In the atmosphere especially in the troposphere which is directly related with human activities, there are atmospheric aerosols that play a very important role in atmospheric processes, which will affect the Earth’s radiative balance directly and indirectly. Under direct aerosol effect, the aerosol will absorb and scatter Earth and Sun radiation. Indirectly, the aerosol will act as cloud condensation nuclei (CCN), which will then affect the concentration of initial droplets, albedo, precipitation formation and lifetime of the clouds [1-5]. Tropospheric aerosols are highly variable in time and space due to the non-uniform source distributions and the strong influence of meteorological conditions on the aerosol concentrations and characteristics . There are different types of aerosols which depend on their sources and regions. Aerosols originate from natural (sea salts, air-borne dust, volcanoes, and storms) and anthropogenic sources (fossil fuels combustion, biomass burning and gas-to-particle conversion) [1-5].
To determine the aerosol type in the atmosphere, we can do so by measuring the Lidar ratio. Lidar ratio is the ratio of the extinction coefficient to the backscatter coefficient. It depends on the microphysical properties of the aerosols, such as aerosol size, shape, refractive index, as well as environmental properties such as relative humidity and therefore can be used for the characterization of the aerosol type [6, 7]. Previously, analytical solutions to the Lidar equation are obtained using Klett's inversion [8, 9]. However, a critical assumption is associated with this method, where we must first assume the value of the Lidar ratio. A wrongly estimated Lidar ratio could lead to a large error of the inversion. An improvement to the Klett's inversion technique is to synergize the Lidar measurement with passive sensors, such as a sunphotometer  or radiometer . This method only permits the retrival of a single-valued, height independent Lidar ratio, which poorly resembles the atmosphere as there might be several layers existing in the atmosphere .However, for the Raman Lidar, which makes use of the inelastic return signal from atmospheric molecules (nitrogen molecules, water vapour and so on), the Lidar ratio can be measured independently [12, 13].
Recent studies such as summarized their Raman Lidar observations carried out in Europe, Asia and Africa during the past 10 years and obtained the mean values of the particle Lidar ratio for different aerosol types and source regions. Similar Raman Lidar observations were also carried out under highly polluted conditions in the Pearl River Delta (PRD) in southern China in October 2004 and at Beijing during a clear period with moderately polluted to background aerosol conditions in January 2005 .
In this study, range-dependent Lidar ratio is observed using backscatter-Raman Lidar during the dry seasons. In this way, we can characterize the aerosols suspended in the atmosphere in Penang Island during the dry seasons. We can also trace pollution outbreak as the Lidar ratio are very much dependent on aerosol types. By using HYSPLIT backtrajectory, we can trace the pollution outbreak back to its possible source, which provides us essential information on regional aerosol transportation.
An eye-safe ground based backscatter-Raman Lidar system, model no. LB100-ESS-D200, manufactured by Raymetrics SA was setup on the roof top of the School of Physics, Universiti Sains Malaysia, Penang. The location is shown in Figure 1. This backscatter-Raman Lidar system is capable of profiling aerosols and clouds by transmitting laser pulses into the atmosphere and measure the return time of the backscatter signal. It operates using a Nd:YAG pulsed laser that emits 355nm laser beam into the atmosphere. The Raman channel of this Lidar system measures the inelastic scattering of the Nitrogen molecules at 387nm. The pulse duration is 5.04ns, the energy emitted in each pulse is 33.4mJ with 20Hz repetition rate. The return signal, which contains the elastic backscattering signal at 355nm and inelastic backscattering signal at 387nm is then collected by a Cassegrain telescope with a diameter of 200mm, 1mrad field-of-view and a complete overlap height of 180m. The captured signal is then spectrally analyzed, filtered and separated by a wavelength separation unit before being focused on two different photon multiplier tube (PMT), with a corresponding spatial and temporal resolution of 7.5m and 1 minute respectively. Next, the current generated from the PMT will be detected by the Licel transcient recorder and the analogue Lidar signal is obtained and recorded. By using a combination of a powerful A/D converter (12 Bit at 40 MHz) with a 250MHz fast photon counting system, the photon counting Lidar signal is obtained and recorded.
The Raman Lidar data were collected at two different periods, corresponding to the dry seasons in Malaysia, one in March 2014, and another from June to August 2014. The Lidar system shot laser pulses at a zenith of 90° and collected 3 hours of Raman channel data starting from 1100 UTC (1900 local time) to 1400 UTC (2200 local time) everyday, except weekends and rainy days. The Lidar system was set to collect an average of 1200 shots into a data file, which means that one profile will be produced every minute.
Since the Raman signal is a weak signal, which is about 103 times weaker than the backscatter signal , the Raman data acquired everyday (180 profiles) were averaged into one profile using Lidar Analysis.exe, which is the processing software provided by Raymatrics SA. Before starting the data processing, the usage of analogue signal, photon counting signal or glued signal had to be determined, according to the conditions stated in . Then, the background radiation was subtracted from the Lidar signal. If the photon counting signal or the glued signal was used, dead-time correction had to be further applied to the photon counting signal as given by . Next, the range corrected signal (RCS) was obtained by applying the distance square law correction (z2) to each data point to compensate for range-related attenuation from the atmosphere. After that, the temporal evolution of the RCS was plotted. This plot showed the general view of the atmospheric conditions. Days with cloud overcast were identified as cloud contaminated and were removed, while the thickness of the aerosol layer was determined in the rest of the data. It was found that the aerosols are generally confined in the first 3km of the atmosphere.
After the pre-processing work, the Raman Lidar data processing was done with the aid of Lidar Analysis.exe, according to method suggested by . In order to obtain the Lidar ratio profile, the software calculated the aerosol extinction coefficient (αaer) and aerosol backscatter coefficient (βaer) independently. The aerosol extinction coefficient (αaer) is given by
where P(z) is the power received from distance Z at Raman wavelength λR if the laser pulse is transmitted at λL, N(z) is the atmospheric number density, αmol is the extinction coefficient due to absorption and Rayleigh scattering by atmospheric gases, and where particle scattering is assumed to be proportional to λ−k. For our Lidar system, λL is 355nm and λR is 387nm. kis assumed to be 1, as according to , it is safe to assumed k = 1 for aerosol particles with diameter comparable with the measurement wavelength.
Then, the aerosol backscatter coefficient (βaer) was determined by using signals from both the backscattered and the Raman channel. Two pairs of measured signal PλL and PλR at height z and at a reference height z0 were needed. Reference height z0 was taken at a height where there is almost no aerosol backscatter signal found and the backscatter signal is mostly molecular signal. Finally, the aerosol backscatter coefficient was given by
Where βmol is the molecular backscatter coefficient, PλR(z)and PλR(Z0) is the Raman signal collected at a height z and reference height z0 respectively. PλL(z) andPλL(z0)is the elastic backscatter signal collected at a height z and reference height z0respectively.
Finally, the Lidar ratio Saer was given by
Results and Discussions
Since the aerosols are mostly confined in the first 3km of the atmosphere, the Lidar ratio is taken from the height where overlap is complete up to 3km. Figure 2(a) and 2(b) show the layer-mean Lidar ratio for different height layers during March 2014 and June to August 2014, respectively. The Lidar ratios obtained ranged between 10 and 70 sr.  stated that aerosols with Lidar ratio around 60-70 sr are highly light-absorbing. These aerosols are most likely to be smoke aerosols produced from biomass burning, since biomass burning aerosol is very light absorbing. Next, Lidar ratios of 30-60 sr generally represents non-absorbing particles, such as ammonium or sulphate particles, which is commonly found in anthropogenic urban aerosols. Finally, Lidar ratios which are smaller than 30 sr are usually produced by clean, non-polluted marine particles.
Besides that, on Figure 2(a), we found that the Lidar ratios are generally higher before 10/3/14, where the Lidar ratios mostly lied around 40-60 sr, whilst after 10/3/14, the Lidar ratios mostly lied around 10-30 sr. This showed that in early March 2014, the atmosphere was more polluted than the middle and end of March 2014. This is mostly caused by smoke intrusion, either within Malaysia or at other neighbouring countries, which is later brought by the monsoon wind into Malaysia. 4 days HYSPLIT backtrajectories (not shown here) showed that in early March 2014, the air masses were blown from Indochina, Southern Thailand, North Malaysia and East Malaysia toward Penang Island. In contrast, the airmasses were mostly blown from The Philippine archipelagos and South China Sea towards Penang Island at the middle and end of March 2014.On the other hand, Figure 2(b) shows no significant difference in the Lidar ratios distribution pattern throughout the period of June to August 2014. The layer mean Lidar ratios are mostly concentrated in the region of 10-30 sr. This happened because the airmass flowing patterns are the same, as shown by the 4 days HYSPLIT backtrajectories (not shown here) which applied to study periods in June to August 2014. The back trajectories all showed that the airmasses are blown from Indian Ocean and Sumatera Island towards Penang Island, thus bringing the same type of aerosol into Penang Island, making the Lidar ratios about the same, unless smoke intrusion occurs, which will then significantly increase the Lidar ratio.
Figure 3(a) and 3(b) shows the time series of the mean Lidar ratio at 355 nm for the lowermost 1000 m in March 2014 and from June to August 2014, respectively. In March 2014, about half of the Lidar ratios lie above 30 sr. More specifically, in most of the observation days in March 2014, either layer (< 500m or 500-1000 m) would return a mean Lidar ratio above 30 sr. This suggested that during March 2014, the lower layer of the atmosphere (from ground level up to 1000 m in height) was polluted by non-marine particle, possibly urban aerosol, as the Lidar ratio for marine particles lies around 20-30 sr, according to . On the other hand, most of the Lidar ratios for the period of June to August 2014 lied below 30 sr. This showed that the period of June to August 2014 was generally less polluted then March 2014. These results also confirmed that marine background environment was present in Penang Island when there was no serious pollution issue.
In 3/3/14, a haze episode struck Penang Island. On this day, the Lidar system was operating throughout the night to acquire data about this haze event. Figure 4 shows the temporal evolution of the range-corrected signal (RCS) plot at 355nm on 3/3/14. The blue and purple colour on the plot indicates weak backscattering; the light blue, green, yellow and orange colour indicates strong backscattering from haze aerosol with increasing intensity; and the red colour shows clouds. The haze aerosols are mostly found around 1000 m in height, confined under the boundary layer, which is also about 1000 m in height. However, from 1600 UTC to 1900 UTC, there was a residual layer found above the boundary layer, at around 1500 m in height. Besides the temporal evolution of the RCS plot at 355nm, the Lidar ratio of the haze event was also calculated every 3 hours and the vertical profile of the aerosol Lidar ratio at 1200 UTC, 1500 UTC and 1800 UTC are shown in Figure 5(a), 5(b) and 5(c).
Based on Figure 5(a), we can see that there was a layer of aerosol with Lidar ratio ranging from 50-60 sr at height around 500-1000 m, while the rest of the Lidar ratio ranged around 10-30 sr. It can be deduced that the aerosol type in the aerosol layer at 500-1000m in height should be smoke aerosol from biomass burning sources and the rest of the aerosol should be marine aerosol. This agrees well with the result shown in Figure 4 from 1200 UTC to 1500 UTC, as there was strong backscattering from haze aerosols at the lowermost 1000 m, and weak backscattering from background marine aerosols at the other height. Then, by referring to Figure 5(b), 3 different aerosol layers were found. The aerosols from the lower layer, located in 500-1000 m in height, had Lidar ratio ranging from 25-30 sr, which was similar to the upper layer of aerosol, situated at 2500-4000 m. The middle layer, however, had a higher Lidar ratio, which ranged from 30-50 sr. This distribution pattern was totally different than that shown in Figure 5(a), which was only 3 hours of time apart. It was suspected that there might be a convection occurring around 1600 UTC, which was shown by the discontinuity at 1600 UTC in Figure 4. This convection would have brought the biomass burning aerosol to a height of 1500m, causing high Lidar ratio detected at that height. Lastly in Figure 5(c), there was only one part of the profile with high Lidar ratio, which was found at a height of 500 m with Lidar ratio ranging from 40-50 sr. This suggested that after the convection at 1600 UTC, the biomass burning smoke again settled down to the lower atmosphere, leaving the marine aerosols at higher altitude. However, Lidar ratio calculated starting from height 1000 m and above were very low, which is above 5-10 sr. Lidar ratio above 1500 m are even lower, which are only around 5 sr. It was suspected that this are cause by the low cloud cover starting around 1900 UTC, as shown in Figure 4. As laser light usually unable to penetrate through cloud, the signal return from these height does not give us correct information about the aerosols at that height. Hence, Lidar ratio at these heights will not be taken into account in our study.
Figure 6(a), 6(b), 6(c) shows 4 days HYSPLIT backtrajectories ending at 3/3/14 1200 UTC, 1500 UTC and 1800 UTC. All of the results from the model showed that the air masses were coming from the South China Sea. Hence, two hypotheses were made on the origin of the biomass burning aerosols. First, the biomass burning aerosol was brought to Penang Island shortly after it was produced in North Malaysia or Southern Thailand, as shown in Figure 6(a), 6(b) and 6(c), where the possible location to produce biomass burning aerosol would be North Malaysia and Southern Thailand. Second, it was also possible that the biomass burning aerosols had already been brought to Penang Island and deposited here in the morning, since the haze event had already started in the morning, around 8 a.m. local time, which corresponded to 0000 UTC. Figure 6(d) shows the HYSPLIT backtrajectories ending at 3/3/14 0000 UTC. It showed that airmasses reaching Penang Island at 1500 m and 2000 m on that particular time were actually coming from Sumatera Island, Indonesia. Hence, it might be possible that these are the aerosols detected by our Raman Lidar, because the aerosols would slowly decent in height when night is approaching. In order to find out which hypothesis is the actually happening here, we have to obtain the Ångström exponent, since it is directly related to the size of the particles. Large particles will have a smaller Ångström exponent . Since long ranged transported biomass burning aerosols would grow in size, according to , biomass burning aerosols from Sumatera Island should be bigger than freshly produced, which would have smaller Ångström exponent. However, this is not possible with our current equipment. To determine Ångström exponent, we must have at least 2 different laser wavelengths at the backscatter channel. Hence, no conclusion about this matter could be drawn here since we do not have enough data to support our hypotheses.
In this study, it was found that the Lidar ratio in Penang Island varied from 10-70 sr. Lidar ratio from 10-30 sr are Lidar ratio of marine aerosols, Lidar ratio from 40-70 sr are most probably the Lidar ratio of biomass burning aerosols, and the other represent Lidar ratio of polluted marine aerosols, where the pollutant should be local produced anthropogenic aerosol. Throughout the study periods, it was found that the atmosphere in Penang Island consisted of marine aerosol most of the time, unless smoke intrusion occurred, which brought in biomass burning aerosols. The results from this study can provide important information on regional aerosol transportation around Southeast Asia.These results are essential to setup Penang Island as a fixed observational site to monitor regional aerosol transportation and provide useful information in the study of regional aerosol climatology. Future studies should involve a synergetic measurement of multiwavelength backscatter Raman Lidar, as well as a particulate meter to provide more data and can more accurately determine the aerosol type and aerosol transportation in Penang Island, Malaysia.
Balis D, Papayannis A, Galani E, Marenco F, Santacesaria V, Hamonou E, Chazette P, Ziomas I and Zerefos C, “Tropospheric LIDAR aerosol measurements and sun photometric observations at Thessaloniki, Greece,” Atmos. Environ. 34, 925 - 932 (2000).Google Scholar
Georgoussis G, Chourdakis G, Landulfo E, Hondidiadis K and Ikonomou A, “Monitoring of air pollution and atmospheric parameters using a mobile backscatter lidar system,” ÓPTICA PURA Y APLICADA 39(1), 35 - 41 (2006).Google Scholar
Landulfo E, Papayannis A, Artaxo P, Castanho A D A, Freitas A Z d, Souza R F, Junior N D V, Jorge M P M P, Sanchez-Ccoyllo O R and Moreira D S, “Synergetic measurements of aerosols over Sao Paulo, Brazil using LIDAR, sunphotometer and satellite data during the dry season,” Atmos. Chem. Phys. 3, 1523 - 1539 (2003).Google Scholar
Chen W-N, Chen Y-W, Chou C C K, Chang S-Y, Lin P-H and Chen J-P, “Columnar optical properties of tropospheric aerosol by combined lidar and sunphotometer measurements at Taipei, Taiwan,” Atmos. Environ. 43, 2700 – 2708 (2009).Google Scholar
Royer P, Chazette P, Lardier M and Sauvage L, “Aerosol content survey by mini N2-Raman lidar: Application to local and long-range transport aerosols,” Atmos. Environ. 45, 7487 - 7495 (2011).Google Scholar
Amiridis V, Balis D, Giannakaki E, Kazadzis S, Arola A and Gerasopoulos E, “Characterization of the aerosol type using simultaneous measurements of the lidar ratio and estimations of the single scattering albedo,” Atmos. Res. 101, 46–53 (2011).Google Scholar
Noh Y M, Kim Y J, Choi B C and Murayama T, “Aerosol lidar ratio characteristics measured by a multi-wavelength Raman lidar system at Anmyeon Island, Korea,” Atmos. Res. 86, 76–87 (2007).Google Scholar
Klett J D, “Stable analytical inversion solution for processing lidar returns,” Appl. Optics 20(2), 211–220 (1981).Google Scholar
Klett J D, “Lidar inversion with variable backscatter/extinction ratios,” Appl. Optics 24(11), 1638 - 1643 (1985).Google Scholar
Chazette P, “The monsoon aerosol extinction properties at Goa during INDOEX as measured with lidar,” J. Geophys. Res. 108(D6), 4187 (2003).Google Scholar
Royer P, Raut J-C, Ajello G, Berthier S and Chazette P, “Synergy between CALIOP and MODIS instruments for aerosol monitoring: application to the Po Valley,” Atmos. Meas. Tech. 3, 893–907 (2010).Google Scholar
Ansmann A, Riebesell M, Wandinger U, Weitkamp C, Voss E, Lahmann W and Michaelis W, “Combined Raman Elastic-Backscatter LIDAR for Vertical Profiling of Moisture, Aerosol Extinction, Backscatter, and LIDAR Ratio,” Appl. Phys. B 55, 18–28 (1992).Google Scholar
Ansmann A, Wandinger U, Riebesell M, Weitkamp C and Michaelis W, “Independent measurement of extinction and backscatter profiles in cirrus clouds by using a combined Raman elastic-backscatter lidar,” Appl. Optics 31(33), 7113 - 7131 (1992).Google Scholar
Muller D, Ansmann A, Mattis I, Tesche M, Wandinger U, Althausen D and Pisani G, “Aerosol-type-dependent lidar ratios observed with Raman lidar,” J. Geophys. Res. 112(D16), 202 (2007).Google Scholar
Tesche M, Ansmann A, Müller D, Althausen D, Engelmann R, Hu M and Zhang Y, “Particle backscatter, extinction, and lidar ratio profiling with Raman lidar in south and north China,” Appl. Optics 46(25), 6302 - 6308 (2007).Google Scholar
Tan F, Beh B C, Tan C H, Lim H S, Abdullah K, Jafri M Z M, Welton E J and Lolli S, “Preliminary analysis of ground based Lidar backscattered signal and performance evaluation in Penang Island,” AIP Conf. Proc. 1528, 140–145 (2012).Google Scholar
Donovan D P, Whiteway J A and Carswell A I, “Correction for nonlinear photon-counting effects in lidar systems,” Appl. Optics 32(33), 6742 - 6753 (1993).Google Scholar
Angstrom A, “The parameters of atmospheric turbidity,” Tellus 16(1), 64 - 75 (1963).Google Scholar