The characterization of Taklamakan dust properties using a multi-wavelength Raman polarization lidar in Kashi, China

The Taklamakan desert is an important dust source for the global atmospheric dust budget and a cause of the dust weather in Eastern Asia. The characterization of the properties and vertical distributions of Taklamakan dust in the source region is still very limited. To fill this gap, the DAO (Dust Aerosol Observation) was conducted in Kashi, China in 2019. Kashi site is about 150 km to the west rim of the Taklamakan desert and is strongly impacted by desert dust aerosols, especially in spring time, i.e. April and May. Apart from dust, fine particles coming from local anthropogenic emissions or/and transported 5 aerosols are also a non-negligible aerosol component. In this study, we provide the first profiling of the 2α+ 3β+ 3δ lidar profiles of Taklamakan dust based on a multi-wavelength Raman polarization lidar. Four cases, including two Taklamakan dust events (Case 1 and 2) and two polluted dust events (Case 3 and 4) are presented. The lidar ratio in the Taklamakan dust outbreak is found to be 51±8 – 56±8 sr at 355 nm and 45±7 sr at 532 nm. The particle linear depolarization ratios are about 0.28±0.04–0.32±0.05 at 355 nm, 0.35±0.05 at 532 nm and 0.31±0.05 at 1064 nm. The observed polluted dust is commonly 10 featured with reduced particle linear depolarization ratio and enhanced extinction and backscatter Ångström exponent. In Case 3, the lidar ratio of polluted dust is about 42±6 sr at 355 nm and 40±6 sr at 532 nm. The particles linear depolarization ratios decrease to about 0.25, with a weak spectral dependence. In Case 4, the variability of lidar ratio and particle linear depolarization ratio is higher than in Case 3, which reflects the complexity of the nature of mixed pollutant and the mixing state. The results provide the first reference for the characteristics of Taklamakan dust measured by Raman lidar. The data 15 could contribute to complementing the dust model and improving the accuracy of climate modeling.

75.93 • E, time zone: GMT+08:00) is in the northwest of the Kashi city and close to the border to Tajikistan, Kyrgyzstan and Afghanistan. Kashi features a desert climate with a big temperature difference between winter and summer. The coldest month is January with average temperature of -10.2-0.3 • C and the warmest month is in July with average temperature of 18.6-32.1 • C.
The annual rainfall in Kashi is about 64 mm. The spring in Kashi is long and comes quickly. The rapidly heated surface sand in the desert could generate ascending currents which result in the frequent dust storm in springtime. This is the main reason 70 that the field campaign was performed in springtime.
Except for desert dust, anthropogenic emission is the other important aerosol source. There are about 4.65 million habitants (predicted for 2017, see the link) in the Kashi prefecture, including the Kashi city and 11 subordinate counties. Kashi prefecture is a very populated region in Xinjiang with more than 1000 persons per square kilometer in the city center (Doxsey-Whitfield et al., 2015). Moreover, there are populated cities in the neighboring countries such as Kyrgyzstan,Tajikistan and Pakistan. 75 Under favorable meteorological conditions, various aerosol, for example, pollution, could be potentially transported to Kashi and mix with dust aerosols.

Lidar system
The multi-wavelength Raman polarization lidar is the main instrument installed in observation site. The lidar system, LILAS 80 (Lille Lidar Atmosphere Study) has been operated in LOA (Laboratoire d'Optique Atmosphérique, Lille, France) since 2013 (Bovchaliuk et al., 2016;Veselovskii et al., 2016;Hu et al., 2019). During the DAO campaign, LILAS was transported from Lille to Kashi (and Beijing in the second session of the campaign) to perform observations. LILAS uses a Nd: YAG laser that emits at three wavelengths: 355, 532 and 1064 nm. The laser repetition rate is 20 Hz. A Glan prism is used to clean the polarization of the laser beam. The emitting power after the Glan prism is about 70, 90 and 100 mJ at 355, 532 and 1064 85 nm, respectively. LILAS system has three Raman channels, including 387 (vibrational-rotational), 530 (rotational) and 408 nm (water vapor). The backscattered light is collected using a 400 mm Newton telescope. The incomplete overlap range of LILAS system is about 1000-1500 m in distance, depending on the selected field of view angle. In the receiving optics, the three elastic channels are equipped with both a perpendicular and a parallel channel in order to measure the linear depolarization ratio at three wavelengths. The polarization calibration is performed following the ±45 • method (Freudenthaler et al., 2009). During 90 the DAO campaign, the polarization calibration has been performed at least once per day. LILAS can provide the profiles of the 2α+3β+3δ (2 extinction coefficients + 3 backscatter coefficients + 3 particle linear depolarization ratios) parameters. The Ångström exponent of the extinction coefficient and backscatter coefficient are calculated by the Equation 1: where p(λ) represents the optical parameters, such as AOD, extinction or backscatter coefficient at wavelength λ,Å represents 95 the Ångström exponent of the corresponding parameters p(λ). The statistical error of lidar derived parameters is estimated to be 10% for the extinction and backscatter coefficient, and 15% for the lidar ratios and particle linear depolarization ratios. The errors for the water vapor mixing ratio and relative humidity are about 20%. The other error sources, such as alignment of laser beam and errors in the selection of reference, are not considered in the error estimate.
Sun/sky photometer 100 Three sun/sky photometers are deployed in the Kashi observation site. One is affiliated to the AERONET (AErosol RObotic NETwork, Holben et al. (1998)) network and the other two are affiliated to SONET (Sun-Sky Radiometer Observation Network). SONET is a ground-based sun/photometer network with the extension of multi-wavelength polarization measurement capability to provide long-term columnar atmospheric aerosol properties over China (Li et al., 2018). The three sun/sky photometers provide complementary measurements by following different measurement protocols. In all, they can measure day-105 time aerosol optical depth (denoted as AOD hereafter) at 340, 380, 440, 675, 870, 1020 and 1640 nm, polarized/unpolarized sky radiances at 440, 675, 870 and 1020 nm and moon AOD as well. The succeeding data treatment and retrieval are performed following the protocols and standards of AERONET or SONET, depending on the affiliation of the instruments.

Satellite data
Satellite data have complementary advantages due to their large spatial coverage compared to ground-based remote sensing 110 technique. In order to show the activity of the desert, we use the UV aerosol index (UVAI hereafter) derived from the OMPS (Ozone Mapping Profiler Suite) onboard the Suomi-NPP (National Polar-orbiting Partnership) satellite (Flynn et al., 2004;Seftor et al., 2014). OMPS provides full daily coverage data and the overpass time for Kashi region is around 06:30 UTC. The UVAI is calculated using the signal in the 340 and 380 nm channels (Hsu et al., 1999): UVAI = −100 × log 10 I 340 I 380 meas − log 10 115 where I ... represents the backscattered radiance at corresponding wavelength. The subscripts "meas" and "calc" respectively represent the real measurements and model simulation in a pure Rayleigh atmosphere. By the definition of UVAI, its positive values correspond to UV-absorptive aerosols such as desert dust and carbonaceous aerosols. Hence, the UVAI from OMPS is a good parameter for monitoring the activity of the Taklamakan desert.
A radiosonde station (39.47 • N, 75.99 • N) in Kashi is 6 km to the observation site. The data are accessible on the website of the Wyoming weather data website (see the link). The radio sounding data are recorded at 00:00 and 12:00 every day at local time.
They provide the vertical temperature and pressure profiles for the calculation of molecule scattering parameters in lidar data Besides, instruments measuring particulate matter (PM10 and PM2.5), gas concentration (SO2, O3 and NOx), particle size distribution, particle scattering and absorption coefficients, solar radiation and a cloud monitor are also deployed in the field campaign. These data contribute to relevant air quality and solar radiation studies within the frame of the DAO campaign. June and July. The Ångström exponent is positively correlated to the FMF and negatively correlated to the AOD. The lowest mean Ångstöm exponent occurs in March and April, indicating that dust particles are dominant due to the seasonal increase of dust activities in this period (Littmann, 1991;Qian et al., 2002). In December and January, the Ångström exponent and FMF increase significantly, which proves that fine particles are an important aerosol source in Kashi. The fine particles are mostly originated from heating, biomass burning, traffic and industrial pollution in the local area. 2000 m. Figure 6 shows the profiles of the optical properties, water vapor mixing ratio (WVMR) and relative humidity ( variations of the lidar ratios and PLDRs are possibly the result of particle sedimentation or/and vertically dependent particle origins.
On 09 April, the Taklamakan desert is covered by a low-pressure zone with easterly and northeasterly wind prevailing over 175 the western part of the desert. It is a favorable condition for the elevation of dust particles. Figure 7 shows the 48-hour back trajectory ending at 20:00 UTC for air mass at 1000, 2000 and 3000 m. The air masses at the three vertical levels are originated from the Taklamakan desert. They all passed over the area where dust plumes have been observed and then diverged when approaching the rim of the desert. In the end, the air masses at 1000, 2000 and 3000 m arrived at the observation site from the northeast, east and southeast respectively, after being lifted from near the surface.

Case 2: 24 April 2019
On 24 April, the observation site was enclosed by floating dust. In the daytime, the sky radiance dropped below the detection limit of the sun/sky photometer, so the AERONET and SONET retrieval can not be applied. A large and intense plume was site. The daily average of AOD is 3.63 and Ångström exponent is about -0.01, according to the daytime sun/sky photometer measurements. The lidar quicklook on 24 April in Figure 5 shows that the boundary layer height rises from about 1200 m to 2000 m from 14:00 to 24:00 UTC. Due to the high dust attenuation in the boundary layer, both sun/sky photometer and lidar cannot detect whether clouds exist on 24 April. Figure

Case 3: 15 April 2019
The daily mean AOD on 15 April was 0.63, and the Ångström exponent was about 0.10. Compared to the previous two cases,  properties could modify due to deposition and mixing with various aerosols. Hence, the complexity of aerosol properties at the boundary layer top is difficult to be resolved.

255
The optical parameters in the 4 cases are summarized in Table 2. In order to distinguish Taklamakan dust observations, we define Taklamakan dust by EAE 355−532 smaller than 0.1 and PLDR 532 greater than 0.32 at 532 nm. Back trajectories are also used as a reference for identifying the aerosol origins. The observations falling beyond this category are classified as polluted dust.
Aerosol source The evidence of pollution in Taklamakan dust has been found in previous in-situ measurements. Huang et al. (2010) sampled aerosol particles in springtime at Tazhong site, which is located in the north rim of Taklamakan desert, and found that the As 270 element was moderately enriched. The As element is a tracer of pollution, originated probably from coal burning. It is also found that the concentration of sulfate in Taklamakan dust is at a high level. The increased concentration of sulfate in the Taklamakan dust could be related to the provenance of the Taklamakan desert, because it is speculated to be ocean 5-7 millions years ago (Sun and Liu, 2006). Sulfate could also come from anthropogenic emission, for example, the uptake of the SO2 gases. It is in agreement with our conclusion, however, in this study we cannot clarify the exact involving aerosol species and the mixing state in the polluted dust. In our study, polluted dust mostly appeared at the boundary layer top, which agrees with  Figure 14 plots    In Saharan dust observations, both spectrally dependent and independent lidar ratios at 355 and 532 nm have been reported. Yumimoto, K., Eguchi, K., Uno, I., Takemura, T., Liu, Z., Shimizu, A., and Sugimoto, N.: An elevated large-scale dust veil from the Taklimakan Desert: Intercontinental transport and three-dimensional structure as captured by CALIPSO and regional and global models, Atmospheric Chemistry and Physics, 9, 8545-8558, 2009.