AMTAtmospheric Measurement TechniquesAMTAtmos. Meas. Tech.1867-8548Copernicus PublicationsGöttingen, Germany10.5194/amt-9-793-2016Global cloud top height retrieval using SCIAMACHY limb spectra: model studies and first resultsEichmannKai-Uweeichmann@uni-bremen.deLelliLucahttps://orcid.org/0000-0002-6698-1388von SavignyChristianSembhiHarjinderBurrowsJohn P.https://orcid.org/0000-0003-1547-8130Institute of Environmental Physics, University of Bremen, Bremen, GermanyInstitute of Physics, Ernst Moritz Arndt University of Greifswald, Greifswald, GermanyEarth Observation Science, University of Leicester, Leicester, UKKai-Uwe Eichmann (eichmann@uni-bremen.de)2March2016927938155May201510August20158January20168February2016This work is licensed under a Creative Commons Attribution 3.0 Unported License. To view a copy of this license, visit http://creativecommons.org/licenses/by/3.0/This article is available from https://amt.copernicus.org/articles/9/793/2016/amt-9-793-2016.htmlThe full text article is available as a PDF file from https://amt.copernicus.org/articles/9/793/2016/amt-9-793-2016.pdf
Cloud top heights (CTHs) are retrieved for the period 1 January 2003 to 7 April 2012 using height-resolved limb spectra measured with the
SCanning Imaging Absorption SpectroMeter for Atmospheric CHartographY (SCIAMACHY) on
board ENVISAT (ENVIronmental SATellite). In this study, we present the
retrieval code SCODA (SCIAMACHY cloud detection algorithm) based on a colour
index method and test the accuracy of the retrieved CTHs in comparison to
other methods.
Sensitivity studies using the radiative transfer model SCIATRAN show that the
method is capable of detecting cloud tops down to about 5 km and very
thin cirrus clouds up to the tropopause. Volcanic particles can be detected
that occasionally reach the lower stratosphere. Upper tropospheric ice clouds
are observable for a nadir cloud optical thickness (COT) ≥ 0.01, which is in
the subvisual range. This detection sensitivity decreases towards the
lowermost troposphere. The COT detection limit for a water cloud top height
of 5 km is roughly 0.1. This value is much lower than thresholds
reported for passive cloud detection methods in nadir-viewing direction. Low
clouds at 2 to 3 km can only be retrieved under very clean atmospheric
conditions, as light scattering of aerosol particles interferes with the
cloud particle scattering.
We compare co-located SCIAMACHY limb and nadir cloud parameters that are
retrieved with the Semi-Analytical CloUd Retrieval Algorithm (SACURA). Only
opaque clouds (τN,c>5) are detected with the nadir passive retrieval
technique in the UV–visible and infrared wavelength ranges. Thus, due to the
frequent occurrence of thin clouds and subvisual cirrus clouds in the
tropics, larger CTH deviations are detected between both viewing geometries.
Zonal mean CTH differences can be as high as 4 km in the tropics. The
agreement in global cloud fields is sufficiently good. However, the land–sea
contrast, as seen in nadir cloud occurrence frequency distributions, is not
observed in limb geometry. Co-located cloud top height measurements of the
limb-viewing Michelson Interferometer for Passive Atmospheric Sounding
(MIPAS) on ENVISAT are compared for the period from January 2008 to March 2012. The global CTH agreement of about 1 km is observed, which is
smaller than the vertical field of view of both instruments.
Lower stratospheric aerosols from volcanic eruptions occasionally interfere
with the cloud retrieval and inhibit the detection of tropospheric clouds.
The aerosol impact on cloud retrievals was studied for the volcanoes
Kasatochi (August 2008), Sarychev Peak (June 2009), and Nabro (June 2011).
Long-lasting aerosol scattering is detected after these events in the
Northern Hemisphere for heights above 12.5 km in tropical and polar
latitudes. Aerosol top heights up to about 22 km are found in 2009 and
the enhanced lower stratospheric aerosol layer persisted for about 7 months.
In August 2009 about 82 % of the lower stratosphere between 30 and
70∘ N was filled with scattering particles and nearly 50 % in October
2008.
Introduction
Clouds are extremely variable in form and size in the Earth's atmosphere.
Thus the physical cloud parameters, for example the
optical thickness, albedo, and the bottom/top heights from near the ground up
to the tropopause, are highly variable. Clouds cover about 66 % of the Earth. The value depends
on the used optical thickness threshold . Clouds are
composed of liquid particles (T>0∘C), ice
(T≈<-38∘C), or a mixed phase for intermediate
temperatures .
Clouds also impact the Earth's radiative balance and play a major, yet still
relatively uncertain, role in the changing energy budget .
The overall cloud radiative effect depends on cloud type and top height,
which is a combination of greenhouse warming (longwave) and reflective
cooling (shortwave). The knowledge of global cloud characteristics is
essential in different fields of numerical analyses, for instance, weather
prediction, circulation, or climate change models.
Optically thick clouds scatter a high percentage of the incoming visible
light back to space due to their high albedo. A cloud radiative effect of
about -50 Wm-2 is expected due to the enhancement of the annual global albedo. Additionally, clouds absorb and emit infrared (terrestrial)
radiation, which strongly depends on cloud temperature. The longwave effect
is annually and globally averaged on the order of 30 Wm-2. Cloud properties and their development in a changing
climate will determine the net effect.
The light scattering ability of cloud droplets in the visible and near
infrared is only weakly wavelength dependent in comparison to light scattered
by molecules. Furthermore, light inside a cloud will encounter enhanced path
lengths as it is scattered multiple times at the water or ice particles
inside the cloud, and light absorption by trace gases in the clouds will be
enhanced.
Tropical clouds can reach heights of roughly 17 km. Either deep
convective clouds or the more common cirrus clouds are observed in the
tropical tropopause layer (TTL) above 14 km. Cirrus
clouds are mainly formed in the upper troposphere (above 8 km) by
condensation nuclei such as mineral dust and metallic particles via
heterogeneous freezing. Larger biological particles are removed by deposition
before reaching these altitudes. Cirrus clouds are either produced by the
uplift of humid air or by a convective blow-off from deep convection. They
are made of small ice crystals due to the low temperatures at these heights
(<-30∘C) .
In 84 % of the measurements of Cloud–Aerosol Lidar and Infrared Pathfinder
Satellite Observations (CALIPSO), cirrus layers have a vertical extension of
less than 1.5 km at heights between 13 and 16 km. Only 1.5 % of the cirrus are more than 3 km
thick. Cirrus layers near the tropopause can be as thin as 0.5 km and
have a horizontal extent up to a few thousand kilometres as detected by Lidar
measurements made on the space shuttle .
Cirrus clouds are categorized as (a) subvisual, i.e. optically very thin
with cloud optical depths (COD) τN< 0.03, (b) thin
(0.03<τN<0.3), or (c) opaque (τN>0.3) .
Subvisual clouds are mainly found in the tropics with higher occurrence
frequencies over oceans and during the night. Thin cirrus clouds are more
frequent at night and are found mainly over equatorial land masses and the
western Pacific. Moreover, opaque cirrus clouds are mainly found during the
day over oceans . Cirrus cloud occurrence frequencies (COFs)
are also coupled to the dynamic circulation patterns.
reported that TTL thin cirrus COFs are partly driven by the quasi-biennial
oscillation (QBO) and the upwelling in the tropical branch of the
Brewer–Dobson circulation (BDC). For example, a reduced BDC upwelling and a
warm QBO westerly phase increase the TTL temperature and reduce the relative
humidity, which in turn leads to a COF reduction.
High-altitude, optically thin clouds are nearly transparent for the incoming
shortwave solar radiation, but they partially absorb Earth's outgoing
longwave radiation. As the cloud particles are much colder than the surface,
they re-emit comparatively less radiation towards space. The net effect is a
warming of the atmosphere below the cloud (“cloud greenhouse forcing”). Low
clouds (zct< 3 km), however, are the main contributors to
an enhancement of the global albedo. found that nearly half
of all clouds over non-polar oceans have low top heights of
zct< 3 km. These clouds have an horizontal extent on the order of
2 km. Approximatively 40 % of all clouds are low-level (above
680 hPa, below 3.2 km), 15 % are mid-level, and 45 % high-level
(below 440 hPa, above 6.5 km) clouds .
Different methods exist to measure cloud properties from space
. Nadir-viewing, passive instruments like GOME or
SCIAMACHY are specifically designed to derive trace gas columns
. However, they can also retrieve cloud top
heights (CTHs) using the strongest absorption band of molecular oxygen (i.e. O2 A
band: 755–775 nm) and infer global trends from the concatenation of a measurement record of 17+ years . The
cloud height accuracy of these methods is within a few hundred metres.
However,
these retrievals are mostly restricted to cloud optical depths larger than 5.
A global COD of 3.9 ± 0.3 is retrieved from the International Satellite
Cloud Climatology Project (ISCCP) D2 data . Thus not all
clouds are detectable from passive nadir observations, like cirrus and
cirrostratus clouds with a mean COD of τN=2.2). These clouds have a
global annual mean cloud occurrence frequency of 16.7 % as measured by the
Cloud–Aerosol Lidar with Orthogonal Polarization (CALIOP) aboard the CALIPSO
satellite .
A number of satellite missions are specially designed for the detection of
cloud properties. For example, CALIOP uses the laser wavelengths of
532 and 1064 nm to measure physical cloud tops with high
accuracy of about 60 m in the troposphere . Thin clouds
with an optical thickness of about 0.01 are still detectable with this
technique. reported an average global cloud amount of
about 68 % for clouds with optical depths higher than 0.1. This increases
to 73 % when also taking the subvisible cirrus clouds into account
.
Optically thin clouds can also be observed by passive sensors in limb or
occultation geometry. For instance, the Stratospheric Aerosol Gas Experiment
II (SAGE) instrument aboard the NASA Earth Radiation Budget Satellite
was able to detect very thin clouds with COD
τN>0.03. SAGE detected a high global cloud amount of 95 %
. A long light path of about 200 km around the
tangent point increases the number of possible cloud measurements. The
Optical Spectrograph and Infrared Imaging System (OSIRIS) on-board Odin is a
limb-viewing instrument that can accurately measure thin cirrus top heights
and COF above 12 km using the 1530 nm IR channel
and the optical thickness of subvisual cirrus clouds at
750 nm. The High Resolution Dynamics Limb Sounder
(HIRDLS) on-board the AURA spacecraft detects cirrus clouds, deep convective
clouds, and polar stratospheric clouds (PSCs) using infrared radiances at
12.1 µm . Michelson Interferometer for Passive Atmospheric Sounding
(MIPAS) on ENVISAT detected similar cloud
parameters using bands in the mid-wavelength infrared and thermal infrared regions .
Clouds in the instruments' field of view influence the retrieval accuracy of
atmospheric trace gas concentrations from limb measurements. The cloud-contaminated part of the measurement profile should be omitted in the
retrieval algorithm to minimize trace gas retrieval errors
. As a consequence, we have developed the SCIAMACHY cloud
detection algorithm (SCODA). This retrieval scheme uses a colour ratio method
to detect cloud top heights in limb-viewing geometry. It was implemented in
the SCIAMACHY level 2 operational processor (version 5.04) to
improve the limb trace gas retrievals towards the troposphere. It is also
used for scientific profile retrievals of upper tropospheric/lower
stratospheric water vapour and trace gases like nitrogen dioxide, bromine
oxide, and tropospheric ozone .
In this study limb observations from the SCIAMACHY spectrometer are used to
retrieve cloud top heights using the SCODA colour index ratio approach. We
investigate the sensitivity of the method to detect clouds in limb view for
different tropospheric cloud types using the radiative transfer model
SCIATRAN. We study the latitudinal and seasonal variations of cloud heights.
Optically thin clouds throughout the free troposphere were detected and even
very thin volcanic aerosol layers in the upper troposphere/lower stratosphere (UT/LS) region. Height-resolved cloud
occurrence frequencies are derived from the cloud top height data set to study
global cloud field distributions. The method is verified by comparing to
other satellite data and literature.
The paper is structured as follows. Section provides
information on the data used for this study. The retrieval method to compute
cloud top heights from the limb-scattered profiles is described in
Sect. . The influence of the limb path and the instrument's
vertical field of view on the results is analysed in Sect. .
Section presents model simulations performed with the radiative
transfer model SCIATRAN to investigate the robustness of the method.
SCIAMACHY retrieval results are discussed in Sect. . The
retrieval method is validated with SCIAMACHY nadir data and MIPAS limb data
in Sect. . Section explains the influence of
volcanic aerosols on retrieval results and conclusions are given in
Sect. .
DataSCIAMACHY
The atmospheric science instrument SCIAMACHY on board of ESA's ENVISAT is a
high-resolution eight channel grating spectrograph . ENVISAT was launched on 1 March 2002 into a polar,
sun-synchronous orbit with a mean local Equator crossing time of 10:00
at the descending node. Contact to the satellite was lost on 8 April 2012.
The SCIAMACHY instrument covered a spectral range from the UV-C
(214 nm) to the near-infrared IR-B (2386 nm), with a channel-dependent spectral resolution between 0.22 and 1.48 nm.
Scattered, reflected, transmitted, and emitted electromagnetic radiation was
observed in solar/lunar occultation, nadir-, and limb-viewing geometry. More
details can be found in .
Limb data
The focus of the SCIAMACHY limb observations are the retrieval of
stratospheric O3, NO3, and BrO profiles using daytime measurements of
scattered solar radiation (see and references therein).
Furthermore, polar stratospheric cloud top heights and stratospheric aerosol
extinction profiles are detectable and mesospheric temperatures are retrieved
using night-time limb measurements.
Figure gives an impression of the atmosphere from a limb-viewing satellite perspective. The picture was taken from the International
Space Station (ISS). Cloud patterns as seen from nadir view are not
detectable anymore. Instead we see a strong increase in intensity from space
towards the ground and a change of colour near the surface from dark blue to
light blue/white, where we expect the influence of clouds.
The Earth's limb as seen from the International Space Station on
16 June 2010. The ISS nadir point is in the Aegean Sea at 36.8∘ N
latitude and 27.1∘ E longitude. Image courtesy of the Image Science
& Analysis Laboratory, NASA Johnson Space Center, Mission-Roll-Frame
ISS024-E-6137 (http://eol.jsc.nasa.gov).
SCIAMACHY's instantaneous field of view was at the tangent point 2.6 km
(vertical field of view, VFOV) in vertical and 110 km in horizontal direction
(across-track) for the nominal limb-scatter mode. The scanning time of each
tangent height was 1.5 s. Typically four distinct measurements were made in
across-track direction from west to east. Each measurement had a horizontal
resolution of 240 km due to the scanning speed and integration time
and a total swath of 960 km. The horizon was
scanned in vertical direction from below the ground to about 92 km
tangent height in more than 30 steps that are 3.3 km apart
. This measurement cycle is called a limb state. A
profile was not fully vertical due to the movement of the satellite and the
scanning scheme. The maximum horizontal displacement between all tangent
height points within one profile was about 60 km in the troposphere.
The atmosphere was undersampled with a vertical gap of roughly 0.7 km
between two adjacent tangent heights. The along-track horizontal resolution
was approximately 410 km for one VFOV shell. The limb horizontal
resolution is analysed in Appendix . Full coverage of the Earth
required 6 days at the Equator with faster coverage towards the polar
regions. Figure shows a schematic view of the limb
geometry. The red solid line depicts the instruments line of sight (LOS) and
the field of view (FOV) as red dotted lines. The blue circles depict the
tangent heights for each LOS, which are numbered from the bottom to the
tropopause.
Schematic view of the limb measurement geometry. The shells are
numbered from the surface to the tropopause (here 1 to 6). The field of view
(FOV), the tangent height step width Δzth, and the cloud
top height zct are superimposed. Examples of shell path lengths
x as blue lines and areas a (grey shaded) are shown. The limb path is a
combination of the paths through each shell. The sub-paths x(3,3) and
x(3,4) (blue line) are depicted belonging to shell 3. The first index
stands for the original tangent height and the second index for the current
shell of the path segment.
We used the SCIAMACHY version 7.04W level 1b spectra. The level 1b data were
calibrated without the polarization correction due to known problems in limb
geometry . The radiometric calibration was omitted as the
colour index ratio is a division of two neighbouring tangent height
measurements and thus no absolute calibration was needed. Height-resolved
limb-scattered spectra were taken from the instrument channels 4
(750 nm) and 6a (1090 nm). These wavelengths were chosen to
avoid molecular absorption bands, e.g. for ozone, oxygen, and water vapour.
The retrieval would be disturbed because of the additional absorption
features with different height distributions. Radiation at wavelengths below
400 nm is not suitable because the atmosphere gets optically thick in
the upper troposphere due to Rayleigh scattering and ozone absorption
increases towards the Hartley–Huggins bands .
Figure displays the limb radiances I (photons s-1 m-2 nm-1 sr-1) of channels 4 and 6 as a function of wavelength
λ (nm) for tangent heights ranging from 2.3 to
22 km. Radiances from channel 5 are not plotted. High intensities are
detected for the four lowest tangent heights, where the slope between the
retrieval wavelength bands is low. At the tangent height of 15.4 km,
the intensities drop significantly compared to 12.1 km and a steeper
gradient between both bands is measured (see black arrows). We conclude that
a cloud layer is in the field of view at the lower tangent height. A larger
quantity of radiance is scattered into the line of sight. Furthermore, the
intensity decreases less with increasing wavelength due to the particle
scattering.
SCIAMACHY limb radiances from channels 4 (700–789 nm) and 6
(1000–1300 nm) for tangent heights (TH) from 2.3 (violet) to
22.0 km (orange). A measurement sequence of the ENVISAT orbit 24382
(AZM 1) from 29 October 2006 (measurement start time 10:16:10.62 UTC) at
46.8∘ latitude, 1.9∘ longitude, and 62.8∘ sun zenith
angle from operational level 1c data version 7.04W (calibration parameters
0–5) has been used. The cloud retrieval wavelength bands around 750 and
1090 nm are shown as vertical dashed-dotted black lines and the channel
borders as grey lines. The largest intensity decrease in both bands is marked
by the black arrows.
Tangent height knowledge
The SCIAMACHY tangent heights are not fixed over time and vary up to
400 m along an orbit. Variations of up to 600 m are possible
within a month. During the early phases of the SCIAMACHY mission the limb
tangent height information was erroneous. Differences between the estimated
and the real tangent height have reached several kilometres
. The main reasons for the tangent
height differences are satellite attitude knowledge errors and a misalignment
between the SCIAMACHY optical axes and the satellite reference frame, which
were later identified and corrected . The geometric
tangent height knowledge of the operational level 1 version 7 limb data is
now accurate within 170 m.
Mean SCIAMACHY limb tangent heights from azimuth index 3 (eastern profile)
and the corresponding 2 σ standard deviations (blue dashed) are
calculated from the first measurements of each orbit per day at the beginning
of each month (Fig. ). Superimposed is the difference between
two adjacent tangent heights (red line), which is ≈ 3.3 km. At
the beginning of the mission in 2003 we detect the largest intra-annual
variations when tangent heights varied between 6.8 km in spring and
4.7 km at the end of the year. The limb measurement cycle was
optimized at this time. The intra-annual amplitude was about 150 m
until the end of the year 2010 with lowest tangent heights at the beginning
of each year. In October 2010, the orbital height of the ENVISAT satellite
has been lowered by about 17.4 km to extend the mission lifetime. As a
consequence, each limb state was truncated by one tangent height step to
maintain the limb/nadir matching . The limb
tangent height pattern was modified again in January 2011 (see also
Fig. ) when tangent heights were lowered by roughly 100 m
and the step width was slightly reduced.
Time series of one limb tangent height (in km) covering the
SCIAMACHY mission lifetime; e.g. TH(3) is the tangent height with index
number 3. Daily means were plotted, taken from the first state of
10 orbits of 1 day per month (usually the first day). The dashed blue lines
depict the corresponding 2 σ standard deviation. The red line shows
the tangent height step width, which is here the difference TH(4)–TH(3)
between two adjacent tangent heights.
Nadir data
SACURA
retrieves the effective radius of cloud
droplets, the liquid water path, and other parameters like the optical
thickness for optically thick clouds (τN≥5). The cloud top height
is determined from spectral measurements in the oxygen A band
(755–770 nm). The total cloud top height error is about
±400 m for most of the cases (Fig. 1). However, it
increases with cloud height and can be up to 1400 m for high clouds
with a low optical thickness. SACURA takes multiple scattering of light
inside and below the clouds into account . Therefore, the
SACURA model is more accurate than retrieval schemes treating clouds as
Lambertian reflectors. showed that passive remote
sensing techniques tend to detect a “radiometric cloud top height”, which can
be up to a few kilometres lower than the physical cloud top height. We used
SACURA cloud top heights retrieved with SCIAMACHY level 1 version 7.04W data.
MIPAS
The limb sounder MIPAS retrieved vertical temperature profiles, trace gas
profiles, and cloud distributions. The Fourier transform infrared
spectrometer measured emissions in the mid-infrared ranging from
4150 to 14 600 nm with high spectral resolution
. While SCIAMACHY scanned the limb in forward flying
direction of the ENVISAT satellite, MIPAS faced backwards. MIPAS had a
similar resolution as SCIAMACHY in limb mode along the LOS
and in vertical direction (3 km). However, the resolution perpendicular to
the LOS of 30 km was considerably higher than SCIAMACHY
(240 km). Due to the specific spectral measurement range, MIPAS was
able to measure day and night and as a pure limb sounder recorded more limb
scans per orbit than SCIAMACHY.
The MIPAS scan measurement time in nominal mode was 56.7 s with 27 floating
altitude grid points from 7 to 72 km after 2005. After severe
in-flight anomalies in 2004, the duty cycle of 100 % was reached again in
December 2007. The vertical sampling step size in the UT/LS region was
1.5 km, which led to an oversampling of the atmosphere by a factor of
2. SCIAMACHY instead undersampled the atmosphere. The lowest tangent height
was not constant along the orbit. It varied from 5 km at the poles to
12 km at the Equator as a function of latitude ϕ with
zth(ϕ)=12-7∗cos(90- abs(ϕ)). Thus only
cloud top heights above 8.5 km were retrieved in the tropics and
extratropics (30∘ N/S).
Different methods to detect clouds have been adopted for MIPAS
. Using a cloud index (CI) method , a fast
detection scheme for the operational level 2 processing was developed to omit
cloud-contaminated measurements. This method was also used to derive cloud
field characteristics like global cloud top distributions and occurrence
frequencies . The standard operational CI
approach used two micro windows from 788.2 to 796.2 cm-1 and from
832.0 to 834.4 cm-1 in the MIPAS CI-A band with a fixed threshold
(see and reference therein).
presented an optimized cloud and aerosol detection method
for MIPAS in which new seasonal-, latitude-, and altitude-dependent cloud
detection thresholds were calculated to maximize the sensitivity to
cloud/aerosol particles in the MIPAS field of view. Based on these
thresholds, a MIPAS CTH data set was calculated for the entire MIPAS mission.
We use this data set from 2008 to 2012 for the study.
Determination of colour index ratios and retrieval of cloud top heights
SCIAMACHY provides measurements of the limb-scattered light integrated along
the line of sight. Choosing only wavelengths where trace gas absorptions and
emissions can be neglected, for instance around 750 and
1090 nm, the detected radiation is dominated by the scattering
processes of molecules, aerosols, and cloud particles
. Rayleigh theory describes scattering on air
molecules, which is strongly wavelength dependent. Scattering on aerosols and
cloud particles, however, is only weakly wavelength dependent and
can be approximated by Mie theory. Water and ice clouds show similar
scattering characteristics for the chosen wavelengths. Only above
1400 nm, the water and ice absorption have different and
non-negligible spectral dependencies , which can be
used to discriminate phases or to distinguish between cloud and aerosol
particle.
Dividing the spectrally averaged radiances of two wavelength bands
(750–751, 1088–1092 nm) by each other, we can
differentiate between measurements where particles are situated in the FOV
and those not affected by particles. This is called the colour index
approach. Colour is defined here as the ratio between radiances from two
different wavelength bands of the same viewing LOS or tangent height. The
technique was previously applied to SCIAMACHY limb measurements for the
detection of polar stratospheric clouds and the
determination of the cloud top height of a hurricane cell
. This work is a generalization of these previous
studies with respect to global particle (cloud) detection and the use of
SCIAMACHY tangent heights from the Earth's surface up to 30 km.
The cloud top height zct (km) is determined using radiance
profiles in different wavelength regions. The height-dependent colour index
CI(zth) is first calculated for all tangent heights zth [km]
from the ratio of two limb-scattered radiance intensities I at different
wavelengths λ:
CI(zth)=Ih(zth,λh)Il(zth,λl),
where h denotes the high and l the low wavelength band. The radiances are
spectrally averaged for the two bands 750 to 751 nm (l) and 1088 to
1092 nm (h). The signal-to-noise ratios of SCIAMACHY are roughly 2000 at 15 km tangent height (Bovensmann et al., 1999) for both the low and the high band, so that noisy spectra have
a negligible effect on the retrievals in general. The colour index ratio
(CIR) Θ [-] can then be defined as
Θ(zth)=CI(zth)CI(zth+Δzth),
where Δzth is the step width between two adjacent tangent heights
zth,i and zth,i+1. The tangent height grid of SCIAMACHY limb
observations, for example the spacing between two adjacent altitude levels,
is about 3.3 km wide. This also determines the retrieval accuracy. The
step width (Fig. , red line) is rather constant throughout the
mission lifetime.
The function Θ(zth) peaks at the tangent height where the cloud top
zct is within the field of view along the line of sight. SCIAMACHY has
undersampled the atmosphere in limb view. Cloud tops up to 0.6 km
above the FOV are thus not detectable at the next higher tangent height.
If Θ exceeds the predefined constant threshold, the cloud top is
allocated to that tangent height. A threshold of 1.4 is chosen to reduce
false detections due to aerosols in the lower troposphere (see
Sect. ). used 1.3 for the detection of
lower stratospheric PSCs. A radiance calibration is not necessary due to the
use of ratios, where all multiplicative and height-independent errors cancel
out. However, additive errors such as stray light can still influence the
retrieval results .
Figure shows the retrieved profiles of (a) the cloud colour
index and (b) the colour index ratio for a SCIAMACHY limb state (29 October 2006, 10:16:10.63 UTC) over Europe with four independent profiles in
across-track direction of the flight path. For visualization purposes the
colour index is normalized by the highest tangent height. A sharp CI decrease
at 12 km is visible for all four measurements. Superimposed is a cloud-free profile (red line) from the same orbit, which exhibits a low surface CI
value and a slow decrease with height. Clouds are detected when the CIR
Θ peak exceeds the threshold of 1.4 (dashed green line). Depending on
the atmospheric conditions, Θ(zth) is an unknown function of the
cloud optical thickness (COT), scattering geometry, albedo, aerosol loading, and
signal-to-noise ratio of the measurements. The sensitivity of Θ to
the scattering geometry and optical characteristics will be discussed in
Sect. .
Normalized colour index (a) and colour index ratio (b) of a
SCIAMACHY limb state (see Fig. ) for tangent heights from the
ground to 30 km. All four limb profiles (AZM 0–3) are measured
simultaneously during a limb state. Superimposed is a cloud-free limb profile
(red line) from the same orbit (start time 10:38:59 UTC, AZM 3). The green
dashed line depicts the cloud detection threshold.
Θ(zth) can peak more than once in a vertical profile, but only the
highest cloud top detected is stored with a corrected quality flag. The
double peaks occur in roughly 10 % of all retrievals, suggesting that two
or more distinct cloud layers are detected. This happens when, for example, a
high thin cloud lies above a low thick cloud. It is not possible to separate
the cloud layers when the highest cloud is optically thick. The limb-viewing
geometry is not well suited to deal with layered clouds as the position of a
cloud along the light path is not known (see also Appendix ).
Limb optical thickness
The main difference between nadir and limb cloud detection is the much larger
limb path length and the much coarser horizontal resolution. Long pathways
greatly enhance the sensitivity of the limb measurements. However, looking
sideways into a cloud field leads to a loss of information on the cloud
amount at the tangent point. Thus retrieved cloud occurrence frequencies will
in general be much higher than what is detected in nadir geometry. This will
be further analysed in Sect. .
The cloud optical thickness τL in limb-viewing geometry is larger
than the vertical optical thickness τN due to the long horizontal light
path. The path length depends on the instrument's VFOV and the tangent height. For example, an enhancement factor of 226 for
the limb optical depth is calculated for a cloud field covering the full VFOV
of 1 km and a top at 15 km tangent height. Due to the curvature
of the atmosphere, a cloud layer drops below the VFOV at a specific point
along the line of sight. Taking the area defined by the lines of the VFOV
that intersect a shell into account, we calculate an area weighted
enhancement factor (AWF) of 151.
This is the highest possible enhancement, because a cloud is not a horizontal
layer and will not cover the VFOV completely. The AWF for the SCIAMACHY
VFOV of 2.6 km is 243. This factor is generally too high. For example,
thin cirrus clouds will only fill a part of the VFOV. We simulate this effect
by reducing the VFOV to 0.1 km and then get an AWF ≈ 47. This is
a more realistic value for the increase of the vertical optical depth in limb
geometry.
As a result, a subvisual cloud layer with an optical thickness τN=0.01
lying within the field of view of SCIAMACHY has a limb τL of about 2.4
(AWF =243) and roughly 0.5 for vertically thin clouds (AWF =47). In limb
view, clouds are detected against the surrounding atmosphere and the
background space. The contrast between a cloudy and non-cloudy part of the
profile is high compared to a nadir scene. The air pressure above a cloud is
lower, less scattering occurs, and the air mass appears darker. The decrease
of brightness towards outer space is also visible in Fig. . The
nadir cloud detection is partly dependent on the contrast at the ground. For
instance, ice and snow lead to larger errors in the cloud optical thickness
determination .
Model simulations
Simulating radiances in cloudy atmospheres is difficult
because 1-D radiative transfer models like SCIATRAN
treat clouds only as layers with a defined vertical
thickness and general stratified physical properties (optical thickness,
effective radius of particles, phase function, cloud phase). This is of
course a strong simplification of the complex 3-D structure of
a cloud, which cannot be resolved with these kind of models.
The software package SCIATRAN version 3.1.28 is used to
study the strengths and limitations of the fast and simple CIR retrieval
method. The CIR sensitivity is tested for typical atmospheric cloud scenarios
like low optically thick or high subvisible clouds. Therefore the influence
of cloud parameters (for example, optical thickness τ, top height, layer
thickness) and geolocation (sun zenith angle, sun azimuth angle) on the
colour index ratio is investigated. The forward model includes a cloud model
and has been described and validated by . The radiance
accuracy is generally better than 1 % compared to other radiative transfer
models.
ENVISAT moved from the north to south along the orbit on the day side, with
varying sun zenith angle (SZA) and relative sun azimuth angle (SAA)
combinations at the limb tangent point. Because of the sun-synchronous orbit
with a descending node crossing time of 10:00, only certain combinations
of these parameters were possible. The largest sun zenith angles were found
near the poles and minima in the tropics (about 26∘). The relative
azimuth angle was generally lower towards the northern high latitudes (SAA
about 22∘) and largest at southern high latitudes (SAA about
157∘).
Pairs of corresponding latitude, longitude, sun zenith angle (SZA),
and relative sun azimuth angle (SAA) of one tangent point in SCIAMACHY limb
measurement geometry (PN: polar north, MN: midlatitude north, ETN: extra-tropical north, TRO: tropics, MS: midlatitude south, PS: polar south, ETS: extra-tropical south). A
measurement from ENVISAT orbit 24381 from 29 October 2006 for the (western)
azimuth index 0 is taken for this example.
Light scattering on aerosols and cloud particles is largest in the north due
to stronger forward scattering at low azimuth angles. Minimum scattering
occurs in tropical regions for SAAs around 90∘. Table
shows typical combinations of the SCIAMACHY limb measurement geometry for
different latitudes of orbit 24381 from 29 October 2006.
The following parameters are chosen for the tests of CIR sensitivity in a
cloudy atmosphere for different cloud top/bottom heights, optical thickness,
and geometry. A spherical atmosphere with refraction is modelled with the
radiative transfer model SCIATRAN . The scalar discrete
ordinate technique is used to solve the radiative transfer equation, where
absorption of the trace gases O3, NO2, and SO2 is taken into account. The
surface is modelled as a Lambertian reflector with a constant albedo
as=0.1, which is roughly twice as high as the albedo of water surfaces.
The tangent heights are placed from 0 to 30 km with a step width of
3 km. Radiances are modelled for the wavelengths 750.0 to
751.0 nm with 0.2 nm step width and 1088.0 to 1092.0 nm
with 0.8 nm step width, corresponding to the SCIAMACHY channels 4 and
6. The SZA varies between 20 and 82∘ and the SAA between
20 and 160∘ to cover the range of values of the SCIAMACHY limb
measurements (see Table ).
Simulated colour index ratios as a function of sun zenith angle and
sun azimuth angle for different cloud layer heights. The radiances are
calculated using the SCIATRAN forward model. The CIR are retrieved on a
3 km height grid. The black and white dots depict zenith and azimuth
angle pairs for a typical SCIAMACHY orbit. Small azimuth angles correspond to
the Northern Hemisphere and large values to the Southern Hemisphere. Please
note the different colour scales for water and ice clouds.
We simulated limb radiances for four typical cloud scenarios to find the
limitations of the cloud retrieval method with respect to cloud height and
optical thickness. The maximum colour index ratios for one retrieval height
are shown as function of sun azimuth and sun zenith angles in
Fig. . Two water cloud layers with (a) the vertical extents
of 2–3 km (lowest cloud case) and (b) 5–6 km (general water
cloud case) are modelled. The spectrally dependent optical thicknesses are
τN,wc=10.0 and 0.1 at 500 nm respectively. Two cirrus cloud
layers made of hexagonal ice crystals with 100 µm height and 50 µm
side length are simulated. We assume (c) an optical
thickness τN,ic=1.0 for the layer 8–10 km (general ice cloud
case) and (d) 0.01 for 14–15 km (subvisual ice cloud case).
The value of CIR is mainly driven by geometry, retrieval height, and cloud
and aerosol optical thickness. Thus, we choose combinations of rather low COTs
and high aerosol optical thicknesses (AOTs) for all cases except (a). The effective radius of the water
droplets is 10 µm , which is in the range of values
from literature . The particle size has a negligible effect
on the CIR of less than 1 % when varying, for instance, the water droplet
size between 10 and 50 µm. In general, ice crystals can vary
substantially in size from about 10 to 2000 µm and exhibit a variety of
different shapes . However, the CIR is not sensitive to the size
of the ice crystals. The CIR varies by less than 1 % when the crystal size
is doubled.
The influence of aerosol contamination on the cloud retrieval is taken into
account for all cases. We use the LOWTRAN aerosol parametrization defining
the height-dependent extinction coefficient and single scattering albedo. The
aerosol optical thicknesses at 790 nm are chosen to be in range with
global means over water and land from measurements of MODIS (Moderate
Resolution Imaging Spectroradiometer). MODIS measured regional annual means
between 0.04 and 0.34 at 550 nm. We selected three
AOTs τN,a=0.07 (clean case a), 0.14
(oceanic case b–d), and 0.25 (polluted land case c) at 790 nm, which
is in the range of global long-term mean values (see ,
Fig. 7.14). The aerosol phase functions are calculated using the single
parameter Henyey–Greenstein analytical formula with an asymmetry factor of
0.772 (0.0–4.0 km), 0.669 (4.0–10.0 km), and 0.657 (above
10.0 km) . The AOT defined at 790 nm
corresponds to a larger AOT at 550 nm. For example, τN,a=0.25
at 790 nm is roughly equal to 0.4 at 550 nm.
The step size for the calculations is 5∘ in SZA and 10∘ in SAA
direction. The black dots depict tangent point SAA/SZA combinations along a
typical SCIAMACHY orbit for the west and east profiles with azimuth angle
index 0 and 3. Cloud tops are in general detectable for all SZA/SAA pairs for
a CIR threshold of 1.4. When the retrieved cloud height differs from the
simulated one, the CIR is not plotted (white areas).
In panel (a) we simulate a low, thick water cloud (2–3 km,
τN,wc=10.0) in a clean area (τN,a=0.07), which is found
mostly over the oceans . Cloud tops are detected at
3 km tangent height for most of the geometry combinations except for
high SZA. If the modelled aerosol loading is set to the global oceanic mean
of 0.14, CTHs at 3 km would not be detectable anymore. Because most of the
aerosol particles are in the lower layer with an optical depth of roughly
0.125 between the ground and 5 km, the CIR is reduced and the
retrieved CTH shifted to the next retrieval tangent height of 6 km.
Low CIR values near the threshold are calculated in sideways (tropics) and
backward direction (southern high latitudes) of the scattered light beam
(panel b). The cloud is between 4 and 5 km with low COT
τN,ic=0.1 and an oceanic global mean τN,a=0.14. High CIR
values are found where the water droplet phase function has a peak in forward
direction. The CIR of a simulated cloud increases when the cloud is shifted
upwards in the model. Thus the threshold is mainly chosen to sort out wrong
cloud top heights due to high aerosol loadings in the lower altitudes. Cloud
tops are correctly retrieved above 5 km for moderate AOT
τN,a≤0.14. Clouds at the lowest possible SCIAMACHY tangent height
between 2 and 3 km are only retrievable for very low aerosol loading
(τN,a≤0.07). The lowest AOTs are found in clean areas over oceans,
especially in the Southern Hemisphere.
Cloud tops consist of ice or mixed phase particles at heights above
6 km. Thus only ice particle clouds are
simulated in the upper troposphere. A CIR map for an ice cloud at
8–10 km and τN,ic=1.0 over polluted land (τN,a=0.25)
is shown in panel (c). The retrieved CIR is well above the threshold for all
cases. High, subvisible ice clouds (14–15 km, τN,ic=0.01,
τN,a=0.14) are always detectable with the CIR method (d). Ice and
water phase functions differ in complexity and asymmetry (see
, Fig. 9). Thus we retrieve a more asymmetric SAA
dependence for the ice CIR with lowest values around 140∘. The water
phase function has a relatively broad minimum around 100∘. This explains
the CIR difference for water (a, b) and ice (c, d) particles between 60
and 120∘, where CIR is more sensitive to ice particles. High CIRs are
found for large SAA/SZA values, making the method also very sensitive to polar
stratospheric clouds as already shown by .
The strong CIR dependence on tangent height is already observable in
Fig. (Fig. S1 in the Supplement). For instance, the
retrieved CIR nearly doubles from about 1.6 (TH =5km) to 3.1
(TH = 18 km) when moving a water cloud (τN,wc=1.0) with a
vertical thickness of 1 km through the troposphere. The retrieval
sensitivity is high enough at lower tangent heights. Under normal aerosol
conditions (AOT < 0.14) clouds are detectable for COTs τN,wc≥0.1
at 6 km retrieval height and τN,wc≥0.005 in the upper
troposphere at 15 km. Sub-visible ice clouds in the upper troposphere
are already detectable for cloud optical thicknesses τN,ic≥0.003
(SZA = 25∘, SAA = 83∘, cloud layer 15–16 km). The influence
of background aerosols and refractive tangent height changes are negligible
at these heights. However, the cloud retrieval can occasionally be obstructed by
enhanced levels of aerosols from volcanic activity in the upper troposphere
when optical thicknesses of clouds and aerosols have the same order of
magnitude. The COT limits have been calculated for a spherical shell cloud
layer and have to be interpreted as theoretical lower limits. In reality, we
expect lower CIR values for very thin cirrus cloud fields that have limited
vertical and horizontal extents.
The SCIAMACHY level 1C tangent heights are geometric ones where refraction is
not taken into account. The real tangent height is lowered due to atmospheric
refraction. The effect is negligible above 22 km, where differences
between geometric and refracted tangent heights are less than 100 m.
However, the light path at zth= 6 km has a refracted tangent height of
about 4.9 km. Furthermore, the vertical field of view increases. For
the geometric VFOV, we calculate the lower edge at 4.7 km and the
upper edge at 7.3 km. The corresponding refracted VFOV edges are
3.4 and 6.4 km respectively. The refractive VFOV is thus
about 3 km wide. The lower VFOV edge at the geometric tangent height
2.3 km above the ground already is below the surface. Thus it is
possible to detect clouds down to the Earth's sea surface. However, the colour
index ratio at the lowest heights is already near the threshold and the CIR
is very sensitive to aerosol contaminations, which increase towards the
surface.
Retrieval results
SCIATRAN was already used for cloud sensitivity studies on the limb ozone
retrieval. concluded that only tangent heights above the
cloud top should be used for ozone profile retrievals in order to reduce
ozone retrieval errors. The same limb measurement mode was used here to
derive cloud top heights in order to minimize differences in time and
geolocation with respect to trace gas profile retrievals. The cloud top
height retrieval model SCODA was developed and then implemented in October 2010 into the SCIAMACHY level 2 version 5 operational processor
to support the retrieval of trace gas profiles in limb view.
SCODA retrieval results are presented in this chapter. We calculated annual
means of cloud top heights and height-dependent occurrence frequencies for
the troposphere. The limb COF is defined as the
number of measurements flagged as cloudy within one grid cell and atmospheric
layer divided by the number of all measurements for that grid cell.
Global map of (a) the annual mean cloud top height (in km) for 2006.
The data are binned in boxes of 2∘ latitude and 2∘ longitude.
The superimposed red rectangles show the approximate size of three
consecutive SCIAMACHY limb scans. The SEVIRI image in (b) shows a part
of the African continent near the Equator to illustrate the coverage of a
cloud field in limb. It was provided by M. Reuter . The
superimposed red rectangles give the sizes of the four limb measurements of one
SCIAMACHY limb scan (limb state) at the tangent point. The arrow indicates
the viewing direction. The annual mean CTH standard deviation (in kilometres) for
2006 is shown in (c).
Annually averaged cloud top heights and occurrence frequencies
Figure a shows the annual mean cloud top height (km) for 2006.
It is calculated by binning the cloudy measurements (roughly 445 000) on a
2∘× 2∘ map. Superimposed is the size and horizontal distance of
about 780 km for three SCIAMACHY limb scan cycles or states (red
rectangles). Each state is composed of four simultaneous measurements in
across-track direction. The approximate footprint size of a limb state is
superimposed in panel (b) to a cloud image from SEVIRI (Spinning Enhanced
Visible and Infrared Imager) on-board the geostationary Meteosat Second
Generation . The lower left corner of the image is at the
Equator over the Gulf of Guinea. The fine horizontal structure of a nadir
cloud field is not detectable in limb view, because a large air volume is
scanned in limb and only the highest cloud tops contribute to the measurement
of that volume. Panel (c) shows the corresponding 1 σ CTH standard
deviation. The largest scatter was found in tropical regions where the high
cloud systems have a seasonal variation changing in north/south direction.
The highest cloud tops with heights of about 13 to 16 km are observed
in the tropics. These are typically deep convective and cirrus clouds. The
lowest clouds are found in the Southern Hemisphere west of the continents
(about 4 km). Also a stream of relatively high clouds is detected
ranging from the Caribbean Sea over the northern Atlantic Ocean towards northern
Europe, which can be related to extra-tropical storm tracks. Similar features
are observed east of Japan and in the southern Pacific Ocean towards the west
coast of South America. An interhemispheric CTH difference towards the poles
is observed that will be further elucidated in Sect. .
We also observe the interhemispheric difference in the CIR annual average in
Fig. a. Especially for latitudes greater than 60∘ N/S,
the CIR differs by about 0.3. This is partly due to dependency of the phase
function on the scattering angle and thus the latitude, as shown in
Sect. . Contrarily, stray light, which is an instrumental
artefact, has a larger impact on measurements in the high Northern Hemisphere
as the sun was shining directly onto the instrument. The SAA changes from
30∘ (NH) to 150∘ (SH) at latitudes where CIR differences are
also found in our model studies (see Fig. a).
Global annual mean of (a) the maximum colour index ratio and (b) the
limb cloud detection rate for 2006. The CIR is multiplied by a factor of 10.
Low CIR regions are furthermore detected in the low cloud top height areas,
which is in line with our expectations from model studies. Panel (b) of
Fig. shows the SCIAMACHY limb cloud detection rate. A
measurement flagged as cloudy is counted when the cloud top height is between 0
and 20 km. Very high values are detected in comparison to cloud
occurrence frequencies retrieved in nadir geometry. More than 93 % of the
391 700 measurements made in 2006 are marked as cloudy. Cloud-free areas are
only rarely detected in SCIAMACHY limb measurements because of the long
tropospheric geometric path length of 1200 km.
called it the “limb-smearing effect”, which makes comparisons to nadir
measurements complicated. Comparatively low cloud detection rates of about
70 % are found only in low cloud top regions, west of the continental
coasts in the Southern Hemisphere. The difference between both polar regions
already seen in other CTH/CIR maps is also detected here.
SCIAMACHY annual mean cloud occurrence frequencies for 2006 and for
tangent heights between (a) 1.0 and 3.75 km, (b) 3.75 and 6.5 km,
(c) 6.5 and 10.0 km, (d) 10.0 and 13.5 km, (e) 13.5 and 16.5 km, and
(f) 16.5 and 19.9 km.
A high cloud coverage over desert areas like Australia or North Africa is not
expected, but we observed a significantly higher cloud measurement rate, for
example, over North Africa in comparison to measurements from high-resolution
nadir instruments (see Fig. ). This can be attributed to
measurements of desert dust transported away from these areas up to heights
of 6.5 km (see Fig. a–b) and to cirrus clouds at higher
altitudes, which are not detectable by passive nadir instruments.
reported consistent high occurrence rates of dust over North
Africa and surrounding areas up to altitudes of 6 km.
Height-resolved cloud occurrence frequencies
Figure shows annual mean limb COF (%) for the year 2006. The COFs are calculated for six altitude
layers. Each layer represents a SCIAMACHY tangent height, which was nearly
constant over time and orbital position. The COF is calculated by comparing
the number of cloud detections in one layer and grid box with the number of
all measurements of that grid box.
In layer 1.0–3.75 km (panel a) the largest amount of clouds are
detected over oceanic regions in the Southern Hemisphere westwards of the
continents. Slightly more clouds are observed between 1.0 and 6.5 km
in the southern polar regions than in the northern counterpart. It seems that
the method is rather insensitive to low clouds for small scattering angles
and large sun zenith angles, which are typical for high northern latitudes.
Stray light might also be a factor that has to be further analysed.
Furthermore, the limb-smearing
effect can lead to higher retrieved CTHs.
The enhanced COFs over North Africa can be attributed to aerosols uplifted by
dust storms over the Sahara, as this region is nearly cloud free. Higher COFs
were also detected over the African continent and west of it at the height
range 3.75–6.5 km (b). CALIOP measured dust plumes from Africa up to
8 km altitude over the Atlantic .
High COF between the midlatitudes and the polar regions were also detected
by CALIOP (see Fig. 3g–h). High COFs at low altitudes
(CTH <3km, >680hPa) were mainly found over the oceans,
especially near the west coasts of the continents, which is in line with our
observations from Fig. a. However, CALIOP observed no enhanced COFs
over North Africa, supporting the argument that SCIAMACHY detected an
aerosol layer in this region.
The largest interhemispheric COF differences are discovered in the layer
6.5–10.0 km shown in panel (c). The northern hemispheric COF at
latitudes north of 60∘ N is above 50 % and in the corresponding
Southern Hemisphere on the order of 40 %. The combination of COF patterns
of layers (a) and (c) explain the interhemispheric CTH differences as seen in
Fig. . observed that the first
measurements of the sunlit part in the Northern Hemisphere are influenced by
stray light and have to be excluded. For this reason only limb states of the
descending orbit phase are used and also the first five states after the
descend are omitted.
Cloud tops between 10 and 13.5 km are mainly detected in areas of
frequent storm activities in the extra-tropical zone for latitudes higher
than 30∘ N/S (see Atlas of Extratropical Storm Tracks
http://data.giss.nasa.gov/stormtracks/). CALIOP also observed high level
clouds above 7.2 km in these regions (Fig. 3c–d).
The occurrence of clouds above 13.5 km is limited to the tropical zone
as seen in panels (e) and (f). Largest COFs of about 70 % are detected over
the western part of the Pacific from Indonesia to the east between 90
and 180∘ longitude. High clouds are also observed over Central Africa
and Middle America (above 40 %). Another feature is the slightly enhanced
COF over Antarctica, which can be attributed to PSCs in the southern
hemispheric winter/spring period from mid-June to early November.
Temporal evolution of CIR and CTH
The global monthly mean colour index ratio is calculated using all cloud
detections for tangent heights between 0 and 25 km. The CIR varies
slightly over time (Fig. ). After a rather stable period from
2003 to 2006 with CIR averages above 2.0, it decreased to 1.84 at the
beginning of 2007 and remained around 1.9 until August 2011, when the largest
decline down to 1.75 occurred. Some of the short-term CIR reductions
corresponded to major volcanic eruptions in 2008, 2009, and 2011. Lower
stratospheric volcanic aerosol layers are optically very thin and
corresponding CIRs are thus smaller than tropospheric cloud CIRs.
Two other wavelength pairs are also tested for their temporal behaviour.
Similar features are found for the 1551/1090 nm CIR, which
is used in the limb water vapour retrievals as an additional constraint
. The CIR is also reduced during the longer instrument
decontamination phases in August/December 2003, June/December 2004, and
December 2008. The 1685/1551 nm CIR is currently under
study, which can possibly discriminate between clouds and aerosols. This
ratio is rather stable for the whole period, indicating that it is insensitive
to aerosols. If the CIR decrease is due to instrumental degradation, it could
reduce the detection sensitivity for very thin cirrus and low-altitude
clouds because of the constant detection threshold. Nevertheless, the mean
CIR is still well above 1.8 at the end of the instruments lifetime in 2012.
Global monthly averages of the SCIAMACHY colour index ratio CIR for
the 1090/750 nm wavelength pair and its 1 σ standard deviation.
Also plotted are two different CIRs using the ratios of 1551/1090 nm
(blue dashed line) and 1685/1551 nm (red dashed line).
Time series of the SCIAMACHY CTH zonal monthly mean for latitudes between
±75∘ show an annual cycle and an interhemispheric CTH difference
(Fig. a). High CTHs are observed in the summer seasons of
each hemisphere with generally lower values in the south. Also the maximum
CTH is slightly shifted to the north. Highest CTHs above 14 km are
measured near the Equator in the first half of each year and lowest CTHs below
6 km in the winter periods around 25∘ S. PSCs are detected
regularly in the southern hemispheric winter with top heights in the lower
stratosphere above 12 km. High PSCs in the northern polar region are
not clearly detectable in the zonal means but are visible in the standard
deviations shown in panel (b). PSCs in the Northern Hemisphere are detected
in the winters of 2005, 2007, 2008, and 2011. The four volcano eruptions (red
dots) that lead to deviations in cloud top height fields will be analysed in
Sect. .
Time series of SCIAMACHY monthly zonal means CTHs (a) and
(b) 1 σ standard deviations (in kilometres) for the period from January 2003 to
April 2012 between 75∘ S and 75∘ N. The red dots depict the start
dates and latitudes of 4 major volcanic eruptions: Kasatochi in August 2008,
Sarychev Peak in June 2009, and both Nabro and Puyehue–Cordón Caulle in
June 2011.
There are features both in the CTH and standard deviation maps that cannot be
attributed yet. For instance, we observed a considerable increase in CTH
between December 2007 and April 2008 around 20∘ N with a large
longitudinal scatter taking the standard deviation in panel (b) into account.
This might be originating from the volcano Tavurvur (4∘ S,
152∘ E), which erupted between August 2006 and January 2007 with a
Volcanic Explosivity Index of 4. Sulfur dioxide and ash clouds reached
the tropopause in October 2006 and spread afterwards northwest and southeast
. However, the largest CTHs and a large longitudinal scatter are
observed at the end of 2011 in the latitude band 0 to 10∘ N.
Mean CTHs up to 21.3 ± 8.2 km are detected in December 2011, while
normal values are on the order of 13 ± 4 km at this location and
time of the year. So far the reason for this phenomenon is unclear and has to
be further analysed. It might be attributed to rising aerosol particles after
the Nabro eruption.
Validation with SCIAMACHY nadir and MIPAS limb cloud top heights
The comparison of limb measurements with nadir retrieved cloud parameters is
no simple task due to long light paths as discussed in
Sect.
and also addressed by . We first compare cloud top heights
retrieved from the two viewing geometries nadir/limb of SCIAMACHY and
interpret the differences. We then validate our results with collocated limb
CTH from MIPAS in the second subsection.
Comparison with SCIAMACHY nadir cloud measurements
A perfect agreement between limb and nadir cloud top measurements cannot be
expected. The two main limiting factors are the low limb horizontal
resolution and the high optical thickness threshold necessary for nadir
retrievals using passive instruments like SCIAMACHY. We used SACURA data
where the following quality checks were applied to constrain the nadir data
(see , Table 4): top height convergence (flag 2), top/bottom
height convergence (5), high geometrical thickness (3), and a cloud optical
thickness larger than 30.
The mean COFs from nadir SCIAMACHY–SACURA data (Fig. )
clearly deviate from the limb COF shown in Fig. . With the
high nadir horizontal resolution a more pronounced land–ocean contrast is
detected. The global COF distribution of both viewing geometries above
10 km altitude is rather similar. A lower percentage of tropical
clouds is found in nadir data (panel e–f). Also PSCs in the southern polar
region are not detected due to optical thickness limitations. The nadir
retrieval is restricted to the detection of clouds with an optical thickness
of τN>5 and the limb retrieval sensitivity degrades
for very low clouds. With a global cloud optical thickness of approximately
3.7 ± 0.3 , thin clouds cannot be detected by nadir
passive sounders, for instance high-altitude cirrus clouds. This has
consequently led to a lower mean top height in the tropics where most of the
cirrus clouds are situated. The active nadir-viewing instrument CALIOP
detected very thin clouds. For instance, 56 % of the total cirrus clouds
are observed at subtropical latitudes of ±30∘.
SCIAMACHY SACURA nadir cloud occurrence for the year 2006, projected on a
grid of 2∘-sided cells. From (a) to (f) values are sorted in altitude layers
as in Fig. . Due to the inherent conceptual difference between limb
and nadir observation geometry, the occurrence is calculated averaging the actual
nadir cloud fraction over time, resorting to the ergodic assumption by
which space and time averages are equivalent.
The COFs for very low clouds over the oceans are enhanced (> 50 %) in both
viewing geometries (a). The limb COF asymmetry between the polar Northern and
the Southern hemispheres especially in the lower layer (a) is not observed in
nadir view. High nadir COFs are detected in the Northern Hemisphere over land
(panel a–b) and over ocean (c). Contrarily, limb COFs are highest
between 6.5 and 10 km north of 60∘ N/S. These high
clouds restrain the detection of low clouds in limb view. The enhanced limb
COFs over North Africa due to aerosols are not observed in nadir data.
Figure summarizes the latitudinal differences between limb and
nadir CTH, where SCIAMACHY zonal mean cloud top heights averaged over 7 years
(2003 to 2009) are shown. The black line with the light grey area depicts the
SACURA nadir mean CTH and the 1 σ scatter. Average CTHs are generally
below 10 km in the tropics and at about 4 km towards the polar
regions. The difference to the SCODA limb averages (orange line) is generally
high. Limb CTHs are up to 4.5 km higher in the tropics near
10∘ N. Smaller differences of about 2 km are observed towards
the southern polar region. When the scatter is added to find the highest nadir
clouds, we still measure differences of roughly 2 km in the tropics
and of up to 2 km at latitudes north of 40∘ N. The
interhemispheric asymmetry is more pronounced in the limb measurements. When
the years of high volcanic activity 2008 and 2009 are excluded from the limb
average (violet dashed line), the cloud heights in the Northern Hemisphere
are only slightly reduced by roughly 0.5 km in the latitude band from
20 to 82∘ N. Mean nadir CTH and standard deviation maps for
2006 are given in Fig. S2.
SCODA zonal mean CTH averaged over 5 (2003–2007, violet dashed
line) and 7 (2003–2009, orange line) years. The gap in the Northern
Hemisphere between both curves shows the influence of the volcanic eruptions
in 2008 and 2009 on the retrieval results. The corresponding SACURA nadir
zonal means (blue dashed and black line) and the 1 σ intervals (grey
area) are superimposed.
Validation with MIPAS limb cloud top heights
Better agreement can be achieved with other limb-viewing instruments that
have a similar field of view and sensitivity to thin cirrus clouds and
aerosol particles (see e.g. ). Co-located
SCIAMACHY and MIPAS limb measurements of cloud/aerosol top heights are used
in this study. The limb tangent point displacement between both sensors is
about 170 km and the temporal displacement is 800 s. The time
difference is nearly constant as both instruments looked into opposite
directions of the satellite flight path. Both instruments had similar
horizontal sampling lengths (Appendix ) due to their VFOVs. A
MIPAS validation data set with measurements from January 2008 to March 2012 is
used. Because of instrumental problems the resolution of MIPAS was changed in
2005. The full data rate was reached again in December 2007. Hence we have used
2008 as the start year for the comparisons.
Figure shows a scatter plot of SCIAMACHY and MIPAS
co-located cloud top heights for April 2010. Only measurements have been used
for the comparison when both instruments detected a cloud or particle layer
between 8.0 and 22 km. The atmosphere is divided into six
latitude bands. The superimposed grey lines depict the SCIAMACHY vertical
resolution of about 3.3 km. As MIPAS has a smaller vertical step size
and tangent heights are varying continuously over time, MIPAS CTHs cover the
whole vertical height range in contrast to SCIAMACHY, which has a rather fixed
tangent height grid.
Scatter diagrams of co-located cloud top heights from SCIAMACHY and
MIPAS measurements for April 2010. The colour-coded measurement pairs are
divided into six latitude bands that are 30∘ wide: north/south polar,
two midlatitude, and two tropical bands.
In general, MIPAS detects cloud tops slightly higher than SCIAMACHY. The
monthly mean difference ΔSM=zct,S-zct,M for 3121 co-locations
is -0.9 km. For SCIAMACHY CTHs around 9 km, we found a high
percentage of MIPAS CTHs that are outside the instruments' vertical field of
view. The largest CTH deviations are observed in the tropics (light blue and
green dots) where the lowest MIPAS tangent heights are at about 10 km.
The main reason seems to be differences in detection sensitivity. MIPAS has a
smaller horizontal across-track field of view, which limits the smearing
effect to some extent and might lead to a higher sensitivity for very thin
clouds. The MIPAS CTH are also expected to be higher by up to 1.0 km
than reality . Furthermore, MIPAS colour indices are
affected by water vapour in the troposphere up to 12 km as reported by
. These effects can explain why MIPAS has measured
consistently higher cloud tops, especially in tropical regions.
Figure summarizes the global CTH differences ΔSM of
SCIAMACHY and MIPAS. Results are zonally and monthly averaged with vertical
differences around ΔSM=-1.1km. This is in the range of the
instrumental field of views that is the limiting factor of the vertical
resolution. The differences can partly be attributed to the different tangent
height step sizes of 3.3 km (SCIAMACHY) and 1.5 km (MIPAS). The
largest differences and 1 σ standard deviations between both data sets
are observed at times of volcanic particle intrusions into the UT/LS region
(see Fig. S3). compared MIPAS top heights
with HIRDLS and CALIOP measurements.
They observed that MIPAS CTHs are up to 1 km higher for altitudes
between 12 and 20 km, which is in line with our comparisons.
Zonal monthly means of co-located cloud top height differences (in
%) between MIPAS and SCIAMACHY (black line with diamonds) and the
corresponding 1 σ standard deviation (dashed black line) for the time
period between 2008 and the end of mission in April 2012. The differences for
four SCIAMACHY tangent heights (TH) between 8.5 and 18.5 km
are also shown as coloured lines.
Taking a closer look at the height-resolved CTH differences, we find the
largest discrepancies at the lowest tangent height of about 8.7 km
that is used for the comparison. Here the vertical displacement ΔSM
for April 2010 is -1.9 ± 1.3 km (415 collocations),
-0.7 ± 1.7 km (1302) at 12 km, -1.1 ± 1.2 km (1071)
at 15.3 km, and +0.8 ± 0.7 km (128) at 18.5 km.
SCIAMACHY cloud top heights are usually higher at the tangent height
18.5 km. This can partially be explained by the coarser tangent height
step size and the undersampling of SCIAMACHY, as 18.5 km is above the
tropopause. The largest differences at this height are observed during
periods of volcanic aerosol intrusions; for example, the difference is
1.9 ± 1.2 km for October 2008 (168 collocations) and
1.7 ± 1.5 km for July 2009 (245 collocations). MIPAS is also
sensitive to volcanic aerosol particles at heights above 19.5 km,
where both instruments detect particles in 2009. However, in this case, the
difference of 4.0 ± 3 km is much larger than the combined vertical
FOVs. This suggests that the instruments have different sensitivities to the
small sulfuric acid droplets in the lower stratosphere. The number of
co-located measurements in September 2009 at the 18 km height (514) is
more than 10 times higher than in 2010 (64).
In the second half of 2011 both instruments detected high cloud/aerosol layer
heights in the tropics (see Fig. ). Starting in June 2011,
the Nabro volcano injected large amounts of sulfur dioxide into the upper
troposphere, comparable to the Sarychev peak eruption 2009. While the top
heights around 18 km nicely agree (0.6 ± 0.8 km) for
September 2011, the disagreement at the lowest tangent height is very large
(-5.3 ± 2.6 km). MIPAS seems to be more sensitive to the Nabro
aerosol layer, as the differences between both instruments in the latitude
band of 30–60∘ N start to increase after this eruption. The CTH differences
increased from 0.7 km in June 2011 to 5.0 km in November 2011,
which can be attributed to the differences at the lower tangent heights. It
has to be noted that the very high CTHs at the end of 2011 as seen in
Fig. are only observed in SCIAMACHY data. The high values
are restricted to a small latitude band in the tropics. A reason for this
event is not found yet.
We also compared zonal mean CTH for heights between 6 and
28 km with results from (Fig. 12) (see Fig. S4). The three monthly averages for March–April–May and
September–October–November 2003 are in relatively good agreement with maximum
differences of about 1.5 km.
Particle detection after volcanic eruptions
The influence of aerosols on the colour index ratio Θ has already been
studied by . High aerosol loading in the upper
troposphere/lower stratosphere can lead to false polar stratospheric cloud
detection. This is also true for the retrieval of cloud top heights
throughout the troposphere.
The lower the cloud, the more an aerosol layer can disturb the retrieval, as
the CIR Θ is strongly height dependent and aerosols tend to reduce the
CIR. Aerosols do not exhibit a sharp top height. Because of lower CIR values
towards the surface, the extra scattering due to aerosols have a relatively
high impact on the CIR. Nevertheless, we detect CTHs at tangent heights below
5 km mainly in oceanic areas and with a high percentage towards the
southern polar regions (see Fig. a). As shown in the height-dependent COF distributions, aerosols also influence the detection over
desert regions. Furthermore, the intrusion of aerosol particles into the
stratosphere is detectable even for very low aerosol optical thicknesses in
the UT/LS region, because the threshold τN for subvisual cirrus
applies to aerosols as well.
There were three main events during the life cycle of SCIAMACHY, where
volcanoes in the Northern Hemisphere ejected a moderate amount of material
into the lower stratosphere. The particles are falsely detected as clouds in
the lower stratosphere over several months on a semi-global scale. The
following eruptions have been observed: Kasatochi (52.1∘ N,
175.3∘ W) starting on 7 August 2008, Sarychev Peak (Matua Island:
48.1∘ N, 153.2∘ E) on 12 June 2009, and Nabro (Eritrea:
13.4∘ N, 41.7∘ E) on 13 June 2011. Mainly for these three events,
MIPAS has observed elevated levels of sulfur in the lower stratosphere
.
The emitted volcanic sulfur dioxide was transported into the UT/LS region.
Over time it was oxidized with water to sulfuric acid and then condensed
into larger droplets. The mechanism is slow and takes weeks for the
transformation. The sulfuric acid droplets were advected around the globe.
Due to the increasing particle diameter, they are detectable in the UT/LS
region with the CIR method when the optical depth is high enough.
The EARLINET lidar site found aerosol optical depths at 532 nm between
0.004 and 0.025 in 2008 for heights above 5 km and
up to 0.07 in 2009 . measured
optical depths up to 0.008 in July 2011 at a wavelength of 780 nm and
aerosol layer top heights of 19 km. OSIRIS measured similar values.
Also the aerosol layer was still detectable in the middle of September 2011
. At the same time in June 2011, the Puyehue–Cordón
Caulle (PCC) volcano (40.6∘ S, 72.1∘ W) emitted large amounts of
mainly ash up to heights of 13 km (see e.g. the global volcanism
program website).
The main components ejected in the Kasatochi and Sarychev event are sulfate
(more than 50 %) and carbonaceous material (less than 43 %)
. The percentages changed over time in the UT/LS region
and estimated a residence time of 45 ± 22 days
for both volcanoes using measurements from the CARIBIC (Civil Aircraft for
Regular Investigation of the atmosphere Based on an Instrument Container)
platform. Residence times in the UT/LS region are the longest for aerosols
with particle diameters around 1 µm. A large spread of residence times
from 9 to 62 days was reported in other studies .
The influence of stratospheric particles on the limb CTH detection is already
observable in Fig. , where the red dots depict the latitude
and estimated start of the breakout from the three volcanoes described above.
We now take a closer look at the impact of the three big eruptions and their
temporal evolution on the CTH retrieval results. The measurements are
sometimes dominated by aerosol particles in the lower stratosphere.
Figure shows the impact of volcanic aerosols can be transported
into the lower stratosphere at latitudes between 30 and 70∘ N. The
lowest tangent height layer between 12.5 and 15.5 km (red line) is
situated in the UT/LS region for this latitude band. While the layer is at
the upper edge of the troposphere for latitudes near 30∘ N, it is
already in the stratosphere towards 70∘ N. Under normal conditions, the
largest occurrence frequencies occur in summer seasons with maxima of nearly
5 % in the lowest layer. The year 2010 is an example of such an undisturbed
year in contrast to the other years. Thus only a small amount of particles are
observed in the lowest layer, while the two highest layers are completely
free of particles for most of the year. Only small areas above 18.8 km
in the wintertime Arctic PSC seasons are filled with particles.
SCIAMACHY monthly zonal mean occurrence frequencies (in %) of
cloud/aerosol scattering layers for the latitude band from 30 to
70∘ N for the years 2008–2011. Three different cloud top height layers
of 12.5–15.5 km (red line), 15.5–18.8 km (green), and
18.8–22.2 km (blue) are depicted. The vertical dashed line show the
start months of the three major volcano outbreaks from Kasatochi, Sarychev
peak, and Nabro.
Most of the volcanic particles are detected in the layer between 12.5 and
15.5 km. Up to 30 % of the area is covered with aerosols in October
2008. The maximum coverage is about 48 % in September 2009 and 17 % in
August/September 2011. The second peak in September 2009 might be due to
sedimentation of particles from the layer 15.5 to 18.8 km, where a
peak occurrence rate 29 % is seen in August. However, still a considerable
amount of aerosols are observed in the intermediate layer with maxima of
19 % in September 2008, 29 % in August 2009, and 23 % in August 2011. A
significant occurrence frequency maximum for the layer 18.8–22.2 km
is only observed for September 2009 (10 %) with lower values in the months
before and after September. Overall, about 45 % of all cloud measurements
are located in the lower stratosphere in October 2008, 82 % in August 2009,
and 40 % in July 2011.
Elevated aerosol levels are observed for about 4 months in 2008, 7 months in
2009, and 5 months in 2011. The time spans for aerosol layer detection in
2008/2009 are in line with measurements from the Infrared Atmospheric Sounding Interferometer of H2SO4 and OSIRIS for both volcano eruptions in 2008/2009
. Atmospheric Chemistry Experiment Fourier transform spectrometer (ACE-FTS) measurements of SO2 and sulfate
aerosols showed a Sarychev particle lifetime of about 7 months
. ACE-FTS detected the plume between 8.5 and
17.5 km with a maximum around 12 km at higher latitudes
(55–70∘ N).
While the volcanoes of Kasatochi and Sarychev Peak are at similar latitudes
in the Northern Hemisphere, the Nabro volcano is in the tropics and the
particles had to be advected into the northern region over time. This might
explain why we detect higher amounts (23 %) of particles in the layer 15.5
to 18.8 km (August 2011) than in the layer below, as particles are
already vertically transported into the lower tropical stratosphere, then
move northwards, and descend again.
The PCC eruption starting in June 2011 is only barely visible in the time
series of Fig. with only a slight increase in mean CTH. However,
a similar CIR reduction in the Southern Hemisphere is observed after the
eruption as for the other volcanoes. The thin aerosol layers in the UT/LS
region lead to lower CIRs compared to clouds and can thus be identified (see
Fig. S5).
Conclusions
In this study, we refined the colour ratio method that was first used for the
detection of PSCs in limb-viewing geometry. SCIAMACHY limb measurements were
used to investigate the global cloud top height and occurrence distributions
for the lifetime of the instrument. The use of only one threshold for all
atmospheric situations is simplistic in nature but sufficient for the
majority of cases, except for the lowest clouds.
For tangent heights between 2 and 5 km, the influence of aerosols and
viewing geometry is not negligible. The retrieved cloud top height can be
wrong, because a shift to the next tangent height is possible for moderate
aerosol loadings in the troposphere. Nonetheless, we are able to detect low clouds
within this altitude range in clean areas, mainly over southern oceans. This
is also supported by comparisons with SACURA nadir COF measurements.
SCIATRAN model studies have shown that the method is very sensitive for a
wide range of cloud optical properties and cloud top heights. The vertical
accuracy of the CTH retrieval is limited to the tangent height step size of
about 3.3 km. Also, the spatial resolution is rather coarse. The
horizontal field of view can be up to 800 km along-track and
240 km across-track, which promotes smearing effects with respect to
the location of the cloud field.
Still we found good agreement, comparing our results with co-located MIPAS
data. MIPAS has a narrow 30 km across-track FOV and a tangent height
step size of 1.5 km. SCIAMACHY CTHs are usually lower by about
1 km. This can be attributed to differences in the tangent height step
size and that MIPAS tends to overestimate the CTH . MIPAS
also has an 1 km offset compared to other satellite measurements
. Furthermore, sensitivity differences of the methods for
thin clouds and small aerosol particles lead to larger discrepancies up to
5 km, as seen in the Nabro case in November 2011.
Aerosols affect the CTH retrieval not only at low tangent heights in polluted
areas but also after intrusions of larger amounts of volcanic sulfuric acid
droplets and ash particles into the UT/LS region. We detected large aerosol-contaminated areas for long periods of several months mainly in the Northern
Hemisphere in 2008 and 2009. After the Sarychev volcano eruption in 2009, the
northern stratosphere was filled with particles from the poles down to the
extra tropics for nearly 7 months. Particles are observed up to heights of
about 22 km.
We analysed monthly-sampled cloud top altitudes measured by SCIAMACHY for the
two different observational geometries limb/nadir. The CTH differences can
partly be attributed to the different sensitivities of the viewing
geometries. To this end, a limb/nadir matching technique proves to be
advantageous in filling the representation gaps for specific cloud
types/regimes, which are only observable in one of both viewing geometries.
SCIAMACHY was the only instrument that could simultaneously make use of the
two viewing geometries.
The cloud top height detection method first used to detect PSCs was the
baseline of the SCODA algorithm. It is now operationally used to improve the
operational limb trace gas retrievals towards the tropopause. Here we have
shown the importance of the method for cloud and aerosol studies. The use of
wavelength pairs in the near IR will possibly enables us to distinguish
between clouds and aerosols in the upper troposphere. Scientific studies are
currently ongoing. Although SCIAMACHY and MIPAS, with their respective
sensitivities, have proven to be of great value for UT/LS studies, new
missions with these types of limb-viewing instruments are currently not
planned.
Horizontal Along-Track Resolution
The along-track horizontal sampling length for one atmospheric shell
2x(i,i) is about 410 km, where i is the index of the tangent
height zth(i) (see Fig. ). The position of the
tropospheric cloud along the line of sight cannot be resolved with the
retrieval technique. Thus the real length, where clouds can be sampled, is much
longer, as they may also appear in the line of sight outside the lowest shell
given by zth(i). The tropopause is taken as the upper edge for clouds.
It is latitude dependent and can reach about 17 km in the tropics.
With this tropopause height (index itp), we calculate the CTH-dependent
extra path length. With a cloud top zth=6km in the tropics, we
found a nearly doubled total horizontal sampling length X=790km,
where clouds can appear along-track within the VFOV. We calculate X as
∑x(i,j=i,…itp). The overall path is 560 km for a cloud
top zct=12km. For instance, a cloud top zct=3km,
measured over Germany, could be anywhere along the countries north/south axis
(about 800 km) at altitudes up to the tropopause.
It should be noted that the horizontal sampling length is on the order of the
distance between two consecutive SCIAMACHY limb states with a geographic
angle of about 7∘ (780 km). For most of the cases, the LOS
above a cloud top of one state will not reach the measurement area of the
adjacent state, as it is then outside the troposphere already. However, for a LOS
from a cloud-free measurement with a tangent point close to the surface, the
along-track horizontal resolution is approximately 1200 km,
corresponding to an angle of roughly 10∘.
To further analyse, whether these long extra paths are relevant for cloud
detection, we also have to take the vertical resolution of the instrument
into account. We calculate the vertical area-of-sight (VAOS) a
(km2), which is defined by a shell segment that is limited in
vertical direction by the VFOV. The across-track LOS dimension can be
neglected, as it is the same for all VAOS. For example, the path length ratio
x(3,3)/x(3,4) is roughly 2.4 and the ratio between the two corresponding
areas a(3,3)/a(3,4) is 1.2 (see Fig. ). The first index
defines the tangent height index, the second the index of the shell. All
shell areas up to the tropopause are important for the limb cloud detection
with area ratios close to 1.
In trace gas limb retrievals, number densities are calculated for each shell
assuming a horizontal continuity in each shell. Thus it is possible to derive
trace gas profiles, which can be attributed to the tangent height points.
This is not the case for discontinuous cloud fields in our retrieval method.
If, for example, the area a(3,3) is free of clouds and a(3,4) is not, the
retrieved limb CTH, which is registered at the tangent point, will be higher
than the nadir CTH. Furthermore, another cloud in area a(2,2) or a(2,3)
is not be detectable in limb. The presence of high clouds outside areas at
the tangent point along the light path thus lead to (a) a CTH overestimation
and (b) an increase of the cloud occurrence frequency.
The Supplement related to this article is available online at doi:10.5194/amt-9-793-2016-supplement.
Acknowledgements
This work was funded in part by ESA within the SCIAMACHY Quality Working
Group (SQWG) project and the Sentinel5-Precursor L2 development project (DLR
grant 50EE1247). The work was supported by the European Space Agency (ESA),
the German Ministry of Education and Research (BMBF), the German aerospace
centre (DLR), the University of Bremen and the Ernst Moritz Arndt University
of Greifswald, Germany. SCIAMACHY was jointly funded by Germany, the
Netherlands, and Belgium.The article processing charges for this open-access publication were covered by the University of Bremen.
Edited by: G. Stiller
ReferencesAndersson, S. M., Martinsson, B. G., Friberg, J., Brenninkmeijer, C. A. M.,
Rauthe-Schöch, A., Hermann, M., van Velthoven, P. F. J., and Zahn, A.:
Composition and evolution of volcanic aerosol from eruptions of Kasatochi,
Sarychev and Eyjafjallajökull in 2008–2010 based on CARIBIC
observations, Atmos. Chem. Phys., 13, 1781–1796,
10.5194/acp-13-1781-2013, 2013.Boucher, O., Randall, D., Artaxo, P. Bretherton, C., Feingold, G.,Forster,
P., Kerminen, V.-M., Kondo, Y., Liao, H., Lohmann, U., Rasch, P.,Satheesh,
S.K., Sherwood, S., Stevens, B., and Zhang, X.Y.: Clouds and Aerosols, in:
Climate Change 2013: The Physical Science Basis. Contribution of Working
Group I to the Fifth Assessment Report of the Intergovernmental Panel on
Climate Change, edited by: Stocker, T. F., Qin, D., Plattner, G.-K., Tignor,
M., Allen, S. K., Boschung, J., Nauels, A., Xia, Y., Bex, V., and Midgley, P.
M., Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA,
571–658, 10.1017/CBO9781107415324.016, 2013.Bourassa, A. E., Degenstein, D. A., and Llewellyn, E. J.:
Climatology of the subvisual cirrus clouds as seen by OSIRIS on Odin,
Adv. Space Res., 36, 807–812, 10.1016/j.asr.2005.05.045, 2005.Bourassa, A. E., Degenstein, D. A., Elash, B. J., and Llewellyn, E. J.:
Evolution of the stratospheric aerosol enhancement following the eruptions of
Okmok and Kasatochi: Odin-OSIRIS measurements, J. Geophys. Res.-Atmos.,
115, D00L03, 10.1029/2009JD013274, 2010.Bourassa, A. E., Robock, A., Randel, W. J., Deshler, T., Rieger, L. A.,
Lloyd, N. D., Llewellyn, E. J. T., and Degenstein, D. A.: Large volcanic
aerosol load in the stratosphere linked to Asian monsoon transport, Science,
337, 78–81, 10.1126/science.1219371, 2012.Bovensmann, H., Burrows, J. P., Buchwitz, M., Frerick, J., Noël, S., Rozanov, V. V., Chance, K. V., and Goede, A. P. H.:
SCIAMACHY: Mission Objectives and Measurement Modes,
J. Atmos. Sci., 56, 127–150, 10.1175/1520-0469(1999)056<0127:SMOAMM>2.0.CO;2, 1999.Bramstedt, K., Noël, S., Bovensmann, H., Gottwald, M., and Burrows, J.
P.: Precise pointing knowledge for SCIAMACHY solar occultation measurements,
Atmos. Meas. Tech., 5, 2867–2880, 10.5194/amt-5-2867-2012, 2012.Burrows, J. P., Hölzle, E., Goede, A. P. H., Visser, H., and Fricke, W.:
SCIAMACHY – scanning imaging absorption spectrometer for atmospheric
chartography, Acta Astronaut., 35, 445–451,
10.1016/0094-5765(94)00278-T, 1995.Chepfer, H., Bony, S., Winker, D., Cesana, G., Dufresne, J. L., Minnis, P., Stubenrauch, C. J., and Zeng, S.:
The GCM Oriented CALIPSO Cloud Product (CALIPSO-GOCCP),
J. Geophys. Res., 115, D00H16, 10.1029/2009JD012251, 2010.Clarisse, L., Coheur, P.-F., Prata, F., Hadji-Lazaro, J., Hurtmans, D., and
Clerbaux, C.: A unified approach to infrared aerosol remote sensing and type
specification, Atmos. Chem. Phys., 13, 2195–2221,
10.5194/acp-13-2195-2013, 2013.Cziczo, D. J., Froyd, K. D., Hoose, C., Jensen, E. J., Diao, M., Zondlo, M.
a, Smith, J. B., Twohy, C. H., and Murphy, D. M.: Clarifying the dominant
sources and mechanisms of cirrus cloud formation, Science, 340,
1320–1324, 10.1126/science.1234145,
2013.Davis, S. M., Liang, C. K., and Rosenlof, K. H.: Interannual variability of
tropical tropopause layer clouds, Geophys. Res. Lett., 40, 2862–2866,
10.1002/grl.50512, 2013.Doeringer, D., Eldering, A., Boone, C. D., González Abad, G., and
Bernath, P. F.: Observation of sulfate aerosols and SO2 from the Sarychev
volcanic eruption using data from the Atmospheric Chemistry Experiment (ACE),
J. Geophys. Res., 117, D03203, 10.1029/2011JD016556, 2012.Ebojie, F., von Savigny, C., Ladstätter-Weißenmayer, A., Rozanov, A.,
Weber, M., Eichmann, K.-U., Bötel, S., Rahpoe, N., Bovensmann, H., and
Burrows, J. P.: Tropospheric column amount of ozone retrieved from SCIAMACHY
limb-nadir-matching observations, Atmos. Meas. Tech., 7, 2073–2096,
10.5194/amt-7-2073-2014, 2014.
European Space Agency: Readme file for SCIAMACHY Level 2 version 5.02
products, Issue 1.2 (ENVI-GSOP-EOGD-QD-13-0118), 2013.Fischer, H., Birk, M., Blom, C., Carli, B., Carlotti, M., von Clarmann, T.,
Delbouille, L., Dudhia, A., Ehhalt, D., Endemann, M., Flaud, J. M., Gessner,
R., Kleinert, A., Koopman, R., Langen, J., López-Puertas, M., Mosner, P.,
Nett, H., Oelhaf, H., Perron, G., Remedios, J., Ridolfi, M., Stiller, G., and
Zander, R.: MIPAS: an instrument for atmospheric and climate research, Atmos.
Chem. Phys., 8, 2151–2188, 10.5194/acp-8-2151-2008, 2008.Fueglistaler, S., Dessler, A., Dunkerton, T., Folkins, I., Fu, Q., and Mote, P.:
Tropical Tropopause Layer,
Rev. Geophys., 47, 1–31, 10.1029/2008RG000267, 2009.Generoso, S., Bey, I., Labonne, M., and Breon, F.-M.: Aerosol vertical
distribution in dust outflow over the Atlantic: Comparisons between GEOS-Chem
and Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observation
(CALIPSO), J. Geophys. Res., 113, D24209, 10.1029/2008JD010154, 2008.Global Volcanism Program: Report on Rabaul (Papua New Guinea), in: Bulletin
of the Global Volcanism Network, edited by: Wunderman, R., 31, 9, Smithsonian
Institution, 10.5479/si.GVP.BGVN200609-252140, 2006.
Gottwald, M., Krieg, E., von Savigny, C., Noël, S., Bovensmann, H., and
Bramstedt, K.: Determination of SCIAMACHY Line of Sight Misalignments,
Proceedings of the Envisat Atmospheric Science Conference, ESA SP-636,
Montreux, Switzerland, 23–27 April 2007.
Gottwald, M., Krieg, E., Lichtenberg, G., Slijkhuis, S., von Savigny, C.,
Noël, S., Bovensmann, H., and Bramstedt, K.: The Status of the SCIAMACHY
Line-of-Sight Pointing Knowledge, Proc. ESA Living Planet Symposium, 28
June–2 July 2010, SP-686, Bergen, Norway, 2010.
Gottwald, M. and Bovensmann, H.: SCIAMACHY – Exploring the Changing Earth's
Atmosphere, Springer, Dordrecht, Heidelberg, London, New York, ISBN: 978-90-481-9895-5, 2011.Greenhough, J., Remedios, J. J., Sembhi, H., and Kramer, L. J.: Towards cloud
detection and cloud frequency distributions from MIPAS infra-red
observations, Adv. Space Res., 36, 800–806, 10.1016/j.asr.2005.04.096,
2005.Haywood, J. M., Jones, A., Clarisse, L., Bourassa, A., Barnes, J., Telford,
P., Bellouin, N., Boucher, O., Agnew, P., Clerbaux, C., Coheur, P.,
Degenstein, D., and Braesicke, P.: Observations of the eruption of the
Sarychev volcano and simulations using the HadGEM2 climate model, J. Geophys.
Res., 115, D21212, 10.1029/2010JD014447, 2010.Höpfner, M., Glatthor, N., Grabowski, U., Kellmann, S., Kiefer, M.,
Linden, A., Orphal, J., Stiller, G., von Clarmann, T., Funke, B., and Boone,
C. D.: Sulfur dioxide (SO2) as observed by MIPAS/Envisat: temporal
development and spatial distribution at 15–45 km altitude, Atmos. Chem.
Phys., 13, 10405–10423, 10.5194/acp-13-10405-2013, 2013.IPCC, 2013: Climate Change 2013: The Physical Science Basis, in: Contribution
of Working Group I to the Fifth Assessment Report of the Intergovernmental
Panel on Climate Change, edited by: Stocker, T. F., Qin, D., Plattner,
G.-K., Tignor, M., Allen, S. K., Boschung, J., Nauels, A., Xia, Y., Bex, V., and
Midgley, P. M., Cambridge University Press, Cambridge, United Kingdom and New
York, NY, USA, 1535 pp., 10.1017/CBO9781107415324, 2013.Kaiser, J. W., von Savigny, C., Eichmann, K.-U., Noël, S., Bovensmann, H., Frerick, J., and Burrows, J. P.:
Satellite-pointing retrieval from atmospheric limb-scattering of solar UV-B radiation,
Can. J. Phys., 82, 1041–1052, 10.1139/p04-071, 2004b.Koelemeijer, R. B. A., Stammes, P., Hovenier, J. W., and de Haan, J. F.: Global
distributions of effective cloud fraction and cloud top pressure derived from
oxygen A band spectra measured by the Global Ozone Monitoring Experiment:
Comparison to ISCCP data, J. Geophys. Res., 107, 4151,
10.1029/2001JD000840, 2002.Kokhanovsky, A. A.:
Optical properties of terrestrial clouds,
Earth-Sci. Rev., 64, 189–241, 10.1016/S0012-8252(03)00042-4, 2004.Kokhanovsky, A. A., Rozanov, V. V., Burrows, J. P., Eichmann, K. U., Lotz, W., and Vountas, M.:
The SCIAMACHY cloud products: Algorithms and examples from ENVISAT,
Adv. Space Res., 36, 789–799, 10.1016/j.asr.2005.03.026, 2005.Kokhanovsky, A. A., Jourdan, O., and Burrows, J. P.:
The cloud phase discrimination from a satellite,
IEEE Geosci. Remote Sens. Lett., 3, 103–106, 10.1109/LGRS.2005.858487, 2006.Kravitz, B., Robock, A., Bourassa, A., Deshler, T., Wu, D., Mattis, I.,
Finger, F., Hoffmann, A., Ritter, C., Bitar, L., Duck, T. J., and Barnes, J.
E.: Simulation and observations of stratospheric aerosols from the 2009
Sarychev volcanic eruption, J. Geophys. Res., 116, D18211,
10.1029/2010JD015501, 2011.Langowski, M. P., von Savigny, C., Burrows, J. P., Feng, W., Plane, J. M. C.,
Marsh, D. R., Janches, D., Sinnhuber, M., Aikin, A. C., and Liebing, P.:
Global investigation of the Mg atom and ion layers using SCIAMACHY/Envisat
observations between 70 and 150 km altitude and WACCM-Mg model results,
Atmos. Chem. Phys., 15, 273–295, 10.5194/acp-15-273-2015, 2015.Leahy, L. V., Wood, R., Charlson, R. J., Hostetler, C. A., Rogers, R. R.,
Vaughan, M. A., and Winker, D. M.: On the nature and extent of optically thin
marine low clouds, J. Geophys. Res.-Atmos., 117, 1–20,
10.1029/2012JD017929, 2012.Lelli, L., Kokhanovsky, A. A., Rozanov, V. V., Vountas, M., Sayer, A. M., and
Burrows, J. P.: Seven years of global retrieval of cloud properties using
space-borne data of GOME, Atmos. Meas. Tech., 5, 1551–1570,
10.5194/amt-5-1551-2012, 2012.Lelli, L.: Studies of global cloud field using measurements of GOME,
SCIAMACHY and GOME-2, PhD thesis, IUP, 2013, available at:
http://elib.suub.uni-bremen.de/edocs/00103470-1.pdf, 2013.Lelli, L., Kokhanovsky, A. A., Rozanov, V. V., Vountas, M., and Burrows, J.
P.: Linear trends in cloud top height from passive observations in the oxygen
A-band, Atmos. Chem. Phys., 14, 5679–5692, 10.5194/acp-14-5679-2014,
2014.Liebing, P., Bramstedt, K., Noël, S., Rozanov, V., Bovensmann, H., and
Burrows, J. P.: Polarization data from SCIAMACHY limb backscatter
observations compared to vector radiative transfer model simulations, Atmos.
Meas. Tech., 6, 1503–1520, 10.5194/amt-6-1503-2013, 2013.
Liou, K. N.: Cirrus clouds and climate, McGraw-Hill 2005 Yearbook of Science
and Technology, USA, 51–53, ISBN: 063-9785511649, 2005.Liu, D., Wang, Z., Liu, Z., Winker, D., and Trepte, C.: A height resolved
global view of dust aerosols from the first year CALIPSO lidar measurements,
J. Geophys. Res., 113, D16214, 10.1029/2007JD009776, 2008.Massie, S., Gille, J., Khosravi, R., Lee, H., Kinnison, D., Francis, G.,
Nardi, B., Eden, T., Craig, C., Halvorson, C., Coffey, M., Packman, D.,
Cavanaugh, C., Craft, J., Dean, V., Ellis, D., Barnett, J., Hepplewhite, C.,
Lambert, A., Manney, G., Strawa, A., and Legg, M.: High Resolution Dynamics
Limb Sounder observations of polar stratospheric clouds and subvisible
cirrus, J. Geophys. Res., 112, D24S31, 10.1029/2007JD008788, 2007.Mateshvili, N., Fussen, D., Mateshvili, G., Mateshvili, I., Vanhellemont, F.,
Kyrölä, E., Tukiainen, S., Kujanpää, J., Bingen, C., Robert,
C., Tétard, C., and Dekemper, E.: Nabro volcano aerosol in the
stratosphere over Georgia, South Caucasus from ground-based spectrometry of
twilight sky brightness, Atmos. Meas. Tech., 6, 2563–2576,
10.5194/amt-6-2563-2013, 2013.Mattis, I., Siefert, P., Müller, D., Tesche, M., Hiebsch, A., Kanitz, T.,
Schmidt, J., Finger, F., Wandinger, U., and Ansmann, A.: Volcanic aerosol
layers observed with multiwavelength Raman lidar over central Europe in
2008–2009, J. Geophys. Res., 115, D00L04, 10.1029/2009JD013472, 2010.Mülmenstädt, J., Sourdeval, O., Delanoë, J., and Quaas, J.: Frequency
of occurrence of rain from liquid-, mixed-, and ice-phase clouds derived from
A-Train satellite retrievals, Geophys. Res. Lett., 42, 6502–6509,
10.1002/2015GL064604, 2015.Normand, E. N., Wiensz, J. T., Bourassa, A. E., and Degenstein, D. A.: Cloud
discrimination in probability density functions of limb-scattered sunlight
measurements, Atmos. Meas. Tech., 6, 3359–3368, 10.5194/amt-6-3359-2013,
2013.
Raspollini, P. and Ceccherini, S.:
Configuration Management of MIPAS L2 Auxiliary Data Files,
Tech note, TN-IFAC-GS0302, issue 6, 2011.Remer, L. A., Kleidman, R. G., Levy, R. C., Kaufman, Y. J., Tanré, D.,
Mattoo, S., Martins, J. V., Ichoku, C., Koren, I., Yu, H., and Holben, B. N.:
Global aerosol climatology from the MODIS satellite sensors, J. Geophys.
Res., 113, D14S07, 10.1029/2007JD009661, 2008.Reuter, M. and Pfeifer, S.: Moments from space captured by MSG SEVIRI, Int.
J. Remote Sens., 32, 4131–4140, 10.1080/01431161.2011.566288, 2011.
Rossow, W. B. and Schiffer, R. A.: Advances in understanding clouds from
ISCCP, B. Am. Meteorol. Soc., 80, 2261–2287, 1999.Rozanov, A., Weigel, K., Bovensmann, H., Dhomse, S., Eichmann, K.-U., Kivi,
R., Rozanov, V., Vömel, H., Weber, M., and Burrows, J. P.: Retrieval of
water vapor vertical distributions in the upper troposphere and the lower
stratosphere from SCIAMACHY limb measurements, Atmos. Meas. Tech., 4,
933–954, 10.5194/amt-4-933-2011, 2011.Rozanov, V. V. and Kokhanovsky, A. A.: Semianalytical cloud retrieval
algorithm as applied to the cloud top altitude and the cloud geometrical
thickness determination from top-of-atmosphere reflectance measurements in
the oxygen A band, J. Geophys. Res., 109, D05202, 10.1029/2003JD004104,
2004.Rozanov, V. V., Rozanov, A. V., Kokhanovsky, A. A., and Burrows, J. P.:
Radiative transfer through terrestrial atmosphere and ocean: Software package
SCIATRAN, J. Quant. Spectrosc. R., 133, 13–71,
10.1016/j.jqsrt.2013.07.004, 2014.
Sassen, K. and Cho, B. S.: Subvisual-Thin Cirrus Lidar Dataset for Satellite
Verification and Climatological Research, J. Appl. Meteor., 31, 1275–1285,
1992.Sassen, K., Wang, Z., and Liu, D.:
Cirrus clouds and deep convection in the tropics: Insights from CALIPSO and CloudSat,
J. Geophys. Res., 114, D00H06, 10.1029/2009JD011916, 2009.Schmetz, J., Pili, P., Tjemkes, S., Just, D., Kerkmann, J., Rota, S., and
Ratier, A.: An Introduction to Meteosat Second Generation (MSG), B. Am.
Meteorol. Soc., 83, 977–992,
10.1175/1520-0477(2002)083<0977:AITMSG>2.3.CO;2, 2002.Sembhi, H., Remedios, J., Trent, T., Moore, D. P., Spang, R., Massie, S., and
Vernier, J.-P.: MIPAS detection of cloud and aerosol particle occurrence in
the UTLS with comparison to HIRDLS and CALIOP, Atmos. Meas. Tech., 5,
2537–2553, 10.5194/amt-5-2537-2012, 2012.Sonkaew, T., Rozanov, V. V., von Savigny, C., Rozanov, A., Bovensmann, H.,
and Burrows, J. P.: Cloud sensitivity studies for stratospheric and lower
mesospheric ozone profile retrievals from measurements of limb-scattered
solar radiation, Atmos. Meas. Tech., 2, 653–678, 10.5194/amt-2-653-2009,
2009.
Spang, R., Riese, M., Eidmann, G., Offermann, D., Pfister, L., and Wang, P. H.:
CRISTA observations of cirrus clouds around the tropopause,
J. Geophys. Res., 107, 8174, 10.1029/2001JD000698, 2002.
Spang, R., Remedios, J. J., Tilmes, S., and Riese, M.: MIPAS observation of
polar stratospheric clouds in the Arctic 2002/2003 and Antarctic 2003
winters, Adv. Space Res., 36, 868–878, 2005.Spang, R., Arndt, K., Dudhia, A., Höpfner, M., Hoffmann, L., Hurley, J.,
Grainger, R. G., Griessbach, S., Poulsen, C., Remedios, J. J., Riese, M.,
Sembhi, H., Siddans, R., Waterfall, A., and Zehner, C.: Fast cloud parameter
retrievals of MIPAS/Envisat, Atmos. Chem. Phys., 12, 7135–7164,
10.5194/acp-12-7135-2012, 2012.
Stubenrauch, C., Rossow, W., Kinne, S., and the GEWEX cloud assessment group:
Assessment of Global Cloud Datasets from Satellites report, WCRP Report No.
23, 2012.Stubenrauch, C. J., Rossow, W. B., Kinne, S., Ackerman, S., Cesana, G.,
Chepfer, H., Di Girolamo, L., Getzewich, B., Guignard, A., Heidinger, A.,
Maddux, B. C., Menzel, W. P., Minnis, P., Pearl, C., Platnick, S., Poulsen,
C., Riedi, J., Sun-Mack, S., Walther, A., Winker, D., Zeng, S., and Zhao, G.:
Assessment of Global Cloud Datasets from Satellites: Project and Database
Initiated by the GEWEX Radiation Panel, B. Am. Meteorol. Soc., 94,
1031–1049, 10.1175/BAMS-D-12-00117.1, 2013.von Savigny, C., Kaiser, J. W., Bovensmann, H., Burrows, J. P., McDermid, I.
S., and Leblanc, T.: Spatial and temporal characterization of SCIAMACHY limb
pointing errors during the first three years of the mission, Atmos. Chem.
Phys., 5, 2593–2602, 10.5194/acp-5-2593-2005, 2005.von Savigny, C., Ulasi, E. P., Eichmann, K.-U., Bovensmann, H., and Burrows,
J. P.: Detection and mapping of polar stratospheric clouds using limb
scattering observations, Atmos. Chem. Phys., 5, 3071–3079,
10.5194/acp-5-3071-2005, 2005.Wang, P.-H., Minnis, P., McCormick, M. P., Kent, G. S., and Skeens, K. M.: A
6-year climatology of cloud occurrence frequency from Stratospheric Aerosol
and Gas Experiment II observations (1985–1990), J. Geophys. Res., 101,
29407, 10.1029/96JD01780, 1996.Weigel, K., Rozanov, A., Azam, F., Bramstedt, K., Damadeo, R., Eichmann,
K.-U., Gebhardt, C., Hurst, D., Kraemer, M., Lossow, S., Read, W., Spelten,
N., Stiller, G. P., Walker, K. A., Weber, M., Bovensmann, H., and Burrows, J.
P.: UTLS water vapour from SCIAMACHY limb measurementsV3.01 (2002–2012),
Atmos. Meas. Tech., 9, 133–158, 10.5194/amt-9-133-2016,
2016.Winker, D. M. and Trepte, C. R.: Laminar cirrus observed near the tropical
tropopause by LITE, Geophys. Res. Lett., 25, 3351–3354,
10.1029/98GL01292, 1998.Winker, D. M., Vaughan, M. A., Omar, A., Hu, Y., Powell, K. A., Liu, Z.,
Hunt, W. H., and Young, S. A.: Overview of the CALIPSO mission and CALIOP
data processing algorithms, J. Atmos. Ocean. Techn., 26, 2310–2323,
10.1175/2009JTECHA1281.1, 2009.