Previous PageTable Of ContentsNext Page


Hydrological analysis of two sub-catchments of the Mareb River (Eritrea)


Roberto Colombo and P. Sarfatti

Istituto Agronomico per l'Oltremare, Firenze, Italy

Abstract

The aim of the present study is the evaluation of surface water resources in Eritrea, by means of remotely sensed data. The study area consists of two sub-catchments of the Mareb river, Shiketi and Emni-Tzellim, situated on the Eritrean highland. These have different morphological and land cover features, while they can be considered quite homogenous from a climatological point of view. They are both closed off by an earth dam, built for irrigation purposes.

A simple model, proposed by the United States Soil Conservation Service (US-SCS), has been utilized to estimate watershed runoff volume. The appropriate Runoff Curve Number has been attributed using, for some factors, remotely sensed data together with "conventional" data. The annual water budget for the two reservoirs has been calculated by taking into account the water inflow, measured by the watershed runoff volume, and the losses, due to evaporation and infiltration. The results have been compared with sedimentation rates of the two reservoirs and with water requirements for irrigation.

Résumé

Le but du présent travail a été l'évaluation des ressources hydriques superficielles, avec l'aide de la télédétection, de deux bassins versants du fleuve Mareb dans les hautes plaines de l'Erythrée : les bassins de Shiketi et d'Emni-Tzellim. Ces deux bassins se différencient par la géomorphologie et par la couverture de leurs sols mais présentent des similarités climatiques. Ils sont fermés en aval par deux barrages.

Un modèle, proposé par le Service de la conservation des sols des Etats-Unis (US-SCS), a été utilisé pour estimer l'écoulement annuel des deux bassins versants. Pour certaines unités de paysage, le numéro des courbes de ruissellement (Runoff Curve Number) a été déterminé en utilisant la télédétection. Pour d'autres unités, la méthodologie "traditionnelle" du Département de l'agriculture des Etats-Unis (US-DA) a été conservée. Les apports annuels dans les deux barrages ont ainsi pu être calculés. Les résultats de cette estimation ont été comparés avec la capacité des réservoirs en tenant compte de leur taux de sédimentation. De plus, la prise en compte des besoins en eau d'irrigation a permis de fixer la superficie actuellement irrigable.


Introduction

FIGURE 1

Study area

 

In order to allow an expansion of irrigation in the highlands, a programme of small dams construction has been undertaken in the past ten years. At the time of dams construction, estimates of catchment yield was not possible and the optimum reservoir capacity could not be determined (FAO, 1994), because of information about water resources, which is a main constraint for planning agricultural development in Eritrea. This has resulted, in some cases, in rapid sedimentation of the reservoir where the catchment was too large for the dam, or in the reservoir not filling during the rainy season if the catchment was too small.

Surface runoff depends on a great number of factors, such as rainfall characteristics, watershed morphometric characteristics, soil physical characteristics (depth, texture, structure, hydraulic conductivity), land cover, land use, and soil moisture conditions prior to rainfall events. There are many well known methods for predicting surface runoff either from empirical formulas or from deterministic models (Hudson, 1971).

In this work the Curve Number method of the US-SCS has been utilized in two different watersheds in Eritrea, testing the potential contribution of satellite data (SPOT XS Nov. 1993 and Landsat TM Nov. 1987) and aerial photos (1:50 000, Nov. 1964) to runoff estimation.

The study area is part of a larger area which had been previously surveyed and mapped with an integrated approach, describing in detail climate, geology, geomorphology, soils and land cover (Ongaro and Sarfatti, 1995). Remotely sensed data have been used to estimate biomass (Viti et al.; 1995) and to calculate soil erosion (Colombo et al., 1995).

Study area

The study area is composed of two sub-catchments of the Mareb river, situated on the Eritrean highland: Shiketi and Emni-Tzellim (figure 1).

Climate and Land Cover

The two watersheds have almost the same climate. Asmara (2360 m asl) is the nearest climatological station, located about 30 km north of the study area. Mean annual temperature is 16.3 °C and annual potential evapotranspiration is 770 mm. Mean annual rainfall is 541 mm. There are two rainy seasons and a long dry season. The so-called small-rains (113 mm) fall between April and June; the so-called great-rains (381 mm) from July to the beginning of September; during Fall and Winter there is a long period with very scarce rainfall (47 mm).

Land cover is extremely variable due to differences in morphology and land use. Natural vegetation is mainly an open formation of shrubs and small trees of Acacia etbaica, utilized for grazing and for fuelwood collection. Percentage of vegetation cover is very poor, ranging from 5 to 30%. Main crops are wheat, barley, teff, maize, sorghum cultivated with traditional low inputs techniques (animal traction). Irrigated agriculture (mainly horticultural crops) is confined to small areas. Most of the Shiketi catchment is covered by natural vegetation, while in the Emni-Tzellim catchment arable land is about 70% of the surface.

Both sub-catchments are affected by severe erosion (Colombo et al., 1995).

Geomorphology and Soils of the Shiketi catchment

The catchment area is delimited between the coordinates of 15°12'04'' and 15°10'10'' latitude, and of 38°50'43'' and 38°51'45'' longitude, covering approximately 585 ha. Elevation is between 2050 and 2300 m asl. Some morphometric parameters are in Table 1.

TABLE 1

Some morphometric parameters of the two sub-catchments.

Parameters SHIKETI EMNI-TZELLIM

Basin area (he)

585 1172

Basin order (Strahler's system)

5 5

Drainage pattern

dendritic subdendritic

Drainage density (km/km2)

4.8 4.8

Bifurcation ratio

3.9 3.5

Relief ratio

0.059 0.025

Gravelius Index

1.2 1.7

Concentration Time (min)

55 130

The Shiketi catchment includes Late Proterozoic rocks and Cainozoic volcanic products, separated by a lateritic formations. The catchment is mainly represented by units of denudational and structural origin with steep to very steep rolling to hilly topography. Rock falls, debris slides, and gullies are the main active processes. Structural terraces in lateritic formation reproduce the oldest peneplain. Valley bottom is characterized by alluvial terraces buried under a variable thick of colluvial sediments.

On the volcanic plateau, on colluvial areas and on laterite there are shallow soils, very gravelly loam to very gravelly sand, well to excessively drained (Leptosols). On volcanic slopes and escarpments there are moderately deep to deep soils, very gravelly loam to very gravelly sand, well drained (Regosols). On alluvial terraces and fans there are moderately deep to very deep soils, gravelly sand to gravelly loam, well to somewhat excessively drained (Fluvisols). On cradle valleys there are deep black soils, clayey, moderately well to poorly drained (Vertisols). On the alluvial plain there are moderately deep soils, gravelly clay to gravelly loam, well to poorly drained (Cambisols in association with Fluvisols and Leptosols).

Geomorphology and Soils of the Emni-Tzellim catchment

The catchment area is delimited between the coordinates of 15°02'58'' and 15°01'01'' latitude , and of 38° 42' 47'' and 38° 47' 06'' longitude, covering approximately an area of 1172 ha. Elevation is between 2 000 and 2 550 m asl. Some morphometric parameters are in Table 1.

The Emni-Tzellim catchment includes Cainozoic volcanic products and a thick sequence of welded pyroclastic flows with interbedded lava flow. The catchment is represented by units of volcanic denudational origin, with steep terrain, and by units of alluvial-colluvial origin with relatively flat to gently sloping topography. Rock falls, gullies, sheets and rills erosion are the main active processes. Accumulation glacis are originated above structural volcanic terraces. The old sheetflood plain is buried under the dissected accumulation glacis and residual alluvial terraces, remnant of an older fluvial landscape, outcrop in the Tzellim plain.

On volcanic escarpments there are shallow to very shallow soils, very gravelly loam to very gravelly sand, well drained (Leptosols in association with Regosols). On volcanic slopes and footslopes there are moderately deep to deep soils, very gravelly loam, well drained (Regosols). On footslopes there are deep soils, loam to sand, well drained (Fluvisols). On older alluvial terraces and on sheetflood areas there are moderately deep to deep soils, gravelly clay to clay, poorly drained (Vertisols). Distributed over the whole area there are moderately deep to deep soils, gravelly loam, moderately to poorly drained (Cambisols).

Runoff estimation

The Soil Conservation Service (SCS) method (US Dept., 1985) has been used to predict the total volume of runoff that may come from a watershed during a design flood (25-year return period) and to predict the total annual runoff volume for daily rainfall, during a period of 28 years.

The SCS Runoff Curve Number (CN) is an empirical description for infiltration and rainfall excess. The SCS runoff equation is:

[1]

where Q = runoff (mm); P = rainfall (mm); Ia = initial abstraction (mm); S = potential maximum retention after runoff begins (mm).

By removing Ia as an independent parameter:

[2]

runoff is:

[3]

The parameter S is related to soil and cover conditions of the watershed through the Curve Number:

[4]

where CN = Curve Number

Runoff volume of the catchment has been calculated using the following formula:

[5]

where Qv = runoff volume (m3); Q = runoff depth (mm); A = catchment area (ha)

The SCS-CN method has been applied to estimate daily and yearly surface runoff volume.

Remote sensing data (aerial photographs and satellite data) have been used, in conjunction with other types of data (field data, laboratory analysis, bibliography) to estimate the Curve Numbers.

Determination of Runoff Curve Number

The Shiketi and the Emni Tzellim sub-catchments have been classified into homogeneous portions of land (land units) by satellite image processing and analysis and aerial photographs interpretation (Table 3).

To select the appropriate CN the following factors have been considered: cover type, hydrologic condition, hydrologic soil group, impervious area.

Cover type and hydrologic condition

As in other applications of the SCS-CN in developing countries (Purwanto and Donker, 1991), also in this case it was difficult to determine the correspondence between land cover types and CN, since the original method is based on land cover types common in North America. Adapting the original tables to local conditions four main categories of land cover have been taken into consideration (Table 2).

TABLE 2

Runoff curve number for main hydrological soil cover-complexes of the study area (for AMC II)

Cover type Hydrologic condition Curve Number (CN) for hydrologic soil group
    A B C D
1 Degraded areas (veg. cover < 5%) Poor 77 86 91 94
2 Open shrubland (mainly Acacia etbaica) Poor 68 79 86 89
3 Irrigated agriculture (furrow irrigation) Poor 72 81 88 91
4 Rainfed agriculture (teff, barley, wheat, sorghum)   65 76 84 88

 

TABLE 3

Runoff curve number (CN) of land units

  SHIKETI     EMNI-TZELLIM  
  Land Unit CN   Land Unit CN
1 Alluvial Terraces 77 1 Alluvial plain 88
2 Cradle Valley 88 2 Alluvial Terraces 88
3 Alluvial fan 68 3 Lower Accumulation Glacis 76
4 Basement footslope 86 4 Higher Accumulation Glacis 76
5 Volcanic footslope 91 5 Sheetflood Plain 88
6 Volcanic Plateau 75 6 Cradle Valley 88
7 Plateau Escarpment 86 7 Structural Volcanic Escarpment 84
8 Volcanic high hills 82 8 Volcanic Ridge 90
9 Volcanic low hills 82 9 Structural Terraces on Slope 76
10 Surface of plantation in Laterite 96 10 Dissected Accumulation Glacis 84
11 Laterite Structural terraces 96      
12 Basement Escarpment 84      

Hydrologic condition indicates the effects of cover type and treatment on infiltration and runoff and is generally estimated from density of plant and residue cover on sample area. The average percentage of vegetation cover has been estimated for each land unit using SPOT Vegetation Index (VI):

[6]

where VI = SPOT Vegetation Index; B2 = SPOT Band 2; B3 = SPOT Band 3

The relationship between SPOT Vegetation Index and percentage of cover has been calculated in a preceding work (Viti et al., 1995):

[7]

where C = Cover (%) and VI = SPOT Vegetation Index

The cover percentage estimated for each land unit, was less than 50%, corresponding to poor hydrologic condition (Table 2); also irrigated areas have been considered with poor hydrologic condition since these are very small plots mixed with other types of land use (fallow, rainfed agriculture).

Hydrologic soil group

Hydrologic soil group depends on soil physical characteristics such as soil texture, soil structure, hydraulic saturated conductivity, soil depth and sealing susceptibility (US Dept., 1993). For each land unit, it has been possible to assign the appropriate hydrologic soil group by using, in conjunction, field data and laboratory analysis, following the US-SCS guidelines.

Soil structure and soil depth have been estimated during field survey; soil texture was determined by laboratory analysis; hydraulic conductivity was determined by bibliographic tables relating soil structure with soil texture (Landon, 1984); sealing susceptibility was calculated using the following formula (FAO/UNEP/UNESCO, 1980):

[8]

where Ss = Sealing susceptibility; Sf = Percentage of fine silt; Sc = Percentage of coarse silt; Cl = Percentage of clay; OM = Organic matter (%)

By using Table 2 it has been possible to evaluate the CN for the assigned hydrological soil group.

Impervious area

Rockiness has been considered as connected impervious area. Percentage of rock outcrops has been estimated in the field for each land unit. For some land units, such as laterite terraces and volcanic hills, CN values have been corrected in relation of the percentage of impervious area using adjustments graphics (US Dept., 1986).

Finally, Curve Numbers of each land unit have been estimated (Table 3).

Daily runoff

The following factors have been considered in order to predict the volume of runoff for the maximum 24-hours rainfall with 25-years return period: daily rainfall, Curve Number and Antecedent Moisture Condition.

Mean duration of a storm is four hours and maximum duration is about seven hours (Griffiths, 1972); July and August daily rainfall distribution is mostly concentrated in a range between 5 to 10 mm. Rainfall intensity has been estimated by the formula of Fletcher (1950): rainfall intensity for 1 hour interval is 27 mm/h. The following formula (Gumbel, 1954) has been used to estimate the return period of the maximum 24-hours precipitation:

[9]

where T = return period in years; N = total number of statistical events; m = rank of events arranged in descending order of magnitude.

Using daily rainfall data of 28 years (Fantoli, 1966), the annual maximum precipitation with a 25-years return period is 102 mm, and it has been used as P in [3].

The Runoff Curve Number has been calculated for each land unit, assuming the type II of Antecedent soil Moisture Condition (AMC II). The CN adjustments table has been used in order to adjust the CN for wet (AMC III) and dry (AMC I) soils (Wanielista M. P.,1990).

Watershed CN (Table 4) has been obtained weighting CN values of each land unit (Table 3), in function of their specific area.

TABLE 4

Curve Number (CN), Storage at saturation (S), Initial abstraction (Ia), Runoff (Q) and Runoff Volume (Qv) obtained with a rain of 102 mm/24 h (AMC II & III)

Catchment CN S (mm) Ia (mm) Q (mm) Qv (m3)
  AMC II AMC III AMC II AMC III AMC II AMC III AMC II AMC III AMC II AMC III
Shiketi 82 92 56 22 11 4.4 56 80 327600 468000
Emni-Tzellim 81 91 59 25 12 5 54 77 632880 902440

Annual runoff

When surface runoff is to be stored in reservoirs, the total runoff volume for a period of several months, usually the annual volume, is of more interest than the runoff for a design storm.

For each month the average Antecedent Moisture Condition (AMC) has been calculated, obtaining: October to June type I; July and September type II; August, type III. The CN values have been consequently corrected in relation with the monthly AMC, before calculating the daily runoff. The annual runoff volume has been obtained adding daily runoff, obtained by equation [3], of each month of the year. Annual runoff volume has been calculated for a period of 28 years, using equation [5]. Mean, maximum, minimum and 75% probability of water volume inflow into the reservoir have been calculated (Table 5).

TABLE 5

Annual volume inflow into the reservoirs.

Annual Volume Inflow (Qv) Unit Shiketi Emni-Tzellim
Initial Reservoir capacity m3 256000 170269
Mean volume inflow m3 557550 939960
Maximum volume inflow m3 1397617 1671806
Minimum volume inflow m3 66416 38998
Volume 75% probability m3 299078 561461

Comparing results of annual runoff estimation with reservoir sizing it comes out that Shiketi dam is well dimensioned, while Emni Tzellim reservoir is far behind the potential water storage. A dynamic annual water balance of the reservoirs has been calculated in order to evaluate the effective water volume available for irrigation.

Sedimentation rate into the two reservoirs (Colombo et al., 1995) has been utilized to estimate their effective volume. Losses due to evaporation and infiltration from the reservoirs have been estimated roughly 10% of the total volume. The irrigable area has been calculated considering an irrigation requirement of 10 000 m3 of water pro hectare (Table 6).

TABLE 6

Reservoirs capacity and irrigable area

Reservoir Capacity Unit Shiketi Emni-Tzellim
Dam construction year 1983 1987
Initial Reservoir Capacity m3 256 000 170 269
Sedimentation Rate % year 3 11
Actual Capacity m3 156 000 0
Actual Available water volume m3 year 141 000 0
Capacity after 10 years m3 180 000 0
Initial Irrigable Surface he 25 17
Irrigable Surface after 10 years he 18 0

TABLE 7

Parameters and methodology considered to compute CN

SCS-CN Factors Parameters used for estimation Method
    Field data Laboratory analysis Emp. formula and bibliog. Aerial photos Satellite images
Hydrologic Soil Group Soil Texture X X      
  Soil Structure X        
  Soil Depth X        
  Topsoil Sealing susceptibility   X X    
  Soil Hydrolog. conductivity X X X    
Cover type Land use and land cover X     X X
Management Cultural practices X        
Hydrologic condition Vegetation cover X       X
AMC Daily rainfall     X    
Impervious area Rockiness X     X  

 

Conclusions

The SCS-CN method has been applied to estimate daily and yearly surface runoff volume. Remotely sensed data have given a helpful contribution to CN approach (Table 7): land unit map has been derived from remotely sensed data, hydrologic condition has been obtained by SPOT Vegetation Index.

As a confirmation, we have calculated the soil water balance after Thornthwaite (1948), assuming an available water capacity of 100 mm, obtaining a surface runoff of 91 mm/year; this figure can be considered in good agreement with data estimated by the SCS-CN method (95 mm/year for Shiketi and 84 for Emni-Tzellim). In the present case, the CN method has proved to be an useful tool for runoff estimation.

A future development is the validation of results with a series of ground measurements. After such a campaign it will be possible to extend the results to other catchments of the Eritrean highland and to supply watershed management projects with reliable data.

Bibliography

Colombo, R., Martucci, A., Rodolfi, G., Sarfatti, P., Kahsay, T. and Yohannes, A. 1995. Stima dell'erosione del suolo nell'alto bacino del fiume Mareb (Eritrea). Atti VII Convegno Nazionale AIT, Chieri (TO): 349-353.

Fantoli, A. 1966. Contributo alla climatologia dell'altopiano Etiopico - Regione Eritrea. Ministero degli Affari Esteri-Cooperazione Scientifica e Tecnica, Roma.

FAO/UNEP/UNESCO. 1980. Méthode provisoire pour l'évaluation de la dégradation des sols. FAO, Rome.

FAO. 1994. Eritrea. Agricultural Sector Review and Project Identification. TCP/ERI/2353, FAO, Rome (3 Vol.).

Fletcher, R.D. 1950. A Relation between Maximum Observed Point and Areal Rainfall Values. Trans. Am. Geophys. Union, 31: 344-348

Griffiths, J.F. 1972. Climates of Africa. World Survey of Climatology, vol 10. Edited by J. F. Griffiths. Elsevier Publishing Company, Amsterdam-London-New York.

Gumbel, E.J.1954. Statistical Theory of Extreme Values and Some Practical Applications. Applied Mathematics Series 33. US Bureau of Standards, Washington, DC.

Hudson, N. 1971. Soil Conservation, Cornell University Press, Ithaca, New York.

Landon, J.R (Editor) 1984. Booker Tropical Soil Manual. Booker Agriculture International Ltd, London.

Ongaro, L. and Sarfatti, P. 1995. Utilizzazione dei dati telerilevati e GIS per la produzione della carta delle Unità di Terre del bacino dell'alto Mareb (Eritrea). Atti VII Convegno Nazionale AIT, Chieri (To): 361-366.

Purwanto, E. and Donker, N.H.W. 1991. Semi-Distributed Hydrologic Modelling of the Humid Tropical Upper Cimandiri Catchment (West Java) Using HEC-1 Model. ITC Journal, 4: 241-253.

US Department of Agriculture. 1985. Soil Conservation Service: National Engineering Handbook. Section 4-Hydrology. Washington, DC.

US Department of Agriculture. 1986. Soil Conservation Service: Urban Hydrology for Small Watersheds. Technical Release 55. National Technical Information Service, Springfield, VA.

US Department of Agriculture. 1993. Soil Survey Division Staff: Soil Survey Manual N.18, Washington DC.

Thornthwaite, C.W. 1948. An Approach a Rational Classification of Climate. The Geographical Review, 38:55-94.

Viti, M.L., Delli, G. and Ongaro, L. 1995. Stima della biomassa legnosa dell'alto bacino del Mareb (Eritrea) utilizzando dati SPOT. Atti VII Convegno Nazionale AIT, Chieri (TO): 499-503.

Wanielista, M.P. 1990. Hydrology and Water Quantity Control. John Wiley & Sons, Inc.

Previous PageTop Of PageNext Page