The highest concentration and greatest seasonal amplitudes of atmospheric CO2 and CH4 occur at 60°–70°N, outside the 30°–60°N band where the main sources of anthropogenic CO2 and CH4 are located, indicating that the northern environment is a source of these gases. Based on the author’s onshore and offshore arctic experimental results and literature data, an attempt was made to identify the main northern sources and sinks for atmospheric CH4 and CO2. The CH4 efflux from limnic environments in the north plays a significant role in the CH4 regional budget, whereas the role of the adjacent arctic adjacent seas in regional CH4 emission is small. This agrees with the aircraft data, which show a 10%–15% increase of CH4 over land when aircraft fly southward from the Arctic Basin. Offshore permafrost might add some CH4 into the atmosphere, although the preliminary data are not sufficient to estimate the effect. Evolution of the northern lakes might be considered as an important component of the climatic system. All-season data obtained in the delta system of the Lena River and typical northern lakes show that the freshwaters are supersaturated by CO2 with a drastic increase in the CO2 value during wintertime. The arctic and antarctic CO2 data presented here may be used to develop understanding of the processes controlling CO2 flux in the polar seas. It is shown that Arctic seas are a sink for atmospheric CO2, though supersaturation by CO2 is obtained in areas influenced by riverine output and in coastal sites. The pCO2 difference between the surface of the Southern Ocean and atmosphere observed in the austral autumn shows that the area east of 7°W might be considered a source of CO2 into the atmosphere, whereas the area west of 7°W is a net sink of CO2. This is corroborated with literature data that indicate an overestimate of the role of antarctic waters as a sink for atmospheric CO2.
Carbon dioxide is the most abundant and most important greenhouse gas (other than water vapor) in the atmosphere. The maximum annual average concentration of atmospheric CO2 is located within 60°–80°N (Tucker et al. 1986; Conway et al. 1994), outside the 30°–60°N band where the main sources of industrial CO2 are located (Rotty 1983). Annually about 0.1 × 1015 g C of anthropogenic CO2 is emitted within the 60°–70°N latitudinal band. The equilibrium greenhouse warming associated with increase of CH4 and chemical feedback (through change of stratospheric water vapor) relative to warming during deglaciation is about 30% of the warning due to CO2 (Chappellaz et al. 1990). Currently, the rate of increase warming due to CH4 is about 38% of the CO2 warming effect (Craig and Chun 1982). The maximum of atmospheric CH4 is also located over the Arctic and subarctic; the value of CH4 over Greenland exceeds that over Antarctic by 8%–10% (Rassmussen and Khalil 1984; Steele et al. 1987; Reeburgh and Crill 1996). Data from the present international network of atmospheric CO2 monitoring sites, located almost exclusively in oceanic areas, cannot be used to resolve longitudinal gradients. Thus identification of the important source–sink areas is difficult (Tans et al. 1990; Conway et al. 1994; IGAC 1994), especially in the Arctic, where only three North American stations for CO2 monitoring have been established. At present the global carbon cycle cannot be balanced to better than about 25% of anthropogenic CO2 emissions. In order to bring the atmospheric budget CO2 and CH4 into balance to the desired closure, improved measurements of CO2 must be combined with measurements of CH4 and with improved oceanic and terrestrial measurements of both. Because the CO2 and CH4 interhemispheric gradients and seasonal amplitudes show that the northern environment is a major contributor to the Northern Hemisphere CO2 and CH4 maxima and seasonal variations (Rassmussen and Khalil 1984; Tucker et al. 1986;Cicerone and Oremland 1988; Nisbet 1989; Tans et al. 1990; Fung et al. 1991; Quay et al. 1991; Reeburgh et al. 1994), the role of Arctic seas and terrestrial ecosystems as sources and sinks of these greenhouse gases should be evaluated.
The interhemispheric gradient of atmospheric CO2 was ∼1 μatm in the 1960s and is ∼3.0 μatm now (Fung 1993). This change in the interhemispheric gradient might be determined by an increase of the Arctic and subarctic natural source of CO2 and/or an increase of a sink in the Southern Ocean. The recent data (1989/90–96), based on year-round measurements of CO2 efflux from the typical northern soil in Siberia, show that only wintertime soil respiration could provide a net source of CO2 into the atmosphere up to 1 Gt C-CO2 and more (Zimov et al. 1993; Zimov et al. 1996). This agrees with the evaluation of Ciais et al. (1995a), which shows that arctic ecosystems north of 65°N are a net source of CO2 of 1–1.2 Gt. The CO2 concentration that was measured beneath the snow in winter in Siberia was similar to that measured in arctic tundra in Alaska (Kelley et al. 1968; Coyne and Kelley 1974) and temperate alpine (Sommerfeld et al. 1993), suggesting also that the northern ecosystems could exhibit large winter CO2 effluxes. The winter CO2 emission from the northern soils would increase the average CO2 concentration in air over the Arctic by approximately 7–16 ppm in different years (Zimov et al. 1993). Such a value is close to 50%–100% of the CO2 seasonal amplitude observed in this area. It was found also that the range of CO2 flux change is in good agreement with the varied double peak or near double peak in atmospheric CO2 wintertime distribution (Semiletov et al. 1993) that was recorded by monitoring stations such as Alert, Mould Bay, Point Barrow, and Sable Island (Wong et al. 1984). Some measurements of summer CO2 flux to the atmosphere were made in the 1990s in northern Siberia (Zimov et al. 1993; Zimov et al. 1996) and in Alaska (Oechel et al. 1993). Oechel et al. (1993) estimate that arctic tundra might contribute 0.19 Gt C-CO2 during summer, though long-term data obtained in northeastern Siberia show a higher value of CO2 emission (Zimov et al. 1996). The conclusion of these works is that the large pools of carbon accumulated in the north have became vulnerable to decay, and carbon is released from soils mainly in CO2 form. It is important that the main input of CO2 into the atmosphere be determined by efflux in winter when respiration of biota adopted for cold soil environment is not balanced by photosynthesis. In spite of the small area of the tundra and northern taiga soils (∼7.3 × 106 km2 and 2.1 × 106 km2, respectively), their carbon contents are among the highest on earth and combined represent an inventory of 455 Gt (Post et al. 1982; Gorham 1991), about 60% of the total atmospheric CO2 burden. Because the large pool of soil carbon in the north becomes vulnerable to decay by an increase in the depth of the seasonal thaw, the carbon released as CO2 might further speed the warming. Based on recent data we can consider the northern soil as a main regional contributor in budget of atmospheric CO2, though the Arctic Ocean might also play a significant role (Kelley and Gosink 1988).
Because CO2 is the end product of CH4 oxidation in the atmosphere (Cicerone and Oremland 1988) we consider here a northern source of atmospheric CH4 (methane hydrate, organics buried in the permafrost, and natural gas) as a CO2 source also.
Anaerobic oxidation of organic matter in the seasonally thawed layer of onshore permafrost (Bartlett et al. 1992; Reeburgh et al. 1994) and peat bogs (Matthews and Fung 1987; Nisbet 1989) may be an important northern source of CH4. Different estimates show a large range of summertime efflux of CH4 from the northern environment. This uncertainty in the efflux of CH4 is related to inhomogenious time and spatial distributions of CH4 flux, and to limited experimental ground-level data (Reeburgh et al. 1994). In general, it was stated that CH4 might be produced in the northern ecosystems only during summertime (100 days), and total evaluation of the northern source of CH4 was based on this assumption (Matthews and Fung 1987; Fung et al. 1991; Reeburgh et al. 1994). However, the dramatic autumn surge in CH4 levels in the northern atmosphere (Steele et al. 1987; Quay et al. 1991) requires us to search for other regional sources of CH4. Early investigations of CH4 emission from a small lake in northwestern Ontario show that lake overturning might be a significant factor in the ultimate evasion of methane to the atmosphere (Rudd and Hamilton 1978).
Other arctic sources of CH4 include methane gas hydrates and natural gas losses, especially in Siberia. One of the largest potential sources of CH4 emission to the atmosphere is natural gas hydrates (Makagon 1982); the shelf and continental slope reservoir is estimated to be roughly 6 × 1018 g (or 6000 Gt), and the onshore permafrost reservoir is about 16 × 1015 to 32 × 1015 g. Usually, onshore CH4 is stable in hydrate in permafrost at depths of 100 m and more, and offshore CH4 hydrate is potentially stable at shallow levels in the seafloor, in water depths in excess of about 250–300 m. At present, the Alaskan records show a warming of permafrost in this century (Lachenbruch and Marshall 1986), and no experimental data indicate a temperature increase of onshore permafrost in Siberia and Canada. Temperature profiles in the offshore permafrost are episodic or limited in timescale (Fartishev 1993; Romanovskiy et al. 1997); that is, long-term representative temperature data are absent today. From an oceanographic point of view, the marine permafrost hydrate is stable at present, because usually the shelf bottom water has a temperature below 0°C (Treshnikov and Salnikov 1985). Thus, no experimental data indicate an unstable hydrate environment, but it would be vulnerable if the permafrost is warming.
A crude estimate of the natural gas losses to the air from gas processing and transport is about 1.5 × 1013 g of CH4 (about 2.5% of gas production) emitted annually in northern latitudes (Cicerone and Oremland 1988). This is probably not the principal cause of the marked seasonal behavior of in northern CH4; an annual midsummer minimum in CH4 is followed by rapid autumn rise (Steele et al. 1987; Quay et al. 1991). Analyses of ice core samples also suggest a strong net northern source of CH4 back to preindustrial times (Rasmussen and Khalil 1984; Chappellaz et al. 1993). The Tropics, too, are a strong source of atmospheric CH4, but there the production of OH, which destroys CH4, is also high. In contrast, the northern source is less balanced by sinks. The net effect is a buildup of CH4 in the Arctic with strong southward transport (Blake and Rowland 1988; Nisbet 1989). Both these processes should lead to low summer and autumn CH4 levels in the northern troposphere; the implication of the CH4 rise during the northern autumn is that there are seasonal sources of CH4 in the north. In contrast, Southern Hemisphere data show much smaller variations, implying that no comparable late summer–autumn source exists there (Fraser et al. 1984).
Here we present our results for the carbonate system study in the Arctic seas and in freshwater ecosystems of Siberia. The role of the northern lakes and offshore permafrost in formation of CH4 maximum in atmosphere is discussed. Based on our recent results and literature data we attempt to identify the main source and sinks for greenhouse gases that are reflected in the spatial and temporal variations of CO2 and CH4 concentration patterns in the atmosphere over the Arctic. Also, we present some Antarctic data that might be used to develop an understanding of the processes controlling CO2 flux in the polar seas. The estimation of CO2 and CH4 fluxes is not the primary concern of this paper; rather, it is to gain insight into the mechanism for formation of CH4 and CO2 maxima over the Arctic through identification of processes that produce the maxima.
Because the interaction between the atmosphere and surface water with respect to the direction of CO2 (CH4) transport can be evaluated in terms of the partial pressure gradient across the air–water interface, we present here the difference between surface water and atmosphere (ΔpCO2); (−) shows undersaturation of the surface water (sink), (+) shows supersaturation (source). Results from these studies may be used to partition the contribution from the sea and terrestrial biosphere as CO2 (and CH4) source or sink of the natural carbon cycle (IPCC 1992; Ciais et al. 1995a; Ciais et al. 1995b).
2. Study regions and measurement
Until very recently, marine chemistry studies in the polar seas were mainly performed in summer. Little or no constituent data were collected during or after autumn convection because of operational problems.
In fall 1994, the concurrent pH25 (at temperature 25°C) and total CO2 (TCO2) and dissolved CH4 were measured on the commercial icebreaker Amderma of the Far Eastern Shipping Company, Vladivostok, Russia. Amderma left Vladivostok in early September, crossed the northern Pacific and Bering Sea, and entered the hard ice field in the Western Chukchi Sea. Amderma reached its westernmost point at 162°E near the mouth of the Kolyma River. Surface water was measured at 23 locations. In late September 1994, the same measurements were made from the research and supply vessel Mikhail Somov, of the Arctic and Antarctic Research Institute (AARI), Saint Petersburg, Russia. Surface water was taken and measured from 30 points in the Laptevs Sea. TCO2 was measured by the stripping gas chromatography (GC) technique using catalysis conversion of CO2 to CH4 in a hydrogen stream (Weiss and Craig 1973; Weiss 1981; Semiletov 1992); CH4 and CO2 were separated on a chromatographic column packed by Poropac T (0.3 cm × 200 cm, 40–60 mesh) and installed in front of the Ni convertor (methanator). The precision of this procedure was about ±1%. Carbon dioxide pressure (pCO2, μatm) was computed from TCO2 and pH (NBS scale) according to a scheme presented in UNESCO (1987) and based on equations and constants advocated by UNESCO (1987). Specifically, the equations used were recommended by Millero (1978) for temperature and salinity dependence of the apparent dissociation constants for carbonic acid in seawater as determined by Meerbach et al. (1973). Measurements of pH were carried out immediately after sampling at 25° ± 0.1°C with precision ±0.01. Total uncertainty in pCO2 values was about ±10 μatm due to experimental errors in carbonate parameters as well in equilibrium constants. Comparison of our calculation with the recent scheme and constants advocated by Millero (1995) shows the results to be similar. The dissolved CH4 was measured by static headspace GC with using a flame ionization detector. The dissolved gas concentrations were calculated using the main static headspace equation (Vitenberg and Ioffe 1982): CL = CG(VG/VL + G), where CL and CG are the concentrations of a gas in the liquid phase (L) and in the equilibrated headspace (G), VG and VL are volumes in the gas (G) and liquid (L) phases of the closed headspace system, and G is a gas partitioning coefficient. The total precision of this technique was about ±1%–3%. The GC methods were presented and discussed in detail recently (Semiletov 1993;Semiletov et al. 1996a). The surveyed areas are presented in Fig. 1. The data obtained in fall 1994 are presented in Table 1 and Fig. 2a. In September 1995, we measured pH25, TCO2, and dissolved CH4 on board Mikhail Somov again (Fig. 1): 67 surface samples were taken in the Kara Sea and Laptevs Sea. The results are presented in Table 1. In September 1996, we measured pH25 (SWS scale) and total alkalinity (TA) on board Alpha Helix (HX 194), U.S. National Science Foundation, in the Chukchi Sea. The expedition was funded by the U.S. National Science Foundation. In this cruise we measured pH at 25° ± 0.1°C with ORION 8103 Ross electrode in SWS scale, using tris-buffer prepared by the prescription of Goyet and Dickson (DOE 1994). The precision was ±0.002 pH unit. Total alkalinity data were obtained by direct indicator titration in an open cell using a 665 Dosimat system with a precision of 0.1%. Comparison of TA determination by direct titration in open cell (Bruyevich’s method) and potentiometric titration in closed cell (Edmond’s method) shows differences within 1% (Rogachev et al. 1996). About 700 samples were analyzed in the Chukchi Sea east of 170°E to map pCO2 values computed from TA and pH from surface to bottom. Values of pCO2 were calculated given TA, pH, T, S, following a scheme and constants advocated by Millero (1995). Distribution of pCO2 in the surface water of the Chukchi Sea during autumn convection is presented in Fig. 2b. The experimental technique and preliminary data are described in Semiletov et al. (1998). In late November 1996, we took water samples from beneath the ice to the bottom in the mouth of the Lena River and adjacent part of the Laptevs Sea. Forty samples were taken and measured to map-dissolved CH4 (Fig. 3) and pCO2 values (Table 1). This work was done mainly to evaluate the efflux of CH4 from offshore permafrost in a shallow shelf.
2) Lakes and rivers
Surface water is a significant part of the landscape of the Arctic coastal plain, composing up to 50%–80% of the land area. Here we consider the typical northern lakes of Siberia. In general, the northern lakes are thermokarst or thaw by origin (Hopkins and Kidd 1988). Thaw lakes, lakes that result from surface collapse caused by the thawing of ice (rich permafrost), are important and conspicuous features of Arctic and subarctic lowland landscapes in both tundra and taiga regions. Thaw lake sediments are underlain by zones of thawed permafrost, called taliks. The depth of taliks increased with the age of thaw lakes and changed from 100 to 102 m (Chekhovskiy and Shamanova 1976). For instance, thaw lakes aged about a few thousand years might cover a layer of thawed permafrost with a depth of ∼100–200 m or more; that is, the vast organic reservoirs immobilized in permafrost became available for anaerobic destruction by way of a lake evolution. Note that thawing of the permafrost under lakes might be a mechanism that is able to involve methane buried in hydrate reservoir into the present biogeochemical cycling. In spring 1992, the study of carbon emission from the northern lakes was started at the lower Kolyma River, about 80–100 km from the east Siberian Sea of the Arctic Ocean. The study area is phytogeographically a transition of main northern landscapes (Zimov et al. 1993): forest–tundra; tall shrub community on floodplain; alpine tundra; southern Khalerchin tundra, which is a mosaic of typical tundra (sedge–dwarf-scrub polygonal mires) and southern tundra (low-shrub–sedge, tussock–dwarf-shrub mire) with patches of fall shrub vegetation; marshy thermokarst depressions; and grass–shrub floodplains. The main study was not very far from the timberline. Within this area containing a range of hills and mountains, ice complex (edoma), alases (depression with gentle slopes and flat bottom), floodplain, and a number of lakes of both alluvial and edomic origin are presented. The typical lakes were presented in a sample. This part of the joint study was based at the northeastern science station located near Chersky (Zelenny Mys); see Fig. 1.
In an effort to improve our understanding of northern aquatic ecosystems as a source of atmospheric CH4 and CO2, in fall 1994 we extended the study area from the Kolyma River lowland to the polar semidesert area near the Lena River mouth, where arctic lakes are in the most harsh environment of the high Arctic. The field surveys were conducted at the Bykovsky Peninsula, the delta of the Lena River, and the north slope of the Verkhoyansky Mountain Ridge (Primorsky Kryazh). Since fall 1994 the estuary of the Lena River has also been surveyed. This study is based at the Polar Geocosmophysical Observatory located near Tiksi, on the coast of the Laptevs Sea (Fig. 1). In summer (late August–early September 1995) the CO2 research was conducted along the Lena River from Yakutsk to the Laptevs Sea (Tiksi) (Table 2). The expedition’s tracks are presented in Fig. 1.
More then 1000 samples were taken during all seasons, though the main study was conducted between fall and early summer. The range of variability of pCO2 and CH4 in the surface waters of north Asia and the Arctic Sea is presented in Table 1. Because the dynamics of the carbonate system and dissolved methane is usually investigated during summertime (Kling et al. 1992; Reeburgh et al. 1994), we focused this paper on our data obtained in the time between autumn and spring. Note that arctic summer is different on land and sea; “land summer” months (no snow covering) are June, July, August, and early September; “sea summer” (main sea aquatory is free from ice) is from late July until late September. Note that it is not possible to obtain reliable oceanographic data in the real high Arctic between October and early July without the very expensive use of icebreakers or organization of drifted stations.
Information about the change in atmospheric CO2 and CH4 is obtained from the National Oceanic and Atmospheric Administration/CMDL, Commonwealth Scientific and Industrial Research Organisation, and other networks, and a number of World Meteorological Organization (GAW) and IGAC Activities (GLOCARB, BIBEX, TRADEX, HESS, etc.) that are summarized in Steele et al. (1987), Blake and Rowland (1988), Cicerone and Oremland (1988), Tans et al. (1990), Quay et al. (1991), Fung et al. (1991), Ciais et al. (1995a), Ciais et al. (1995b), Conway et al. (1994), Prinn (1994), Reeburgh and Crill (1996), and other literature.
The study area includes most northern landscapes, making it representative of the variety found in the Arctic.
b. Southern Ocean
The atmospheric gradient constrains the combined uptake by the Southern Ocean gyres and Antarctic waters to be from 0.6 to 1.4 Gt of C per year (Tans et al. 1990). An analysis of NOAA/CMDL Global Air Sampling Network shows that a large sink of CO2 must be between 30° and 60°S (Conway et al. 1994), though there are not enough oceanographic data to understand process of CO2 exchange in this region. According to Ciais et al. (1995b), the latitude band south of the Antarctic convergence might be a small sink of atmospheric CO2 (in 1992) or source of CO2 into the atmosphere (in 1993). To understand distributions of sources and sinks additional oceanographic data are needed. For this reason, we have investigated the distribution of pCO2 in the surface waters south of 40°S in the Atlantic and Indian sector of the Southern Ocean. Note that until the late 1980s only seven oceanographic stations collected carbonate data in the entire Indian Ocean south of 30°S (Chen 1988).
Concurrent measurements of pH25 (NBS scale) and TCO2 were made on board the research vessel Professor Vize of AARI during austral summer–autumn (February–March) 1989. After a stay in Montevideo, Uruguay, Professor Vize twice crossed the subantarctic convergence (SC), Antarctic convergence (AC), and Antarctic divergence (AD), and reached its southernmost point within the ice field at the Antarctic Circle near the Mirniy observatory in the Davis Sea. The experimental technique is the same as used in the Arctic research and described in Weiss and Craig (1973) and Semiletov et al. (1995). The study area of this expedition and difference of pCO2 between ocean and atmosphere are shown in Fig. 4. The calculated flux of CO2 (F) in the atmosphere–ocean system is presented in Fig. 5.
3. Results and discussion
1) Arctic seas
The Arctic Ocean is the most sensitive link in the climatic system; only a thin layer of freshwater beneath the ice prevents the ice cover from melting due to heat advection by warm water of the North Atlantic through the Fram Strait (Aagaard and Carmack 1989; Broecker 1997). The Arctic Ocean receives about ∼10% of the global river runoff but composes only ∼5% of the area and ∼1.5% of the volume of the global ocean. In fact, the combined freshwater inflow (∼3500 km3 yr−1) is determined by the great north Asian rivers, Yenisey (602 km3 yr−1), Lena (513 km3 yr−1), and Ob (451 km3 yr−1), and additional runoff (estimated at 1670 km3 yr−1, when referenced to salinity of 34.8 psu) enters indirectly through the Bering Strait (Antonov and Morozova 1957;Aagaard and Carmack 1989). Clearly, any attempt to understand the effect of the Arctic Ocean on global change or the effect of global change on the Arctic Ocean requires a thorough knowledge of the riverine influence on hydrochemistry of the arctic adjacent seas, especially on the carbonate system of water and dissolved and solid substances. Investigations for the MacKenzie and Yukon (whose discharge enters the Arctic via the Bering Strait inflow) are currently under development and ongoing data are available. Otherwise, the Siberian rivers, especially the Lena River, which contributes most of the freshwater input to the Amerasian basin, are poorly investigated.
(i) Inorganic carbon system
Here we present our recent results that were obtained mainly in the Laptevs Sea, which is influenced strongly by inflow of the Lena River (Fig. 2a). Also we briefly consider the CO2 data obtained in the Chukchi Sea (Fig. 2b).
The shallow Chukchi Sea with a typical depth of about 40 m is a controlling area for the fluxes of the freshwater, nutrients, and carbon into the Arctic from the North Pacific. Late September and October mark the transition time between summer and winter seasons in the Arctic. During this time, the magnitude of river discharge is decreased significantly, about 2- to 3-fold for the Yukon and Mackenzie Rivers (Coachman et al. 1975) and 10- to 20-fold for the Siberian Rivers (Antonov and Morozova 1957) and the regional thermohaline regime is switched from summer estuarine environment to a winter “reverse estuary” outflow of high-density saline waters (Aagaard and Carmack 1989). Distribution of ΔpCO2 (Fig. 2b) between the surface waters and atmosphere shows that the aquatory is strongly undersaturated by atmospheric CO2, which should result in absorption of CO2 from the atmosphere. Supersaturation by CO2 was found in a small area near Cape Barrow, which might be governed by coastal upwelling. The increase in the magnitude ΔpCO2 (Fig. 2b) from the Bering Strait toward the ice edge (∼72°N) is controlled mainly by temperature. Evaluation of the temperature effect on the carbonate system shows (Millero 1995) that the observed temperature decrease of the surface water ∼6°–7°C should cause a decrease in equilibrated value of pCO2 of about 100 μatm, which is similar to observations (Fig. 2b). Our results for the dynamics of the carbonate system in the Chukchi Sea are presented in detail elsewhere (Semiletov et al. 1998). The ΔpCO2 data presented in Fig. 2a show that the aquatory of the Laptevs Sea was also undersaturated strongly by CO2 relative to the atmosphere, though the area adjacent to the delta of the Lena River is supersaturated significantly by CO2. Our year-round study of the CO2 system in the Lena River–Laptevs Sea system shows that the riverine waters are a source of CO2 for the atmosphere during all seasons (Semiletov et al. 1996b). Likewise, a long-term hydrochemical investigation of AARI demonstrates that the spatial distribution of dissolved oxygen is in opposition to pCO2 variability obtained in our research (Rusanov et al. 1979). The high value of pCO2 agrees with the low concentration of dissolved oxygen. Measured concentrations of dissolved CO2 and oxygen are a result of the interaction between different physical and biological processes as cooling–warming, photosynthesis–respiration, and conservative mixing of different waters (Park et al. 1974; Skirrow 1975; Weiss et al. 1982; Bordovsky and Ivanenkov 1985; Chen 1988). In different seasons and latitudes physical or biological processes might dominate the distributions of pCO2 and dissolved oxygen in surface water.
Other data show that the sea aquatory influenced strongly by the riverine input is also two to three times or more supersaturated by CO2 (Table 2). For instance, the value of pCO2 near the sandbar of the Kolyma River, the east Siberian Sea was 716–1779 μatm in fall 1994 (Amderma cruise). Our experimental data obtained in fall 1995 on board R/V Mikhail Somov also shows that not far from the mouth of the Ob and Yenisey, the surface waters of the Kara Sea were higher in comparison with an aquatory moved offshore (Table 2). This is corroborated by the data of Kelley (1970), which were obtained in the Kara Sea during the late summer of 1967.
Existing data show that the surface pCO2 in the Arctic Sea varies mainly in the range 150–500 μatm. It is affected by changes in water temperature, photosynthesis of marine plants, CO2-rich river runoff, biochemical oxidation, coastal upwelling, upward divergence of deep water by cyclonic gyres, freezing processes, changes in depth of the surface mixed layer, and by changes in advection of water determined by variability in the atmospheric circulation. From Figs. 2a,b and Table 1, we can see that the whole aquatory of the Arctic Sea, even during and after convection and freeze-up, seems to be a sink of atmospheric CO2 with undersaturation up to 40%–60%. Supersaturation by CO2 is obtained only in the restricted areas near mouths of the Siberian rivers and in the coastal sites (Table 1), related mainly to upwelling of waters more enriched by CO2. Our recent data obtained in September 1998 and 1997 indicate that summertime coastal retreatment (from 2 to 40 m yr −1) might be an important source of terrestrial organics into the shelf waters. And oxidation of these organics could be a cause of supersaturation of bottom water by CO2 (up to 2000 μatm). We did not calculate the value of CO2 flux here because to date there are no representative data for the transfer velocity of CO2 during and after freeze-up time.
Our CO2 data agree well with literature data obtained in different seasons (Kelly 1970; Park et al. 1974; Codispoti et al. 1982; Chen 1993). Previous studies of ΔpCO2 between the atmosphere and the surface waters of the North Atlantic Ocean and the Barents and Kara Seas show that the aquatory of the Kara, Barents, and Norwegian Seas during the late summer of 1967 was undersaturated by CO2 regard to the atmosphere in range of ΔpCO2 from 100 to 150 μatm (Kelly 1970). Likewise, the surface water near the mouths of the Ob and Yenisey Rivers were supersaturated in CO2 with respect to air. Carbonate data obtained in September–October 1993 near the mouths of the Ob and Yenisei also indicate supersaturation by CO2 from 10 to 200 μatm (Makkaveev 1994). As the surface water cools in its flow northward, the CO2 pressure is decreased in a similar manner as observed for the Chukchi Sea in Fig. 2b. Because the uptake of CO2 by photosynthesis is high only in the delta waters of the Siberian rivers (in the Lena River, up to 0.1–0.3 g C per m−2 per day, integral value) and decreases 10–100 times toward the shelf edge (Sorokin and Sorokin 1996), we can assume that the general decrease in CO2 saturation from Arctic coast toward the north is determined mainly by cooling, as was stated previously by Kelley (1970) in agreement with our data. Recent Japanese pCO2 data obtained in the Greenland Sea (August 1993; April and May 1994) and in the Barents Sea (June 1995; July and August 1996), presented at the BASIS Conference (February 1998, Saint Petersburg) by H. Ito also indicate significant undersaturation of the surface layer by CO2 with a decrease of pCO2 value toward the ice edge (down to 200 μatm). It is probable that relatively high photosynthesis near the outlet of the riverine waters plays a negligible role in a local budget of atmospheric CO2, because during all seasons the delta area is a source of CO2 into the atmosphere. Likewise, the riverine input of inorganic carbon might play a significant role in the hydrochemical regime of halocline waters in the Arctic basin. For instance, our data (Semiletov et al. 1996b), corrected with the most recent data (September 1998), show that only the Lena River brings about 3–4 Tg C-CO2 in the subice water annually. Our recent measurements show that the Lena River brings also about 6 Tg C yr−1 in the form of dissolved organic matter that might be oxidized in CO2 form.
There are currently few observations for the Arctic and subarctic seas during ice-covered periods, which might be 10–11 months per year. Therefore we can only speculate about the net CO2 flux between atmosphere and arctic waters. Because young arctic ice is permeable for gas flux (Gosink et al. 1976) yet there are no representative experimental data for the transfer rate of CO2 across the ice cover, we attempt here to answer qualitatively the following question: is the Arctic Shelf a source or sink for atmospheric CO2? In general, the arctic measurements show that surface water near the wintertime marginal ice zone is undersaturated by CO2 because of cooling, and because biological respiration is negligible. Photosynthesis might decrease the pCO2 (and TCO2) value, because C-CO2 and C-HCO−3 (bicarbonate ion) are utilized during photosynthesis (Smith and Sashaug 1990). For this reason, unusually high oxygen concentrations (sometimes greater than 150% of saturation values) observed in near-freezing waters (Codispoti and Richards 1971) should indicate the existence of low values of pCO2. Our data and data in the literature show that cooling of water might be considered a process that decreases the pCO2 distribution in the polar seas. Weiss et al. (1982), based on observations, demonstrate that in the low latitudes and other oligotrophic zones temperature might greatly influence the distribution of pCO2 in the surface waters. In winter (February–March) 1983 Gosink (1983) found that the surface water of the Barents, Norwegian, and Greenland Seas were undersaturated in CO2 with respect to air, and ΔpCO2 varied from −20 to −50 μatm. The CO2 system data obtained by Chen (1985, 1993) in February–March indicate undersaturation near and beneath ice across the marginal ice zone of the central and southeastern Bering Sea Shelf, that is, the greatest sea shelf in the subarctic. The undersaturation was found from the surface to a depth of about 75 m. In this research the biological factor is negligible, and undersaturation of CO2 is due to cooling of the eastern shelf water. Carbonate data of Codispoti et al. (1982) obtained before and during the spring bloom in a similar area show that before the bloom the surface water of the eastern shelf is near equilibrium with air by CO2. Near the shelf edge the water was supersaturated by CO2, probably due to upwelling of the deep water (Pacific in origin) more enriched by nutrients and CO2 (Park et al. 1974). Thus, a limited dataset shows that the shelf waters of the Arctic and subarctic seas are mainly a sink of atmospheric CO2 during wintertime. Alhtough there are no available data for the CO2 system beneath ice in the Arctic seas located between 90°E and 170°W, we can consider an oxygen dataset obtained by AARI from the drifted ice stations NP1–NP21 during 1948–73 (Rusanov et al. 1979). The mean average data show that the subice water in the Amerasian sector (Canadian Basin) are supersaturated by dissolved oxygen up to 5%, though the aquatory, which is influenced strongly by riverine input, is enriched by organic and TCO2 and undersaturated by dissolved oxygen about 5%–10%. The rise in oxygen supersaturation beneath the ice might be governed by mixing with more warm water, not by biochemical processes (Chen 1988). The most recent research obtained on board icebreaker Des Grosielliers in winter 1997–98 shows that under the influence of global warming, the salinity in the upper 50-m water column in the Amerasian sector (between 74° and 77°N) decreased significantly in comparison with previous measurements (10 and 20 yr ago), and the water beneath the ice was supersaturated by oxygen during polar night when the biological factor is negligible (I. Melnikov 1998, personal communication). Using the anticorrelation in biogeochemical cycling of CO2 and O2 influenced strongly by temperature factor (Bordovsky and Ivanenkov 1985) we assume that the surface water enriched by oxygen within a stable temperature regime should be undersaturated by CO2. In this case, following Treshnikov and Salnikov (1985) we consider a temperature regime beneath the ice as a relatively stable in comparison with the open aquatory. Previously, Kelley (1970) compared the surface water oxygen concentrations to the pCO2 values and found mutual variations with temperature. Otherwise, we can assume that the shelf water influenced by riverine input should be supersaturated by CO2, because the riverine waters are enriched by dissolved organic carbon that oxidized to CO2. This agrees with our experimental data: during fall the open (fall 1995) and ice-covered (fall 1994) aquatory of the Laptevs Sea is a sink for atmospheric CO2, except the water adjacent to the Lena delta and some coastal sites (see Fig. 2a and Tables 1 and 2). Data for pCO2 distributions near the delta of the Ob and Yenisei Rivers also demonstrate supersaturation by CO2 (from +30 μatm to +140 μatm) in late summer and autumn (Kelley 1970; Makkaveev 1994).
Thus, the available wintertime data obtained during the late winter–spring bloom and early winter–fall convection and freeze-up show that the Arctic shelf waters are a sink rather than a source of CO2 into the atmosphere. Due to the lack of data for the Arctic Basin and kinetics for CO2 transfer across the ice cover, at present we cannot evaluate the overall role of the Arctic Ocean in regional CO2 cycling. Based on data in the literature (Kelley 1970; Park et al. 1974; Bordovsky and Ivanenkov 1985; Smith and Sakshaug 1990; Codispoti and Richards 1971) we can assume that during summertime aquatory of the Arctic Sea is mainly undersaturated by CO2 and supersaturated by O2 due to photosynthesis.
(ii) Dissolved methane
Our fall (1994, 1995, 1997) study of CH4 in the surface water of the Arctic Sea from 70°E to 170°W shows that the concentration of CH4 is usually less than 0.015 μM (μM = 10−6 moles per liter) (i.e., analytical zero in our GC measurement). Because of interference from the ship’s electrical system, it would not have been possible to detect lower concentrations of CH4 in the samples. Because equilibrated concentration of dissolved CH4 (Weisenburg and Guinasso 1979) in arctic air–water system ranges from 0.003 μM to 0.004 μM with temperatures of 10°C and −1°C, respectively (for constant salinity = 20‰), we can assume that uncertainty in our measurements is limited by possible supersaturation or undersaturation by a factor of 5 in summer and a factor of 4 in winter. In comparison with CH4 lake data presented in Table 1, which corresponded to a supersaturation factor range between 102 and 105, we consider the possible CH4 efflux from arctic seas as“small,” though this assumption is crude and needs additional data. Note that our assumption agrees with a 10%–15% increase of tropospheric CH4 over land (Kelley and Gosink 1988; Harriss et al. 1992).
To evaluate the role of shallow offshore permafrost in late September 1994 we have sampled water above the subwater ice complex (near the Bykovskiy peninsula) enriched by organics. The CH4 concentration was detected at the level 0.030–0.050 μM (Semiletov et al. 1996b), which is similar to the range of CH4 measurements in Canadian Beaufort Shelf waters (MacDonald and Thomas 1991).
The recent (November 1996) wintertime measurement of CH4 in the shallow waters above the offshore permafrost near the Lena River Delta (Bykovskiy peninsula) shows that the bottom sediment/thaw permafrost might be a source of CH4 in the atmosphere, because the subice water is enriched significantly by CH4 (up to 103 times) relative to the atmosphere. The vertical distribution of dissolved CH4 shows that the main maximum is located just beneath the ice, not in the bottom layer. The distribution of CH4 in the subice layer is presented in Fig. 3. Because the bottom is the source of CH4, it is evident that transfer of CH4 toward the surface is determined mainly by ebullition. The same vertical distribution with a subice maximum was obtained in the thaw lakes of the Kolyma lowland during or after a cyclonic system crossed the study area (Semiletov et al. 1996a). It is interesting that we did not find dissolved CH4 in such high concentration in summertime in the same place. We assume that the following factors might influence an increase in CH4 concentration beneath the ice. First, since the sea ice cover significantly decreases the transfer rate of any gas into the atmosphere (Gosink et al. 1976), we can consider the ice cover a barrier trapping the methane below. Second, the bacterial activity for methane oxidation is reduced strongly due to the drop of temperature to 0°C and less. Third, it might be important that summertime warming reaches the greatest depths of the active layer during wintertime; that is, the temperature of the deep layer of the bottom sediment and thaw permafrost is increased during fall and early winter. Increase of the sediment temperature dominates in production of CH4 (Reeburgh et al. 1994), which was detected in the northern lakes in early wintertime (Zimov et al. 1997b). Note that the range of variability of the CH4 concentration in the thermokarst lakes of the Bykovsky peninsula (onshore permafrost) during wintertime 1994/95 (Semiletov et al. 1996a; Semiletov et al. 1996b) is similar to the range of CH4 variability obtained over shallow offshore permafrost (Fig. 3) that is the same by origin (Pleistocene ice complex) with onshore permafrost. An anomalous high value of CH4 is observed at point 17 (Fig. 3), where CH4 concentration beneath ice is 20 μM. We explain this by an episodic increase in CH4 ebullition from this site due to a previous drop in the atmospheric pressure, which might induce the CH4 ebullition (Semiletov et al. 1996a). Such a correlation was obtained previously in the temperate latitudes at Mirror Lake, New Hampshire (Mattson and Likens 1990).
2) Arctic lakes and rivers
Usually, levels of dissolved methane in the northern lakes and rivers were measured during summer (Bartlett et al. 1992; Kling et al. 1992; Reeburgh et al. 1994). It was found that in summer the Alaskan arctic lakes are a source of CH4 for the atmosphere that is similar in magnitude to the CH4 source from wet soils. Our all-season study for dissolved CH4 dynamics and efflux into the atmosphere shows that the northern lakes play a significant role in the regional CH4 budget during all seasons (Semiletov et al. 1994; Semiletov et al. 1996a;Semiletov et al. 1996b).
Here we discuss mainly the data for the pCO2 and CH4 distributions obtained from fall until springtime, because the behavior of both gases in the summer season is described in detail in the literature (Bartlett et al. 1992; Kling et al. 1992; Reeburgh et al. 1994). Our wintertime data show the vast range of CH4 values in the subice layer of water, from 4 to 360 μM in the Kolyma River lowland. These values of CH4 are significantly higher in comparison with the summertime Alaska data (North Slope region). The corresponding diffusive fluxes into the atmosphere might be significantly higher, although the ice covering might resist gas transfer from the water body to the atmosphere, and wintertime efflux of CH4 (and CO2) is going through“koshkas” (see below) or ice trenches and ice overflow water (Semiletov et al. 1994). The data obtained in the Kolyma lowland were taken mainly in the area adjacent to the timberline. Most of the data are representative for the vast region of the northern taiga, woodland tundra, and low Arctic tundra. To obtain the data in the high Arctic we did the same study in the most severe conditions of the high Arctic tundra, in the Tiksi area (see Fig. 1). We found that in fall the content of CH4 in the thaw lakes located in the high Arctic ranged from 0.06 to 0.4 μM in the subice layer and from 0.07 to 2.2 μM in the bottom layer of lakes. These values are about 102 lower than in the Kolyma lowland (Table 2). It is probable that the lower production of CH4 is related to the difference in the thermic regime and environment for formation of the ice covering. Also, strong winds are typical for the coastal arctic zone and usually snow is blown up from the ice surface, that is, the thermoisolation effect from the snow covering is absent here, whereas the snow thickness usually ranged between 0.2 and 0.4 m on the lakes of the Kolyma lowland. Therefore, at the Tiksi area the temperature in the water body drops to 0.1°–0.3°C in late October when the thickness of the ice is about 0.4 m, whereas in the Kolyma lowland these values vary ∼1.0°–2.4°C and ∼0.2 m, respectively. In late wintertime the ice thickness at the Tiksi area reaches 2.2–2.3 m, whereas at the Chersky area only 1.3–1.5 m. The CH4 production decreases in the upper layer of limnic sediment in the Tiksi area. Likewise, the thaw zone underneath the water body of the lakes is shallower here [about 15–30 m, by Kunitskiy (1989)] than in the more mild arctic–subarctic environment (Chekhovskiy and Shamanova 1976), such as the Kolyma lowland near and south of the timberline where through taliks might exist. For this reason, in the high Arctic CH4 is produced mainly in the younger surface sediments. Indeed, CH4 sampled from the North Slope region of Alaska was only 200 years old (Martens et al. 1992). In contrast, in two lakes located near the treeline in the Kolyma Lowland about half of the current annual methangenesis is fueled by Pleistocene C (Zimov et al. 1997b), which is involved in the modern biochemical cycling through deep thaw of permafrost underneath the lakes.
In order to examine factors controlling methane and CO2 concentrations in various lake depths in different seasons, we have measured the vertical distribution of methane in different seasons in the typical northern lakes. The vertical profiles show that the dissolved methane concentrations were usually elevated drastically toward the bottom. Usually the CH4 concentrations in bottom water increased during the wintertime CH4 accumulation period from fall (October) to late spring (May). Furthermore, the bottom layer, which contained the highest CH4 concentration, became thicker during wintertime. In general, the dynamics of methane concentration profiles is corroborated by that in Canadian lakes (Rudd and Hamilton 1978). We found that the nitrogen content is near-equilibrated with air and relatively high oxygen concentrations were observed in the water column under ice. For example, until 18 January 1993 the oxygen concentrations dropped to the mean value about 2.5–3.0 ml L−1 in the thaw lakes with a typical depth of 3–10 m (the Kolyma lowland).
The springtime dynamics of dissolved CH4 and pCO2 show (Figs. 6a,b) that during the period of observations, the distribution of pCO2 between 1- and 6-m depth was quasi-uniform, whereas CH4 in the bottom water decreased with time. Note that dynamics of the CH4 profile might be considered to be controlled by partial upwelling of bottom water due to springtime overturning. It correlates with the existence of subice maximum of CH4 obtained on 29 May 1994 in the thaw lake near Chersky in the Kolyma lowland.
The methane profiles observed in the lake suggest two questions: what type of transfer provides the recorded CH4 distribution, and what amount of CH4 might escape to the atmosphere due to hydrodynamics?
The methane diffusive flux might be equal to or exceed the bubble flux in freshwater lakes (Cicerone and Oremland 1988). As was shown in Semiletov et al. (1996a), the formation of the near-bottom CH4 maxima might be determined by complete dissolution of small sediment bubbles with a radius of about 0.01 cm. Bartlett et al. (1992) found that in summer the surface CH4 concentrations differed significantly between large and small lakes. Methane concentrations in large lakes ranged from 0.01 to 0.31 μM. Concentrations in smaller lakes were higher and ranged from 0.21 to 10.4 μM. Calculated diffusive fluxes from open water varied with the lake’s size. The large lakes emitted 3.8 mg CH4 m−2 day−1, and small lakes emitted an average of 77 mg m−2 day−1. Thus summertime diffusive methane fluxes from the thaw small lakes and ponds are nearly 20 times higher than those from large thaw lakes and are approximately equal to one-half of fluxes from wet meadow sites.
Emissions from lakes by sporadic bubbling from the sediment are difficult to quantify, since measurement depends upon different factors, such as sedimentary environments, organics content, evolution of thaw lakes, and the amount and distribution of ground ice in the substrate, which depend upon local deposition and thermal history. To improve the study of factors controlling lake bubble composition we surveyed many different thaw lakes located from Dyvanny Yar to the mouth of the Kolyma River (Fig. 1). We sampled visible bubbles at the surface using a hand funnel. The bubble sizes were varied within 0.2–0.6 cm in diameter at water surface. Usually the methane content was no less than 43%–85% by volume of ebullient gases (Semiletov et al. 1996a). It was found that most intensive fluxes of methane-rich gas bubbles were obtained along the lake edge environment, particularly near steep banks of ice complex or “edoms” (the northern hills contained great quantities of ground ice, about 40%–80% of the volume of perennially frozen eolian sand, silt, and peat rich in organic matter).
The winter data obtained in the Kolyma lowland show the high CH4 concentrations in the surface layer under ice. The subice maximal CH4 value was obtained in January 1993, after drastic CH4 ebullition related to a low air pressure event. All sites, including shallow-water sites, show a remarkable synchrony in the bubble release correlated with changes in local air pressure. We assume that the subice CH4 maxima is associated with rising CH4 bubbles accumulated partly under the ice surface. It was observed visually through October when the ice surface was not covered completely by snow. In winter 1995/96 the subice maxima in the dissolved CH4 distribution induced by ebullition in the lakes of the southern Alaska was detected also by A. Phelps (1996, personal communication).
Figure 7 shows methane ebullition rates at the site in typical thaw lake with depth 10 m (Kolyma lowland) obtained during two weeks synchronously with air pressure (P) changes. During summertime the CH4 ebullition from shallower sites was larger, probably due to enhancement of CH4 production causing significant warming of the sediment surface. Low air pressure events associated with storm systems appeared to induce ebullition, whereas high pressure inhibited ebullition. Sometimes the CH4 ebullition attained value of tens of mM CH4 m−2 per day, when the air pressure dropped to 740 torr. The proportion of the methane in the bubbles was changed. The CH4 portion was increased simultaneously with the total gas ebullition (probably due to enhancement in bubble size and upward velocity). Thus, during the period of air pressure drop the net CH4 flux to the atmosphere is increased significantly. Such a correlation was obtained previously at Mirror Lake, New Hampshire (Mattson and Likens 1990). A similar phenomenon has been known to colliery ventilation engineers in the United Kingdom for more than 250 years (McQuaid and Mercer 1991). We found that the net CH4 emission to the atmosphere is provided mainly by the large bubbles that rise quickly and dissolve in the water column in small amounts. In this case methane oxidation in the water column is negligible. Likewise, the CH4 enrichment of a water body is provided mainly by the small bubbles that disappear when rising from the sediment (Semiletov et al. 1996). Note that low air pressure events increase ebullition in all seasons.
In fall 1992 and spring 1993 the main transport of CH4 through the ice covering in the Kolyma lowland was realized via the holes or koshkas (local name). These holes or koshkas are produced by relatively constant gas ebullition from each site. The size of the koshkas ranges from 100 to 102 cm in diameter (Semiletov et al. 1994), which is determined by gas ebullition rates from sediment. Usually a gas ebullition flux obtained at each site ranged from 5 to 10 cc min−1. Sometimes the sporadic large bubble ebullition ranged from 40 to 70 cc min−1. When the ice thickness is increased, a koshka begins to collect gas bubbles from a larger area like a funnel, because risen bubbles are rolled along the ice–water interface toward the koshka. For example, in mid-October, when mean ice thickness in the Chersky area (Kolyma lowland) is about 15–20 cm, the koshka of about 10 cm in diameter collected different bubbles injected from a distance of about 2–3 m; that is, one“typical” koshka collects risen bubbles from an area of about 12–30 m2. A positive feedback mechanism is at work here. Usually in the Kolyma lowland the largest koshkas exist until the coldest months (January–February). Periodically water above ice is formed when settling occurs due to an accumulation of snow on the ice covering. We assume that accumulated gas arrives in the atmosphere with water above ice. This event is found usually one or two times per winter. It should be noted that the ice settling is found usually close to the lake’s edge where the snow accumulation is more effective. Simultaneously, in the central parts of lakes the large-scale ice fracturing takes place and headspace gas under ice is liberated into the atmosphere. In the Tiksi area (coast of the Laptevs Sea) we found that the koshkas are closed in early or mid-October (1994), because strong winds blow snow from the ice surface, which increases the rate of freeze-up significantly. Also, in the Tiksi area in mid-October we observed synchronous large fracturing of ice covering in the different lakes due to supertension of ice during quick freeze-up. This is accompanied by the ventilation of the under-ice headspace, which also favors the CH4 and CO2 emission via diffusion transfer. However, our observations show that during the coldest months the number of active koshkas is decreased significantly, even in the relatively mild subarctic environment. Consequently, the total CH4 (CO2) flux to the atmosphere is decreased also. This correlates with the monitoring data for atmospheric CH4 (Quay et al. 1991).
In the Lena River delta the concentration of dissolved CH4 was usually too small to detect (<0.015 μM) during summertime, but during the ice cover season the value of CH4 varied from trace up to 0.1 μM in the Bykovsky Channel and Neelovsky Gulf (Fig. 3) and in the Olenek Channel (Semiletov et al. 1996b); at seaside near delta concentration of CH4 reached 20 μM (Fig. 3). In summer the riverine and lake water was usually supersaturated by CO2 two–three times; the supersaturation increased during winter significantly (up to 10–20 times). Our measurement shows that the Lena River water is supersaturated by CO2 from the middle stream (Yakutsk) downstream to the delta (Table 2). This result agrees with pCO2 data obtained in the deltas of the Ob and Yenisey Rivers in September–October 1993 (Makkaveev 1994).
3) Evaluation of CH4 emission from the Arctic lakes
Our direct measurement of CH4 ebullition shows that the CH4 ebullition rate might be increased drastically when atmospheric pressure drops (Fig. 7). Sometimes the CH4 ebullition achieved values of tens of mM CH4 per day near or south of the timberline. Is this CH4 emission significant for the regional CH4 budget or not? And how much CH4 might be emitted into the atmosphere from the northern lakes by other processes?
Taking into account that during polar night the supply of the hydroxyl radical is reduced dramatically in a comparatively closed arctic air mass (Vowinckel and Orvig 1970; Cicerone and Oremland 1988; Nisbet 1989), we assume that in winter the CH4 emission from lakes via holes and cracks in ice is a net source to the atmosphere. We consider here the arctic atmosphere as a comparatively closed arctic air mass, that is, a very simple approach, but the existence of the maxima of CH4 and CO2 over the Arctic demonstrates that the main sources of CH4 and CO2 are at northern areas rather than midlatitudes. Midlatitudes play only an episodic role in the increase of CH4 and CO2 through northward transport events (Worthy et al. 1994). The recent latitudinal and seasonal variations of the atmospheric methane show a very broad maximum in winter and minimum in midsummer (Fung et al. 1991; Quay et al. 1991; Reeburgh and Crill 1996). Thus, the maximum interhemispheric gradient is obtained during Northern Hemisphere winter. If we assume that during winter there is no gas exchange between the northern atmosphere and midlatitudes and methane is not destroyed in the troposphere, it is possible to estimate the magnitudes of the CH4 flux, which is adequate to explain the interhemispheric CH4 gradient. An air column of 1 m−2 with a CH4 concentration of 1.8 ppmv would contain ∼10 g CH4. Then approximately 0.8 g · m−2 of CH4 would arrive in the atmosphere for an observed increase of the CH4 concentration of ∼150 ppbv (Steele et al. 1987; Quay et al. 1991). Following Fung et al. (1991) we assume that the northern wetland area between 50° and 70°N is ∼2.65 × 1012 m2 and the total area of the tundra and northern taiga is about 9.4 × 1012 m2. Then the wetlands cover ∼25% of the total area north of 50°N. Taking into account that in winter the soil source of CH4 is absent, the limnic unfrozen sediment should produce more than 3.2 g CH4 m−2 to produce the observed interhemispheric gradient. Our current observations show that this value is not unreasonable. For instance, the cross movement of a cyclone with an air pressure drop to 740 torr increases the mean value of CH4 ebullition from background values ranging between ∼15–50 mg CH4 m−2 day−1 to 0.2–0.3 g CH4 m−2 day−1 (Semiletov et al. 1994; Semiletov et al. 1996a). Hence, the cross movement of cyclonic systems might play the role of triggering the CH4 efflux from lakes. Also, the CH4 emission due to the spring and fall overturn, when bottom waters enriched by CH4 rise to the surface might be superimposed on CH4 ebullition. For instance, if about 1 m3 of typical bottom water with CH4 concentration of ∼8 mg CH4 per liter reaches the surface, then ∼8 g CH4 m−2 might escape from the lake to the atmosphere. Here we assume that water overturning is a “quick” process, which precludes consumption of CH4 by the methane oxidants. Our initial observations show that such a quick process of overturning takes place in aquatory of lakes with typical depths less then 3–5 m. It was shown that large quantities of CO2 might escape into the atmosphere due to diffusity, especially during overturning. Indeed, in spring and fall surface waters of lakes are supersaturated significantly by CO2 and CH4 (Rudd and Hamilton 1978; Semiletov et al. 1996a; Semiletov et al. 1996b), in agreement with the timing of the rise in atmospheric methane over the Arctic (Steele et al. 1987; Quay et al. 1991). Concentrations of dissolved CH4 and CO2 obtained in the high Arctic (Tiksi area) are lower by about two orders of magnitude in comparison with low Arctic/subarctic (Cherskiy area) but are 10–100 times more than values equilibrated with air. Note that the range of pCO2 values of lakes located in the Tiksi area is similar to that near Barrow, Alaska (Coyne and Kelley 1974), and the North Slope region of Alaska (Kling et al. 1992).
Measurements of CH4 flux from the Kolyma lowland lakes (Zimov et al. 1997a) show that the annual flux is about 11 g CH4 m−2, of which 76% is released during ice cover season (October–May). This might be enough to maintain the interhemispheric gradient (see text above). The total winter efflux from Siberian lakes accounts for 2.5 Tg CH4 yr−1 (Zimov et al. 1997a). This is a lower evaluation, because it does not contain the CH4 emission produced by fall (and spring) overturning that was detected in atmosphere over Barrow, Alaska, as a strong signal (Quay et al. 1991). If we use the result of our simplistic evaluation of CH4 emission due to autumn lake overturning, the total atmospheric input from the Siberian lakes will be about 4–5 Tg CH4 per winter. This somewhat crude estimate should be corrected for the northward decrease in CH4 production by limnic sediment and underlying thaw permafrost once more data are obtained. At any rate, this crude estimate based on our year-round data shows that the northern lakes may play a significant role in formation of the atmospheric methane maximum over the Arctic. This assertion is supported by the space–time changes in the δ13C of atmospheric CH4 (Quay et al. 1991) that show a seasonal cycle with minimum δ13C values in fall and maximum values in the summer (increased income of“isotopically heavy” CH4 from the midlatitudes), with the strongest season amplitude in the north. A meridional gradient shows that the lowest values of δ13C in atmospheric CH4 are also in the north. Thus, the most biogenic and isotopically light methane is obtained in the north during fall–winter, indicating a northern source of biogenic CH4. The fall increase in atmospheric concentration of biogenic CH4 might be associated with the increase in the CH4 emission due to the lake’s overturn and a decrease in the meridional air transport of air masses enriched by nonbacterial methane from the midlatitudes.
Recent δ13C values of CH4 collected from two thaw lakes in the Kolyma lowland and measured at U.S. facilities were −70 to −72 in summer, and −65 to −80 in winter (Zimov et al. 1997b). This value is less than produced in summer by the Alaskan tundra (Martens et al. 1992) and less than detected over Barrow (Quay et al. 1991). It implies either that the Siberian CH4 was not as oxidized as in these other environments or that there was a difference in substrate or a different pathway of methanogenesis. It was shown also that the hydrogen isotopic composition of the CH4 was variable, but most samples from the northeastern Siberian were low (δD = −370), indicative of a biotic source for CH4, low oxidation rates in the water column, and CH4 production by fermentation (Zimov et al. 1997b).
Our initial results show that total wintertime (October–May) emission of CH4 from lakes into the atmosphere might be more then 16 g CH4 m−2 (about 8 g m−2 due to autumn overturn and 8.4 g m−2 due to CH4 ebullition), five times more than needed to form the CH4 atmospheric maxima over the Arctic, if the Arctic is treated as an isolated box and with negligible oxidation of atmospheric CH4 in winter.
4) Evaluation of the role of the northern lakes in the past
It has been shown that the concentration of methane in the Greenland ice core is about 10% (±4%) higher that in the Antarctic ice cores over the Holocene (Rasmussen and Khalil 1984). At present, the NOAA/CMDL network shows that the annually averaged methane concentration is highest at Barrow (71°N), which is about 8% above that of the South Pole (90°S) (Steele et al. 1987; Blake and Rowland 1988; Fung et al. 1991). Comparison of historical data obtained in the works of Jouzel et al. (1993) and Chappellaz et al. (1993) shows that the interhemispheric gradient is decreased from ∼10% in warm epochs to practically negligible value during glacial epochs. Then it might be assumed that during global warming the north is a source of CH4 to atmosphere (Semiletov et al. 1994). What is the significance of such global changes?
The joint analysis of the data obtained from the new Greenland Ice Core Project ice core (Jouzel et al. 1993) and measurements from Vostok (Antarctica) (Chappellaz et al. 1993; Semiletov et al. 1994) shows that during two main glacial–interglacial transitions, the CH4 concentration increased from ∼350–360 ppbv to ∼650–750 ppbv. Hence, it could imply that ∼0.8 Gt C-CH4 moved into the atmosphere during the glacial–interglacial transition. During glacial periods and “short time” cold events such as the Younger Dryas, the interhemispheric gradient between poles was practically absent; that is, the hypothetical northern source was blocked. When the climate was warmed the interhemispheric gradient increased to ∼10% in the optimums of interglacial periods and warm stages. Therefore, we can suggest that in warm epochs only limnic sediments and taliks underneath the northern lakes are able to maintain relatively high CH4 concentration in the north. We also can state that during winter, limnic sediment and talik are sole sources of CH4 into the atmosphere, because in winter the soil source of CH4 is dormant.
To estimate how much organic carbon in CH4 form might be available for anaerobic destruction by lake evolution we use here a simple calculation (Semiletov et al. 1996a). Because thaw lakes with ages of about a few thousands years might be underlain by a layer of thawed permafrost with a depth ∼100–200 m or more (Chekhovskiy and Shamanova 1976; Tomirdiaro 1980), the vast organic reservoirs immobilized in permafrost became available for anaerobic destruction by way of lake evolution. Concentrations of organics in the permafrost of north Siberia vary, usually in range of 0.5%–5% of C by weight [Zvyagintsev (1992)], which is similar to our measurement in the study area. Assuming that the permafrost density is about 2 × 103 kg m−3 and the content of organic C ∼0.5% by weight, the upper 100-m column of permafrost contains about 103 kg of organic carbon per square meter. Therefore, the upper layer of tundra and northern taiga (total area ∼9.4 × 106 km2) contains no less than 9400 Gt of organic carbon, which might became available for biotic activity via thaw lake evolution. Due to the existence of ancient thaw lakes, with ages of 3000–5000 years or more, carbon stock in the clathrate stability zone might be involved in the recent methane cycle, because such lakes might be underlain by talik extended through permafrost in the CH4 clathrate stability zone. Thaw lakes migrated across the north Siberian plains during the Holocene (Tomirdiaro 1980), releasing to the atmosphere a huge amount of CH4; half of this CH4 was derived from Pleistocene carbon: CH4 collected from thaw lakes in winter had an average 14C age of 27 Kyr (Zimov et al. 1997b). This age indicates that Pleistocene sediments deposited 20–40 Kyr ago contributed 68%–100% of the CH4 flux from these lakes. In contrast, CH4 emitted in the summer had an average 14C age of 9.2 Kyr, indicating that Pleistocene C fueled 23%–46% of summer methanogenesis and thus that more CH4 was produced in the younger surface sediments, which are warmer in summer than winter. Thus, about half of current annual methanogenesis is fueled by Pleistocene carbon. The current atmospheric CO2 and CH4 burdens are ∼750 Gt C-CO2 and ∼3.6 Gt C-CH4 (Quay et al. 1991). Then small changes in the current carbon stock of permafrost might affect significantly the growth of the main greenhouse gases in the arctic/subarctic atmosphere. The main end product of anaerobic destruction of organics in the subwater environment is methane; the atmosphere content is less than 0.5% of CO2. Therefore conversion of a small part of the carbon buried in permafrost might cause a large change in the growth of atmospheric CH4 in comparison with CO2, though a part of CH4 might be oxidized and converted in the form of CO2. Note that we consider a potential source of atmospheric CH4 as a source of CO2 also, because atmospheric CH4 is oxidized to CO2 during a few years (Cicerone and Oremland 1988).
Our study of CH4 in the surface Arctic waters shows that the adjacent Arctic seas are not a significant source of CH4, whereas the CH4 emission from the limnic environment play a significant role in the regional CH4 budget. It confirms the aircraft data of Kelley and Gosink (1988) and Harriss et al. (1992), which show a 10%–15% increase of CH4 over land. It is probable that similar processes were important during previous interglacial epochs, when the “pole-to-pole” gradient of CH4 was about 10% (Rasmussen and Khalil 1982; Chappellaz et al. 1990; Chappellaz et al. 1993), similar to the present gradient.
Due to lack of representative carbonate data, the World Ocean was considered for many years to act as a giant pump, removing atmospheric CO2 from the polar regions and liberating it in warmer latitudes, especially at equatorial divergence. This concept was based on a limited dataset for the meridional distribution of atmospheric CO2 (Bolin and Keelling 1963). This approach has been incorporated into the block-diffusion and open-diffusion models of the ventilation of deep waters in the high latitudes (Siegenthaler 1983). Such models assume that North Atlantic Deep Water (NADW) rise in the southern high latitudes to the surface, absorb excess CO2, undergo rapid mixing, and descend as a result of a nonlinear density increase during mixing. However, this hypothesis has not invariably been confirmed by sea measurements. For instance, the results of the DOWNWING and MONSOON expeditions in the Southern Ocean are contradictory (Skirrow 1975). Other measurements of pCO2 in the atmosphere and surface layer of the Southern Ocean indicate that this region is not a permanent sink of excess atmospheric CO2, at least during austral summer (Inoue and Sugimura 1986). Analysis of pCO2 data (1970–82) based on pH–TA technique and O2–pCO2 correlation technique by V. N. Ivanenkov, Yu. I. Lyakhin, and P. N. Makkaveev shows that net evasion of CO2 should be during the autumn and winter in the temperate and high latitudes in both the Northern and Southern Hemispheres (Bordovsky and Ivanenkov 1985; Bordovsky and Makkaveev 1991).
The autumn (austral) distribution of ΔpCO2 is presented in Fig. 4. We use an annual average value of pCO2 = 350 μatm in the atmosphere over the Southern Ocean in February–March 1989 that is based on the datasets used by Tans et al. (1990) and Erickson (1989). The resulting distribution of ΔpCO2 shows that the surface waters of the south Atlantic Ocean west of ∼7°W are undersaturated by CO2 relative to the atmosphere. It is correlated with the early experimental carbonate data discussed by Keeling (1968). The undersaturation by CO2 obtained in this area might be related to either physical or biological factors. The governing physical factor might be a cooling of the warm Brazil Current due to movement toward the Antarctic. Based on the recent experimental data of Millero (1995), we found that a recorded water temperature decrease about 1°C is enough to explain the obtained undersaturation, because a temperature coefficient provides a decrease of pCO2 by 4.3% per 1°C drop in the water temperature. It is difficult to estimate the influence of biological factors, because there is a lack of data for primary productivity in our cruise. We cannot use literature data, because satellite sensing of pigments has shown that significant large-scale variability exists in the Southern Ocean (Smith and Sakshaug 1990). The difference of pCO2 shows that the area east of 7°W might be considered as a source of CO2 into the atmosphere, because the surface waters are supersaturated by CO2. It is interesting that the supersaturation was obtained near the ice edge, though many researchers found anomalous high productivity near the ice edge (Bordovsky and Ivanenkov 1985; Smith and Sakshaug 1990). It should be noted here that any sea expedition obtains a dataset only for the restricted area during limited time; that is, the distribution of ΔpCO2 in the study area is a “photo” of the real situation. Indeed, we have found previously that mesoscale space–time variability of the CO2 system might change a photo significantly. For instance, our detailed investigation of the CO2 system in the megapoligon area located in the Pacific subarctic frontal zone shows that net invasion of CO2 predominates in June, whereas in August the reverse situation is found due to a change in the mesoscale circulation (Semiletov and Pipko 1991). Our investigation of the CO2 system during 3 days of drifting near the ice edge in the Davis Sea shows that the surface water was in a nearly equilibrated condition, with reversal from slight undersaturation (−6 ÷ 7 ppm), to supersaturation (+30 ÷ 36 ppm) that could be related to the outcropping of deeper waters enriched by CO2 and nutrients.
To evaluate the CO2 flux (F) across an air–water interface the film model is used following the equation F = K × αΔP, where K is a measured value determined by real hydrometeorological conditions (wind stress, temperature and etc.), α is solubility of the gas (CO2), and ΔP is partial pressure gradient. Here we use the mean value of CO2 transfer velocity, determined by Erickson (1989) for different latitude belts. We used K = 120 cm day−1 for the aquatory south of 60°S, K = 240 cm day−1 for latitudes between 50° and 60°S, and K = 300 cm day−1 for latitudes between 40° and 50°S. The distribution of calculated fluxes of CO2 between the atmosphere and water is presented in Fig. 5, which shows a general source of atmospheric CO2 east of 7°W and south of 60°S. Because the southern boundary of the study area is located near Antarctic divergence we can assume that upwelling of old North-Atlantic Deep Waters (NADW) resulted in partial enrichment by CO2 for surface Antarctic waters that is mainly a mix of shelf Antarctic waters and NADW. This assumption, based on our limited dataset, agrees with atmospheric data that indicate a rise in the CO2 value at the edge of Antarctic (Tans et al. 1990; Conway et al. 1994). Indeed, the annual average concentrations of CO2 monitored at Antarctic coastal stations Halley Bay (76°S, 26°W) and Palmer Station (65°S, 64°W) are higher than at the South Pole (90°S) or at midlatitude (30°–50°S) in the Southern Hemisphere stations (Cape Grim, Amsterdam Island), where undersaturation of CO2 in surface waters occurred. The presence of a CO2 source near Antarctic divergence also agrees with anomalous high concentrations of phosphate (P-PO4) in the zone of Antarctic divergence and south of it (Keeling 1968), which could indicate penetration of NADW (enriched by CO2 and nutrients) in the surface layer.
Comparison of Figs. 4 and 5 shows a good correlation between distribution of ΔpCO2 and F, because the gradient of pCO2 is a driving force for the CO2 exchange between air and water. Likewise, we can see that the spatial distribution of the flux, F (Fig. 5), is more homogeneous in comparison with the pCO2 gradient distribution (Fig. 4). This is explained by “compensation”;there are higher values of the pCO2 gradient in the southern part of the study area, but higher values of transfer velocity (K) in the northern part. The mean value of K is enhances from south to north due to an increase of the wind stress and diffusion coefficient (D), influenced strongly by warming. For instance, the real meridional temperature gradient across the study area might be responsible for a 1.5 times increase of K value near the northern boundary in comparison with the southern boundary, if the wind stress is similar at the southern and northern boundaries.
The oceanographic CO2 data shows that the distribution of atmospheric sinks or sources might be varied significantly, which should be recorded in atmospheric variations of CO2 and δ13C in CO2. Nakazawa and Morimoto (1997), based on the approach of Ciais et al. (1995a), calculated oceanic CO2 fluxes for a non-ENSO year: the total CO2 source at latitudes south of Antarctic convergence (∼50°–52°S) was found. The average latitudinal distribution of CO2 fluxes based on a worldwide long-term database of pH and TA also indicates that the aquatory of the Southern Ocean is a small source throughout the year, rather than a sink of atmospheric CO2 (Bordovsky and Makkaveev 1991; P. Makkaveev, 1998 personal communication). Previous observations of pCO2 show that during winter, cooling of the surface water reduces pCO2, but upwelling and entrainment increase pCO2 (Chen 1988). These observations suggested that the Antarctic surface waters are likely to be in equilibrium with the atmosphere.
An evaluation of the annual mean oceanic CO2 fluxes shows that net transfer between 53°S and Antarctic might be reversed from sink (in 1992) to source (in 1993), which may indicate a change in the balance between photosynthesis and respiration in this area (Ciais et al. 1995b) or a small oscillation in the present-day large-scale thermohaline circulation pattern of the ocean (Broecker 1997).
Comparison of the autumn ΔpCO2 distribution for Arctic (Figs. 2a,b) and Antarctic (Fig. 4) shows that in both regions the water cooling increases invasion of atmospheric CO2, but different factors influence supersaturation of the surface waters by CO2. In the Antarctic, outcropping of NADW enriched by CO2 probably increases pCO2 at the surface; in the Arctic, riverine output and upwelling of bottom coastal waters enhance CO2 evasion. Note, that near the arctic ice edge a large undersaturation by CO2 was obtained in the different seasons, but near the antarctic ice edge (in the Davis Sea) the surface water was supersaturated or quasi-equilibrated by CO2 with air, though photosynthesis is increased near the ice edge in the both regions (Park et al. 1974; Rusanov et al. 1979; Smith and Sakshaug 1990).
The fall study in the Laptev and Chukchi Seas shows a significant northward decrease in surface pCO2 that is governed mainly by temperature. The distribution of ΔpCO2 shows that the aquatory of the Arctic Seas is mainly a sink of atmospheric CO2, though the water near the mouths of the Siberian rivers and coastal sites are supersaturated significantly by CO2. The literature data for pCO2 distribution in the surface waters show also that the Barents, Kara, and Greenland Seas are undersaturated significantly (down to 40%–50%) with regard to atmospheric CO2.
All-season data obtained in the delta system of the Lena River and typical northern lakes show that the freshwater is supersaturated by CO2 with a drastic increase in the CO2 value during wintertime.
The CH4 efflux from the limnic environment in the north plays a significant role in the CH4 regional budget, whereas the role of the arctic adjacent seas in regional CH4 emission is small. This agrees with the aircraft data, which show a 10%–15% increase of CH4 over land when the airplane is flown from the Arctic Basin toward the south (Kelley and Gosink 1988; Harriss et al. 1992). Offshore permafrost might add CH4 into the atmosphere, though the preliminary data available are not sufficient to estimate the source. Evolution of the northern lakes might be considered as an important component of the climatic system.
The pCO2 difference between the surface of the Southern Ocean and atmosphere observed in the austral autumn shows that the area east of 7°W might be a source of CO2 into the atmosphere, whereas the area west of 7°W is a net sink of CO2. This is corroborated by literature data that indicate on overestimation of the role of Antarctic waters as a sink for atmospheric CO2.
This study was supported by the Russian Fund for Basic Research (Grants 93-05-8258, 96-05-66350, 96-05-79143, 97-05-79064), the International Science Foundation (ISF) and the Joint Program of ISF and RF Government (Grants RJD000 and RJD300), and the Federal Program “Integration between Academy of Sciences and High School” (Grant 726). Irina Pipko and Svetlana Pugach did the main work of processing for data and technical preparation of the manuscript. We thank Alexander Gukov, Nikolay Pivovarov, Pavel Tishchenko, Sergei Zimov, Yura Voropaev, Evgeny Radaev, Sergei Daviodov, Sasha Skotnikov, and others who have participated directly in the Arctic research. Sergei Ryabchuk, Vadim Zlobin, Nikita Molochushkin, Alexander Matvienko, Boris Bombelo, and Anatoliy Devyatkin helped us strongly with work in the Tiksi area. We thank also the administration of AARI, especially Alexander Danilov and Sergei Pryamikov and the crews of R/Vs Professor Vize, Mikhail Somov, Alpha Helix, and Dunay for logistical support. Special thanks to John Christensen, who involved us in the Chukchi Sea Cruise on board the R/V Alpha Helix. We are grateful to Victor Ilyichov, Georgui Golitsyn, and Oleg Bordovsky, who helped us to save the Arctic research in the early 1990s. I thank three anonymous reviewers and the editorial board for useful recommendations and comments and editing of the manuscript. Special thanks for Patricia Golden, who helped with the English version.
Corresponding author address: Dr. I. P. Semiletov, Arctic Regional Center, Pacific Oceanological Institute, 43, Baltiyskaya Street, Vladivistok 690041, Russia.