All current atmospheric models of how the atmosphere will respond to increasing emissions of carbon dioxide from fossil fuel burning predict global warming. The extent to which emissions must be limited will depend critically on how rapidly that warming develops. The models necessarily contain uncertainties that make it difficult for scientists to give firm advice to governments on desirable limitations to carbon dioxide emissions.

One of the largest uncertainties relates to the effects of clouds on the Earth’s radiation balance (IPCC, 2001). Satellite measurements have shown that clouds have a net cooling effect that is many times larger than the warming due to all trace gases. This cooling effect due to clouds results from the difference between the sunlight they reflect back to space (mostly in the visible and ultraviolet part of the spectrum) and the radiation (infrared) that they prevent from escaping to space. The difference is dependent on how finely the water in the clouds is divided. The more cloud drops there are for a given water content, the greater the cooling effect. The concentration of cloud drops in turn depends on how many soluble aerosol particles larger than a particular size is present in the air in which the cloud forms. In clouds colder than 0°C the presence of ice crystals can profoundly modify the size of the cloud elements, creating fewer but much larger particles and greatly decreasing the sunlight returned to space. Thus the particle population on which cloud drops or ice crystals can form is important in determining the cooling imposed by clouds. If the concentrations of those particles are themselves dependent on temperature (for example, if they are derived from or influenced by biological processes) then climatic feedback processes involving e.g. the upper ocean, ice, clouds, and radiation can occur that will either augment or oppose warming.

On a global basis, one of the main barriers to improving future climate predictions arises from the difficulty of separating the natural and man-made components of the particles on which cloud drops form (cloud condensation nuclei, CCN) or on which ice crystals can form (ice-forming nuclei, IFN). This is due to the multiplicity of local sources and the possibility of transport from more distant sources. Oceanic regions remote from land and the polar regions offer the best possibilities for reducing the number of sources and minimising human influences.

In summer, most of the particles originating over land are usually lost through scavenging by cloud drops and precipitation during their transport into the pack-ice region of the central Arctic Ocean (north of about latitude 80°N). Thus the remote central Arctic, representing more than 4% of the planet’s surface, appeared to offer a relatively simple system for studying the sources of CCN and IFN, their nature and their environmental effects. The cooling effect is most pronounced for optically thin low stratiform clouds, which are extremely common in the Arctic summer. In addition, the change in reflectivity of the ice cap when melting occurs (enhancing warming) makes the central Arctic Ocean a region particularly sensitive to climate change. These considerations lay behind the Atmospheric Programme of the International Arctic Ocean Expedition 1991 organised by the Swedish Polar Research Secretariat and a follow-up expedition in 1996 (Arctic Ocean 1996).

Results from the 1991 and 1996 expeditions that dictated the programme for the expedition of 2001

There were many unexpected findings. In spite of an apparently very stable lower atmosphere and prolonged periods without obvious changes in the weather, there were frequent large changes in the concentrations of particles and of the trace gas dimethyl sulfide (DMS).

DMS is released into seawater by marine organisms. It then escapes into the air and becomes oxidised to gaseous compounds such as sulfur dioxide (SO₂), sulfuric acid (H₂SO₄) and methane sulfonic acid (MSA). The role of oxidised DMS and ammonia (NH₃), also from marine biological sources, in allowing growth of pre-existing particles (initially too small to be active in cloud formation) was, however, confirmed through the observed chemistry of the aerosol. This was in spite of the fact that over the pack ice area (between 70–95% ice cover) local contribution of DMS (and further oxidation products) from the open leads to the atmosphere was shown to be of very little importance, compared to that transported from the substantial marine source south of the ice edge zone. Thus marine life near the ice edge of the mostly ice covered central Arctic Ocean seemed very important in controlling the available gaseous and particulate sulfur and consequent cloud formation north of 80°N.

Limited helicopter measurements in the second expedition showed vertical stratification of particles, sometimes intense. Before these expeditions it was generally believed that particle formation (nucleation) of water vapor and sulfuric acid gas originating from oxidation of DMS took place exclusively above the cloud layer. There the particles grew by further accretion of sulfuric and methane sulfonic acids and neutralization by ammonia gas. When mixed into the sub-cloud layer, the larger ones formed cloud drops and accreted further sulfur mass through aqueous phase oxidation of SO₂. While this picture was entirely consistent with the observed chemistry of the aerosol and the probable lifetimes of particles above and below cloud, there were many indications that it was not generally true. There were episodes of the formation of particles (less than 5nm in diameter) that appeared to be confined to the lowest 200 m or so and to be accompanied by particles sometimes in discrete size ranges up to about 50 nm. Moreover, these episodes were linked to the simultaneous enhancement of still larger particles that were plainly of marine origin and had a large organic content. The suspicion arose that bubbles produced in the open water between ice floes (”open leads”), either by wave action or from microbiological processes, were collecting organic material from the surface microlayer of the open water and injecting it into the atmosphere when they burst. There was the possibility that not only did this bypass the need for growing particles to CCN size above clouds, but that nucleating material released in the process could produce new particles near the surface and enlarge them by condensation of condensable gases. The weak points of these deductions were that there were no data on the organisms present in the water (other than visual evidence of large colonies of ice algae) and a complete lack of knowledge of the chemistry and biological debris present in the surface microlayer. Whatever the exact processes may be, it is evident that the unexpected nature of the particles found requires a revision of the existing theories either of particle formation or of their early stages of growth in the Arctic boundary layer (the layer of air closest to the surface extending in vertical to about 500 m). Perhaps the most intriguing aspect of the findings is that the role of biological processes in regulating climate is enhanced rather than diminished, since a greater part of the evolution of the aerosol relies on organic material than in the currently accepted sulfuric acid water vapor particle nucleating theory.

A further weakness was the very incomplete picture of mixing processes both near the surface and at higher altitudes. It was not possible to say whether mixing from higher levels to the surface might still have been important in spite of the evidence that there was surface production of particles. It could not be deduced that the measurements of CCN and IFN near the surface could be safely used to deduce cloud properties.

The Atmospheric Programme on the Arctic Ocean 2001

In spite of the advances made in the past, many questions concerning aerosol formation and growth, their interactions with clouds and mixing processes in the boundary layer remain unanswered. Eight key questions are listed in table 1. The answer to question 8 was a crucial part of our overall objectives but will only be possible when answers to all the other questions have been obtained.

The decision of the Swedish Polar Research Secretariat to have a further expedition in 2001 to the same region and to allow a greatly increased multi-disciplinary atmospheric programme made it possible to tackle these questions. The programme comprises three sub-programmes: marine ecology, gas/aerosol chemistry/aerosol physics, and meteorology.

The work onboard and on the ice

To give the best possible answers to the eight key questions the following basic needs were identified:

• A. to provide a complete description of the microbiology of the water and ice, the nutrients present, productivity and sedimenting material
• B. to determine the properties of the surface microlayer and its probable role in influencing the nature of particles produced by bubble bursting
• C. to make shipboard measurements of the chemistry and physical properties of the aerosol, trace gases, cloud-active particles, radioactive tracers and electrical conductivity
• D. to have a lengthy and as continuous as possible series of measurements of mixing processes in the atmosphere from the surface to above cloud top
• E. to assess horizontal homogeneity of the near-surface atmosphere
• F. to obtain as many vertical profiles as possible of aerosols in various size ranges and of trace gases

The main part of these requirements was best performed at a distance from the ship. The ship therefore moored to an ice sheet at about latitude 89°N and drifted with the ice for a period of about three weeks. During this time intensive measurements or samplings were carried out for activity (A, B, D, E, and F) at a distance from the ship, while almost continuous measurements (C and D) were made on board.

Microbiological studies (A)

Ship based

Water samples were obtained with a CTD rosette that was lowered to an approximate depth of 200 m. On its way back up to the surface, water bottles attached to it were closed. This allowed water samples to be collected at 10 different depths. Sub-samples were analysed for many different parameters to help answer question 1. In the upper water column we measured DMS concentration, production and consumption, nutrient, organic compound and suspended biomass concentrations, and bacterio- phyto- and micro-zooplankton abundance. Primary and bacterial production was measured at 5–7 depths using the ¹⁴C method. Meso-zooplankton were sampled vertically with a net and their fecal pellet production determined. The amount of fecal pellets and the flux of particles out of the surface layers, as measured using sediment traps, helped to determine how much biological activity is occurring in the ocean.

Sampling of sea ice

Ice cores were taken to find how many and what kind of microorganisms was present within the ice, particularly in brine pockets within the ice. DMS precursors were also measured in ice and brine. This was in an attempt to seek answers to question 1. In addition spectral properties of the ice and melt ponds were measured and chemical analysis of melted ice performed.

The surface microlayer (B)

On-ice activities

Two radio-controlled boats patrolled open leads. They were preceded by Teflon rollers which collected the surface film (just a few millionths of a metre). It was scraped off and collected in bottles for further treatment and analysis. Analysis of bacterial abundance and production, as well as of other particles and gases (DMS and precursers) within it can help determine rates of production and air-sea exchange (question 1).

To examine the capacity of the open leads to emit particles to the air (question 1), due to bubble bursting at the water surfaces, a so-called eddy correlation technique was applied at two locations on the ice. One system was mounted in the central mast (see section D) and the other on an arm stretching out over an open lead at a height of 0.5 m.

Shipboard measurements of gases and aerosol particles (C)

The sampling of gases and particles was housed in three laboratory containers (one of them connected to a pump house located at the port side) on the 4^th deck. Gases and particles must be sampled with a minimum interference from the ship and from the sea/ice surface surrounding the ship. For that purpose an inlet and air sampling system was set up on top of one of the laboratory containers, facing in the forward direction so as to maximise both distance from sea level and from the superstructure of the ship. The inlet system consisted of two masts extending at an angle of 45° to about three metres above the container roof, so that the height of the inlet system was about 25 metres above sea level. Additional air sampling lines ran from this location to the other two laboratories on the starboard side. Direct contamination from the ship was excluded to a large extent by a particle pollution alert system with a wind sector controller. To further ensure the use of unperturbed data, a set of instruments was brought along to reveal possible pollution episodes by either local or far distant sources. The radioactive tracer gases ^220/222Rn, ⁷Be and ²¹²Pb and combustion gases as carbon dioxide, acetonitrile and benzene were measured. Radon (Rn) and lead (Pb) is used to give an indication of the time since the air last contacted land while beryllium (Be) will be used to give  a measure on mixing from higher levels and therefore be intimately linked to question 7.

Measurements of the gases that might participate in photochemical reactions either to form new particles or to influence particle growth included automated techniques with time resolution typically better than 15 minutes. The gases measured were e.g. DMS, SO₂, NH₃ and ozone and measurements were primarily directed towards answering question 2–4, but because of their marine biological origin DMS and NH₃ also relate to question 1. Also linked to question 1 are the detection of acetaldehyde and acetone, in that they could indicate possible degradation of organic material originating from the surface film of the open leads.

An array of cascade particle impactors and filters was used to collect particles in different size ranges to study the size-segregated chemistry of the aerosol. Ion chromatography, PIXE (proton induced X-ray fluorescence) and techniques to determine the content of organics such as proteins and amino acids were used for analysis. Particles were collected directly onto transmission electron microscope grids for subsequent analysis by electron microscopy.

One of the most valuable data sets from both previous expeditions was a record at 10 minute intervals of particle number size distribution from about 6 nm to 900 nm diameter. A similar system was provided for this expedition, this time also including a ”pulse height analyser” for looking at the smallest sized particles smaller than 5 nm with the addition of an aerodynamic particle sizer for extending the range of measurement beyond 900 nm. Conductivity measurements were made because it had been suggested that bubble production at the surface of the open leads could be accompanied by ion production.

One way of estimating the particles that might be active in cloud formation is to measure the CCN concentration. This is done by comparing particle concentrations emerging from an isothermal chamber and an identical one operated at supersaturation, in which CCN are precipitated. CCN concentrations were also deduced from a tandem differential mobility analyzer system, TDMA. Here, the extent to which a selected size of particle grew when the relative humidity was raised from 10% to a higher value (90% for example) allowed their cloud-forming ability to be estimated. IFN were measured by collecting all particles on filters, then subjecting them to a temperature of -15°C and water saturation and counting the ice crystals that grew. To study the interaction of cloud drops and aerosols we continuously monitored fog drop size distributions up to 47 µm diameter using a forward scattering spectrometer (FSSP-100) probe with a 10 min resolution.

Answering questions 2, 3 and 4 was the main application but the potential charge and organic content of individual particles related to question 1 as well.

Mixing processes in the atmosphere (D)

On ice activities

To study mixing processes closest to the surface an 18 m high meteorological mast was erected several hundred metres from the ship. This is because such types of measurements are sensitive to the flow distortion around the ship hull. The high frequency measurements (Hz) allowed mean vertical profiles at five levels of wind speed and direction, temperature, humidity and at two levels of turbulence (mixing) and turbulent fluxes to be measured. To study the radiation balance at the surface the temperature in the ice at several levels down to 1 m depth was monitored at the same site.

In addition measurements of high frequency fluctuations in wind and temperature were made from an instrumented package (a Tethered Lifting System (TLS)) carried on an aerodynamic kite (at wind speeds > 5 ms^-1) or tethered (Kevlar line) blimp-like balloon (at wind speeds < 5 ms^-1) to heights as great as 2 km. The TLS was hosted in a simple probe attached to the tether ”line”. To measure high frequency fluctuations in the horizontal wind and in temperature, a newly developed platform including a sonic anemometer and a motion sensing pack was similarly deployed. This system was developed specifically for this expedition. In addition the system also hosted a size resolved aerosol probe (more details below).

Furthermore an array of microbarographs was deployed in a triangular pattern on the ice to detect gravity waves (wave number and phase speed) which previous experience had suggested to be sometimes related to the sudden changes observed in gas and particle concentration through their influence on atmospheric mixing.

Ship based

Continuous information from greater heights was obtained using several shipboard remote sensing devices. An acoustic sounder (sodar) produced profiles of wind speed and direction and detected regions of turbulence. It sent out audible pulses and received return signals from small turbulent motions in the atmosphere. During the ice drift the sodar operated from the ice to minimize sound disturbances. A scanning microwave radiometer and a wind-profiler provided further vertical information on temperature and wind (in addition to that coming from the sodar).

The wind, temperature and humidity measurements were augmented by rawinsonde releases from the helicopter deck of the ship at six hour intervals during the ice drift operation (at 0600, 1200, 1800 and 2400 UTC). The ship sent the data from these launches to the Swedish National Weather Service in order to improve the weather forecasts sent back to the ship.

The provision for such an extensive meteorological operation, including a ship attached to an ice floe near the North Pole, was a novelty. No comparable programme has been attempted north of latitude 75°N. The application of these methods was mainly in answering questions 5, 6 and 7.

Horizontal homogenity of the atmosphere (E)

An important question in interpreting changes in gas or particle concentrations is whether the changes were travelling with the wind or resulted from downward mixing of air from higher levels. Two portable automatic meteorological (PAM) stations reporting data to the ship by radio link were deployed by helicopter at distances up to 8 km from the ship. Some horizontal traverses of particle concentrations in several size ranges were also made by helicopter.

Vertical profiling of aerosols and gases (F)

In addition to the vertical profiles of meteorological parameters, the ship’s helicopter carried three aerosol instruments providing particle concentrations in sizes >3 nm, >20 nm and >300 nm up to a few micrometres in diameter sizes.  The concentration difference between the first and second of these allowed an estimate of relatively newly-formed particle concentrations, while the third would be useful if large particles carried the nucleating material with them, as had been speculated from the 1996 expedition results. Temperature and humidity sensors were also used. A further novel instrument capable of detecting a wide range of atmospheric trace gases such as acetone, acetaldehyde, acetonitrile and benzene was also carried on the helicopter. Many of these gases were exclusively of continental origin and their concentrations relative to those over the continents in conjunction with their atmospheric lifetimes could provide information on the extent to which Arctic air had been influenced by transport from continents.

The helicopter was unable to fly during fog or low cloud. The TLS-kite and balloon systems could, however, operate under these conditions and were a successful complement to the helicopter platform directed towards answering questions 2–7. A small particle counter (particles >20 nm) and several size resolved aerosol screens were used to obtain vertical particle profiles similar to those gained with the helicopter. In addition a profile of trace gases was obtained to about 100 m by using these lifting systems to raise an air intake tube to the gas detection instrument used on the helicopter.

Preliminary results

A large data set was obtained which will require an immense effort to post-process and analyze and relate to the multitude of problems that we set out to study and it will probably take some years to reveal its scientific content and to make full use of the information. Below follow some impressions from observations made during the cruise and also some remarks on the logistics of the operations.

The meteorological sub-programme

Comparison of the conditions encountered during the three expeditions to the central Arctic

The extent to which measurements in any one season give a typical representation of the properties of the Arctic atmosphere is obviously an important question in applying the results obtained from these expeditions. There were several obvious differences both in the ice cover encountered and in weather.

The lower troposphere was continuously very moist and conditions were overcast almost all the time. The boundary layer was typically a few hundred metres deep and often relatively well mixed. As a consequence, motion systems expected in a moist and stable stratified boundary layer capped by a relatively dry free atmosphere, e.g. low-level jets and Kelvin-Helholtz instabilities, were much more rare than in previous expeditions.

Instead the meteorology was far more dominated by synoptic scale weather systems. We saw much more snow and even an appreciable amount of rain than on the other two voyages. The 1996 expedition also had far more sunshine than the other two expeditions.

The surface temperatures remained semi-constant close to the melting temperature of the ice. There were far more ”melt pools” (small blue fresh water ponds on the ice surface) during the 2001 expedition than in the other two. There was also much more sediment (even some tree trunks) on the ice at latitudes 86–88°N and longitudes 60–150°E than on the earlier expeditions. There were much larger open leads in the eastern longitudes in 1996 than in the other two expeditions but the ice was closely packed west of longitude 0°.

Comprehensive meteorological data from the Arctic basin as far north as during this expedition are very rare. If we had had the very wide range of measurements made in 2001 on the earlier expeditions it might be possible to decide whether these differences will seriously affect the applicability of our conclusions. The past lack of microbiological information and the relatively simple meteorological measurements made will make this assessment difficult.

Logistics

The meteorological measurements, in particular those made on the ice during the ice drift, were a logistic challenge. In conclusion, most systems worked as expected and the logistical challenges were met successfully.

• The central mast was erected and supplied with power from the ship (backed up by batteries on the site) as planned, although adverse weather delayed deployment by about 2 days. Once erected, almost all systems worked more or less without problem throughout the operation.
• The same is true for the sodar system, also powered from the ship with battery backup.
• Tethered soundings were to be carried out from the helipad of Oden. Unfortunately this could not be realized due mostly to the very turbulent wake behind the superstructure of the ship. All soundings with the TLS were thus made from the ice, mostly during the ice drift. The system, once deployed on the ice, worked nicely.
• The deployment time of the PAM-systems was cut short by roughly a week compared to the original plan (~30% of the planned time). This was due to adverse weather prohibiting the use of the helicopter. However, through working with the helicopter crew in setting up the sling-load arrangements, once the stations could be flown out the actual deployment time was record short – only about 1.5-2 hours, compared to the six hours planned for.
• Remote sensing instruments onboard the ship worked mostly without problems.
• A microbarograph array was deployed and worked marginally, due to adverse weather and ice conditions.
• Rawindsoundings launches were handled expertly by the Polar crew.

The atmospheric chemistry and physics sub-programme

The record of particle size distributions often showed how very complicated the changes can be during any day in the high Arctic. There were frequent events of brief fog lasting for a few hours that coincide with episodes of small particle formation (nucleation). Also present on many days was a full sequence of incursions of particles smaller than 20 nm and a major production of particles larger than about 100 nm occurred near local noon. This was present on many other days as well. We will have to see if any of these events can be related to mixing of air from above, to horizontal air motions (from patches of o pen water for example) or to changes in microbiological activity on the lead surface microlayer.

At a first glance the vertical helicopter profiles of the smallest particles shows strong indications of a source of the newly formed particles confined to the surface. Their plume type horizontal appearance suggests the nucleating material to be derived from the surface film on the open leads. An important step towards confirming the deductions made during Arctic Ocean 1996 appears to been taken.

The marine ecology sub-programme

Phytoplankton was present during the entire drift, but low in abundance, mostly in the upper 20 m of the water column.  The concentration and production of DMS and precursor compounds were low but detectable during most of the drift and heterogeneous in space and time, this apparent variability should still be considered preliminary as some samples have yet to be analysed. However, as seen in previous expeditions, concentrations of DMS and precursors were significantly higher in the marginal ice edge zone waters than in pack ice waters.

Concluding remarks

It is remarkable that in spite of often severe weather and icing due to supercooled fogs, almost all items of equipment were serviceable for nearly all of the available time and every possible opportunity for microbiological sampling was taken. As a result there is an extremely large pool of data. Relating it all will be a formidable task that will probably take years but it should result in firm conclusions about the sources and nature of the Arctic aerosol and their environmental significance.

When we compared the conditions encountered during the three expeditions to the remote high Arctic there were several large differences both in the ice cover encountered and in weather. This identifies a future need to obtain longer records of the parameters studied before we can assess if the natural aerosol and CCN production and transformation processes are important for climate change, and if so, whether they will constitute a positive or negative climate feedback.