22]

THE MORPHOLOGY AND DOWNSTREAM HYDRAULIC GEOMETRY

RELATIONS OF ALLUVIAL STREAM CHANNELS IN A HUMID

TROPICAL ENVIRONMENT, SOUTHWESTERN NIGERIA

Dr. Fola S. Ebisemiju,

Department of Geography,

Ondo State University,

P.M.B. 5363, Ado-Ekiti,

Ondo State, Nigeria

ABSTRACT The study of the morphology of alluvial channels in the

Elemi River Basin, southwestern Nigeria, reveals that stream channel

capacity is only moderately related to stream discharge indexed by

basin area, but is more strongly controlled by bank cohesion which

accounts for 65% of the variance. Analysis of the downstream hydraulic

geometry of the streams reveals that, while the coefficients and expo­

nents are consistently high and closest to 1.0 for small ephemeral

headwater streams, both decrease in the middle and lower reaches of

larger streams and vary appreciably between streams of different

sizes. The relationship between basin area and channel capacity is

curvillinear (negative exponential) with two threshold basin areas at 2 2

about 20 km and 100 km . The observed downstream changes in the hyd­

raulic geometry relations are attributed to downstream changes in flow

regime and increase in environmental heterogeneity as drainage basin

area increases. The spatial variation in the downstream hydraulic geo­

metry relations between streams is also strongly related to the varia­

bility in the channel perimeter cohesion index. These findings suggest

that (a) site specific rather than areal factors are the main deter­

minants of channel morphology in the area, (b) truly consistent and

predictable relationships between channel size and basin area exist

only along streams with homogeneous catchment and hydrological charac­

teristics; and (c) the results of small catchment experiments cannot

be meaningfully extrapolated to larger basins because of greater vari­

ability in channel perimeter stability, catchment characteristics, and

flow regime.

222

Introduction

A stream channel adjusts its morphology to optimize transport of water

and sediment within the constraints imposed by stream discharge,

channel bed slope, grain size, and channel bank stability (Carson

1986). It has been suggested, therefore, that an understanding of

spatial variations in channel form can facilitate the estimation of

channel processes from channel forms (Hedman 1970), and point up the

ways in which changes can occur over time (Rango 1970). Investigations

of the spatial variations of channel geometry within a drainage basin

also provides a basis for identifying the patterns of channel form

which exist and also for indicating the magnitude of adjustments which

may have been achieved by the stream system or the modifications

induced by man. Adjustments of river channels are complex; the rate,

magnitude, and direction of response, therefore, may vary not only

with regard to time but more importantly to different parts of the

river system. In order to understand the complex response of a river

system, therefore, it is necessary to evaluate the significance of the

spatial patterns in the magnitude, rate, and direction of response

along a single river, and to explain the observed variations between

river systems.

A cursory look at literature on hydrogeomorphology reveals dearth of

data on channel forms, processes, hydraulic geometry relations, and

channel form responses and changes in the humid tropics (Gupta 1984).

The need for such studies, cannot be over-emphasized in view of the

increasing awareness that the results of hydrological and hydrogeomor-

phic investigations conducted in humid temperate and other regions

cannot be meaningfully applied or extrapolated to humid tropical en­

vironments in view of the significant differences in the energetics

and dynamics of fluvial systems, the environmental conditions specific

to each area, and the varying geological and geomorphic evolution of

their landscapes.

This paper attempts to bridge this information gap by analysing and

discussing the spatial patterns of hydraulic geometry relations and

channel morphological controls in a humid tropical environment.

223

The Study Area

The investigation was conducted within the 276 km Elemi River Basin

in southwestern Nigeria (Fig.l).

The Basin is underlain by three metamorphic rock groups : the migma-

"tite-gneiss in the eastern part, charnockitic rocks in the west-

central areas, and the Older Granites which occupy the northern,

western and southern parts (Fig.l). The area can be divided into three

erosional surfaces. The lower erosion surface which occupies the

central and eastern parts of the basin and develops across the migma-

tite-gneiss rocks, is a gently undulating to flat land which slopes

gently from north to south, and occurs at elevations of 380 m -4l0 m

above sea level. The slopes, with angles hardly exceeding 3 ~5 . are

generally very long and are frequently interrupted by ruwares and in­

dented by a few intermittent stream channels. This plain is separated

from a higher, well-dissected surface at 530 m - 606 m by an irregular

scarp about 130 m high; the scarp is well pronounced in the north and

west of the basin. A few massive inselbergs are associated with the

older Granites and charnockites in the western margin of the basin.

The main tributaries of River Elemi originate in the higher erosion

surface and descend the scarp through joint-aligned channels. The

heads of first-order valleys are dambos, and narrow floodplains, less

than 800 m wide, occur along most of the streams which are incised to

a depth of 2 m to 5 m-

The area falls within the seasonably wet humid tropics with a mean

annual rainfall of 1267 n»n and a uniformly high mean monthly tempera­

ture of 25 C. The rainy season which lasts from March to November is

characterised by high rainfall intensity and erosivity. The primeval

forest vegetation has been almost completely removed through cultiva­

tion, lumbering, and urbanization, and is found today only on the

steep slopes of the scarps and inselbergs, and along some stream cour­

ses. Fallow vegetation and cultivated land dominate the landscape, and

the fallow vegetation is shrub dominated by the fast-growing herba­

ceous plant eupatorium odoratum.

224

Methodology

The conventional hydraulic geometry approach (Leopold & Maddock 1953)

in which a channel morphometric parameter is correlated with a process

variable (usually stream discharge) is employed in this study. Due to

the lack of stream discharge recourds, drainage basin area above

channel cross-section was employed as a surrogate for stream dis­

charge .

Bankfull channel cross-sections were surveyed along the principal

stream of River Elemi and its 28 tributaries. In all 423 channel

cross-sections were measured and the shoulder width, average depth,

channel capacity, hydraulic and form ratio of each cross-section were

computed. Data were also collected on parameters controlling perimeter

resistance to the stresses applied by flowing water and its sediments.

Bank cohesion was indexed by four variables: % silt/clay, soil shear

strength, bank vegetation density index, surface incrustation index.

The two important vegetation properties that influence channel bank

cohesion and stability are plant roots and cover. Hey & Thome (I986)

used a vegetation code (1-4) as a dummy variable to define increasing

bank vegetation density in Colorado, U.S.A. In drainage basin studies,

Melton (1957) proposed a bare area index which is perhaps a useful

index of cover in arid and semi-arid regions, while Ebisemiju (1979)

evaluated tree density through stereoscopic examination of a network

of 2 mm - square grids (0.0064 km ) superimposed on a 1:40000 -scale

aerial photographs. In order to avoid the subjectivity inherent in Hey

and Thornes method, the line intercept method widely used by biogeo-

graphers was employed to provide data on plant cover density. The

number of plant stems intercepted by the tape stretched along channel

bank was divided by the total length of the three lines spaced at 1 m

intervals. Soil samples were collected at 50 cm intervals along the

central line, and analysed for percentage weight of plant roots, per­

centage silt/clay (#SC), percentage sand (#S), and percentage latérite

gravels. Soil shear strength (SSR) was measured by a proving ring cone

penetrometer. The product of the % weight of plant roots and plant

cover density (VI) was employed to index the degree of channel bank

225

cohesion and stability induced by the presence of vegetation. The mean

% lateriate gravel in the soil samples collected along each cross-

section provide data on the degree of surface incrustation (SI). A

perimeter cohesion index (PCI) was then obtained from the ratio of the

product of %SC, SSR, VI and SI to % sand. Simple correlation and step­

wise multiple regression analyses were then performed to evaluate

channel morphological and hydraulic geometry relation controls in the

area.

Data Analysis and Results

Fig. 2 reveals a moderately strong positive relationship between

channel capacity and basin area (r=0.6556) which is significant at the

0.001 level. The scattergram and the allometric equation, however,

highlight two important features of this downstream relation. Firstly,

the exponent term in the equation is extremely low and not signifi­

cantly different from zero (b = 0.03*0- This suggests that although

stream discharge explains about ^3# of the downstream variation in

channel capacity. The average rate of change of channel capacity rela­

tive to increase in basin area or stream discharge along the principal

stream of River Elemi is extremely low. Dury (1973) in his study based 2 2

on drainage areas varying from 1 km to 232 km in the Ouse-Twin

basins reached a similar conclustlon: "the rate of downstream increase

in width and depth on the Ouse-Twin, and thus of cross-sectional area,

is distinctly less than theory predicts". Secondly, and more import­

antly, the exists two threshold basin areas at which marked changes

occur in the static allometric relationship. For channel cross-sec­

tions with contributing areas below 20 km , the correlation coeffi­

cient is 0-955 and the exponent of 0.429 means that at this spatial

scale channel capacity increases as a function of basin area raised to

the 2.33 power. In middle reaches with moderate sized basins (20 - 100

km ), Fig. 2 reveals some wide scatter of points, and the moderately

low correlation coefficient of 0.5^5 and coefficient of determination

of 30# suggest that factors other than basin area or discharge are

more important in controlling channel size at this spatial scale. In

the lower reaches with larger catchment areas above 100 km , there are

wide scatter of points as well as some clustering. There is no rela­

tionship whatever between channel capacity and discharge (r=0.083),

226

and the regression equation is not significant. These findings suggest

that the downstream relationship between channel capacity and dis­

charge is curvillinear (negative exponential) rather than linear, that

the relationship cannot be meaningfully described by a single regres­

sion equation but three, and that the exponent 0.034 is an average of

spatially varying allometric relations along the principal stream of

River Elemi.

Analysis of the downstream hydraulic geometry of twenty-eight tribu­

taries of River Elemi reveals considerable variations in the correla­

tion coefficients and the exponents of the relations between channel

capacity and basin area (Table 1). The regional pattern based on data

from these tributaries (n=432) is basically similar to that exhibited

by the downstream data for the prnicipal stream. For example, while

the coefficients and exponents are consistently high and closest to

1.0 for small headwater streams with catchment areas less than 10 km ,

both vary appreciably between the relatively larger catchments. Also

when channel capacity at the outlet point of each tributary stream was

related to basin area, thresholds or break points also occur at areas 2 2

of about 5 km and 10 km (Fig.3); the relationship is curvillinear

and can be best described by three rather than one regression equation

(Table 2).

Some previous studies carried out in humid temperate as well as arid

areas have shown that the hydraulic geometry relations do not conform

to a simple power function (Thornes 1970, 197*», Faulkner 1974, Klein

1976a, 1981). This has been attributed to downstream variations in

types of flow and the peakedness of discharge (Klein 1976b, 1981,

Gregory and Park 1976) and to environmental constraints on channel ad­

justment (Park 1978). These studies and the southwestern Nigeria case

study reported in this paper suggest, therefore, that the spatial loc­

ation of channel cross-sections within a stream network has an impor­

tant control on the morphology and therefore, the hydraulic geometry

relations in individual streams. Simple correlation and regression

analyses were performed, therefore, to examine the point-specific

factors controlling channel morphological variations in the area.

227

The analyses reveal that channel bank vegetation and the degree of

surface incrustation of channel perimeter exercise very strong con­

trols on channel sizes along the principal stream of River Elemi

(r=-0.721, - 0.822 respectively) and the correlation of the channel

perimeter cohesion index with channel capacity is -O.806. A multivari­

ate regression model with basin area and the channel perimeter cohe­

sion index as independent variables explains 89X of the spatial vari­

ations in channel capacity, 65% of being attributable to the channel

perimeter cohesion index. It is logical, therefore, to expect that

this index should exercise some control also on the hydraulic geometry

relations of individual streams.

A stepwise multiple regression model relating the exponents in the

allometreic equations for the twenty-eight tributaries to basin area,

relief ration, channel bed slope and the coefficient of variation of

the PCI reveals that, although all the independent variables exercise

varying degrees of control on channel allometric relations, the

spatial variations in the downstream hydraulic geometry of the streams

in the area is controlled primarily by the variability in the channel

perimeter stability index, which alone accounts for 75% of the vari­

ance in the exponent term of the allometric equations (Table 3)-

Discussion

Drainage basin area has been aptly described by Anderson (1957) as

'the devils' own variable because of the strong control it exercises

on the hydrological, morphometric and topologic attributes of streams.

It has been widely used, therefore, as a surrogate for stream dis­

charge where data on the latter are unavailable. Stream channel capa­

city should increase, therefore, in direct proportion to the down­

stream increase in basin area. Two major factors, however, are likely

to complicate the simple power function relationships expected between

catchment area and channel capacity. These are the changing nature of

stream flow in the downstream direction, and increase in environmental

heterogeneity as basin area increases downstream.

228

The effect of the nature of discharge on the allometric relation

between basin area and channel capacity has been summarized by Park

(1978) as follows:

"the value '(b) will be closest to 1.0 in the case of low

flows. This is because variations in low flow discharge per

unit area throughout the basin will be lower than those of

peak flow discharge per unit area. Peak discharge to which

channel capacity is adjusted, will be relatively reduced

with increasing size of catchment areas as a result of

flood-routing effects".

Consequently, the - b - values should be highest for small headwater

catchments and decrease downstream as peak discharge is reduced and

variations in low flow per unit area increases. As Fig. 4 reveals, the

- b - values decrease with increase in catchment area in the manner

postulated by Park (1978). It is pertinent here to note also that

Klein (1981) identified "change in the flow regime (peakedness) with

increasing drainage area (as) as key for an explanation of the change

in the - b - value" in this analysis of the variation of channel width

downstream.

Small headwater catchments are more homogeneous in terms of lithology,

soil, vegetation, land use, and the areal distribution of rainfall and

runoff. As catchment area increases in a downstream direction, the

degree of environmental homogeneity is likely to decrease even in

drainage basins with uniform lithology. This will be manifested in

downstream changes and variations in (a) flow regime, (b) the range of

channel types, (c) the variability of bank materials both along the

river and vertically within a bank, (d) the resistance of perimeter

sediments with the growth of bank vegetation and differences in ante­

cedent moisture conditions prior to a potential erosion event, (e) the

processes involved in bank retreat, and (f) in the calibre of stream

load related to variations in lithology along the stream, the nature

of materials brought into the main stream by tributaries, and from

upland sites with different land uses and therefore different rates of

sediment production and delivery to streams.

229

Since these factors do not change downstream in any predictable

manner, it is clearly difficulty to formulate a general model of their

influence on channel form. Thus as environmental heterogeneity increa­

ses with downstream increase in basin area, channel form and sizes

should exhibit increasing variability. The curillinear nature of the

regression line, and the wide scatter and clustering of points in the

middle and lower reaches of the river (Fig.2) suggest, therefore, that

stream channel capacity and the allometric relations between it and

basin area are controlled by point-specific rather than areal factors.

This has been confirmed by the stepwise multiple regression analysis

which identifies the perimeter cohesion index as the main control of

channel size, while the variability of the - b -values is strongly

related to the variability of the PCI along the individual streams.

In the study area, channel perimeter sediments vary considerably along

the major streams because of differences in geological formations and

soils. They also vary vertically within channel bank in the middle and

lower reaches of the larger streams which are deeply incised below

floodplains. Channel bank and flood plain vegetation types also vary

partly as a result of variations in land use on adjacent floodplains

and upland sites. It is not uncommon to find bare banks alternating

with grass or shrub-covered banks along a reach while the bank should­

ers are either lined with trees whose roots are exposed along channel

banks or covered with thick fallow vegetation. More importantly, is

the presence of indurated groundwater latérites on channel banks and

beds, especially in the middle and lower reaches of the larger streams

where their presence suggests that the concentration of iron in the

soil profile occurs at considerable depths below the ground surface.

As channels are deepened, the stream eventually cuts into the ground­

water latérite layer which, as a result of periodic exposure during

the dry seasons, become indurated. Although they are subsequently

broken down into blocks or disintegrate into lateriate gravels, they

impart considerably stability to channel banks and beds wherever they

occur. Their general absence in the headwater reaches is attributed to

the fact that channel incision in these areas has not yet reached the

groundwater latérite layer. The channels of headwater streams are cut,

therefore, into the upper horizons of the soil which are "exclusively

of hillwash origin, having no stones or gravel" (Smyth and Montgomery,

230

1962, p.2^8). Below this at depths exceeding about 2 m are ironpans or

soil with high content of concretions related to the iron-rich nature

of the underlying rocks in the area.

Where present, the indurated latérites inhibit channel widening or

deepening. Since they ocur irregularly along a given stream, the down­

stream increase in channel size is not in direct proportion to increa­

ses in drainage area. Their absence in the headwater areas which are

also characterised by a high degree of environmental homogeneity and

small variations in low flow discharge per unit area accounts for the

stronger correlation between channel capacity and basin area and the

higher - b - values in these areas.

Conclusions

The preceding analyses of data on the downstream hydraulic geometry of

streams in a humid tropical environment in southwestern Nigeria

suggest that stream channel capacity is not a simple power function of

stream discharge at the large spatial scale; the exponent value at

such a scale is an average for aggregated data and masks the true re­

lationship that exist at different spatial scales. At the small

spatial scale of headwater catchments with basin areas less than 20 2

km where drainage basins are homogeneous and streamflow is predomi­

nated by intermittent low flows with high peakedness, the relation of

channel capacity to basin area is strong, and the - b - value of 0.429

indicates a state of negative allometry, that is, the ratio of cannel

capacity to basin area decreases as catchment area increases. The cor­

relation is moderately strong for medium-sized catchments character­

ized by ephemeral flow. In the downstream reaches with peremial flow 2

and catchment area above 100 km however drainage basin area exercises

no control whatsoever on channel size. The exponent values in the all-

ometric relations between channel capacity and basin area, therefore,

decreases as basin area increases downstream. The hydraulic geometry

relation at the large spatial scale, therefore, is negative exponen-2

tial and, for streams with catchment areas about 300 km , it is best

described by a three line model rather than the widely-reported single 2

line power function, the break points being at basin areas of 20 km 2

and 100 km . Channel capacity is more strongly controlled by channel

231

perimeter cohesion than by stream discharge. Additionally, the varia­

bility of the - b - value between streams is strongly related to the

variability of perimeter cohesion.

These findings suggest that:

(a) site-specific rather than areal factors are the main determi­

nants of channel size in the study area;

(b) truly consistent and predictable relationships between channel

capacity and stream discharge exist only along streams with

homogeneous catchment and hydrological characteristics; and

consequently,

(c) the results of small catchment experiments cannot be meaning­

fully extrapolated to larger basins because of greater varia­

bility in channel bank stability, catchment characteristic,

and flow regime.

References

Anderson, H.W. (1957) Relating sediment yield to watershed variables.

Trans. AGU 38 (b), 921 - 924.

Carson, M.A. (1986) Transport of gravel in alluvial channels: a review

with special referance to the Canterbury plains. North canterbury

Catchment Board and Regional water Board, Christchurch, NE.

Dury, G.H. (1973) Magnitude frequency analysis and channel morphology.

In: Fluvial Geomorphology, Proceedings of the Fourth Annual

Geomorphology Symposium Binghamton, New York. 91-121.

Ebisemiju, F.S. (1979) Analysis of drainage density and similar para­

meters in relation to soil and vegetation characteristics.

Nig. Geog. J. 22 (1), 33-43.

Faulkner, H. (197*0 An allometric growth model for competitive

gullies. Zeit. fur Geomorph. Suppl. Bd 21, 16-87-

Gregory K.J. & Park, C.C. (1976) stream channel morphology in

Northwest Yorkshire. Rev, de Geomorph. Dyn. XXV, 63-72.

232

Gupta, A. (1984) Urban hydrology and sedimentation in the humid

tropics. In: Developments and Applications of Geomorphology

(ed. by Costa, J.E. & Fleischer, P.J.). 2*40-267.

Springer-verlag, Berlin.

Hedman, E.R. (1970) Mean annual runoff as related to channel geo­

metry of selected streams in California. USGS Water Supply

Paper 1999E.

Hey, R.D. & Thome, C.R. (1986) Stable channels with mobile gravel

beds. J. Hydraulic Engn. 112, 67I-689.

Klein, M. (1976a) Hydrograph peakedness and basin area.

Earth Surf. Processes 1, 27-30.

Klein, M. (1976b) The influence of drainage area in producing

thresholds for the hydrological regime and channel character­

istics of natural rivers. Working Paper, No. 1^7, Department

of Geography, University of Leeds.

Klein, M. (1981) Drainage area and the variation of channel geometry

downstream. Earth Surf. Processes and Landforms 6, 589~593-

Leopold, L.B. & Madduck, T. (1953) The hydraulic geometry of stream.

channels and some physiographic implications.

USGS Prof. Pap. No. 252.

Melton, M.A. (1957) An analysis of the relations among elements of

climate, surface properties and geomorphology. Office of Naval

Research, Geog. Brach, Project NR 389-042: Technical Report 11,

Columbia University.

Park, C.C. (1978) Allometric analysis and stream channel morphology.

Geogr. Analysis x, 211-228.

Rango, A. (1970) Possible effects of precipitation modification on

stream channel geometry and sediment yield.

Water Resour. Res. 6, 1765 - 1770.

Smyth, A.J. & Montgomery, R.F. (1962) Soils and Land Use in Central

Western Nigeria. Government Printer, Ibadan.

Thornes, J.B. (1970) The hydraulic geometry of stream channels in

the Xingu-Araguaia headwaters. Geog.J. 136 (3). 376-382.

Thornes, J.B. (1974) Speculations on the behaviour of stream channel

width. Discussion Paper, No. 49, Graduate School of Geography,

London School of Economics.

233

Correlation Coefficients and Exponents of Allo-

metric Equations of Bankfull Channel Capacity

versus Basin Area in Twenty-Eight Tributary

Streams of River Elemi, South western Nigeria.

Basin Area {km ) n r b

1.00

0.72

1.56

2.26

0.70

2.15

5.39 1.60

0.84

8.80

3-85

3-95

26.12

3-07

2.25

31.75

9.15

7-00

1.48

1.95

8.40

33.25

3-85

14.25

2.68

68.37

3.15

0.93

6

5

7

11

5 10

18

6

5

18

13

11

23

13

9

29

16

11

6

8

13

25

7

15

7

35

8

6

0.93*t

O.966

0.9^5

0.938

0.913

0.9W

0.892

0.904

0.937

0.889

0.953

0.948

O.672

0.819

0.923

0.608

0.793

0.887

0.935

0.961

0.903

0.663

0-992

0.715

0.986

0.687

0.921

0.956

1.015

0.975

O.893

0.931

1.003

0.834

0.818

0.917

0.938

0.938

0.893

0.875

O.613

0.981

1.123

0.489

0.888

0.913

0.945

0.956

0.875

0.567

0.972

O.815

0.958

0.415

0.933

0.945

234

Table 2. Allometric Equations of Bankfull Channel Capa­

city versus Basin Area for Three Spatial Scales.

Limits of Basin Areas (km ) n r a b

A.

B.

Tributary Streams

Less than 5

5 - 1 0

10 - 100

Regional Data

Less than 20

20 - 100

100 - 300

18

5

5

157

227

44

.9685

.5007

.4615

.9153

.6072

.0513

0.758

2.096

3-560

3.02

6.26

II.5I

0.819

0.239

0.052

0.501

0.082

0.003

Table 3- Summary of Stepwise Multiple Regression Analysis

of Factors Controlling the Exponents in the

Allometric Equations for 28 Streams, Elemi

Drainage Basin.

Step

1.

2.

3-

4.

Variable

Perimeter cohesion

Relief ratio

Basin area

Channel bed slope

r

0.866

0.803

0.674

0.759

Cumulative

R 2 W

75-00

90.33

96.73

97-08

Increase

75.00 ***

15-33 **

6.40 *

0.35 NS

*** Significant at 0.001 level

** Significant at 0.01 level

* Significant at 0.05 level

NS Not significant at 0.05 level

235

7°45'

- 7°40'

7°35

S°K)' 5°15 5°20

fig- 1- Geological formations and stream network, Etemi drainage basin, southwestern ÈUgeria.

236

20

10

8

6

4

2

1

- 8

-

-

-

(o)

(b) /

*/

1 1 I

«

/ *

•

• i i

. .{c)J^^.

n loi AH dolo 7 7

(b) Undtf 20kn?13

!c) 20-100km2 20 (a) 100-300kfl?44

i i i i i i t

_

• M) ••

• " , ' . V :> .—•

-

-r A B

•656 5-89 .034

•955-2-48 -429

.56 1 6-5 7 »0 60 ~_ • 051 11-51 -0 03 . i i i

20 40 60 80 100 200 300

Catchment Area (km2)

Fig-2 Relationship between chonnel capocity and basin area along the principal streom of River Bemi, Southwestern Nigeria.

C » 2.611A0-079

r = &671 n » 28

1 [ I I I I I ' i i i l l 8 10 20

_l 1 I I I U 60 eo loo

Basin Area (km2)

Fig-3- Relationship, between channel capacity at basin mouth and basin area for tributaries of River Eletni, Southwestern Nigeria-

K>

•8

•6

•4

i l

1

1 1. 1

- 0-9703

0

I I I .. 1 1

0

1 1 1

•

1 1 1

.

1 1

-

1 1

, 5 8 10 20 40 60 80 100 200 400

Catchment Area (km2)

Fig- 4 - Relationship between correlation coefficients of channel capacity versus basin area and downstream increase in catch­ment area, River Elerni, Southwestern Nigerio