Statistical validation of Aeolus L2A particle backscatter coefﬁcient retrievals over ACTRIS/EARLINET stations on the Iberian Peninsula

. The Global Observing System (GOS) has encountered some limitations due to a lack of worldwide real-time wind measurements. In this context, the European Space Agency (ESA) has developed the Aeolus satellite mission, based on the ALADIN (Atmospheric Laser Doppler Instrument) Doppler wind lidar; this mission aims to obtain near-real-time wind retrievals at the global scale. As spin-off products, the instrument retrieves aerosol optical properties such as particle backscatter and extinction coefﬁcients. In this work, a validation of Aeolus reprocessed (baseline 10) co-polar backscatter coefﬁcients ( β partAeolus ) is presented through an intercom-parison with analogous ground-based measurements taken at the ACTRIS (Aerosols, Clouds and Trace gases Research InfraStructure Network)/EARLINET (European Aerosol Research Lidar Network) stations of Granada (Spain), Évora (Portugal) and Barcelona (Spain) over the period from July 2019 until October 2020. Case studies are ﬁrst presented, followed by a statistical analysis. The stations are located in a hot spot between Africa and the rest of Europe, which guarantees a variety of aerosol types, from mineral dust layers to continental/anthropogenic aerosol, and allows us to test Aeolus performance under different scenarios. The so called Aeolus-like proﬁles ( β part Aeolus like , 355 ) are obtained from total particle backscatter coefﬁcient and linear particle depolarization ratio ( δ part linear ) proﬁles at 355 and 532 nm measured from the surface, through a thorough bibliographic review of dual-polarization measurements for relevant aerosol types. Finally, the study proposes a relation for the spectral conversion of δ part linear , which is implemented in the Aeolus-like proﬁle calculation. The statistical results show the ability of the satellite to detect and characterize signiﬁcant aerosol

Station (ISS) (Yorks et al., 2015Rodier et al., 2016) was performed within the EARLINET community (Proestakis et 100 al., 2019). Regarding Aeolus, several cal/val studies have focused on Aeolus wind products (known as L2B products), including wind profiles and their biases, with frequent intensive campaigns (e.g. Baars et al., 2020;Lux et al., 2020;Witschas et al., 2020, Guo et al., 2020. Recently, some applications of Aeolus optical products (known as L2A) have been reported (Dai et al., 2021;Feofilov et al., 2021). However at the time of writing this article , no studies assessing the calibration and validation of Aeolus aerosol products with ground-based stations have been published. 105 The present study presents the intercomparison of Aeolus L2A aerosol optical products, in particular, the particle backscatter coefficients at 355 nm, with analogous ground-based lidar measurements from three different ACTRIS/EARLINET stations in the Iberian Peninsula, namely Granada, Évora and Barcelona. This article is structured as follows. Section 2 presents the experimental setup of the different systems, addressing the characteristics of Aeolus and ALADIN, along with the main aspects of the ground-based ACTRIS/EARLINET stations involved in the study. Section 3 is devoted to the methodology 110 followed in the intercomparison process, the peculiarities of Aeolus products and ground -based measurements, as well as the criteria set for the comparison. Section 4 gathers the results and discussion. Finally, section 5 summarizes the main findings of the study, with special attention to the most relevant aspects for the satellite mission.

Aeolus satellite 115
Aeolus was launched to space on the 22nd August 2018 from the French Guiana. It is placed on a Sun -synchronous orbit at an altitude of 320 km and inclination angle of 97º. It moves at 7.71 km s -1 and completes an orbit in 90 minutes (16 orbits per day) (Reitebuch et al., 2018;Flamant et al., 2020). The revisit time is 7 days (Flamant et al., 2020). The orbit setting of Aeolus allows it to overpass the equator at 06:00 and 18:00 Local Solar Time (LST) during ascending and descending modes, respectively, in order to minimize the background noise caused by solar radiation, as it avoids solar zenith. 120 The Aeolus satellite is equipped with a single instrument, i.e. the ALADIN lidar. The instrument, based on the Doppler wind lidar technique, acquires profiles of wind speed and particle optical properties in the troposphere and low stratosphere up to 30 km Flamant et al., 2008). The ALADIN lidar points towards the Earth's surface with an angle of 35º compared to nadir, although due to the planet's curvature this angle changes to 37.6º at surface level (Reitebuch et al., 2018). Due to the orbit configuration and the instrument design, ALADIN only retrieves the projection of the horizontal 125 wind speed over the line-of-sight (HLOS), a variable sufficient for the characterization of the wind field (ESA, 2008). This variable is known as Level 2B (L2B) product. As spin-off products, ALADIN also provides particle optical properties, namely the particle backscatter and extinction coefficients among others, known as Level 2A (L2A) products, which are retrieved through the Standard Correct Algorithm (SCA), Standard Correct Algorithm middle bin (SCAmb), Iterative Correct Algorithm (ICA) and Mie Channel Algorithm (MCA) separately (Flamant et al., 2020). The retrieval employs the 130 High Spectral Resolution Lidar (HSRL) technique (Wandinger, 1998). Additionally, L2C products consist of wind fields https://doi.org/10.5194/acp-2021-388 Preprint. Discussion started: 10 May 2021 c Author(s) 2021. CC BY 4.0 License.
after the assimilation of L2B profiles by the forecast models of the European Center for Medium-Range Weather Forecasts (ECMWF) (Ingmann an Straume, 2016). On 12th May 2020, L2B products were made available for the general public (aeolus-ds.eo.esa.int) after going through bias correction procedures. Currently, L2A products access is still limited until a more confident version of the data products is achieved. 135 The instrument emits UV radiation at 355 nm and acquires the backscattered radiation through a dual receiver, co nsisting of two spectrometers, with Rayleigh and Mie channels, for molecule and particle backscattering, respectively (Ingmann and Straume, 2016). Consequently, two independent wind profiles can be retrieved. However, a single measurement of aerosol optical products is obtained from the combination of both signals. The emitted radiation is circularly polarized whilst the receiver only detects the parallel (co-polar) component, resulting in underestimated particle backscatter coefficients and 140 overestimated extinction coefficients (Flamant et al., 2020), especially under conditions with highly depolarizing particles.
ALADIN emits light pulses at a repetition frequency of 50.5 Hz. A single observation, i.e. profile, is generated by averaging shots over a 12 s period corresponding to an horizontal resolution of 87 km. Each profile is divided into 24 vertical bins. The vertical resolution of each bin depends on the altitude: 500 m between 0 and 2 km (roughly the atmospheric boundary layer), 1 km between 2 and 16 km (roughly free troposphere) and 2 km between 16 and 30 km (roughly the lowermost stratosphere) 145 (Ingmann and Straume, 2016).

ACTRIS/EARLINET stations
The EARLINET (European Aerosol Research Lidar Network;Pappalardo et al., 2014) network, in the framework of ACTRIS (Aerosols, Clouds, and Trace gases Research Infrastructure Network; actris.eu), aims to generate a vast database of quality-assured vertical profiles of aerosol measurements under homogeneous standards around Europe. Thanks to its large 150 spatial coverage of the European continent, ACTRIS/EARLINET actively has participated and participates in the validation/calibration of satellite missions measurements (Pappalardo et al., 2010;Amiridis et al., 2015;Papagiannopoulos et al., 2016;Proestakis et al., 2019). In this study, three ACTRIS/EARLINET lidar stations from the Iberian Peninsula, namely Granada, Évora and Barcelona, are considered.
The ACTRIS/EARLINET Granada station (37.164ºN, 3.605ºW, 680 m asl) is located in the Southeastern part of Spain, in a 155 fairly populated region. The city lies in a geographic depression, at the foot of Sierra Nevada, with altitudes up to 3479 m asl to the east of the station. Aerosol particles from anthropogenic origin can be detected during the whole year, mainly released by fossil fuel burning (Lyamani et al., 2006(Lyamani et al., , 2010(Lyamani et al., , 2012. Due to the station proximity to the north of Africa, mineral dust intrusions from the Sahara Desert are often detected during the yea r, mainly in the summer season (Guerrero-Rascado et al., 2008Bravo-Aranda et al., 2015;Granado-Muñoz et al., 2016;Mandija et al., 2016), although winter dust intrusions 160 are more frequent in the last years (Cazorla et al., 2017;Fernández et al., 2019). The concentration of continental aerosols from the European continent are also significant along the year (Lyamani et al., 2010). Notable events of local wildfires smoke (Alados-  and long-range transported smoke from North-America (Ortiz-Amezcua et al., 2014, at specific periods of the year (Cariñanos et al., 2021). The station is equipped with a multispectral Raman lidar system, 165 MULHACEN (LR331D400, Raymetrics S.A.), operated by the Atmospheric Physics research group in the Andalusian Institute for Earth System Research (IISTA-CEAMA) of the University of Granada. The lidar system is based on a Nd:YAG radiation source with a receiver at 355, 532 and 1064, as elastic channels, as well as 354 (N2), 407 (H2O) and 530 (N2) nm as rotational and/or vibrational Raman channels, with a repetition frequency of 10 Hz. A polarization cube in the 532 -nm optical path enables to split the parallel (532p) and perpendicular (532s) components. Furthermore, the lidar has a nominal vertical 170 and temporal resolution of 7.5 m and 1 min, respectively, with a full overlap height at around 800 m agl for all channels (Guerrero-Rascado et al. 2010;Navas-Guzman et al., 2011). The depolarization channel is routinary calibrated using the +45º method (Bravo-Aranda et al., 2013). Further details about MULHACEN and subsequent upgrades can be found in Guerrero -Rascado et al. (2008) and Ortiz-Amezcua et al. (2020. The ACTRIS/EARLINET Évora station (38.568ºN, 7.912ºW, 293 m asl) is located in the Southern part of Portugal, at 175 around 100 km from Lisbon and the border with Spain, in a mainly flat and rural region with relatively low in dustrialization and low concentrations of anthropogenic aerosol (Pereira et al., 2009;Preißler et al., 2013). Smoke particles from nearby wildfires or industrialized regions are regularly transported over the station (Preißler et al., 2013;Pereira et al., 2014), as well as long-range transport from North America Baars et al., 2019). Furthermore, due to the proximity to the Sahara Desert, mineral dust layers are frequent over the city (Pereira et al., 2009;Preißler et al. 2011Preißler et al. , 2013 with many 180 extreme events (Preißler et al., 2011;Valenzuela et al., 2017;Couto et al., 2021). Rare events of simultaneous biomass burnings and mineral dust intrusions have been characterized (Salgueiro et al., 2021). Significant bioaerosol concentration events can occur at specific stages of the year (Galveias et al., 2021). The lidar station is operated by the Institute of Ea rth Sciences, associated with the University of Évora. Among other atmospheric research instruments, the station holds a multispectral Raman lidar of the POLLY XT type  named PAOLI. The Raman lidar is based on a Nd:YAG 185 laser source, which enables it to emit radiation on 355, 532 and 1064 nm with a repetition frequency of 20 Hz. The same wavelengths are detected in reception along with two Raman channels at 387 (N2) and 607 (N2) nm. A depolarization filter allows the instrument to obtain the perpendicularly polarized component of the backscattered signal at 532 nm. Nominal measurements are acquired with a vertical and temporal resolution of 30 m and 30 s, respectively. The full overlap height is around 800 m agl. Further details can be found in Preißler et al. (2011). 190 The ACTRIS/EARLINET Barcelona station (41.393ºN, 2.120ºE, 115 m asl) is located on the northeastern coast of Spain, in a highly populated and industrialized region. Due to its location, the different types of present aerosols are significantly diverse. The background aerosol load, mostly made of urban, traffic-related particles and marine aerosols, is located in the lowermost part of the boundary layer . Saharan mineral dust intrusions are frequent along the year due to its relative proximity to North Africa (Pérez et al., 2006). The variability of the aerosol optical properties in Barc elona in the 195 atmospheric column has been studied by Sicard et al. (2011). The Barcelona  and linear polarization. In reception, the current configuration includes three elastic channels at 355, 532 and 1064 nm, one pure-rotational Raman channel at 354 nm, one vibro-rotational channel at 607 nm and one water vapor channel at 407 nm. 200 In addition the system has also two channels for depolarization measurements at 355 and 532 nm, measuring the light passing through linear polarizers perpendicularly aligned with respect to the linear polarization sent by the laser. The measurements are acquired with a vertical resolution of 3.75 m and the approximate full overlap height being between 400 and 500 m agl. General details about the system can be found in Kumar et al. (2011); details about the depolarization channel set up and the pure rotational retrievals can be found in Rodríguez -Gómez et al. (2017) and Zenteno-Hernández et 205 al. (2021), respectively.
The variety of locations, surroundings and orography of the three stations enables the study to explore coastal/inland, rural/urban and flat/mountainous effects on the quality of the comparisons between the ground station and the satellite.
Aeolus overpasses Évora during an ascending mode (south to north) at around 52 km east every Friday at 18:17 UTC. In the case of Barcelona, the satellite overpasses the station during an ascending mode at 26 km west every Tuesday at 17:39 UTC. 210 The station in Granada lies at the intersection of two Aeolus overpasses: every Thursday at 06:24 UTC (ascending orbit) and 18:04 UTC (descending orbit), both of them at about 24 km west. All three stations fulfill ESA's requirement that only satellite overpasses with ground-track distance less than 100 km should be considered (Straume et al., 2019). Figure 1a shows Aeolus overpasses over a wide part of Europe. Figure 1b

Database and intercomparison methodology
For the intercomparison between ground-based measurements and satellite products, a series of spatio-temporal criteria was established taking into account the location of each station and the overpasses times. On the one hand, the Aeolus observation closest in distance to the station was chosen. For the location of the Aeolus overpass, the middle point of the 87 220 km horizontal average forming each single profile was considered. On the other hand, the temporal co-location was established according to the measurement protocols at each station. For Granada, a 1.5 -hour interval containing the morning overpass time (i.e. 05:30 -07:00 UTC) and a 1-hour interval containing the evening overpass time (i.e.17:30 -18:30 UTC) were chosen. For Barcelona, a 1-hour range centered at the overpass time (i.e. 17:09 -18:09 UTC) was considered. For Évora, a 1.5-hour interval containing the overpass time (i.e. 17:30 -19:00 UTC) was considered to take into account the 225 larger distance between the Aeolus ground track and the lidar site.
Aeolus measurements are generated under specific data processing, called baselines, which are constantly being improved and updated. In October 2020 Aeolus products from July 2019 to December 2019 and from 20th April 2020 to 6th October 2020 were reprocessed under a single baseline, Baseline 10 (B10), aiming to homogenize the processing of the products and encourage Aeolus cal/val teams to proceed with B10 products. The Aeolus database conside red in this work is exclusively 230 composed of B10 Aeolus products and covers different seasons and atmospheric conditions. The current study is developed under cloud-free scenarios. Cloud screening was performed by visual inspection of the ground-based profiles. Aeolus observations have been carefully and individually checked, and atmospheric conditions have been studied along each overpass to ensure cloud-free conditions. Aeolus products are automatically processed by ESA. In the current study, only aeroso l products (L2A) are considered, and 235 in particular particle backscatter coefficients derived from the Standard Correct Algorithm (SCA) and Standard Correct Algorithm middle bin (SCAmb). These algorithms employ the information from the Rayleigh and Mie channels and the derived profiles are divided into a series of vertical bins. The difference between them is that SCAmb bins (or middle bins) are obtained from two halves of adjacent original SCA bins, aiming to reduce noise in the products. These products com e with quality flags that mark individually the validity of each bin measurement. The quality flags assess the signal-to-noise 240 ratio of both Mie and Rayleigh channels, as well as retrieval uncertainties (known as error estimates). A full description of L2A products and their implemented algorithms is given by Flamant et al. (2020).
The ground-based measurements are processed by the Single Calculus Chain (SCC) (D'Amico et al., 2015, the standardized tool that allows to automatically process the lidar data acquired at each station within EARLINET. Very few measurements of the Barcelona station were not inverted successfully by the SCC. In those cases, and 245 after checking the cloud-free condition, the measurements were inverted manually with an algorithm validated in previous algorithm intercomparisons at network level (Böckmann et al. 2004;Pappalardo et al., 2004;Sicard et al., 2009). All data used in this work are level 1.0 and 2.0 data from the EARLINET/ACTRIS database (actris.nilu.no).

Aeolus-like conversion of ground-based lidar particle backscatter coefficients
Aeolus underestimates particle backscatter coefficient as the receiver only detects the co -polar component of circular 250 polarized backscattered radiation at 355 nm (Flamant et al., 2020), while ground -based lidars retrieve particle backscatter coefficients with the total backscattered radiation at 355 nm or 532 nm. For comparison purposes, the co -polar component of the ground-based observations at 355 nm must be extracted from the total particle backscatter coefficient, through the expression: (1) 255 where ,355 is the co-polar component of the particle backscatter coefficient at 355 nm (henceforth labeled as Aeolus -like coefficient, ,355 ), ,355 is the total component of the particle backscatter coeff icient at 355 nm derived from the ground-based lidar, and ,355 is the circular particle depolarization ratio at 355 nm, which is not directly measured by the considered ground-based lidars. The linear particle depolarization ratio can be easily converted into circular particle depolarization ratio from (Mishchenko and Hovenier, 1995): 260 where ,355 is the linear particle depolarization ratio at 355 nm. The stations in Évora and Granada do not hold a depolarization channel at 355 nm and lack the possibility to measure ,355 . A conversion of ,532 in ,355 is proposed in the form of: ( 3)  265 where is the spectral conversion factor. Thus, a thorough bibliographic review of previous multispectral depolarization studies has been conducted and discussed in Section 4.1 to estimate such a conversion factor. The third station, Barcelona, does measure both depolarization ratios but for the sake of consistency of the data processing, Barcelona ,355 was calculated the same way than the other two stations. In Section 4.1 the measurements of ,355 and ,532 in Barcelona are superimposed onto the literature results in order to validate our methodology. 270

Statistical parameters
A key point of the intercomparison is the vertical resolution of each profile. Aeolus divides each profile in a set of 24 vertical bins not homogeneously distributed. The resolution of these bins depends on the altitude range: 500 m between 0 and 2 km asl, 1 km between 2 and 16 km asl and 2 km between 16 and 30 km asl (Ingmann and Straume, 2016). Because the groundbased lidars present a much finer resolution, of the order of a few meters, the resolution of each ground -based profile has 275 been degraded to the Aeolus vertical resolution. Thus, the different ground-based vertical values within a given Aeolus bin are averaged into a single value, permitting a bin-to-bin intercomparison. This degradation process is performed on the ground-based Aeolus-like profiles in the last stage, prior to the statistical analysis calculations. Ground -based vertical coverage depends on the station and on each particular case, typically up to 10 km or up to the top of the highest aerosol layer, while Aeolus profiles extend way beyond 10 km. Hence, only statistical comparisons below 10 km, where most of the 280 aerosols are, is presented.
The statistical results are presented in vertical ranges of 1 km. A pair of values Aeolus/Aeolus-like will fall in a given 1 km range if the middle point of the bin lies within the vertical range (for instance, if a bin ranges from 1900 m to 2400 m, its middle altitude 2150 m lies within the 2 km vertical range). Three statistical parameters are calculated to assess the intercomparison results: bias, root-mean-square error (RMSE) and linear fit. Bias profiles are calculated as follows: 285 where is the vertical range, is the middle altitude of the bin's range that lies within the vertical range and is the number of pairs of values Aeolus/Aeolus-like whose lies within . This parameter indicates if Aeolus underestimates ( ( ) < 0) or overestimates ( ( ) > 0) the co-polar particle backscatter coefficient in each region . The RMSE profile is also obtained as follow: Finally, the linear regression of ,355 ( ) against ,355 ( ) is performed under a series of different settings, in order to test if they lie close to the 1:1 relation. The Pearson correlation coefficient, , is calculated in all cases.

Estimation of the depolarization spectral conversion factor from ground-based profiles to Aeolus-like products 295
For the calculation of the ground-based Aeolus-like profile (Eq. 1), ,355 is needed (Eq. 2). PAOLI (Évora) and MULHACEN (Granada) hold only one depolarization channel at 532 nm, while the lidar system at Barcelona retrieves depolarization information at 355 and 532 nm channels. Therefore, the estimation of ,355 from ,532 is required.
Pairs of ( ,355 , ,532 ) obtained from a thorough review of the literature and for different aerosol types are listed in Table 1. The literature provides a modest, but significant, dataset for different well characterized aerosol types, including 300 mineral dust (fresh, aged, mixed), marine and mixed anthropogenic. These three aerosol types are the predominant aerosol types in Barcelona and the pairs ( ,355 , ,532 ) from the literature will be compared to measurements in Barcelona. Although not listed in Table 1, the literature also offers data for volcanic, bioaerosol and smoke particles.
A linear fit has been applied to the pairs ( ,355 , ,532 ) in order to set up a simple relationship to estimate ,355 from ,532 (Eq. 3) through the factor , called the depolarization spectral conversion factor. Figure 2a shows the 305 scatterplot of ( ,355 , ,532 ) for dust, marine and mixed anthropogenic aerosol from the literature, that are the aerosol types present in the cases used for the intercomparison. The best linear fit for these types together is obtained for = 0.82 ± 0.02 , with a fairly acceptable statistical significance (Pearson correlation coefficient = 0.99 ). The scatterplot for the other aerosol types (volcanic, bioaerosol and smoke) is shown in Figure 2b, and the linear fit is calculated separately for each aerosol type. For biomass burning particles, ,355 is higher than ,532 and = 1.36 ± 0.08, 310 with a reliable statistical significance ( = 0.97). The relationships obtained between ,355 and ,532 (dust plus non-dust, biomass burning, volcanic and bioaerosols; see Figure 2) aim to serve as a look-up table for any station where only the depolarization channel at 532 nm is available, which is a frequent handicap for many lidar systems worldwide. In the case of the Aeolus overpasses considered in our study, only dust and non-dust (marine and anthropogenic) particles have been identified, with no evidence of smoke, 325 volcanic or bioaerosol particles in significant concentrations. For dust and non -dust types, equals to 0.82 ± 0.02 and is implemented from now on in the calculation of the Aeolus-like profile.

Case studies
A set of case studies are given for the different stations under relevant atmospheric conditions. These case studies illustrate the intercomparison process and serve as graphic examples of the Aeolus performance. Sun -photometer measurements are 330 taken into account for the sake of completeness aerosol typing, through the study of the aerosol optical depth at 675 nm (AOD675), the AOD-related Ångström exponent calculated between the channels at 440 and 870 nm (AOD-AE440-870), the fine/coarse mode AOD fraction at 500 nm, the particle size distribution and the single scattering albedo at 440 and 1020 nm (SSA440 and SSA1020) (e.g. Dubovik et al., 2002;Gobbi et al., 2007;Lee et al., 2010;Shin et al., 2019;Foyo -Moreno et al., 2019). AERONET level 1.5 or level 2.0 products, depending on availability, computed from the version 3 algorithm (Giles 335 et al., 2018) are used.
The location of the stations is highly interesting due to their proximity to the Sahara Desert and mainland Europe, so frequent events of mineral dust and anthropogenic particles could be detected by the satellite. In addition, Barcelona lies just in the coastline, and both Barcelona and Granada present high concentrations of anthropogenic aerosol, while Évora aerosol concentrations could be classified as rural. Thus, Aeolus operation can be tested under a complete set of atmospheric 340 scenarios.

Case study of anthropogenic aerosol: Granada, 5th September 2019
Aeolus overpassed Granada at 18:04 UTC on the 5th September 2019 with a horizontal distance of 14 km (from Aeolus observation middle point). The daily range corrected signal time series and the ground track of the satellite are presented in Figures 3a and 3b, respectively. As observed in the time series, a significant particle concentration is detected during the 345 whole day up to 4 km asl approximately, with a few cirrus clouds between 6 and 14 km asl. Thus, following the ESA requirements, and for the sake of homogeneity, this case is not included in the statistical analysis presented in Section 4.3, but due to its interesting features it is included as a case study of Aeolus performance. Figure 3b shows Aeolus SCAmb backscatter along the orbit, considering Aeolus quality flags. A significant aerosol layer over North Africa and the South of the Iberian Peninsula can be observed. 350 The HYSPLIT model (Figure 4) revea ls that the air masses over Granada at 18:00 UTC come from mainly two differentiated regions: above roughly 4.3 km agl (equivalent to 5 km asl, not shown) the air masses traveled the Atlantic from North America, while below 4.3 km agl the air masses were m ostly stagnant over the Iberian Peninsula. Thus, mostly continental/anthropogenic particles are expected over Granada in the lowermost region.
The measurements from the co-located Sun-photometer (not presented here) suggest a predominance of fine mode particles 355 during the whole available period. The AOD-AE440-870 values agree with the presence of small particles (continental/anthropogenic aerosol), with a mean value of 1.30 ± 0.08. Furthermore, the AOD675 slightly varies throughout the day, with a mean value of 0.15 ± 0.01. The columnar particle size distribution displays a bimodal distribution in the early morning with a mean effective radius of 0.39 ± 0.06 μm and the SSA around 0.99 for all wavelengths, indicating the presence of non-absorbing particles (Shin et al., 2019). 360 Figure 5 presents the most relevant vertically-resolved quantities measured by Aeolus and the ground-based lidar system in Granada. According to the particle backscatter coefficients at 355 and 532 nm (Figu re 5a) a significant aerosol layer is observed up to 4 km approximately and a clear free troposphere above with a thin cirrus cloud at around 11 km asl. The  Figure 5c), in terms of co-polar particle backscatter coefficient values and vertical layering, with an excellent agreement for 370 these bins. Second, for this particular case, SCAmb retrievals present a better agreement with ground -based measurements than SCA retrievals. Third, the satellite performance presents a surface-related effect for their lowermost bins, retrieving a large and unreasonable co-polar particle backscatter coefficient. Fourth, the cirrus cloud shown in Figure 3a is well retrieved by SCA but not by SCAmb. The implementation of Aeolus quality flags (Figure 5d) produces a notable decrease in the amount of available data points. In this case, quality flags do not seem to help in cloud screening of the satellite data (at 11 375 km asl, Aeolus retrievals are the same with and without quality flags). However, SCA retrievals are improved since the quality flags exclude the negative particle backscatter coefficients found between 6 and 9 km asl.

Case study of mineral dust: Évora, 28th June 2019
Aeolus overpassed Évora at 18:17 UTC on the 28th June 2019 with a horizontal distance of 61 km (from Aeolus observation middle point). The time series of the daily lidar range corrected signal at 355 nm and the ground track of the satellite are 380 presented in Figure 6a and 6b, respectively. A notable and homogeneous layer can be identified throughout the whole day below 3 km asl. Furthermore, there is no evidence of cloud presence above the station. Figure 6b shows the Aeolus SCAmb backscatter retrievals along the orbit with the quality flags applied. A homogeneous layer can be seen over the western side of the Iberian Peninsula, with a wider vertical extension over Morocco.
The HYSPLIT model indicates that the 12:00 UTC air masses over Évora at 1.7 and 2.7 km agl (equivalent to 2 and 3 km 385 asl) are coming directly from lower altitudes in Northern Africa (Figure 7a). The back trajectories of the air masses over Évora at 18:00 UTC (Figure 7b), closer in time to Aeolus overpass, still indicate an origin over lower altitudes in the African continent for the air masses at 1.7 km agl but no longer for 2.7 km agl. Both BSC -DREAM8b and NAAPS models (not shown here) indicate the presence of low but non-negligible concentrations of mineral dust particles over the region at 18:00

UTC. 390
The co-located Sun-photometer measurements (not presented here) suggest the predominance of coarse-mode particles until 09:30 UTC approximately. From 09:30 UTC on, fine-mode particles dominate. The AOD-AE440-870 agrees with the presence of a mineral dust layer over the station during the first half of the day, with a mean va lue of 0.70 ± 0.07. This value maintains between 0.88 and 1.12 approximately during the second half of the day, indicating that the dust episode is vanishing over Évora. The AOD indicates that the possible mineral dust layer over the first half of the day does not present large 395 concentrations of mineral particles, although these values are far from representing a clean atmosphere. The columnar size distribution endorses this hypothesis, with a large predominance of large particle radii during the morning and a decrease in the concentration after noon, with mean total effective radii of 0.60 ± 0.02 μm and 0.47 ± 0.03 μm, respectively. The SSA corroborates the presence of mineral dust during the day, with a positive difference SSA1020-SSA440 of +0.04 in the morning and a difference of almost zero (-0.001) in the late afternoon. Thus, the models and the Sun -photometer measurements 400 indicate the presence of a minor mineral dust episode over Évora that vanishes in the afternoon. The satellite overpass takes place at 18:17 UTC, when the dust episode is practically finished and the dust concentration is low. At that time, AOD675 takes a value of 0.11 while AE reaches 1.04. Figure 8 presents the most relevant vertically-resolved quantities measured by Aeolus and the ground-based lidar system in Évora. A well defined layer in the lowermost atmosphere is detected by the lidar at both 355 and 532 nm channels (Figure  405 8a). The lidar ,532 (Figure 8b) agrees with the presence of mineral dust particles mixed with other non-polarizing particles. Furthermore, the backscatter-related Ångström exponent profile calculated with the 355 and 532 nm channels (not presented here) takes values close to zero, corresponding to mineral dust particles (e.g. Müller et al., 2007;Guerrero-Rascado et al., 2009;Preißler et al., 2011;Fernández et al., 2019), in the whole vertical range of the detected layer. Aeolus detects this layer under both SCA and SCAmb (Figure 8c). First, a fair agreement between the satellite and ground-based systems is 410 achieved in the whole profile. Second, this case study leaves no doubts that the proposed Aeolus -like conversion of the total component ground-based backscatter profiles to co-polar component profiles has to be considered. Third, no significant difference is detected between SCA and the SCAmb intercomparison results for this case. Fourth, Aeolus behaves stably above the layer, measuring particle co-polar backscatter coefficients close to zero in the free troposphere, although sometimes it retrieves negative and meaningless values. Fifth, the satellite presents a surface -related effect for the lowermost bin, 415 retrieving large (and unrealistic) co-polar particle backscatter coefficients. In the final stage, quality flags are applied to Aeolus measurements (Figure 8d), presenting a notable decrease in the amount of available Aeolus values to perform the intercomparison. These quality flags limit the intercomparison to the layers with significant aerosol loads, preventing the intercomparison of Aeolus behavior in the free troposphere. Moreover, current preliminary quality flags do not prevent surface-related effects on the final Aeolus measurements. 420

Case study of smoke: Barcelona, 2nd July 2019
Aeolus overpassed the city of Barcelona at 17:39 UTC on the 2nd July 2019 with a horizontal distance of 35.17 km (from Aeolus observation middle point). Figure 9a presents the time series of the daily lidar range corrected signal at the 1064 nm channel, and the ground track of the satellite is presented in Figure 9b. A significant aerosol layer, which is itself stratified in thinner layers, is detected up to 2.5 km asl, as well as a sparse small layer above, between 2.5 and 4 km asl. Figure 9b also 425 presents the Aeolus SCAmb co-polar backscatter retrievals along the orbit with the quality flags applied. Figure 9b displays Aeolus SCAmb co-polar backscatter coefficients along the considered orbit applying the quality flags. A significant lay er is captured by the satellite above France, the north-western part of the Iberian Peninsula and the Mediterranean Sea.
The HYSPLIT model indicates that the air masses over Barcelona at 18:00 UTC between 1.9 and 2.9 km agl (equivalent to 2 km and 4 km asl), approximately, come directly from Southeastern France/northwestern Italy ( Figure 10). In particular, the 430 air masses at 1.9 km agl have the typical pattern of local recirculation and might carry pollutants from southern France.
Additionally, the NAAPS model (not shown here) yields the presence of significant smoke concentrations over southeastern France/northwestern Italy during the previous days. Furthermore, both Aqua -MODIS and Terra -MODIS measurements (not shown here) reveal the existence of wildfires in southeastern France/northwestern Italy.
The co-located Sun-photometer retrievals indicate the predominance of fine mode particles throughout the day. The daily 435 mean AOD-AE440-870 is 1.43 ± 0.13, while the AOD675 also remains constant throughout the day, with a mean value of 0.15 ± 0.02. The particle size distribution presents two distinct modes, with a mean effective radius of 0.39 ± 0.04 μm approximately, where the fine mode dominates. This distribution is constant over the day. After noon, all of the retrieved SSA1020 lie below 0.95, suggesting the prevalence of absorbing particles (Shin et al., 2019). On the other hand, for all of the available sets of SSA1020 and SSA440, the difference SSA1020-SSA440 is negative, so the presence of mineral dust is discarded 440 (Dubovik et al., 2002). These Sun-photometer measurements suggest the presence of a smoke layer over the station during the whole day. Aeolus overpassed the station at 17:39 UTC, when the values of AOD675 and AOD-AE440-870 are 0.15 and 1.47, respectively.
The most relevant vertically-resolved properties measured by Aeolus and the ground-based lidar in Barcelona are represented in Figure 11. It detects the presence of several layers: a non -depolarizing aerosol layer up to 2.5 km asl approximately and a 445 depolarizing layer above 2.5 km asl (Figures 11a and 11b). From 2.5 to 6 km the particle backscatter coefficient decreases Second, the satellite clearly detects a layer up to 5 km asl approximately. Third, the SCA retrieves stable close to zero values in the free troposphere, although some are negative and meaningless. Fourth, the surface -related effect is present in the lowermost bins. In this case, Aeolus quality flags (Figure 11d) seem to remove the surface-related effect, but again they do not allow for investigating the Aeolus performance in the free troposphere. 460

Statistical analysis
This section assesses the intercomparison of Aeolus SCA and SCA middle bin products with ground -based measurements from a statistical point of view. The process is performed considering Aeolus quality flags to achieve a further understanding of the products. Moreover, ground-based measurements are cloud screened (Granada case study for the 5th September 2019, see details in Section 4.2.1., is removed as well). Taking into account the requirements and considerations presented in 465 Section 3, the initial database is largely reduced, in order to ensure the reliability of the intercomparison. From the initially available measurements, i.e. 101 B10-overpasses for Granada, 51 for Évora and 52 for Barcelona, and after applying the set of requirements, the intercomparison has been performed with 24 cases for Granada, 15 cases for Évora and 16 cases for Barcelona, leading to enough statistical significance.
First, we address the general performance of the satellite, with emphasis on the domain of Aeolus co -polar particle 470 backscatter coefficient retrievals. On the one hand, Figure 12a shows that Aeolus SCA retrievals range from approximately -2 Mm -1 sr -1 , to large and unrealistic values (up to 86 Mm -1 sr -1 ) which are associated with the surface-related effect shown by the satellite. On the other hand, Aeolus SCAmb retrievals range from 0 Mm -1 sr -1 to similarly large and unrealistic va lues (up to 68 Mm -1 sr -1 ) (Figure 12b). With the implementation of the quality flags (Figure 12c and 12  were tested and no valid model was found. Quality flagged data (Figures 13c and 13d) worsen the linear relationship ( smaller than 0.25 in both cases). The same analysis has been performed fo r each station separately (not presented here) in order to search potential particularities of each site. Analogous and unsatisfactory results were found with the dataset of each station.
Aeolus backscatter coefficient uncertainties (known as Aeolus error estimates) are addressed through the biases between 490 satellite and ground-based measurements. Figure 14 reveals that the larger the Aeolus uncertainties, the larger the bias. In this case, it can clearly be seen that quality flags implementation does not remove Aeolus retrievals with large uncertaint ies in absolute terms. Quality flags assess Aeolus errors relative to the backscatter coefficient retrievals. Therefore, the lowermost measures of Aeolus, which generally present large and unrealistic co -polar backscatter coefficients and large uncertainties, are still flagged as valid. Furthermore, Figures 14a and 14b shows that the SCAmb retrievals present smaller 495 errors.

Granada
The statistical results for Granada are obtained from the 24 selected cases (309 SCA data points and 246 SCAmb data points). Aeolus retrieves co-polar particle backscatter coefficients from approximately 32 km to the ground level (downward view). However, due to the station's altitude (680 m asl) and the lidar full overlap height, no matching measurements are 500 available between 0 and 1 km asl ( Figure 15). On the one hand, Aeolus products present a significant surface -related effect for the lowermost regions, between 1 and 2 km asl. Thus, the satellite strongly overestimates the co-polar particle backscatter coefficient in the 1 to 2 km asl vertical range (with no quality flag implementation, Figures 15a, 15b, 15c and 15d), with a SCA bias around 9 Mm -1 sr -1 and RMSE around 19 Mm -1 sr -1 along with a SCAmb bias around 8 Mm -1 sr -1 and RMSE around 11 Mm -1 sr -1 . This surface effect may affect as well the 2 to 3 km asl range to a lesser extent. Figures 15a and 15b show that 505 the general performance of the SCA underestimates particle backscatte r coefficients from 3 to 11 km asl, with a fair bias value, smaller than 0.4 Mm -1 sr -1 in any case, and RMSE lower than 1 Mm -1 sr -1 (average values of -0.18 ± 0.07 and 0.6 ± 0. 2 Mm -1 sr -1 respectively). On the other hand, Figure 15c shows that the SCAmb do es not present any trend between 2 and 11 km asl, with the bias values oscillating around 0, between -0.11 and 0.17 Mm -1 sr -1 (average value of 0.07 ± 0.11 Mm -1 sr -1 ).
For this algorithm, the RMSE (Figure 15d) lies below 0.7 Mm -1 sr -1 in every range above 3 km asl (average value of 0.40 ± 510 0.13 Mm -1 sr -1 ). SCAmb derived RMSE values are smaller than those obtained with the SCA in every vertical range. Thus, a better agreement is found between the satellite and the ground -based measurements with the SCAmb. Furthermore, with the quality flags implementation (Figures 15e-h), the number of available measurements flagged as valid is largely diminished (only 1 out of 7 SCA values and 1 out of 6 SCAmb values), especially above 3 km asl. Therefore, the statistical significance of the results is also reduced and the reliable results are limited to the lowermost ranges. Additionally, after the quality flags 515 consideration between 2 and 4 km asl, i.e. the statistically significant ranges, SCA and SCAmb RMSE values increase a 45 and 61 %, respectively. The use of the quality flags worsens the average agreement between the satellite and the groundbased system and does not avoid the surface-related effect on the measurements.

Évora
In the case of Évora, the statistical results are derived from 15 selected cases (150 SCA data points and 108 SCAmb data 520 points). Figure 16 indicates that a few matching measurements in the vertical range from 0 to 1 km asl could be found for this lidar system, due to the station's altitude (293 m asl). On the one hand, the surface -related effect is present in all cases in the lowermost regions as well, from 0 to 2 km asl, with the vertical range from 0 to 1 km clearly more affected by this effect.
In these regions Aeolus largely overestimates co-polar particle backscatter coefficient (Figures 16a -d) with a SCA bias around 11 Mm -1 sr -1 and RMSE around 13 Mm -1 sr -1 along with a SCAmb bias around 5 Mm -1 sr -1 and RMSE around 6 Mm -1 525 sr -1 . Therefore, SCAmb retrieval is less affected by the surface effect. On the other hand, an inhomogeneous performance is observed for the SCA above 2 km asl (Figure 16a), with bias values ranging from -0.2 to 0.5 Mm -1 sr -1 (average value of 0.02 ± 0.24). Additionally, an average RMSE value of 0.5 ± 0.3 Mm -1 sr -1 is obtained for the vertical ranges above 2 km asl ( Figure 16b). The SCAmb (Figures 16c and 16d) seems to overestimate particle backscatter coefficient from 2 to 11 km asl, although a fair agreement with ground-based measurements is observed (average bias value of 0.11 ± 0.08 Mm -1 sr -1 and 530 RMSE of 0.36 ± 0.17 Mm -1 sr -1 ). Therefore, the SCAmb retrievals present a better agreement with ground-based measurements than the SCA. The number of selected cases is smaller for Évora than the ones used in the case of Granada, but still statistically significant. However, with the implementation of the quality flags (Figures 16e -h) the amount of valid matching measurements is drastically reduced (only 1 out of 5 SCA values and 1 out of 6 SCAmb) and the results cannot be considered statistically significant in any range. The only improvement in the agreement between the systems is observed 535 between 1 and 2 km asl, but the statistical significance has to be taken into account. Again, the application of the quality flags does not avoid the surface-related effect on the final results.

Barcelona
The statistical results for Barcelona are derived from 16 selected matching ca ses (80 SCA data points and 76 SCAmb data points). In Barcelona, the validation process of the particle backscatter coefficient cuts the profiles where the aerosol layers 540 end, so the vertical coverage usually does not extend higher than 5 or 6 km asl (depending on the atmospheric conditions).
Thus, no statistical intercomparison could be performed above this altitude. Furthermore, the station lies at a very low altitude above sea level (115 m asl) and its full overlap height (between 400 and 500 m agl) allo ws us to work with a significant amount of matching values between 0 and 1 km asl, the vertical range which is most affected by the surface https://doi.org/10.5194/acp-2021-388 Preprint. Discussion started: 10 May 2021 c Author(s) 2021. CC BY 4.0 License.
( Figure 17). Between 0 and 1 km asl Aeolus largely overestimates co -polar particle backscatter coefficients (with no quality 545 flag implementation, Figures 17a and 17b), with an approximate SCA bias of 15 Mm -1 sr -1 and RMSE of 21 Mm -1 sr -1 along with a SCAmb bias around 9 Mm -1 sr -1 and RMSE around 18 Mm -1 sr -1 . However, possible surface-related effects can be observed in the case of the SCA retrievals between 1 and 4 km asl (Figure 17a), where an average RMSE of 1.6 ± 0.1 Mm -1 sr -1 is observed. This could be explained by the complex terrain orography below the Aeolus ground -track, mostly affected by the transition from sea to land with the Central Prelitoral System situated only 15 km from the coast and reaching almost 550 1000 m asl. On the contrary the SCAmb seems to be partially affected between 1 and 2 km asl (Figure 17b) and to a lesser extent (1.4 Mm -1 sr -1 ). Therefore, the SCAmb is more robust to the surface effects. Nevertheless, Aeolus does not present a trend above 1 km asl neither under SCA, nor under SCAmb (Figures 17a and 17b), and the bias values ranges from -0.5 to 0.8 Mm -1 sr -1 and from -0.2 to 0.3 Mm -1 sr -1 respectively. In the rest of the available vertical ranges, Aeolus presents a slightly better agreement with the ground-based system under SCAmb, with RMSE values below 0.5 Mm -1 sr -1 between 2 and 7 km 555 asl. Finally, when quality flags are applied (Figures 17c and 17d), the amount of valid matching measurements is reduced (almost 2 out of 5 SCA values and 2 out of 7 SCAmb values), affecting the statistical significance of the results.
Additionally, after the quality flags consideration between 1 and 3 km asl, i.e. the statistically significant vertical ranges, SCA RMSE values increase a 40 % and SCAmb RMSE values a 65 % between 1 and 2 km asl. Thus, quality flag filtering of the dataset worsens the statistical results and does not avoid the surface-related effect. 560

Conclusions
Aeolus satellite was launched in 2018. At the time of writing of this article, the longest, fully homogeneous product dataset has been reprocessed in baseline 10 (reprocessed products, B10 version). In this study we evaluated Aeolus B10 optical products with a thorough analysis of Aeolus co-polar backscatter coefficients under the standard correct algorithm (SCA) and the standard correct algorithm middle bin (SCAmb), and an effective testing of Aeolus quality flags. This process was 565 performed taking into account the ESA and the cal/val community recommendations through the intercomparison of Aeolus products with analogous ground-based measurements taken at the ACTRIS/EARLINET stations of Granada, Évora and Barcelona (Southwestern Europe), matching temporally and spatially the satellite's overpasses (55 cases). However, Aeolus overpasses at each station were analyzed separately, aiming to characterize Aeolus performance under different and relevant atmospheric conditions, aerosol types and orographic features. 570 We assessed the so-called Aeolus-like conversion of ground-based measurements. Aeolus retrieves the co-polar component of the backscatter coefficient, which is not directly comparable to the total component measured at the surface stations. Thu s, the co-polar component of the ground-based measurements has to be derived from the total one. In this work, an approach based on a thorough bibliographic review of dual-polarization measurements for relevant aerosol types, aiming to estimate Aeolus-like conversion of the ground-based measurements, and the implementation of the spectral relationship if needed, which has proven to be effective in our case studies.
Several types of linear and nonlinear relations were tested and no valid model was found for ,355 and .
Also, a relation between high Aeolus uncertainties and bias differences was noted. These results were observed at the three 580 stations, suggesting that they were related to the satellite data characteristics and/or the methodology employed and difficulties inherent to satellite cal/val activities rather than to a specific feature of one particular station.
Aeolus SCAmb retrievals presented a better agreement with respect to ground -based measurements than the SCA ones. For the Granada station, the difference between algorithms was less significant, while the SCAmb presented much better results than the SCA for the stations in Évora and Barcelona. Évora presented the highest agreement with the SCAmb retrievals, 585 although the results for Barcelona and Granada were quite satisfactory as well. RMSE profiles obtained with the SCAmb are fairly similar for the three stations, providing consistency to the results obtained. Aeolus quality flags implementation entailed a strong reduction of the amount of Aeolus measurements initially available. For both SCA and SCAmb approximately 20 % of the data remains after the quality tests. This substantially affected the statistical significance of t he results of the filtered dataset. Additionally, Aeolus measurements over all of three stations presented a critical surface-related 590 effect that caused Aeolus to drastically overestimate the co -polar backscatter coefficients. Depending on the station and the orography of the region, this effect extended up to higher altitudes. Finally, the statistical intercomparison was not improved after the quality flag application, e.g. the surface effect was not mitigated and even an increase of the RMSE (between a 40 and a 65 %) was observed.
It has been seen that under significant cloud conditions the satellite experiences saturation and retrieves inconsistent and 595 invalid results. Even cirrus conditions can affect the results. However, the presented case study for Granada (5th September 2019) shows that the satellite is able to characterize thin cirrus clouds with a fairly acceptable agreement.
Despite the distance between the overpasses and the stations, and the fact that Aeolus products are generated by averaging horizontally over 87 km, a good agreement was found between Aeolus retrievals and ground-based lidar measurements, demonstrating that Aeolus has a high potential for the worldwide characterization of the aerosol vertical distributions.