SCOPE 13 - The Global Carbon Cycle

ABSTRACT

Annually 1.08 x 10²⁰ g of water precipitates on to the continents, of which 0.376 x 10²⁰ g runs off in rivers. This water is the motor of erosion and terrigenous biological activity alike. The amount of CO₂ in rain is small (0.065 x 10¹⁵ g C/year) compared with the amount which enters the-water in the soil (net flux reaching the oceans: 0.23 x 10¹⁵ g C/year). Some 0.8 % of all CO₂ produced by root respiration and microbial activity reaches the ocean by water transport. The gross flux from soil to groundwater, however, is larger, because much of the CO₂ is given off when groundwater reappears in springs. CO₂ pressures are 100 times larger in soil air, and 10 times larger in shallow groundwater and in rivers, than in the atmosphere. Both ground and river waters show pronounced seasonal variability of CO₂ pressure and alkalinity. An estimated 20% of carbonate rocks occurs in the crust, but only 4% of the continental area shows karst features. These areas play a distinct role in the carbonate dissolution rate; Europe, the continent with the highest karst percentage, has the largest carbonate concentrations in its rivers.

12.1 INTRODUCTION

The continental fresh waters contain the most diversified carbon pool on earth. Not only does fresh water act as a storage for various inorganic and organic compounds, but it is also a chief transport medium between some of the sizewise more important carbon storages in the biota, lithosphere, and oceans. Furthermore, water is a necessary reaction partner of carbon dioxide, both in the carbonate system and in the biological formation of organic matter. To carry out photosynthesis, land biota is totally dependent on the availability of water (see Chapter 8, this volume, Figure 8.2).

This chapter provides an introduction to the global water cycle, and a discussion of the different water compartments and their respective carbon contents.

12.2 THE WATER CYCLE

Fresh water is available as precipitation (rain, snow, or dew), ice, rivers, lakes, and ground water. The best values of the global water cycle currently available have been calculated by Baumgartner and Reichel (1975). Figure 12.1 illustrates the global water cycle: annually 4.25 x 10²⁰ g of water evaporates from the oceans; this is equivalent to a water layer, 117.8 cm thick, spread over the total ocean surface. A total of 0.397 x 10²⁰ g is transferred from the oceans to the continents (11 cm). The continents receive a precipitation of 1.081 x 10²⁰ g annually, equivalent to a layer of 74.6 cm covering the total land area. Of this amount, two-thirds re-evaporate and one-third returns in rivers and glaciers to the sea. The global ratio between evaporation and runoff is 0.357.

 Figure 12.1 The global water cycle (drawn after data by Baumgartner and Reichel, 1975). P = precipitation, E = evaporation, r.t. = residence time

Table 12.1a lists some estimates of the sizes of various water reservoirs. It must be noted that there is no accepted value for the total volume of rivers and lakes available; estimates deviate by as much as a factor of 10. But it is this relatively tiny fraction of water which man most urgently needs, and which he most criminally misuses. Even less information is available for groundwater volumes. Estimates depend on assumptions of mean porosity, mean sediment depth, and mean depth of groundwater table.

Table 12.1b compiles the data for continental runoff and its carbon load. The most recent attempt to make a global calculation of the amount of river loads was made by Livingstone (1963). In Section 12.10 some corrections are made to the Livingstone calculations: these corrections include updated values for river discharge. Corrections on mean dissolved matter concentrations of the major rivers also seem to be necessary. For many rivers, long-term weighted means are available today, while Livingstone used only data from random sampling.

As rivers provide the bulk transportation from land to ocean (Table 12.2), their average loads are of primary interest for the carbon cycle. The second largest discharge rate of eroded material is due to glaciers, while minor amounts are transported by ground water, atmospheric dust, volcanic events, and shore erosion.

Table 12.1a Water volumes of the earth in solid, liquid, and gaseous forms. (Table 1, Baumgartner and Reichel, 1975. Reproduced by permission of the R. Oldenbourg-Verlag GmbH, München)

12.3 CARBON IN PRECIPITATION

Table 12.1b Annual river discharge and carbon content

	Livingstone (1963) (see Section 12.10)				Baumgartner and
					Reichel (1975)
	Discharge	Salinity  HCO3		C	Discharge
Continent	1015 g H2O	ppm	ppm	ppm	1015 g H2 O
Europe	2 498.5	182	95	18.7	2 800
Asia	11 108.5	142	79	15.5	12 200
N. America	4 557.3	142	68	13.3	5 900
S. America	8008.4	69	31	6.1	11 100
Africa	5 901.3	121	43	8.5	3 400
Australia	316.3	59	31.6	6.2	2 400
(Antarctica as ice					2000)
Recalculated sums and	32 390.1	122.4	59.8	11.8	37 700
weighted means
Dissolved Organic Carbon (DOC):				3.28	(Garrels et al., 1975)
Particulate Organic Carbon (POC):				1.76
Discharge of carbon (in 1015 g/year) as			HCO3	DOC	POC
from Livingstone:			0.382
Baumgartner and Reichel:			0.445	0.123	0.066
Including new data from Amazon River:			0.454

With an atmospheric CO₂ pressure (PCO₂) of 0.0003 parts CO₂ in 1 part of air, we can, according to Henry's law, expect the CO₂ content of rainwater to be between I ppm at 0 ^°C and 0.5 ppm at 20 ^°C. Because continents and precipitation are unevenly distributed over the climatic zones of the globe (Figure 12.2 a, b), it is difficult to assess a global mean precipitation temperature. At 15 ^°C (0.6 ppm CO₂), the annual flux with precipitation to the continent surface should amount to

0.6 x 10^-6 CO₂ x 1.08 x 10²⁰ g H₂ O = 0.065 x 10¹⁵ g CO₂ (0.018 x 10¹⁵ g C/year or 0.0014 x 10¹⁵ moles CO₂ )

Part of the CO₂ is instantaneously lost again into the atmosphere, when the precipitation evaporates from rocks, soil, or vegetation. In this context, the ratio between physical evaporation and biological evapotranspiration should be of interest.

In the more densely populated humic zones of the earth, industrial fossil-fuel burning increases the CO₂ content of air locally and rain may contain larger CO₂ concentrations as compared with the global average.

Actual measurements of CO₂ in rainwater (Table 12.3) show, in fact, a much larger content than expected from the equilibrium argument. The real flux of CO₂ from atmosphere to land with precipitation may, therefore, be several times larger than previously anticipated. No global data are available on rain concentrations of organic carbon.

Table 12.2 Agents of material transport to oceans (After Garrels et al., 1975, Table 10. Reproduced by permission from Chemical Cycles and the Global Environment, by Garrels, Mackenzie and Hunt. Copyright © 1975 by William Kaufman, Inc., Los Altos, California. All rights reserved)

CO₂ in rain alone cannot account for the HCO₃ found in rivers. According to 

H2O + CO2 + CaCO3 Ca(HCO3)2 Ca 2+ + 2HCO3

at least half of the stream HCO₃ must originally be derived from the atmosphere, i.e. 0.23 x 10¹⁵  g C (Table 12.1b).

Figure 12.2 (a) Areas of oceans and continents versus latitude in 5° intervals. (b) Precipitation, evaporation, and runoff from continents versus latitude in 5 ^° intervals (drawn after data by Baumgartner and Reichel, 1975)

Table 12.3 CO₂ content of precipitation (Miotke 1968. Reproduced by permission of the author)

	1. Rainwater	Mg CO2 /1
	McLeod (1869)	2.7
	Frankland (1874)	2.5
	Baumert (1877) in Paul, B. H.	0.7
	Peligot (1877) in Paul, B. H.	1.0
	H. Lehmann (1956) Cuba	2.53.5
	Bögli (1961) Switzerland	2.22.6
	Miotke (1965) Picos, Spain	2.2
	2. Meltwater from snow
	Bögli (1951) fresh meltwater	0.851.76
	Bögli (1961) from vegetated soil	2.643.63
	Miotke (1965) Picos, fresh meltwater	0.4

12.4 VOLCANIC CO₂

Whether or not volcanic CO₂ or CO₂ from hydrothermal sources play a quantitatively important role is not yet known. Recycled CO₂, emitted by volcanic vents or moffettes, directly adds to the atmophere. The estimates of this process differ widely. In certain areas, however, there exists a diffuse CO₂ flow through the crust. These areas are marked by rivers and lakes which carry more bicarbonate than alkaline earth metals. Soda lakes have developed by this process in Ethiopia, Kenya, Tanzania, and in Eastern Turkey (Lake Van; Kempe, 1977). Crater lakes often have sub-lacustrine CO₂ sources, adding volcanic carbon (e.g. Laacher See, Federal Republic of Germany; Lake Kivu, East Africa).

12.5 SOIL AIR

Additional atmospheric CO₂ dissolves in rainwater, only if rain comes into direct contact with alkaline earth metal carbonate rocks. Rain running down bare limestone creates karren by gradual dissolution of CaCO₃ and slow uptake of atmospheric CO₂ Bögli, 1975). The final carbonate concentration, however, which these runoff films reach, is far below that of ground or river water.

The major part of CO₂ responsible for weathering of carbonates and silicates (see Chapter 13, this volume) enters precipitation water when it percolates through soil. Due to the immense total surface of soil particles, soil water is in equilibrium with soil air. At this point a substantial amount of CO₂ enters the freshwater cycle.

In soils, bacterial oxidation decomposes the photosynthetically produced organic matter to CO₂. Root respiration is an equally important source of CO₂ in soil. CO₂ is generated in the. upper few metres of soil, and diffuses upward to the atmosphere and downwards to groundwater. Concentrations and diffusion rates vary with seasons, climate, type of soil, and type of'vegetation cover. CO₂ partial pressures (PCO₂) of up to 0.2 have been measured in soils, and an average PCO₂ of 0.03 is often quoted (Miotke, 1974). To monitor CO₂ emission from soils, two methods have been followed: (i) the CO.₂ increase in a plastic dome covering the soil surface is measured, and (ii) the apparent diffusion constants of the respective soil profile is determined in the laboratory, together with CO₂ profiles under regular field conditions. Albertsen (1977) has obtained annual averages of CO₂ emissions by this second method for four sites with different vegetation types in Northern Germany: coniferous forest 31 g C/m² year; deciduous forest 86 g C/m² year; forest clearing 196 g C/m² year; and agricultural land 404 g C/m² year. The bulk CO₂ emission is directed towards the atmosphere; only in spring and early autumn was a concentration gradient towards the groundwater noticed.

The total CO₂ content of soil air may be estimated by assuming an average groundwater table depth of 6 m (Garrels et al., 1975), an average PCO₂ of 0.003 (which is certainly on the low side), an average gas effective porosity of 0.2, and a vegetation-covered total continental area of 100 x 10⁶ km² (see Chapter 5, this volume, Table 5.2, total continental area minus deserts, semideserts and ice). This calculation yields 36 x 10⁹ m³ of CO₂ of 9.7 x 10¹² g C. As the annual net primary productivity amounts to about 60 x 10¹⁵ g C/year (Chapter 5, this volume), the residence time of CO₂ in soil is approximately one hour. This result is obtained on the assumption that the total primary production is degraded in the soil. This is obviously not the case, as a residence time of one hour is in contradiction with the slow diffusion rate through soil. Therefore, large quantities of litter must degrade under direct contact with the free atmosphere.

The annual net flux of CO₂ from soil to groundwater must at least equal the difference between half the HCO₃ load of the streams and the CO₂ in rain: 0.23 x 10¹⁵ g C 0.018 x 10¹⁵ g C = 0.21 x 10¹⁵ g C. Garrels et al. (1975) assume that this amount of CO₂ used in the weathering of rocks is derived directly from the atmosphere. As discussed above, this does not seem to be the case. The CO₂ in the process of weathering is derived from 26.5 x 10¹⁵ g C of CO₂ in the model by Garrels et al. (1975) produced by respiration and decay, which then diminishes to about 26.3 x 10¹⁵ g C returning directly to the atmosphere. Consequently the flux of CO₂ used in weathering from the atmosphere to land also diminishes.

The fractionation of degradation CO₂ between atmosphere and groundwater is 122/1; 0.77% of the CO₂ produced by degradation of terrestrial organic matter is transferred to groundwater.

With respect to the interaction of water and soil, one other aspect should be noted. If, indeed, most of the CO₂ in groundwater is soil-derived, then this CO₂ must be much older than that furnished by rain. The turnover of humus lapses over a long period, often several hundred years. Reiners (1973) has presented a good review of detritus sizes and turnover masses. Sinter chronology by ¹⁴C is limited by the inability to define the time of humus turnover. Double analyses of wood and its sinter cover show the flowstone cover to be much older than the wood, even after adjusting for the dead CO₂ derived from the country rock of the cave roof. It should be possible, by this method, to estimate the fractionation of soil CO₂ between percolated water and air.

12.6 TYPES OF RUNOFF

Three types of runoff have to be discerned: overland flow, interflow, and subsurface runoff by groundwater.

Surface runoff is mostly low in dissolved matter, but carries almost all of the suspended particles of the runoff. At times of maximal runoff, the concentration of dissolved substance decreases while suspended load increases.

Interflow occurs through the very upper part of the soil cover, through cracks in the soil, and small surface voids. Interflow behaves hydrologically much like surface runoff, but has geochemical similarities to groundwater.

Groundwater passes totally through the soil and can be clearly discerned geochemically from surface runoff. Its discharge peak after thunderstorms is much delayed or cannot be identified at all. Hydrologists have developed several methods to divide river stage curves into surface and groundwater runoff. These methods, however, are not widely used, and global estimates on the ratio of surface to groundwater discharge have not, as yet, become available.

Three types of groundwater aquifers can be discerned: cleft, porous, and conduit aquifers. The first two have diffuse flow patterns and occur in igneous rocks and shales (cleft), and in sandstones and unconsolidated alluvial rocks (porous). Conduit flow occurs mostly within carbonate karst rocks. Some 80% of the continental surface is covered by sediments, of which 60% are shales, 20% carbonate rocks, 15% sandstones, and 5% evaporite deposits. The average composition of the sediment cover is slightly different, with 65% shale and 15% carbonates (Garrels and Mackenzie, 1971; Garrels et al., 1975).

12.7 GROUNDWATER IN KARST ROCKS

The area governed by conduit flow is probably less than 20% of the continental surface, as many calcareous sandstones and shales do not develop conduits. Northern Germany, for example, is underlain by calcareous glacial till, and rivers show high alkalinity and alkaline earth contents similar to karst areas, though hydrologically and morphologically no criteria of karst have been developed.

Balázs (1977a) estimated the total area with morphological karst features to be 4% of the continents (Table 12.4). The bulk of the karst occurs in Europe and Asia between 60^° N and 20^° N latitude, with smaller areas in North America and even less in the southern hemisphere (Figure 12.3). Carbonates occur either in the orogenic belts, or as epicontinental sediments deposited from shallow seas. These areas can form vast karst plateaus once they are uplifted, as in the case of Southern China where the largest single karst area exists. The most pronounced karstification occurs where limestone and humid climate coincide. Figure 12.4 is a map which shows the main areas of karst: the AppalachianCaribbean, the European, the South-East Asian and Pacific Karst Provinces (Balázs, 1977b).

Table 12.4 Distribution of karst areas (Balázs, 1977a. Reproduced by permission of the 7th International Speleological Congress and by permission of the author)

	Area of	Area of karst/1000 km2			Percentage of karst/
	continent				continent =100%
Continent	106 km2	orogenic	epeirogenic	Total	orogenic	epeirogenic	Total
Europe	10.5	528	890	1418	5.0	8.5	13.5
Asia	43.9	808	793	1601	1.8	1.8	3.6
Africa	30.3	67	923	990	0.2	3.0	3.2
North and Central	24.2	760	100	860	3.1	0.4	3.5
America
South America	17.9	90		90	0.5		0.5
Australia and the	8.5	132	250	382	1.6	2.9	4.5
Pacific
Total without	135.3	2385	2956	5341	1.8	2.2	4.0
Antarctica

Figure 12.3 Karst areas in comparison to orogen and platform areas in western and eastern hemispheres (top figure). Latitudinal distribution of karst areas in 5^° intervals both for total and individual continents (Balász, 1977a. Reproduced by permission of the 7th International Speleological Congress and by permission of the author.)

Figure 12.4 Regions of largest geoclimatic potential for karstification (Balázs, 1977b. Reproduced by permission of the 7th International Speleological Congress and by permission of the author.)

These karst areas play an important role in the bicarbonate budget. Subsurface runoff is the main mode of discharge in karst. Here, high soil PCO₂ (Miotke, 1974) can be neutralized to saturation with calcite (CaCO₃) or dolomite (CaMg(CO₃)₂ ), binding large amounts of CO₂. As limestone areas are more common in the humid areas of North America, Europe, and Asia than in the southern continents, we find a very low mean alkalinity in Australian and South American rivers (see Table 12.1) where hardly any limestone exists. In fact, Europe (with a karst area of 13.5%) has the largest average total dissolved solid (TDS) concentration (182 ppm) of all the continents. Half of this load is HCO₃ (95 ppm).

The complex flow patterns within a carbonate conduit aquifer are illustrated by Figure 12.5. Very often, air-filled caves intersect the path of seepage water. Most caves exchange their air quite rapidly with the atmosphere. Cave air has, therefore, a lower PCO₂ than soil water percolating from above. The water loses CO₂, thus increasing the PCO₂ of cave air. Air currents transport the CO₂ back to the surface, where cave mouths are sources of CO₂ . When the water loses CO₂, sinter is formed. Where there is no soil above the cave, sinter cannot precipitate, a fact which is best illustrated by ¹⁴C flowstone analyses, which show an interruption of sintering in glacial times (Franke and Geyh, 1972).

Figure 12.5 Water types encountered in a carbonate aquifer and their interconnecting flowpatterns (Harmon et al., 1972. Reproduced by permission of the British Cave Assocation and by permission of the authors.)

Table 12.5 Karst water analyses (Harmon et al., 1972. Reproduced by permission of the British Cave Research Association and by permission of the authors)

With the availability of computer programs for calculating the carbonate equilibria in fresh waters, karst waters have been investigated intensively for their CO₂ pressure and mineral saturation. Variation in PCO₂, alkalinity, and saturation is due to the length of contact between water and rocks, to the kind of flow encountered, to the mean temperature (and hence to latitude and altitude), and to the seasons.

Table 12.5 compares soil water (top) with water from vadose and phreatic zones (Harmon et al., 1972). The PCO₂ (expressed as a logarithm) of soil water is 0.10, while the saturation index of calcite (SI_c) shows strong undersaturation (negative value). Deeper water still has a PCO₂ of 0.05, while the SI_c has reached values near saturation. This water, upon surfacing, loses CO₂ until it reaches the atmospheric PCO₂ of 0.000 33 (log PCO₂ = -3.48) and will precipitate calcite.

The type of flow determines several parameters of the carbonate system. Parizek et al. (1971), who investigated karst springs with diffuse and conduit flow, found lower total hardness and lower calcite saturation in the case of rapid conduit flow. The variability of hardness is much larger in the conduit compared with the diffuse flow, but CO₂ pressure stayed around 0.0032 (log = -2.5) in both groups.

Average composition of karst water is a function of mean temperature (Figure 12.6, Table 12.5). CO₂ pressure and hardness increase at lower latitudes. This is due to a higher microbial CO₂ production in soils of warmer climates, compared to those of cooler regions. For example, PCO₂ increases from the northern United States (0.0032, log = -2.5) to Mexico (0.032, log = -1.5) by a factor of 10.

Figure 12.6 (a) Relation between average temperature of karst water samples in Table 12.5 and latitude (Harmon et al., 1972). (b) Variation of hardness and PCO₂ with water temperature for averages of all samples within a region (compare Table 12.5) (Harmon et al., 1972). (c) Variation of hardness and PCO₂ with temperature for spring samples only, averaged within the given regions (Harmon et al., 1972. All figures reproduced by permission of the British Cave Research Association.)

Seasonal variations in PCO₂ of karstic waters have been monitored by Hess (1974) in a Pennsylvanian limestone karst and by the Arbeitsgemeinschaft für niedersächsische Höhlen (unpublished data) in a dolomite and gypsum karst in Germany. Figure 12.7 gives two log PCO₂ curves from the German karst, depicting a spring from dolomitic source rock with a diffuse flow (annual mean log PCO₂ 1.7) and a gypsum karst spring with a more or less conduit type of flow (annual mean log PCO₂ 2.4). The main features of seasonal CO₂ variation in Germany and Pennsylvania show a minimum in CO₂ pressure in March-April, due to the beginning of the vegetation period, and a maximum in late summer and autumn, when respiration and decay of organic matter prevail. The amplitude is quite large, over log values of 1.0 in the German and 0.7 in the Pennsylvanian cases. These figures reveal a pronounced seasonal effect in the CO₂ liberation rate from soil. It is, however, not yet possible to calculate from these figures the net flux of inorganic carbon and its seasonal variation. It is known from experience that alkalinity and saturation levels are not only a function of the PCO₂, but also of the amount of water present.

Figure 12.7 Seasonal variation of PCO₂ in springs of South Harz mountains, Germany. Top curve: Diffuse flow spring from a dolomite aquifer. Bottom curve: conduit flow spring from a gypsum aquifer (drawn from own data)

12.8 DEEP GROUNDWATER

So far, we have only been concerned with groundwater of the uppermost water-filled zone. Springs discharge water as long as the reservoirs behind them are filled. This gravitational resurging water is dependent on renewal by precipitation in the source area and by the transmissibility of the rock. Low transmissibility will cause slow outflow. The water discharged by springs is mainly recent percolation water, yet a certain amount of older deep groundwater is also involved.

Deep groundwater has very long residence times, up to 1000 or even 10 000 years. The chemistry of deep groundwater (White et al., 1963) is well known in certain areas because of exploration for oil or thermal water. Estimates on the mass of water within the crust differ widely (Table 12.1).

Many waters (conate waters) have been buried together with the sediments. In tectonically active regions a constant pore-water flux is maintained by compaction. By lowering fresh mud, with 80% porosity, 2 km down into the crust such compaction will occur, and only 10% of the porosity remains (Wunderlich, 1966). Under such pressures a variety of mineral exchange reactions occur, especially with carbonates. The resurgence of these pore waters is one of the factors which stimulate the circulation of deep groundwater.

12.9 RIVERS

In most cases, groundwater gives off CO₂ at its reappearance. If the water stems from limestones, and is near saturation with respect to calcite, calcium carbonate precipitates downstream. Because of the influence of freshwater biota, most rivers never attain equilibrium of PCO₂ with the atmosphere. A characteristic log value for PCO₂ in rivers is 2.5. Figures 12.8 and 12.9 depict seasonal variations in PCO₂ and alkalinity for the rivers Elbe (23-year mean) and the Amazon (for the year 1975). In the Elbe River, CO₂ pressure and alkalinity decrease with the beginning of the phytoplankton bloom in early spring, with an annual minimum in April. Heterotrophic processes cause the CO₂ pressure to rise quickly to a maximum in July, while alkalinity only slowly regains its former concentrations. A small minimum, caused by a late-summer plankton bloom, is encountered around August and September. The long-term mean of the log PCO₂ in the Elbe River is 2.4.

The seasonal PCO₂ and alkalinity variation in the Amazon is less regular, though large changes occur during the year. Alkalinity is high during times of spring floods, when more country rocks are exposed to erosion by water. PCO₂ is low during periods of high waters, when nutrients are available to phytoplankton, while larger PCO₂ values are encountered during periods of lower waters and the ceasing of photosynthesis.

Figure 12.8 Twenty-three-year mean of seasonal variation in PCO₂ and alkalinity of the Elbe River at Hamburg (calculated from hydrochemical river records by the Hamburger Wasserwerke)

Figure 12.9 Seasonal variation of PCO₂ and alkalinity of Amazon River at Obidos 1975 (drawn from data M.Sc. thesis M. L. Nossar Simöes de Dalgo, personal communication Dra. A. de Luca Rebello, Rio de Janeiro) 

Since the CO₂ pressure of rivers is approximately 10 times larger than that of the atmosphere, there must be a considerable flux of CO₂ from stream surfaces into the air. This flux could be estimated according to the laws of diffusion, if we knew the total surface area of streams and their average boundary layer thickness. However, neither figure is available at the moment.

Groundwater also discharges directly into the oceans. This is especially true for limestone areas bordering the sea; half of the drainage of the Yugoslavian Dinarides, for instance, is said to discharge below sea-level (personal communication, Dr. R. Gospodaric, Postojna). Water from caves, which developed during glacial times at low sea-level, continues to be transmitted directly to the sea. This water does not come into contact with air again, and, since it stems from carbonate rocks, it carries a substantially greater load of carbon than the global river mean. The size of this flux has not yet been estimated (see Table 12.2).

With respect to the various parts of the water cycle and the carbon flow interconnected with it, it becomes apparent that the gross flux of carbon from soil to fresh water is probably much larger than that of river loads discharged to the ocean (Table 12.1 b). How much more carbon that is remains obscure; it may be 50 or even close to 100 per cent of the final discharge.

An attempt is made in Figure 12.10 to show schematically the various fluxes found in the freshwater part of the carbon cycle. Instead of reservoir sizes and flux figures, however, only the PCO₂ values and their range in the different compartments are given.

12.10 REVISION OF LIVINGSTONE'S (1963) RIVER LOADS

In 1963 Livingstone published the paper, `Chemical composition of rivers and lakes', which is still the fundamental work for estimating the erosion rates of the continents. By recalculating his world means, several errors, however, became apparent. Both the weighted mean of river salinities of the individual continents and of the world, as well as the HCO₃ figure for world runoff are inaccurate.

The weighted mean is calculated by multiplying the volume of discharge with the salinity, adding these figures for continents (or total earth), and dividing this sum by the total discharge for all continents (or total earth):

		(river discharge x salinity)
	M (continents)=
		river discharge

 Livingstone obtained a mean world river salinity of 120 ppm. By using his continental means, this figure should read 122.4 ppm. By using recalculated continent values, the salinity is 116.4 ppm and the HCO₃ content becomes 59.8 ppm (instead of 58.4 ppm).

The runoff values quoted by Livingstone have been thoroughly revised by Baumgartner and Reichel (1975) (Table 12.1b). Using their continent discharge values, in combination with the revised continent salinities of Livingstone, we obtain a mean total salinity of only 111.06 ppm, i.e. 9 ppm lower (7.5%) than the original value of Livingstone. This is especially due to the present larger discharges of the Amazon and Orinoco in South America. On the other hand, the total discharged volume has increased (Livingstone, 32.4 x 10³ km³ /year; Baumgartner andReichel, 37.8 x 10³ km³ /year),which gives a total dissolved load of 3.77 x 10¹⁵ g/year (revised Livingstone value) and 4.20 x 10¹⁵ g/year (new value) respectively. For HCO₃ the Livingstone value is 1.94 x 10¹⁵ g/year (= 0.381 x 10¹⁵ HCO₃C g/year) and the new estimate becomes 2.26 x 10¹⁵ g/year (= 0.444 x 10¹⁵ HCO₃C/year). There is new chemical data available for some of the larger rivers as well. Livingstone used a mean of 26.4 mg of HCO₃ for the Amazon River. Data from the M.Sc. thesis of Miss Maria Lucia Nossar Simões de Dalgo (courtesy Dra. Angela de Luca Rebello, Rio de Janeiro, Brazil) indicates an annual weighted mean of 30.4 mg/l. As the Amazon supplies 15.1% of the world runoff, the new mean yields.

		30.4
	2.26x1015 +2.26 x1015 x0.151x		1   = 2.31x1015  g
		26.4

(or 0.454 x 10¹⁵ g HCO₃-C/year). This newer value is 19% higher than the revised Livingstone calculations.

Figure 12.10 -log PCO₂ of different freshwater compartments. The CO₂ pressure is expressed as its negative decadic logarithm: one unit less means a ten times larger pressure

12.11 POSSIBLE SINKS OF CARBON IN THE FRESHWATER CYCLE 

Industry accounts for approximately 5 x 10¹⁵ g of fossil carbon emitted per year into the atmosphere, and possibly the same amount is liberated by land management practices. Of this amount, only 2.4 x 10¹⁵ g stay in the atmosphere while 1.4 x 10¹⁵ g are believed to dissolve in the oceans. The rest enters the carbon cycle at points not yet known. Checking the freshwater cycle for such unnoticed `leaks', we have to keep in mind that the amount of carbon to be looked for surpasses the total flux rates discussed here. We can, therefore, expect only minor sinks in this part of the cycle. Nevertheless, these might exist at several points.

In the most heavily industrialized zones, SO₂ emission has decreased the pH of rain below the value of 5.6, which is the normal value, due to the presence of CO₂ and carbonic acid. At a lower pH, the solubility of CO₂ increases (one pH unit causes a 100-fold solubility increase; Cole, 1975). CO₂ could, therefore, be lost from the atmosphere faster in polluted areas than in non-polluted regions.

The atmospheric CO₂ pressure has risen  from a preindustrial value of 0.000 29 to 0.000 33 today (see Chapter 3, this volume). The rise in CO₂ pressure will certainly increase carbonate dissolution on the continents, though the increase due to carbonate equilibria is only very small. Nevertheless, there is some evidence of an increase in alkalinity and in CO₂ pressure as in the case of the Elbe River (see Chapter 13, this volume), and other rivers can be checked for similarly rising trends. These increases, however, may also be induced by CO₂ released from organic pollution.

In mature woods, CO₂ resulting from respiration in soil is quickly recycled by plant uptake, thus keeping the bottom air low in CO₂ . If a forest is cut, this balance is upset in two ways: (i) the total CO₂ production increases (see examples in Section 12.4), and (ii) the CO₂ content of bottom air rises due to lack of large plants to recycle the emitted CO₂. Both processes should result in an enlarged CO₂ diffusion to groundwater and hence to an enhanced weathering of rocks.

Also, groundwater usage causes a disturbance in the carbon cycle. Most pumping of groundwater involves the liberation of CO₂. On the other hand, younger groundwater, with possibly larger amounts of CO₂, is diverted from the upper levels of the subsurface water body into deeper parts, possibly carrying more CO₂ downwards than the amount of CO₂ liberated by the pumping of older ground water.

The eutrophication of fresh water is another process by which we can permanently fix CO₂. Increasing amounts of organic matter are buried in lakes, marshes, and coastal seas.

Probably none of these fluxes will be larger than 10¹⁴ g C. Nevertheless, our attempt to investigate them shows how ignorant we are about the flow patterns and flux rates of carbon in connection with the freshwater cycle.

REFERENCES

Balázs, D. (1977a) The geographical distribution of karst areas. Proc. 7th Intern. Congr. Speleology, Sheffield 1977, 13-15, Sheffield.

Baumgartner, A. and Reichel, E. (1975) The World Water Balance, 1-179. R. Oldenbourg-Verlag, München, Wien.

Cole, G. A. (1975) Textbook of Limnology, 1-283. The C. V. Mosby Company, St. Louis.

Franke, H. W. and Geyh, M. A. (1972). Tropfsteinwachstum and Datierung. Mitt. Verb. Dt. Höhlen-u. Karstforscher 18(3), 59-60.

Garrels, R. M., Mackenzie, F. T. and Hunt, C. (1975) Chemical Cycles and the Global Environment, Assessing Human Influences, 3rd ed., 1-206. W. Kaufmann, Inc., Los Altos, Calif.

Hess, J. (1974) Hydrochemical investigations of the Central Kentucky Karst aquifer system. Ph.D. thesis, 1-219, Pennsylania State University.

Livingstone, D. A. (1963) Chemical composition of rivers and lakes. U.S. Geol. Surv. Prof. Pap. 440G, 1-64.

Lvovich, M. I. (1970) Publ. Assoc. Intern. Hydro. Scient. 93, 401. (Quoted in Flohn, H., 1973.)

Miotke, F. D. (1974) Carbon dioxide and the soil atmosphere. Abh. Karst- u. Höhlenkunde, R. A 9, 1-49.

Reiners, W. A. (1973) Terrestrial detritus and the carbon cycle. In: Woodwell, G. M. and Pecan, E. V. (eds), Carbon in the Biosphere. AEC Symp. Ser. 30, 303-327, NTIS U.S. Dept. of Commerce, Springfield, Virginia.

Wunderlich, H. G. (1966) Wesen and Ursachen der Gebirgsbildung, 1-367 Hochschultaschenbücherverlag, Mannheim.