Near-surface and columnar measurements with a micro pulse lidar of atmospheric pollen in Barcelona, Spain

We present for the first time continuous hourly measurements of pollen near-surface concentration and lidarderived profiles of particle backscatter coefficients and of volume and particle depolarization ratios during a 5-day pollination event observed in Barcelona, Spain, between 27 and 31 March 2015. Daily average concentrations ranged from 1082 to 2830 pollen m−3. Platanus and Pinus pollen types represented together more than 80 % of the total pollen. Maximum hourly pollen concentrations of 4700 and 1200 m−3 were found for Platanus and Pinus, respectively. Every day a clear diurnal cycle caused by the vertical transport of the airborne pollen was visible on the lidar-derived profiles with maxima usually reached between 12:00 and 15:00 UT. A method based on the lidar polarization capabilities was used to retrieve the contribution of the pollen to the total aerosol optical depth (AOD). On average the diurnal (09:00– 17:00 UT) pollen AOD was 0.05, which represented 29 % of the total AOD. Maximum values of the pollen AOD and its contribution to the total AOD reached 0.12 and 78 %, respectively. The diurnal means of the volume and particle depolarization ratios in the pollen plume were 0.08 and 0.14, with hourly maxima of 0.18 and 0.33, respectively. The diurnal mean of the height of the pollen plume was found at 1.24 km with maxima varying in the range of 1.47–1.78 km. A correlation study is performed (1) between the depolarization ratios and the pollen near-surface concentration to evaluate the ability of the former parameter to monitor pollen release and (2) between the depolarization ratios as well as pollen AOD and surface downward solar fluxes, which cause the atmospheric turbulences responsible for the particle vertical motion, to examine the dependency of the depolarization ratios and the pollen AOD upon solar fluxes. For the volume depolarization ratio the first correlation study yields to correlation coefficients ranging 0.00–0.81 and the second to correlation coefficients ranging 0.49–0.86.

PM10 measurements were acquired at the "Eixample" station of the Xarxa de Vigilància i Previsió de la Qualitat de l'Aire (XVPCA, the Catalonian network for monitoring and forecasting the air quality). It is located at 1.2 km to the southwest of the pollen sampling instrumentation.
Meteorological data were recorded in the "Zona Universitaria" area of Barcelona, at approximately 0.6 km south-southeast of the lidar site. network. The MPL system is a compact, eye-safe lidar designed for full-time unattended operation (Spinhirne, 1993;Campbell et al., 2002;Flynn et al., 2007). It uses a pulsed solid-state laser emitting low laser pulse energy (~6 μJ) at a high pulse rate (2500 Hz) and a co-axial "transceiver" design with a telescope shared by both 15 transmit and receive optics. The Barcelona MPL optical layout uses an actively controlled liquid crystal retarder which makes the system capable to conduct polarization-sensitive measurements by alternating between two retardation states (Flynn et al., 2007). The signals acquired in each of these states are recorded separately and called "co-polar" and "cross-polar". In nominal operation the raw temporal and vertical resolutions are 30 s and 15 m, respectively.

20
The linear volume depolarization ratio, V  , is defined as: In the case of a linear polarization lidar, d = 0 indicates that no depolarization occurs, while d = 1 indicates that the returned beam is completely depolarized. By adapting the notations of Flynn et al. (2007), especially in Eqs (1.4) and (1.6), to ours one can formulate the linear volume depolarization ratio for the MPL system as:

30
( ) ( ) ( ) ( ) The linear particle depolarization ratio, p  , can then be determined by (Freudenthaler et al., 2009): where β m and β p denote the molecular and particle backscatter coefficient, respectively, of the total (perpendicular + parallel) returned signal. According to the MPL optical requirements in the receiving system the spectral filtering is performed by filters with a spectral band ≤ 0.2 nm. This number produces a temperatureindependent molecular depolarization ratio of 0.00363 m   according to Behrendt and Nakamura (2002).
The particle backscatter coefficient, β p , was retrieved with the two-component elastic algorithm (also known as 10 the Klett-Fernald-Sasano method;Fernald, 1984;Sasano and Nakane, 1984;Klett, 1985) with a constant lidar ratio of 50 sr and applied to the total lidar signal, P, reconstructed from the MPL lidar signals as (Flynn et al., 2007): ( ) ( ) 2 ( ) co cr P z P z P z   The value of 50 sr is motivated by two previous studies. First, it falls in the range of the mean columnar lidar 15 ratios, 46 -69 sr, found in Barcelona during the period from February to April and calculated over a period of 3 years (Sicard et al., 2011). In that work the columnar lidar ratio was retrieved with the two-component elastic lidar inversion algorithm constrained with the aerosol optical depth from a sun-photometer (Landulfo et al., 2003;Reba et al., 2010). Second, Noh et al. (2013b) used the same method and found a mean columnar lidar ratio of 50 ± 6 sr during a 6-day pollination event (mostly dominated by Pinus and Quercus pollen) in South where no pol   and pol  are the particle depolarization ratios of all particle types except pollen (nondepolarizing) and the pollen (

15
Section 4, the pollen plume is characterized during the whole pollination event by a near-constant or slightly decreasing profile of pol  . From this aspect the structure of the pollen plume is much simpler than the ABL structure usually found in Barcelona (Sicard et al., 2006). This allows us to use a simple threshold method such as the one used to estimate the ABL height by Melfi et al. (1985) and Boers et al. (1988). After several tests we empirically set a threshold of 0.055 Mm -1 sr -1 and defined pol h as the first height at which 20 in Figure 1a for a period surrounding the peak of the pollination event under study. Figure 1b shows the fraction of each one of the four most abundant taxa to the total (pollen + spore
The temporal evolution of the profiles of the particle backscatter coefficient and the volume depolarization ratio during the pollination event is shown in Figure 2. Aerosols are present everyday up to 2.5 -3 km. However most of the aerosol load is found below approximately 1.5 km. Near the ground (< 0.5 km) high values of β p (4 Mm -1 sr -1 ) are found on almost all days. Between 0.5 and 1.5 km the green color code indicates values of β p 25 not higher than 2 -2.5 Mm -1 sr -1 (except on 26 March on which day clouds are present below 2 km before 08 UT). In general two regimes are observed everyday: an increase in amplitude and height starting around 10 UT which persists until the night, and a less pronounced nighttime regime starting usually after midnight. On 31 March one sees a layer appearing after 11 UT with very large values of β p (> 5 Mm -1 sr -1 ) and confined in the first 0.5 km of the ABL. This increase of β p in the bottom part of the ABL has no impact on the volume 30 depolarization ratio vertical distribution ( Figure 2b) which suggests that it is due to non-depolarizing particles.
The green color code volume depolarization ratio shown in Figure 2b indicates values of δ V near 0.02 -0.03.
It is the usual value of δ V for background, local aerosols near the surface in Barcelona. Everyday around 08 UT a plume with δ V > 0.08 (yellowish) appears, raises up to 1.0 -1.7 km in a few hours and starts decreasing before 16 UT at a lesser rate than it raised. This diurnal pattern of δ V is observed on each single day of the

5
-31 March a counting on an hourly basis. The method used is described in Section 2.1. Although all pollen and spore taxa were counted, in the following we will only show the results of the total pollen (spore is no longer taken into account) and of the two most abundant pollen types: Platanus and Pinus. The two main reasons for that choice are that 1), as found earlier, Platanus and Pinus pollen represent more than 80 % of the total (pollen + spore) taxa and 2) the ratio of total spore to total pollen is less than 13 % during the period 27 -10 31 March.
Many works have investigated the influence of the meteorological conditions, such as relative humidity, temperature, wind speed, the number of sunshine hours and rainfall, on the release and transport of pollen in the atmosphere (Raynor et al., 1973, Mandrioli et al., 1984Hart et al., 1994;Alba et al., 2000;Jato et al., 2000;Bartková-Šcevková, 2003;Vázquez et al., 2003;Latorre and Caccavari, 2009, among others). On the one 15 hand, relative humidity and temperature greatly affect the release of pollen in the atmosphere by influencing the extent to which individual pollen grains dehydrate. For example a low relative humidity associated with a high temperature will tend to increase the number of airborne pollen grains by decreasing their specific gravity.
The relation of pollen with water comes from its hydrophilic properties and from the fact that it is prone to harmomegathic movement (accommodation of volume change when it absorbs water; Wodehouse, 1935). The 20 duration of the sunshine has also proved to have an influence on the pollen release (Alba et al., 2000). On the other hand, wind speed plays a major role in the transport and dispersion of airborne pollen: high daytime wind speed may facilitate the dispersion of airborne pollen in the atmosphere (Latorre and Caccavari, 2009;and references therein). The effect of rainfall is to reduce the number of airborne pollen grains by washing out the atmosphere. During the pollination event presented in this work, no rain was detected.

25
In Figure 3 we present the hourly temporal variations of 1) Platanus, Pinus and total pollen concentration during the period 27 -31 March, together with 2) relative humidity (RH), 3) temperature (T) and 4) wind speed. In

Pollen near surface vs. columnar properties: day-by-day analysis
The daily temporal variation of some lidar-derived range-resolved and columnar parameters are investigated 30 and further compared to pollen concentrations in order to find possible correlations. Figure 4 shows the diurnal  the lowest relative humidity and the highest temperature observed during the whole pollination event (see Figure   3b). We have checked that the relative humidity and temperature profiles (not shown) measured daily by radiosoundings launched close to the lidar site were, respectively, the lowest (RH < 40 % up to 1 km) and the highest (T > 15ºC up to 1 km) on 30 March at 12 UT. It also follows a strong peak of pollen release at ground

15
We now move on to the analysis of the lidar-derived columnar parameters and their possible correlations with the near-surface pollen concentration. The daily mean AOD which varies in the range 0.14 -0.24 (Table 1)  from the ABL. The fact that it is associated to a strong increase of the AOD but to any increase of the nearsurface PM10 level suggests that the new non-depolarizing aerosol plume seen in Figure 2 with high values of β p (> 5 Mm -1 sr -1 ) and confined in the first 0.5 km is not from local origin.
We have calculated the correlation coefficients between the near-surface pollen concentration (total, Platanus and Pinus) and the columnar properties discussed above in this Section (Table 2,  In the concern to find a possible proxy of the pollen (be it total, Platanus or Pinus) near-surface concentration with some columnar parameter easily measurable by remote sensing instrument, we examine the correlation between pollen concentration and AOD. The AOD is a columnar parameter which is relatively easily measurable, e.g. with a sun-photometer, and which can be retrieved with a relatively high accuracy. For   Atmos. Chem. Phys. Discuss., doi:10.5194/acp-2016-212, 2016 Manuscript under review for journal Atmos. o V  seems to be an appropriate proxy for the total pollen concentration.
It is important to recall that these conclusions have to be taken in a general sense, and that, depending on the meteorological conditions, there may be cases for which these statements do not apply. In the following Section we seek the possible reasons which could explain the aforementioned correlations between pollen near-surface concentration and columnar depolarization ratios.

Influence of the solar radiation on the pollen vertical transport in the atmosphere
For the pollen to be dispersed in the atmosphere, vertical transport is needed. According to Mandrioli et al. (1984) the main mechanism driving the vertical movement of atmospheric particles is the atmospheric turbulence. In the ABL atmospheric turbulences result from the vertical movement of air masses due to the heating and cooling of the ground by the sun and to the flow of air (wind) over the ground. At the end of Section  March. In the first row of Figure 8 we represent all together the solar fluxes, V  and p  as a function of time.
Everyday a diurnal pattern is clearly visible on all curves with an increase in the morning and a decrease in the 30 afternoon. On 27 March the temporal evolution of V  and p  seems to follow especially well the pattern of the solar flux. In all three cases a time delay is observed between the diurnal evolution of the solar fluxes and the depolarization ratios. As one could intuitively expect, the pollen vertical transport pattern, triggered by the turbulences caused by the heating/cooling of the ground, should follow with a given time delay (i.e. start after) Atmos. Chem. Phys. Discuss., doi:10.5194/acp-2016-212, 2016 Manuscript under review for journal Atmos. Chem. Phys. that of the solar flux. In order to quantify that time delay, t, for each day and for each of the two depolarization ratios, we have searched the value of t that maximizes the correlation coefficient defined as: where F is the solar flux and δ either V  or p  . The optimized value of t that maximizes the correlation coefficient is called topt. The second and third rows of Figure 8 present the scatter plots of the solar flux vs. V  5 and p  , respectively. We have represented the scatter plots without time delay (red colour, r(t=0)) and the scatter plots with a time delay equal to topt (blue colour, r(t=topt)). In the first row of Figure

Conclusion
For the first time near-surface and columnar measurements of airborne pollen have been performed continuously at a temporal resolution of one hour during a 5-day pollination event in a large European city. At the peak of the event 2830 pollen and fungal spore grains were counted per cubic meter per day. Platanus and

25
Pinus pollen types represented together more than 80 % of the total concentration. Hourly concentration maxima of 4700 and 1200 m -3 h -1 were found for Platanus and Pinus, respectively. Except on one day, the total pollen concentration at ground level followed a clear diurnal cycle and was correlated positively with Atmos. Chem. Phys. Discuss., doi:10.5194/acp-2016-212, 2016 Manuscript under review for journal Atmos. temperature (r = 0.95) and wind speed (r = 0.82) but negatively with relative humidity (r = -0.18). These results indicate a strong dependence of pollen release upon the meteorological conditions, especially temperature and wind speed. As far as pollen AOD is concerned, its peaks were systematically associated with minima of relative humidity and maxima of temperature but they did not present a systematic dependence upon wind speed.

5
The pollen AOD showed a clear diurnal cycle with maxima between 12 and 15 UT. The diurnal (9 -17 UT) mean of AODpol was 0.05 over the whole event and represented 29 % of the total AOD. However peaks of AODpol and AODpol / AOD of, respectively, 0.12 and 78 % were found on the hourly data. The diurnal mean volume and particle depolarization ratios in the pollen plume were 0.08 and 0.14, with hourly maxima of 0.18 and 0.33, respectively. The diurnal height of the pollen plume was found at 1.24 km on average with maxima 10 varying in the range 1.47 -1.78 km.