Auxiliary material for 2014GL061792
Where and when will we observe cloud changes due to climate warming?
H. Chepfer(1), V. Noel (2), D. Winker(3) and M. Chiriaco (4),
(1) LMD/IPSL, Univ. Pierre and Marie Curie, France
(2) LMD/IPSL, CNRS, France
(3) NASA/Langley, USA
(4) LATMOS/IPSL, UVSQ, France
Geophysical Research Letter, 2014
This dataset contains optically thin cloud profiles and optically thick cloud profiles over
seven years (2007-2012) built from CALIPSO-GOCCP SR-histograms (Fig. S01) and
COSP/lidar + climate models (Fig. S03). It also contains High, Mid, Low-level cloud covers
from CALIPSO-GOCCP observations and COSP/lidar + climate models (Fig. S02).
As the goal of this study is to identify where cloud changes associated with climate warming
would be unambiguously observable with space borne lidar, we have selected simple and
robust variables. We use cloud variables that are observed directly by the lidar, rather than
variables which are retrieved indirectly. As a consequence, changes in these variables can
safely be considered as due to either changes in the cloud characteristics (which we are
looking for) or to changes in the attenuated backscatter profile data due to instrumental
changes such as laser wavelength or field of view.
The altitude-intensity histogram of occurrence over 40 vertical levels (z = 480m), is
contained in the CALIPSO-GOCCP dataset (Chepfer et al. 2010), and contains the full
information on the detailed vertical structure and the intensity backscattered by the cloud. The
intensity of the profile is split into 15 scattering ratio (SR) bins. SR = ATB/ATBmol where
ATB is the Attenuated Backscatter profile and ATBmol the molecular Attenuated Backscatter
profile. (Fig. S01a)
The optically thin cloud fraction profile is the ratio between the number of occurrences with
SR>5 divided by the total number of occurrences at each altitude level. (Fig. S01b in blue).
The optically thick cloud fraction profile is the ratio between the number of occurrences with
SR<0.01 (i.e. an arbitrary low value) divided by the total number of occurrences at each
altitude level (Fig. S01b in green). Although these clouds are opaque to the lidar beam, the
plane albedo can be as low as about 25% and diffuse radiation can be transmitted through
them. Since cloud particles are much larger than the CALIPSO laser wavelength (0.532 m),
their extinction efficiency (Qext) in the thermal infrared is approximately half that in the
visible. Thus the maximum visible optical depth penetrated by the laser can be converted to
an equivalent infrared optical depth (ir= vis /2 = 1.25 to 2.5). The corresponding infrared
emissivity  is then = 1 – exp(-ir) = 0.8 to 0.9. Thus the altitude where the cloud becomes
opaque to the lidar describes an altitude of constant cloud longwave emissivity of about = 0.8
(0.9) for water (ice) clouds. As the atmospheric column beneath the lidar attenuation level is
counted as cloudy and contributing towards the optically thick fraction, even if not cloudy,
the "thick cloud fraction" monotonically increases towards the surface.
The altitude intensity histograms have been described in details in Chepfer et al. 2010,
whereas the optically thin and thick cloud fraction profiles are new diagnostics specifically
developed for the current study.
Figure S01: How we built cloud profiles from SR histograms.
First row, left: Average histogram of SR for all latitudes over 2007-2012 as a function of
altitude. right: Corresponding vertical profiles of optically thick (green, SR < 0.01) and
optically thin (blue, SR > 5) cloud fractions.
Second row: Average vertical profiles of optically thick (green) and thin (blue) cloud
fractions in five latitude regions (>60°, 30°-60°, and +/30°) over 2007-2012.
Figure S02:
a-c) High, Mid, Low-level cloud covers from CALIPSO-GOCCP observations and
COSP/lidar + models. As in Fig. 1b, the green area contains seven years (2007-2012) of
observed annual zonal mean anomalies, and the green dotted line represents 3 times this
envelope. The solid lines show the difference between the +4K scenario and the current
climate predicted with CanAM4 (blue) and HadGEM2 (red) models.
Figure S03: Cloud fraction profiles for optically thin (first row) and optically thick
clouds that fully attenuate the laser (second row). Horizontal dashed lines divide altitudes
of low, mid and high-level clouds. The observations (mean values over 2007-2012) are in
grey line.