Can Ultra-violet Mie Lidar Be an Effective Instrument During High Pollution Episode?

Vertical profiles of particulates were measured by a 355 nm Mie scattering lidar during a dust-storm event. A high energy pulse laser was employed as the light source to detect the extinction coefficient in the altitude up to 6 km in the day and 9 km at night. The extinction profiles showed layers of high aerosol concentrations in good agreement with ground-based pollution measurements, which indicated that such ultra-violet lidar is a very useful remote sensing instrument for monitoring extinction profiles during extreme high aerosol loading and low visibility atmospheric conditions when low energy lidar systems cannot obtain effective results.


INTRODUCTION
Air quality is always of great concerns. The particulate concentration keeps increasing in China over the past 20 years with the rapid economic development. Numerous ground stations make continuous round-the-clock point measurements of a range of pollutants, but they seldom provide the pollutants vertical profile which is important for modelling the transport of pollution. With good spatial and temporal resolution, real-time measurements can now be obtained from remote sensing technique such as lidar. Lidar has been widely employed to study aerosol spatial distribution [1] and explore the transport of aerosols over larger areas [2] . During episodes of dust event, lidars are important monitoring instruments [3,4] .
There is an internal relation between aerosol extinction and mass concentration. Without information on the particle size distribution and complex refractive index, the aerosol mass concentration can be retrieved from Mie lidar data using statistical correlations [5] . It is possible to extract aerosol mass concentrations within a reasonable accuracy of 20-30% without supplementary information on the aerosols [6,7] . With varying refractive index, Del Guasta [8] obtained aerosol mass concentrations 40 m above the ground from lidar measurements.
In this paper, rough proportionality between accumulation-mode aerosol mass concentration and lidar extinction is reported. The current work aims to infer vertical profiles of PM10 using a single channel lidar. Mass concentrations are inferred by establishing empirical correlations between PM10, visibility and extinction coefficient. In Section 2, the lidar instrumentation and data analysis method are described. Then, in Section 3, some results and discussions from dust case measurement are presented. And Section 4 is the conclusion.

METHODOLOGY
The lidar system used for the study, is a single wavelength Mie lidar located on top of the Taipa Grande in Macao at 112 m above sea level. The third harmonics of Nd:YAG laser is chosen as the emitter laser source with a pulse width of 5.7 ns at 50 Hz. The maximum pulse energy of the laser is 160 mJ, which is set to at most 50% of the maximum capacity in routine aerosol monitoring. The system layout is depicted in Figure 1. In the daytime, the sky background radiation in ultraviolet wavelength range is much weaker than that in the visible wavelength range. Under extremely high atmospheric extinction conditions this lidar can still obtain aerosol profiles up to 6 km in the day and 9 km at night. The Fernald's method [9] is used to retrieve the extinction coefficient. The molecular Rayleigh scattering values referred were taken from the US Standard Atmosphere (1976), the lidar ratio was assumed to be 50 sr for all data presented here, based on data from a Raman lidar operated in the Pear River Delta region near Hong Kong [10] . The boundary condition was the assumption that the lidar could transmit far enough and the far range aerosol extinction was zero. In general, aerosol concentrations are higher and vary sharply near the surface, therefore, capturing values of the near range is important for air quality studies. Near range correction of the lidar signal followed the methods proposed by A.Y.S Cheng et al. [11] and Liu et al. [12] . Figure 2 presents the PM10 and PM2.5 concentration in the dust episode in March, 2010. A significant increase of PM10 and PM2.5 concentration was found after March 21 st . Before that day, the PM10 concentration was much lower, and the average value was only 60.40g/m 3 . PM10 concentration had a weak growth on 21 st , however, at 18:00 on 21 st , it increased rapidly. At 5:30 on 22 nd it was up to the maximum value 616.83 g/m 3 , which was ten times more than the average value before. Then the high concentration maintained for a period of time and slowly decreased subsequently. Data showed the concentration decreased to 30.45 g/m 3 on 24 th . By comparing the two lines in figure 2, it can be found that during the dust period, particles of large size (2.5 ~ 10μm) are three times more than the small particles (2.5μm or less). The extinction coefficients inverted from Mie lidar in suit measurement are present in Fig.3. Each of these profiles is the average result in an hour, with the starting time on the top. It can be seen that on 22 nd , when the pollution is heavy, the particles congregation in the low altitude was evident. The extinction coefficients near the ground exceeded 0.7 km -1 ， which decreased rapidly from the maximum value about 0.9 km -1 at the low height to no larger than 0.3 km -1 at about 0.9 km. However, on 26 th , the moderate pollution day, the downtrend of extinction was not as sharp, and the near-surface extinction coefficients on 22 nd was larger than that on 26 th . Comparing the extinction coefficients with PM10 concentrations in heavy polluted episode with that of moderate polluted episode, we find an interesting phenomenon. That is, though the PM10 concentration has fallen from high pollution episode value (average value of 527.49µg/m 3 on the 22 nd ) to a normal level (average value of 97.25µg/m 3 on 26 th , which is less than one-fifth of the high pollution episode) before the dust-storm event, the near-surface extinction coefficients from lidar measurements didn't show such difference. Lidar measurements showed that the impact by dust particles had not yet been over on 26 th . There are two main reasons for explaining the disagreement between PM10 concentrations and extinction coefficients. On one hand, as is known, the variation of return signals caused by different particle size-distribution will impact the extinction coefficient retrieval from lidar measurement. In the heavy polluted episode, the bigger particles are the major component of aerosol. But they have a weaker influence on extinction in our back-scattering Mie lidar system, for their scattering energy is mainly concentrated in the forward. On the other hand, compared to the rapidly dropped bigger particles, the smaller particles can stay longer in the atmosphere (about one week). So extinction coefficients on 26 th are still at a high level.

RESULTS
Based on these two reasons, extinction coefficients at 355nm are not sufficient to show the impact of dust particles on atmosphere. Figure  4 shows the near-surface extinction coefficients at 1064nm converted from 355nm results [13] and PM10 concentration values at the same time. The correlation is helpful for inferring the vertical profiles of PM10 concentration in the dust episode and tracing the possible aerosol sources.

CONCLUSIONS
Aerosol extinction coefficients were measured by using the ultra-violet Mie lidar in a high pollution episode, which is the first study of aerosol extinction vertical profiles in a dust-storm event using lidar in Macao. The results show that the ultra-violet lidar has obtained good vertical profiles during the high pollution environment when other low energy lidar systems are ineffective.

ACKNOWLEDGEMENT
The work described in this paper is supported by the Shandong province excellent young and middle-aged scientists research award fund with