Advanced Search
Article Contents

Multiyear Observations of Deposition-Mode Ice Nucleating Particles at Two High-Altitude Stations in India

doi: 10.1007/s00376-017-7048-8

Get Citation+


Share Article

Manuscript History

Manuscript received: 09 March 2017
Manuscript revised: 19 May 2017
Manuscript accepted: 22 June 2017
通讯作者: 陈斌,
  • 1. 

    沈阳化工大学材料科学与工程学院 沈阳 110142

  1. 本站搜索
  2. 百度学术搜索
  3. 万方数据库搜索
  4. CNKI搜索

Multiyear Observations of Deposition-Mode Ice Nucleating Particles at Two High-Altitude Stations in India

  • 1. Department of Atmospheric and Space Sciences, Savitribai Phule Pune University, Pune, India 411007
  • 2. Indian Institute of Tropical Meteorology Pashan, Pune, India 411008
  • 3. Skymet Weather Services Pvt Ltd., Vashi, Navi Mumbai, India 400703

Abstract: Ice nucleating particle (INP) measurements were made at two high-altitude stations in India. Aerosols collected on filter paper at Girawali Observatory, Inter University Center for Astronomy & Astrophysics (IGO), and at the Radio Astronomy Center, Ooty (RAC), were activated in deposition mode using a thermal gradient diffusion chamber to determine the INP concentrations. The measurement campaigns at IGO were conducted during 2011, 2013 and 2014, and at RAC during 2013 and 2014. When the aerosol samples were exposed to an ice supersaturation of between 5% and 23% in the temperature range -17.6°C to -22°C, the maximum INP number concentration at IGO and RAC was 1.0 L-1 and 1.6 L-1, respectively. A maximum correlation coefficient of 0.76 was observed between the INP number concentration and ice supersaturation. The airmass trajectories analyzed for the measurement campaigns showed that the Arabian Desert and arid regions were the main INP contributors. Elemental analysis of particles showed the presence of Na, Cl, Si, Al, Fe, Cu, Co, Cd, S, Mn and K, as well as some rare-Earth elements like Mo, Ru, La, Ce, V and Zr. When aerosols in the size range 0.5-20 μm were considered, the fraction that acted as INPs was 1:104 to 1:106 at IGO, and 1:103 to 1:104 at RAC. The higher ratio of INPs to aerosols at RAC than IGO may be attributable to the presence of rare-Earth elements observed in the aerosol samples at RAC, which were absent at IGO.

摘要: 本文通过印度两个高海拔站开展冰核观测. 在天文与天体物理校际中心(IGO)的吉徕瓦里(Girawali)天文台和乌蒂(Ooty)射电天文中心(RAC), 气溶胶通过滤膜收集后在热力梯度扩散云室中测量凝华核化形成的冰核数浓度. IGO站的观测时间为2011年、2013年和2014年, RAC站的观测时间为2013年和2014年. 在冰面过饱和度为5%至23%、温度为?17.6 °C至?22 °C的条件下, 观测到的最大冰核数浓度在IGO站和RAC站分别为1.0 L?1和1.6 L?1. 冰核数浓度和冰面过饱和度之间的最大相关系数达到0.76. 气团轨迹分析表明冰核主要来自阿拉伯沙漠和干旱区. 元素分析显示气溶胶粒子中含有Na、Cl、Si、Al、Fe、Cu、Co、Cd、S、Mn和K元素以及稀有元素Mo、Ru、La、Ce、V和Zr. 气溶胶粒径为0.5至20 μm时, 活化冰核占气溶胶比例在IGO站为1:104 至1:106, 在RAC站为1:103 至1:104. 相比IGO站, RAC站高的冰核占比归因于气溶胶粒子中含有IGO站缺少的稀有元素. (翻译: 赵震)

1. Introduction
  • Predicting the formation and occurrence of ice crystals in clouds is one of the biggest challenges for climate and regional models (Boucher et al., 2013) and is a source of prediction uncertainty. One of the causes for the uncertainty in models is aerosol particles, which aid the formation of ice crystals in the atmosphere (Boose et al., 2016). Clouds can contain liquid droplets and ice crystals in coexistence between the levels of 0°C and -38°C in the atmosphere (Pruppacher and Klett, 1997). Cloud droplets freeze homogeneously below -38°C. Another mechanism of ice formation in the atmosphere is heterogeneous nucleation, where ice formation is triggered by special types of aerosol particles known as ice nucleating particles (INPs) (Pruppacher and Klett, 1997). (Vali et al., 2015) defined INPs as particles, material, substances, and objects etc. that are responsible for heterogeneous nucleation. INPs can alter cloud microphysics, and hence precipitation, cloud lifetime, and the radiative properties of clouds (DeMott et al., 2003a, 2010). An order-of-magnitude increase in INP concentration can increase the radiative forcing of clouds by approximately 1 Wm-2 (DeMott et al., 2010). The role played by aerosols as INPs in the formation of precipitation and the modulation of cloud radiative properties has long been an important topic of research globally, but there has been a particular resurgence in this field with in the last decade (DeMott et al., 2011). Heterogeneous nucleation can take place via four different mechanisms: deposition nucleation, condensation freezing, immersion freezing, and contact freezing. Deposition nucleation takes place when water vapor is directly deposited on INPs, when the surrounding air is supersaturated with respect to ice. Condensation freezing takes place when INPs act as cloud condensation nuclei (CCN), and ice nucleation occurs when droplets freeze during the CCN activation stage. In immersion freezing, INPs act as CCN at warmer temperatures, with the freezing of droplets taking place at lower temperatures; and in contact freezing, INPs collide with super cooled droplets, leading to droplet freezing (Pruppacher and Klett, 1997; Lopez and Ãvila, 2013; Vali et al., 2015). The efficiency of ice nucleation also depends upon the temperature and supersaturation to which the INPs are exposed (Pruppacher and Klett, 1997; Cziczo et al., 2009; Hoose and Möhler, 2012). The role played by the chemical composition of INP aerosols, the chemical ageing of aerosol particles, the total particle surface area of INPs, and the active sites and their importance, have been discussed in several works (Pruppacher and Klett, 1997; Kanji et al., 2008; Möhler et al., 2008; Phillips et al., 2008; Cziczo et al., 2009). The importance of some of these factors is still under investigation (Lohmann et al., 2016).

    To understand the characteristics of INPs and other related aspects, various field campaigns, laboratory experiments and modeling studies have been carried out across the world during the last 60 years. Among these, several of the experimental studies have used expansion cloud chambers to study INP activation, and sugar solution is used to count the ice crystals that develop (Lopez and Ãvila, 2013). Chamber equipment like AIDA (Aerosol Interaction and Dynamics in the Atmosphere) is used to simulate atmospheric conditions in laboratory. AIDA is an 84 m3 and 4-m diameter actively cooled chamber placed in a thermally insulated box. The temperature of the air inside the chamber can be cooled by adiabatic expansion. A background particle-free situation with a background count of 0.1 cm-3 can be achieved in AIDA (Möhler et al., 2008; Cziczo et al., 2009).

    With the development of diffusion chambers and membrane filters, aerosol samples from various locations have been collected by drawing known volumes of air through the membrane filter. These filters are exposed to freezing temperatures and supersaturations in a controlled manner to activate the INPs (Kulkarni et al., 2009; Santachiara et al., 2010; Patade et al., 2014). Although this method is relatively easy to perform experimentally, it has several disadvantages, including its failure to provide real-time INP data in the air sample. Another category of equipment is the continuous flow diffusion chamber (CFDC), which is used for online measurements of INPs in an air sample (e.g., Rogers, 1988; DeMott et al., 2003a). Many CFDCs in use are laboratory versions, with only a few having gone into the field to measure INPs. A complete review of INP measurements and the challenges involved is available in (DeMott et al., 2011).

    Air masses over North America and central Europe are generally of African origin (DeMott et al., 2003a, Conen et al., 2015). (DeMott et al., 2003a) observed that, within a dust layer—— and for aerosol particle sizes of less than 1 μm, at a temperature of -36°C, and at an ice supersaturation of 23% —— the concentration of INPs exceeds 1 cm-3. (Rogers et al., 1998) found INP concentrations ranged from <0.1 to ∼500 L-1 and, whilst overall wide variability was observed, they were generally greater at colder temperatures and higher supersaturation. Similarly, (Sassen et al., 2003) reported Saharan dust as an effective INP source able to nucleate ice at temperatures ranging from -5.2°C to -8.8°C.

    The INP size range, which is most active for heterogeneous freezing, provides another dimension for study. (Santachiara et al., 2010) reported that PM1 particles account for 50% of INPs in the atmosphere. In their study, they considered size ranges of PM1, PM2.5 and PM10, and further described particles with diameters of <10 μm as being involved in ice nucleation, with positive correlation between INP concentration and supersaturation. They reported an average INP concentration of 337 139 m-3 at an ice supersaturation of 32% and a temperature of -18°C. (Chou et al., 2011) observed that the INP number concentration increases with the concentration of larger-sized particles, and found an INP number concentration of >25 L-1on dusty days, and an average of 14 L-1 on non-dusty days, at Jungfrajoch, Switzerland. (Klein et al., 2010) reported a similar observation; specifically, they found a high correlation between INPs and aerosol surface area, and reported an INP concentration of 40 L-1 at -18°C. (Boose et al., 2016) conducted a field campaign in Tenerife, Spain (2373 m MSL), during July and August of 2013 and 2014. They reported that, for temperatures between -40°C and -20°C,and a relative humidity of 100% with respect to ice and water, dust particles were effective INPs. Furthermore, they reported that the concentration of INPs varied from 0.2 L-1 to 2500 L-1 at -33°C and 105% relative humidity with respect to water. (Jiang et al., 2015) studied the characteristics of atmospheric INPs at the top of the Huangshan Mountains, China, and observed reduced concentrations of INPs with increasing altitude. They also found that the changes in the INP concentration with time were correlated with the change in aerosol number concentration within the size range of 0.5-20 μm. They reported an INP concentration of 11 L-1 at -20°C and an ice supersaturation of 22%. (Jiang et al., 2016) measured the INP activity of desert dust aerosols in a dust episode and dust-free episode in north western China. The average INP concentration at -20°C and an ice supersaturation of 22% on non-dust days was found to be around 11 L-1, whereas this value during a dust event was several hundred per liter, with a correlation between aerosols diameters of >0.5 μm and INP concentration also reported.

    To date, very few studies on INPs have been carried out in India (e.g., Prabhakar and Ramana Murty, 1962; Paul et al., 1985; Paul, 2000, Patade et al., 2014). From aircraft observations, (Patade et al., 2014) reported an average INP concentration of 1.12 L-1, with a maximum of 5 L-1, at ice supersaturations varying from 6% to 24% and a temperature range from -18.5°C to -13.5°C. These measurements were highly variable in space and time, and the sample collection at any point in the atmosphere was for a few minutes only. Although measurements were reported for three years during the Indian summer monsoon, the data were limited.

    The present study is an extension of the study conducted by (Patade et al., 2014), reporting more observations of the type of aerosols that activate as INPs over the Indian region. Measurement campaigns were conducted at two high-altitude stations, with the aim to: (1) measure the INP concentrations; (2) study the diurnal variation of INPs ; (3) obtain information on the ratio of INPs to aerosols and investigate the elemental composition of aerosols at the two high-altitude stations during the campaign period.

2. Method
  • The first sampling site, at Girawali Observatory (19.07°N, 73.40°E; 1005 m MSL), Inter University Center for Astronomy and Astrophysics (IGO), is located in the western part of India and on the leeward side of the Western Ghats (Sahyadri mountain range). The second site, at the Radio Astronomy Center, Ooty (RAC) (11.38°N, 76.66°E; 2240 m MSL), is located in the southern part of India in the Nilgiri mountain range, also a part of the Western Ghats. Both sites are remote and situated away from local vehicular and industrial pollution.

  • The sampling of aerosol particles for INP measurement was carried out in a similar way to (Patade et al., 2014). The aerosol samples were collected on hydrophobic filter paper (11806-25-N, Sartorius Stedim Biotech GmbH, Germany) with a pore diameter of 0.45 μm. The sampling filters were placed inside the cartridges. Sampling cartridges were mounted on a tripod stand at 2.5 m from ground level and 90 L of air was drawn through at a flow of 2 L min-1. The cartridges were closed and kept in a cartridge carrier (developed in-house) to avoid any particle losses due to tilting and shaking during transport. A single cartridge was opened only at the time of processing inside the thermal gradient diffusion chamber TGDC. The sampling was done at the highest point of the sampling site, away from trees and buildings. Aerosol concentrations were measured with a portable aerosol spectrometer (Model 1.108, GRIMM, Germany) throughout the campaign period. This instrument is a battery-operated light weight and portable optical particle counter. The measuring principle is that the light scattering of a single particle with a semiconductor laser as a light source and particles can be detected over a size range from 0.3 μm to 20 μm. In the present study, the instrument was operated in count mode, to obtain the aerosol number concentration at 1-min intervals. At RAC, the aerosol concentration measurements were performed from 0900 to 1800 local time (LST; UTC+0530), and at IGO they were performed from 0700 to 2300 LST. Four to five samples of INP measurements were collected every day of the campaign. The number of samples collected during each campaign is listed in Table 1.

    To understand the elemental composition of aerosol particles, separate sampling was carried out and the filters were analyzed for their elemental composition. The elemental composition of aerosol particles was determined using scanning electron microscopy coupled with energy dispersive x-ray spectroscopy. Separate filters were used for sampling the aerosol particles. More than 240 L of air was drawn through the Polytetrafluoroethylene (PTFE) filter (11806-25-N, Sartorius Stedim Biotech GmbH, Germany), with a pore size of 0.45 μm. The aerosol samples were taken during the afternoon between 1200 and 1430 LST.

  • The filter samples were analyzed in the TGDC described in detail by (Patade et al., 2014). However, a brief description of the TGDC is also given here. The TGDC consists of two aluminum plates separated by a Teflon block. Each of these plates has a cavity and is kept inside a walk-in cold room. Water is poured inside the cavity of both plates to prepare ice layers. As the temperature of both plates reaches -25°C, the filter paper is placed on a mesh located at the center of both plates and the TGDC is closed. The temperature of the upper plate is now raised by a heating mat, and heating continues till the plate reaches a steady temperature. This steady temperature is fixed between -19°C and -10°C, depending on the supersaturation required in the TGDC. The temperature of the lower and upper plates is used to calculate the saturation ratio with respect to ice at the center of the TGDC via the method given in (Saxena et al., 1970) and (Kulkarni et al., 2009). After 5 to 10 min, the TGDC is opened and ice crystals formed on the filter paper are counted with the help of a magnifying glass. The calibration methods and limitations of this TGDC are given in (Patade et al., 2014). The lower detection limit was determined by the background counts of 15 unexposed filters. The average back ground count was found to be 6.4 INP per filter, with a standard deviation of 1.3. The filters were processed at the highest ice supersaturation attainable in the TGDC (22%). The background was subtracted from the total number of ice crystals that developed on an exposed filter.

    To determine the effect of the rate of sampling, 120 L of air was sampled at two flow rates: 2 and 10 L min-1. Four samples per day (two each) in the morning and evening were collected for five days. All samples were processed at -18°C and at an ice saturation of 20%. The average INP concentration for the 2 and 10 L min-1 flow rates was 0.269 L-1 and 0.242 L-1, respectively, with standard deviations of 0.145 and 0.150. The Student's t-test showed there was no significant effect of the two flow rates on the INP count. Therefore, we assumed that the effect due to the rate of sampling was negligible for these two flow rates.

    A picture of a typical ice crystal-activated filter after processing in the TGDC is shown in Fig. 1a. Figure 1b shows an overpopulated sample where ice crystals could not be measured with our method. The possible reasons for samples with overpopulated ice crystals are discussed later in the paper.

  • Five-day back trajectories were calculated using the Hybrid Single-Particle Lagrangian Integrated Trajectory (HYSPLIT) model (Draxler and Hess, 1998; Li et al., 2014). Details of the back trajectories during the campaign period are discussed in sections 3.2 and 3.3. During our measurement periods, the five-day back trajectories showed that the air masses originated from land (arid or desert). (DeMott et al., 2010) reported that the mode diameter of particles found at the center of ice crystals is ∼0.5 μm. As the pore size of the filters used in several previous studies in different regions was 0.45 μm (Patade et al., 2014), the same was used in the present study. However, using this pore size biased our study towards examining the role of particles larger than 0.45 μm in size as INP ——a range where mineral and soil dust particles dominate the free troposphere (DeMott et al., 2010). This is one of the limitations in using filters with a pore size of 0.45 μm.

3. Results and discussion
  • The INP concentrations were determined as a function of supersaturation over ice. The supersaturation with respect to ice varied from 5% to 23% over a temperature ranging from -17.6°C to -22°C. Figures 2a-c shows the variation in INP concentration with supersaturation with respect to ice (SSi) at IGO during the years 2011, 2013 and 2014, respectively. Figures 3a and b show a similar variation, but at RAC during the years 2013 and 2014, respectively. Figures 2b and c and Figures 3a and b show that the INP concentration increased with ice supersaturation. This illustrates a good agreement with other studies (Meyers et al., 1992; DeMott et al., 2003a; Kulkarni et al., 2009; Welti et al., 2009; Kulkarni and Dobbie, 2010; Lopez and Ãvila, 2013; Patade et al., 2014; Jiang et al., 2015). However, several factors apart from supersaturation with respect to ice, such as aerosol particle surface area, time for which the particles are exposed to supersaturation, their chemical composition, contact angle between particles and water, crystallographic structure of particles, and the mineralogy of aerosols, also affect the ice nucleation activity of aerosol particles (Kanji and Abbatt, 2006; Eastwood et al., 2008; Kanji et al., 2008; Kulkarni et al., 2009; Welti et al., 2009; Klein et al., 2010; Kulkarni and Dobbie, 2010). Consequently, the observed variation may be associated with variation in surface chemical composition or surface physical features like cracks, steps, cavities and pores. Such irregularity over surfaces might be responsible for the change in surface energy and reactivity with different chemical composition (Kulkarni and Dobbie, 2010).

    Figure 1.  Representative photographs of sample filters: (a) ice crystals developed on a sample filter; (b) large concentration (over populated filter) of small crystals developed on the filter. The filters with over populated ice crystals were not counted in the analysis.

    Figure 2.  Variation in INP concentration with SSi at different temperature ranges at IGO during (a) 2011, (b) 2013 and (c) 2014. The y-axis error bar is the standard deviation calculated in for the supersaturation bin shown by the x-axis error bar.

    Figure 3.  Variation in INP concentration with SSi at different temperature ranges at RAC during (a) 2013 and (b) 2014. The y-axis error bar is the standard deviation calculated in the supersaturation bin shown by the x-axis error bar.

    Figure 4 shows the variation in mean aerosol particles number concentration with size at IGO for the years 2011, 2013 and 2014, as well as RAC for the year 2013 measured with the GRIMM spectrometer. At both locations, a dominance of aerosol particles in the range of 0.3 to 1 μm was observed with the concentration of higher size ranges rapidly decreasing. In the present study, we did not find any correlation between total INP concentration and total aerosol number concentration. Aerosols are observed to be externally and internally mixed in any airmass (Abbatt et al., 2006; Kanji et al., 2008), so the degree to which ice nucleates upon this complex mixture will depend upon the relative surface areas of each aerosol present and their intrinsic ice-forming capabilities.

    Figure 4.  Variation in mean aerosol concentration with different aerosol size range at stations IGO and RAC. Error bars represent the standard deviation of data.

    The correlation coefficient between ice supersaturation and INP's concentration (Pearson's R) was 0.55 and 0.56 at IGO and RAC, respectively. Data of all years were used to calculate the correlation coefficient. Meanwhile, in individual years, the calculation of R at IGO yielded values of 0.63 and 0.76 for 2013 and 2014, respectively. At RAC, for the year 2014, the coefficient was 0.76, while no significant correlation was found for the year 2011 at IGO, or for 2013 at RAC.

  • Figures 5a-c show the diurnal variation in INP concentration at IGO for the measurement campaigns in the years 2011, 2013 and 2014, respectively. The results show the INP count in the temperature range -17.6°C to -22°C. For the years 2011 and 2014, we did not observe any trend in diurnal variation of INP concentration. In 2011 and 2013, the air masses originated mostly from the Arabian Peninsula or Arabian Sea. In this figure, the solid filled symbols indicate the days when the airmass originated from the Arabian Desert or surrounding areas. We call these days "dust-impact days". In 2014, the airmass originated over the Bay of Bengal but had travelled for most of the time over the Indian Peninsula. The air mass originating over the Arabian Desert carried ample amounts of fine dust, and this could be observed in the formation of an excessive number of small ice crystals on the filter ——a representative image of such an over populated filter is shown Fig. 1b.

    Figure 5.  Variation in INPs with time at station IGO during (a) 2011, (b) 2013 and (c) 2014. Error bars represent the standard deviation of data for each day.

    Such anomalies of a rapid increase in INP concentration, termed "IN-storms", are characterized by a sudden increase in INP concentration within a day or less (Pruppacher and Klett, 1997). The data for such days were not included in our analysis, mainly due to the difficulty in counting these numbers of ice crystals and, hence, the INP concentration. Our samples on those days were saturated to such an extent that it was impossible to manually count the ice crystals that formed on the filter. Figures 6a and b show the variation in INPs with time in the temperature range -17.6°C to -22°C at RAC, for the campaigns in the years 2013 and 2014. It can be seen from the figure that there was no trend at this station. However, in 2013, on five successive days (17-19 January 2013 and 21-22 January 2013), during the afternoon sample period, large concentrations of ice crystals that over populated the filters were observed, indicating high concentrations of INPs, and such filters were not counted. However, unlike at IGO, the back trajectory analysis suggested that only on the last two days had the airmass originated from the Arabian Peninsula; on the other days, it originated from the Bay of Bengal. Dust-impact days for the year 2013 are denoted by solid filled symbols in Fig. 6a. The filters were overpopulated on these days, and so were not used for determining the INP concentrations. Such high concentrations of INPs may be due to the passage of the airmass over coal-fired power plants on the east coast of India (Kumar et al., 2008; Srinivas and Sarin, 2013). Figure 7 shows the five-day back trajectory of the airmass arriving at RAC. We can see that, in 2013, the airmass passed over coal-fired power plants located along the east coast of India. Studies on stack emissions from power plants have shown that, during transport, aged fly-ash diffuses in the atmosphere and contains particles that are separated from their volatile surface layer, which become active INPs; whereas, freshly formed fly-ash shows no significant contribution (Parungo et al., 1967; Schnell et al., 1976). In 2014, such a large concentration of INPs (over populated filter) was observed only on 16 January 2014. The airmass arriving at RAC on this day originated from the Bay of Bengal, but spent a long time over the Indian Peninsula.

    In 2014, a morning and evening peak was observed in the INP activation, with a minimum in the afternoon. The morning and evening peaks may have been due to domestic activities from the two nearby villages. However, we did not have access to local wind measurements at the RAC site, and the portable aerosol spectrometer was not available for the 2014 campaign.

    Figure 6.  Variation in INPs with time at station RAC during (a) 2013 and (b) 2014. Error bars represent the standard deviation of data for each day.

    Figure 7.  Five-day back trajectories in 2013 and 2014 plotted for station RAC at three different altitudes of 2200, 2700 and 3200 m MSL [color bar represents variation in altitude (m) corresponding to each trajectory, and black dots represent coal-fired power plants along the east coast of India].

    The cut off size of aerosol particles for the present study was 0.5 μm. So, there was underestimation of total INPs due to the elimination of particles in the lower size range. The portable aerosol spectrometer gives the aerosol concentration from 0.3 to 20 μm. The under estimation in the size range of 0.3 to 0.5 μm was found to be 13.52 L-1 in 106 aerosol particles at RAC in 2013, while at IGO it was 1.59 L-1 in 106, 2.8 L-1 in 106, and 2.74 L-1 in 106 for the years 2011, 2013 and 2014, respectively (Hiranuma et al., 2015; Boose et al., 2016). The large value of the INP under estimation at RAC clearly indicates that a high concentration of small particles was present in the atmosphere.

    The INP concentration at RAC in 2014 was mostly equal to or greater than 1 L-1, whereas at IGO the maximum was 1.0 L-1 and the minimum was 0.4 L-1. The altitude at RAC from sea level is almost twice that at IGO. RAC is surrounded by mountains and four different reserve forests, whereas IGO is on a standalone peak with a large gap between two mountains. So, the local winds may play a role in modulating the INP count.

  • The collected samples were processed for SEM-EDX analysis to determine the chemical composition of the samples. However, the elemental analyses carried out using SEM-EDX can be used for qualitative purposes only. During the SEM-EDX analysis, elements like C and F are suppressed, as the filter contributes these elements. The results of the analysis are summarized in Table 1.

    Few studies have been conducted over the trajectory origin areas, such as the Arabian Sea and the Bay of Bengal (Budhavant et al., 2010; Saxena et al., 2014). One study conducted over the Arabian Sea and the Bay of Bengal by (Kumar et al., 2008) showed the presence of crustal elements like Al, Fe and Si, and these crustal elements were present at both RAC and IGO. The presence of Al in samples is typically an indicator of mineral aerosol of crustal origin (Kumar et al., 2008). They also reported that the origin of the airmass over the Arabian Sea is likely from the Arabian Desert. Trajectory analysis at IGO showed that, from14 to17April 2011, the airmass originated in the Arabian Desert and was dust-rich with Si and Al. (DeMott et al., 2003b) and (Klein et al., 2010) reported that dust is the major source of INPs. (DeMott et al., 2003b) reported that INPs are composed of mineral dust and fly-ash aerosols, and (Klein et al., 2010) reported the contribution from silicate or Ca-carbonate as INPs can reach 90%.

    The back trajectory analysis in Fig. 8 indicates that the airmass at IGO originated mostly from the Arabian Desert region and travelled over the Arabian Sea. The elemental composition of aerosols showed the presence of elements such as Na, Si, Cl, Mn, Al, Fe and Cu, with S observed in 2011 and 2014 And Zr observed in 2013 and 2014. Some elements, such as Co, Cr and V, were observed only in the year 2011; and elements like Mg, K and Ti were seen in 2014.

    From Fig. 7 it can be seen that the airmass arriving at RAC was of mixed origin; on some days it was continental in origin, and on several other days it was marine in origin—— from the Bay of Bengal. Elements such as Na, Si, Cl and Fe were common at RAC and IGO, in all years. In 2013 at RAC, other elements seen were Cd, Mo, Ru, La and Ce. The airmass trajectories of the two stations showed that it originated over land but transited over the sea. Hence, the effect of marine aerosols cannot be neglected. Although the SEM analysis showed the presence of Na and Cl, we are unable to give a quantitative value. (Wilson et al., 2015) suggested that marine aerosols of biogenic origin and less than 0.2 μm in size have good ice nucleating ability. In light of the results of (Wilson et al., 2015), it is possible that we may be missing the INP contribution from marine aerosols because of the higher cut off size of 0.45 μm used in our samples.

    Figure 8.  Five-day back trajectories in 2011, 2013 and 2014 plotted for station IGO at three different altitudes of 1005, 1500 and 2000 m MSL [color bar represents variation in altitude (m) corresponding to each trajectory].

  • Aerosol concentrations measured using the GRIMM aerosol spectrometer were averaged over the 45-min observation periods when the sample collection was carried out. The fraction of total aerosol population that acted as INPs was very small. Earlier studies have reported a ratio of INPs to aerosol concentration of 1:106, with some other studies having reported even lower values. (Santachiara et al., 2010) found a ratio of between 1:108 and 1:107 when they exposed samples to temperatures of between -17°C and -19°C at SSi and water supersaturation (SSw) in the range of 20%-32% and 2%-10%, respectively. They used different size classes of aerosols: PM1, PM2.5, PM10 and total suspended particles. The fraction of total aerosol that acts as INPs depends upon temperature, supersaturation and other properties of the INPs.

    The ratio of total INP to average aerosol concentration, the so-called activated fraction, at IGO during the years 2011, 2013 and 2014, was 1:105 to 1:106 during 2011 and 2013, whereas for the year 2014 it was 1:104. At RAC, this ratio for the year 2013 was 1:103 to 1:104. The fraction of aerosol particles that acted as INPs at RAC was found to be more in comparison to IGO. One of the reasons could be the presence of rare-Earth elements like La and Ce, which were detected in the SEM-EDX analysis of the aerosol particles. The ice nucleating properties of oxides of La and Ce have been described by (Matsubara, 1973), and the critical temperatures for ice nucleation lie in the range of the temperatures we used in this study. Another reason may be the presence of different metals at RAC. It has also been reported that oxides of different metals released from steel mills and power plants possess good ice nucleating activity ——for example, CdO and Mn3O4 at a threshold temperature of -5°C to -12°C; Fe3O4 at -12°C to -20°C; and FeO and Fe2O3 at below -20°C. (Rasool, 1973). Another hypothesis is that, since RAC is surrounded by four dense reserve forests, particles of biological origin could be activating as INPs. However, the SEM-EDX analysis of the filter samples could not confirm the presence of bioaerosol.

    The activated fractions measured at IGO and RAC were higher than those reported by (Santachiara et al., 2010), as their observations were ground-based and samples were exposed to water saturation as well. (Roberts and Hallett, 1968) reported that, at a temperature of 255 K and relative humidity of 120%, one particle in 104 nucleateas ice (1:104); the sizes of particles they considered were in the range of 0.5 to 3 μm. (Rogers et al., 1998) carried out measurements of deposition-as well as condensation-mode INP activation and estimated the INP fraction to be 1:105 to 1:103 when the ratio of INPs to condensation nuclei is considered.

4. Summary and conclusions
  • Field campaigns in which deposition-mode INPs were measured at two high-altitude stations in India were conducted during two different seasons. Aerosol particles deposited on the filter were analyzed for elemental composition. The major findings and conclusions of our study are as follows:

    (1) The INP concentration increased with SSi. A positive correlation between SSi and INP concentration (L-1) was found when temperature varied from -17.6°C to -22°C and supersaturation from 5% to 23%.

    (2) No trend in INP variation with time was observed during the observation period. The maximum at RAC and IGO was 1.6 L-1 and 1.0 L-1, respectively. Altitude difference, season and activity in the path of the airmass may have been modulators of the INP number concentration.

    (3) SEM-EDX analysis showed the presence of sea-salt particles like Na and Cl, as well as crustal elements like Si and Al, and metals such as Fe, Cu, Co and Cd. The presence of Al showed the dominance of mineral aerosols. Several rare-Earth elements, such as Mo, Ru, La, Ce and Zr were found in the analysis. Metals may have originated from industrial activity along the east coast of India and oil refineries along the coast of Gujarat and the Arabian Peninsula.

    (4) The fraction of aerosols that acted as INPs was 1:103 to 1:106, when only the deposition mode was considered, and aerosol particles in the size range of 0.5-20 μm.

    The limitations of the study are that only the deposition mode was investigated, and in the temperature range of -17.6°C to -22°C. We filtered out INPs of less than 0.5 μm in size, so the study was biased towards coarse-mode particles. Also, as one of the measurement sites was RAC, which is surrounded by forests, the contamination by biological particles cannot be ruled out. Despite our measurements having been made at the surface, they were at high-altitude locations where the measurements were unaffected by local pollution. Our observations give important insights into the INP properties over this region, where data are otherwise quite sparse. In future, the study can be improved by measuring INPs in more than one mode and at wider temperature and supersaturation ranges during different seasons of the year.




    DownLoad:  Full-Size Img  PowerPoint