FLUORESCING AEROSOLS AND CLOUDS: INVESTIGATIONS OF CO-EXISTENCE

RAMSES of the Lindenberg Meteorological Observatory, Germany, is the ﬁrst multi-purpose lidar to routinely measure the ﬂuorescence spectra of atmospheric aerosols. Combined with the other measurement parameters (cloud water content and optical properties, moisture and temperature), this capability al-lows one to study the co-existence of clouds and ﬂuorescing aerosols for the ﬁrst time. The ﬂuorescence receiver is brieﬂy described, and measurement examples are presented and dis-cussed.


INTRODUCTION
Its scientific relevance notwithstanding, lidar studies of cloud-aerosol co-existence and, prospectively, interaction have not been attempted so far because conventional elasticbackscatter or Raman lidars cannot distinguish between scattering from different particles, and so the aerosol signal is masked in the presence of clouds. With its new fluorescence spectrometer, the spectroscopic Raman lidar RAMSES [1,2] is the first instrument to make such a distinction: Pure condensed water does not fluoresce while experience shows aerosols in the vast majority of cases do. So fluorescence observed within clouds unveils the occurrence of aerosol particles. In this contribution the integration of the fluorescence spectrometer into the RAMSES instrument is outlined. Furthermore, aerosol observations are discussed, particularly with regard to the interdependence between aerosol type, spectral characteristics and source region; and with regard to the coexistence of aerosols and clouds.

FLUORESCENCE SPECTROMETER
Similarly to the water spectrometer [2], the fluorescence spectrometer has been added to the RAMSES far-range receiver. Like the former, it consists of four main components, (1) the fiber coupler which images the atmospheric return signals transmitted by a dichroic beamsplitter used to suppress visible light in the vibrationalrotational Raman detection channels (BS7, see Fig. 4(b) of [1]) onto the circular front face (2.3mm diameter) of a, (2) quartz/quartz fiber bundle (100-µm cores, 0.12 numerical aperture). The fiber bundle, rectangularly reshaped, is coupled into a, (3) 1/8-m grating spectrograph (600 grooves per millimeter, 500-nm blaze). In its focal plane a, (4) 32-channel multianode photomultiplier single-photon-counting detection system is mounted (SP32-20, Licel GmbH, Berlin, Germany), yielding a spectral resolution of about 12.8 nm per detection channel. The wavelength range of the spectrometer covers the near UV and the visible light spectrum, however, because of the spectral characteristics of the above-mentioned beamsplitter, fluorescence spectra are only attainable > 440 nm. Fortunately, the transmitted signal of the 387nm Raman band of molecular nitrogen is sufficiently strong to be used for spectrum calibration. As a first step, relative calibration of the spectrum measurement is obtained by inserting a light source of known spectral characteristics into the telescope hall and illuminating the farrange receiver field stop. An absolute calibration is then achieved by comparing measured and calculated (from spectroscopic and concurrent radiosonde data) molecular-nitrogen Raman signals. Light extinction by particles and molecules is taken into account.

Saharan dust
As a first measurement example, the Saharan dust event in early April 2016 is presented in Figs. 1-3. RAMSES was operating continuously, but only nighttime measurements are discussed when data from the fluorescence spectrometer are available. Figure 1 depicts the advent of the dust plume which originated from Algeria and Mali according to simulations with the FLEXPART Lagrangian particle dispersion model [3]. Fluorescence is low, and, with a spectral maximum missing > 440 nm, the spectrum resembles those usually observed for boundary-layer aerosols under unperturbed conditions (see Fig. 7). Optical properties are homogeneous. Particle depolarization ratios, δ par , of about 30-35%, and lidar ratios, S par , between 60 and 80 sr (not shown) indicate par- ticle diameters of about 0.8 µm, according to finite-difference time-domain scattering computations [4] for quartz-rich irregular aerosol particles with an aspect ratio of 1. A shortlived change in particle properties to values typical of ice clouds (δ par = 15%, S par = 25 sr) is observed at the leading edge of the aerosol plume at 20:45 UTC (Fig. 1), but the elastic backscatter coefficient, β par , is rather small (∼ 3.5 Mm −1 sr −1 ), and the relative humidity with respect to ice, RHi, low (about 60%). Similar events occur at later times and slightly higher altitudes at much higher RHi values. Note that sporadic cirrus clouds were present above 9 km. The strong fluorescence below 1.5 km is indicative of a polluted boundary layer.
The next day, a mostly frozen cloud appeared in a moistened layer between 3.5 and 5.5 km. Interestingly, obscured by elastic cloud scattering otherwise, the fluorescence measurement reveals the presence of the weakly-fluorescing dust aerosol in the cloud-top layer (Fig. 2). Origin and spectral characteristics of the aerosol are very similar to what was found the day before. As an example of a combined measurement of elastic and fluorescence cloud and aerosol properties, Fig. 3 shows the RAMSES observation at 22 UTC. RHi hardly reaches ice saturation within the cloud, elastic-backscatter peak and fluorescence spectral backscatter coefficient maxima roughly coincide in height. Cloud optical properties are characteristic of horizontally aligned ice particles. Note that fluorescence capacity is several times larger below 3 km, which suggests differences in aerosol composition even though the fluorescence spectra (not shown) in the wavelength range currently accessible with RAMSES is similar.

Biomass-burning aerosol (BBA)
On 9 September 2016 aerosols from Africa were measured in the lower free troposphere (2-3.7 km; (δ par < 10%, S par = 60 sr)) again (Fig. 4), but the source region was slightly more to the southwest. Interestingly, the fluorescence properties differ vastly from those of the Apr 2016 cases: The aerosol fluoresces much more strongly (e.g., blue fluorescence capacity of about 5-7 Mm −1 nm −1 as compared to ≤ 1 Mm −1 nm −1 ), and the spectrum exhibits a distinct local maximum at 490-500 nm (Fig. 7). This may indicate the presence of BBA because fluorescence spectra of local Easter bonfires show a similar peak, albeit at longer wavelengths (Fig. 7). Relative humidity with respect to liquid water, RHw, is between 35% and 80%, β par correlates with RHw, but humidity is insufficient for cloud formation.
Hardly visible in Fig. 4, but yet confirmed by a Lindenberg in-situ optical backscatter sonde launched at 23 UTC, is another BBA layer at heights around 5 km. It is remnant of a larger BBA event that could be observed for several days starting on 6 September 2016 (Fig. 5). While β par is near the detection limit, the BBA shows a brilliant fluorescence display. The spectrum closely resembles the spectrum measured near the ground during the bonfires. The air masses of the vertically well-defined layer were advected from North America, very dry (< 20% RHi) and thus void of clouds. Incidentally, a thin cirrus was observed aloft (> 9 km) during the whole night.
Finally, Fig. 6 illustrates a BBA incident dur- ing which cirrus clouds occurred. The measurement was taken in the night of 7-8 September 2015, note the fine fluorescence streaks. A remarkable observation is that the descending cloud-top layer is strongly correlated with a BBA filament. Unfortunately, data quality is insufficient to determine RHi within the cirrus. In conclusion, the fluorescence spectrometer has proven to be a promising tool to study aerosol-cloud co-existence. The information extracted from the fluorescence spectra is sufficient to allow for a rough aerosol classifica- tion, which is an important result per se because backtrajectory calculations are often inconclusive and not meaningful enough to determine aerosol type. Experimental work is underway to extend measurements of the fluorescence spectrum to wavelengths < 440 nm.