Introduction
In general, the contribution of aerosols to atmospheric processes is not
fully documented. In particular, an important gap needs to be filled
to clarify the role of aerosols in the Earth radiation budget and in
climate change . The aerosols' high variability in terms of type, time and space makes
it quite difficult to understand the atmospheric processes in
which aerosols are involved . Therefore, there is a strong need from the scientific community to have access to
comprehensive aerosol data sets in which vertically resolved aerosol
optical parameters can be found. Lidar measurements, providing high-resolution
profiles (in both space and time) of aerosol optical properties, meet
this demand entirely as they allow the full characterization of each layer
present in the atmosphere.
Another important aspect for the study of aerosols on a planetary scale is the increased spatial coverage. To
support this need, several coordinated lidar networks have been established in the
last years e.g.,. In particular, EARLINET (European Aerosol
Research Lidar Network) has been operated in Europe since the year 2000 and provides
the scientific community with the most complete database of vertically
resolved aerosol optical parameters across Europe . The EARLINET data can be used for
several purposes including model evaluation and assimilation, full
exploitation of satellite data, the study of aerosol long-range transport
mechanisms, and the monitoring of special events like volcanic eruptions, large forest
fires or dust outbreaks.
Within the EARLINET-ASOS (European Aerosol
Research Lidar Network – Advanced Sustainable Observation System) project, great attention was
paid to the optimization of lidar data processing (http://www.earlinetasos.org). The core of this
activity was the development of the EARLINET Single Calculus Chain (SCC), a tool
for the automatic evaluation of lidar data from raw signals up to the final
products. The main advantage of this approach is that it increases the rate of population of the aerosol database (which is the
main outcome of any lidar network) and to promote the usage of
vertically resolved aerosol parameters within the scientific community.
This paper is the first of three publications about the SCC and it
presents an overview of the SCC and its validation. Two separate
papers are used to describe the technical details of the SCC
pre-processing module and of the optical processing
module , respectively.
A general overview of the SCC is provided in Sect. of this paper. Section
illustrates the SCC structure by providing technical details of all
SCC modules. The strategy adopted to validate the SCC is described in Sect. , and, finally, an example of the application of the SCC
as a tool to provide network lidar data in near-real time is given in Sect. .
SCC description
The SCC is an official EARLINET tool. It has been developed to accomplish the fundamental need of any
coordinated lidar network to have an optimized and automatic tool
providing high-quality aerosol properties. Currently, it has been used
by 20 different EARLINET stations which have submitted about 2600 raw data files covering a very large time
period (2001–2015). Moreover, more than 5000 SCC optical products (about
3600 aerosol backscatter profiles and 1400 aerosol extinction profiles) have been
calculated and used for different purposes like analysis of instrument
intercomparisons , air-quality model assimilation
experiment , and ongoing long-term comparisons with
manually retrieved products . The large usage and the
long-term plan for the centralized processing system make the SCC the
standard tool for the automatic analysis of EARLINET lidar data.
General considerations
Main concepts at the base of the SCC are automatization
and fully traceability of quality-assured aerosol
optical products. At network level, the SCC ensures high-quality products by
implementing quality checks on both raw lidar data and final optical
products. Such quality checks are part of a rigorous quality assurance
program developed within EARLINET. In many specific situations, it is also quite important that the retrieved
products are available in real time or in near-real time for large
geographical areas (on a continental scale). For example, this is the
case when vertically resolved lidar products are used to improve the
forecast of air-quality models, to validate satellite sensors
or models, or to monitor special events. Without a common analysis
tool it could be difficult to assure at the same time homogenous
high-quality products and short-time availability of the data, because
high-quality manual lidar data analysis usually requires time and
manpower. Moreover, different groups within the network may use
different retrieval approaches to derive the same type of aerosol
parameter with a consequent loss in the homogeneity of the network data set.
At the same time, in order to make the use of the SCC really
sustainable, expandability and flexibility should be assured to
guarantee the analysis of the data measured by new or upgraded lidar
systems. Excluding few exceptions, a lidar network is usually formed by different
and not standardized lidar systems ranging from single-wavelength elastic-backscatter lidar to advanced
multi-wavelength Raman systems. A system is frequently improved or
upgraded from a basic configuration to a more complex one by adding,
for example, new detection channels. As a consequence, the SCC must be able to handle data acquired by different instruments which usually require different
instrumental corrections and also different approaches to get quality-assured
products. EARLINET is a good example showing how heterogenous the lidar systems
forming a network can be. Most of the EARLINET lidar systems are home-made
or highly customized, and typically they differ in terms of
emitted or detected wavelengths, acquisition mode (analog and/or
photon-counting), space and time resolution, and detection
systems. A network like AERONET does not suffer from this problem as it is based on the same standardized
instrument. Therefore, a common scheme for
the analysis of raw data does not need to take many different
instrumental aspects into account and, thus, allows for reduced development complexity.
In addition, the EARLINET quality assurance program
on both instrumental and
algorithm levels puts more
constraints on the SCC development. In particular, it is required that
each SCC product has been measured with a lidar system that passed the instrumental quality
assurance tests, and it has been calculated applying certified
algorithms.
With the SCC it is possible to calculate aerosol extinction
and backscatter coefficient profiles. Especially in case
of multi-wavelength lidar measurements, this set of optical parameters can provide a full characterization
of atmospheric aerosol from both quantitative and qualitative
point of view .
Moreover, these products can be used
as input to infer microphysical properties of atmospheric
particles .
It is important to stress that two independent SCC modules for the
retrieval of microphysical properties of the atmospheric aerosols have
been already developed . The main products of these modules are
particle effective radius, volume concentration, and refractive index,
which are calculated with a semi-automated and unsupervised algorithm. Although operational
versions of these modules have been released, they are not
included in the automatic structure of the SCC yet. Mainly,
instability problems make the full automatization of lidar
microphysical retrievals a quite challenging task.
The high flexibility and expandability of the SCC also makes it possible
to use the tool in a more general context. As EARLINET already represents a quite complete example of
all available lidar system types, it is expected to adapt the SCC
easily to run in more extended networks like GALION (GAW Aerosol
LIdar Observation Network).
To our knowledge, the SCC is the first tool that can be used to analyse
raw data measured by many different types of lidar systems in a fully automatic way. Other existing automatic tools for the analysis
of lidar data are usable only by specific lidar systems and cannot
be easily extended to retrieve aerosol properties of whole lidar
networks composed by different instruments. Another
unique characteristic of the SCC is that its aerosol optical products are
delivered according to a rigorous quality assurance program to
provide always the highest possible quality for products at network level.
Requirements
In this section the requirements to accomplish all key
points explained in the previous section are described.
In the framework of the EARLINET quality assurance program several algorithms for the retrieval of aerosol optical parameters
have been inter-compared to evaluate their performances in providing
high-quality aerosol optical products .
This inter-comparison was mainly addressed
to asses a common European standard for the quality assurance of lidar
retrieval algorithms and to ensure that the data provided by each individual
station are permanently of the highest possible quality according to
common standards. All different quality-assured analysis algorithms developed within EARLINET have been collected, critically evaluated with respect to their general applicability, optimized to make them fully automatic, and finally implemented in the SCC. A critical point was the implementation of reliable and robust algorithms to assure accurate
calibration of aerosol backscatter profiles. In a fully automatic
analysis scenario, particular attention should be devoted to this issue to avoid large inaccuracy in the final
optical products. Noisy raw lidar signals or the presence of aerosol
within the calibration region can induce large errors in the lidar
calibration constant .
The SCC has been developed having in mind the following concepts:
platform independency, open-source philosophy, standard data format (NetCDF), flexibility through the implementation
of different retrieval procedures, expandability to easily include new
systems or new system configurations. All libraries and compilers
needed to install and run the SCC are open source and freely available. The SCC can operate on
a centralized server or on a local PC. The users can connect to the
machine on which the SCC is running and use or configure the SCC
retrieval procedures for their data using a web interface. The
centralized server solution (which is the preferred way of using the
tool) has many advantages compared to local installation, especially when the SCC is used within a coordinated
lidar network as EARLINET. First of all, it is possible to keep track of all
system configurations of all systems and also to certify which
configurations are quality assured. Moreover, in this way it is always
guaranteed that the same and latest SCC version is used to produce optical
products.
Particular attention has been paid to the design of a suitable NetCDF structure for the SCC input file as it needs to
fulfill the following constraints.
It should contain the raw lidar data as they are measured by the
lidar detectors (output voltages for analog lidar channels, counts for
photon-counting channels) without any correction earlier applied by
the user. This is particularly important to assure the quality
of the final products: all necessary instrumental
corrections should be applied by the SCC using quality-assured
procedures. For this reason a specific pre-processing SCC
module has been developed.
It should contain additional input parameters needed for the
analysis. As it will be explained in the next section, the main part
of the required input parameters is efficiently stored in a SCC database. However, there are some parameters easily changing from
measurement to measurement (e.g., electronic background or
number of accumulated laser shots) that usually cannot be stored in a database. The
only way to pass such parameters to the SCC is via the
input file. To improve the self-consistency of the SCC input file, it
has been allowed to include in the file some
important parameters already stored in the SCC database. In case
these parameters are found in the input file these values will be
used in the analysis.
It should also contain a unique method to link the information
contained in the input file with the ones included in the SCC
database. As it will be explained in the next section, this is
assured by the definition of unique channel IDs which identify the
different lidar channels.
It should allow efficient data processing. As the SCC has been
designed to be a multi-user tool it is important to improve the
computational speed as much as possible to avoid long delay in getting the final
products. This has been accomplished by putting the time series of all channels available for a lidar
configuration in a single SCC input
file.
Finally, as the SCC
products need to be uploaded to the EARLINET database, the output file structure is fully
compliant with the structure of EARLINET e-files and
b-files. The e-files contain the particle
extinction coefficient profiles and optionally the backscatter coefficient profiles derived from Raman
observations at the same effective vertical resolution. The
b-files contain the particle backscatter coefficient
profiles derived either from elastic-backscatter signals (Klett
or iterative method) or from the ratio of elastic-backscatter
and nitrogen Raman signals (Raman method) at highest possible vertical resolution. More details about EARLINET
e- and b-files are provided elsewhere
.
SCC structure
Figure shows the general structure of the SCC which
consists of several independent but inter-connected modules. Basically
there is a module responsible for the pre-processing of raw lidar data,
a module for the retrieval of the aerosol extinction and backscatter
profiles, a daemon (computer program running as a background process
without direct control of an interactive user) which automatically starts the pre-processing or the
processing module when it is necessary, a database to collect all
input parameters needed for the analysis, and finally a web
interface. Once the new raw data file is submitted to the SCC via the
web interface, the daemon automatically starts the pre-processing module
and in succession the processing module. The status of the analysis in
each step can be monitored using the web interface and the
pre-processed or the optical results can be downloaded.
Block structure of the Single Calculus Chain.
SCC database
The retrievals of aerosol optical products from
lidar signals require a large number of input parameters to be used in both
pre-processing and processing phase. Two different types of parameters are needed: experimental (which are mainly used
to correct instrumental effects) and configurational (which define the way
to apply a particular analysis procedure). An example of experimental
parameter is the dead time of a photon-counting system . Once measured,
the value of the dead time for a particular photon-counting lidar
channel can be included in the database among the other parameters
that characterize the channel and, consequently, it will be used to
correct the corresponding raw lidar data. The dead time is an example of an experimental parameter that, in general,
changes from channel to channel. There are other experimental
parameters which may be shared by multiple channels, e.g.,
telescope or laser characteristics (several lidar channels usually
share the same laser or the same telescope).
Configuration parameters are the ones used to identify which
algorithm, among the implemented ones, has to be used to calculate
a particular product. In general, there are multiple
quality-assured algorithms in the SCC to calculate a particular aerosol
product. For instance, for the particle backscatter coefficient
profiles derived from elastic-backscatter signals both
the iterative and the Klett method
have been implemented. The
data provider can choose which one to use by setting a corresponding parameter
in the database.
In general, both configuration and experimental parameters can change
from one lidar system to another and, even for the same lidar system,
they can change for the different configurations under which the lidar
can run. For example, a lidar can deliver extinction and backscatter
coefficient profiles from Raman observations in night-time
configuration, whereas elastic-backscatter methods are applied under
daytime conditions.
In this complex context, a relational database represents an optimal
solution to handle, in an efficient way, all this information. For
this reason, a SCC database has been implemented to store the input
parameters for all EARLINET systems and, at the same time, to
access the subset of all parameters associated to a particular lidar
configuration.
A multiple-table MySQL database has been used for that purpose.
In the SCC database, the experimental parameters are grouped in terms
of stations, lidar configurations and lidar
channels. Figure shows a simplified version
of the SCC database structure. Each station is linked to one or
more lidar configurations which in turn are linked to one or more
lidar channels. Moreover, each lidar configuration is
associated also to a set of products that the
SCC should calculate. Basically, the products are specified in
terms of type (e.g., aerosol extinction, backscatter by Raman
method, etc.) and “usecase” which, as it will be explained later,
represents the way to calculate the product. Additionally, for a particular product, it is
possible to fix a set of calculation options, e.g.,
the pre-processing vertical resolution, the backscatter
calibration method, the maximum statistical error we would like to
have on the final products and so on.
Finally, when lidar measurement sessions are submitted to the SCC they are linked
to a specific lidar configuration. In this way, with specific
SCC database queries, it is possible to get any detail needed for
the analysis of the lidar measurements.
On one hand, a so structured database allows us to keep track of all
information used to generate a particular SCC product assuring
the full traceability; on the other hand, it guarantees the
implementation of a reliable and rigorous quality assurance
program at network level.
Pre-processor module (ELPP: EARLINET Lidar Pre-Processor)
The ELPP module implements the corrections to
be applied to the raw lidar signals before they can be used to derive
aerosol optical properties. As the details of this module are described in
here just the main characteristics are reported.
The main reason for which we implemented a pre-processor module
along with a optical processing module is that the EARLINET
quality assurance program does not apply only to the retrieval of
aerosol optical properties but also to the procedures needed to correct
instrumental effects. Moreover, by handling the raw data it
is possible to identify problems in lidar signals that may be not
so evident in already pre-processed signals. The raw lidar
signals have to be submitted in a NetCDF format with a well-defined structure
. In particular, the raw lidar data should consist of
the signal as detected by the lidar detectors. In case of analog detection mode
the signal should be provided in mV, while for photon-counting mode it
should be expressed in pure counts.
According to the specific lidar system and to the input parameters
defined both in the SCC database and in the NetCDF input file, different types of
operations can be applied on raw data. To make the SCC a useful tool
for all EARLINET systems it is required that the pre-processing module
implements all different instrumental corrections defined for the
different EARLINET lidars. The complete description of all
these corrections is given in , here we just report a list
of the most common ones: dead-time correction, trigger-delay correction,
overlap correction, background subtraction (both atmospheric and
electronic).
Besides these corrections, the pre-processor module is also
responsible for generating the molecular signal needed to calculate the
aerosol optical products. This can be done by using a standard
model atmosphere (e.g., US 1976) or correlative radiosounding profiles.
Finally, the pre-processor module implements near- and far-range automatic signal gluing, vertical interpolation, time
averaging and statistical uncertainty propagation .
The outputs of the pre-processor module are intermediate pre-processed NetCDF files
which will be the input files for the optical processor
module. These files contain the pre-processed range-corrected lidar
signals, the statistical uncertainties, and the corresponding molecular atmospheric profiles. As
these quantities can be used in many different fields of application
(quick-look generation, model assimilation, inter-comparison
campaigns) the intermediate NetCDF files can be considered
additional (non-calibrated) products provided by the SCC.
Simplified version of the SCC database structure. Multiple
arrows indicate one-to-many relationship while single arrows
represent one-to-one correspondence.
Two examples of SCC usecases corresponding to the
calculation of particle backscatter coefficient determined
by the use of Raman signals. In particular, the usecase 0 (on the left) can be used for
a lidar system measuring only the elastic backscattered signal
(elT) and the corresponding N2 Raman backscattered signal
(vrRN2). The usecase 13 (on the right) refers to a more complex lidar configuration
in which there are two different telescopes. Four lidar
channels are detected by each telescope: one elastic backscattered
signal split in analog (elTan) and photon-counting (elTpc)
detection channels and one N2 Raman backscattered signal split in analog
(vrRN2an) and photon-counting (vrRN2pc) detection mode.
Optical processor module (ELDA: EARLINET Lidar Data Analyzer)
The ELDA module applies the algorithms for the retrieval of aerosol
optical parameters to the pre-processed signals produced by the
pre-processor module. All details of the ELDA module are described in
, therefore only a very brief overview of its main
functionalities is given here. ELDA can provide aerosol products in a flexible way choosing from a set of possible pre-defined analysis
procedures (usecases). ELDA enables the retrieval of particle
backscatter coefficients by using both the Klett method
and the iterative algorithm
, the calculation of particle extinction
coefficient profiles after the Raman method
, and finally the computation of
particle backscatter coefficient profiles after the Raman method
. An automatic vertical-smoothing and time-averaging
technique selects the optimal resolution as a function of altitude on
the basis of different thresholds on product uncertainties fixed in
the SCC database for each product . The final optical
products are written in NetCDF files with a structure fully compliant
with the EARLINET e-files and b-files.
Usecase
To improve the flexibility of the SCC, the concept of “usecase”
has been introduced. The SCC utilizes the usecases to adapt the analysis
of lidar signals to a specific lidar configuration. Each usecase
identifies a particular way to handle lidar data. An example on how
the usecases are defined is illustrated in Fig. . In
the left part of the figure usecase 0 for the calculation of the
backscatter coefficient after the Raman method is schematically shown. This usecase refers to
a basic Raman lidar configuration where only an elastic
signal (elT) and the corresponding vibrational-rotational N2 Raman
signal (vrRN2) are detected. These two signals are pre-processed by the SCC
pre-processor module and the results are saved in a NetCDF
intermediate file. Then ELDA ingests the preprocessed signals and
delivers the particle backscatter coefficient profile as final result. In the
right part of Fig. a more complex
usecase (the usecase 13) for aerosol backscatter
calculations after the Raman method is reported. It corresponds to a lidar system that uses two different
telescopes: one optimized to detect the signal backscattered by the
near-range atmospheric region and another one optimized to detect the
atmospheric signal from the far range. Moreover, for both
telescopes the elastic and the vibrational-rotational N2 Raman signals are
detected in analog and photon-counting mode. In this case, the SCC should
combine eight raw signals to get a unique particle backscatter
coefficient profile. Looking at Fig. we can see the details
of this combination for the usecase 13. First, the analog and the
corresponding photon-counting signals are combined by the pre-processor
module. This step results in four signals being reported in the
intermediate NetCDF file. These signals correspond to the combined (analog and photon-counting)
elastic and vibrational-rotational N2 Raman signals
detected by the near-range and far-range telescopes. The ELDA
module combines these four pre-processed signals and retrieves two different backscatter
coefficient profiles (one for the near-range and the other for
the far-range). Finally, these products are glued together to get a single
particle backscatter coefficient profile.
A total of 34 different usecases have been defined and implemented
within the SCC for the calculation of all optical
products. A schematic description of all implemented usecases is
provided in the Appendix. This set of usecases
assures that all different EARLINET lidar setups can be processed by
the SCC. Moreover, we may have further flexibility choosing among the
different usecases compatible for a fixed lidar configuration.
Finally, the concept of usecase improves also the expandability of the
SCC: to implement a new lidar configuration in the SCC it is sufficient to
implement a new usecase, if the ones already defined are not compatible
with it.
SCC daemon module
The SCC database, the ELPP and ELDA modules are well separated objects
that need to act in a coordinated and synchronized way. When
a set of raw lidar data is submitted to the SCC a new entry is created in the
SCC database. As soon as this operation is completed, the
pre-processing module should be started to treat the submitted
measurements. As soon as there are pre-processed data available, the
ELDA module should be started to retrieve the aerosol optical
products. All of these operations are performed by the module SCC daemon.
This module is a multithread process running continuously in the
background, and it is responsible to start thread instances for the
pre-processor or the optical processor module when it is necessary.
Another important function of the SCC daemon is to monitor the status
of started modules and to track the corresponding exit status in the
SCC database.
As the SCC is mainly designed to run on a single server where
multiple users can perform different lidar analyses at the same time,
the SCC daemon has been developed to act in a multithread
environment. In this way, different processes can be started in
parallel by the SCC daemon enhancing the efficiency of the whole
SCC.
Comparison of backscatter coefficient profiles at
1064 nm derived with the iterative method for five lidar
systems participating in the EARLI09 inter-comparison campaign. All
profiles refer to the measurement session taken from
21:00 to 23:00 UT on 25 May 2009. The profiles in blue are the analyses provided by
the originator of the data using his/her own analysis software. The
profiles in red are the ones retrieved by the SCC. From left to
right, upper panel: RALI, MARTHA; middle panel: PollyXT,
MSTL-2;
bottom panel: MUSA.
Web interface
This module represents the interface between the raw data provider and the
SCC. In particular, the SCC end-user needs to interact only
with the SCC database because, as already mentioned, all other
analysis procedures are handled by the SCC daemon automatically.
The web interface provides a user-friendly way to interact with
the SCC database by using any of
available web browsers. Via the web interface it is possible
to do the following:
change or visualize all input parameters for a particular
lidar system or add a new system;
upload data to the SCC server and register the measurements in
the SCC database. Along with the raw lidar data it is also possible to
upload ancillary files, e.g., correlative sounding
profiles and overlap correction functions which can be used in the
analysis. All of these files should be in NetCDF format with a well-defined structure. The interface does not allow the upload
of files that are in wrong format or not compliant with the defined
structure;
visualize the status of the SCC analysis. In case of failure a specific error message is shown so that the user can easily figure
out the reason for the failure;
download the pre-processed or the optical processed data from the
server. In particular, it is possible to visualize the calculated
profiles of aerosol optical products;
re-apply the SCC on an already analysed measurement.
The web interface has been developed in a way that the above actions
can be performed depending on different types of accounts. For
instance, users belonging to a particular lidar station cannot modify
any input parameters for a lidar system linked to a different
lidar station. It is also possible, e.g., to define users that can only
perform analysis and cannot change input parameters.
Moreover, the processing status of each measurement can be also
monitored using a web API (application programming interface). Using
this API, the SCC can be tightly integrated to each station
processing system making the process of submission of the raw data and
the corresponding analysis fully automatic.
Finally, using the web interface it is possible to have access to the EARLINET Handbook of Instrumentation (HOI) where all
instrumental characteristics of the lidar systems registered
in the SCC database are reported. The main goal of the HOI is to collect all characteristics
of all EARLINET lidar systems and to make this information available
for the end-user in an efficient and user-friendly way. For this
reason, the information in the HOI is grouped in terms of different
subsystems composing a complete lidar system: laser source,
telescope, spectral separation, acquisition system. Additional
information concerning the station running the lidar system is also
provided, including a history of any changes made to the lidar in question.
Comparison of backscatter coefficient profiles at
355 nm derived with the Raman method for five lidar
systems participating in the EARLI09 inter-comparison campaign. All
profiles refer to the measurement session taken from
21:00 to 23:00 UT on 25 May 2009, and they have been
retrieved combining elastic-backscatter signals at 355 nm
and the corresponding N2 Raman backscatter signals at 387 nm. The profiles in blue are the analyses provided by
the originator of the data using his/her own analysis software. The
profiles in red are the ones retrieved by the SCC. From left to
right, upper panel: RALI, MARTHA; middle panel: PollyXT,
MSTL-2;
bottom panel: MUSA.
Comparison of backscatter coefficient profiles at
532 nm derived with the Raman method for five lidar
systems participating in the EARLI09 inter-comparison campaign. All
profiles refer to the measurement session taken from
21:00 to 23:00 UT on 25 May 2009, and they have been
retrieved combining elastic-backscatter signals at 532 nm
and the corresponding N2 Raman backscatter signals at 607 nm. The profiles in blue are the analyses provided by
the originator of the data using his/her own analysis software. The
profiles in red are the ones retrieved by the SCC. From left to
right, upper panel: RALI, MARTHA; middle panel: PollyXT,
MSTL-2; bottom panel: MUSA.
Comparison of extinction coefficient profiles at
355 nm derived with the Raman method for five lidar
systems participating in the EARLI09 inter-comparison campaign. All
profiles refer to the measurement session taken from
21:00 to 23:00 UT on 25 May 2009, and they have been retrieved using the N2 Raman
backscatter signals at 387 nm. The profiles in blue are the analyses provided by
the originator of the data using his/her own analysis software. The
profiles in red are the ones retrieved by the SCC. From left to
right, upper panel: RALI, MARTHA; middle panel: PollyXT,
MSTL-2; bottom panel: MUSA.
Comparison of extinction coefficient profiles at
532 nm derived with the Raman method for five lidar
systems participating in the EARLI09 inter-comparison campaign. All
profiles refer to the measurement session taken from
21:00 to 23:00 UT on 25 May 2009, and they have been retrieved using the N2 Raman
backscatter signals at 607 nm. The profiles in blue are the analyses provided by
the originator of the data using his/her own analysis software. The
profiles in red are the ones retrieved by the SCC. From left to
right, upper panel: RALI, MARTHA; middle panel: PollyXT,
MSTL-2, bottom panel: MUSA.
Validation
A validation strategy to prove whether the SCC can provide quality-assured
aerosol optical products has been implemented. The performance of the
SCC has been evaluated on both synthetic and real lidar data.
As a first step, the SCC has been tested with synthetic lidar signals
used during the algorithm inter-comparison exercise performed in the
framework of the EARLINET project . This set
of synthetic signals was simulated with really realistic experimental
and atmospheric conditions to test the performance of specific
algorithms for the retrieval of particle extinction and
backscatter coefficient profile. By comparing the calculated profiles with the corresponding
input profiles used to simulate the signals it is possible to verify, if
an implemented algorithm returns reliable results. As the details of
this exercise are provided in we just mention here
that all algorithms implemented within the SCC produce profiles
that agree with the solutions within the statistical uncertainties.
As second validation level, we have evaluated the SCC performance
when it is applied to real lidar data by comparing the optical products
calculated by the SCC with the corresponding optical products
generated by the analysis software developed by different EARLINET lidar
groups and used so far to provide lidar profiles to the EARLINET
database. This comparison has been performed using two different
approaches. First, we compared the analysis for lidar
measurements taken by several lidar systems at the same
place and at the same time as in the case of the EARLI09
(EArlinet Reference Lidar Intercomparison 2009) campaign
. Secondly, we have used
climatological data of two EARLINET stations to evaluate possible biases in the SCC analysis not
visible from the comparison of one single case.
Absolute (d¯) and relative (d¯r)
mean deviations between SCC and corresponding manual analysis. The
parameters d¯ and d¯r are calculated
according to the Eqs. () and (), respectively,
and considering all the profile altitude bins shown in
Figs. –. Backscatter
coefficient absolute differences are expressed
in Mm-1sr-1, while extinction coefficient absolute
differences are given in Mm-1.
355 nm
532 nm
1064 nm
System
d¯
d¯r [%]
d¯
d¯r[%]
d¯
d¯r[%]
Backscatter coefficient
RALI
-0.065
-10.3
-0.040
-9.3
-0.024
-10.5
MARTHA
-0.146
-20.9
-0.028
-6.0
-0.012
-5.9
PollyXT
0.037
6.9
-0.008
-1.9
-0.027
-9.5
MSTL-2
0.080
14.6
0.068
18.2
0.010
4.4
MUSA
0.025
3.7
0.008
1.7
-0.005
-2.0
Extinction coefficient
RALI
-3.162
-6.3
1.362
3.3
MARTHA
-0.511
-1.5
-1.172
-4.5
PollyXT
-6.754
-9.5
-6.002
-11.9
MSTL-2
4.126
17.6
1.511
6.5
MUSA
-0.574
-2.1
1.571
3.0
Validation based on EARLI09 data
The EARLI09 measurement campaign held in Leipzig, Germany, in May
2009 gave us the possibility to test the
SCC with measurements taken by different lidar systems under the same atmospheric
conditions. Eleven lidar systems from ten different EARLINET stations
performed one month of co-located, coordinated measurements under
different meteorological conditions.
During the campaign, the SCC pre-processor module was successfully used
to provide, in a very short time, signals corrected for instrumental
effects for all participating lidar systems . In this way, all
signals were pre-processed with the same procedures and, consequently,
discrepancies among pre-processed signals could be only due to
unknown system effects.
The data set of the EARLI09 campaign gives us a good opportunity to test not only the
pre-processor module but also all other SCC modules. After the
campaign, a few cases were selected which were characterized by data availability
from all participating systems and stable atmospheric conditions. All participants were asked to
produce their own analysis for these cases allowing us to compare
these profiles with the corresponding results of the SCC. The cases differ in terms of
atmospheric conditions and refer to both night-time and daytime measurements.
For the SCC validation we focus on the case of 25 May 2009 from 21:00 to 23:00 UT
when a Saharan dust event occurred over Leipzig. To
allow for a complete evaluation of the SCC retrieval algorithms, we first selected only the EARLI09 lidar
systems able to measure at same time backscatter coefficient profiles at three wavelengths (1064,
532 and 355 nm) and extinction coefficient profiles at 532 and
355 nm. Among these advanced systems, we made a further selection on the
basis of their differences in terms of technical characteristics. In
particular, we considered the Multiwavelength Raman Lidar (RALI) from
Bucharest as an example of a commercial lidar system;
the MARTHA (Multiwavelength Atmospheric Raman Lidar
for Temperature, Humidity, and Aerosol Profiling) system from Leipzig
as an example of a home-made lidar ; the PollyXT from Leipzig
as representative of the PollyNet network
; the CIS-LiNet Lidar Network for Commonwealth of Independent States
countries, reference system MSTL-2 from Minsk
and, finally, the MUSA (Multiwavelength System for Aerosol) from
Potenza as an EARLINET network reference system .
Figure shows the backscatter coefficient profiles
at 1064 nm obtained from the elastic-backscatter signals
measured by the five lidar systems mentioned above. The profiles
obtained by the SCC are plotted in red, while the corresponding profiles provided by each group with its
own analysis software are shown in blue. The same colour convention is valid
for all other figures in this paper. The agreement between the two
analyses is generally good for all lidar systems indicating
the good performance of the algorithm for the retrieval of the aerosol
backscatter coefficient from elastic-backscatter signals implemented in the SCC.
The red profiles shown in Fig. are obtained using the iterative
method. However, we found that the SCC profiles obtained using the Klett
approach are practically indistinguishable from the ones calculated by
the iterative technique.
A more quantitative comparison between SCC and manual retrievals can
be performed by calculating the mean deviation d¯ and the mean
relative deviation d¯r defined as
d¯=〈si-mi〉,
d¯r=〈si-mimi〉,
where si and mi are the values of the SCC and the manually retrieved
profile at altitude bin i, respectively. The symbol 〈⋅〉 refers to the average over the altitude scale.
The values obtained for the parameters d¯ and
d¯r starting from the profiles shown in
Fig. are summarized in the last two columns of
Table . For the backscatter retrieval at
1064 nm, the mean relative deviations range from a maximum underestimation of
-10.5 % for the RALI system to a maximum overestimation of 4.4 % for the MSTL-2 system. The EARLINET quality
requirements allow a maximum deviation of 30 % or
0.5 Mm-1sr-1 for the backscatter
coefficient at 1064 nm . Consequently, the SCC
backscatter coefficient retrieval at 1064 nm meets the EARLINET
quality requirements for both d¯ and d¯r for
all the considered systems. The highest relative mean deviation
observed for the RALI system is probably due to slightly different
calibration input parameters used in the two analyses as the infrared wavelength is quite
sensible to the calibration procedure .
The backscatter coefficient profiles at 355 nm (at
532 nm) derived with the Raman method from the same lidar systems
are shown in Fig. (Fig. ). The manually obtained profiles agree quite well with the corresponding SCC ones,
considering the reported error bars. As shown in Table , for all systems the mean deviations are larger at
355 nm than at 532 nm. In particular, at 355 nm the relative
mean deviation ranges from -20.9 to 14.6 %, while at 532 nm
the range is from -9.3 to 18.2 %. According to the EARLINET requirements
deviations of aerosol backscatter coefficients at 355 and
532 nm have to be below 20 % or smaller than
0.5 Mm-1sr-1. The EARLINET requirements on
d¯r and d¯ are met at both wavelengths by the
majority of the systems. The only exception is the MARTHA system
for which the SCC retrieval shows a relative mean deviation slightly above
the maximum. However, at the same time, the EARLINET requirements on
d¯ are clearly below the maximum allowed value for all the
systems. In general, the discrepancies can be explained by small
differences in the reference value and in the height range used for the
calibration and also by the depolarization correction ,
which is taken into account in some of the manual analyses but not implemented yet in
the SCC. This is, for example, the reason for the discrepancies observed between 2
and 4 km in the backscatter profiles at 355 nm for
PollyXT (leftmost plot in the middle panel of
Fig. ). This lidar system is equipped with optics exhibiting quite
different transmissivity at 355 nm for the two components of
light polarization. In this case, if the depolarization correction is not
considered and, at same time, strong depolarizing aerosol is observed (like in this case where Saharan dust was
present between 2 and 4 km) an overestimation of the
aerosol backscatter coefficient is made. This effect is clearly visible in the mentioned
plot. The correction of the depolarization effect is not
implemented in the SCC because its application requires the
measurement of the particle linear depolarization-ratio which is not yet a standard
SCC product. However, the next SCC release will include the
correction for depolarization effect as the implementation of
quality-assured procedures to calculate the particle linear depolarization
is planned.
Figures and are examples
of comparisons of the Raman extinction retrievals.
The curves in Fig. are the aerosol extinction
profiles at 355 nm obtained from the nitrogen vibrational-rotational Raman
signal at 387 nm for the five different lidar systems, while Fig. shows the aerosol extinction profiles at 532 nm
calculated from the nitrogen vibrational-rotational Raman signal at 607 nm for
the same systems. The agreement between the two independent analyses is good for both wavelengths.
However, the extinction coefficient profiles at 532 nm are noisier than the ones at 355 nm, and so, in some cases, it is not easy to clearly evaluate the agreement between
manual and SCC analysis. Nevertheless, for all
systems the atmospheric structures are present with very similar and consistent shape
in the manually and the SCC retrieved profiles. The good agreement is
also confirmed by the values of d¯ and d¯r
reported in the bottom part of Table . For
extinction coefficients, the maximum allowed relative and absolute
deviations according to EARLINET requirements are 20 % and
50 Mm-1, respectively . As a result, the SCC
aerosol extinction retrieval meets the EARLINET requirements for all
the considered systems at 355 and 532 nm.
For all the profiles shown in
Figs. –, the molecular contribution to atmospheric
extinction and transmissivity has been calculated using the atmospheric
temperature and pressure profiles measured by a radiosounding
correlative to the lidar measurement session.
Number of MUSA (Potenza) and PollyXT (Leipzig) measurement cases included in the
calculation of the mean profiles shown in
Figs. –. The quantity b1064 indicates the
backscatter coefficient profile derived with the iterative method at
1064 nm while b532 (b355) and e532 (e355) represent
backscatter and extinction coefficient profiles at 532 nm
(355 nm), respectively.
Night-time
Daytime
MUSA
PollyXT
MUSA
PollyXT
b1064
23
15
12
9
b532
20
15
12
9
b355
24
15
10
9
e532
16
15
–
–
e355
14
15
–
–
Mean night-time analysis comparison for the Potenza
station (MUSA system). In the upper graph the mean profiles obtained using
the manual analysis are shown, while in the bottom graph the results obtained by the
SCC are presented. Several measurement cases (see Table ) have been analysed, and the corresponding
backscatter and extinction profiles have been averaged (left two
panels of each graph). The
other two panels of each graph show the lidar
ratios and the Ångström exponents, respectively
as calculated from the mean aerosol extinction and backscatter profiles.
Mean daytime analysis comparison for the Potenza
station (MUSA system). On the left the mean analysis obtained using
the manual analysis is shown, while on the right the results obtained by the
SCC are presented. Several measurement cases (see Table ) have been analysed, and the corresponding
backscatter profiles have been averaged (left two panels of each
graph). The other panel of each graph shows the backscatter related Ångström exponents
as calculated from the mean backscatter profiles.
Comparison of the mean values and standard
errors of the mean for the profiles of the Potenza station shown in Figs. and . Mean values and
standard errors of the mean (reported in parentheses) were
calculated by averaging the mean profiles within Range 1 (0–2 km) and
Range 2 (2–4 km).
Night-time
Daytime
β [Mm-1sr-1]
α [Mm-1]
LR [sr]
β [Mm-1sr-1]
λ [nm]
Manual
SCC
Manual
SCC
Manual
SCC
Manual
SCC
Range 1
355
2.01(0.10)
1.97(0.12)
86.42(3.52)
79.53(4.32)
47.23(1.65)
45.48(1.04)
1.58(0.07)
1.60(0.09)
532
1.35(0.04)
1.38(0.07)
100.00(4.57)
108.35(6.99)
76.64(1.78)
85.17(2.99)
0.85(0.03)
0.87(0.04)
1064
0.65(0.02)
0.69(0.03)
–
–
–
–
0.53(0.01)
0.57(0.02)
Range 2
355
0.62(0.06)
0.60(0.05)
34.74(2.04)
32.28(1.31)
61.71(2.13)
59.76(2.46)
0.50(0.04)
0.48(0.04)
532
0.54(0.03)
0.53(0.03)
43.81(2.17)
41.73(1.39)
84.39(2.52)
81.01(2.31)
0.37(0.02)
0.36(0.02)
1064
0.29(0.01)
0.29(0.01)
–
–
–
–
0.27(0.01)
0.26(0.01)
Relative differences between SCC and corresponding manually retrieved mean profiles for
the Potenza station (MUSA system). On the
left (upper part) the deviation between night-time mean aerosol
backscatter profiles at 355, 532, and 1064 nm (see
Fig. ) are shown. On the right (upper
part) the deviations between night-time mean aerosol extinction profiles at 355 and
532 nm (see Fig. ) are reported. In the
bottom part the deviation between daytime mean aerosol backscatter
profiles at 355, 532, and 1064 nm (see
Fig. ) are shown.
Validation based on climatological data
In the previous section, comparisons of the SCC
analysis with the corresponding manual ones for a single measurement
case were shown, considering several different lidar systems. This comparison
allowed us to investigate the ability of the SCC to provide aerosol
optical properties for different systems, but it did not assure that the
algorithms implemented in the SCC are not affected by systematic
errors or that they work well under different atmospheric
conditions. To prove this ability, mean SCC profiles have been compared to the
corresponding mean profiles obtained by an independent analysis
procedure. In particular, several measurement cases have been inverted
with both the SCC and the manual analysis software (the same manual software used
so far to provide profiles to the EARLINET database). The results have
been averaged and finally compared. Two representative EARLINET lidar
systems have been taken into account for this comparison: MUSA
and PollyXT
operating at the Potenza and Leipzig stations, respectively.
For the Potenza station we have compared the mean profiles obtained by
averaging the measurements made with the MUSA system in correlation with CALIPSO
Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observations,
overpasses between March 2010 and November 2011.
In Table , we summarized the number of single profiles
that have been considered in calculating the mean profiles for both
SCC and manual analysis. The quantity b1064 indicates the
backscatter coefficient profile at 1064 nm while b532 (b355)
and e532 (e355) represent the backscatter and extinction coefficient
profiles at 532 nm (355 nm), respectively.
The number of averaged profiles are not the same for all quantities
as it was not possible to get optical products for
all lidar channels for all cases. For night-time conditions, backscatter
coefficients at 532 and 355 nm have been obtained using the
Raman method. For daytime conditions, the backscatter coefficients at all
wavelengths are calculated with the iterative method.
Figure summarizes the results of the comparison
made for night-time conditions. For each analysis three mean
backscatter coefficient profiles (first plot on the left)
at 1064 nm (red curve), 532 nm (green curve), and 355 nm (blue curve)
and two mean extinction coefficient profiles (second plot from the left) at
532 nm (green curve) and 355 nm (blue curve) are reported. In the
same figure other important aerosol parameters are plotted which
are directly derived from the extinction and backscatter coefficient profiles: the
extinction-to-backscatter ratio (lidar ratio) and the
Ångström exponents. As it is well known that these parameters
depend only on the type of aerosol, it is quite interesting to test the
SCC performance with respect to these parameters.
In general, the agreement between the two analyses is good for all
profiles shown in Fig. . Table and Fig. provide a more
quantitative comparison. In particular, two separate altitude ranges were selected in order to
allow a direct comparison of statistical quantities. As most of the
aerosol load is trapped below 4 km height, the first one (Range 1) extends up to 2 km and the second one (Range 2) from 2 up to 4 km
height. For all vertical profiles plotted in Fig. , mean values and standard
errors of the mean within Range 1 and Range 2 have been
calculated and reported in Table . The agreement on
the backscatter-related mean values and standard errors is quite good
for both Range 1 and
Range 2. The mean values calculated within Range 2
for the aerosol extinction mean profiles agree slightly better than
the ones calculated within Range 1.
The general good agreement is also confirmed by the two plots in the upper part of Fig. showing the
relative difference (below 4 km height) of the aerosol backscatter and extinction mean profiles
displayed in Fig. .
In Fig. the comparison for the MUSA system
under daytime conditions is shown. As already mentioned, in this case the two Raman
channels are not available and so it is only possible to compare
backscatter-related quantities. As it can be seen from Table , also for daytime conditions we have a good agreement
between the two analyses. The same conclusion is supported also by the
mean deviations shown in the bottom part of Fig.
which, typically, vary between ±10 %.
For the Leipzig station, we have compared all regular EARLINET
climatology and CALIPSO measurements made by PollyXT from September
2012 to September 2014 for which the complete data set of three
backscatter coefficient and, at night-time, two extinction coefficient profiles
were available.
The numbers of PollyXT single profiles that have been included in the
calculation of mean profiles are reported in Table .
Figures and show the results
of the comparison for the PollyXT system made under night-time
and daytime conditions, respectively. All quantities displayed in these
figures are the same already described in Figs. and . The agreement between the two analyses is good in
both cases. All manually calculated profiles plotted in
Figs. and agree well with
the corresponding ones calculated by the SCC. Moreover, the same
quantitative comparison made for the
MUSA system has been carried out also for the PollyXT
lidar. The results are summarized in Table
and Fig. . In particular, Table
shows a very good agreement of both mean values and standard errors
calculated within Range 1 and Range 2. The
differences of the backscatter coefficient mean profiles at
355 nm are mainly due to the polarization sensibility of the
PollyXT system at this wavelength. As already mentioned in the
previous section, this effect is corrected in the manual analysis but not yet in the
SCC. The deviations of the aerosol
extinction mean profiles below 1.5 km are probably related to small differences in
handling the correction for not complete overlap.
From the comparison discussed in this section we can
conclude that the SCC performs well under different atmospheric
conditions and for different systems. Of course, further comparisons
and evaluations of SCC products are planned in the near future
especially when more statistical data will be available.
Mean night-time analysis comparison for the Leipzig
station (PollyXT system). In the upper graph the mean profiles obtained using
the manual analysis are shown, while in the bottom graph the results obtained by the
SCC are presented. Several measurement cases (see Table ) have been analysed, and the corresponding
backscatter and extinction profiles have been averaged (left two
panels of each graph). The
other two panels of each graph show the lidar
ratios and the Ångström exponents, respectively
as calculated from the mean aerosol extinction and backscatter profiles.
Mean daytime analysis comparison for the Leipzig
station (PollyXT system). On the left the mean analysis obtained using
the manual analysis is shown, while on the right the results obtained by the
SCC are presented. Several measurement cases (see Table
) have been analysed, and the corresponding
backscatter profiles have been averaged (left two panels of each
graph). The other panel of each graph shows the backscatter related Ångström exponents
as calculated from the mean backscatter profiles.
Comparison of the mean values and standard errors of the mean
for the profiles of the Leipzig station shown in Figs. and . Mean values and
standard errors of the mean (reported in parentheses) were
calculated by averaging the mean profiles within Range 1 (0–2 km)
and Range 2 (2–4 km).
Night-time
Daytime
β [Mm-1sr-1]
α [Mm-1]
LR [sr]
β [Mm-1sr-1]
λ [nm]
Manual
SCC
Manual
SCC
Manual
SCC
Manual
SCC
Range 1
355
3.16(0.22)
3.03(0.19)
168.93(13.4)
157.51(11.3)
52.21(0.59)
51.09(0.57)
2.30(0.23)
2.45(0.26)
532
1.56(0.10)
1.55(0.09)
88.81(9.13)
85.33(7.96)
52.85(1.85)
52.54(1.62)
1.00(0.08)
0.98(0.07)
1064
0.58(0.01)
0.56(0.01)
–
–
–
–
0.48(0.03)
0.47(0.03)
Range 2
355
1.39(0.05)
1.47(0.06)
75.81(2.70)
76.80(2.52)
55.37(0.67)
54.15(1.22)
0.20(0.01)
0.24(0.02)
532
0.86(0.02)
0.94(0.02)
45.84(1.50)
45.33(1.39)
53.09(0.75)
48.17(0.58)
0.08(0.01)
0.11(0.01)
1064
0.32(0.02)
0.31(0.02)
–
–
–
–
0.06(0.01)
0.06(0.01)
Relative differences between SCC and corresponding manually retrieved
mean profiles for the Leipzig station (PollyXT
system). On the left (upper part) the deviation between night-time mean aerosol
backscatter profiles at 355, 532, and 1064 nm (see
Fig. ) are shown. On the right (upper panel) the deviations between
night-time mean aerosol extinction profiles at 355 and
532 nm (see Fig. ) are reported. In
the bottom part the deviation between daytime mean aerosol backscatter
profiles at 355, 532, and 1064 nm (see
Fig. ) are shown.
Example of near-real-time applicability
In this section the main objectives of this 72 h
operationally exercise are briefly recalled and some specific
technical details about how the SCC has been used during that period are described.
In July 2012, 11 EARLINET stations performed an intense period of
coordinated measurements with a well defined measurement protocol. The
measurements started on 9 July at 06:00 UT and continued without interruption
for 72 h whenever the atmospheric conditions allowed lidar
measurements. The details of this quite intensive observation period are
provided in .
The main aim of the 72 h operationally exercise was to provide a large set of
aerosol parameters obtained in a standardized way for a large number
of stations in near-real time. Especially the SCC was
used to retrieve both pre-processed products in real time (mainly
range-corrected lidar signals) and optical products in near-real time for all stations participating in the exercise. The
outputs of the SCC produced in that way can be used for a large variety of applications
like the assimilation of lidar data in air-quality or dust transport models, model validation,
or monitoring of special events like volcano eruptions. In particular, the
SCC pre-processed data measured during the 72 h operationally exercise
have been successfully assimilated in the air-quality model Polyphemus
developed by the Centre d'Enseignement et de Recherche en Environnement
Atmosphérique (CEREA) to improve the quality of PM10 and
PM2.5 forecast on the ground .
All participating stations agreed to provide raw data in SCC
format containing 1 h time series of raw lidar signals
synchronized to the start of each hour.
Starting from these raw data files the SCC was configured to
provide 30 min time-averaged range-corrected signals (pre-processed
files) for all involved lidar systems. During the exercise the SCC was an important tool toward the
standardization of lidar products as the participating lidars
operate at different raw time resolutions (from 1 to 5 min) and they also differ in many other characteristics requiring
different instrumental corrections.
To make the SCC outputs available as soon as possible, an infrastructure
was set up to automatically submit the data to the SCC. To start
the retrieval of the SCC on a particular measurement the user needs to
register the measurement into the SCC database using the web
interface. This operation needs time and also the presence of an
operator. To improve that, a fully automatic uploading system has been
implemented and used during the 72 h measurement exercise. Once the system has
detected the presence of a new measurement, a check on the format of
the uploaded data file is automatically performed and in case of success the measurement is automatically registered to the SCC database and
consequently the SCC is started on it. The results of the SCC analysis
are sent back to the originator for their evaluation as soon
as they are available. With such a system it was
possible to automatically retrieve the needed aerosol optical products and
make them available within 30 min from the end of measurement.
Conclusions
The SCC, an automatic tool for the analysis of EARLINET lidar data,
has been developed and made available to all EARLINET
stations. The SCC has been installed on a centralized server where the
user can submit data using a pre-defined NetCDF structure. The SCC is
highly configurable and can be easily adapted to new lidar systems. In
particular, a user-friendly web interface allows the user to change all
instrumental and configuration parameters to be used in the
analysis. The products of the SCC are
all quality certified in terms of the EARLINET quality assurance program.
The SCC can provide different levels of output:
pre-processed signals (range-corrected lidar signals
corrected for all instrumental effects) and aerosol optical
products (aerosol backscatter and extinction coefficient profiles).
The pre-processed and the aerosol optical products are calculated by
two different SCC modules: the ELPP module which accepts as input the
raw lidar data and the ELDA module which takes as inputs the outputs of the
ELPP module. The actions of the two modules are automatically
synchronized and coordinated by another module called SCC daemon. All parameters
required by the ELPP and ELDA modules are stored in an
efficient way in the SCC database.
The SCC has been validated by using synthetic lidar
signals used during the EARLINET algorithm inter-comparison exercise
and, in this paper, using real lidar data. In particular, the
validation with real lidar data was accomplished by comparing the
SCC optical products with the corresponding products retrieved with
independent manual quality-certified procedures. The
validation was carried out in two different steps. First, considering a case study selected from the EARLI09
inter-comparison campaign, it was proved that the SCC is able to provide
optical products in good agreement with the corresponding manual analysis
for all EARLI09 lidar systems considered. Second, it was checked that the SCC can
provide reliable results in different atmospheric conditions. This was
achieved by performing a statistical analysis of the long-term data
set of two EARLINET stations. The comparisons indicated a good performance of the
SCC as well.
An example of the applicability of the SCC was provided
by describing the use of the SCC during the 72 h EARLINET measurement
exercise. In this case, the SCC delivered high-quality
aerosol properties at different levels (pre-processed signals and aerosol
optical products) in near-real time. Such products can be
assimilated in models or can be used for model validation purposes or
to monitor special events at network level.
The development of the SCC modules is continuing. New
features like particle depolarization-ratio calculation, automatic
determination of aerosol layer properties from both geometrical and
optical point of view, and cloud masking are under investigation and will
be included in the SCC in the framework of the ACTRIS and ACTRIS-2 projects
(http://www.actris.eu). Due to its flexibility the SCC could be easily extended
to GALION to evaluate lidar data of networks different from EARLINET.