• Albrecht, B. A., 1989: Aerosols, cloud microphysics, and fractional cloudiness. Science, 245 , 12271230.

  • Albrecht, B. A., , C. S. Bretherton, , D. Johnson, , W. H. Schubert, , and A. S. Frisch, 1995: The Atlantic Stratocumulus Transition Experiment—ASTEX. Bull. Amer. Meteor. Soc, 76 , 889904.

    • Search Google Scholar
    • Export Citation
  • Benkovitz, C. M., , C. M. Berkovitz, , R. C. Easter, , S. Nemesure, , R. Wagener, , and S. E. Schwartz, 1994: Sulfate over the North Atlantic and adjacent continental regions: Evaluation for October and November 1986 using a three-dimensional model driven by observation-derived meteorology. J. Geophys. Res, 99 , 2072520756.

    • Search Google Scholar
    • Export Citation
  • Benkovitz, C. M., , M. A. Miller, , S. E. Schwartz, , and O-U. Kwon, 2001a: The influence of cut-off lows on sulfate burdens over the North Atlantic during April 1987. Preprints, A Millennium Symp. on Atmospheric Chemistry, Albuquerque, NM, Amer. Meteor. Soc., 170–174.

    • Search Google Scholar
    • Export Citation
  • Benkovitz, C. M., , M. A. Miller, , S. E. Schwartz, , and O-U. Kwon, 2001b: Dynamical influences on the distribution and loading of SO2 and sulfate over North America, the North Atlantic and Europe in April 1987. Geochem. Geophys. Geosyst, 2 .doi:10.1029/2000GC000129.

    • Search Google Scholar
    • Export Citation
  • Brenguier, J-L., , H. Pawlowska, , L. Schuller, , R. Preusker, , J. Fischer, , and Y. Fouquart, 2000: Radiative properties of boundary layer clouds: Droplet effective radius versus number concentration. J. Atmos. Sci, 57 , 803821.

    • Search Google Scholar
    • Export Citation
  • Cahalan, R. F., , W. Ridgway, , W. J. Wiscombe, , T. L. Bell, , and J. B. Snider, 1994: The albedo of fractal stratocumulus clouds. J. Atmos. Sci, 51 , 24342455.

    • Search Google Scholar
    • Export Citation
  • Chin, M., and Coauthors, 2002: Tropospheric aerosol optical thickness from the GOCART model and comparisons with satellite and sun photometer measurements. J. Atmos. Sci, 59 , 461483.

    • Search Google Scholar
    • Export Citation
  • Coakley Jr., J. A., , and F. P. Bretherton, 1982: Cloud cover from high-resolution scanner data: Detecting and allowing for partially filled fields of view. J. Geophys. Res, 87 , 49174932.

    • Search Google Scholar
    • Export Citation
  • Greenwald, T. J., , and S. A. Christopher, 2000: The GOES I–M imagers: New tools for studying microphysical properties of boundary layer stratiform clouds. Bull. Amer. Meteor. Soc, 81 , 26072619.

    • Search Google Scholar
    • Export Citation
  • Greenwald, T. J., , S. A. Christopher, , and J. Chou, 1997: Cloud liquid water path comparisons from solar reflectance and passive microwave satellite measurements: Assessment of sub-field-of-view cloud effects for microwave retrievals. J. Geophys. Res, 102 , 1958519596.

    • Search Google Scholar
    • Export Citation
  • Han, Q., , W. B. Rossow, , and A. A. Lacis, 1994: Near-global survey of effective droplet radii in liquid water clouds using ISCCP data. J. Climate, 7 , 465497.

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

    • Search Google Scholar
    • Export Citation
  • Han, Q., , W. B. Rossow, , J. Chou, , and R. M. Welch, 1998: Global variation of column droplet concentration in low-level clouds. Geophys. Res. Lett, 25 , 14191422.

    • Search Google Scholar
    • Export Citation
  • Harshvardhan, , B. A. Wielicki, , and K. M. Ginger, 1994: The interpretation of remotely sensed cloud properties from a model parameterization perspective. J. Climate, 7 , 19871998.

    • Search Google Scholar
    • Export Citation
  • Harshvardhan, , S. E. Schwartz, , C. M. Benkovitz, , and G. Guo, 2002: Aerosol influence on cloud microphysics examined by satellite measurements and chemical transport modeling. J. Atmos. Sci, 59 , 714725.

    • Search Google Scholar
    • Export Citation
  • Haywood, J., , and O. Boucher, 2000: Estimates of the direct and indirect radiative forcing due to tropospheric aerosols: A review. Rev. Geophys, 38 , 513543.

    • Search Google Scholar
    • Export Citation
  • Jiang, H., , G. Feingold, , and W. R. Cotton, 2002: Simulations of aerosol-cloud-dynamical feedbacks resulting from entrainment of aerosol into the marine boundary layer during the Atlantic Stratocumulus Transition Experiment. J. Geophys. Res.,107, 4813, doi:10.1029/2001JD001502.

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

  • Kaufman, Y. J., , D. Tanré, , and O. Boucher, 2002: A satellite view of aerosols in the climate system. Nature, 419 , 215223.

  • Kawamoto, K., , T. Nakajima, , and T. Y. Nakajima, 2001: A global determination of cloud microphysics with AVHRR remote sensing. J. Climate, 14 , 20542068.

    • Search Google Scholar
    • Export Citation
  • King, M. D., 1987: Determination of the scaled optical thickness of clouds from reflected solar radiation measurements. J. Atmos. Sci, 44 , 17341751.

    • Search Google Scholar
    • Export Citation
  • Loeb, N. G., , and J. A. Coakley Jr., 1998: Inference of marine stratus cloud optical depths from satellite measurements: Does 1D theory apply? J. Climate, 11 , 215233.

    • Search Google Scholar
    • Export Citation
  • Martin, G. M., , D. W. Johnson, , and A. Spice, 1994: The measurement and parameterization of effective radius of droplets in warm stratocumulus clouds. J. Atmos. Sci, 51 , 18231842.

    • Search Google Scholar
    • Export Citation
  • Nakajima, T., , and M. D. King, 1990: Determination of the optical thickness and effective particle radius of clouds from reflected solar radiation measurements. Part I: Theory. J. Atmos. Sci, 47 , 18781893.

    • Search Google Scholar
    • Export Citation
  • Nakajima, T., , A. Higurashi, , K. Kawamoto, , and J. E. Penner, 2001: A possible correlation between satellite-derived cloud and aerosol microphysical parameters. Geophys. Res. Lett, 28 , 11711174.

    • Search Google Scholar
    • Export Citation
  • Nakajima, T. Y., , and T. Nakajima, 1995: Wide-area determination of cloud microphysical properties from NOAA AVHRR measurements for FIRE and ASTEX regions. J. Atmos. Sci, 52 , 40434059.

    • Search Google Scholar
    • Export Citation
  • Platnick, S., , and S. Twomey, 1994: Determining the susceptibility of cloud albedo to changes in droplet concentration with the Advanced Very High Resolution Radiometer. J. Appl. Meteor, 33 , 334347.

    • Search Google Scholar
    • Export Citation
  • Platnick, S., , M. D. King, , S. A. Ackerman, , W. P. Menzel, , B. A. Baum, , J. C. Riédi, , and R. A. Frey, 2003: The MODIS cloud products: Algorithms and examples from Terra. IEEE Trans. Geosci. Remote Sens, 41 , 459473.

    • Search Google Scholar
    • Export Citation
  • Ramaswamy, V., and Coauthors, 2001: Radiative forcing of climate change. Climate Change 2001: The Scientific Basis, J. T. Houghton et al., Eds., Cambridge University Press, 349–416.

    • Search Google Scholar
    • Export Citation
  • Rosenfeld, D., , and I. M. Lensky, 1998: Satellite-based insights into precipitation formation processes in continental and maritime convective clouds. Bull. Amer. Meteor. Soc, 79 , 24572476.

    • Search Google Scholar
    • Export Citation
  • Schwartz, S. E., , Harshvardhan, , and C. M. Benkovitz, 2002: Influence of anthropogenic aerosol on cloud optical depth and albedo shown by satellite measurements and chemical transport modeling. Proc. Natl. Acad. Sci. USA, 99 , 17841789.

    • Search Google Scholar
    • Export Citation
  • Szczodrak, M., , P. H. Austin, , and P. B. Krummel, 2001: Variability of optical depth and effective radius in marine stratocumulus clouds. J. Atmos. Sci, 58 , 29122926.

    • Search Google Scholar
    • Export Citation
  • Twomey, S., 1977: The influence of pollution on the shortwave albedo of clouds. J. Atmos. Sci, 34 , 11491152.

  • Twomey, S., , and C. F. Bohren, 1980: Simple approximations for the calculations of absorption in clouds. J. Atmos. Sci, 37 , 20862094.

  • View in gallery

    Simulation of reflected radiances in AVHRR channels 1 and 3 as a function of optical thickness and effective radius for θ = 40°, θ0 = 60°, and φ = 50° (from Nakajima and Nakajima 1995). Straight lines from the origin denote possible locations of radiance pairs for partially cloudy pixels

  • View in gallery

    Channel-1 and channel-4 AVHRR images of the eastern North Atlantic Ocean at dates and times indicated. Study areas are marked with letters and numbers

  • View in gallery

    (a) Scatterplot of AVHRR channel-1 reflectance (%) and channel-4 radiating temperature (K) on 3 Apr 1987 for region B in Fig. 2; (b) scatterplot of channel-4 local standard deviation (mW m−2 sr−1 cm) for the same region and time period; (c) retrieved cloud optical depth for all pixels having reflectance in excess of 10% plotted vs the inferred cloud-top temperature (gray dots) and the subset of pixels that also have channel-4 local standard deviation less than 0.5 mW m−2 sr−1 cm; and (d) as in (c), but for cloud effective droplet radius (μm)

  • View in gallery

    Histograms of pixel frequency distribution of (a) cloud optical depth, (b) effective droplet radius, (c) liquid water path, and (d) the liquid water path frequency for 3 Apr 1987, region B. The shaded boxes represent all pixels satisfying the visible reflectance threshold criterion, and the clear outlined boxes represent all pixels that satisfy the standard deviation criterion as well

  • View in gallery

    Scatterplot of retrieved effective droplet radius (μm) vs retrieved cloud optical depth for 3 Apr 1987, region B. Gray dots represent all pixels passing the 10% visible reflectance threshold, while black dots represent pixels that, in addition, have channel-4 local standard deviation less than 0.5 mW m−2 sr−1 cm

  • View in gallery

    As in Fig. 3, but for 5 Apr 1987, region 8

  • View in gallery

    As in Fig. 4, but for 5 Apr 1987, region 8

  • View in gallery

    As in Fig. 5, but for 5 Apr 1987, region 8

  • View in gallery

    Liquid water path (kg m−2) calculated from Eq. (4) plotted vs inferred cloud-top temperature for all pixels in the labeled regions on 3 Apr 1987 that satisfy both the reflectance threshold and standard deviation criteria. Lines show adiabatic LWP from an assumed cloud base. The dashed line in (g) corresponds to the distinct higher cloud deck

  • View in gallery

    As in Fig. 9, but for the labeled regions on 5 Apr 1987

  • View in gallery

    Mean droplet number concentration (cm−3) plotted vs inferred cloud-top temperature for all pixels in the labeled regions on 3 Apr 1987 that satisfy both the reflectance threshold and standard deviation criteria

  • View in gallery

    As in Fig. 11, but for the labeled regions on 5 Apr 1987

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 8 8 0
PDF Downloads 7 7 0

Remotely Sensed Microphysical and Thermodynamic Properties of Nonuniform Water Cloud Fields

View More View Less
  • 1 Department of Earth and Atmospheric Sciences, Purdue University, West Lafayette, Indiana
  • | 2 JAXA Earth Observation Research and Application Center, Tokyo, Japan
© Get Permissions
Full access

Abstract

Visible and near-infrared reflected radiances have been used to estimate the cloud optical depth and effective radius of cloud-filled global area coverage (GAC) pixels from the Advanced Very High Resolution Radiometer (AVHRR) for two cases in the North Atlantic Ocean. One is representative of clouds having low concentrations of cloud condensation nuclei (CCN), while the other is an example of maritime clouds forming in continental air, in this case, intruding from Europe around a cutoff low pressure system. It is shown that an estimate of the cloud drop concentration can be obtained from remotely sensed cloud radiative properties and standard meteorological analyses. These concentrations show very clearly the influence of enhanced CCN on cloud microphysics. However, conclusions regarding the indirect radiative effect of aerosol on cloud must wait for the development of a framework for analyzing changes in cloud liquid water path (LWP). It is shown that estimates of LWP are greatly influenced by the scheme that is used to identify cloudy pixels at the AVHRR GAC resolution. Application of a very strict thermal channel spatial coherence criterion for identifying cloud-filled pixels yields mean LWP estimates for cloudy pixels alone that are 40%–75% higher than mean LWP estimates for the much larger sample of possibly cloudy pixels identified by a reflectance threshold criterion.

Corresponding author address: Dr. Harshvardhan, Dept. of Earth and Atmospheric Sciences, Purdue University, 550 Stadium Mall Dr., West Lafayette, IN 47907-2051. Email: harshvar@purdue.edu

Abstract

Visible and near-infrared reflected radiances have been used to estimate the cloud optical depth and effective radius of cloud-filled global area coverage (GAC) pixels from the Advanced Very High Resolution Radiometer (AVHRR) for two cases in the North Atlantic Ocean. One is representative of clouds having low concentrations of cloud condensation nuclei (CCN), while the other is an example of maritime clouds forming in continental air, in this case, intruding from Europe around a cutoff low pressure system. It is shown that an estimate of the cloud drop concentration can be obtained from remotely sensed cloud radiative properties and standard meteorological analyses. These concentrations show very clearly the influence of enhanced CCN on cloud microphysics. However, conclusions regarding the indirect radiative effect of aerosol on cloud must wait for the development of a framework for analyzing changes in cloud liquid water path (LWP). It is shown that estimates of LWP are greatly influenced by the scheme that is used to identify cloudy pixels at the AVHRR GAC resolution. Application of a very strict thermal channel spatial coherence criterion for identifying cloud-filled pixels yields mean LWP estimates for cloudy pixels alone that are 40%–75% higher than mean LWP estimates for the much larger sample of possibly cloudy pixels identified by a reflectance threshold criterion.

Corresponding author address: Dr. Harshvardhan, Dept. of Earth and Atmospheric Sciences, Purdue University, 550 Stadium Mall Dr., West Lafayette, IN 47907-2051. Email: harshvar@purdue.edu

1. Introduction

The retrieval of some measure of the optical depth and cloud drop size of water clouds from satellite-borne instruments using the simultaneous measurement of reflected radiances at nonabsorbing and absorbing wavelengths in the water vapor windows is now fairly standard (Platnick and Twomey 1994; Han et al. 1994; Nakajima and Nakajima 1995; Rosenfeld and Lensky 1998; Greenwald and Christopher 2000; Kawamoto et al. 2001). All retrieval techniques utilize forward calculations of reflectance based on one-dimensional plane parallel models. It is therefore not surprising that the techniques are beset with errors when the pixel under consideration is only partially filled with cloud or the cloud field has significant geometric structure and internal inhomogeneity (Han et al. 1994; Loeb and Coakley 1998).

The problem is compounded when the optical depth and effective radius are combined to yield an estimate of the pixel-averaged liquid water path (LWP; Han et al. 1995; Greenwald et al. 1997). The above-mentioned studies have estimated that effective radius biases are probably not more than 1–2 μm for overcast pixels, although optical depths could be significantly underestimated when a partially filled pixel is assumed to be completely cloud covered. The significance of biases in LWP has not been articulated, although its estimate is crucial to any discussion of the indirect cloud radiative forcing of aerosol, which in turn has been identified as a radiative forcing with a very low level of scientific understanding and a large range of uncertainty in the formulation of climate models (Haywood and Boucher 2000; Ramaswamy et al. 2001).

The indirect effect of aerosol on cloud radiative forcing involves both changes in cloud-effective radius (and concomitant change in drop concentration), the so-called first indirect effect (Twomey 1977), as well as changes in LWP and cloud lifetime (Albrecht 1989). Moreover, area-integrated radiative forcing is due to all clouds that are present, even clouds in partially filled pixels. Therefore, global surveys of cloud microphysical properties made with the express purpose of elucidating the indirect aerosol effect (Nakajima et al. 2001) need to be interpreted with caution. Here we use two case studies to illustrate the extent to which global satellite retrievals of cloud microphysical properties can be used to estimate the aerosol indirect effect and also to identify a few of the roadblocks on the path to this quest. The analysis is limited to the somewhat coarse spatial resolution of Advanced Very High Resolution Radiometer (AVHRR) global area coverage (GAC).

2. Retrieval theory

An excellent exposition of the theory underlying cloud microphysical property retrievals is given in Nakajima and King (1990). Here we will only present the bare essentials. If a horizontally homogeneous cloud layer is illuminated from above by a parallel beam of radiation of flux, Fo, then the reflected radiance, I(−μ, φ), may be written as (King 1987):
Iμ,ϕμoFoπRτcμ,μoϕ
In the above equation, τc is the total optical depth of the cloud, μo is the cosine of the solar zenith angle, μ is the absolute value of the cosine of the zenith angle of the reflected radiance measured with respect to the positive τ direction (τ increases downward), and φ is the relative azimuth angle between the direction of propagation of the reflected radiation and the incident solar radiation. The reflection function, R(τc; μ, μo, φ) is defined by (1) and has a value of 1.0 when the reflecting medium is a perfect, conservative, isotropic reflector.
At wavelengths for which scattering is conservative, corresponding to a single scattering albedo, ωo = 1, R(τc; μ, μo, φ) increases monotonically with increasing τc. However, when there is absorption and ωo < 1, R reaches a maximum value and then remains constant even as τc is increased indefinitely (e.g., Fig. 1 of King 1987). Physically, all solar energy at that wavelength has either been reflected back upward or absorbed within the cloud, and additional cloud matter added below does not participate in the scattering process. Therefore, for clouds of moderate to large optical depth, reflected radiances at a nonabsorbing wavelength provide information on the optical depth, while a measurement at an absorbing wavelength provides an estimate of the single scattering albedo. At a fixed wavelength, ωo is in turn related to the drop radius. Twomey and Bohren (1980) have shown that a good approximation is given by
ωorκ,
where r is the drop radius and κ is the bulk absorption coefficient at that wavelength, when 2 ≪ 1.0 and the particle radius is large relative to the wavelength. In practice, retrieval schemes do not rely on such approximations but make detailed scattering calculations followed by radiative transfer computations to generate lookup tables (LUTs). For an assumed representative size distribution, one can therefore relate the reflected radiance to an effective radius, re, of a polydisperse water cloud;
i1520-0469-61-21-2574-e3
where n(r, z) is the number size distribution per increment of radius at level z above the cloud base.

Liquid water absorbs in several bands at wavelengths λ > 0.7 μm but is a conservative medium at the shorter visible wavelengths. Therefore, the two wavelengths chosen for retrievals of the optical depth and effective radius are usually in the visible and near-infrared spectral regions, respectively. Since water vapor also absorbs in the near-infrared, the sensing channels are selected such that they are in the water vapor windows at around 1.6, 2.2, and 3.7 μm. Some instruments, such as the Moderate Resolution Imaging Spectroradiometer (MODIS), have all three near-infrared window channels available for cloud property retrievals (Platnick et al. 2003). The AVHRR on operational polar orbiters has only one of the channels at 3.7 μm (channel 3), and information from this channel and the visible channel centered at around 0.64 μm (channel 1) are used in the current and many previous studies.

The 3.7-μm channel suffers from one drawback. There is significant thermal emission, primarily from cloud top, at that wavelength. This component of radiation is removed by utilizing the emission from a thermal channel centered at around 11 μm (channel 4). We have followed the procedure of Kawamoto et al. (2001), using meteorological reanalyses (Kalnay et al. 1996) to furnish the atmospheric profile above cloud for the region and day of interest. The profile for the particular grid box on that day above cloud top estimated from channel-4 radiance is used. The reflected radiances from channels 1 and 3 of AVHRR are then used for the cloud microphysical retrieval from a cloudy pixel. This is done by first generating LUTs created by carrying out a forward calculation for assumed size distributions of cloud drops and vertical distribution. The LUTs used here were generated assuming a lognormal size distribution and vertically uniform cloud properties (Nakajima and Nakajima 1995).

An example of such a forward calculation is shown in Fig. 1, reproduced from the above for cloud overlying a nonreflective surface. Simultaneous measurements in channels 1 and 3 provide τc and re, given by the intersection of isolines of optical depth and effective radius. Since the forward problem assumes a completely cloud-filled pixel of uniform properties, even error-free retrievals will have biases for inhomogeneous cloud cover and particularly for partially filled pixels that are included in an analysis. A very simple illustration is provided by the straight lines superimposed on the isolines in Fig. 1. They represent plausible loci of retrievals of τc and re when the pixel has fractional cloud cover of uniform cloud properties given by the terminus of the lines and the rest of the pixel is clear and nonreflecting. So, for example, a pixel half-filled with cloud of τc = 64 and re = 8 μm will have retrieved properties of τc ≈ 7 and re ≈ 15 μm for the particular sun–satellite configuration. Another example can be found in Fig. 3 of Han et al. (1994).

The large bias in effective radius illustrated above points to the need for some consistency in defining cloudy pixels. If the fraction of ambiguous retrievals is small, then satellite-derived cloud microphysical properties may be used to provide global estimates of the cloud indirect forcing attributable to anthropogenic aerosol. The two case studies below explore the matter further. It should be noted that these examples use AVHRR GAC; higher spatial resolution data are expected to reduce ambiguity in cloud screening.

3. Satellite case studies

We have chosen two cases in April 1987 to examine the consequences of the selection of cloudy pixels for any analysis that addresses issues pertaining to changes in cloud microphysical properties, particularly increases in cloud droplet number concentration forced by increasing cloud condensation nuclei (CCN) concentrations. The choice was based on prior work in which we had identified low-level marine clouds that had significantly different drop sizes on different days (Harshvardhan et al. 2002; Schwartz et al. 2002). Over the period 2–8 April 1987, a broad area of high sulfate column burden developed and then dissipated in a region of the North Atlantic west of the British Isles. This was attributed to the incursion of continental air from Europe (Benkovitz et al. 2001a,b).

a. 3 April 1987

The region identified for study is delineated in Figs. 2a and 2b, which show, respectively, AVHRR visible (channel 1) and thermal (channel 4) images of a portion of the National Oceanic and Atmospheric Administration-9 (NOAA-9) orbit. There are extensive low-level clouds, which appear as darker shades of gray in the thermal image. The boxes marked A–H are each 2.5° × 2.5° in latitudinal and longitudinal extent, respectively, and were chosen to coincide with the meteorological analysis grid. The crucial importance of identifying cloudy pixels for analysis will be illustrated with an example from box B.

Figure 3a is a scatterplot of channel-1 reflectance plotted versus the channel-4 radiating temperature for all global area coverage pixels in box B. The nominal horizontal resolution at nadir of GAC pixels is 4 km × 1 km with a 3-km gap between pixels across the scan lines (Han et al. 1994). Pixels of low reflectance and warm radiating temperature are clear ocean and pixels of high reflectance and cold radiating temperatures are cloud filled. However, there is a continuous range of both quantities, and it is necessary to specify rules for selecting cloudy pixels for analyses.

We have chosen to follow the spatial coherence technique of Coakley and Bretherton (1982) to identify cloud-filled pixels. First, we apply a reflectance threshold of 10% to channel 1 and then select pixels that are part of a larger area of nearly uniform radiating temperature. This is done by plotting the local standard deviation of thermal emission in channel 4 versus the local mean emission of a 2 × 2 array of GAC pixels. The resulting scatterplot for box B is shown in Fig. 3b, where mean emission has been converted to radiating temperature for ease of interpretation. Each point on the plot represents four neighboring GAC pixels from Fig. 3a. The scatterplot takes the shape of an arch with the “warm foot” representing pixels that are cloud-free and the “cold foot” identifying cloud-filled pixels. The radiating temperature of the warm foot is an estimate of sea surface temperature (SST). The body of the arch represents pixels that are possibly only partially filled with cloud or are surrounded by pixels of significantly different radiating temperature, and hence altitude, indicating complex cloud-top geometry. It is possible, but unlikely, for partially cloudy pixels to produce low standard deviations as calculated by this method, as neighboring pixels would then have to exhibit nearly identical subpixel scale variability. The points in the arch that are at warmer radiating temperatures represent pixels that probably contain clouds of optical depth lower than those at the cold foot of the arch.

In what follows we have selected two groups of “cloudy” pixels for microphysical retrieval—those that pass the threshold criterion of channel-1 reflectance in excess of 10% and those that, in addition, have channel-4 local standard deviation of less than 0.5 mW m−2 sr−1 cm, that is, the cold foot of the arch. Cloud optical depth and effective radius were retrieved using the procedure of Kawamoto et al. (2001), and the results are presented as scatterplots in Figs. 3c and 3d, respectively. The radiating temperature given by the warm foot of the arch in Fig. 3b is assumed to be the surface temperature required for the retrieval. It is illustrative to plot the retrieved quantities versus cloud-top temperature, which is also one of the retrieved products of the analysis. In the figures, black dots are for pixels that satisfy both the reflectance threshold and channel-4 standard deviation criteria, while gray dots represent all pixels satisfying the reflectance criterion alone. Points colder than 270 K representing overlying high clouds have been removed.

Figure 3c shows two fairly well-defined cloud layers in which individual cloud-filled pixels have optical depths ranging from about 5 to over 60 and effective radii from 8 to 16 μm. The gray area in the optical depth plot is occupied by pixels that are bright enough to be at least partially cloud filled or are part of cloudy regions with significant geometrical features. The three-dimensional nature of the radiative transfer in the latter case makes the application of the LUTs used here inappropriate. Pixels with small retrieved optical depth and warm radiating temperatures are almost certainly only partially filled with cloud but could be thin semi-transparent layers. The cloud-screening technique used here is probably too conservative and excludes many pixels that are essentially cloud covered. Of 2645 pixels that passed the reflection threshold, only 457 also passed the channel-4 standard deviation criterion.

We can look to Fig. 3d to provide further information. The gray points at cloud-top temperatures colder than about 276 K with retrieved optical depths up to 20 and effective radii in the 8–12-μm range could represent either cloud-filled or partly cloudy pixels. However, points that are warmer and have smaller optical depths are almost certainly partly filled. Some of the thinnest clouds have drops of retrieved effective radii in excess of 12 μm. Since cloud drops tend to grow as air parcels ascend (Brenguier et al. 2000), pixels with small optical depth and large effective radii have grossly biased retrieved properties and represent points on the straight lines shown in Fig. 1.

The representation of all clouds in box B and by extension, for any global satellite-based method, is therefore influenced by the cloud-screening process. The degree to which information could potentially be distorted is shown in Figs. 4a and 4b, in which histograms of pixel frequency are shown for pixels passing the reflectance threshold alone (gray) and for those passing the thermal standard deviation criterion as well (black outline). The latter are, of course, a subset of the former.

As expected, the median optical depth shifts to smaller values (from 16.7 to 8.1) when all cloudy pixels are considered, relative to the stricter selection criterion of cloud-filled pixels. What is somewhat surprising is that the median effective radius also shifts to smaller radii (from 11.8 to 10.7 μm). This would not occur if the majority of pixels that do not meet the strict selection criterion were filled partially by cloud of moderate-to-high optical depth. There are two possible explanations.

The cloud optical depth of the cloudy portion of partly filled pixels tends to be small, and the drop size is also correspondingly small. The loci of such partially filled pixels would still yield small optical depths but also small drop radii with considerable uncertainty attached to the retrieval. One possible path is shown in Fig. 1, connecting τc = 4 and re = 4 μm with the origin. Another possibility is that the pixels are not partially cloudy but are simply covered by thinner clouds of great variability. Figure 5 shows the data from Figs. 3c and 3d as a scatterplot of effective droplet radius versus cloud optical depth. The numerous gray dots trending toward smaller optical depths and smaller droplet radii are probably thin clouds. These pixels did not pass the channel-4 standard deviation criterion because of excessive heterogeneity at GAC spatial scales. The gray dots showing small optical depth and large cloud drop radius almost certainly represent partially filled pixels. Although not shown here, simply relaxing the standard deviation criterion does not separate thin nonuniform clouds from partially cloudy pixels. The addition of nonuniformly emitting pixels in this case results in the inclusion of retrieved cloud droplet radii that are predominantly smaller than for more uniformly cloud-covered pixels, hence the shift in median drop size to slightly smaller values.

The retrieved values of τc and re are combined to obtain LWP, and the pixel distribution is shown in Fig. 4c. We have chosen to follow Szczodrak et al. (2001) and calculate LWP from
i1520-0469-61-21-2574-e4
where ρw is the density of liquid water, τc is the total cloud optical depth, as before, and re(H) is the effective radius at cloud top identified by height H above cloud base. The standard assumption has been made that the extinction efficiency Qext = 2. Our choice of (4) is based on the sense that satellite-retrieved properties are representative of the top portions of the cloud (Nakajima and King 1990).

The significant shift in the LWP distribution shown in Fig. 4c can be explained by inspection of (4). LWP is linearly proportional to the product of cloud optical depth, τc, and effective radius, re. Owing to the convexity of the relationship between the reflectance function and optical depth (King 1987; Cahalan et al. 1994), the retrieved optical depth from smeared pixels (such as partly cloudy pixels assumed to be cloud covered) is reduced from the value of the optical depth of the cloudy portion alone by a ratio that is far greater than the cloud fraction. For instance, for the example shown in section 2, cloud fraction of 0.5 reduces the retrieved optical depth from 64 to about 7. The increase in retrieved effective radius from 8 to 15 μm does not compensate for this effect when (4) is used to calculate LWP. The inferred LWP would therefore invariably be biased low unless the cloudy portion of the pixel had a low optical depth in the range where the R-versus-τc relationship is less convex and nearly linear. Such biases have been reported earlier, for example, by Harshvardhan et al. (1994). It may be noted that the linear range may not be important because the combination of low optical depth of the cloudy portion and additionally low cloud fraction would yield a pixel reflectance that would probably not pass the reflectance threshold and thus would be excluded from any analysis of cloud properties. Additionally, as pointed out earlier, many of the pixels included by relaxing the channel-4 standard deviation criterion are thinner nonuniform clouds, and the retrieved droplet radius is small, as is the LWP calculated from (4).

Although there is a considerable increase in the proportion of pixels having the smallest inferred LWP, a better measure of biases introduced by cloud-screening criteria is the LWP-weighted histogram shown in Fig. 4d. This would provide a measure of biases in the total amount of liquid water in the grid box, which is the relevant quantity when comparing retrievals with model generated LWP or retrievals at coarser resolution obtained directly from microwave measurements (Greenwald et al. 1997). The fraction of LWP in the lowest bin (0–0.05 kg m−2) is only 0.03 when completely cloud-covered pixels alone are considered, but increases to 0.20 when all potentially cloudy pixels are analyzed. Although all pixels do not have the same areal footprint, the estimates are representative of the potential undercount of total LWP because one would expect a random spatial distribution with respect to the satellite track. For the case shown in Fig. 4d, the mean LWP of the cloudy portion of the grid box is 0.077 kg m−2 when all pixels passing a reflectance threshold are counted as being cloudy but is 0.130 kg m−2 when the stricter spatial coherence screening is used to identify pixels as cloudy.

Of greater importance for detecting an indirect radiative forcing is the problem of ascertaining whether there are differences in LWP over areas of O(100 km2), perhaps forced by meteorological conditions or CCN concentration. For example, in their global analysis, Nakajima et al. (2001) show that there is no trend in LWP with increasing aerosol column number density. This result would stand on firmer ground if it was certain that the distribution of LWP was not altered by changes in CCN. There is the possibility, for instance, that any cloud-screening algorithm could select more cloudy pixels in one case than in another, there by reducing the mean LWP for the region, such as is shown in Fig. 4d. The hope and expectation is that a sufficiently large sample will average things out but that will lead to an erroneous conclusion if there are systematic differences caused by identified physical processes.

b. 5 April 1987

The second case chosen here is shown in Figs. 2c and 2d. The cloud system is somewhat different than on 3 April, but the boxed area has considerable low-level cloudiness. Figures 6a and 6b show scatterplots of all GAC pixels in region 8, which is a 2.5° × 2.5° latitude– longitude grid box. As explained earlier, we can choose cloudy pixels from Fig. 6b to satisfy the stringent spatial coherence criterion or simply the reflectance threshold from Fig. 6a.

The results of the retrievals are shown in Figs. 6c and 6d. We have excluded all pixels indicating cloud top colder than 265 K; 1779 pixels passed the reflection threshold, and 503 of them satisfied the channel-4 standard deviation criterion. The cloud layers are more widely distributed in height but the general observations made earlier still hold. The range of drop effective radii is now from 5 to 11 μm, and there is a distinct increase in retrieved effective radii for warmer pixels that are probably only partially filled. There are also a few cloud-filled pixels at low levels that could be part of another cloud system.

The distribution of microphysical properties is shown in Figs. 7a–d. There is the expected shift to lower optical depths and also lower drop radius in spite of the additional larger drop sizes added to the population when the cloud-screening criteria are loosened. This case is for higher CCN and the most notable feature is the absence of cloud drops larger than 11 μm. However, the most telling difference between this case and the one on 3 April 1987 is the rather modest redistribution of LWP in Fig. 7d. Our interpretation is that the cloud system is more evenly spread out with proportionately fewer partially cloudy pixels. The contribution by the smallest bin is 3% when completely filled pixels are considered and 11% when all cloudy pixels are included. The mean LWP of pixels passing the reflectance threshold is 0.108 and 0.149 kg m−2 when in addition the thermal spatial coherence criterion is applied to the cloud-screening process. Figure 8 shows the scatterplot of effective droplet radius versus cloud optical depth for 5 April 1987, region 8. A distinction between partially cloudy pixels and thin nonuniform clouds is clearer in this particular case, but one can still not separate overcast and partially cloudy pixels unambiguously. The addition of thinner clouds to the population also results in the addition of pixels with smaller retrieved droplet radius and, the median again shifts to smaller drop sizes. However, in both cases, the LWP distribution shifts to lower values.

4. Cloud drop concentration

Although satellite retrievals only provide τc and re in a pixel-averaged sense, reasonable assumptions can be made to carry the analysis further, apart from the standard computation of LWP. Of great significance is an estimate of the cloud drop concentration, N(cm−3),
i1520-0469-61-21-2574-e5
since an increase in N(z) resulting from an increase in CCN is the physical basis of the indirect effect. Han et al. (1998) and Nakajima et al. (2001) have derived, on a global basis, the cloud column droplet concentration, Nc(cm−2),
i1520-0469-61-21-2574-e6
from which one can estimate N if H, the geometrical thickness of the cloud, is known and assuming that N is constant with height (Szczodrak et al. 2001). The key step in obtaining Nc from τc and re is the application of a relationship between re and the volume mean radius, rυ, which is given by
i1520-0469-61-21-2574-e7
with w(z) being the liquid water content in the cloud, which is in turn related to the liquid water path through
i1520-0469-61-21-2574-e8
In situ measurements suggest a relationship of the form k = r3υ/r3e, with k ranging from 0.67 ± 0.07 for continental air masses to k = 0.80 ± 0.07 for maritime clouds (Martin et al. 1994). The actual value of k depends on the shape of the cloud drop size distribution. We have adopted a constant value of k = 0.735 for this study. Assuming that the column-integrated effective radius is five-sixths of the value at cloud top [the basis for Eq. (4)], the column-integrated volume mean radius used to calculate Nc is [e.g., Eq. (3) of Harshvardhan et al. 2002]
i1520-0469-61-21-2574-e9

The final step is to estimate the cloud thickness, H, an impossible task using reflected radiances alone. In the absence of a meteorological sounding, other assumptions need to be made. The model we use here is that of air parcels in the grid box rising from a common base. If the process is adiabatic, then the condensed liquid water content w(z) is very nearly a linear function of height from the base (Brengiuer et al. 2000) and can be calculated from a thermodynamic diagram. It follows from (8) that the adiabatic liquid water path is a quadratic, increasing with height from the base. A scatterplot of pixel LWP versus height should then be constrained by this condition, with each value of LWP less than the adiabatic value to some extent.

Figures 9 and 10 show scatterplots of LWP versus cloud-top temperature for all eight grid boxes on 3 and 5 April, respectively. Only pixels satisfying the visible reflectance and thermal standard deviation criteria have been plotted, so we are minimizing the biases alluded to earlier. Although cloud-top temperature is being used as a surrogate for height, it should be kept in mind that there is a horizontal temperature gradient across the grid box. We propose that, within the large area of clouds, there are some pixels that have undergone adiabatic lifting from the base. Using surface air temperature and surface pressure information from the meteorological reanalysis for the particular grid box and time, we have computed adiabatic LWP from several assumed cloud-base levels on a thermodynamic diagram. The lines shown in Figs. 9 and 10 correspond to realizations that approximately touch the highest value of LWP at temperatures near cloud top, representing the few pixels that are free of nonadiabatic processes such as entrainment and mixing with drier air. The temperature (and corresponding height) at the zero LWP intercept is the location of the cloud base.

One can now compute Nc from τc and re(H) with the assumption of a constant factor k and then proceed to estimate the cloud drop number concentration, N, as simply Nc/H. This procedure has been applied to all the plotted data. Note that box G of 3 April appears to have two distinct low-level cloud layers, and we have assigned two separate locations for the base. Results for the two days are shown in Figs. 11 and 12, respectively. The gray dots in Fig. 11 (box G) represent the lower layer. The dramatic difference in N on the two days is quite evident. The range and values are consistent with those measured by Brenguier et al. (2000) on two days during the second Aerosol Characterization Experiment (ACE-2), one with pure marine air and the other with contamination by pollution from Europe. The above procedure can, in principle, be used to generate maps of cloud drop concentration of marine water clouds that are not obscured by overlying higher cloud layers, which would aid in the identification of the aerosol indirect effect if information on the aerosol concentration was available, for example, from an aerosol model (Benkovitz et al. 1994; Chin et al. 2002) or remote sensing (Kaufman et al. 2002).

5. Conclusions

We have shown through a simple analysis and two case studies using AVHRR GAC that it is crucial to examine changes in cloud LWP when investigating the indirect forcing of cloud by aerosol. Moreover, LWP retrieved from remotely sensed cloud microphysics is subject to significant errors due to cloud inhomogeneity and patchiness, which may not be eliminated simply by averaging over a number of realizations. The errors invariably lead to underestimates in LWP, but more importantly, the degree of error depends on the degree of inhomogeneity of cloud cover. This raises the intriguing possibility that a causal relationship between CCN and cloud macrostructure would introduce a hitherto unnoticed complication in the study of aerosol indirect effect.

An example of just such a situation can be found in the Albrecht et al. (1995) survey of the Atlantic Stratocumulus Transition Experiment (ASTEX). Their Fig. 7, a visible satellite image obtained on 16 June 1992, shows clouds with uniform tops forming in continental air in the ASTEX study region, while clouds forming in maritime air have considerable vertical structure and horizontal patchiness. Profiles made by aircraft on that date indicate that LWP in continental air was about twice that in the maritime air, which together with smaller cloud droplets resulted in continental clouds appearing to be more reflective. This difference is actually amplified by the pronounced structure of the maritime clouds, which would have even lower albedo at the scale of the entire cloud field. Another example is an ASTEX simulation by Jiang et al. (2002), who used an eddy-resolving model to simulate stratocumulus evolution for different CCN concentrations. They found that higher CCN inhibited cumulus convection, resulting in lower LWP relative to the clean case. Such dynamical feedbacks can result in pronounced differences in cloud field macrostructure. We therefore propose that in addition to the usual analysis of LWP in the typical aerosol cloud-forcing study, an effort should be made to note some measure of cloud inhomogeneity as given by, for example, distribution of LWP.

This study has also presented a method for inferring the cloud droplet concentration using only remotely sensed estimates of optical depth and effective radius and standard meteorological analyses by making some reasonable thermodynamic assumptions regarding the vertical structure of cloud layers. The method can, in principle, be used to generate global fields of cloud drop concentration over the oceans from currently available satellite data such as that from the MODIS instrument on Terra and Aqua (Platnick et al. 2003).

Acknowledgments

We thank Jim Coakley for advice and encouragement and an anonymous reviewer for very helpful suggestions. This work was supported by NASA Grant NAG5-7727 and ONR Grant N00014-02-1-0182. Portions of the study were conducted while the lead author was a visiting senior research scientist at the Goddard Earth Sciences and Technology Center, UMBC.

REFERENCES

  • Albrecht, B. A., 1989: Aerosols, cloud microphysics, and fractional cloudiness. Science, 245 , 12271230.

  • Albrecht, B. A., , C. S. Bretherton, , D. Johnson, , W. H. Schubert, , and A. S. Frisch, 1995: The Atlantic Stratocumulus Transition Experiment—ASTEX. Bull. Amer. Meteor. Soc, 76 , 889904.

    • Search Google Scholar
    • Export Citation
  • Benkovitz, C. M., , C. M. Berkovitz, , R. C. Easter, , S. Nemesure, , R. Wagener, , and S. E. Schwartz, 1994: Sulfate over the North Atlantic and adjacent continental regions: Evaluation for October and November 1986 using a three-dimensional model driven by observation-derived meteorology. J. Geophys. Res, 99 , 2072520756.

    • Search Google Scholar
    • Export Citation
  • Benkovitz, C. M., , M. A. Miller, , S. E. Schwartz, , and O-U. Kwon, 2001a: The influence of cut-off lows on sulfate burdens over the North Atlantic during April 1987. Preprints, A Millennium Symp. on Atmospheric Chemistry, Albuquerque, NM, Amer. Meteor. Soc., 170–174.

    • Search Google Scholar
    • Export Citation
  • Benkovitz, C. M., , M. A. Miller, , S. E. Schwartz, , and O-U. Kwon, 2001b: Dynamical influences on the distribution and loading of SO2 and sulfate over North America, the North Atlantic and Europe in April 1987. Geochem. Geophys. Geosyst, 2 .doi:10.1029/2000GC000129.

    • Search Google Scholar
    • Export Citation
  • Brenguier, J-L., , H. Pawlowska, , L. Schuller, , R. Preusker, , J. Fischer, , and Y. Fouquart, 2000: Radiative properties of boundary layer clouds: Droplet effective radius versus number concentration. J. Atmos. Sci, 57 , 803821.

    • Search Google Scholar
    • Export Citation
  • Cahalan, R. F., , W. Ridgway, , W. J. Wiscombe, , T. L. Bell, , and J. B. Snider, 1994: The albedo of fractal stratocumulus clouds. J. Atmos. Sci, 51 , 24342455.

    • Search Google Scholar
    • Export Citation
  • Chin, M., and Coauthors, 2002: Tropospheric aerosol optical thickness from the GOCART model and comparisons with satellite and sun photometer measurements. J. Atmos. Sci, 59 , 461483.

    • Search Google Scholar
    • Export Citation
  • Coakley Jr., J. A., , and F. P. Bretherton, 1982: Cloud cover from high-resolution scanner data: Detecting and allowing for partially filled fields of view. J. Geophys. Res, 87 , 49174932.

    • Search Google Scholar
    • Export Citation
  • Greenwald, T. J., , and S. A. Christopher, 2000: The GOES I–M imagers: New tools for studying microphysical properties of boundary layer stratiform clouds. Bull. Amer. Meteor. Soc, 81 , 26072619.

    • Search Google Scholar
    • Export Citation
  • Greenwald, T. J., , S. A. Christopher, , and J. Chou, 1997: Cloud liquid water path comparisons from solar reflectance and passive microwave satellite measurements: Assessment of sub-field-of-view cloud effects for microwave retrievals. J. Geophys. Res, 102 , 1958519596.

    • Search Google Scholar
    • Export Citation
  • Han, Q., , W. B. Rossow, , and A. A. Lacis, 1994: Near-global survey of effective droplet radii in liquid water clouds using ISCCP data. J. Climate, 7 , 465497.

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

    • Search Google Scholar
    • Export Citation
  • Han, Q., , W. B. Rossow, , J. Chou, , and R. M. Welch, 1998: Global variation of column droplet concentration in low-level clouds. Geophys. Res. Lett, 25 , 14191422.

    • Search Google Scholar
    • Export Citation
  • Harshvardhan, , B. A. Wielicki, , and K. M. Ginger, 1994: The interpretation of remotely sensed cloud properties from a model parameterization perspective. J. Climate, 7 , 19871998.

    • Search Google Scholar
    • Export Citation
  • Harshvardhan, , S. E. Schwartz, , C. M. Benkovitz, , and G. Guo, 2002: Aerosol influence on cloud microphysics examined by satellite measurements and chemical transport modeling. J. Atmos. Sci, 59 , 714725.

    • Search Google Scholar
    • Export Citation
  • Haywood, J., , and O. Boucher, 2000: Estimates of the direct and indirect radiative forcing due to tropospheric aerosols: A review. Rev. Geophys, 38 , 513543.

    • Search Google Scholar
    • Export Citation
  • Jiang, H., , G. Feingold, , and W. R. Cotton, 2002: Simulations of aerosol-cloud-dynamical feedbacks resulting from entrainment of aerosol into the marine boundary layer during the Atlantic Stratocumulus Transition Experiment. J. Geophys. Res.,107, 4813, doi:10.1029/2001JD001502.

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

  • Kaufman, Y. J., , D. Tanré, , and O. Boucher, 2002: A satellite view of aerosols in the climate system. Nature, 419 , 215223.

  • Kawamoto, K., , T. Nakajima, , and T. Y. Nakajima, 2001: A global determination of cloud microphysics with AVHRR remote sensing. J. Climate, 14 , 20542068.

    • Search Google Scholar
    • Export Citation
  • King, M. D., 1987: Determination of the scaled optical thickness of clouds from reflected solar radiation measurements. J. Atmos. Sci, 44 , 17341751.

    • Search Google Scholar
    • Export Citation
  • Loeb, N. G., , and J. A. Coakley Jr., 1998: Inference of marine stratus cloud optical depths from satellite measurements: Does 1D theory apply? J. Climate, 11 , 215233.

    • Search Google Scholar
    • Export Citation
  • Martin, G. M., , D. W. Johnson, , and A. Spice, 1994: The measurement and parameterization of effective radius of droplets in warm stratocumulus clouds. J. Atmos. Sci, 51 , 18231842.

    • Search Google Scholar
    • Export Citation
  • Nakajima, T., , and M. D. King, 1990: Determination of the optical thickness and effective particle radius of clouds from reflected solar radiation measurements. Part I: Theory. J. Atmos. Sci, 47 , 18781893.

    • Search Google Scholar
    • Export Citation
  • Nakajima, T., , A. Higurashi, , K. Kawamoto, , and J. E. Penner, 2001: A possible correlation between satellite-derived cloud and aerosol microphysical parameters. Geophys. Res. Lett, 28 , 11711174.

    • Search Google Scholar
    • Export Citation
  • Nakajima, T. Y., , and T. Nakajima, 1995: Wide-area determination of cloud microphysical properties from NOAA AVHRR measurements for FIRE and ASTEX regions. J. Atmos. Sci, 52 , 40434059.

    • Search Google Scholar
    • Export Citation
  • Platnick, S., , and S. Twomey, 1994: Determining the susceptibility of cloud albedo to changes in droplet concentration with the Advanced Very High Resolution Radiometer. J. Appl. Meteor, 33 , 334347.

    • Search Google Scholar
    • Export Citation
  • Platnick, S., , M. D. King, , S. A. Ackerman, , W. P. Menzel, , B. A. Baum, , J. C. Riédi, , and R. A. Frey, 2003: The MODIS cloud products: Algorithms and examples from Terra. IEEE Trans. Geosci. Remote Sens, 41 , 459473.

    • Search Google Scholar
    • Export Citation
  • Ramaswamy, V., and Coauthors, 2001: Radiative forcing of climate change. Climate Change 2001: The Scientific Basis, J. T. Houghton et al., Eds., Cambridge University Press, 349–416.

    • Search Google Scholar
    • Export Citation
  • Rosenfeld, D., , and I. M. Lensky, 1998: Satellite-based insights into precipitation formation processes in continental and maritime convective clouds. Bull. Amer. Meteor. Soc, 79 , 24572476.

    • Search Google Scholar
    • Export Citation
  • Schwartz, S. E., , Harshvardhan, , and C. M. Benkovitz, 2002: Influence of anthropogenic aerosol on cloud optical depth and albedo shown by satellite measurements and chemical transport modeling. Proc. Natl. Acad. Sci. USA, 99 , 17841789.

    • Search Google Scholar
    • Export Citation
  • Szczodrak, M., , P. H. Austin, , and P. B. Krummel, 2001: Variability of optical depth and effective radius in marine stratocumulus clouds. J. Atmos. Sci, 58 , 29122926.

    • Search Google Scholar
    • Export Citation
  • Twomey, S., 1977: The influence of pollution on the shortwave albedo of clouds. J. Atmos. Sci, 34 , 11491152.

  • Twomey, S., , and C. F. Bohren, 1980: Simple approximations for the calculations of absorption in clouds. J. Atmos. Sci, 37 , 20862094.

Fig. 1.
Fig. 1.

Simulation of reflected radiances in AVHRR channels 1 and 3 as a function of optical thickness and effective radius for θ = 40°, θ0 = 60°, and φ = 50° (from Nakajima and Nakajima 1995). Straight lines from the origin denote possible locations of radiance pairs for partially cloudy pixels

Citation: Journal of the Atmospheric Sciences 61, 21; 10.1175/JAS3301.1

Fig. 2.
Fig. 2.

Channel-1 and channel-4 AVHRR images of the eastern North Atlantic Ocean at dates and times indicated. Study areas are marked with letters and numbers

Citation: Journal of the Atmospheric Sciences 61, 21; 10.1175/JAS3301.1

Fig. 3.
Fig. 3.

(a) Scatterplot of AVHRR channel-1 reflectance (%) and channel-4 radiating temperature (K) on 3 Apr 1987 for region B in Fig. 2; (b) scatterplot of channel-4 local standard deviation (mW m−2 sr−1 cm) for the same region and time period; (c) retrieved cloud optical depth for all pixels having reflectance in excess of 10% plotted vs the inferred cloud-top temperature (gray dots) and the subset of pixels that also have channel-4 local standard deviation less than 0.5 mW m−2 sr−1 cm; and (d) as in (c), but for cloud effective droplet radius (μm)

Citation: Journal of the Atmospheric Sciences 61, 21; 10.1175/JAS3301.1

Fig. 4.
Fig. 4.

Histograms of pixel frequency distribution of (a) cloud optical depth, (b) effective droplet radius, (c) liquid water path, and (d) the liquid water path frequency for 3 Apr 1987, region B. The shaded boxes represent all pixels satisfying the visible reflectance threshold criterion, and the clear outlined boxes represent all pixels that satisfy the standard deviation criterion as well

Citation: Journal of the Atmospheric Sciences 61, 21; 10.1175/JAS3301.1

Fig. 5.
Fig. 5.

Scatterplot of retrieved effective droplet radius (μm) vs retrieved cloud optical depth for 3 Apr 1987, region B. Gray dots represent all pixels passing the 10% visible reflectance threshold, while black dots represent pixels that, in addition, have channel-4 local standard deviation less than 0.5 mW m−2 sr−1 cm

Citation: Journal of the Atmospheric Sciences 61, 21; 10.1175/JAS3301.1

Fig. 6.
Fig. 6.

As in Fig. 3, but for 5 Apr 1987, region 8

Citation: Journal of the Atmospheric Sciences 61, 21; 10.1175/JAS3301.1

Fig. 7.
Fig. 7.

As in Fig. 4, but for 5 Apr 1987, region 8

Citation: Journal of the Atmospheric Sciences 61, 21; 10.1175/JAS3301.1

Fig. 8.
Fig. 8.

As in Fig. 5, but for 5 Apr 1987, region 8

Citation: Journal of the Atmospheric Sciences 61, 21; 10.1175/JAS3301.1

Fig. 9.
Fig. 9.

Liquid water path (kg m−2) calculated from Eq. (4) plotted vs inferred cloud-top temperature for all pixels in the labeled regions on 3 Apr 1987 that satisfy both the reflectance threshold and standard deviation criteria. Lines show adiabatic LWP from an assumed cloud base. The dashed line in (g) corresponds to the distinct higher cloud deck

Citation: Journal of the Atmospheric Sciences 61, 21; 10.1175/JAS3301.1

Fig. 10.
Fig. 10.

As in Fig. 9, but for the labeled regions on 5 Apr 1987

Citation: Journal of the Atmospheric Sciences 61, 21; 10.1175/JAS3301.1

Fig. 11.
Fig. 11.

Mean droplet number concentration (cm−3) plotted vs inferred cloud-top temperature for all pixels in the labeled regions on 3 Apr 1987 that satisfy both the reflectance threshold and standard deviation criteria

Citation: Journal of the Atmospheric Sciences 61, 21; 10.1175/JAS3301.1

Fig. 12.
Fig. 12.

As in Fig. 11, but for the labeled regions on 5 Apr 1987

Citation: Journal of the Atmospheric Sciences 61, 21; 10.1175/JAS3301.1

Save