Figure 1. Central Arctic boundary layer structure and potential sources of INPs. The lowest part of the boundary layer (surface mixed layer) is often decoupled from the rest of the boundary layer [Brooks et al. , 2017]. In this paper we report INP measurements both in the surface mixed layer and in the cloud mixed layer and use these measurements to infer information about the dominant sources of INPs in the central Arctic boundary layer.
2 Methods
To determine the INP concentration spectra relevant for mixed-phase clouds in the central Arctic, 48 days of sampling were conducted aboard the Swedish icebreaker Oden during Arctic summertime and into the early freeze-up period (August and September). Filter samples were collected and analysed during the journey towards the North Pole from Svalbard whilst ice breaking, and whilst moored to an ice floe. The dates for the respective periods are: Marginal ice zone (MIZ) 02/08/18-03/08/18, Clean-air station 10/08/18-11/08/18, Ice-breaking 03/08/18-16/08/18, Ice floe 16/08/18-15/09/18, Ice-breaking 15/09/18-19/09/18, MIZ 19/09/18.
2.1 Aerosol sampling from the ship and the balloon borne platform
For the ship-based aerosol sampling, filters (0.4 μm pore size, polycarbonate, Nuclepore Track-Etched Membrane Filters, Whatman) were collected by subsampling from a heated whole-air inlet at a flow rate of 9 L min-1 (standard temperature and pressure). The inlet was mounted on the 4th deck of the ship, 25 m above mean sea level. This type of filter has been used previously for INP sampling and has a low background INP count and high particle recovery rates [Adams et al. , 2020; O’Sullivan et al. , 2018; Sanchez-Marroquin et al. , 2021]. In addition, these filters collect aerosol across the full atmospheric size distribution with high efficiency, despite having pores of 0.4 µm (smaller aerosol particles are efficiently lost to the filter surface through diffusional processes) [Adams et al. , 2020].
Aerosol samples from a balloon-borne sampler, the selective-height aerosol research kit (SHARK) [Porter et al. , 2020], were collected above the surface mixed layer. All inlets were covered until sampling was started via a radio signal from the ground. Two cascade impactors (100 L min-1, MSP Model 128, TSI, USA and 9 L min-1 Sioutas, SKC Ltd., UK) sampled aerosol. See [Porter et al. , 2020] for details of the size bins and how data from SHARK are treated. A radiosonde (S1H2-R, Windsond, Sweden) was used to measure the temperature, pressure and relative humidity. In order to choose an appropriate altitude for sampling, the radiosonde was constantly operating to provide information to the user on the ground about the SHARK altitude and boundary-layer temperature and humidity structure as the SHARK was ascending. In addition, sampling was paused if the relative humidity increased above 80 %, and was stopped completely before the SHARK was brought back down.
2.2 INP analysis
Filters were analysed for INP content as soon as possible after sampling, usually within 1 - 12 h of being removed from the inlet. The filter samples were not frozen before offline INP analysis, due to concerns this may affect the INP activity, but were stored at +4 °C. Performing the analysis on ship soon after sampling minimised the chances (risks?) of changes in the INPs on the filter since storage at any temperature is expected to affect the activity of the samples [Beall et al. , 2020]. The aerosol particles on the filters were washed into either 5 or 10 mL of ultra-pure water (Millipore Alpha-Q, with a resistivity of 18 MΩ cm at 25 °C) to suspend the collected aerosol particles. These particle suspensions were then pipetted to form an array of 1 µL droplets on a cold stage, the Microlitre Nucleation by Immersed Particle Instrument, µL-NIPI [Whaleet al. , 2015]. The uL-NIPI is a standard INP measurement instrument that has been benchmarked alongside a range of other INP instruments during a number of intercomparison studies [DeMottet al. , 2018]. The cold stage cooled at a controlled rate of 1 °C min-1 until all droplets had frozen, and the freezing events were recorded in order to determine the concentration of INPs with respect to the volume of air that had been sampled through the inlet. Heat sensitivity of the collected INP samples was determined by heat treatment, where subsamples of the particle suspensions in 50 mL conical centrifuge tubes were immersed in a water bath at 100 °C for 30 min, before being reanalysed using the µL-NIPI [Daily et al. , 2021].
The INP concentration data presented here is shown with the contribution from the background accounted for. The background influence on the INP concentration was determined by collating the differential nucleus concentrations for water and handling blanks, and subtracting this from the sample differential nucleus concentrations. The differential concentrations were then summed to produce the cumulative INP spectra [Sanchez-Marroquin et al. , 2021].
2.3 Other measurements at ship level
To evaluate the concentration of dimethyl sulfide (DMS), filter samples of DMS were collected and analysed onboard. Equivalent black carbon (eBC) concentrations were obtained from a multi-angle absorption photometer (MAAP, Model 5012, Thermo Fisher Scientific Inc.). Particle size distribution measurements were made continuously using an aerosol spectrometer (WELAS 2300HP, Palas GmbH) for particles of size 0.15 - 9.65 µm, and a differential mobility particle sizer (DMPS) with a custom-built medium Vienna-type differential mobility analyzer (DMA) with a mixing condensation particle counter (MCPC, Model 1720, Brechtel Manufacturing Inc.) for particles of size 10–921 nm.
An ion chromatography system (ICS-2000, Thermo Fischer Scientific, previously Dionex) was used to determine the chemical composition of the samples. Using certain standards, the concentration of chloride, nitrate, sulphate, mesylate, methane sulfonic acid, sodium, ammonium, potassium, magnesium, calcium in the sample were determined from the ion chromatograms. A synthetic sample (QC Rainwater Standard, Inorganic Ventures, USA) was used to estimate the random percentrage error, which is up to 3 %. More details on the method can be found in Leck and Svensson [2015].
2.4 Prevention of ship stack pollution
Combustion products in a ship’s exhaust may influence INP populations [Thomson et al. , 2018]. In order to ensure that the INP concentrations measured were not affected by the ship stack emissions, rigorous sampling procedures were put in place. The aerosol sampling inlets faced the ship’s bow and the ship was manoeuvred to face into the wind whenever the wind direction changed, which minimised the probability of sampling ship stack emissions. In addition, an auto-stop for the inlet pumps was operated if aerosol concentrations increased suddenly (which would be indicative of sampling the ship stack plume), halting the sampling until aerosol size distributions returned to normal. As a precaution, the direction and speed of the wind was monitored closely, and sampling was stopped when there was a chance that the wind might introduce ship stack to the sampled aerosol. Finally, sampling was stopped if any activity that could produce aerosol was planned, including the movement of the ship, ice coring, and helicopter flights (this involved the operators being on call 24 hours a day to respond to any potential contamination). Smoking of cigarettes was also only allowed in certain areas of the ship, to ensure there was no influence on aerosol sampling.
2.5 Backward trajectories
In order to define the potential origin of measured INPs, backward trajectories of the air reaching the sampling location was conducted. The 10-day (only 7 days of which are used here) back trajectories were calculated using the Lagrangian analysis tool LAGRANTO [Sprenger and Wernli, 2015] with wind fields from 3-hourly operational ECMWF analyses, interpolated to a regular grid with 0.5° horizontal resolution on the 137 model levels. The trajectory data contains the hourly positions (longitude, latitude, pressure) along the trajectory. To focus on the segments of the trajectories that can potentially be affected by surface aerosol emissions, the trajectories are only included when they were within the model boundary layer. Additionally, removal of aerosol by precipitation, which may remove the signature of upwind aerosol sources via wet deposition, has been considered by removing all the trajectory points before the precipitation event (using a threshold of 0.1 mm h-1). The overall relationship with origin is unchanged by the addition of this filter, which indicates that the results were insensitive to precipitation events.
3 Results and Discussion
3.1 Ice-nucleating particle concentrations within the surface mixed layer
We first present our INP concentrations derived from samples collected on the ship, which was within the surface mixed layer (Figure 2a). The concentrations of INPs measured in the surface mixed layer were highly variable, and ranged from < 6 × 10-3 INP L-1 to 2 INP L-1 at −15 °C. This resulted in INP activation temperatures ranging from −9 to −30 °C for a concentration of 0.1 INP L-1. This is clearly contrary to what we expected in this remote location based on measurements in other remote oceanic loactions around the world. For example, in the Southern Ocean, INP concentrations are systematically at the low end of what we observe here [McCluskey et al. , 2018a; Murray et al. , 2021; Welti et al. , 2020].
The vast majority of INP measurements made in the Arctic were made on land or at least some distance from the Pole. A summary of these measurements is given Figure 2b. These measurements clearly show that there are strong sources of INP between around 65 to 80°N [Hartmannet al. , 2021; Sanchez-Marroquin et al. , 2020; Toboet al. , 2019; Wex et al. , 2015]. Our measurements demonstrate that the INP concentrations can also sporadically be very high in the pack at the North Pole. Previous measuremets at close to the Noth Pole also reveal substantial variability in INP concentrations active at -15°C [Bigg, 1996; Bigg and Leck, 2001]. Our results indicate sporadically higher concentrations than those results and also demonstrate that the concentration of INP can be in exces of 0.1 L-1 at temperatures up to around -10°C. We come back to the question of where these highly active INP come from later in the paper.