Chemical Exchange Between the Atmosphere and Polar Snow

Ice cores have now become established as one of the primary archives of paleoclimatic information, covering timescales from seasonal up to 100,000 years or more. In the polar ice sheets, where there is little or no melting, snow layers build up year by year. Included in them are samples of the atmosphere: trace gases in air bubbles, particles and adsorbed gases, and the water molecules themselves. By drilling into the ice at suitable places, we can collect cores that give profiles of chemical content and physical properties of the ice. These are then used to infer the state of the atmosphere in the past.

In spite of their relative remoteness, the atmosphere of the poloar region is already disturbed by human activities. The development of the so called ″ozone hole″ over Antarctica in spring and the Arctic haze in winter represent good examples of the fragility of these remote atmospheres. One interesting characteristic of these high latitute regions lies in solid precipitation which has been accumulating over polar ice caps for the last several hundred thousands years. Assuming a sufficient knowledge of the relation lingking the composition of the snow and that of the atmosphere at the deposition, ice cores extented from Grrenland and Antractica offer a unique possibility to reconstruct the chemical composition of our pre-industrial atmosphere. They can also help to investigate the variability of our natural atmosphere over time periods as long as several thousands of years.

Since the start of the ice core drilling some 30 years ago, a whole zoo of species and parameter have been measured and reported in the literature. Those records per se are of minor relevance. Of great interest is the link to the corresponding atmospheric concentrations, which are related to basic parameter of the Earth System.

Dispersion of a gravitationally neutral tracer in the lower 50 km of the atmosphere is generally governed by the macroscopic motion of the air itself. Molecular diffusion can be disregarded except for laminar layers of air with thickness on the order of a few centimetres close to rigid surfaces. The macroscopic perception of fluid flows are based on the concept that fluid parcels can be regarded as thermodynamically homogeneous. The total mass of each parcel is assumed to be constant; i.e. no molecular diffusion of total mass is permitted, and a unique velocity of the air-parcels can be defined as the derivative of parcel positions with respect to time.

The chemical composition of the troposphere (0 to ~8 km above msl) in the Arctic is distinctly different than it is in the Antarctic. As pointed out in previous reviews (Barrie, 1986; Barrie et al, 1992; Barrie, 1993; Barrie, 1995), the Arctic is surrounded by populated continents from which pollution is released to the atmosphere and is transported readily to the north. In contrast, the Antarctic region is entirely surrounded by the southern Pacific ocean and is remote from human activities. Thus, it comes as no surprise that the tropospheric concentration of many anthropogenic aerosols and gases is much higher in the Arctic than in the Antarctic. What may be less obvious is that atmospheric trace constitutents of natural origin are found to have a different chemical climatology in the Arctic than in the Antarctic. Substances derived from sea spray, wind blown dust, marine biogenic activity and volcanoes generally have different seasonal variations and concentrations in the Arctic compared to the Antarctic.

Aerosol samples were collected continuously in two size fractions from June 1993 to June 1994 at a site 6 km north of the GRIP camp at Summit in central Greenland. The battery-driven sampling equipment was designed for automatic operation and low energy consumption. The samples were analyzed for elements using Particle Induced X-ray Emission (PIXE). The elements Al, Si, S, Cl, K, Ca, Ti, Mn, Fe, Zn, Br, Sr and Pb were measured well above the detection limits in many samples. The contents of most of these elements can be attributed to three sources, a crustal, a sea salt and an anthropogenic source. Additional contributions from other sources seem to be of importance for S and especially for Br. The crustal source had a peak incidence in the spring, while the anthropogenic source was present all year, strongest in winter and spring. The sea salt signal was very weak and difficult to resolve from the data, but was strongest in the winter. The results from Summit are compared with published data from Dye 3 on the southern part of the Greenland Ice Sheet, and with measurements from Station Nord, a sea level site in northeastern Greenland.

Central Antarctica, by virtue of its remoteness and the lack of local anthropogenic pollution sources, can probably be considered the cleanest location on earth. It is a true pristine background environment in the most fundamental sense. Indeed, it may be the last true pristine background environment. Yet, airborne materials of anthropogenic origin are detected there, albeit at very small concentrations, thanks to long-range transport and mixing processes in the atmosphere. The Antarctic troposphere plays an important role in global climate because it links the global atmosphere and oceans with the Antarctic ice sheet and stratosphere. Tropospheric transport processes occur on fairly short time scales, of the order of days to weeks; whereas stratospheric processes (e.g., interhemispheric exchange processes) may occur on the order of years.

Snow fields of the coastal Antarctic region are situated intermediately between the vast environmental regimes of the high antarctic plateau and that of the South Polar Ocean. Since they are directly exposed to moist maritime air masses, a relatively high snow accumulation rate, associated with the deposition of mainly marine derived aerosol species, is expected here. These characteristics make this region unique to recover high resolution ice core records that reveal the environmental and climatic history of the South Polar Ocean. Indeed, several extensive ice core studies have been or are currently performed in coastal antarctic regions (see Figure 1). To “calibrate”, in particular, their glacio-chemical information in terms of the corresponding atmospheric changes, representative long term records of the coastal antarctic aerosol chemistry (including the relevant precursor gases) are needed.

The chemistry of the atmosphere is driven by solar radiation which can give rise to photodissociation. The amount of radiation reaching the troposphere is attenuated by scattering and absorption processes in the upper layers of the atmosphere. In polar regions, the path to the troposphere is longer than at other parts of the globe, and hence less radiation is available for photochemical processes. In fact, during the winter months the lower parts of the polar troposphere are entirely excluded from receiving dired solar radiation. In addition, temperature regimes are different. In the winter they drop to levels of -25 to -50°C in the Arctic and lower in the Antarctic. Consequently the absolute humidity is very low. Finally, due to the low level of incoming solar radiation and loss of energy due to infrared cooling at the surface the lower troposphere is stratified and vertical mixing is hindered. These natural differences between the polar troposphere and other regions of the globe cause changes in chemical reactions that take place.

There are two principal aims of this chapter. (1) To present a framework for the use of field observations, including transfer coefficients among snow-ice-atmosphcre, in defining current atmospheric chemical processes. (2) To present a framework for interpreting ice core concentrations of key spccies in the past with a focus on the deduction of pre-industrial global hydroxyl (OH) radical concentration. The highly reactive hydroxyl radical determines the lifetime of major C, N, S and halogen species. These frameworks define sets of measurements that must he made to constrain atmospheric models for interpreting oxidant (0₃₊ OH₊ and H₂0₂) and other trace gas levels in present-day and past atmospheres.

This contribution presents a review of recent field experiments investigating the relationship between the composition of snow falling onto polar glaciers and the composition of aerosols in the overlying atmosphere. The limited data exisiting prior to the late 1980s indicated that aerosol removal processes should cause fractionation between the composition of aerosols and snow. Uncertainties regarding the relative importance of ice-nucleation scavenging, in-cloud riming of snow flakes and dry deposition over polar ice sheets precluded assessment of the impact the likely fractionation would have on efforts to reconstruct temporal variations in aerosol chemistry from ice core chemistry records. Two large international experiments on the Greenland ice sheet after 1988 focused on these, and other, issues central to understanding air-snow exchange processes. These experiments confirmed that ice-nucleation scavenging is the major process incorporating aerosol-associated species into polar snow. This process enriches snow in large aerosols relative to the aerosol population aloft. Dry deposition was found to be of minor importance at the present time, but also enriches the snow in large aerosols and the species associated with them. Since the large aerosols over Greenland are predominantly derived from sea-salt and dust, while SO₄=. NH₄+ and several pollutant trace metals are concentrated in submicron aerosols, snow chemistry presents a biased view of aerosol composition. However, it would appear to be possible to incorporate such a bias into efforts to reconstruct aerosol chemistry records from ice cores, were it not for several complicating factors. Spatial variability in snow chemistry will likely impose an inherent limit on the temporal resolution that will be possible in such reconstructions.

The glacial record has proven to be a most valuable source of information about climate change, major geologic events, and variations in global cycling of chemical species (Barnola et al., 1987; Dansgaard et al., 1989; Mayewski et al., 1990; Taylor et al., 1993). This information has been obtained by noting variations in chemical concentrations and in ice properties with depth in ice sheets worldwide (Delmas and Legrand, 1989). Virtually all of the chemical constituents found in the world’s glaciers, aside from those near bedrock, originally came from the atmosphere. Yet published studies of ice core data have seldom incorporated information about the atmosphere or air-to-snow transfer into their interpretations.

The analysis of hydrogen peroxide and formaldehyde in ice cores offers the potential to reconstruct the oxidation capacity of the atmosphere in the past [Neftel and Fuhrer, 1993], since these species are closely linked to the odd H and the odd O budget of the atmosphere [Thompson, 1992]. These species are so-called reversibly deposited on the snow surface since they show a strong exchange with the gas phase after deposition and their ice concentrations are dominated to a large extent by postdepositional changes in the firn. Therefore, the transfer from the atmosphere into the snow is an important step in the overall transfer function of these species [Neftel, this issue]. To get a better understanding of this step several field experiments were carried out during the 1993 and 1994 summer seasons at Summit in Central Greenland, in the 1993/94 Austral summer in Antarctica along the Swedish traverse (SWEDARP), and at South Pole in November and December, 1994. We will give an overview of the data and point out possible contradictions and knowledge gaps.

Polar firn and ice contain the most detailed, long-term, complete record of past climatic conditions on the Earth. Gas enclosed in bubbles and released from ice when it is brought to atmospheric pressure are samples of ancient atmospheres. Impurities in the ice matrix are related to atmospheric composition at or shortly after the time the snow that forms the ice was deposited. Climate reconstruction from polar ice cores offers the only possible indication of historical concentrations for many atmospheric chemical species that are critical to understanding how the Earth’s atmosphere responds to changes in emissions of the various carbon, nitrogen and sulfur species. These historical records provide the data against which to evaluate chemical/climate models that are designed to simulate how the Earth’s atmosphere will respond to changing patterns of anthropogenic emissions.

Uptake of gases onto ice and reactions of species on the ice surface are discussed in the context of ice-cores. Because of a lack of information that is directly applicable to ice cores, occurrence of certain processes are inferred from the results of lower temperature studies carried out to understand polar stratospheric clouds (PSCs). It is suggested that strong acids will be efficiently taken up. Easily-hydrolyzable species such as N ₂ O ₅ and CIONO ₂ will be converted to HNO ₃ on ice. The extent of absorption of weak acids and other species such as O ₃ will depend on the heat of absorption and the gas phase partial pressure of the gas. The reactions of a few other species on ice at warmer temperatures of the ice sheets are proposed to be similar to those for PSCs.

Over the last decade there has been a growing interest in the chemical composition of the snow packs in the polar regions (Bales and Wolff, 1995). Delmas (Delmas, 1992; Delmas, 1994) has noted that “information recorded in polar ice cores over the last several hundred millennia is invaluable to studies aimed at understanding the pre-industrial environmental system and anticipating the future evolution of the climate and the atmosphere.” For example, the isotopie composition of the ice (H₂0) matrix is a reliable paleothermometer. From the analysis of deep Antarctic and Greenland ice cores the ice age environmental conditions appeared to correspond to about 6 °C cooler temperatures and atmospheric CO₂ and CH₄ levels lower by factors of nearly 2 and 4, respectively. The biogeochemical cycles of S and N also appear to be affected by climatic changes that result in modifications in the source intensity and the transport of gaseous precursors. Even though atmospheric sulfate is derived principally from marine biogenic sources (i.e., dimethyl sulfide emission), cataclysmic volcanic eruptions can contribute sporadically to the atmospheric sulfur budget through large point source emissions of SO₂. These events are ultimately detected in polar ice as H₂SO₄ spikes. Nitrate, which is the next most abundant anion found in polar snowfall, exhibits concentration changes that are poorly understood, but which could be linked with the polar ozone hole formation. In addition to ions derived primarily from gas-to-particle conversions,

Energy, mass transfer and grain recrystallization processes show the most dynamic variation in the upper few meters of polar firn. Our understanding of processes in dry snow and polar firn shows feedback mechanisms exist between microstructure changes and energy and mass transfer coefficients. Differences between the predominant processes in the short polar summer and the winter cause sharp textural discontinuities in the stratigraphic columns at many polar locations. This distinctive layering forms the basis of studies on accumulation rates, layer ages and other investigations pertinent to firn and ice core analysis. Processes forming the stratification of the firn are not completely understood, nor are the associated loading patterns of chemical species. Past research on energy and mass transfer in near-surface polar firn is briefly surveyed. Current research focuses on processes controlled by the geometry of the ice and pore phases in the snow and firn. An overview is given of some of this work.

Ice cores contain a suite of chemical tracers which, when correctly interpreted, can reveal properties of the paleo-atmosphere and the climate system. These chemical constituents have been subjected to a series of physical and chemical processes at each step on their journey from the remote paleo-atmosphere into the ice (see Neftel, this volume; Waddington, this volume). Current research in polar snow chemistry focuses on the transfer functions, i.e. the chemical and physical processes that control the concentrations and fluxes of these atmospheric components at each step. Understanding the transfer functions is essential to the ice core - paleoclimate Inverse Problem, i.e. the derivation of the paleo-climate from the ice core data (Waddington, this volume). In this paper we focus mainly on one step in the Forward Problem; this is the transfer of dry-deposited chemical tracers from the local atmosphere into the snow by air flow through the snow (windpumping). Nearly two decades ago in a seminal paper entitled The filtering effect of snow, Gjessing (1977) summarized this topic as follows:

In their hallmark work on the Antarctic blowing snow phenomenon, Budd et al. (1966) recounted a quote of Peary (1898) that still has relevance to interpretation of glacial phenomena:

“There is one thing of special interest to the Glacialist — the transport of snow on the ice-cap by the wind”

Following Peary’s reasoning, the blowing snow phenomenon was studied intensively in the Antarctic several decades ago (Lister, 1960; Mellor and Radok, 1960; Dingle and Radok, 1961; Budd et al, 1966) with more recent studies by Kobayashi (1978), Takahashi (1985) and Moore et al. (1994). Other work of similar nature has been conducted in the Arctic (Dyunin, 1959; Benson, 1982; Tabier et al, 1990b; Benson and Sturm, 1993). Blowing snow is quite frequent in the Antarctic, Dalrymple (1966) noted snow transport occurred from 30–55% of the time in the South Pole region, increasing to 55–65% of the time at Byrd Station. Fujii (1981) found snow accumulation to occur in only one or two out of every three years in East Antarctica because of wind erosion and sublimation. At Mzuho Station, East Antarctica, Takahashi et al. (1994) estimated that of an annual snowfall of 140–260 mm water equivalent, 100 rnm/year is eroded from surface snow and transported away by blowing snow and 50 mm/year sublimates.

Vertical profiles of wind speed, temperature and humidity were used to estimate the roughness lengths for momentum (z.₀), heat (z._H) and moisture (z._Q) over Antarctic surfaces. The vertical profile-measurements were performed on and near a blue ice field in Queen Maud Land, East Antarctica. The roughness lengths are generally used to compute the fluxes of momentum, heat and moisture from standard meteorological observations using the flux-profile relationships. The value of z₀ is easily evaluated from the wind speed profile. The scalar roughness lengths are evaluated using a new method, which circumvents the difficult measurement of the surface temperature. It is found that the vertical exchange processes are strongly influenced by the saltating snow particles close to the surface during snow-drift events. The scalar roughness lengths seem to be approximately equal to z.₀ for a large range of roughness Reynolds numbers, despite the frequent occurrence of drifting snow. It is suggested that snow-drift processes are important for the turbulent transport of scalar quantities such as heat and water vapour, and presumably also for the transport of other atmospheric constituents.

The question of gas adsorption or absorption at the ice-air interface as function of temperature is central to understanding how and to what extent ice cores preserve a representative record of atmospheric chemistry. Gases can either adsorb at the two-dimensional interface between air and the ”surface region” on ice, partition to the three-dimensional surface region, partition into the bulk ice, or some combination of these. Ice has a ”rough” surface near 00C, and as it approaches 00K should become perfectly faceted. At some intermediate temperature much of the rough surface, which has some liquid-like properties, becomes small. Acidic gases such as SO₂ [Conklin et al., 1993] and CO₂ [Ocampo and Klinger, 1982] show greater uptake with increasing temperature, whereas non-reactive species such as NO [Sommerfeld et al., 1992] show the opposite trend; H₂O₂ shows both trends depending on the temperature range. There is currently no method of predicting this behavior based on known properties of gases and our limited knowledge of the nature of the air-ice interfacial region. It is thought that the increase observed with acidic gases is due to trapping of solute molecules at the surface by mobile water molecules (i.e. a surface liquid-like layer) as the temperature approaches the melting point. If the ice surface is curved, as in the case of snow, the freezing point is also a function of grain curvature. The addition of impurities can allow liquid layers to be present in a snowpack at temperatures below 00C.

Different components and properties of the ice deposits on the earth (glaciers, ice sheets) store information on the climate and environment. We can mainly distinguish between three storage types: a) properties and isotopie composition of the ice itself, b) solid (and liquid) trace substances in the ice, and c) gaseous and volatile components. This article reviews some aspects of the transfer of gaseous components from the atmosphere to the ice.

The aim of drilling ice cores is to obtain paleoatmospheric information of wide significance. However, the concentrations of chemicals found in the- ice are determined both by the atmospheric concentrations, and by depositional and post-depositional processes. Other papers in this volume discuss the depositional processes, ami the processes that subsequently alter the concentrations in near-surface (up to a metre or so depth) snow. However, there arc a number of documented cases where further changes occur below the surf act snow. Additionally, there is the possibility that chemical changes can occur, at least to more complex species, over tlx: long timescales that arc relevant for ice. Diffusion is bound to take place to some extent, affecting the apparent rate of temporal change inferred from ice core profiles. Discussion of all the factors requires an understanding of the way in which impurities are held in the ice This paper discusses all these items, conccntrating on processes in solid ice. and in firm below the top metre or two.

The processes by which chemical species in the atmosphere become incorporated in the firn depend both upon the nature of the forcing from the atmosphere and upon the properties of the firn itself. These processes include both diffusion and advection (the transport of heat, vapor, and chemical species by air flow within the snow and firn). In this paper we present recent field measurements of firn properties relevant to the transport processes, and use simplified model calculations to investigate the possibility of advection at Summit.

The analysis of ice cores from polar ice caps has the potential to yield information on the evolution of atmospheric composition during the past 400,000 years (Legrand, 1995). Before such information is obtained, however, we need to understand the relationship linking atmospheric composition to snow composition, i.e. the transfer function. For gases that do not interact with ice, such as CH₄, the transfer function is simple, and the CH₄ mixing ratio in air bubbles trapped in ice is the same as in the atmosphere (Chappellaz et al, 1990). Gases such as HCl and HNO₃, however, interact strongly with ice, and occluded gases contribute to the chloride or nitrate content of ice in a negligible manner. These compounds are then mainly included in the ice volume, but their mechanism of incorporation is not understood. HCl is an indicator of sea salt fractionation. Its mixing ratio yields information on atmospheric transport and acidity. HNO₃ is the final stage of oxidation of NO_y species. Its determination is useful to understand the NOy budget and the oxidizing capacity of the atmosphere.

The global distribution of atmospheric constituents (wether major or minor species, gas or aerosol, or even energy and momentum) is determined by the distribution of the sources, sinks, and conversion within the atmosphere, and by transport and mixing by the atmospheric circulation. General circulation models (GCMs) simulate, on a global scale, the transport and mixing of air, heat, moisture and momentum. All numerical formulations for these processes should be universal, but of course they are not, or otherwise one would never have to use the terms “parameterization” and “tuning”. However, if a process parameterization is physically-based, it should be usable for all species equally affected by this process. Implementing transport and mixing of atmospheric species within a GCM should thus be fairly straightforward.

Records from polar ice sheets contain invaluable information about past variations in the climate and the composition of the atmosphere. Past temperatures and precipitation rates can be deduced from stable water isotope records. Soluble (ion) and insoluble (dust) compounds forming the atmospheric aerosol are scavenged by the precipitation from the air mass above the ice sheet, or dry-deposited on the ice sheet surface, and become incorporated in the ice. Concentration records of ions and dust from ice cores can reveal the past aerosol composition and concentration in the atmosphere provided that the air-snow transfer functions are known. These are, at present, poorly known and the topic of this workshop. However, the atmospheric aerosol has a high spatial variability due to short residence times in the atmosphere of most compounds forming the aerosol. This complicates any attempt to extrapolate information from polar ice cores to airborne concentrations on a global scale.

There are several incentives for improving our understanding of the role of snow and ice as determinants of the behaviour of organic chemicals in the environment. Snow scavenges organic chemicals from the atmosphere both by adsorbing gaseous chemical to the ice surface and by scavenging aerosol particles with their associated chemical. Snow undoubtedly influences soil-air transfer of chemicals, and ice probably prevents water-air transfer in lakes, rivers and oceans. The period of snow melt may cause a pulse in organic chemical loading to receiving waters and to groundwater. This is obviously of particular importance in arctic and sub-arctic regions. Finally, it is possible that the glacial record of concentration may contain valuable information about chemical concentrations in past atmospheres, similarly to that of lead. Unlike lead, organic chemicals have an appreciable volatility. It is thus likely that when they are present in snowpack they are subject to evaporation back to the atmosphere. This complicates their behaviour and the interpretation of glacial records.

Measurements of the multielemental composition of size-fractionated aerosols provide information on the source processes of the various elements and they allow one to examine the physico-chemical transformation processes that take place during the atmospheric transport or as a result of local meteorology and atmospheric conditions. Furthermore, such measurements provide the necessary input data for estimating the scavenging probabilities and dry deposition velocities of the elements. Several studies on elemental mass size distributions have already been performed in the Arctic, e.g., by Pacyna et al. [1984], Li and Winchester [1990], Hillamo et al. [1993] and Barrie et al. [1994], but virtually all former studies were done in the winter and were limited to time periods of a few months. To investigate changes in sources, source processes and/or particle size modification processes over the course of the year, in late 1990 we started long-term (and still ongoing) samplings with a cascade impactor at the Zeppelin mountain station in Spitsbergen. Selected preliminary results of these samplings are presented and briefly discussed.

Ice cores are commonly used to reconstruct the chemical composition of historical deposition, and there is great interest to link the concentrations of various trace components in the kc with their concentrations in the atmosphere at the time of deposition Therefore it is important to understand the atmosphere/snow transfer mechanisms for the different spccies. In mixed phase clouds, i.e. clouds containing icc crystals, it is assumed that supercooled cloud droplets arc first fonrjed. After ice nucleation, ice crystals grow by water vapour deposition at the expense of the supercooled cloud droplets (Wegener-Bcrgcron-Fintleiswi mechanism) and/or by direct accretion of cloud droplets (called rinung) onto the crystal’s surface. The first mechanism is often the dominant physical process but insignificant wiih respect to the removal or transfer of pollutants resulting in very dean ke crystals. The latter mechanism determines to a great extent the chemical composition of the kc crystals in polluting them with activated aerosol panicles. However, the degree of riming depends on a variety of parameters, such as cloud droplet size, snow crystal size and snow crystal settling velocity.

In recent years the study of how atmospheric aerosols and gases are deposited in the snow pack of the interiors of the Greenland and Antarctic ice sheets has become a growing field, exemplified by the proceedings of this volume and the proceedings of the NATO ARW in Annecy in 1993 (Delmas, 1995). If the functions of transfer of various aerosols and gases from the atmosphere to snow/ice strata were known, then ice core data could be used to calculate past atmospheric compositions.

As snow accumulates on ice sheets it is slowly compressed and sintered, ultimately forming solid ice, typically at depths of between 50 and 100 m. Above this depth, in the unsolidified firn, the interstitial air is still in connection with the surface atmosphere. Molecular diffusion, however, retards the attainment of equilibrium at the base of the firn with any changes in surface concentration (Raynaud et al. (1993); Schwander et al. (1993)). As a result, concentrations at depth in the firn may reflect atmospheric conditions years, or even decades, earlier.

Hydrogen peroxide is a major trace constituent of polar ice cores, and is the only archived species so far detected that reflects atmospheric oxidation capacity (Neftel & Führer 1993). As has been noted by Neftel (this volume) and Bales & Choi (this volume) our ability to interpret this archived record is limited by our lack of understanding of the atmosphere to snow transfer function. The Law Dome site, and in particular the Dome Summit South (DSS) ice core, offers distinct features which may well shed light on the difficult issue of the atmosphere/ice relationship. Firstly, the high accumulation rate (0.7 m/yr ice equivalent) leads to well preserved seasonal cycles which, in conjunction with atmospheric measurements or models, should provide useful constraints on exchange processes. Secondly, the climatology of the site (mean temperature −20.8°C) leads to a near total absence of summer melt, which is known to interfere with both measured peroxide levels and diffusion processes in the firn. Thirdly, and related to the first point, there is a significant gradient in accumulation across Law Dome. This means that the timescale of firnification differs significantly between the DSS core and other sites which sample the same air mass (e.g. DE08, 16 km from DSS, with approximately twice the accumulation rate). Consequently if post-industrial changes in H₂O₂ levels are detected (Sigg & Neftel 1991), these can be distinguished from firnification effects.

Significant progress has been made recently (e.g. Bales and Choi, this volume; Neftel this volume) toward understanding the transfer functions that describe how stable isotopes, chemical species, and other impurities get from the polar atmosphere into deep polar ice. This research is motivated by the reverse process, i.e., a need to derive quantitative paleo-climate information from ice core data. Even when the transfer functions from atmosphere to ice are known, the Inverse Problem,i.e. inference of concentrations in the paleo-atmosphere from concentrations in ice cores, may be neither straightforward nor simple. The need to solve Inverse Problems in many different fields has lead to the creation of Inverse Theory. Inverse Methods (a) provide a conceptual framework in which to understand the limitations common to all efforts that use data to infer properties of physical systems, (b) determine how much information can be obtained from the data, and (c) assess the reliability of that information. The ability of Inverse Methods to assess the Resolving Power of the data (i.e. to determine how well the Inverse Problem can be solved) is a key feature of great value to the ice core Inverse Problem, Inverse Theory takes many forms; this paper outlines a few concepts to show how Inverse Methods can provide a useful focus for ice core - atmosphere paleoclimate studies.

This group focused on aerosol species that deposit irreversibly on an ice sheet. Irreversibly depositing species include both primary aerosols emitted directly from sources and also secondary aerosols formed in the atmosphere from precursor gases. These species can be conveniently divided into several categories: 1.Soil-derived inorganic aerosols, including those containing Al, Ti, Si, Ca, Fe and several other elements. Al, Ti and Si are considered to be the best tracers of soil dust, although some of the Si in the polar regions may originate from coal combustion.2.Seasalt inorganic aerosols, including those containing Na, CI and other elements. Na is considered to be one of the better tracers for this category. CI is prone to chemical reactions and subsequent losses, and a fraction of the CI in the polar regions may be from anthropogenic sources.3.Anthropogenic inorganic aerosols, including those containing S, Pb, V, In, Zn, As, Cd and several other elements. The seven elements listed are all suitable tracers under many conditions.4.Sulphate in older snow, and in Antarctica, derives mainly from marine biogenic emissions of DMS (producing also MSA), and from volcanic emissions.5.Organic aerosols, including natural organics such as biogenically emitted aerosols as well as anthropogenic organics such as combustion products. However, not all organics can be considered irreversible. The group considered three steps in the overall transfer of chemical species from source regions to their final resting place in the deep ice. The first step involves long-range atmospheric transport to the air over the ice sheet. This is followed by the second step in which chemical species reach the ice sheet surface. Finally, the third step involves physical and chemical changes that occur as the chemical species become buried by successive snowfall, and the snow changes to firn and ultimately to ice.

Nitric acid studies may help to evaluate natural (eg lighting, stratospheric N₂O oxidation, terrestrial biogenic) and anthropogenic contributions to the NO_x budget in remote regions. For instance, the role of the stratosphere in contributing via the denitrification process to the nitric acid burden of the polar boundary layer, if confirmed, could be useful to investigate minimum temperatures of the lower stratosphere in the past. In polar ecosystems nitrate is the limiting nutrient. In addition, nitrate paleodata are useful in the study of the carbon cycle because of the interactions between the C and N cycles.

The goal of this working group was to determine what data and analyses are needed to develop a transfer function between atmospheric concentrations and snow/ice concentrations of H₂0₂ and HCHO. The proposed suggestions are directed towards the interpretation of ice cores collected in Greenland and Antarctica.