Fluorescent biological aerosol particle measurements at a tropicalhigh-altitude site in southern India during the southwestmonsoon season

Abstract. An ultraviolet aerodynamic particle sizer (UV-APS) was continuously operated for the first time during two seasons to sample the contrasting winds during monsoon and winter to characterize the properties of fluorescent biological aerosol particles (FBAPs), at a high-altitude site in India. Averaged over the entire monsoon campaign (1 June–21 August 2014), the arithmetic mean number and mass concentrations of coarse-mode (> 1 µm) FBAPs were 0.02 cm−3 and 0.24 µg m−3, respectively, which corresponded to  ∼  2 and 6 % of total aerosol loading, respectively. Average FBAP number size distribution exhibited a peak at  ∼  3 µm, which is attributed to the fungal spores, as supported by scanning electron microscope (SEM) images, and these results are consistent with previous studies made for FBAPs. During 11 weeks of measurements the variability of the total coarse-mode particle number (TAP) concentration was high compared to that observed in FBAP number concentrations. The TAP and FBAP number concentrations measured at this site were strongly dependent on changes in wind direction and rainfall. During periods of westerly/southwesterly winds with heavy persistent rainfall, the TAP and FBAP concentrations exhibited very low values (1.3 and 0.005 cm−3, respectively) with no significant diurnal variations, whereas during periods of northerly winds with scattered rainfall FBAPs exhibited relatively high concentration values (0.05 cm−3) with pronounced diurnal variations, which were strongly coupled with diurnal variations in meteorological parameters. The campaign-averaged FBAP number concentrations were shown to correlate with daily patterns of meteorological parameters and were positively correlated with relative humidity (RH; R2  =  0.58) and negatively with temperature (R2  =  0.60) and wind speed (R2  =  0.60). We did not observe any significant positive correlation with precipitation as reported by previous researchers from selected areas. These measurement results confirm the fact that the ratio of PBAPs to TAP is strongly dependent on particle size and location and thus may constitute a significant proportion of total aerosol particles.


Introduction
terminology. The Western Ghats, one of the eight mountain ranges in India and identified as one 150 of the hottest hot spots of biodiversity (Myers et al., 2000) in the world, originates near the   between two He-Ne red lasers (=633 nm). Once the particle size is determined, the same 188 particle is further excited using a third ultraviolet Nd:YAG laser (=355 nm) and emissions are 189 measured in the range of 420 -575 nm. The spectrally unresolved total fluorescence is recorded 190 for each individual particle in to one of the 64 channels with increasing order of fluorescence 191 intensity. Huffman et al., (2010) described that the counting efficiency of the instrument drops 192 below 100% at D a < 0.7 m (counting efficiency ~50% at 0.54 m), hence, the particle number  The UV-APS was placed next to the window inside a room in the College of Engineering, 204 Munnar, Kerala located on a hill. A stain-less steel tubing with ¾" OD (outer diameter) and TSP 205 inlet was used to construct the inlet unit for air sampling, which was ~9 m and ~2 m above the 206 ground and rooftop, respectively. Thus the sampled air masses were expected to have minimal 207 influence caused by the dynamics associated with the building structure. To minimize the 208 particle losses due to impaction resulting from sharp bends, the electrically conductive silicon 209 rubber tubing (~1.5 m; 12 mm inner diameter) was attached to the stain-less steel tube just  For the present study we derived number size distribution of fluorescence biological aerosol 217 particles, dN F /dlog D a , for each size bin by summing up the particle number concentration from the fluorescence channel numbers 3 -64 and similarly the total particle number size distribution, 219 dN T /dlog D a , was derived from channel numbers 1 -64. In the present study we have used 1.0 220 m as a cut-off diameter for given dN F /dlog D a and dN T /dlog D a to calculate the fluorescence 221 biological aerosol number and total aerosol number concentrations, N F and N T , respectively. This 222 is mainly due to the fact that particle counting efficiency of the UV-APS drops below unity at 0.7 223 m and the interferences due to fluorescence from non-biological aerosol particles below 1.0 m 224 can at times be very high (Huffman et al., 2010). Also note that the cutoff at 1 m moreover 225 represents the border between fine (<1m) and coarse (>1m) modes of the particle number size 226 distribution. The subscripts throughout this manuscript text "F" and "T" refer to fluorescent and 227 total coarse mode particles, respectively. Please refer to Table 1 for the abbreviations, notations, 228 and symbols used in this manuscript. The particle mass size distributions (dM/dlogD a ) for total as 229 well as fluorescent biological aerosol particles were calculated for each size bin by multiplying 230 dN/dlogD a with volume of an aerodynamically equivalent sphere with the geometric midpoint 231 diameter (D a,g ) and assuming the unit density (1 g cm -3 ) and unit shape factor. The integral mass 232 concentrations of coarse fluorescent biological aerosol particles and total coarse particles, M F and 233 M T , respectively were calculated by integrating the particle mass distribution for D a >1m; but 234 should be viewed as first approximation as a result of uncertainty associated with the density and 235 shape of the particles (Huffman et al., 2010).

236
Fluorescence of submicron particles 237 It has been reported by previous researchers that UV-APS is known to exhibit fluorescence for 238 some fraction of non-biological aerosol particles including soot, PAHs, and cigarettes smoke, 239 which could be erroneously counted as FBAP (Huffman et al., 2010;Pan et al., 1999aPan et al., , 1999b. It 240 has also been emphasized that such interference can mostly occur for particles less than 1 m as  From these analyses we infer that the contribution of non-biological aerosol particles exhibiting 266 fluorescence was negligible in both submicron and supermicron (except during "dusty period"; 267 discussed below) size ranges. Thus we hypothesize that due to persistent rainfall the submicron  The meteorological parameters in parallel with the UV-APS measurements were recorded during 279 the entire campaign using an ultrasonic weather sensor (Lufft WS600-UMB) installed on a 280 rooftop at the same height and a few meters away from the UV-APS inlet (Fig. S1). The weather 281 station was capable of recording temperature, dew point temperature, relative humidity, 282 precipitation intensity, wind speed, wind direction, and air pressure and was set to record these 283 meteorological parameters with every 5 minutes interval with time synchronized to UV-APS 284 measurement clock. The data from the weather sensor was stored by using an in-house 285 developed external data logger. The obtained meteorological data was compared with another    Overall the meteorological conditions during the campaign at Munnar can be summarized as 310 follows: The predominant wind direction was observed to be Westerly/Southwesterly (Fig. 1

326
The total of more than five months of bioaerosol measurements in high time and size resolution 327 were performed at this site comprising two contrasting seasons, monsoon (dominated by 328 Southwesterly winds) and winter (dominated by Northeasterly winds). In this study we present 329 the results from the field campaign carried out during the SW monsoon season whereas the 330 detailed results from the winter campaign from the same measurement site will be presented in 331 the follow up study. We first discuss the characteristic features of the time series as a broad

Statistical distribution of mass concentration 444
Basically UV-APS measures the particle number; the average mass of size-resolved particles can 445 be derived as first approximation by assuming the particle density equal to 1 g cm -3 (unit particle mass was ~3 times higher (~6%) than its contribution to coarse mode particle number corresponding to N F . However, the prominent peak in dM F /dlog D a was observed at higher 477 diameter (~3 -4 µm), which is due to the fact that dM F /dlog D a has been derived from dN F /dlog 478 D a assuming unit density. As observed for N F during the month of June, the consistent morning    The size-resolved ratio of FBAP to TAP averaged over the course of measurement is shown in showed that the first peak in dN F /dN T was more or less consistent at ~22% during June and 578 August except for the July when second peak in N F /N T ratios contributed more (~12%) than the 579 first peak (~10%).  here did not represent equal duration of the observations. The average total particle number 617 concentration, N T , showed a decrease of ~70% from dusty period to clean period (~4.2 cm -3 and 618 ~1.3 cm -3 respectively), whereas the N T concentration during high bio period was ~1.8 cm -3 . The 619 high N T concentration during the dusty period caused the high variability between 5 th and 95 th 620 percentile in N T when averaged over entire campaign period (Fig. 3a). The fraction of dust in 621 coarse mode aerosol, which is observed to be very high during pre-monsoon and first few days 622 from the onset of monsoon rainfall, gradually decreased as the monsoon progressed as a result of 623 wash out and wet deposition due to persistent rainfall in the path of air masses (Hirst 1953;  As expected, the N F was highest during the high bio period (Fig. 9b)   As anticipated the relative contribution of FBAP in TAP during dusty and clean periods was 646 almost negligible with N F / N T ratio of ~1%. Whereas during the high bio period the relative 647 FBAP number and mass contribution to corresponding TAP was ~5% and 12% respectively.  The FBAP mass size distribution (Fig. 11) during dusty period was dominated by bimodal peaks 675 with prominent peak at ~3 m and relatively less pronounced peak in the range of ~4 -6 m 676 showing broader tail on the right side of the distribution curve. The dM F /dlog D a , during clean 677 period, exhibited similar bimodal peaks with extended shoulder in the diameter range from ~4 to 678 7 m. The dM F /dlog D a distribution during high bio period was distinct compared to two other 679 focus periods discussed above with a prominent monomodal peak at ~3 m. The primary peak 680 observed in dM F /dlog D a in the range of ~3 to 4 m was consistent during individual months and 681 different focus periods. TAP mass size distribution (Fig. S15) exhibited similar qualitative 682 pattern to that of campaign averaged dM T /dlog D a with peak between ~2.5 to 3.5 m with an 683 extended tail on the right side, which gradually increased for D a >13 m.

684
The size resolved ratio of FBAP to TAP particles averaged for three distinct focus periods is

Diurnal patterns 704
A prominent early morning peak in N F during high bio period in the diameter range of 1.5 -3 705 m was observed from 06:00 hr to 08:00 hr, which clearly reflected in campaign averaged 706 diurnal patterns at the same hour of the day. The diurnal variations in N F during dusty and clean 707 periods were not so pronounced (Fig. 13) as compared to the variations during high bio period.

708
During dusty period N F showed slightly high concentration starting from ~20:00 hrs and 709 persistently remained high until early morning without any variations, whereas during clean 710 period N F concentration consistently remained flat throughout 24 hrs. The evening peak 711 observed during dusty period, however, was clearly absent during high bio period. A moderately 712 pronounced peak in N F during evening hours at ~20:00 hr during dusty periods might indicate 713 that releasing mechanism of bioaerosols was efficient as a result of nocturnal sporulation. This   Figure 14 shows the exemplary SEM images of biological particle types often observed during 737 the SW monsoon season at Munnar. The details about the sampling techniques, instrument used, 738 etc. for obtaining these bioaerosol images are discussed in details by Valsan et al., (2015). Note 739 that these images are not being presented here for any quantitative purpose and to draw any 740 specific scientific conclusions but mainly to showcase the particle types consistently observed   17. The FBAP number concentration >0.1 cm -3 was associated with calm (0 -1 m s -1 ; ~80% 806 frequency of occurrence) and predominant Northerly winds (Fig. 17a) as opposed to high wind 807 speed (2 -5 m s -1 ) and predominant Westerly/Southwesterly winds for the FBAP number 808 concentration <0.1 cm -3 (Fig. 17b). The calm northerly winds coming over from densely of Basidiosppres. However, due to technical difficulties associated with sampling we could not 823 establish the fact that spores observed at this observational site during Westerly/Southwesterly 824 winds were dominated by Ascospores and these details will be addressed in follow up studies.

825
The corresponding wind rose scaled by D g,T obtained from dN T /dlog D a is shown in Fig. S21.

826
As shown in Tab. 5 the wind speed was observed to be negatively affecting the N F during entire   The average observed dN F /dlog D a exhibited a peak at ~3 μm, which was consistent even during  (Schumacher et al., 2013). In the pine forest of Colorado the 950 distinct peaks were observed at ~1.5 μm and ~5 μm (Schumacher et al., 2013). The mode at ~3 951 μm is likely due to the fungal spore whose release mechanism is strongly governed by the scale, however, a significant correlation between N F and meteorological parameters was 971 observed. We observed that N F followed the similar diurnal trend to that of relative humidity and between N F and wind speed was found to be strongly negative. Since majority of the spore          time) illustrating the distinct and contrasting wind patterns during two contrasting seasons; Southwest monsoon season (representative month Jul) and Northeast monsoon season (representative month Jan) when field measurement campaign was carried out. It is evident that predominant wind pattern during Southwest monsoon season was Westerly/Southwesterly bringing the clean marine influx. Also shown in inset is wind rose diagram prepared using the data obtained using the ultrasonic weather station (a). The inlet system prepared for sampling the air using Ultraviolet Aerodynamic Particle Sizer (UV-APS) for bioaerosol number size distribution measurement. Inset shows the arrangement made for installing the ultrasonic weather station (b). The map shown is color-coded by topography (meters).