Hydrological Sciences-Bulletin-des Sciences Hydrologiques, XXI, 2 6/1976

TOTAL MINERAL DISSOLVED TRANSPORT BY WORLD MAJOR RIVERS

MICHEL MEYBECK Laboratoire de Géologie Dynamique, Université de Paris 6, France *

Received 17 November 1975

Abstract. World variations of dissolved transport Td [t k m - 2 year - 1 ] and major ion content S [mg 1. ] with runoff g [1. s - 1 k m - 2 ] and solid transport 77s [t k m - 2 year - 1] are examined on the basis of 40 of the world's biggest rivers (basin area/4 > 4 0 0 00 km2 or discharge g > 5 000 m 3 s - 1 ) . For each river esti­mates of A, Q, S, Td and 77s are given and two subsamples are considered according to the size of the riv­ers. For each subsample it is found that S is inversely related to q, Td is directly related to q and to Ts. The ratio Ts/Td is highly variable but generally increases when Ts increases. Climate dominates S varia­tion and relief Td variation; wide scatterings of S and Td are due to the other factor. On a world scale relief is the main factor controlling variations of Td with 77s. The direct relationship between Td and Ts confirms the work of Alekin and Brazhnikova in the Soviet Union, while the inverse relationship found by Judson and Ritter in the conterminous USA seems to be a minor trend due to the influence of cli­mate surpassing relief. The total sample of rivers considered covers 48 per cent of the continental area draining to the ocean and represents 44 per cent of the world's runoff. If this sample is taken as repre­sentative, dissolved transport of material appears to be dominant or equal to solid transport for about 35 per cent of the world's continental surface. A typology of dissolved transport by major rivers is given according to their morphoclimatic features. If chemical transport is a major process on the earth's surface, the solid material carried by the rivers to the ocean as a whole is five times more than the dissolved trans­port: 20 x 109 t year - 1 according to Holeman as compared to 3.25 x 109 t year"1 on the basis of these 40 rivers.

Transport en sels dissous des plus grands fleuves mondiaux

Resume. Les variations du transport dissous Td [t k m - 2 a n - 1 ] et de la concentration totale en sels dissous S [mg 1_1], en fonction de l'écoulement q [1. s - 1 k m - 2 ] et du transport solide 77s [t k m - 2 a n - 1 ] sont étudiées sur 40 des plus grands fleuves mondiaux (d'une superficie drainée supérieure à 400 000 km2

ou d'un débit supérieur a 5 000 m 3 s - 1 ) . Sur l'ensemble des 40 fleuves les régressions suivantes ont été trouvées: S= 392<r°-62; Td= 12.3<?0-39; Td= 6.6 7s0-3 7 . Le rapport Ts/Td varie de 0.1 à plus de 30 et augmente avec Ts. Le climat domine les variations de S et le relief celles de Td; les dispersions impor­tantes remarquées sont dues à l'autre facteur. Par contre le relief est le principal facteur contrôlant les variations de Td en fonction de Ts. La relation directe trouvée entre 77d et Ts confirme les travaux sovié­tiques d'Alekin et de Brazhnikova; la tendance inverse décrite pour les Etats-Unis par Judson et Ritter n'est pas représentative du phénomène à l'échelle mondiale. L'ensemble des fleuves considérés représente 48 pour cent de la superficie continentale drainée vers les océans et 44 pour cent de l'écoulement exoréi-que, si cet échantillon est représentatif les transports dissous sont supérieurs aux transports solides sur plus de 35 pour cent de la superficie continentale drainée vers les océans. Les variations importantes de q, S, Td permettent de proposer une typologie des transports effectués par les grands fleuves suivant des critères morphoclimatiques. Si le transport dissous car les rivières est un processus fondamental, les ap­ports solides à l'océan restent globalement bien supérieurs aux apports dissous: 20 x 109 t an - 1 de ma­tières solides d'après Holeman et 3.25 x 109 t an - 1 de matières dissoutes, estimation basée sur les 40 grands fleuves étudiés.

Until 1960 the dissolved t ranspor t by rivers had been m u c h less s tudied by geomorphologis ts , hydrologists and geochemists than solid t ranspor t . The la t ter was generally considered as the major process of mater ial supply to the ocean and therefore some studies of denuda t ion

* Present address: Laboratoire de Géologie, Ecole Normale Supérieure, 46 rue d'Ulm, 75005 Paris, France.

265

rates only took mechanical erosion into account. There is now a new interest in river water quality studies and these are proceeding in two main directions: (1) geochemical processes are thoroughly studied on small representative basins (some 100 km2 or less) where environ­mental factors can be well defined (this kind of work was generally undertaken in the Inter­national Hydrological Decade), (2) on the other hand the overall geochemical balance be­tween land and ocean is established on the basis of inputs of the world's biggest rivers. This approach was attempted by Alekin and Brazhnikova (I960, 1962) who performed intensive work on rivers in the Soviet Union. The first attempt to make a world survey of river water quality was by Durum et al. (1960) for the International Association of Scientific Hydrology, but this work, which concerned both major and trace elements, was never completed. In 1963, Livingstone published the most complete literature review of surface water quality. Some other papers concerned the river quality of only one country, such as Japan (Kobayashi, 1960), or particular elements in many rivers (Turekian, 1969; Konovalov, 1970).

Since Livingstone's review a lot of important work has been attempted on the world's biggest rivers: the Amazon (Gibbs, 1967, 1972), Parana (Depetris and Griffin, 1968; Bonnetto et al, 1969), Danube (Liepolt, 1967), Mackenzie (Brunskill et al, 1975), Chari (Roche, 1975), Mekong (Meybeck and Carbonnel, 1975). The studies of river mineralization and dis­solved transport have been undertaken mainly on rivers located in the same country where basins are sometimes similar, or they were based on worldwide river water quality data (Alekin and Brazhnikova, 1962, 1968; Langbein and Dawdy, 1964; Corbel, 1964; Judson and Ritter, 1964; van Denburgh and Feth, 1965; Leopold etal, 1964; Gibbs, 1970). Many relationships between mineralization, dissolved and solid transport rates, and environmental factors have been described by these authors. The principal assumptions can be summarized as:

(1) The mineralization of water is inversely related to specific discharge (Langbein and Dawdy, 1964; Durum et al, 1960).

(2) Dissolved transport is directly related to specific discharge (van Denburgh and Feth, 1965; Langbein and Dawdy, 1964).

(3) Dissolved transport is inversely related to solid transport (Judson and Ritter, 1964). This assumption has been quoted many times in American literature (Strahler, 1971, p. 647).

(4) Dissolved transport is directly related to solid transport (Alekin and Brazhnikova, 1962). This is mainly the view of Soviet authors (Strakhov, 1967). While Corbel (1964) found no relationship between these two forms of transport.

(5) The ratio of solid transport to dissolved transport increases when solid transport increases (Alekin and Brazhnikova, 1962) and when specific discharge decreases (Leopold et al, 1964).

(6) For Alekin and Brazhnikova (1962, 1968), Strakhov (1967) and Gibbs (1967), re­lief is the first environmental factor that controls river chemistry, but climate is also very important, while lithology has a minor influence. Gibbs (1970) found that river chemistry is influenced by precipitation, rock interaction and evaporation.

On the basis of up-to-date quality data for about 40 of the world's major rivers the ob­jective of this paper is (1) to set up a review as complete and reliable as possible of river characteristics, namely discharge, water mineralization, solid transport, (2) to test the prev­ious assumptions with a sample from the major rivers of the world, (3) to examine the qual­itative importance of dissolved transport by rivers on a world scale.

As in many previous studies only the sum of the major ions (Ca++, Mg++, Na+, K+, CI", SO4""", HC03~, N03~) and silica will be considered here. Throughout the rest of the paper the term 'salinity' is used for these quantities. As a matter of fact as it can be seen from Livingstone (1963) and Alekin and Brazhnikova (1960, 1962), the ionic ratios in waters of the major rivers are very constant: Ca++ and HCO3 are the most abundant. This is the case for more than 90 per cent of the rivers in this study which have the 'rock dominated' type of

266

water described by Gibbs (1970). For the remaining rivers Na+, CI" or S04"~ are dominant; these waters are either of the 'rain dominated' type or of the 'evaporation-crystallization type' (Gibbs, 1970). Apart from these exceptions, the variations observed for salinity and mineral transport can be extrapolated to transport and concentrations of elements, except for silica, 10 per cent of S, of which peculiar variations will not be discussed here.

GENERAL CHARACTERISTICS OF MAJOR RIVERS

About 40 rivers were selected for which either the discharge was more than 5 000 m3 s~' or the drainage area was more than 400 000 km2. Discharge, drainage area, total mineral con­tent and solid transport were determined for each river (Table 1). Basins less than 100 000 km2, where local influences can outweigh major environmental factors were eliminated, and middle-sized basins (area between 100 000 and 400 000 km2) were not numerous enough to be considered. It is obvious that the scattering of data decreases when the size of the rivers in­creases and at the limit there would be only one set of data if the whole continental surface were drained by a single giant river. Thus the effect of river size on the derived relationships is important, and the sample of rivers has therefore been divided into two subsamples ac­cording to either discharge Q or basin area A :

big rivers continental rivers 5 000 m3 s-i < Q < 20 000 m3 s~i Q > 20 000 m3 s-i or or 400 000 km2 < A< 2 400 000 km2 A > 2 400 000 km2

The 'continental' rivers drain part of their continents: i.e. the Amazon, Congo, Lena, Missis-sipi, Nile, Ob, Orinoco, Parana, Yang Tse Kiang and Yenissei. The Yang Tse Kiang is the only continental river for which water quality data were not available. Some continental basins are not homogeneous as regards their geological or climatological features and it was thought better to split these into sub-basins. This has only been carried out for the world's largest river, the Amazon, on the basis of Gibbs' data (1967). Therefore the Madeira (Q = 32 000 m3 s~* ) has been taken as a continental river but not the Negro (Q = 45 300 m3 s-1, A - 755 000 km2) for which the drainage basin was considered as too small. The River Amazon itself has been represented in the figures but it has not been taken into account in the regression computations.

On the other hand rivers from Japan, Finland and the Quebec Canadian Shield were added that were similar enough to be considered collectively as single rivers to which weight­ed values were attributed as had been done before (Durum et ai, 1960; Judson and Ritter, 1964). Finally, in order to illustrate the general scattering of the phenomena, and their var­iability with river size, Figs. 1, 2 and 3 present the different ranges of variations for (1) con­tinental rivers, (2) big rivers, (3) smallest rivers (10 000 km2 < A < 400 000 km2). Hydrological features A complete list of hydrological features of the rivers in the sample is given in Table 1. Data for other major rivers, e.g. the Yang Tse Kiang, for which no quality data could be found are also given. The figures in Table 1 have been derived from numerous sources. First of all the up-to-date data published by UNESCO (1969) or by the UN ECAFE Commission (1968) were referred to. Then values published by scientists doing national studies were chosen (Geographic Atlas of the World published by the Soviet Academy of Science, 1964; US Geo­logical Survey publications for the USA). When no data could be found from these sources other reviews, e.g. those of Leopold (1962), Gibbs (1972), Holeman (1968), and Lopatin quoted by Lisitzin (1972), were considered. It should be noted that these data are sometimes different from one author to another: for instance for Asian rivers Coleman (1968) attributes a very high discharge to the Huang Ho: 20 000 m3 s-1, compared to 1 480 m3 s - 1 , a much

267

as nu

al m

iner

al

<

% ea

r-i]

C O

—1

w

TA

BL

^ fe a.

?> S* & s £ t j

& â

s of

the

ure,

• K .

^ "5î

fe l

<w * + j

ta-.

5" ai « t a > "S ° " o o •*>

23 ^ ' a S S

•S^ i i

min

erai

en

t,S

1.

-1]

tal:

ont

[mg

o o — H m

s-T S

ET B!

32

ç»

o

l n H H M ^ \ O t C O H H H r-4 U-) VO r-H t-H _. V I . CM so so r~- v> T-H co CM -^f so so r- ?t | ' T ' O CM CM CM r-T-H 3 CM I l O H H O ^ ^ j - M , ^ H H O -," • - v" '"T-^cr. v i . ^ V I O H - « l o o M SO sO 'O uo cO f-~ c3 - f SO so uo 1 JD I 1 so . * CM ,0 CM U") rH © CM CM . „ . « . . . . . . . . . « . « . . t . . . . . . . . • - . ~ . ~ . . . .«ON . - . . - . * N • - • -l O H H H ^ r f l N H f n H H H O O-H rH H ,„lf)H>nrHSO ' ^ lO

CM so so t-- *o co t - iN "3- SO so r- co m t ^ D ^ a s M - t M O H I CM^t

O O r - ^ r ) f S ^ T - o - H r n s O ON CM CM t-» FH f n t t so -H H r-

M N ^ i o t n H HH^fON r-j t> r* oqoqcoso© ©t-HcôoNCMincMcÔTH

^ m ^ o s o O r t r n t t M n r -r— I—* o r- r--«a* i—* oo n m

CM co 00 v j

CO CO t-H o o o o N s o v - i r - o o i o - c j - o o uo o o so vo © uo co ON co T-H T-H wo r - - H O M ^ t | CM r - l

so © © © CO CM T-H uo CM CO 00 ^ H s o m CM uo i-H t- CM co r- I

r i r - O r f - ^ - O O N C O O O < o o o « ^ l o O t s s o O c n m O H O s O

, soONCMt>-©uo©oo©c-I co co •«*• f—( ON r~-rt H O )

uo©©©uowocoouo© s O r o s o ( S O O M o o 00 r-HCM H H H M f O CM

0 ( N r^ © ON co O co T-H © T t q c j o o r ^ r n OO SO CO CO CM t-H r-H ON 00 CO CM CM CO ! > H l O S D v i f O fNj CO t-H f—f CO

uoco SOSO T-H

. . . . r-~co uo © © c o CM COCMCM so

© © O © © uo © © uo © O wo uo © in uo oo © co r - © © © CM r» ^r co oo Tf uo so so so © oo wo Tj- ON r -

© © u o O © © O r - - u o t - - O u o © i o c o ^ f c o © o O ' « 5 r © O N s o r - w o t o ^ ^ S O T f O O C O C M r f r r - C M © ^ ^

© o o © © © © © © © © © © o © © © © o o © © © © © © O © uo © CM-3-so © co uot-~© oo © o uo O © © O o o o ^ t © O 0 © ' = t - c o c O S O O N C N ' ! 3 - s o o o s o -< - C ^ - < 3 - T ~ H C O S O © U O © C O , < 3 - I > - C O U O UOr-HrHON—H f - ON SO r-H T-H r-< S O C O m S O O > ( N i r - H C O C O UOCO f-. T-H i H CO T-H i—! t-H CO H r t r H T f

H O > t O S O ( N r H C O N r ~ - « t V T T t CO T-H CM CM T-H H f N t t t - l

&* o.S

ess 2^3-s § | as

m i n t s ^ H C O T Û O - H IOOO<N

a S o) _» S 60.B" . ,

c^ S b*â 2 c s >>••— =

•a c P3

T3t3 ^c « - a

3 3 « 3 3 g cucu-3

see

c

.CO c . N ^ ^

S S <*<! S3 < u ^

I!

268 Iiiiss3iisiiicii^iiîiïii^

Ï1= o

-H "S

pa g ^ 1

C5

g

S S I Ê ^ J S

5"

Q SE

1 1^3 —

c -3 (2 m

X t f <1)

^

s ^ m O

CD

"* •*

© <o -"d" <*> t—

, „ r -<CO m . -4> \ p rH Tj* r - 20

;'24

1 ;b;6

5 ;2

5;2

5 i n r» t N , H

T-H i n

- in *nr* in 1 C J r t *-C> n ;-

;54

rtM(S

i i S T-H^O 00

M M y3 " * VO-( l> 13 ^D r- ^ ( S t M r H > O o J -5-

o r- O C Ï © v© co ^

r- Tt* ro in r-- j H

\©co mos^r

00 CO in o

Os O " 00 V)

, coco l r

© O C O C O C N C O O O

CScO *n j j <N

©cs\© o

r - o o s © o uocom O h H os o r -eO T f t-(

^o co r es , c o ^ H N o o T t

C O Q Q © C O C N © C S T - H

tn©.cocoin©oo\o fS T-H *H H T J - ^

© 0 0 . © © © ' O , *n r f © © *sO Os es I ooioos I co c--c-co oo r CO r-i CS T-HT-H

00

in oo os os in © ^ es

^O Tj-CS 00 CO Tt <N © CO CS Tf <N t~; © CO OS CO CS H \ O Î © V J D m* - ^ " < t r 5 ^ ^ s © t t «-* y D o ô « n t ^ T - H O \ CO T-H ,«H r-H CS r-H »H rH -H CS

8© © © © © © M O © © 0 © O Ç > © © 0 © O © © © © © Ô O O i o Q h M i > MO '-H © © © 4o in •<* in ©r -^ foo inoo

-H © in O osoo MD© rf co "«t v> os ^j- co co in os ^ r -cooo inco

S™

i

© © © © o © © © © © c o i n © © © © © © T -«oocoos©0' -<r-Os %£>eSCS<N©00 ©CS

wH CO T-H r H

fO«*soooinc--^ifN

© © © O © © © © i n © © o o © © © r -cs© i>©Tf oo ©

© © © o © o © o o * n © © es es T-H t--. r- in

© C S *-H CO oo <n T-H os

»nco es©* escoooe

rttr.l

JlfaPIJ! •o

cl ^

S1! —i H-l

g o

"S S

g o. O 2 9. 3

2 >> "•a >, c

<D 0 O

Ji,£jl 3 5 , 3 c <-. c

-o c oj

8 ^ 6 S

« ï w ' S l H 1 " 's °^ S S g

« S o C * S? O G o

269

more credible value, given by Holeman (1968). The same confusion exists for the Brahma­putra for which the UN ECAFE value was chosen. As for the Mekong the discharge value of 18 300 m3 s - 1 generally given by ECAFE (United Nations, 1968) was chosen instead of the higher value (21 200 m3 s -1) chosen previously (Meybeck and Carbonnel, 1975) that was perhaps a misprint in the ECAFE report (1968, p.6).

Water quality data Apart from some Soviet data, most of the river quality values have been published since Liv­ingstone's review (Amazon, Parana, Mackenzie, St. Lawrence, Yukon, Niger, Orange, Ganges, Mekong, Danube) or were not taken into account by this author (Nile). Complete references are given in Table 1. In most cases the total mineral content has to be computed. For some rivers for which complete analyses were not available, we took the ignition loss. For Soviet rivers the total ionic content was calculated as the ratio of the annual ionic transport over annual discharge, as given in the Soviet Atlas, plus 10 per cent in order to take into account the mean silica content for Soviet surface waters. As the annual ionic transport in these riv­ers was computed with bicarbonates expressed as C03~~, we have previously corrected these values by the factor 1.364 as suggested by Alekin and Brazhnikova (1960). Expressed this way the data given for Soviet rivers are in good agreement with the last publication of Zverev and Rubeikin (1970) where bicarbonates are expressed as HCO^.

As only the general process of mineral dissolved transport was the subject of this study, neither the dissolved organic matter, which may represent up to 20 per cent of the total dis­solved content (Alekin and Brazhnikova, 1960), nor the various colloidal forms that are at the limit between dissolved and particulate states and are generally not quantitatively im­portant, were considered.

Whenever it was possible annual weighted means have been computed: it is now well known that river mineralization is highly dependent upon discharge and hydrological events. This variation exists even for the major rivers (Gibbs, 1967; Roche, 1975). For example in the Mekong River at Phnom Penh total ionic contents ranging from 90 to 190 mg lr1 were measured (Carbonnel and Meybeck, 1975) and the mean arithmetic total ionic content was 121 mg lr1 versus 95 mg lr1 for discharge weighted mean. Nevertheless some water quality data are based only on a very few analyses, such as for the Orinoco, Magdalena, Brahma­putra, Congo, Niger and Indus. Accuracy is very poor in these cases (30 to 50 per cent) but these values were still taken into account because the geographical variations of salinity are much more important (two orders of magnitude for these rivers).

Corrections to quality data When dissolved transport and erosion rates are considered many corrections are usually sug­gested (Douglas, 1964; Meade, 1969;Janda, 1971; Gorham, 1961). Most of them are very relevant but difficult to accomplish. First dissolved transport and chemical erosion must be clearly distinguished, as many of the dissolved elements carried by the rivers do not come from land erosion. On the other hand the dissolved elements can precipitate in some rivers in semiarid and arid environments (Blanc and Conrad, 1968) and the dissolved transport may not be complete. This phenomenon which exists for the evaporation-crystallization type of rivers of Gibbs (1970) is very rare for the major rivers studied here. Much more important is the addition to the river of elements such as recycled marine salts, volcanic emanations, and all kinds of land and atmospheric pollutants. Recycled marine salts can be very important for very dilute waters such as those in Finland (Viro, 1953), Quebec or lower Amazonia (Gibbs, 1970). This type of river is, however, uncommon as when salinities increase to more than 20 mg lr1 (more than 90 per cent of the major rivers) the proportion of marine salt decreases. Zverev (1971) has estimated the proportion of ions from atmospheric origin (re­cycled marine salts, other natural sources, pollutants) for each major drainage basin in the Soviet Union. There is a maximum for the Berhing and Okhotsk sea basins (28 per cent of

270

the total ionic content, here around 60 mg lr1). For the whole Soviet Union this proportion is estimated to be 14.4 per cent. This value can be taken as a first estimate of the global im­portance of the atmospheric elements in the world's rivers.

Man's activities in industrial countries have sometimes greatly affected the river quality either by direct dumping of wastes or by indirect atmospheric or soil pollution. Drastic changes have been described by Ackermann et al. (1970) for the Mississipi, Weiler and Chawla (1969) for the St. Lawrence and the Rhine. These changes are mainly caused by in­dustrial wastes and generally affect CI", Na+ and SO4" and as a consequence the total ionic content. For instance the salinity of Lake Ontario shifted from 140 to 190 mg lr1 between 1900 and 1960. There is a similar variation for the Mississipi River (from 200 to 300 mg lr1). These examples may represent extreme changes of river quality and in these cases pre-indust-rial water quality data were sought. For other rivers it was assumed that the natural quality has not been altered by pollution by more than 20 per cent. Another of man's influences on river quality is through the use of water for irrigation in semiarid countries, where evapora­tion and deposition of salt deposits has greatly increased the salinity (Gibbs, 1970; Feth, 1971). This is the case for the Colorado River for which the author took Judson and Ritter's (1964) data rather than present values published by the US Geological Survey.

Finally, it must be pointed out that an important part of the bicarbonates is not derived from land erosion but from atmospheric C02 . This is the reason why Alekin and Brazhnikova usually expresssed their data in terms of carbonates. This correction has not been realized because the exact proportion of atmospheric C02 in rivers cannot be accurately known as it depends on each weathered mineral. But when dissolved transport rates are taken for chemic­al denudation rates this correction may be very important, as bicarbonates usually constitute more than 50 per cent of the dissolved mineral content of rivers. For these reasons river transport rather than erosion has been discussed in this paper.

It is known that the quality of the data on dissolved content is still very poor, with errors up to 50 per cent in the worst cases. However, it has already been pointed out that the salinity range in major rivers varies from a few milligrams per litre to 1 000 mg lr1, and this variation greatly exceeds the range of errors.

Finally, transport rates have been expressed in metric units of tonnes per square kilo­metre per year as this is more useful than the units used by geomorphologists [m3 km-2

year-1 or mm/1 000 years] that are based on assumed mean rock densities. The new chemic­al denudation unit [equivalent m~2 year-1 ] recently introduced by Reynold's and Johnson (1972) is very convenient but not suitable for the comparison of chemical and solid transport.

Solid transport data Most of the solid transport data come from previous reviews of Parde (1953), Holeman (1968), Coleman (1968), Gibbs (1967), UN ECAFE Commission (1968), Lopatin quoted by Lisitzin (1972), and the Soviet Geographic World Atlas (1964). As natural values were need­ed the data before river damming were selected where possible, e.g. for the Niger (Lopatin in Lisitzin, 1972), and Colorado (Curtis et ai, 1973). Even more so than for dissolved trans­port, for some rivers there are large differences between data from various sources, especi­ally for some Asian rivers such as the Brahmaputra, Ganges and Mekong for which the ECAFE data were taken. Despite corrections some of these values are not representative of natural erosion but represent erosion which has been greatly increased by man (deforesta­tion, agricultural practises, urban development). In some regions of the United States mech­anical transport has been increased more than four fold since European settlement (Meade, 1969). Moreover solid transport is not a continuous process and solid materials released by mechanical erosion are not always carried all the way to the ocean. There is very often a deposition of the coarser materials in the river bed or in the flood plain.* The proportion of

* Or in lakes and swamps, therefore the Nile and St. Lawrence Ts values were not taken into account.

271

solid material carried by the river that is redeposited reached 50 to 65 per cent for the Yel­low River (Huang Ho) before dam construction. Solid transport is usually estimated by sus­pended load, while bed load, which is generally supposed to be 10 per cent of the total transport, is not taken into account. However, these variations from the natural solid trans­port values are still beyond an order of magnitude of the world range of solid transport which varies from a few tonnes per square kilometre per year up to some 1 000 t km- 2

year-1 for the major rivers.

VARIATIONS OF TOTAL MINERAL CONTENT WITH RUNOFF

Major rivers Figure 1 represents the variation of river 'salinity' S with runoff q for the two samples con­sidered: big rivers and continental rivers. There is a definite relationship between S and q:

S = aqb with - 1 < 6 < 0 (1)

when the two samples are considered regression values are slightly different but this differ­ing I."1 Salinity 10000

1 Range tor all rivers

- | Z Z | Range lor rivers Q > 5 0 0 0 m 3 s - ' or A > 400000 km 2

, | 3 8 3 R a n 9 e , o r rivers Q > 2 0 0 0 0 m 3 s - ' or A > 2 400000 km2

— Langbein and Dowdy - Variation tor the USA

*93

JW

100 l.s"1 krt Specitic discharge

Fig. 1 — Salinity (major ion content plus silica) versus specific discharge.

272

ence is not significant. If S is expressed in mg lr1 and q in 1. s-1 km-2, the following relation­ships were found:

big rivers S = 442q-°-66 (r = -0.56 for 32 values) continental rivers S = 256g-»-45 (r = -0.73 for 9 values) all rivers S = 392q~-°-(>2 (r = -0.57 for 41 values)

The correlation factors are significant at the 0.95 and 0.99 confidence limits. This relationship was expected but variation is less marked than in the studies of Lang-

bein and Dawdy (1964) for 168 American rivers arranged in groups and Durum et al. (1960) for the five major basins in the USA. This variation with runoff indicates a strong relation­ship between salinity and climate. However, there is an important scattering of salinities because of relief effect: the salinity is always greater for mountain rivers than for rivers in plains. In regions where the climatic range is limited, relief may become the first environ­mental factor. This is very well exemplified by the Amazon basin (Gibbs, 1967) where runoffs range only from 20 to 60 1. s-1 km-2. The tributaries from the Andes, namely the Maranon and Ucayali have much higher loads (90 and 150 mg lr1) than the rivers that enter lower in the basin (a few milligrams per litre for the Xingu and Tapajos). Gibbs calculated that relief accounted for 85 per cent of the variance of the salinity in the 16 Amazon sub-basins and climate for only 4 per cent. It must be noted that water quality for the whole Amazon is greatly influenced by the upper basin tributaries and the weighted average that is taken does not truly correspond to any one tributary. According to Gibbs (1967) the Ucayali and Maranon, representing only 13 per cent of the Amazon basin, contribute 45 per cent of the dissolved material. This is not the case for the Mekong River where the transport rate was found to be constant along the river (Meybeck and Carbonnel, 1975)—for this river climate is probably the most important environmental factor controlling the salinity.

World range of variation As expected the scattering of data is greater when any particular river is included whatever its size. S varies from less than 5 mg lr1 in the lower Amazon basin to more than 20 000 mg lr1 for the Pecos River, New Mexico (US Geological Survey). Data are much more scattered at higher runoffs (q > 201. s-1 km-2) than at lower runoffs (q < 0.1 1. s-1 km-2) where the maximum and minimum salinity values tend to converge. Unfortunately the data for semi-arid rivers are only for rivers in the southwest USA. If this trend was confirmed in other regions it would mean that under semiarid climates relief becomes a very minor factor: this is similar to Gibbs' assumption (1970) that for such rivers the salinity is controlled by the evaporation and crystallization processes.

VARIATIONS OF DISSOLVED TRANSPORT

Influence of runoff

Major rivers Dissolved transport rate Td is the ratio between the dissolved matter Md carried by a river during a given period over the basin area A. It can be expressed as Td=Md/A=S Q/A=S q where q is the specific discharge or runoff (q = Q/A). When S is derived from relation (1), the dissolved transport is linked to runoff according to relation (2):

Td = aqc with 0 < c < 1 (2) This relation is strictly derived from relation (1). Figure 2 represents this variation, together with the previous figures of Langbein and Dawdy (1964) and van Denburgh and Feth (1965).

273

I km-'yeor-' Specific dissolved transport 10000 -i

I I Ronge for ail rivers

h E S Ronge for rivers Q > 5 0 0 0 m 3 s " 1 or A > 400000 km2

> j j^x^ j Range for rivers Q > 20000rn3 s"1 or A > 2 400000 km2

~ZT.'^ Van Denburgh and Feth for western USA

Langbein and Dowdy - Variation for the USA

100 l.s-' km"2

Specific discharge

Fig. 2 - Specific dissolved transport versus specific discharge.

The regressions are not very different from the observations from the USA.

big rivers continental rivers all rivers

Td = 13.8^0-34 (r = 0.34 for 32 values) Td = B.l5qo-s4 Q. = 0.78 for 9 values) Td = 12.3co.39 Q, = g 40 for 4i values)

(2a) (2b) (2c)

The correlation factor is only significant for (2b) and (2c). There is an important influence by the lower Amazon tributaries. All regressions are more significant if only the Amazon itself is considered.

As expected maximum transport rates occur for mountain rivers where both runoff and salinity are high-the Brahmaputra's rate is the highest of the major rivers of the world (140 t km-2 year-i)~while the minimum transport rates have different origins. They are caused either by very low runoff (q < 0.5 1. s - 1 km-2), such as in the Murray for which the salinity increase does not balance the runoff decrease, or by a very low mineral content despite high runoff such as in the lower Amazon basin (Xingu), or by a combination of low salinity and low runoff such as for rivers in a semi-arctic environment (Yana).

274

World range of dissolved transport The minimum transport rate is always around 3 t km-2 year-1. This can represent the mini­mum atmospheric input in the case of areas of very low erosion. For non-perennial rivers (q < 0.01 1. s"1 km' 2 ) this minimum should be less than 3 t km-2 year-1, but it is difficult to compare this type of river to perennial rivers as chemical transport is not a continuous process. Maximum rates are inversely related to basin size. For middle-sized basins the max­imum could be reached by the River Rhone (190 t km-2 year-1 for 96 000 km2, unpublished data). For small streams draining calcareous deposits the transport reaches 500 t km-2 year-1

for the Dranse River, France, a small tributary of Lake Leman (Meybeck, 1972) and for the Bregenzer Ach, a tributary of Lake Constance, also in the Alps (Muller, 1964). Corbel (1964) quoted a rate of 420 t km-2 year-1 for Gold Creek, Alaska. Strakhov (1967) proposed a similar range, from 3.9 t km-2 year-1 for the River Yana to 290 t km-2 year-1 for some mountain tributaries of the Caspian Sea. These ranges are much more significant than the one quoted by Stoddart (1971) (between 20 and 60 t km-2 year-1).

Comparison of dissolved and solid transport

General characteristics of solid transport It is far beyond the scope of this paper to study solid transport 7s for which some of the well-known characteristics are as follows:

(1) solid transport is highly variable in time and space, (2) it is not a continuous process, (3) suspended concentration increases with discharge.

Many authors have tried to relate solid transport with environmental factors but their findings are sometimes conflicting. The results of studies are greatly dependent on the size of river basins, on their homogeneity and on the range of morphoclimatic features in the study. For instance Langbein and Schumm (1958) observed a maximum transport rate for low runoffs (around 10 in. year-1) on small US rivers. This relationship was not found by Tabuteau (1960) who considered widespread data from Europe and North Africa, and found some rough ranking of rivers according to their morphoclimatic characteristics. The author also did not find any relationship between Ts and runoff for the sample of the world's major rivers considered here.

Once again relief and climate are the main environmental factor controlling solid trans­port but they sometimes act in opposite ways. This causes a great scattering of data when only one parameter is considered. Most authors agree that relief is the most important factor (Schumm, 1954; Fournier, 1960; Corbel, 1964) but climate is also very important. Recently Jansen and Painter (1974) proposed four main relationships between Ts and morphoclimatic parameters according to four types of climate. In every case they found Ts directly related to runoff, altitude, precipitation, relief, temperature and inversely related to area and some­times protective vegetation cover. In most cases the influence of geology and vegetation cov­er was not significant at the 90 per cent level. This work is so far the most complete attempt to quantify solid transport and clearly shows the complexity of this process which involves four or five major factors.

Solid transport by glaciers is another source of material but it is not included here. Trans­port rates can be 20 times higher than transport rates in rivers. Corbel (1959) quoted a rate of 60 000 t km-2 year-1 for Hidden Glacier in Alaska, and Borland (1961) measured a max­imum rate of 80 000 t km-2 year-1. As it does not generally affect solid transport for big rivers, it will not be studied here.

Relation between dissolved and solid transport for major rivers This is a much debated question as American authors following Judson and Ritter (1964)

275

believe Td is inversely related to Ts while Soviet authors following Alekin and Brazhnikova (1960) believe Td is directly related to Ts. When the 40 rivers of this study are considered (Fig. 3) it is obvious that Td and Ts are directly related but that the ratio Ts/Td is not con­stant and generally increases towards high Ts values. This variation for major rivers can rough­ly be expressed as

Td = lTs™ with 0 < m < l (3) The regression parameters are

big rivers Td = 4.8 Ts°-** (r = 0.71 for 28 values) (3a) continental rivers Td = 15.6 Ts0-19 (r = 0.51 for 8 values) (3b) all rivers Td = 6.2 Ts°-^ (r = 0.67 for 36 values) (3c) Regressions (3a) and (3c) are significant at the 0.99 confidence levels.* The reason for the different conclusions by the above mentioned authors is due to the types of river taken into account. Soviet authors took 30 large rivers that cover a wide range of environmental char­acteristics while the American authors considered five main US basins that were not so var­ied as the Soviet ones. Therefore when the Judson and Ritter's data are presented in Fig. 3 their trend appears to be an exception with regard to the world variation and this also applies to Langbein and Dawdy's (1964) data recomputed from their figures 1 and 2. Their samples of rivers are not representative of the world average so that their trend cannot be extrapolat­ed outside the USA.

This major trend is obviously linked to relief as both transport rates are at a maximum for high relief basins and a minimum for low relief basins. To this influence is superimposed the effect of the climate which is responsible for the scattering of data, especially for the big rivers sample. This action is clearly shown when the relief factor disappears: for basins with similar reliefs Ts and Td are inversely related as Td is directly linked to runoff while Ts is inversely linked to runoff. This trend is the one observed by Judson and Ritter (1964) for US basins. This double influence of climate and relief has already been described by Corbel (1964), but as he did not select the rivers according to their size, no definite trend can be found when his Td values are plotted versus Ts values, despite a very broad sample of rivers. The size effect, particularly for Ts, surpasses the influence of relief and climate.

The Ts/Td ratio has been studied by Leopold et.al. (1964) who found a direct relation­ship with runoff. The same variation was observed by Langbein and Dawdy (1964). Once again the climatic influence was not observed on the sample of rivers studied here because it is completely masked by the relief. For example at a given runoff of 15 1. s-1 km-2, Ts/Td is around 10 for the River Ganges while it is only 0.3 for the Tapajos River in lower Amazo­nia. But it is evident that Ts/Td increases with relief (Fig. 3) as already noted by Alekin and Brazhnikova (1962).

As before the world range of Td versus Ts has been drawn up in Fig. 3. This range should be extended towards high solid transport when values for Chinese rivers, specially the Yel­low River, are reported. But these high 7s values (more than 2 000 t km-2 year-1) are excep­tions and most rivers are within the range presented in Fig. 3. The Ts/Td ratio varies from 0.1 for rivers in plains in semi-arctic regions up to 30 for rivers in semiarid regions.

WORLD IMPORTANCE OF DISSOLVED TRANSPORT

When the world balance of river material supplied to the oceans is considered there is about 5 times more suspended material than dissolved material. Alekin and Brazhnikova (1968) found a ratio of 4.9 and when the Livingstone (1963) and Holeman (1968) budgets are com-

* Regression (3b) is significant if the Amazon is taken instead of its tributaries.

276

t km'^ year" 1000

Dissolved transport

Ronge for all rive

+ L Range {or rivers

• PYvVj R°n9^ for rivers

-5000m'' s or A > 400000 k m '

-2 400000 km '

Judson and Ritter for the USA

1000 t km"2 year' Solid transport

Fig. 3 — Dissolved specific transport versus solid specific transport. (Author note: Value no. 10 for the Mackenzie has been wrongly plotted.)

bined this ratio is 5.0. But solid and dissolved transport are not evenly geographically distrib­uted over the continental surface. The mechanical transport is greatly affected by phenom­enal rates in southeast Asia, particularly for the Yellow River, this region contributing around 80 per cent to the world's solid discharge (Holeman, 1968). When the geographical distribu­tion of Ts and Td is studied for the sample of rivers considered here which covers a total area of 48.5 X 106 km2, that is 48 per cent of the world's surface where river transport is effect­ive, the dissolved transport is dominant for 35 per cent of this area, for 25 per cent the Ts/Td ratio is between 1 and 2, and for 40 per cent it is more than 2. For this computation the Amazon basin has been split into sub-basins. This geographical importance of Td is certainly greater as for most of the major rivers the solid transport is only dominant in the upper basin. If the continental surface were divided into unit areas of 100 000 km2 , the dissolved transport would be for most of the time the major form of material transport by the rivers. As an example the Soviet Union has an average Ts transport much lower than the world average which is influenced by record rates. This territory covers a broad range of morpho-climatic features and can be taken at a first estimate as representative of the world continent­al area, except southeast Asia, and Makarenko et al. (1970) have estimated that subsurface

277

chemical denudation exceeded mechanical denudation for 64 per cent of the USSR area, i.e. except for the area of the latest tectonic movements. This confirms the assumption made by the author.

TYPOLOGY OF MATERIAL TRANSPORT BY MAJOR RIVERS

Table 2 shows the material transport by the major rivers. These rates correspond to average values for extended areas of more than 400 000 km2. When it was possible the basins of the continental rivers have been divided into more homogeneous sub-basins. This typology is only semi-quantitative as were the previous ones of Alekin and Brazhnikova (1960) and Corbel (1964). No attempt was made to compute quantitative relationships between Td and environmental factors such as Jansen and Painter (1974) did for Ts. This was for several rea­sons: (1) quality data on the world's rivers are not yet sufficiently reliable and accurate, (2) environmental factors such as mean relief or average precipitation are not always well known, (3) the sample is too hetero geneous, i.e. the rivers chosen were generally too big to correspond to a single morphoclimatic environment, thus the data did not represent a well-defined environment but were weighted averages for basins exposed to many different influ­ences. Any attempt to quantify chemical transport by rivers should be made on smaller more homogeneous basins but not ones so small that local environmental factors are eliminated. A basin size of 100 000 km2 should be convenient.

DISSOLVED MINERAL MATTER CARRIED TO THE OCEANS

It is possible to estimate from the sample considered the total dissolved mineral matter car­ried by the rivers to the oceans. The continental area drained by rivers flowing into the oceans is generally estimated to be 101 X 106 km2 (Alekin and Brazhnikova, 1960) and the total runoff to be around 36 300 km3 year-1. Glaciated areas have not been taken into ac­count. If we disregard the Volga, Amu-Daria and Chari Rivers which flow to inland bodies of water the sample of rivers-including Finland, Quebec and Japan-represents a total dis­charge of 15 900 km3 year-1 for a total area of 48.5 X 106 km2, that is 44 and 48 per cent of the world values. The average runoff value of the sample is 10.4 1. s_1 km - 2 for the whole drainage to the oceans, so the sample of rivers can be considered as well representative of the world's river runoff. There is still an exception for the Indonesia and New Guinea archipel­agos, which are not represented in our sample and are assumed to have high rates of dissolved discharge (Meybeck and Carbonnel, 1975).

The total dissolved transport for the sample is 1 540 X 106 tyear-1 and this corresponds to a specific transport of 32 t km - 2 year-1. As suggested by Alekin and Brazhnikova (1960) specific transport and drainage area were extrapolated rather than dissolved content and dis­charge. Thus the total dissolved discharge to the ocean—major ions plus silica—is estimated to be around 3 250X 106 t year-1. If a runoff value of 36 300 km3 is taken, the average sal­inity of rivers would be around 89 mg lr1. Previous budgets by Clarke (1924), Alekin and Brazhnikova (1960, 1968), Livingstone (1963), and Corbel (1964) are presented in Table 3. The comparison is sometimes difficult because of different methods of computation and various values for world drainage area and runoff. The values from this study are very close to those of Corbel, computed by the same method but based on only 12 rivers. The Soviet authors computed their balance on the basis of Td determinations for various morphoclim­atic environments in the USSR extrapolated to the whole world. Their estimate is similar to that of the author unlike Livingstone's value which is slightly higher. However there seems general agreement for a mean value of dissolved transport in rivers flowing to the oceans of

278

o •s:

e-o S

w 1-1 sa <

S

o

?3

(A.

£ 5

Cd

api

CL c C. C3

c3 ^

Q 3 6 < 6 •a

o

O .-H

r-M CO

CN ^ H

O O

o o

y Q c_

i

^ rt o

2 -r -r -r O O O

o o m o co

o o ~? " • } <—1

o op CD

-*

O

f i n i - H

o o o o ( N —< S 2 2

o

s i 3 S

° & G j ' 3 <u

S B

3.2 3 3 2'E. c •-*

^oun

tai

iw p

rec

SX

o

5 o *o 0 *-* + J 4> cd

as. < s

tD

imat

"u

ô >> •s Î*_T OJ

i

- i

o <+~r _0> "o>

O h J

<D

imat

"o

bar>

^ =*H"

o 13

O J

CD

imat

"o

trop

ic

t*-T _0J

13 o

o .s J —

279

30 to 35 t km^ 2 year - 1 . It will now be very difficult to improve this computation as the number of rivers to be taken into account will increase greatly: if the 10 greatest rivers re­present about 32 per cent of the world discharge, the 40 rivers next in size will only increase this proportion to 50 per cent.

TABLE 3 World mineral dissolved discharge to the oceans

Author Specific

dissolved transport Total

discharge [x 106 tyear-ij

Comments

Clarke

Alekin and Brazhnikova

Livingstone

Corbel

This work

36

31.4

35.6

37.2

31.6

32

3 700

3 160

3 600

3 760

3 250

Silica included; COJ converted into HCO3 Without silica; COjj" converted into HCO3" Including silica according to Livingstone's estimate Silica included; recomputed from Livingstone's data Silica included; 12 major rivers Silica included; based on 44 per cent of world river runoff and 48 per cent of the world continental area drained by rivers All budgets have been computed taking 101 x 106 km2 as the world continental area drained by rivers.

CONCLUSIONS

Dissolved transport by rivers has been sometimes underestimated as regards solid transport. It is true that as a whole rivers carry to the ocean 5 times more material in the solid form than in the dissolved form, but chemical transport is a major process that is geographically as important as solid transport. On a sample of very different rivers representing about 48 per cent of the earth's surface exposed to erosion by running waters (glaciated and arid areas being excluded), chemical transport dominates solid transport in at least 35 per cent of the area of the basins.

Chemical transport is much more regular than solid transport and there is only a varia­tion of dissolved transport rates of two orders of magnitude as compared to three orders of magnitude for the solid transport rates. The maximum dissolved transport rate of 500 t k m - 2

year-1 was observed for small alpine rivers. As for solid transport this maximum rate is in­versely related to the basin size and when large basins are considered it decreases to 200 t km- 2 year-1 for middle-sized basins (A= 100 000 km 2) and to 140 t k m - 2 year - 1 for major rivers (A > 400 000 km2). There is no phenomenal rate of dissolved transport (more than 1 000 t km- 2 year - 1 ) as there is for mechanical transport by rivers in semiarid countries or by glaciers. Maximum rates are always observed in mountainous regions such as the Hima­laya, Andes or the Alps. Minimum rates are noted for semiarid regions where runoff is very low, in tropical rainy regions where salinity is a minimum and in semi-arctic plain regions where both salinity and runoff are low.

When only major rivers are considered there are some significant relationships:

(1) Salinity S is inversely related to runoff:

S = aqb with -Kb<0 (1)

280

(2) dissolved transport Td is directly related to runoff: Td = aqc with 0 < c < l (2)

(3) dissolved transport Td increases with solid transport 7s: Td = lqm with Q<m<\ (3)

(4) Ts/Td is highly variable according to climate and relief but in any case Ts/Td in­creases when 7s increases. When every river is taken into account, whatever its size, there is a wide scatter of data, but as a whole these trends are still observed.

Although relating dissolved transport to environmental factors such as mean relief, rain intensity, vegetation cover has not been attempted, some major trends clearly appear. Relief and climate, as exemplified by runoff, are the major factors. Climate dominates the salinity variation and relief the dissolved transport variation. Wide scattering of data is due to the effect of the other factor—relief or climate respectively. When both dissolved and solid transport are concerned, relief clearly becomes the major factor. Total river transport is a maximum in mountainous regions and a minimum in plains especially under semi-arctic conditions. This importance of relief implies that even for chemical transport the upper part of a river basin greatly affects the river's quality downstream, and for this reason it is sometimes difficult to interpret the average quality and transport rates for inhomogeneous basins. The choice of rivers used to study the dissolved transport may influence the relation­ships found and the major trends in some countries can be secondary ones on a world-scale basis.

Finally it must be pointed out that as present river transport rates are widely used by geomorphologists and geochemists, some important restrictions must be remembered:

(1) Only a fraction of the dissolved matter supplied by rivers to the ocean comes from chemical denudation.

(2) This material can be trapped in the estuarine system, as for silica and other elements (Martin et ai, 1970; Kharkar et al., 1968). Though these regions only represent a small part of the whole oceanic area, they can affect the material balance between land and ocean for these elements.

(3) Present transport rates are determined for a very unusual geological period charac­terized by important eustatic variations, tectonic movements and climatic oscillations that differ from most of the past geological conditions. Furthermore the erosion process is not always at equilibrium as in some places both dissolved and solid transport rates may be high­ly dependent on past climatic or morphologic conditions.

(4) There have been wide changes of the continental area, of the pétrographie composi­tion of the surface rocks exposed to erosion as well as of the chemical composition of sedi­mentary rocks through geological time, (Ronov, 1972; Ronov etal., 1970) that restrict the extrapolation of present data.

(5) The study of present denudation and transport rates should be extended by concen­trating on middle-sized homogeneous basins (A = 100 000 km2) where major environmental factors can be well defined rather than on the world's biggest rivers. If a broad range of cli­matic and morphologic features is studied on various rivers a quantification of dissolved transport can be realized that will constitute a basis to compute past denudation rates with paleogeographic features.

Acknowledgements. The author is most grateful to R. Laaksonen, G.K. Seth, J.M. Martin, P. Hubert, J.F. Jarrige, M. Grellières and Mrs E. Benedetti-Crouzet for communication of unpublished analyses and for their precious water samples that were analysed at the Centre de Recherches géodynamiques de Thonon-les-Bains, France.

281

REFERENCES 1. Ackermann, W.C., Harmeson, R.H. and Sinclair, R.A. (1970) Some long-term trends in water quality

of rivers and lakes. Trans. Amer. Geophys. Un. 51 , no. 6, 516-522. 2. Alekin, O.A. and Brazhnikova, L.V. (1960) A contribution on runoff of dissolved substances on the

world's continental surface (in Russian). Gidrochim. Mat. 32, 12-34. 3. Alekin, O.A. and Brazhnikova, L.V. (1962) The correlation between ionic transport and suspended

sediment (in Russian). Dokl. Akad. Nauk SSSR 146, no. 1, 203-206. 4. Alekin, O.A. and Brazhnikova, L.V. (1968) Dissolved matter discharge and mechanical and chemical

erosion. Geochemistry, Precipitation, Evaporation, Soil Moisture, Hydrometry-Proceedings of the General Assembly of Bern, vol. IV, pp. 35-41 : IAHS Publ. no. 78.

5. Anderson, V.G. (1941) The origin of dissolved inorganic solids in natural waters with special refer­ence to the O'Shannassy River catchment, Victoria. Austral. Chem. Inst. J. Proc. 8, 328-350.

6. Blanc, P. and Conrad, G. (1968) Evolution géochimique des eaux de l'Oued Saoura (Sahara nord-occidental). Rev. Géogr. Phys. Géol. Dyn. 10, no. 5, 415-428.

7. Bonnetto, A., Dioni, W. and Pignalterri, C. (1969) Limnological investigations on biotic communities in the middle Parana River Valley. Verh. int. Ver. Limnol. 17, 1035-1050.

8. Borland, W.M. (1961) Sediment transport on glacier-fed streams in Alaska. /. geophys. Res. 66, no. 10, 3347-3350.

9. Brunskill, G.J. et al. (1975) The chemistry, mineralogy and rates of transport of sediments in the Mackenzie and Porcupine rivers watersheds, NWT and Yukon 1971-73. Technical report no. 566, Fisheries and Marine Service, Environment Canada.

10. Carbonnel, J.P. and Meybeck, M. (1975) Quality variations of the Mekong river at Pnom Penh, Cambodia, and chemical transport in the Mekong basin. J. Hydrol. XXVII, 249-265.

11. Clarke, F.W. (1924) Data of geochemistry, Vth edition. US Geol. Survey Bull. 770, 841 pp. 12. Coche, A.G. (1968) Description of physico-chemical aspects of Lake Kariba, an impoundment in

Zambia-Rhodesia. Fish. Res. Bull. Zambia 5, 200-267. 13. Coleman, J.M. (1968) Deltaic evolution. In Encyclopedia of Geomorphology (edited by R.W.

Fairbridge): Reinhold Book Corp, New York, 1265 pp. 14. Corbel, J. (1959) Vitesse de l'érosion. Z. Géomorphologie 3, 1-28. 15. Corbel, J. (1964) L'érosion terrestre. Etude quantitative. Ann. géographie 398, 385-412. 16. Curtis, W.F., Culbertson, J.K. and Chase, E.B. (1973) Fluvial sediment discharge to the oceans from

the conterminous United States. US Geol. Survey Circ. 670, 14 pp. 17. Depetris, P.J. and Griffin, J.J. (1968) Suspended load in the Rio de la Plata drainage basin. Sediment-

ology 11,53-60. 18. de Villiers, P.R. (1962) The chemical composition of the Orange River at Violsdrif, Cape Province.

Ann. Geol. Survey, Pretoria V, no. 1, 198-206. 19. Douglas, I. (1964) Intensity and periodicity in denudation processes with special reference to the

removal of material in solution by rivers. Z. Géomorphologie 8, no. 4, 453-472. 20. Durum, W.H., Heidels, G. and Tison, L.J. (1960) Worldwide runoff of dissolved solids. Surface wat­

ers-Proceedings of the General Assembly of Helsinki, pp. 618-628: IAHS Publ. no. 51 . 21. FAO (1971) FAO Fisheries Circular no. 134. 22. Feth, J.H. (1971) Mechanisms controlling world water chemistry: evaporation-crystallization process.

Science 172, 871-872% 23. Fournier, F. (1960) Débit solide des cours d'eau. Essai d'estimation de la perte en terre subie par l'en­

semble du globe terrestre. Land Erosion, Precipitation, Evaporation-Proceedings of the General As­sembly of Helsinki, pp. 19-25: IAHS Publ. no. 53.

24. Frenette, M. and Larinier, M. (1973) Some results of the sediment regime of the St. Lawrence River. Proceedings of the 9th Canadian Hydrology Symposium on Fluvial Processes and Sedimentation, Edmonton.

25. Gibbs, R.J. (1967) The geochemistry of the Amazon River System, Part 1 : The factors that control the salinity and the composition and concentration of the suspended solids. Geol. Soc. Amer. Bull. 78, 1203-1232.

26. Gibbs, R.J. (1970) Mechanism controlling world water chemistry. Science 170,1088-1090. 27. Gibbs, R.J. (1972) Water chemistry of the Amazon River. Geochim. cosmochim. Acta 36, 1061-1066. 28. Gorham, E. (1961) Factors influencing the supply of major ions to inland waters with special refer­

ence to the atmosphere. Geol. Soc. Amer. Bull. 72, 795-840. 29. Grove, A.T. (1972) The dissolved and solid load carried by some west African rivers: Senegal, Niger,

Benue, and Chari. J. Hydrol. XVI, no. 4, 277-300. 30. Holeman, J.N. (1968) The sediment yield of major rivers of the world. Wat. Resour. Res. 4, no. 4,

737-747. 31 . Hurst, H.E. (1957) The Nile: Constable, London, 331 pp. 32. Janda, R.J. (1971) An evaluation of procedures used in computing chemical denudation rates.

Geol. Soc. Amer. Bull. 82, no. 1, 67-80.

282

33. Jansen, J.M.L. and Painter, R.B. (1974) Predicting sediment yield from climate and topography. J.Hydrol. 21,371-380.

34. Judson, S. and Ritter, D.F. (1964) Rates of regional denudation in the United States. / . geophys. Res. 69, no. 16, 3395-3401.

35. Kharkar, D.P., Turekian, K.K. and Bertine, K. (1968) Stream supply of dissolved silver, molybdenum, antimony, selenium, chromium, cobalt, rubidium and cesium to the oceans. Geochim. cosmochim. Acta 32, 285-298.

36. Kobayashi, J. (1960) A chemical study of the average quality and characteristics of river waters of Japan. Ber. Ohara Inst. Landwirt. Biol. 11, no. 3, 313-357.

37. Konovalov, G.S. (1970) The dynamics of rare and trace element content and their drainage from the USSR territory into the sea basins. Proceedings of the 1st Symposium on Hydrogeochemistry and Biogeochemistry, Clarke Cy, Washington, DC, 1973, pp. 606-612.

38. Laaksonen, R. (1971) Quality of inland waters in Finland. Aqua Fennica 1971, 47-57. 39. Langbein, W.B. and Dawdy, D.R. (1964) Occurrence of dissolved solids in surface waters in the

United States. US Geol. Survey Prof. Paper 501 D, D115-D117. 40. Langbein, W.B. and Schumm, S.A. (1958) Yield of sediment in relation to mean annual precipitation.

Trans Amer. Geophys. Un. 39, no. 6, 1076-1084. 4L Leopold, L.B. (1962) Rivers. Amer. Sel 50, no. 4, 511-537. 42. Leopold, J.B., Wolman, M.G. and Miller, J.P. (1964) Fluvial Processes in Geomorphology: W.H.

Freeman, San Francisco and London, 522 pp. 43. Liepolt, R. (1967) Limnologie des Donau: Schweizerbart'sche Verlagbuchhandlung, Stuttgart. 44. Lisitzin, A.P. (1972) Sedimentation in the world ocean. Soc. Econ. Paleo. Min., SpecialPubl. 17,

Tulsa, Oklahoma. 45. Livingstone, D.A. (1963) Chemical composition of rivers and lakes. Data of geochemistry. US Geol

Survey Prof. Paper 440 G, G1-G64. 46. Makarenko, F.A., Zverev, V.P. and Kononov, V.I. (1970) Subsurface chemical runoff in the USSR

area. Proceedings of the 1st Symposium on Hydrogeochemistry and Biogeochemistry, Clarke Cy, Washington, DC, 1973, pp. 567-573.

47. Martin, J.M., de Groot, A.J. and Kulbicki, G. (1970) Terrigeneous supply of radioactive and trace elements to the ocean. Proceedings of the 1st Symposium on Hydrogeochemistry and Biogeochemis­try, Clarke Cy, Washington, DC, 1973, pp. 463-483.

48. Meade, R.H. (1969) Errors in using modern stream load data to estimate natural rates of denudation. Geol. Soc. Amer. Bull. 80, no. 7, 1265-1274.

49. Meybeck, M. (1972) Bilan hydrochimique et géochimique du Lac Léman. Verh. int. Ver. Limnol. 18, 442-453.

50. Meybeck, M. and Carbonnel, J.P. (1975) Chemical transport by the Mekong River. Nature, Lond. 255, 134-136.

51. Ministère des Richesses Naturelles, Province de Québec. Annuaire Hydrologique, Qualité des Eaux. 52. Moore, G.T. (1969) Interaction of rivers and oceans-Pleistocene petroleum potential. Amer. Ass.

Petrol. Geol. Bull. 53, no. 12, 2421-2430. 53. Muller, G. (1964) Zur Hydrochemie von Alpenrhein und Seerhein. Gas und Wasserfach 105, 810-814. 54. Parde, M. (1953) La turbidité des rivières et ses facteurs géographiques. Rev. Geogr. Alpine 41 , 399-

421. 55. Reynolds, R.C. and Johnson, N.M. (1972) Chemical weathering and temperate glacial environment

of the Northern Cascade mountain. Geochim. cosmochim. Acta 36, 537-554. 56. Roche, M.A. (1975) Geochemistry and natural ionic and isotopic tracing; two complementary ways

to study the natural salinity regime of lake Chad. / . Hydrol. XXVI, 153-171. 57. Ronov, A.B. (1972) Evolution of rock composition and geochemical processes in the sedimentary

shell of the earth. Sedimentology 19, no. 3-4, 157-172. 58. Ronov, A.B., Migdisov, A.A. and Yaroshevsky, A.A. (1970) The main stages of geochemical history

of the outer shells of the earth. Proceedings of the 1st Symposium on Hydrogeochemistry and Bio­geochemistry, Clarke Cy, Washington, DC, 1973, pp. 40-53.

59. Schumm, S.A. (1954) The relation of drainage basin relief to sediment loss. Land Erosion, Precipita­tion-General Assembly of Rome, vol. I, pp. 216-219: IAHS Publ. no. 36.

60. Simaika, Y.M. and Sherbini, H.A. (1957) Some aspects of erosion in Egypt. Land Erosion, Instru­ments, Precipitation-General Assembly of Toronto, vol. I, pp. 381-386: IAHS Publ. no. 43.

61. Soviet Academy of Science (1964) Fisiko-Geograficheskiy Atlas Mira. Translated in Soviet Hydrol, 1965,6,5-6.

62. Spronck, R. (1941) Mesures hydrographiques dans la région divagante de bief maritime du fleuve Congo. Mem. Inst. roy. Colon, sect. Sci. Technol. 3, no. 1, 56 pp.

63. Stoddart, D.R. (1971) World erosion and sedimentation. In Introduction to Fluvial Processes, chap­ters I-III: Methuen, London.

64. Strahler, A.N. (1971) The earth sciences. In Systems of Fluvial Denudation, chapter 36 : Harper and Row, New York.

283

65. Strakhov, N.M. (1967) Principles of Lithogenesis, vol. 1: Oliver and Boyd, Edinburgh. 66. Symoens, J.J. (1968) La minéralisation des eaux naturelles. Hydrobiological Survey of Lake Bang-

weulu and Luapula River basin, II, 1, Cercle Hydrobiologique de Bruxelles. 67. Tabuteau, M. (1960) Etude graphique pour les conséquences hydro-érosives du climat méditerranéen.

Ass. Géogr. Français Bull. 294-295, 130-142. 68. Turekian, K.K. (1969) Oceans, streams and atmosphere. In Handbook of Geochemistry (edited by

K.H. Wedepohl), chapter 10: Springer Verlag. 69. UNESCO (1969 and 1970) Discharge of Selected Rivers of the World, vol. I-III, series on Studies and

Reports in Hydrology. 70. UN ECAFE (Committee for Coordination of Investigations of the Lower Mekong River Basin) (1962-

1966) Lower Mekong Hydrologie Yearbooks, Bangkok. 71 . UN ECAFE (1968)/4 Compendium of Major International Rivers in the ECAFE Region: Water Re­

sources series no. 29. 72. US Geological Survey, Quality of surface waters of the United States. In US Geol. Survey Water

Supply Papers. 73. van Denburgh, A.S. and Feth, J.H. (1965) Solute erosion and chloride balance in selected river basins

of the western conterminous United States. Wat. Resour. Res. 1, no. 4, 537-541. 74. Viro, P.J. (1953) Loss of nutrients and the natural balance of the soil of Finland. Communications,

Instituti Forestalls Fenniae 42, no. 1, 1-45. 75. Weiler, R.R. and Chawla, V.K. (1969) Dissolved mineral quality of the Great Lakes waters. Proceed­

ings of the 12th Conference on Great Lakes Research, pp. 801-818: Int. Ass. Great Lakes Res. 76. Zverev, V.P. (1971) Hydrochemical balance of the USSR territory. Dokl. Akad. Nauk SSSR 198,

no. 1, 161-163. 77. Zverev, V.P. and Rubeikin, V.Z. (1970) The role of atmospheric precipitation in circulation of chem­

ical elements between atmosphere, hydrosphere and lithosphère. Proceedings of the 1st Symposium on Hydrogeochemistry andBiogeochemistry, Clarke Cy, Washington, DC, 1973, 613-620.

Author Note Since the manuscript for this paper was submitted a book called The World Water Balance by A. Baum-gartner and E. Reichel has been published (Elsevier 1975). The data of these authors on river discharges and drainage areas are very similar to those presented here except for the Yang Tse Kiang (Q = 35 000 m 3 s-1) , Ganges (0= 15 500 m 3 s"i) and Si Kiang (£?= 11 000 m 3 s-i, ,4 = 435 000 km^).

284