Overview of SLOPE I and II campaigns: aerosol properties retrieved with lidar and sun-sky photometer meuasurements

The Sierra Nevada Lidar aerOsol Profiling Experiment I and II (SLOPE I and II) campaigns were intended to determine the vertical structure of the aerosol by remote sensing instruments and test the various retrieval schemes for obtaining aerosol microphysical and optical properties with in-situ measurements. These campaigns deployed a set of in-situ and 20 remote sensing instruments at the stations include in AGORA observatory (Andalusian Global ObseRvatory of the Atmosphere) in the Granada area (Spain) along summer in 2016 and 2017. In this work, using the in-situ measurements performed at a high-altitude station, Sierra Nevada station, and airborne flights, we evaluate the retrievals of aerosol properties by GRASP code (Generalized Retrieval of Atmosphere and Surface Properties) combining lidar and sun-sky 25 photometer measurements. Besides, we show an overview of aerosol properties retrieved by GRASP during SLOPE I and II campaigns. We evaluate the GRASP retrievals of total aerosol volume concentration (discerning between fine and coarse modes), extinction and scattering coefficients, and for the first time we present an evaluation of absorption coefficient. The statistical analysis of the aerosol optical and microphysical properties, both 30 column-integrated and vertically-resolved, from May to July 2016 and 2017 shows a large variability in aerosol load and types. The results show a strong predominance of desert dust particles due to the North African intrusions. The vertically-resolved analysis denotes a decay of the atmospheric aerosols with altitude up to 5 km a.s.l. Finally, two events of desert dust and biomass burning were used to show the high potential of GRASP to retrieve and study the 35 aerosol properties profiles such as absorption coefficient and single scattering albedo for https://doi.org/10.5194/acp-2021-66 Preprint. Discussion started: 5 February 2021 c © Author(s) 2021. CC BY 4.0 License.


Introduction
The characterization of atmospheric aerosol optical and microphysical properties is difficult due to their high spatial and temporal variability in the atmosphere. These together with the complexity of the aerosol-radiation interaction (scattering and absorbing incident solar and 50 outgoing thermal radiation) and the cloud-aerosol interaction (modifying cloud properties), results in a large uncertainty in the radiative forcing of climate due to aerosols (IPCC, 2013).
During the last decades, a good number of field campaigns has been carried out for studying atmospheric aerosol properties (e.g., Tanré et al., 2003;Mallet et al., 2016;Veselovskii et al., 2016;Vandenbussche et al., 2020) using observatories with in-situ 55 measurements and included in global networks, based on passive and active remote sensing instruments, such as AERosol RObotic NETwork (AERONET; Holben et al., 1998) and European Aerosol Research LIdar NETwork (EARLINET; Pappalardo et al., 2014). On the one hand, the in-situ ground-based observatories only represent limited atmospheric sample in the layer closest to the surface. The passive remote sensing instruments, such as sun-sky 60 photometers or satellites provide aerosol properties in entire atmospheric column, while have very limited information about variations within the column. Hence, vertically-resolved aerosol observations are needed to discern between the different aerosol layers and to study their radiative properties. In these regards, the lidar systems are used for aerosol optical and microphysical properties profiling. Advanced lidar systems have information on the microphysical properties by inversion algorithms using the 3 + 2 configuration (e.g . Müller 70 et al., 1999;Böckmann, 2001;Veselovskii et al., 2002).
The main drawback of these algorithms is the scarcity of Raman lidar measurements during the daytime that represents a limitation to the retrieval of the extinction coefficient data (Veselovkii et al., 2015;Ortiz Amezcua et al., 2020). As an alternative, during the last years, several synergetic retrievals algorithms have been developed to retrieve aerosol optical and 75 microphysical properties combining data from sun-sky photometers and backscatter lidar measurements such as LIRIC (LIdar-Radiometer Inversion Code) by Chaikovsky et al. (2008, Granados-Muñoz et al, 2020 and GARRLiC (Generalized Aerosol Retrieval from Radiometer and Lidar Combined data) by Lopatin et al. (2013). One of the most popular advanced inversion algorithms is the Generalized Retrieval of Atmosphere and Surface 80 Properties code (GRASP; Dubovik et al., 2011Dubovik et al., , 2014. It should be noted here that GARRLiC is a branch of GRASP. The versatility of GRASP allows the retrieval of aerosol vertical and surface properties combining different types of measurements, such as sun-photometers, lidar, ceilometers, satellite, sky-cameras, nephelometers, etc. (e. g. Lopatin et al., 2013;Espinosa et al., 2017;Román et al., 2017;Torres et al., 2017;Benavent-Oltra et al., 2017;Titos et al., 201985 Herreras et al., 2019. In addition, GRASP retrievals have been used to evaluate forecast models, as constrains for global models and as inputs for radiative transfer models (e.g. Tsekeri et al., 2017;Chen et al., 2018Chen et al., , 2019Granados-Muñoz et al., 2019). It is important to explore the potential of this kind of algorithms by applying them to different input data and for different atmospheric conditions. In these regards, the extensive measurement 90 dataset obtained during Sierra Nevada Lidar aerOsol Profiling Experiment I and II (SLOPE I and SLOPE II) campaigns in May, June and July 2016 and 2017, respectively, allow an evaluation of the atmospheric aerosol properties retrieved by GRASP code combining lidar and sun-sky photometer measurements. This database was successfully utilized in several previous studies of the atmospheric aerosol (e.g. de Arruda Moreira et al., 2018Bedoya-95 Velásquez et al., 2018;Horvath et al., 2018;Casquero-Vera et al., 2020).
The main objective of this work is to provide an overview of the aerosol optical and microphysical properties during SLOPE I and II campaigns using the GRASP code. We check the GRASP retrievals versus in-situ measurements performed at the Sierra Nevada Station (SNS, Spain; 2500 m a.s.l.) and instrumented flights. In contrast to previous studies by Román 100 et al. (2018) and Titos et al. (2019), which mainly evaluated long-term vertical profiles retrieved by GRASP combining sun-sky photometer and ceilometer measurements here, for https://doi.org/10.5194/acp-2021-66 Preprint. Discussion started: 5 February 2021 c Author(s) 2021. CC BY 4.0 License.
the first time, we study aerosol properties such as absorption coefficients and volume concentration for fine and coarse modes, separately. In addition, a statistical analysis of both total column and vertically-resolved aerosol properties is performed, and two extreme events 105 of desert dust and biomass burning are evaluated.

Sites and measurements
The SLOPE I and II campaigns took place in Granada (Spain) during the summers of 2016 and 2017 and were designed to determine the vertical structure of the aerosol by remote sensing instruments through the application of various retrieval schemes for obtaining aerosol 110 microphysical and optical properties. The main objective of this campaign was to perform a closure study by comparing remote sensing system retrievals of atmospheric aerosol properties with various in-situ measurements (Román et al., 2017;Benavent-Oltra et al., 2019). The study area typically presents very variable aerosol loads and type, with large presence of anthropogenic aerosols mainly in winter (e.g., Lyamani et al., 2010;del Aguila et al., 2018;115 Casquero-Vera et al., 2021) and frequent Saharan dust intrusions (e.g., Perez-Ramirez et al., 2012;Valenzuela et al., 2012) and primary aerosol associated to the local phenology (Cariñanos et al., 2020). The region is often affected by episodes of aerosol stagnation due to its complex geography (e.g., Lyamani et al., 2010), while Atlantic air masses are usually responsible for cleaning the atmosphere (Perez-Ramirez et al., 2016).

120
During SLOPE I and II the instrumentation was deployed at the three stations of the during SLOPE I and II were sun-sky-lunar photometer Cimel CE318-T, which also perform lunar direct irradiance measurements to retrieve the AOD during night-time between the first and third Moon quarters (e.g., Barreto et al. 2016. In this work, we 155 used AERONET Version 3 Level 1.5 (cloud-screened) data (e.g., Giles et al., 2019;Sinyuk et al., 2020).
The ground-based MWR (RPG-HATPRO G2, Radiometer physics GmbH) located at UGR station as part of the MWRnet (Rose et al., 2005;Caumont et al., 2016), is used here for retrieving temperature profiles. MWR is a passive remote sensor that performs measures 160 unattended of the brightness temperatures of oxygen and water vapor in the atmosphere. The oxygen is measured in the K-band (51-58 GHz) and the water vapor in the V-band from 22 to 31 GHz with a radiometric resolution between 0.3 and 0.4 rms errors at 1.0 s integration time.
The retrievals of temperature profiles from the measured brightness temperatures are performed using a standard feed forward neural network (Rose et al., 2005). A detailed

In-situ instrumentation
The integrating nephelometer (model TSI 3563) at SNS measures the particle light scattering coefficient ( ) at three wavelengths (450, 550 and 700 nm) with 1-min temporal resolution.

170
The aerosol flow in the nephelometer was set to 30 lpm. The nephelometer measurements are within the angular range 7-170º, so the data were corrected for truncation and non-Lambertian The SMPS measurements followed ACTRIS and GAW recommendations (Wiedensohler et 185 al., 2012(Wiedensohler et 185 al., , 2018 and high-quality data were guaranteed after the successful participation of the instrument in the ACTRIS inter-comparisons workshops (TROPOS, Leipzig, Germany) and in-situ intercomparison (ACTRIS Round Robin Tour). The Aerodynamic Particle Sizer (APS; TSI Mod. 3321) provides the coarse particle number size distribution within the 0.5-20 µm aerodynamic diameter range. The APS also measures number aerosol concentrations up to 190 1000 particles·cm -3 with coincidence errors inferior to 5% and 10% at 0.5 and 10 μm diameters, respectively. By the combination of SMPS and APS measurements, total aerosol volume concentrations were obtained in the 0.05-10 μm radius range with 5-min time resolution. To that end, Q-value=1 has been assumed for conversion from aerodynamic to mobility size distribution (Sorribas et al., 2015).

Methodology
In this work, we use the GRASP algorithm following the scheme proposed by Lopatin et al. (2013), which combines lidar and sun-sky photometer measurements to retrieve the optical and microphysical properties of aerosol particles. This scheme uses normalized backscattered range corrected signal at 355, 532 and 1064 nm and the AOD and sky radiance (almucantar scan) 225 both at 440, 675, 870 and 1020 nm from AERONET version 3 level 1.5. It should be noted that GRASP retrievals were performed during daytime with solar zenith angles larger than 40º and clear-sky conditions. This configuration of GRASP allows the retrieval of aerosol properties for both fine (radii range 0.05 to 0.576 µm) and coarse (radii range 0.33 to 15 µm) modes separately, the complex refractive index, single-scattering albedo (SSA) and lidar ratio (LR).

Aircraft data
In order to make comparable the profiles from the aircraft data and the remote sensing retrievals, there are some corrections to consider. Remote sensing data are provided at ambient conditions (temperature and pressure), but the aircraft data is registered at different conditions.

250
Nephelometer data from the aircraft were recorded at cabin temperature and ambient pressure, and aethalometer data were registered at 0ºC and 1013.25 hPa. The cabin temperature used was the nephelometer sampling temperature (Ts), i.e. temperature inside the nephelometer, and the profile atmospheric pressure used was the nephelometer pressure sensor (Ps). The cabin on the aircraft was not pressurize so the pressure inside the nephelometer can be consider the outside 255 pressure. The aircraft did not register the outside temperature, so an external source of temperature profile was required. We used a temperature profile from a microwave radiometer MWR (Tmwr) as described in section 2.1., using an average profile during the time of the entire aircraft profile and interpolated to the exact altitudes of the aircraft profile.
Aircraft profiles show some noise, especially at higher altitudes, so a convolution with 260 a mean filter was applied to the aircraft in-situ data in order to smooth the profiles. We observed that using 100 meters for the nephelometer and 200 meters for the aethalometer data in the vertical profiles reduced noise while preserving the profile features. Finally, Aurora nephelometer wavelengths were converted to the TSI wavelengths using the Ångström exponent law to make the aircraft and ground based in-situ data comparable.  and total volume concentrations, respectively. The results show that GRASP retrievals overestimate in-situ measurements with a mean difference (±standard deviation) of 4 ± 4 µm 3 /cm 3 and 6 ± 8 µm 3 /cm 3 for fine and total volume concentrations, respectively. In contrast, better correlation is observed for coarse mode volume concentrations (slope equals to 1) with 285 a lower mean difference (2 ± 6 µm 3 /cm 3 ). In terms of absolute concentrations, 65% (91%), 70% (88%) and 45% (71%) of the differences are observed within ±5 µm 3 /cm 3 (±10 µm 3 /cm 3 ) for fine, coarse and total volume concentrations, respectively. These results are similar to those found in previous GRASP assessments by Benavent-Oltra et al. (2017) and Tsekeri et al. (2017). Those authors also showed an overestimation of VCF compared with in-situ data, while 290 for VCC similar GRASP retrievals to in-situ data was found for cases with coarse particles predominate. The observed overestimation is lower than the obtained by Román et al. (2018) using GRASP with ceilometer data, and by Benavent-Oltra et al. (2019)  intercepts (from 0.5 to 1.5 Mm -1 ) of the regressions. The mean differences (±standard deviation) of 2 ± 6 Mm -1 , 1 ± 3 Mm -1 and 0.8 ± 1.7 Mm -1 at 370, 520 and 880 nm, 310 respectively. Furthermore, the differences between GRASP and in-situ measurements are less than ±2.5 Mm -1 for 61%, 81% and 90% of the data at 370, 520 and 880 nm, respectively. The results from Figure  Finally, the comparison between GRASP retrievals and in-situ data for extinction coefficient showed in Figure 4a evidence better agreement. The GRASP retrievals and in-situ data show good agreement (slope equals to 1) and are highly correlated ( = 0.9). Figure 4b shows the frequency histogram of the differences in extinction coefficient (∆ ) between 320 GRASP and in-situ, showing a skewed histogram to positive differences that implies slightly overestimation by GRASP (75% of these differences within ±15 Mm -1 ). These overestimations can be associated with the differences in scattering coefficient. [ Figure 5] For , both GRASP and airborne measurements follow the same pattern where 340 GRASP overestimates the airborne data with a mean absolute difference of 14 ± 10 Mm -1 .
During SLOPE I, these mean absolute differences are lower than 8 Mm -1 and there is a good agreement between GRASP and SNS measurements (differences <4 Mm -1 ). However, during SLOPE II, the differences between GRASP and in situ measurements (both airborne and SNS) are larger, reaching values of 30 Mm -1 . In the case of , GRASP and airborne profiles show 345 large differences during SLOPE I with mean absolute differences between 0.5 and 3 Mm -1 reaching differences around 6 Mm -1 on 18 th June 2016. On the other hand, the absorption coefficients retrieved by GRASP show good agreement within situ measurements (both airborne and SNS) with a mean absolute difference of 0.7 ± 0.4 Mm -1 during SLOPE II. In general, the differences between GRASP and in situ measurements are close to the detection 350 limit for the aethalometer on-board the airplane and SNS. The differences obtained both for and can be explained due to the low AOD440 (below 0.40) that represents a challenge for the retrieval of the aerosol properties both for AERONET  and inversion algorithms as GRASP (Lopatin et al., 2013   [ Figure 8]

Vertically-resolved
The extinction, scattering and absorption coefficients profiles at 532 nm (Fig 8 c,  profile approximately 10% of total extinction corresponds to absorption. Thus, GRASP retrieval combining several remote sensing instruments presents a step forward to aerosol characterization because permits characterizing aerosol absorption with vertical resolution and for lower aerosol loads than classical AERONET inversion. For SAE (Figure 8h), which is more related to the predominant particle size, the highest value is found at the lowest altitude, suggesting larger predominance of fine particles closer to the surface. This pattern agrees with the assumption of higher anthropogenic aerosol loads at these levels are dominated by fine mode particles, while at altitudes above the atmospheric boundary

Special Events
During the SLOPE I and II campaigns two extreme events with AOD440 ~ 1.0 were registered.
The first one was a Saharan mineral dust outbreak (DD) in July 2016, and the second one was a biomass burning transport event (BB) in July 2017 with fires origin in Portugal. Figure 9 and GRASP (Li et al., , 2020 and to be explored in the future.

Conclusions
In this study, we presented an overview of aerosol optical and microphysical properties retrieved with GRASP code during SLOPE I and II field campaigns. The measurements from lidar and sun-sky photometer performed on May, June and July 2016 and 2017 were used as 495 input data in GRASP to retrieve these aerosol properties.
The in-situ measurements performed at Sierra Nevada Station during SLOPE I and II campaigns, and the airborne measurement gathered during special periods on both campaigns allowed the assessment of aerosol properties retrieved by GRASP code at 2.5 km a.s.l. and for the whole profile, respectively. The volume concentration comparison shows better agreement 500 for coarse mode (R>0.8) than for fine mode due to the few cases (15%) with predominating fine particles. For the scattering and absorption coefficients, the differences between GRASP data at 2.5 km a.s.l. and in-situ measurements are lowest for longest wavelengths, with differences of 11 ± 17 Mm -1 at 450 nm and 2 ± 6 Mm -1 at 370 nm for and , respectively. The agreement between GRASP and in-situ measurements at SNS is solid for 505 both for scattering and absorption coefficients. In general, GRASP somewhat overestimates the in-situ data at 2.5 km a.s.l.. These differences (14 ± 10 and 1.2 ± 1.