Californian Wildfire Smoke Over Europe: A First Example of the Aerosol Observing Capabilities of Aeolus Compared to Ground‐Based Lidar

In September 2020, extremely strong wildfires in the western United States of America (i.e., mainly in California) produced large amounts of smoke, which was lifted into the free troposphere. These biomass‐burning‐aerosol (BBA) layers were transported from the US west coast toward central Europe within 3–4 days turning the sky milky and receiving high media attention. The present study characterizes this pronounced smoke plume above Leipzig, Germany, using a ground‐based multiwavelength‐Raman‐polarization lidar and the aerosol/cloud product of ESA’s wind lidar mission Aeolus. An exceptional high smoke‐AOT >0.4 was measured, yielding to a mean mass concentration of 8 μg m−3. The 355 nm lidar ratio was moderate at around 40–50 sr. The Aeolus‐derived backscatter, extinction and lidar ratio profiles agree well with the observations of the ground‐based lidar PollyXT considering the fact that Aeolus’ aerosol and cloud products are still preliminary and subject to ongoing algorithm improvements.

profiles since 2006 (e.g., Young et al., 2018). While the main focus of the Aeolus mission is to obtain atmospheric horizontal winds, the lidar is also capable to retrieve aerosol and cloud optical properties as spin-off products (Ansmann et al., 2007;Flamant et al., 2008) via the HSR lidar (HSRL) technique. This technique is a spaceborne novelty. Thus, one of the goals of this Earth Explorer Mission is to prove the concept of this new technology in space.
While the exploitation of the aerosol products and algorithm development of Aeolus is still ongoing, we want to present a first glimpse on the Aeolus' capabilities to measure vertical aerosol profiles. For this purpose, we investigated dense smoke layers observed near Leipzig, Germany, originating from very strong and intense wildfires in the western United States (US) of America. In 2020, wildfires, for example in California, have been dramatically intensive, also triggering the discussion of an increased risk of wildfires due to climate change (Brando et al., 2019;Wehner et al., 2017). The resulting smoke plumes from such fires, once lifted to the free troposphere and/or lower stratosphere (UTLS) can travel several thousand kilometers and influence the aerosol load even in very remote regions (Baars et al, 2011(Baars et al, , 2012Fromm et al., 2010;Kloss et al., 2019). In late summer 2017, an intense event of smoke-plume advection from Canadian wildfires was monitored in central Europe Haarig et al., 2018;Hu et al., 2019;Khaykin et al., 2018). At that time, tropospheric smoke plumes were remarkable but stratospheric smoke was even more intense, lasting for more than half a year above the tropopause over the European continent (Baars et al., 2019). Similar observations were made in the Southern Hemisphere during the recent Australian fires (austral summer 2019/2020, Ohneiser et al., 2020;Khaykin et al., 2020).
Here, we present an example of the most recent smoke event originating from fires in south-western North America (mainly in California) and observed over Leipzig, Germany in the frame of the German initiative for the Experimental Validation and Assimilation of Aeolus observations (EVAA, e.g., Baars, Geiβ, et al., 2020;Geiβ et al., 2019). The observations were performed on September 11, 2020 at the Leibniz Institute of Tropospheric Research (TROPOS) in Leipzig. Our objective is to present a first assessment on the Aeolus aerosol observing capabilities and to show the dramatic impact of fires in the western United States of America on atmospheric conditions over central Europe-it was the highest perturbation of the troposphere in terms of Aerosol Optical Thickness (AOT) ever observed at TROPOS since the start of the regular lidar observations in 1997. Aeolus is currently the only spaceborne lidar able to measure independently backscatter and extinction of aerosol and clouds and thus deliver the particle-type-depending extinction-to-backscatter ratio, called lidar ratio. It therefore provides valuable information on the radiative properties of these smoke aerosols. Furthermore, Aeolus is the only satellite in space that measures simultaneously aerosol properties together with profiles of the horizontal wind speed. This unique configuration is expected to improve the predictions of such extreme events as well as air quality forecasts (European Space Agency, 2008;Horányi et al., 2015a;Letertre-Danczak et al., 2020). Figure 1 shows the atmospheric conditions over Leipzig, Germany on September 11, 2020 as measured with the ground-based multiwavelength-Raman-polarization lidar PollyXT  organized within the voluntary, scientific network PollyNET  and part of EARLINET/ACTRIS (Pappalardo et al., 2014). The time-height plot of the uncalibrated attenuated backscatter coefficient at 1064 nm is presented to indicate the backscatter intensity of atmospheric particles. While blueish colors represent no significant particle concentration and thus almost pure molecular scattering, green and reddish colors indicate a moderate to high aerosol load. White colors represent very strong backscattering caused by clouds. Besides some low-level clouds and local aerosol below 2.5 km before 05:00 UTC, first aerosol layers, namely smoke from Californian fires, arrived at an altitude of about 12 km at 03:00 UTC as seen in Figure 1. The arrival at this height range is most probably a result of the higher wind speeds in the jet stream region. Later through the day, smoke layers were observed throughout the whole troposphere down to about 4.5 km altitude and thus well above the local planetary boundary layer (PBL). The total AOT in Leipzig increased to values of 0.6 at 380 nm and 0.4 at 532 nm between 14:00 and 17:00 UTC as observed with an AERONET sun photometer (Giles et al., 2019) on this day. Smoke layers from this event could be observed at TROPOS until 14 September.

Californian Smoke Over Central Europe
To characterize the atmospheric state over Leipzig on September 11, 2020, we exemplary analyze the airmasses arriving at noon (12 UTC) at about 8.5 km-the height of the estimated center of the smoke layer according to Figure 1. A 10-day FLEXPART (Pisso et al., 2019;Stohl et al., 2002) backward simulation of 500 air parcels originating at Leipzig between 8.3 and 8.7 km height on September 11, 2020 at 12 UTC was performed. Five clusters were generated by grouping the simulated air parcels based on their location at three-hourly intervals. The result is shown in Figure 2. Cluster size is indicated by the size of the dots, with larger clusters representing more air parcels. Colors are assigned to clusters that are closest to each other BAARS ET AL.   at two consecutive times. However, they do not represent a continuous trajectory, as the particles for each cluster might have changed.
This backward simulation proves that the airmasses arriving over Leipzig at 8.5 km height at 12:00 UTC originated from the west coast of North America, where intense fires were taking place during this time as indicated by the very high fire radiative power as shown in Figure 2, left (Giglio, 2016). Fire frequency and strength in California increased during the first week of September. Slightly weaker fires were observed in Oregon, Washington and Montana. Due to the prevailing winds, the travel time of the smoke from the US West Coast to Europe was only 3-4 days. The airmass was transported across the Atlantic Ocean in a long filament at approximately 51°N on 1 day only (9 September).
The time-height cross-section of airmass source ( Figure 3, Radenz et al., 2021) generalizes the findings from the single time and height analysis as shown in Figure 2. Air parcel positions are calculated every 3 h and for height intervals of 500 m. At each step, the residence time is summed up for each region, whenever particles were below the reception height during the backward simulation. The reception height of 5 km was set to agree to with typical injection heights caused by wildfires of similar strength (Val Martin et al., 2018) and shall not be confused with arrival height of the plume at Leipzig. The grouped residence time profiles are then visualized in the time-height-cross-section. According to Figure 3, airmasses originating at the west coast of North America were present over Leipzig during the whole day, being especially prominent above 4 km after 9 UTC. The results suggest that in this case, smoke injection into the middle troposphere was sufficient to be subject to long-range transport. Please note, that the actual injection could also occur at lower altitudes, but would still be covered by the 5 km threshold. As shown in Figure 2, the lifting to 8.5 km occurred during the transport on the 8 September.
It can therefore be concluded that the strong wildfires in California in 2020 produced large amounts of smoke, which were then transported toward Europe.

Aerosol Optical Profiles of Aeolus Compared to Ground-Based Lidar
Aeolus reached central Europe around 16:45 UTC on September 11, 2020, as shown in Figure 4. Here, the backscatter coefficient of Aeolus retrieved with the HSRL technique (called Standard Correction Algorithm [SCA] product; Flamant et al., 2017;Huber et al., 2016) is visualized with VirES (Santillan et al., 2019) for this overpass. Despite the noise in the backscatter product, this intense aerosol plume is clearly visible ranging from Leipzig to southern Norway (greenish colors). The very strong backscatter values (bright yellow) over central Norway were caused by clouds also indicated by deep blue colors below (no signal due to complete attenuation of laser beam in optically thick clouds). Note that Aeolus aerosol profiles are currently only available with a horizontal resolution of 87 km and vertical resolution from 500 to 2000 m depending on altitude (500-1000 m in the troposphere). We used the ground-based Raman lidar observations made 2-3 h later at Leipzig to compare them with the Aeolus observations made around 100 km north-west of Leipzig (indicated with "+1" in Figure 4) instead of presenting the comparison with the closest overpass. Due to the strong solar background at the time of the overpass (around 17:45 local solar time), Raman retrievals (Ansmann et al., 1992) of the ground-based lidar delivering the backscatter and extinction coefficient independently were not possible. Backward trajectories indicate that airmasses traveled around 3 h from the Aeolus track 100 km north toward Leipzig. For this reason, we used the night time observation made between 20:00 and 21:30 UTC for the comparison as visualized in Figure 1. Considering no significant changes in the aerosol distribution within the smoke plume over this short time and thus only advection as result of the mean wind fields, a good collocation criterion is found to allow the comparison of the ground-based Raman retrievals with the HSRL retrievals of Aeolus. A similar procedure (Papagiannopoulos et al., 2016;Tesche et al., 2013) has also been applied for Calibration/ Validation activities of the spaceborne lidar on-board CALIPSO (Winker et al., 2009). Figure 5 shows the comparison of the ground-based Raman-polarization lidar PollyXT with the Aeolus observation as described. The particle backscatter and extinction coefficient, along with the corresponding lidar ratio (extinction-to-backscatter ratio), are shown (Raman methodology for PollyXT and HSR methodology for Aeolus). For Aeolus, the HSR product SCA and a HSR product obtained by smoothing two vertical bins called "SCA mid bin" are presented. These two products are standard output.
As the Aeolus lidar data are very noisy and the provided uncertainties are currently overestimated, the quality flags included in the Aeolus aerosol spin-off (L2A) product are not yet mature (Reitebuch et al., 2020). For this reason, we applied the following quality control criteria for Aeolus data: 1. Exclude negative backscatter and extinction coefficient values 2. Absolute value of the provided extinction error must be less than 100 and 275 Mm −1 for SCA and SCA mid bin, respectively As seen in Figure 5, left, an excellent agreement regarding the backscatter coefficient (at 355 nm) between the ground-based lidar (black thick line) and the Aeolus SCA algorithms (red and blue lines) is found for the lofted smoke layer between 4 and 12 km. In the PBL, Aeolus obtains higher values than the lidar at Leipzig, probably due to the coarse horizontal resolution of Aeolus and the spatio-temporal variations in terms of optical particles properties within the PBL (e.g., due to hygroscopic growth). In addition, sporadic contamination of the 87 km aggregated Aeolus observation with small boundary layer clouds cannot be ruled out.
BAARS ET AL.
10.1029/2020GL092194 5 of 10 It should be noted that since Aeolus is only able to detect the co-polar backscatter component of the circularly polarized emitted light (Ansmann et al., 2007;Flamant et al., 2008), the ground-based PollyXT observations at 355 nm (dotted, black line) had to be transformed to an Aeolus-like profile by means of the measured particle linear depolarization ratio -for more details see, for example, Baars, Geiβ, et al. (2020). Nevertheless, as the smoke particles arriving from the US did almost not depolarize the light during the backscattering process (particle linear depolarization ratio at 355 nm below 10% according to PollyXT), almost no difference between the total particle backscatter profile and the Aeolus-like (co-polar) backscatter profile is evident.
The extinction coefficient profiles of Aeolus are in reasonable agreement with the ground-based ones (Figure 5,center panel). Due to the utilization of molecular scattering processes for the extinction retrievals, the profiles are naturally much noisier compared to the backscatter ones. Nevertheless, both shape and magnitude of the extinction profile agree well within the pronounced smoke layer between 4 and 8 km for both retrievals. The SCA mid-bin algorithm provides more robust results due to averaging of two neighboring height bins and should be prioritized for extinction analyses. Sporadic values for extinction above 8.5 km are provided by both algorithms, an effect that is attributed to instrumental noise. Compared to the groundbased lidar profiles, these values above 8.5 km are unrealistically high. A series of algorithm improvements (including quality flags, improved calibration and error estimates etc.) are still under development and are foreseen to avoid such artifacts in future. Therefore, the current Aeolus aerosol optical data is still flagged as preliminary.
The message for this example case is that Aeolus is able to measure -as a first spaceborne lidar -backscatter and extinction independently in pronounced aerosol layers, and thus is able to directly derive the lidar ratio as shown in Figure 5, right. We thus focus in the following on the intensive aerosol properties within the main smoke layer below 8 km. Within this pronounced aerosol layer, the agreement between the lidar ratio observed by the ground-based lidar (and converted to the co-polar (Aeolus-like) component) and the one observed by Aeolus is reasonable (please remind, that current Aeolus uncertainties are overestimated and thus unrealistically high). The PollyXT observations yield a lidar ratio of the smoke of about 40-50 sr at 355 nm, which is in excellent agreement with former observations of long-range transported smoke (Haarig BAARS ET AL. 10.1029/2020GL092194 6 of 10 Figure 5. Comparison of the optical profiles provided by Aeolus (red and blue) and the ground-based EARLINET lidar PollyXT (black lines). The particle backscatter coefficient (left), the particle extinction coefficient (center), and the corresponding lidar (extinction-to-backscatter) ratio is shown. The groundbased backscatter observations (dashed line) are converted to Aeolus-like, that means, co-polar, profiles (solid line) for comparability. The mean coordinates and distance of the 87 km long Aeolus observations to the ground-based lidar are given as well. Ohneiser et al., 2020). Aeolus measured values within the same range, even though some outliers in the smoke region exist (at around 6 km). These outliers are mainly caused by the noisy signals but algorithm development to reduce these effects are ongoing. As for the extinction, the SCA mid bin algorithm is to be preferred against the SCA for the lidar ratio analysis.
The missing depolarization component of Aeolus did not restrict the characterization of the dominant aerosol types in the lofted smoke layer, since the particles were mainly spherical and thus non-depolarizing. However, in cases of a more heterogeneous aerosol field, polarization capabilities would have been highly desirable also with respect to a potential aerosol type identification (e.g., polarizing mineral dust and volcanic ash) and aerosol/cloud discrimination Baars et al., 2017;Pappalardo et al., 2013). Such capabilities are implemented in the future EarthCARE mission (Wandinger et al., 2016).
From the multiwavelength Raman polarization PollyXT measurements, it is also possible to derive the smoke mass concentration as described by Mamouri and Ansmann (2016) and Ansmann et al. (2020). For this purpose, an extinction-to-volume conversion factor of 0.169 × 10 −12 Mm at 532 nm and a particle density of 1.3 g cm −3 have been used for the smoke layer. These values, which are required for the conversion of particle extinction coefficient to mass concentration, correspond to semi-aged smoke according to Ansmann et al. (2020). The procedure yields to a maximum smoke mass concentration of 22 μg m −3 in the smoke layer peak around 5 km (maximum extinction coefficient of 160 and 100 Mm −1 at 355 and 532 nm, respectively). Integrating the mass concentration over the entire smoke layer leads to a total mass of 64,757 μg m −2 that has been observed at Leipzig between 20:00 UTC and 21:30 UTC (corresponding to a mean concentration of 8 μg m −3 ). The AOT of the smoke layer (from 3.8 to 11 km) was 0.45 and 0.29 at 355 and 532 nm, respectively, and thus about 2/3 of the total AOT, showing the dramatic impact of the Californian wildfires on the Central European troposphere.

Conclusions
The potential of Aeolus to independently retrieve aerosol backscatter and extinction coefficients was demonstrated in this study by means of a comparison to a high-quality ground based lidar. An extreme event of smoke advection from the US west coast which was observed over Leipzig, Germany, Central Europe, served as an example. Besides Aeolus' observing capabilities, this study demonstrated that strong wildfires, such as the ones occurred in western US in 2020, can emit huge smoke amounts, which are comparable to volcanic eruptions (Peterson et al., 2018). These extraordinary smoke plumes are able to travel relatively fast over large distances between the continents and thus might influence radiation, clouds, and weather at even remote places usually not directly affected by wildfires and several thousand kilometers away from the source.
The intense smoke plume, which arrived at the measurement location of Leipzig on September 11, 2020, caused a perturbation of the troposphere for 4-5 days. A smoke-related AOT of 0.45 at 355 nm was measured above Leipzig with estimated maximum and mean mass concentrations of the smoke of 22 and 8 μg m −3 , respectively. Aeolus captured this smoke plume near the ground-based station of Leipzig on the first day of the event during its regular overpass. Despite the coarse horizontal (≈ 87 km) and vertical (500-1000 m) resolution of Aeolus' aerosol optical profiles, it was shown that the HSRL technique can be successfully applied from space. Within the main smoke layer, the Aeolus-derived backscatter, extinction and lidar ratio profiles compare well with the high-quality ground based lidar considering the still ongoing algorithm improvements and uncertainty estimation refinements.
Measuring the lidar ratio directly from space is a spaceborne novelty and boosts aerosol-cloud interaction research into a new era. The results presented here indicate that Aeolus is partly capable of filling the gap between the ending CALIPSO and the upcoming EarthCARE mission (Illingworth et al., 2015). However, further sensitivity studies on how well Aeolus is able to correctly determine extinction and backscatter coefficients in less pronounced aerosol layers are still ongoing and one main focus of current Cal/Val activities.

Data Availability Statement
The ground-based Polly lidar data is visualized at polly.tropos.de and will become publicly available via the data portal of the European Aerosols, Clouds and Trace gases Research Infrastructure (ACTRIS) when its implementation phase is completed. Aeolus aerosol spin-off data used in this publication is not yet publicly available, but, after reprocessing in spring 2021, including improved radiometric calibration and quality flags, an updated data set will become freely available via ESA services (i.e., https://aeolus-ds. eo.esa.int/oads/access/ and the online VirES tool at https://aeolus.services/). Nevertheless, to allow the FAIR data principles, all data used for this study are available at zenodo.org under https://doi.org/10.5281/ zenodo.1115995.

Acknowledgments
This research has been supported by the German Federal Ministry for Economic Affairs and Energy (BMWi) under grant no. 50EE1721C. Authors also acknowledge support through ACTRIS-2 under grant agreement no. 654109 from the European Union's Horizon 2020 research and innovation program, and through PoLiCyTa from the German Federal Ministry of Education and Research (BMBF) under grant no. 01LK1603A. We also appreciate very much the fruitful discussions within the EVAA consortium (LMU, DWD, DLR), with the "Aeolus DISC" team, and the Aeolus Team of ESA. The processor development, improvement and product re-processing preparation are performed by the Aeolus DISC (Data, Innovation and Science Cluster), which involves DLR, DoRIT, ECMWF, KNMI, CNRS, S&T, ABB, and Serco, in close cooperation with the Aeolus PDGS (Payload Data Ground Segment). The analysis has been performed in the framework of the Aeolus Scientific Calibration & Validation Team (ACVT).