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.

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.

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 .

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.

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.

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.

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.

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.

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;

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.

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.

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.