The version 3.00 release of the lidar level 3 aerosol profile product contains
several improvements to the original version 1 release. New definitions for sky
condition now partition the atmosphere into a consistent set of monthly mean
aerosol extinction profiles which yield more useful representations of vertical
aerosol loading compared to sky conditions in version 1 which were difficult to
interpret. Single-species averages for polluted dust and smoke have been added
using an improved averaging strategy which generates level 3 aerosol extinction
profiles with shapes that more faithfully represent the input level 2 extinction
profiles. A revised aerosol optical depth (AOD) calculation removes low-biases
that existed in level 3 monthly mean AOD. Mean aerosol extinction near the
planetary surface has been improved by (a) mitigating the impact of negative
signal anomalies, which cause negative spikes in aerosol extinction at the onset
of the surface, and (b) correctly compensating for high biases (i.e., premature
base determination) in the level 2 layer base detection algorithm. Two new
extinction scale height data sets have been added to identify altitudes below
which 63% and 90% of the total mean AOD is found. Several rarely used statistics
have also been removed. The major version number for this release (version 3)
reflects that version 3 level 2 data products have been used as input for the
level 3 product. These improvements are described in greater detail below.
Level 2 Input Data
The following table lists the version numbers of level 2 data used as input
for the V3.00 level 3 product. Note that the major version number (3) matches
between level 2 and level 3. Minor version number changes to the level 2 products
occur at boundaries between months, so a given level 3 file will only contain
input for a single version of the level 2 product.
Lidar Level 2 Input Data For Level 3 Aerosol Profile Product, V3.00
Level 2 Version
Data Date Range
March 2013 - present
November 2011 - February 2013
June 2006 - October 2011
Level 2 Input Data
Level 3 averages are reported for four different sky conditions. “Sky
condition” refers to occurrence and location of clouds within the level 2
profiles which are included in the level 3 average. The definitions for sky
conditions have changed in V3.00:
Sky Condition (V3)
Description of Level 2 Data Usage
All level 2 columns are averaged, regardless of cloud occurrence
Only cloud-free level 2 columns are averaged
Only level 2 columns containing transparent clouds are averaged
Only level 2 columns containing opaque clouds are averaged
An example of how an individual level 2 granule is parsed into the four sky
conditions is shown in Figure 1. Note that it is possible to derive mean extinction
for all sky from the cloud-free, cloudy-sky transparent and cloudy-sky opaque sky
conditions by weighting the mean extinction values by the “Samples Averaged”
statistic for each sky condition.
Figure 1. Level 3 Version 3 sky conditions applied to a single level 2
granule. White indicates columns which are excluded from the particular sky
New single aerosol species averages
Extinction, AOD and sampling statistics are now included for polluted dust-only
and smoke-only in addition to dust-only which was reported in version 1.
Revised averaging method for single aerosol species averages
When averaging level 2 extinction for single aerosol species (for example dust-only),
the aerosol extinction for all other aerosol species are now assigned
extinction = 0.0 km-1 prior to calculating the average. In version 1
of the level 3 product, extinction for those other aerosol species were ignored
in the average, which introduced a high bias into the mean extinction at altitudes
dominated by other aerosol species (Amiridis et al., 2013).
Figure 2 below demonstrates this effect. Saharan dust transporting westward
during Boreal summer initially remains elevated and then descends gradually into
the marine boundary layer (MBL) towards the Caribbean (Liu et al, 2008).
Dust-only extinction should therefore peak above the MBL just west of the North
Africa coast. In level 3 version 1, the dust-only extinction profile in this region
shows two peaks - one at 4 km and a second, larger peak in the MBL. The lower peak
occurs because the fraction of dust samples decreases rapidly near the ocean
surface where marine aerosol is the dominant species (Figure 1, right panel).
Since the marine aerosol extinction was ignored during the averaging process, the
mean dust-only extinction near the surface is magnified because numerous
“dust-free” extinction values of 0.0 km-1
(i.e., where the aerosol is marine, not dust) are not included in the averages.
Crucially, the profile shape for dust-only extinction becomes badly deformed
solely because a small amount of dust existed at the same altitude as marine aerosol.
This deformation is not an accurate portrayal of the profile shapes reported by
the CALIOP level 2 aerosol profile product, but is instead an artifact of the
single aerosol species averaging procedure used in the version 1 analyses. In
the version 3 product, all non-dust aerosol extinction is set to 0.0 km-1
during the averaging process which yields a dust-only extinction profile with a
single peak near 4 km that is more representative of dust-only extinction in this
region. A similar methodology is implemented for polluted dust-only and smoke-only
averages in level 3 version 3.
Figure 2: Level 3 mean dust-only extinction averaged according to level 3
version 1 and version 3 methodologies (left panel) and the fraction of dust
samples out of all possible aerosol subtypes (right panel) for July 2008 at
night. Inset map indicates region examined (red box).
Amiridis, V., U. Wandinger, E. Marinou, E. Giannakaki, A. Tsekeri, S. Basart,
S. Kazadzis, A. Gkikas, M. Taylor, J. Baldasano and A. Ansmann, 2013:
“Optimizing Saharan dust CALIPSO retrievals”, Atmos. Chem. Phys.,
13, 12089-12106, doi:10.5194/acp-13-12089-2013.
Liu, Z., et al. (2008), “CALIPSO lidar observations of the optical
properties of Saharan dust: A case study of long-range transport”, J. Geophys. Res.,
Revised AOD mean computation
Mean aerosol optical depth (AOD) reported in version 3 of the level 3 product
is computed as the vertical integral of the mean level 3 extinction profile. In
contrast, version 1 mean aerosol optical depth was computed as the average of
level 2 column AODs. The version 1 method caused a low-biased mean AOD because
the columns being averaged together have unequal geometric depths due to cloud
cover, opacity or data quality filtering (Figure 3). A simple example demonstrates
this. Imagine two level 2 aerosol extinction profiles - one profile having valid
aerosol extinction all the way down to the surface and the other profile having
an opaque cloud with a small amount of aerosol extinction above. Computing the
vertical integral for these two profiles would yield one relatively large AOD
and one small AOD since there is less aerosol above cloud. Averaging these AODs
together would give a rather small AOD. Conversely, averaging these two extinction
profiles together and then integrating the mean extinction would yield a larger
AOD that is representative of the aerosol loading at all altitudes. This is the
version 3 method. It also has the benefit that the mean AOD is consistent with
the mean extinction profiles reported.
Figure 3: Version 1 Level 3 AOD for one month calculated according to the
version 1 method (average of column AODs) and the version 3 method (integral of
New extinction scale heights
New science data sets, “AOD 63% Below” and
“AOD 90% Below” indicate the altitude which contains 63% and 90%
of the total mean aerosol optical depth below.
Revised low-bias filter for undetected surface-attached aerosol
When the CALIPSO feature detection algorithm does not detect the entire vertical
extent of an aerosol layer and/or the aerosol base extension algorithm is invoked,
a gap of “clear-air” can exist beneath the layer base and the surface.
If left as “clear-air”, level 3 mean extinction would exhibit a low
bias in range bins near the surface. To guard against this in version 1 of the
level 3 product, “clear-air” was ignored below the lowest layer in a
column whose base was below 2.5 km. However, setting this threshold altitude at
2.5 km proved to be overly aggressive, and would occasionally introduce high
biases into the level 3 profiles. With version 3, the lowest layer base altitude
threshold is lowered to 250 meters, which is much more consistent with the
“clear-air” gap of 90-120 meters from the surface that is seen in
the level 2 products. This revision focuses the filter application more precisely
where it is needed rather than at higher altitudes which, as mentioned, can
otherwise yield a high bias in level 3 mean extinction.
New negative signal anomaly mitigation strategy
A recently discovered phenomenon, dubbed a “negative signal anomaly”
sometimes occurs when the level 1B attenuated backscatter becomes anomalously
negative at the abrupt onset of a strongly scattering target such as the planetary
surface or a dense cloud. Aerosol extinction can also become anomalously negative
when attenuated backscatter affected by the anomaly is included in the extinction
retrieval. Subsequent level 3 mean extinction would therefore exhibit a low bias
near the surface where the negative signal anomaly is present. To mitigate the
impact of this anomaly, all level 2 aerosol extinction coefficients within 60
meters of the planetary surface are excluded from level 3 calculations. Globally,
this yields a 2-5% decrease in AOD, but increases AOD in regions strongly affected
by the negative signal anomaly. Crucially, anomalously low spikes in level 3
extinction in lowest altitude range bins have been removed.
The following statistics have been removed from the level 3 aerosol profile
product compared to V1.x.
AOD and AOD dust statistics were previously reported for all four sky
conditions in every level 3 file. In V3.00, AOD is only reported for the sky
condition of the level 3 file.
The following sections describe issues with the level 3 product known at the
time of its release in October 2015.
Undetected aerosol influence on level 3 aerosol extinction statistics
Regions classified as “clear air” by the CALIPSO feature detection
algorithms are assumed to have aerosol extinction = 0.0 km-1 when
calculating level 3 averages of aerosol extinction. CALIOP has a limited detection
sensitivity however, and weakly scattering layers go undetected, resulting in a
low bias. This occurs most often in the upper troposphere where aerosol layers
tend to be optically weak, in the daytime since the SNR is smaller than at
nighttime resulting in more layer misdetection, and beneath optically thick
clouds due to significant signal attenuation. More details of why some aerosol
features may go undetected can be found in the level 2 aerosol layer data quality
summary. Biases due to missed detections are believed to be small at night in most
regions, though potential biases have not yet been comprehensively quantified
by validation intercomparisons. Biases can be more pronounced in the daytime
when strongly scattering layers are not detected due to enhanced solar noise over
bright surfaces. The impact of this effect has not been quantified at this time.
Meteorological context statistics not calculated for sky condition
Currently, each meteorological context parameter is generated for every grid
cell by averaging together all profiles of the parameter that exist within the
grid cell together, regardless of cloud cover or aerosol quality screening. As
a consequence, there is not a one-to-one correspondence between the region of
sky used to generate aerosol extinction statistics and the region of sky used
to generate the meteorological context statistics. For example, the aerosol
extinction reported for the “Cloud-Free” sky condition will not
include extinction samples in cloudy columns, but the Temperature statistics
(for example), and all other meteorological parameters would still contain those
samples in cloudy columns.
Samples Searched statistic does not include cloudy samples
The “Samples Searched” statistic is intended to represent the
number of atmospheric samples searched by the level 3 algorithm regardless of
whether they were accepted, rejected or ignored in the averaging process. This
statistic should include cloudy samples, but it does not in the V3.00 release.
This will be corrected in a forthcoming release.