![Comparison of the DRIFT observations with airborne in situ measurements at different midlatitude sites (data are taken from Fig. 14 of Korolev et al. [2003]). Cloud observations of Peppler 1940 (P40), Borovikov et al. 1963 (B63), Isaac and Schemenauer 1979 (I79), and Korolev et al. [2003] are considered.](https://www.researchgate.net/profile/Patric-Seifert/publication/228607204/figure/fig1/AS:669491520348172@1536630625858/Comparison-of-the-DRIFT-observations-with-airborne-in-situ-measurements-at-different_Q320.jpg)
![Comparison of the temperature dependence of the fraction of ice-containing clouds over the tropics (Cape Verde, [Ansmann et al., 2009]) and the European mid latitude station Leipzig. Thick and thin curves indicate Cape Verde observations and Leipzig observations (3-8.5 km only), respectively. The dashed curves show respective observations after screening the data sets for cloud seeding effects.](https://www.researchgate.net/profile/Patric-Seifert/publication/228607204/figure/fig2/AS:669491520348173@1536630625873/Comparison-of-the-temperature-dependence-of-the-fraction-of-ice-containing-clouds-over_Q320.jpg)
Content uploaded by Patric Seifert
Author content
All content in this area was uploaded by Patric Seifert
Content may be subject to copyright.
Lidar-based profiling of the tropospheric cloud-ice distribution to
study the seeder-feeder mechanism and the role of Saharan dust as
ice nuclei
Patric Seifert1, Albert Ansmann1, Ina Mattis1, Dietrich Althausen1, Matthias Tesche1, Ulla
Wandinger1, Detlef M¨uller1,2 , and Carlos P´
erez3
1Leibniz Institute for Tropospheric Research, Permoserstr. 15, 04318 Leipzig, Germany.
Email: seifert@tropos.de, albert@tropos.de, ina@tropos.de, dietrich@tropos.de, tesche@tropos.de,
ulla@tropos.de, detlef@tropos.de
2Atmospheric Remote Sensing Laboratory, Gwangju Institute of Science and Technology, Republic of
Korea. Email: detlef@gist.ac.kr
3Barcelona Supercomputing Center, Barcelona, Spain. Email: carlos.perez@bsc.es
1. INTRODUCTION
Heterogeneous ice formation in clouds at temperatures
between -40◦C and 0◦C plays a major role in the produc-
tion of precipitation in mid–tropospheric clouds and deter-
mines their radiative properties as well as their life time.
However, the role of ice multiplication effects as ice parti-
cle splintering and cloud seeding, the so–called seeder–
feeder mechanism (Rutledge and Hobbs 1983), to sup-
port the process of ice–formation in the troposphere re-
mains unresolved. Also the initial formation of ice crys-
tals at temperatures between 0◦C and -40◦C is still poorly
understood (Cantrell and Heymsfield 2005). In this tem-
perature range heterogeneous freezing of supercooled
droplets must be initialized by aerosol particles acting as
ice nuclei. In laboratory studies Saharan dust was found
to be a favorable ice nuclei (Field et al. 2006) that is
available in comparably high concentrations in the tropo-
sphere.
This study presents the investigation of heterogeneous
freezing in natural mid-tropospheric clouds by means
of polarization lidar measurements at a midlatitude site
(51.3◦N, 12.2◦E). From a statistics based on obser-
vations of pure water and ice–containing clouds we in-
vestigated the influence of the seeder–feeder mecha-
nism and of Saharan dust on the temperature–distribution
of ice–containing clouds. Information about instruments
and measurements used for the study are introduced in
Sec. 2. Results are presented in Sec. 3. The impact of
cloud–seeding and of Saharan dust on the freezing tem-
perature of clouds is discussed in Sec. 4.
2. INSTRUMENTATION
The results presented in this abstract are based on mea-
surements with polarization lidar. Here, a laser is utilized
to emit pulses of co–polarized light. By detecting the re-
turned light in the co– and cross–polarized direction, the
depolarization ratio can be obtained which is a measure
of the non–sphericity of the atmospheric scatterer. Thus,
spherical liquid water droplets cause no depolarization
whereas non–spherical ice crystals cause significant de-
polarization to the scattered laser light. Hence, the depo-
larization ratio can be used to distinguish between liquid–
phase clouds and clouds that contain ice crystals.
At Leipzig (51.3◦N, 12.2◦E), the vertically pointing three–
wavelength Raman polarization lidar MARTHA (Multi-
wavelength Atmospheric Raman lidar for Temperature,
Humidity, and Aerosol profiling) (Mattis et al. 2004) has
been employed for regular measurements since 1997.
Besides other parameters it provides profiles of the de-
polarization ratio at 532 nm with a primary vertical reso-
lution of 60 m and a temporal resolution of 30 s. 2344
hours of measurements performed with MARTHA be-
tween April 1997 and June 2008 built the basis for the
DRIFT project (Dust–Related Ice Formation in the Tropo-
sphere). In the scope of DRIFT all measurements were
screened for clouds which were then classified as de-
scribed in Seifert et al. (2007). If a cloud layer was
separated by more than 500 m in the vertical and 5 min
in time from another cloud it was classified as a single
cloud case. This ensures the classification of a cloud
according to the meteorological conditions which led to
its formation. Otherwise, long-lasting cirrostratus or alto-
stratus cloud systems would impact a time-of-occurrence
based statistic too strongly. Temperature information for
each cloud case were obtained from regular radiosonde
launches of the German meteorological service at Op-
pin, 30 km northwest of Leipzig. For times not cov-
ered by the radiosondes FNL/GDAS model data of the
U.S. Weather Services National Center of Environmen-
tal Prediction (NCEP) was used (Information available at:
http://www.arl.noaa.gov/archives.php).
Because MARTHA is pointed to the zenith, the signal re-
ceived from ice–containing clouds can be significantly af-
fected by specular reflection caused by horizontally ori-
ented ice crystals. Specular reflection results in the mea-
© Proceedings of the 8th International Symposium on Tropospheric Profiling, ISBN 978-90-6960-233-2
Delft, The Netherlands, October 2009. Editors, A. Apituley, H.W.J. Russchenberg, W.A.A. Monna
S01 - O04 - 1
0
100
200
300
Number of Cases
Ice-ContainingWater Unclear Cases No Cloud Top
68 134
-70 -60 -50 -40 -30 -20 -10 0 10 20
000 0 00 1 3 4 7 21 61 203 230 80 45 0
36 183 195 152 98 89 63 73 47 36 49 13 4 0 0 0 0
29263 4 1 1 024 13 37 58 48 32 3 1
11 28 29 22 16 9 45 2 35 9 713 15 1 05
Cloud Top Temperature [°C]
Figure 1: Numbers of well-defined cases of observed water clouds (blue) and ice-containing clouds (red),
and unclear cases (undefined cloud phase), and cases with undefined cloud top for 5 K temperature inter-
vals.
surement of depolarization ratios that are in the range of
those produced by liquid water droplets. Thus, an un-
ambiguous discrimination between ice and water layers
within a cloud is not possible when a zenith–pointing li-
dar is applied. Therefore it was decided to categorize
all observed clouds only into pure water clouds and ice–
containing clouds which includes both, mixed–phase as
well as pure ice clouds (Seifert et al. 2008).
In order to study the potential role of Saharan dust as
ice nuclei for heterogeneous freezing (Field et al. 2006)
information about columnal dust load at the grid point
of Leipzig was obtained in 12–hourly intervals from the
DREAM (Dust Regional Atmospheric Modeling System)
model system (P´
erez et al. 2006). This data was used to
assign a measure of dust concentration to every observed
cloud.
Table 1: Cloud observation statistics of the DRIFT
dataset.
April 1997 – June 2008 Cases Fraction
All Observed cloud layers 2319
Well-defined cloud layers 1899 100%
Pure water clouds 789 42%
Ice–containing clouds 1110 58%
Clouds with HTop >8 km 774 41%
Clouds with TTop <-40 ◦C 747 39%
Ice–containing clouds <8 km 363 19%
Cloud phase undefined 236
Cloud top undefined 184
3. RESULTS
Table 1 gives an overview to all cloud observations. Out
of 2319 observed cloud layers 1899 were well–defined,
i.e. their phase and cloud top could be determined ac-
curately. In 184 cases the cloud top could not be deter-
mined because of strong attenuation of the laser light. In
236 cases the cloud phase could not be determined. Rea-
sons for that were technical problems with the depolariza-
tion channels or unclear depolarization signatures due to
strong specular reflection or multiple scattering effects.
The distribution of the observed cloud cases with respect
to cloud top temperature is shown in Fig. 1. The dis-
tribution is separated into contributions of water clouds
(blue), ice–containing clouds (red), and undefined clouds
for which either the cloud phase (orange) or the cloud top
(green) could not be determined.
The fraction of ice–containing clouds with respect to all
well–defined clouds of the DRIFT dataset is shown in
Fig. 2 (asterisk). For comparison, results of four previ-
ous studies based on airborne in–situ measurements are
shown (Korolev et al. 2003). All curves agree well. The
slight shift of the DRIFT values to lower temperatures can
be explained by the fact that the airborne in–situ observa-
tions are related to temperature at flight level whereas the
DRIFT results are given as function of cloud top tempera-
ture which is usually the coldest point of a cloud.
4. DISCUSSION
In order to prevent an influence of low boundary layer
clouds as well as of homogeneously nucleated cirrus
clouds all subsequent investigations of the DRIFT data
S01 - O04 - 2
P40
B63
K03
I79
DRIFT97-08
Fraction of Ice-containing Clouds [%]
0
20
40
60
80
100
Temperature [°C]
010-10-20-30-40
Figure 2: Comparison of the DRIFT observations
with airborne in situ measurements at different mid–
latitude sites (data are taken from Fig. 14 of Ko-
rolev et al. [2003]). Cloud observations of Peppler
1940 (P40), Borovikov et al. 1963 (B63), Isaac and
Schemenauer 1979 (I79), and Korolev et al. [2003]
are considered.
set were restricted to clouds that were detected between
3 and 8.5 km height.
4.1. Effect of the seeder–feeder mechanism
The vertical tropospheric distribution of cloud ice is ex-
pected to be in part determined by the seeder–feeder
mechanism (Rutledge and Hobbs 1983). When several
cloud layers occur, ice crystals falling out of the higher,
colder cloud can act as ice nuclei in the lower, warmer
cloud. This causes immediate glaciation of the seeded
cloud that would be too warm for heterogeneous freezing
to occur. The seeding effect therefore alters the radiative
properties as well as the life time of the seeded cloud.
Whereas airborne observations can only probe the atmo-
sphere at flight level, active remote sensing with lidar is
capable of detecting multiple cloud layers. This allows the
observation of the seeder–feeder mechanism. Ansmann
et al. (2009) studied heterogeneous freezing in altocu-
mulus clouds that were observed at Cape Verde (15◦N,
23.5◦W). A strong impact of cloud seeding on the dis-
tribution of ice–containing clouds in dependence of tem-
perature was found. In a seeding–corrected data set all
ice clouds were assigned to be water clouds if they were
obviously seeded by crystals falling out of an ice cloud
above. By applying this approach the fraction of ice–
containing clouds of the Cape Verde dataset decreased
to almost zero at temperatures above -20◦C. The results
of Ansmann et al. (2009) are shown as the thick solid and
dashed curves in Fig. 3. The solid line represents the un-
corrected data set whereas the dashed line shows the dis-
Fraction of Ice-containing Clouds [%]
Temperature [°C]
-30 -25 -20 -15 -10 -5 0
0
20
40
60
80
100
Figure 3: Comparison of the temperature depen-
dence of the fraction of ice–containing clouds over
the tropics (Cape Verde, [Ansmann et al., 2009])
and the European mid latitude station Leipzig. Thick
and thin curves indicate Cape Verde observations
and Leipzig observations (3-8.5 km only), respec-
tively. The dashed curves show respective observa-
tions after screening the data sets for cloud seeding
effects.
tribution of the ice–containing clouds of the data set that
was corrected for the seeder–feeder effect. The same ap-
proach was applied to the DRIFT data set, shown as thin
curves in Fig. 3. Here, an ice–containing cloud was as-
signed to be a water cloud when another ice–containing
cloud was observed within 2 km above it. The difference
between the uncorrected solid curve and the seeding–
corrected dashed curve is smaller compared to the one
of the Cape Verde data set. This is most probably be-
cause of the different meteorological conditions that lead
to the formation of the clouds. Ansmann et al. (2009)
observed predominantly thin layers of altocumulus clouds
that are produced by weakly ascending air masses. An
influence of frontal systems can be excluded in the region
of the tropics. The clouds of the mid latitudinal DRIFT
data set are to a high fraction influenced by meteorolog-
ical processes that go along with the passage of frontal
systems. Stronger updrafts, wind shear, and thus tur-
bulence can be expected during the development of the
midlatitude clouds. As Hobbs and Rangno (1985) pointed
out, a broad drop–size spectrum is needed in order to trig-
ger heterogeneous ice formation at temperatures >-10◦C.
The higher fraction of ice–clouds and the smaller impact
of cloud seeding at mid latitudinal sites compared to the
tropics can therefore be best explained by the more turbu-
lent conditions at mid latitudes which lead to the produc-
tion of broader cloud–droplet spectra.
S01 - O04 - 3
4.2. Role of Saharan dust as ice nuclei
In order to study the effect of Saharan dust on the
heterogeneous–freezing temperature of clouds we used
data of DREAM which, among other parameters, pro-
vides the columnal dust load at 00 and 12 UTC for the
whole time period of the DRIFT dataset. By assigning a
value of modeled columnal dust load to every observed
cloud case we were able to separate clouds that were ob-
served during dusty conditions from rather clean scenar-
ios. The thresholds of dust load for the dust–influenced
cases and for the presumably dust–free cases were set
in such a way that a statistical significant number of cloud
cases was available for both classes. Thus, a cloud case
was categorized as presumably dust–free when the cor-
responding modeled dust load was <0.03 gm−2. At a
columnal dust load ≥0.05 gm−2a cloud was catego-
rized as a dust–influenced cloud. With these thresholds
189 dust–free cloud cases and 124 dust–influenced cloud
cases were found. Figure 4 presents curves of the dis-
tribution of ice–containing clouds in dependence of tem-
perature for the dust–free cases (circles) and the dust–
influenced cases (stars). Vertical bars show the statisti-
cal error. Between temperatures of -20◦C and 0◦C the
curves of the dust–free and dust–influenced clouds are
clearly separated from each other. At all temperatures
above -20◦C the dust–influenced clouds show a higher
fraction of ice–containing clouds compared to the dust-
free clouds. Therefore, the presented findings corroborate
the hypothesis that Saharan dust is an effective ice nuclei
(Field et al. 2006). It should be noted that the curve of
the presumably dust–free cases has a similar shape to
the distribution of the only 90 cloud cases for which the
columnal dust load was exactly 0.0 gm−2.
11 11 16 9 8 32 19 18
26 30 34 29 23 17 24 6
dustload < 0.03 g/m²
dustload ≥ 0.05 g/m²
-40 -35 -30 -25 -20 -15 -10 -5 0
Temperature [°C]
0
20
40
60
80
100
Fraction of Ice-Containing Clouds [%]
Figure 4: Fraction of ice–containing clouds for cases
weakly affected by Saharan dust (circles) in com-
parison to cases strongly affected by Saharan dust
(stars). Vertical bars show the standard error.
Number of cases for each temperature interval are
shown in the top area of the figure. Only clouds be-
tween 3-8.5 km were considered.
REFERENCES
Ansmann, A., M. Tesche, P. Seifert, D. Althausen,
R. Engelmann, J. Fruntke, U. Wandinger, I. Mat-
tis, and D. M ¨
uller (2009). Evolution of the ice phase
in tropical altocumulus: Samum lidar observations
over Cape Verde. Journal of Geophysical Re-
search, in press.
Cantrell, W. and A. Heymsfield (2005, June). Produc-
tion of ice in tropospheric clouds: A review. Bulletin
of the American Meteorological Society 86, 795–
807.
Field, P. R., O. M¨
ohler, P. Connolly, M. Kr¨
amer, R. Cot-
ton, A. J. Heymsfield, H. Saathoff, and M. Schnaiter
(2006, July). Some ice nucleation characteristics
of Asian and Saharan desert dust. Atmospheric
Chemistry & Physics 6, 2991–3006.
Hobbs, P. V. and A. L. Rangno (1985, December). Ice
particle concentrations in clouds. Journal of Atmo-
spheric Sciences 42, 2523–2549.
Korolev, A. V., G. A. Isaac, S. G. Cober, J. W.
Strapp, and J. Hallett (2003, January). Microphysi-
cal characterization of mixed–phase clouds. Quar-
terly Journal of the Royal Meteorological Soci-
ety 129, 39–65.
Mattis, I., A. Ansmann, D. M ¨
uller, U. Wandinger, and
D. Althausen (2004, July). Multiyear aerosol obser-
vations with dual-wavelength Raman lidar in the
framework of EARLINET. Journal of Geophysical
Research (Atmospheres) 109, 13203.
P´
erez, C., S. Nickovic, G. Pejanovic, J. M. Baldasano,
and E. ¨
Ozsoy (2006, August). Interactive dust-
radiation modeling: A step to improve weather
forecasts. Journal of Geophysical Research (Atmo-
spheres) 111, 16206.
Rutledge, S. A. and P. V. Hobbs (1983, May). The
mesoscale and microscale structure and organiza-
tion of clouds and precipitation in midlatitude cy-
clones. VIII: A model for the ‘Seeder-Feeder’ pro-
cess in warm–frontal rainbands. Journal of Atmo-
spheric Sciences 40, 1185–1206.
Seifert, P., A. Ansmann, D. M¨
uller, U. Wandinger,
D. Althausen, A. J. Heymsfield, S. T. Massie, and
C. Schmitt (2007). Cirrus optical properties ob-
served with lidar, radiosonde, and satellite over the
tropical indian ocean during the aerosol-polluted
northeast and clean maritime southwest monsoon.
J. Geophys. Res. 112, D17205.
Seifert, P., R. Engelmann, A. Ansmann, U. Wandinger,
I. Mattis, D. Althausen, and J. Fruntke (2008). Char-
acterization of specular reflections in mixed–phase
and ice clouds based on scanning, polarization,
and Raman lidar. In Reviewed and revised papers
presented at the 24th International Laser Radar
Conference, 23–28 June 2008, Boulder, Colorado,
USA. Edited by the Organizing Committee of the
24th ILRC, ISBN 978-0-615-21489-4, pp.571–574.
S01 - O04 - 4
