus formation, by forcing vertically propagating planetary waves, rather than a direct influence through the spreading of cumulonimbus anvils.
  • The characteristics of TTL cirrus clouds and the planetary-scale environments with which they are associated are not uniform throughout the tropics. TTL cirrus above equatorial Africa and South America does not exhibit a Kelvin–Rossby wave signature (Fig. 4), nor is it accompanied by a Kelvin–Rossby wave signature in the 100-hPa temperature or wind fields (Fig. 5). It does not exhibit an eastward tilt with height or any obvious relation to convective activity in the Western Hemisphere (Fig. 7), but it does appear to vary out of phase with deep convection and TTL cirrus over the Maritime Continent (Fig. 15). In the companion paper, we show that TTL cirrus over equatorial Africa and South America is significantly correlated with the MJO.

The conclusion of this study that TTL cirrus is formed largely in situ in regions of planetary-scale ascent is in contrast to studies such as those by Spang et al. (2002) and Massie et al. (2002), who used back trajectories to relate tropical cirrus [identified by the Cryogenic Infrared Spectrometers and Telescopes for the Atmosphere (CRISTA) and the Halogen Occultation Experiment (HALOE), respectively] to deep convection and found that most tropical cirrus originated in cumulonimbus anvils. We would expect to find anvil cirrus near the altitude of maximum convective detrainment, which lies near the 250-hPa level (∼11 km; Highwood and Hoskins 1998). Our results indicate that convective activity may be directly involved in the formation of lower-altitude cirrus, as evidenced by the statistically significant correlations between 300-hPa ω and nearby clouds with bases between 9 and 12 km (Virts 2009), but not the higher-altitude cirrus with which we are concerned. Further reconciliation of these results could be accomplished through a more comprehensive trajectory analysis.

The use of TTL cirrus data from the CALIPSO satellite has proven both illuminating and challenging. On one hand, this dataset provides an unprecedented amount of information regarding the occurrence and characteristics of TTL cirrus, including details about layer reflectivity and optical depth that have not been examined in this study. Its orbital paths provide a geographic sampling of the tropical belt that is adequate for many purposes. Using the CALIPSO dataset has allowed us to identify factors involved in TTL cirrus formation and to analyze the relationship of the TTL cirrus field to planetary wave activity in the TTL.

The chief limitation of the CALIPSO dataset has proven to be its irregular spatial and temporal sampling. A given point along one of CALIPSO’s orbital paths is sampled once every 16 days; consecutive daytime or nighttime passes cross the equator ∼25° longitude apart with a time difference of ∼1.6 h. As a result, most tropical clouds are sampled only once by the CALIPSO satellite, and a meaningful, representative daily map of TTL cirrus occurrence cannot be obtained from it. These sampling characteristics limit our ability to describe the characteristics of the TTL cirrus field and relate it to other atmospheric variables—for example, the analysis of the fields as sampled by CALIPSO does not capture the full strength of the relationship with the fields in the ERA-Interim data (Virts 2009). The quality of the results presented here demonstrates that even with these limitations, the CALIPSO dataset is a valuable tool for the study of the tropical atmosphere, including TTL cirrus.

Acknowledgments

The authors thank Todd Mitchell for his assistance, Holger Vömel for providing the Biak radiosonde profiles, and Robert Wood and three anonymous reviewers for their helpful recommendations. This work was supported by the National Science Foundation under Grant 0812802. QF is in part supported by DOE Grant DE-FG02-09ER64769 and by NASA Grant NNX08AF66G; TA is in part supported by DOE Grant DE-PS02-08ER08-23.

REFERENCES

  • Adler, R. F., and Coauthors, 2003: The version-2 Global Precipitation Climatology Project (GPCP) monthly precipitation analysis (1979–present). J. Hydrometeor., 4 , 11471167.

    • Search Google Scholar
    • Export Citation
  • Boehm, M. T., , and J. Verlinde, 2000: Stratospheric influence on upper tropospheric tropical cirrus. Geophys. Res. Lett., 27 , 32093212.

    • Search Google Scholar
    • Export Citation
  • Buck, A. L., 1981: New equations for computing vapor pressure and enhancement factor. J. Appl. Meteor., 20 , 15271532.

  • Clark, H. L., 2005: Longitudinal variability of water vapor and cirrus in the tropical tropopause layer. J. Geophys. Res., 110 , D07107. doi:10.1029/2004JD004943.

    • Search Google Scholar
    • Export Citation
  • Dima, I. M., , and J. M. Wallace, 2007: Structure of the annual-mean equatorial planetary waves in the ERA-40 reanalyses. J. Atmos. Sci., 64 , 28622880.

    • Search Google Scholar
    • Export Citation
  • Dinh, T. P., , D. R. Durran, , and T. P. Ackerman, 2010: Maintenance of tropical tropopause layer cirrus. J. Geophys. Res., 115 , D02104. doi:10.1029/2009JD012735.

    • Search Google Scholar
    • Export Citation
  • Durran, D. R., , T. Dinh, , M. Ammerman, , and T. Ackerman, 2009: The mesoscale dynamics of thin tropical tropopause cirrus. J. Atmos. Sci., 66 , 28592873.

    • Search Google Scholar
    • Export Citation
  • Fu, Q., , Y. Hu, , and Q. Yang, 2007: Identifying the top of the tropical tropopause layer from vertical mass flux analysis and CALIPSO lidar cloud observations. Geophys. Res. Lett., 34 , L14813. doi:10.1029/2007GL030099.

    • Search Google Scholar
    • Export Citation
  • Fueglistaler, S., , A. E. Dessler, , T. J. Dunkerton, , I. Folkins, , Q. Fu, , and P. W. Mote, 2009: Tropical tropopause layer. Rev. Geophys., 47 , RG1004. doi:10.1029/2008RG000267.

    • Search Google Scholar
    • Export Citation
  • Fujiwara, M., and Coauthors, 2009: Cirrus observations in the tropical tropopause layer over the western Pacific. J. Geophys. Res., 114 , D09304. doi:10.1029/2008JD011040.

    • Search Google Scholar
    • Export Citation
  • Gill, A. E., 1980: Some simple solutions for heat-induced tropical circulation. Quart. J. Roy. Meteor. Soc., 106 , 447462.

  • Guichard, F., , D. Parsons, , and E. Miller, 2000: Thermodynamic and radiative impact of the correction of sounding humidity bias in the tropics. J. Climate, 13 , 36113624.

    • Search Google Scholar
    • Export Citation
  • Highwood, E. J., , and B. J. Hoskins, 1998: The tropical tropopause. Quart. J. Roy. Meteor. Soc., 124 , 15791604.

  • Holton, J. R., 1979: An Introduction to Dynamic Meteorology. Academic Press, 391 pp.

  • Holton, J. R., , and R. S. Lindzen, 1968: A note on “Kelvin” waves in the atmosphere. Mon. Wea. Rev., 96 , 385386.

  • Holton, J. R., , P. H. Haynes, , M. E. McIntyre, , A. R. Douglass, , R. B. Rood, , and L. Pfister, 1995: Stratosphere–troposphere exchange. Rev. Geophys., 33 , 403439.

    • Search Google Scholar
    • Export Citation
  • Immler, F., , K. Krüger, , M. Fujiwara, , G. Verver, , M. Rex, , and O. Schrems, 2008: Correlation between equatorial Kelvin waves and the occurrence of extremely thin ice clouds at the tropical tropopause. Atmos. Chem. Phys., 8 , 40194026.

    • Search Google Scholar
    • Export Citation
  • Jensen, E. J., , O. B. Toon, , H. B. Selkirk, , J. D. Spinhirne, , and M. R. Schoeberl, 1996: On the formation and persistence of subvisible cirrus clouds near the tropical tropopause. J. Geophys. Res., 101 , 2136121375.

    • Search Google Scholar
    • Export Citation
  • Jensen, E. J., , L. Pfister, , T-P. Bui, , P. Lawson, , and D. Baumgardner, 2010: Ice nucleation and cloud microphysical properties in tropical tropopause layer cirrus. Atmos. Chem. Phys., 10 , 13691384.

    • Search Google Scholar
    • Export Citation
  • Leith, C. E., 1973: The standard error of time-averaged estimates of climatic means. J. Appl. Meteor., 12 , 10661069.

  • Massie, S., , A. Gettelman, , W. Randel, , and D. Baumgardner, 2002: Distribution of tropical cirrus in relation to convection. J. Geophys. Res., 107 , 4591. doi:10.1029/2001JD001293.

    • 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
  • Matsuno, T., 1966: Quasi-geostrophic motions in the equatorial area. J. Meteor. Soc. Japan, 44 , 2542.

  • McFarquhar, G. M., , A. J. Heymsfield, , J. Spinhirne, , and B. Hart, 2000: Thin and subvisual tropopause tropical cirrus: Observations and radiative impacts. J. Atmos. Sci., 57 , 18411853.

    • Search Google Scholar
    • Export Citation
  • Robinson, G. D., 1980: The transport of minor atmospheric constituents between troposphere and stratosphere. Quart. J. Roy. Meteor. Soc., 106 , 227253.

    • Search Google Scholar
    • Export Citation
  • Roundy, P. E., 2008: Analysis of convectively coupled Kelvin waves in the Indian Ocean MJO. J. Atmos. Sci., 65 , 13421359.

  • Spang, R., , G. Eidmann, , M. Riese, , D. Offermann, , P. Preusse, , L. Pfister, , and P-H. Wang, 2002: CRISTA observations of cirrus clouds around the tropopause. J. Geophys. Res., 107 , 8174. doi:10.1029/2001JD000698.

    • Search Google Scholar
    • Export Citation
  • Virts, K. S., 2009: Cirrus in the tropical tropopause transition layer: Formation mechanisms and influence of local and planetary-scale environment. M.S. thesis, Department of Atmospheric Sciences, University of Washington, 99 pp.

  • Virts, K. S., , and J. M. Wallace, 2010: Annual, interannual, and intraseasonal variability of tropical tropopause transition layer cirrus. J. Atmos. Sci., 67 , 30973112.

    • Search Google Scholar
    • Export Citation
  • Wallace, J. M., , and V. E. Kousky, 1968: Observational evidence of Kelvin waves in the tropical stratosphere. J. Atmos. Sci., 25 , 900907.

    • Search Google Scholar
    • Export Citation
  • Wang, P-H., , P. Minnis, , M. P. McCormick, , G. S. Kent, , and K. M. Skeens, 1996: A 6-year climatology of cloud occurrence frequency from Stratospheric Aerosol and Gas Experiment II observations (1985–1990). J. Geophys. Res., 101 , 2940729430.

    • Search Google Scholar
    • Export Citation
  • Webster, P. J., 1972: Response of the tropical atmosphere to local, steady forcing. Mon. Wea. Rev., 100 , 518541.

  • Wheeler, M., , G. N. Kiladis, , and P. J. Webster, 2000: Large-scale dynamical fields associated with convectively coupled equatorial waves. J. Atmos. Sci., 57 , 613640.

    • Search Google Scholar
    • Export Citation
  • Winker, D. M., , and C. R. Trepte, 1998: Laminar cirrus observed near the tropical tropopause by LITE. Geophys. Res. Lett., 25 , 33513354.

    • Search Google Scholar
    • Export Citation
  • Winker, D. M., , W. H. Hunt, , and M. J. McGill, 2007: Initial performance assessment of CALIOP. Geophys. Res. Lett., 34 , L19803. doi:10.1029/2007GL030135.

    • Search Google Scholar
    • Export Citation
  • Yang, Q., , Q. Fu, , J. Austin, , A. Gettelman, , F. Li, , and H. Vömel, 2008: Observationally derived and general circulation model simulated tropical stratospheric upward mass fluxes. J. Geophys. Res., 113 , D00B07. doi:10.1029/2008JD009945.

    • Search Google Scholar
    • Export Citation
  • Yin, X., , A. Gruber, , and P. Arkin, 2004: Comparison of the GPCP and CMAP merged gauge–satellite monthly precipitation products for the period 1979–2001. J. Hydrometeor., 5 , 12071222.

    • Search Google Scholar
    • Export Citation

Fig. 1.
Fig. 1.

(a) Seven-day running mean cold point temperature at Manus (gray; inverted scale at left) and CALIPSO cloud fraction with base above 15 km within a 10° latitude × 10° longitude region centered on Manus (black; scale at right) from June 2006 to June 2009. (b) As in (a), but an 80-day high-pass Lanczos filter has been applied to both variables.

Citation: Journal of the Atmospheric Sciences 67, 10; 10.1175/2010JAS3412.1

Fig. 2.
Fig. 2.

Cross sections of (a) zonal-mean, annual-mean cloud fraction [contour interval (CI) = 0.05], (b) ERA-Interim relative humidity (CI = 10%) and (c) vertical velocity (CI = 1 mm s−1; zero contour in red). Cloud fractions calculated from CALIPSO profiles from June 2006 to June 2009. Latitude resolution is 5°.

Citation: Journal of the Atmospheric Sciences 67, 10; 10.1175/2010JAS3412.1

Fig. 3.
Fig. 3.

(a) Mean 100-hPa temperature (contours at … −80, −79, −76, −73, … °C; outermost dark contour is −79°C), (b) cloud fraction with base above 15 km (CI = 0.05), (c) cloud fraction in the 11–12-km layer (without reference to cloud-base altitude; CI = 0.05), and (d) GPCP precipitation (CI = 1.5 mm day−1). Latitude resolution is 5°, with values every 2.5°; longitude resolution is 10°.

Citation: Journal of the Atmospheric Sciences 67, 10; 10.1175/2010JAS3412.1

Fig. 4.
Fig. 4.

Correlations between filtered ERA 100-hPa temperatures within black reference boxes and filtered CALIPSO TTL cirrus index throughout the tropics (CI = 0.1). The strongest correlation is indicated in lower right corner of each panel.

Citation: Journal of the Atmospheric Sciences 67, 10; 10.1175/2010JAS3412.1

Fig. 5.
Fig. 5.

Correlations between filtered CALIPSO TTL cirrus index within black reference boxes and filtered 5° latitude × 5° longitude ERA 100-hPa temperatures (colors and contours) and winds (vectors) throughout the tropics (CI for temperature–cloud fraction correlations = 0.1). The strongest correlation is indicated in lower right corner of each panel.

Citation: Journal of the Atmospheric Sciences 67, 10; 10.1175/2010JAS3412.1

Fig. 6.
Fig. 6.

As in Fig. 4, but with filtered variables within the Maritime Continent reference box as indicated in the upper left corner of each panel.

Citation: Journal of the Atmospheric Sciences 67, 10; 10.1175/2010JAS3412.1

Fig. 7.
Fig. 7.

Correlations between filtered ERA 100-hPa temperatures within 10° latitude × 10° longitude reference boxes (centered on white stars) and filtered CALIPSO height-dependent tropical cloud index from 5°S to 5°N (CI = 0.1). The strongest correlation is indicated in lower right corner of each panel.

Citation: Journal of the Atmospheric Sciences 67, 10; 10.1175/2010JAS3412.1

Fig. 8.
Fig. 8.

As in Fig. 7, but for filtered ERA 300-hPa ω in Maritime Continent reference box.

Citation: Journal of the Atmospheric Sciences 67, 10; 10.1175/2010JAS3412.1

Fig. 9.
Fig. 9.

Correlations between filtered CALIPSO TTL cirrus index within 10° latitude × 10° longitude reference box (black vertical lines) and filtered ERA temperatures (colors and contours) from 2.5°S to 2.5°N. ERA longitudinal resolution is 5°; CI = 0.1.

Citation: Journal of the Atmospheric Sciences 67, 10; 10.1175/2010JAS3412.1

Fig. 10.
Fig. 10.

(left) Correlation and (right) regression coefficients between CALIPSO-derived cloud fraction and ARM radiosonde temperatures at Manus as functions of height. Correlations indicating 95% significance level are plotted as dotted lines in the left panel.

Citation: Journal of the Atmospheric Sciences 67, 10; 10.1175/2010JAS3412.1

Fig. 11.
Fig. 11.

(left) Composite “clear” (red) and “cloudy” (blue) temperature profiles at Manus as a function of height, constructed from CALIPSO cloud profiles and radiosonde temperatures, along with Monte Carlo–derived 2-standard deviation confidence intervals (red and blue dashed; see text for methodology). (right) As at left, but composite “warm” (red) and “cold” (blue) cloud fraction profiles based on radiosonde cold point temperatures from Manus.

Citation: Journal of the Atmospheric Sciences 67, 10; 10.1175/2010JAS3412.1

Fig. 12.
Fig. 12.

(left) Scatterplot of mean CALIPSO TTL cirrus index within regularly spaced bins of ERA 100-hPa temperatures (°C, both variables averaged over 10° latitude × 10° longitude regions over 7-day periods). (right) Scatterplot of mean cloud fraction within regularly spaced bins of ARM temperatures [°C, both variables averaged over 15–16 km (triangles), 16–17 km (dots), and 17–18 km (asterisks), and cloud fraction averaged over 5° latitude × 5° longitude region centered on Manus]. Error bars indicate 10th and 90th percentiles of cloud fraction for each interval and are plotted for the middle layer only in the right panel.

Citation: Journal of the Atmospheric Sciences 67, 10; 10.1175/2010JAS3412.1

Fig. 13.
Fig. 13.

(left) Threshold temperatures below which mean cloud occurrence greater than 10% (pluses), 30% (dots), 50% (asterisks), and 70% (circles) are observed (cloud fractions are calculated over a 5° latitude × 5° longitude region centered on Manus). Mean temperature (T; dash–dotted) and frost point temperature (Tf; dashed) from Manus and mean frost point temperature, measured by frost point hygrometers, from Biak (Tfb; gray dashed; see text for details) are also plotted. (right) As at left, but best linear fit to each cloud fraction threshold is plotted.

Citation: Journal of the Atmospheric Sciences 67, 10; 10.1175/2010JAS3412.1

Fig. 14.
Fig. 14.

Time–height section of radiosonde-derived temperature anomalies from Manus (colors and contours; CI = 2°C) and CALIPSO-derived cloud fraction within successive 200-m layers, calculated for a 5° latitude × 5° longitude region centered on Manus (vertical black dots, scaled by cloud fraction observed). Cold point tropopause height plotted as thick black line.

Citation: Journal of the Atmospheric Sciences 67, 10; 10.1175/2010JAS3412.1

Fig. 15.
Fig. 15.

(top) Filtered CALIPSO TTL cirrus index and ERA 100-hPa winds throughout tropics regressed onto PC time series of the first EOF of filtered TTL cirrus index (EOF calculated using only points with mean cloud fraction ≥0.1). The variance explained is indicated in the lower right corner. (bottom) Cross section of filtered CALIPSO height-dependent tropical cloud index (from 5°S to 5°N) and ERA winds (plotted every 15° and at selected vertical levels) regressed onto PC time series from the top panel. The vertical component of arrows represents the heating rate (see text); wind vector scaling is arbitrary, and the vertical scale of the gray arrows is smaller than that for the black arrows.

Citation: Journal of the Atmospheric Sciences 67, 10; 10.1175/2010JAS3412.1

Save