Observation of non-spherical ultragiant aerosol using a microwave radar



[1] Observations of ultragiant aerosol particles performed at the CNR-IMAA Atmospheric Observatory using a Ka-Band Doppler radar in four different periods from 19 April to 13 May 2010 are presented. In the reported cases, the aerosol radar signatures are characterized by a similar scenario. In particular, the linear depolarization ratio shows values higher than −4 dB probably related to the effect of bulk density and to the non-sphericity of the ultragiant particles. During the same period, volcanic aerosol layers coming from Eyjafjallajökull volcano were observed over most of European countries, including Southern Italy, using lidar technique. The observation of volcanic layers over Potenza by multi-wavelength Raman lidar measurements suggests a volcanic origin of the ultragiant aerosol particles observed by the radar, revealing that these particles might travel in the atmosphere for more than 4000 km after their injection in the atmosphere.

1. Introduction

[2] Observations of large aerosol particles injected in the atmosphere by explosive volcanic eruptions are already reported in the literature related to paleo-climatological and volcanological studies. For example, Lacasse [2001] reports a list of evidences of long range transport of Icelandic tephra detected by coring in lakes, marine sediments and glacier ice. Long range transport of ultragiant particles of basic or silicic origin (tephra) with diameter ranging within 30 μm–10 mm and over distances up to 2200 km from the volcanic vent are reported. The observation of volcanic ash using lidar remote sensing instruments is also largely reported in literature. Lidar techniques are particularly appealing in monitoring the dispersion of a volcanic plume in the atmosphere because they are able to characterize the optical properties of volcanic ash, like backscattering and extinction coefficient [e.g., Pappalardo et al., 2004], with a high vertical and temporal resolution, as well as to provide measurements of the aerosol depolarization ratio. Their coupling with in-situ techniques is also a powerful approach for the development of retrieval algorithms for the particle size estimation.

[3] Recently, microwave weather radars have been also used for monitoring the microphysical and dynamical features of volcanic plumes during the eruption phase using a physical–statistical approach [Marzano et al., 2006a, 2006b, 2010]. These studies are referred to the observation of a high density of large size tephra particles emitted into the atmosphere that can be accurately studied also using the radar reflectivity. The performance of retrieval techniques for the estimation of particles size using the radar reflectivity in the S-, C-, X-, and Ka-band have been simulated and also successfully tested on experimental data [Marzano et al., 2006a, 2006b]. However, as far as we know, radar has been never used as a technique for monitoring the dispersion of the volcanic ash in the atmosphere over long spatial ranges. Millimetre microwave radars have been only applied to the study of hydrometeors and their sensitivity to ultragiant aerosol is mainly unknown. Studies focused on cirrus clouds report that the detection of ice particle with an effective radius lower than 30 μm using Ka-band radars should be not feasible, due to the their limited sensitivity to these particle sizes [Intrieri et al., 1993]. Nevertheless, this sensitivity threshold depends on the experimental design of the radar system.

[4] On 14 April 2010, the Eyjafjallajökull volcano in Iceland started an eruption phase characterized by an ash cloud injection into the troposphere, up to about 10 km of altitude. The space-time evolution of the distribution of the volcanic plume has been observed by EARLINET (European Aerosol Research Lidar NETwork) that performed almost continuous measurements since 15 April 2010 [Pappalardo et al., 2010] and for the whole event until 22 May 2010. The volcanic ash was monitored over The Netherlands and Northern Germany on 16 April [Ansmann et al., 2010] and then it moved toward south reaching Italy on 19 April and Greece on 21 April. In May, the volcanic plume was observed over Portugal and Spain and then over Italy, Greece and Southern Germany again. Volcanic aerosol layers have been observed at the EARLINET station in Potenza (40.60N, 15.72E, 760 m a.s.l), Southern Italy, at the CNR-IMAA Atmospheric Observatory (CIAO), using the multi-wavelength Raman lidar systems [Mona et al., 2010a; L. Mona et al., Observation of Eyjafjallajökull volcanic plume over Potenza, Southern Italy: EARLINET multiwavelength Raman lidar measurements, submitted to Atmospheric Chemistry and Physics, 2010, hereinafter referred to as Mona et al., submitted manuscript, 2010b] from 19 April to 19 May 2010. During this period, non-spherical ultragiant particles have been observed at CIAO by the Ka-band MIRA-36 Doppler microwave radar (see http://metekgmbh.dyndnds.org) operating at 8.45 mm (35.5 GHz) in four separate days.

2. Observations

[5] Since 15 April 2010, when the Eyjafjallajökull volcano eruption started a phase characterized by the creation of an ash cloud, the CIAO radar observed in different days signatures characterized by a clear spectral behaviour with a high variability of the reflectivity signal-to-noise ratio (SNR) with respect to the other momentum of the spectrum. This variability could be consistent with the detection of ultragiant aerosol particles. In the following, we show the radar observations corresponding to the four case studies collected at CIAO during the period of the volcanic eruption. The radar profiles reported in this paper have a vertical resolution of 30 m (200 ns pulse length) and a temporal resolution of 10 seconds. The radar sensitivity is −40.3 dBZ at 5 km (0.1 sec time resolution) and the Doppler velocity resolution is 0.02m/s. The linear depolarization ratio (LDR) accuracy is within +/−2.0 dB.

[6] On 19 April 2010 from about 19:30 to 22:30 UTC, the radar clearly observed a layer in the range between 1.5 and 2.0 km above the ground (Figure 1). The black line in the reflectivity signal-to-noise ratio plots is the melting layer height obtained combining radar and surface temperature measurements [Bauer-Pfundstein and Görsdorf, 2007]. The layer detected by the radar is characterized by a low reflectivity signal-to-noise ratio (SNR), lower than −10 dB, a Doppler velocity within −0.5 and 0.2 m/s and a LDR higher than −2 dB. The layer slowly descents from 2.0 to 1.5 km above the ground, around 22:30 UTC approaches the height level where a few scatter clouds are observed and, finally, disappears.

Figure 1.

Time series of the vertical profile of the (top left) reflectivity SNR, (middle left) linear depolarization ratio and (bottom left) Doppler velocity measured by MIRA-36 radar from 19:00 to 22:00 UTC on 19 April 2010; (top right) time series of the vertical profile of the SNR, (middle right) linear depolarization ratio and (bottom right) Doppler velocity measured by MIRA-36 radar from 22:00 UTC on 6 May 2010 to 01:00 UTC on 7 May 2010.

[7] From 23:50 UTC on 6 May 2010 to 00:50 UTC on 7 May 2010, the radar detected a more dense layer descending from 4.0 to 2.0 km above the ground with values of the reflectivity SNR lower than −10 dB, a Doppler velocity within −2.0 and 0.1 m/s and a LDR higher than −1 dB (Figure 1). In this case, the layer rapidly moves down to 2 km approaching a cloud layer that causes a moderate rain. The radar signature also shows that the descending layer seems to remain separate from the cloud, above its top: neither signs of interaction between them is revealed nor evidence of transition phase are observed in the LDR.

[8] On 10 May 2010, from 14:00 to 14:30 UTC, the radar measurement again shows a layer in rapid descent from about 6.0 to 2.0 km above the ground (Figure 2). The observed scenario is also characterized by the presence of altostratus between 4.0 and 5.5 km over the whole time series. At the beginning of its observation, the descending layer observed by the radar is interleaved by the altostratus, while after 14:15 UTC remains below the cloud base. Below the clouds, the descending layer is characterized by a Doppler velocity ranging within −2.0 and 0.5 m/s and a LDR higher than −4 dB. The reflectivity SNR is again lower than −10 dB up to 14:15 UTC, but from 14:15 to 14:30 UTC, the layer shows a higher reflectivity in its upper part (<10 dB) and values similar to those observed before 14:15 UTC in the lower part (−20 ÷ −10 dB). However, the LDR values are not significantly modified by this change in the reflectivity SNR values. Moreover, there is again no evidence of a phase transition in the LDR.

Figure 2.

Time series of the vertical profile of the (top left) reflectivity SNR, (middle left) linear depolarization ratio and (bottom left) Doppler velocity measured by MIRA-36 radar from 13:00 to 16:00 UTC on 10 May 2010; time series of the vertical profile of the (top right) reflectivity SNR, (middle right) linear depolarization ratio and (bottom right) Doppler velocity measured by MIRA-36 radar from 11:00 to 14:00 UTC on 13 May 2010.

[9] A longer and more evident signature has been observed on 13 May 2010 (Figure 2). The radar measurement shows a thin layer around 11:30 UTC rapidly coming down from about 7.0 km up to about 3.5 km above the ground, where a nucleation mechanism could occur with the formation of a cloud, observed by the radar itself. The observed descending layer is again characterized by an average reflectivity SNR lower than −10 dB, with a peak of about −5 dB in the middle of the layer, a Doppler velocity within −2.0 and 0.5 m/s and a LDR higher than −3 dB. The temporal evolution of the layer is again characterized by a rapid descent of the layer in the free troposphere but over a longer temporal window.

[10] Radar observations are co-located with observations performed with two multi-wavelength Raman lidar systems. The lidar systems observed volcanic aerosol layers over Potenza in the period from 19 April to 19 May 2010. Radar and lidar measurements are available during the whole period of the volcanic eruption except in presence of low clouds or rain, when the lidar systems are not operative. The aerosol layers observed by the multi-wavelength Raman lidar have been classified combining lidar measurements with Lagrangian dispersion models that allow us to trace the path followed by the observed air masses revealing that they have been originated in the region surrounding the Eyjafjallajökull volcanic area [Mona et al., 2010a; Mona et al., submitted manuscript, 2010b; Pappalardo et al., 2010]. An example of co-located radar and lidar observations is reported in Figure 3. Figure 3 reports the vertical profile of the volume backscatter coefficient, as measured from the lidar at 355 nm on 19 April at 20:20 UTC, and the corresponding profile of the LDR obtained from the radar. Both lidar and radar profiles are integrated over of 10 minutes time resolution and they have a vertical resolution of 30 m. In the lidar profile, two aerosol layers, above the nocturnal boundary layer at 800 m above the ground level (a.g.l.), have been identified between 1.1 and 3.1 km a.g.l and 3.5 and 4.4 km a.g.l. and classified as of volcanic origin. The ultragiant aerosol layer detected by the radar, characterised by high LDR values, is located between about 1.4 and 2.1 km above the ground. This layer is completely included in the lower volcanic aerosol layer observed by the lidar and characterized by a maximum value of the 355 nm volume backscattering coefficient of 2.3 Mm−1 sr−1. The reported lidar data provide a description that is consistent with the scenario observed by the radar and they qualitatively strengthen the presented analysis of radar observations.

Figure 3.

Volume backscatter coefficient at 355 nm (black line) retrieved from lidar observations at 20:20 UTC on 19 April 2010 and the corresponding profile of the linear depolarization ratio (LDR) obtained from the radar (red dots).

[11] The availability of water vapour mixing ratio measurements, provided by the lidar, and of temperature and RH profiles retrieved from a co-located microwave profiler (F. Madonna et al., CIAO: The CNR-IMAA advanced observatory for atmospheric research, submitted to Atmospheric Measurement Techniques, 2010), allows us to further confirm the absence of ice and droplets. The aerosol layers observed using the radar in the three cases reported in May are related to the presence of a humidity field with values lower than 60%, obtained by combining lidar water vapour measurements with microwave temperature profiles as well as by the relative humidity profile retrieved from the radiometer. Only in the case of 7 May 2010, after 00:30 UTC, the aerosol layer detected by the radar intrudes a high humidity structure (RH > 90%) located below 2 km above the ground but this does not seem to particularly affect the dynamical evolution of the aerosol layer. Moreover, on 13 May 2010, RH values between 4 and 7 km above the ground are lower than 20%, indicating unfavourable atmospheric conditions for the occurrence of nucleation. Therefore, the retrieved RH values allow us to exclude that, for the cases reported in May, the aerosol particles might have nucleated generating ice crystals or water droplets. Moreover, there is no evidence of water phase transition in the LDR. On 19 April 2010, the layer is located at a lower height level and is characterized by larger RH values (<88%). Nevertheless, no evidence of water phase transition in the time series of LDR is observed.

3. Discussion and Conclusion

[12] The described case studies basically show similar scenarios and the observed radar signatures are consistent with the observation of ultragiant aerosol particles. The main common element is the very high LDR values (>−4 dB). These LDR values are usually addressed in literature as due to the presence of particular ice formation, as graupel, to multiple scattering effect [Battaglia et al., 2006] or, as we suppose in our case, to the effect of aerosol bulk density on the microwave propagation, in combination with the large non-sphericity of the observed particles [Intrieri et al., 1993; Matrosov et al., 2001]. Actually, MS usually occurs in high density medium, while in the described cases the radar reflectivity SNR is consistent with a low volume particle concentration. The observed LDR might be generated by the effect of the bulk density and of the non-sphericity of the ultragiant aerosol particles. Bulk density typically occurs in granular solids: it is not an intrinsic property of a material and it can change depending on the turbulent mixing of the observed particles.

[13] Another possible effect that could produce large depolarization values in spring is the presence of birds migrations that, as a whole, could be compatible only with the radar observation on 19 April 2010, when the observed layer is lower in altitude (below 3 km), while for the other cases the presence of birds can be excluded because the layers are located at higher altitudes and because of their rapid dynamical evolution in the troposphere. Also on 19 April 2010, the observed high depolarization values, ranging out of the values reported in literature for bird observations, typically about −10 ÷ −7 dB [Dinevich and Leshem, 2007], and the lacking of an evident increase of the Doppler velocity allow us to exclude that we are observing birds migration.

[14] It is also worth to be mentioned that similar high values of the LDR have been never observed at CIAO since March 2009, when the radar became operational. This also allows us to suppose that all the identified aerosol layers are not related to local sources. Moreover, the reported signatures of the radar LDR values could be also considered as one of the parameters to be used for the identification of such kind of events.

[15] The investigation of the collected radar spectra also allows us to exclude that the observed layers are related to the observation of cloud layers, because they are characterized by a higher variability with respect to clouds. Indeed, the automatic peak width analysis, routinely performed on the radar spectra mainly based on thresholds applied to LDR, suppresses the observed layer classifying them as plankton [Bauer-Pfundstein and Görsdorf, 2007]. However, ice nucleation processes involving the ultragiant particles occurred under certain conditions along their path from the source to Potenza, and subsequent sublimation/evaporation during their descent in the atmosphere cannot be excluded at this stage of the analysis. On the other hand, volcanic particles not enhanced by hydrometeors may result in especially long-lived volcanic clouds [Webley and Mastin, 2009] as long as 20 days after eruptive events [Tupper et al., 2003].

[16] The difference in the falling velocity between the ultragiant particles observed by the radar on 19 April 2010 and those observed in May could be related to the different meteorological conditions occurred over Europe and, therefore, to the significant difference in the path followed by the volcanic air mass injected into the atmosphere during two different eruption periods. According to meteorological analysis and chemical dispersion models, during the first eruption period, started in mid April, the volcanic aerosol was directly transported from Iceland to Central Europe and, after a few days of circulation over the continent, they reached over South Italy. During a second period, started at the beginning of May, instead, the volcanic aerosol was mainly transported over the Atlantic Ocean, passing over Ireland, and then transported West off the Iberian Peninsula before reaching the Mediterranean Basin and Southern Italy.

[17] In conclusion, the importance of the presented observations is twofold. As far as we know, observations of ultragiant particles using a Ka-band radar are not reported in literature. Moreover, according to the described scenario occurring over Europe in the considered time period, the reported case studies represent also a potential evidence that non-spherical ultragiant tephra particles injected into the upper troposphere from Eyjafjallajökull eruption could have been transported over more than 4000 km and over a temporal scale larger than 72 hours depending on the different case studies and on the meteorological synoptic situation. This spatial range is longer than the expected and reported in literature, considering both the cited paleo-climatic studies and the forecasts inferred from simulation of the tephra sedimentation mechanisms. The described observations, coupled with air mass trajectory analysis could be also helpful for the identification of geographical areas of potential interest for volcanological and paleo-climatological studies.

[18] The sensitivity of a Ka-band radar in the detection of giant and ultragiant aerosol implies that the observed signatures are consistent with the presence of non-spherical ultragiant particles, with an effective radius probably higher than about 30 μm [Matrosov et al., 2001] consistently with the results reported in literature for long range transport of volcanic tephra [Lacasse, 2001]. The possibility to detect this type of ultragiant aerosols using a Ka-band radar could strongly contribute to change the observation strategy of these events. Radar observation could also close an important observation gap in the study of the impact of ultragiant aerosol on the weather and climate system with the added values that they are available in all weather conditions. Giant aerosol particles or giant Cloud Condensation Nuclei (GCCN), e.g., sea salt and dust, are supposed to be a potential element to expedite the warm-rain process (especially in anthropogenic polluted clouds). Rosenfeld et al. [2002] suggested that the GCCN invoke an effective collection process within the cloud leading to larger drops and, therefore, precipitation. Future investigations will be focused on the use of the simultaneous co-located radar and lidar observations with the aim to retrieve the effective radius of the observed ultragiant particles.


[19] The financial support of the national project “Programma Operativo Nazionale (PON) – Regione Basilicata 2000/2006” is gratefully acknowledged. The financial support for EARLINET by the European Commission under grant RICA-025991 is gratefully acknowledged. The authors wish to thanks Matthias Bauer and Simone Tanelli for their very helpful comments.