Read PII: S0341-8162(02)00131-5 text version

Catena 50 (2003) 353 ­ 368

Geomorphological investigation on gully erosion in the Rift Valley and the northern highlands of Ethiopia

P. Billi a,*, F. Dramis b


` Dipartimento di Scienze della Terra, Universita di Ferrara, Ferrara, Italy b Dipartimento di Geologia, Universita di Roma 3, Roma, Italy

Received 28 March 2000; received in revised form 19 February 2001; accepted 30 March 2001

Abstract Gully erosion phenomena are very common in Ethiopia. They affect large areas with different morphological, pedological and climatic characteristics. The amount of soil loss due to gullying has become a very serious problem in the recent decades as it was associated to remarkable depletion of cultivated land. Field investigations on gully morphology and its genetic processes were carried out in two study areas of Ethiopia, representative of different geo-environmental conditions: the Lakes Region in the Rift portion north of Shashamene and the area surrounding the town of Mekele in Tigray. Two main types of gullies were identified on the basis of their morphological and hydraulic geometry characteristics: (1) discontinuous gullies which generally develop on low gradient slopes (1 ­ 5j on average) and the hydraulic radius of which increases from an upstream minimum to a maximum, at approximately their mid length, and decreases again to a relative minimum at their downstream end; and (2) stream gullies, formed by deep erosion processes typically migrating upslope. In order to investigate the main causes originating the different types of gullies, shear stress data were collected in the field from their hydraulic geometry. Hypotheses on the mechanisms responsible for both discontinuous and stream gullies development and for their different characteristics are discussed considering the pattern of shear stress variation in the downstream direction. D 2003 Elsevier Science B.V. All rights reserved.

Keywords: Gully erosion; Hydraulic geometry; Shear stress; Ethiopia

* Corresponding author. 0341-8162/03/$ - see front matter D 2003 Elsevier Science B.V. All rights reserved. PII: S 0 3 4 1 - 8 1 6 2 ( 0 2 ) 0 0 1 3 1 - 5


P. Billi, F. Dramis / Catena 50 (2003) 353­368

1. Introduction Gullying is one of the most important erosion processes which largely contribute to the sculpturing of the earth surface. The development of gullies has many negative impacts as it normally involves the loss and (in some cases) the deposition of a great amount of soil. For many countries, the loss of large soil masses by gully erosion often stands for the depletion of a basic natural resource. Moreover, the formation of gullies implies an alteration of overland flow, a shortening of runoff lag time and an increase in runoff volume. Once a gully is formed, it tends to develop further and this process is seldom inverted or halted naturally. This results in a substantial damage to the economy and may represent a relevant constrain to the development of poor countries. In spite of the many efforts made to understand the main factors and processes originating gullies, they are not yet well understood (Boardman, 1998). The first authors facing the problem of gully formation postulated a development from rills (e.g. Ireland et al., 1934), but gullies have different geomorphic and hydraulic features as reported by Oostwoud Wijdenes and Gerits (1994) and partly in this paper. Gullies can also develop from headward stream retreat affecting unincised, upstream slopes (De Oliveira, 1989). Soil tunnelling, too, is considered a common factor in triggering gully formation (Oostwoud Wijdenes and Gerits, 1994). A very important role is also played by man through disturbance to vegetation. Its removal by logging or cropland expansion, in humid areas, or by overgrazing, in semiarid zones, favours the development of gullies (Trimble, 1974; De Ploey, 1990). In fact, a sparse vegetation cover results in a diminished boundary roughness opposing a reduced resistance to overland flow. The erosive capability of runoff is therefore increased, while a deficit of organic matter in the soil decreases its aggregate stability. Unless the causes of gully formation and development can be discerned, any efforts in designing countermeasures to gullying will be frustrated. Gullies are probably initiated and develop through different, concurrent factors, though, we agree with Oostwoud Wijdenes and Gerits' (1994) statement that in all cases runoff response and sediment transport on slopes play a crucial role. According to Parsons et al. (1990) interrill flow tends to concentrate downslope into pathways leading to a more efficient flow and to a decrease in flow resistance. It follows that the interaction between overland flow and ground surface roughness is a process relevant to gully initiation (Poesen and Govers, 1990; Grissinger, 1996). As a gully forms, its development keeps on being controlled by different factors (Heede, 1976) including the relationship between flow and flow resistance which, in the end, are responsible for its morphological features. The analysis of their hydraulic geometry can therefore contribute to the understanding of the processes acting on the development of gullies. In order to investigate the effectiveness of hydraulic factors in gully development and shaping, the hydraulic geometry of 16 gullies has been measured and expressed in terms of spatial distribution of flow energy (i.e. downstream variation of boundary shear stress). These measurements were coupled by field observations on the local geomorphological setting and by direct inspection during intense rainstorms in the rainy season. The gullies studied are located in Ethiopia, in areas with different geo-environmental characteristics. The study sites are included in the Rift Valley bottom (Langano Lake), in its northern

P. Billi, F. Dramis / Catena 50 (2003) 353­368


margin, west of Ziway lake (Tora-Butajira area), and in the northern highlands, in a small area south of Mekele.

2. Description of the study areas The study area in the Rift Valley is located southeast of Butajira (Fig. 1) and comprises a few tens of kilometres wide belt stretching from the northern Rift margin to the great lakes in the Valley bottom. The physiography is given by a sequence of a few fault scarp benches, placed at different altitudes and connected by gentle slopes giving the area a large scale stepped morphology. Almost the entire area is underlain by Quaternary volcanic rocks (namely tuffs, volcanic ash, ignimbrites and lava flows) with the exception of a narrow belt around the lakes on the valley bottom where Quaternary lacustrine deposits outcrop. Extensive slope deposits are also very common on foothills. Below the elevation

Fig. 1. Location map.


P. Billi, F. Dramis / Catena 50 (2003) 353­368

of 1800 m a.s.l., the natural vegetation consists of scanty grass and acacia trees, reflecting the lack of sufficient rainfall characterising the Rift Valley floor (Fig. 2a). The land is cultivated mainly around the lakes while on the surrounding hills animal husbandry prevails. At higher elevations the larger quantity of rainfall (Fig. 2b) favours a much denser vegetation cover. Almost all the land is cropped, though animal husbandry is carried out as well. The study sites in the northern highlands are located between the village of Adi Gudum and the town of Mekele. Here Mesozoic (mainly Jurassic) sedimentary rocks outcrop. They consist of clays, marls, limestones and sandstones, while scattered dolerite domes punctuate the area. Slope deposits are common as well. Faults, pertaining to the main rifting system, have dislocated the area into block structures bounded by steep fault scarps. The very sparse vegetation reflects the semi-arid conditions affecting the area (Fig. 2c). In the study site animal husbandry is common and the cultivated land consists of rainfed crops of grains. The water scarcity is as marked as in the Rift Valley bottom. In both these areas, decades of overgrazing and uncontrolled exploitation of forestry resources seem to have influenced the water resources depletion and the erosion processes intensification (Ethiopian Mapping Authority, 1988). In the Rift Valley study area, a neat differentiation can be observed in both the annual and monthly rainfall distribution between the ``lowland'' of the valley floor and the upland of the rift margin. The rainfall data measured at Ziway (1650 m a.s.l.) and Butajira (2000 m a.s.l.) gauging stations were considered respectively (Fig. 2a and b). The annual precipitation of Ziway amounts to 769 mm, while at Butajira it is 1194 mm. In Ethiopia, traditionally, the rainfall distribution is subdivided into three periods: the dry season (from October to February), with almost no rain; the short rains, from March to May (locally known as Belg--Ethiopian Mapping Authority, 1988) and the summer long rains, from June to September (known as Keremt--Ethiopian Mapping Authority, 1988). At Butajira, two relative monthly peaks are recorded: the first is in March with 135 mm, while the second, corresponding to the yearly maximum, occurs in July with 171 mm. The driest month is November with only 10 mm. At Ziway, Belg and Keremt are not so well distinct as the monthly values gradually increase from the lowest value of November (1.7 mm) to the annual maximum recorded in July with 145 mm. In Mekele (2070 m a.s.l.) the precipitation is concentrated in July and August since 70% of the annual rainfall occurs during these two months, while the Belg is characterised by very small rain amounts (Fig. 2c). The rapid expansion of gullied areas that seems to have occurred in Ethiopia during the recent decades has been associated to deforestation and overgrazing (Sagri,1998), but also a change in the rainfall regime was postulated. Data records referred to the last 3­ 4 decades seem to indicate that Belg rain amounts tend to increase, while those of Keremt show a decreasing pattern for several Ethiopian stations (Billi and Dramis, 1999, 2000). An increase in Belg rains may have some relevance in favouring gully development since this precipitation occurs after the long dry period, when soil dissection cracks are at their maximum extension (tunnelling prone soils) and vegetation cover is at its minimum. Moreover, though Belg precipitation is smaller than Keremt, their maximum intensities for hourly values (i.e. the shortest time interval available for the study areas from the Ethiopian Meteorological Agency) are of the same order of magnitude, i.e. around 40 and 43 mm hÀ 1 at Ziway and Mekele, respectively.

P. Billi, F. Dramis / Catena 50 (2003) 353­368


Fig. 2. Monthly rainfall distribution for the study areas. Ziway is on the shore of the homonymous lake on the Rift Valley floor. Butajira is located near the northern Rift margin, while Mekele is on the northern Ethiopian highlands.


P. Billi, F. Dramis / Catena 50 (2003) 353­368

3. Field observations on gully morphology In Ethiopia, gullies can be found everywhere, no matter the climate, the top soil characteristics, the physiography or the lithology of the substratum. Normally, they develop on low gradient slopes that seldom exceed 10j (1 ­5j on average). Two main types of gullies were recognised. They have some similarities with those described by De Oliveira (1989). The first type resembles the discontinuous gullies of Leopold et al. (1964), but commonly were found as individual, isolated gullies that entirely develop within a single slope stretch. They originate at a highly variable distance from the divide (as observed by Heede, 1976) and reach their maximum depth at a variable distance from their downslope end. Width is not uniform, as assumed by Poesen and Govers (1990), since it may either constantly increase in a downstream direction or, less commonly, reach a maximum around its mid length to decrease again to a relative minimum at the downstream end. The 10 discontinuous gullies considered in this study are from a few tens to more than a hundred metres long (Table 1). Their upstream edge may be very narrow (less than 1 m) and not deeper than a couple of metres. They have a width/ depth ratio typically comprised between 1 and 3 and the distinction based on this parameter and high-intensity low-frequency and low-intensity high-frequency rainstorms proposed by Poesen and Govers (1990) does not seem to apply here. In their upstream half, these gullies become deeper and wider downstream and rectangular in cross-section, with almost vertical banks, but in places where collapsed bank material locally gives rise to trapezoidal channel geometry. In the downstream half width/depth ratio may either increase or be constant around an average value, though bank failure is not so extensive as

Table 1 Main morphological features of the study gullies Gully name Discontinuous gullies Gully 1 Gully 3 Gully 5 Gully 9 Agoddo1 Agoddo2 Langano North Mt. Lanfuru Lamgano South Kile Stream gullies Maikei Tora 2 Ejersa Lele Kedida Kedida North Hondolesa Ltot 113 36 19 20 55 54 33 90 35 30 J 0.0026 0.1843 0.0682 0.1117 0.0570 0.0550 0.0488 0.5022 0.0194 0.0330 H 1.35 1.29 0.59 0.75 0.98 1.94 0.64 1.43 0.83 0.44 H Sdev 0.23 0.7 0.15 0.33 0.36 0.49 0.22 0.69 0.33 0.1 W 7.9 1.67 1.22 1.33 5.43 5.88 0.97 0.83 1.81 0.88 W Sdev 2.14 0.97 0.39 0.56 1.23 0.96 0.33 0.37 0.52 0.20

127 60 76 38 37 28

0.0291 0.0393 0.0145 0.1104 0.0366 0.0734

1.79 2.32 0.62 0.88 1.23 0.47

0.26 0.46 0.21 0.30 0.50 0.10

5.30 1.76 2.20 0.94 1.12 0.69

1.36 0.67 1.31 0.20 0.41 0.14

Ltot = gully total length (m); J = average bed gradient; H = average depth (m); H Sdev = standard deviation of gully depth (m); W = average width (m); W Sdev = standard deviation of gully width (m).

P. Billi, F. Dramis / Catena 50 (2003) 353­368


in the upstream half. The slope gradient of the discontinuous gullies studied varies between 0.046 and 0.093, while their average bed gradient is commonly lower (0.003 ­ 0.184). The second type of gully investigated in this study are named stream gullies. Their characteristics (Table 1) are very similar to those observed for the upstream portion of discontinuous gullies but for the size and the fact that they pertain to a well-established river system. Like slope gullies, river gullies originate as a small incision at a variable distance from divide and grow downstream with depth and width increasing at an almost constant rate. Their width/depth ratios range between 1 and 4, but there is no evidence of a clear downstream trend. Bank failure is common and the collapsed material is removed by the flow that normally has energy in excess to entrain and transport all the supplied sediment. The stream gullies studied developed on slopes with almost the same gradient (range 0.003 ­ 0.084) of discontinuous gullies, but the bed gradient of stream gullies (range 0.014 ­0.123) is commonly higher than slope gradient. This suggests that the stream gullies studied were nothing but former slope gullies captured by upslope growing of river network headwaters (i.e. continuous gullies in the sense of Leopold et al., 1964). Both slope and stream gullies enlarge by headward expansion and widening and in places they may coalesce to form wide belts of severely degraded land. Field observations indicate that, in the study areas, gullies develop on different settings and under different conditions: lithological substratum, geomorphology, soils with variable characteristics and depth, annual precipitation, vegetation cover, land use, etc. This led to consider the hydraulic processes (i.e. the flow capacity to entrain and transport sediment) as the main factors in providing both discontinuous and stream gullies with their actual geomorphological characteristics.

4. Field methods The slope gradient, the bed gradient and the hydraulic geometry of 16 gullies were measured in the field. The hydraulic geometry was obtained by cross-section surveying at 1- to 4-m interval along the entire length of discontinuous gullies and down to stretches where bankfull condition could be clearly identified for stream gullies. Tributary junction sectors and reaches extensively affected by recent bank failure were skipped. The bed gradient was measured by levelling at each cross section or, more often, for individual reaches with uniform gradient within the same gully. By these data, at each cross-section it was possible to calculate the hydraulic radius and the shear stress, assuming a condition of bankfull flow. This approach was justified by the fact that both the discontinuous and the stream gullies studied experienced bankfull discharge during high intensity rainfalls. This was ascertained by direct inspection during a summer field campaign and confirmed by local people. Shear stress was calculated by the classical Chezy formula: s ¼ cRS where S is the energy slope (assumed equal to bed gradient); R is the hydraulic radius and c is the specific weight of water. The specific weight of water was kept constant and


P. Billi, F. Dramis / Catena 50 (2003) 353­368

assumed equal to 1000 kg mÀ 3 as no data on its variation is available. Moreover, since the relative variation of shear stress was considered (see next section), keeping the specific weight of water constant does not imply a relevant bias to the analysis proposed here. In order to compare gullies with different size, dimensionless expressions for shear stress and downstream distance were used. Dimensionless shear stress is defined as T ¼ s0 =smax where s0 is the actual shear stress calculated at a given cross-section and smax is the highest value of shear stress obtained for the same gully. The dimensionless downstream distance was instead defined as L ¼ Lp =Ltot where Lp is the distance from the gully headcut and Ltot is the total length of the gully.

5. Data analysis The downstream variation of shear stress for the gullies studied is reported in Figs. 3 and 4. For discontinuous gullies (Fig. 3), shear stress tends to increase up to a maximum, typically coinciding with the gully midpoint, beyond which it again decreases to a minimum at the gully mouth. Although the diagram of Fig. 3 shows some scatter, all

Fig. 3. Downstream variation of shear stress for 10 discontinuous gullies. T and L are dimensionless shear stress and distance, respectively (see text for their definition). In the diagram, the best fit curve (solid line) and the boundary curves (dashed line) are reported as well.

P. Billi, F. Dramis / Catena 50 (2003) 353­368


the data are distributed within a curved belt and also the data of individual gullies follow a parabolic function whose general, best fit equation is T ¼ 1:36L2 þ 1:24L þ C with the constant C ranging from 0.2 to 0.8. Gully morphology and, consequently, shear stress distribution reflects the erosion and deposition conditions observed in the field. In fact, in the upstream half of slope gullies erosion processes prevail as excess flow energy is dissipated through gully upslope migration, bottom scouring and sediment transport. Flow energy is therefore decreased in the downstream portion and in-channel deposition of part of the material supplied from upstream takes place. The resulting decrease in bed gradient and the relative increase in flow resistance causes the flow to spread laterally and, in places, induces the widening of the downstream end of discontinuous gullies. Bank height is normally smaller at the gully mouth and the material coming from bank failure is relatively limited compared to that entrained in the gully upstream and middle portions. The latter can be considered the main sediment source in a discontinuous gully. Beyond its downstream end, the lack of flow confinement, the increase in flow resistance and the decrease of sediment transport capacity lead to further deposition with sediment accumulating in elongated lobes or, more commonly, in fan-shaped splays ahead of the gully mouth. Part of the discontinuous gullies studied did not show significant variation of bed gradient, while others are composed by different reaches with distinctive gradients. Schematically, three main reaches can be distinguished: (1) from the upstream end to about the gully midpoint we have the highest bed gradient that can be associated to scouring processes; (2) the following reach is commonly very short and characterised by intermediate values of bed gradient. Here, boundary material entrainment and deposition are more or less in equilibrium (transport reach); (3) the downstream reach is normally

Fig. 4. Downstream variation of shear stress for six stream gullies. T and L are dimensionless shear stress and distance, respectively (see text for their definition). R is the correlation coefficient.


P. Billi, F. Dramis / Catena 50 (2003) 353­368

longer than the intermediate (but shorter than the upstream one), has the lower bed gradient and depositional processes prevail. As a result, the downstream portion of a discontinuous gully is generally more stable than the upstream and middle reaches. In these latter reaches, in fact, bank failure is rather common since bottom scouring results in high, unstable banks that can be also undermined by fast, shallow flows during the receding phases of the gully flooding. Fig. 4 reports the downstream changes of dimensionless shear stress for six stream gullies. Its increase with distance is linear, with a relatively high coefficient of determination (R = 0.70), for both whole and individual gully data sets. In stream gullies, as well as in the upstream part of discontinuous gullies, flow energy is partially consumed by gully upslope migration, channel scouring, boundary material entrainment and transport. The relative uniformity of bed gradient (commonly higher than slope gradient) and the downstream addition of flow from tributary gullies and runoff contribute to maintain an excess of flow transport capacity and, therefore, the conditions for erosion and transport of the material supplied from upstream. The width/depth ratios of stream gullies are a little lower than those of discontinuous gullies, indicating some more hydraulic efficiency of the former. Stream gullies may largely contribute to the sediment yield of a river system, while in discontinuous gullies scouring is partially counterbalanced by deposition, with both these processes mainly restricted to an individual slope.

6. Gully formation In the study areas, gullies seem to be initiated by soil piping and the formation of soil hollows such as the headcuts reported by Leopold et al. (1964) and by Bull (1997). The

Fig. 5. Converging flow to headcuts during a high-intensity rainstorm near Butajira.

P. Billi, F. Dramis / Catena 50 (2003) 353­368


analysis of soil characteristics was beyond the aim of this study and no soil investigation was carried out. Nevertheless, soil piping was found very common in areas where multiple slope gullies developed. Headcutting is a poorly known processes though rather common in the study areas. It consists of an upslope migrating terracette whose cuspate morphology is given by contiguous arcuate niches that are very similar in shape to the headcuts of discontinuous ephemeral streams reported by Bull (1997) (Fig. 5). The front height is normally controlled by harder soil horizons (e.g. calcrete) typically occurring a few tens of centimetres below the ground surface. No satisfactory explanation of the main factors originating the initial soil step was found and further investigation is needed. However, once an incipient soil step is formed, sheetflow converges to it (Bull 1997, his Fig. 12c)

Fig. 6. South Langano discontinuous gully. This gully formed during a short (2 ­ 3 h) high-intensity rainstorm. Notice the gully sinuosity, the aggradation in the downstream reach and the high rate of bank failure in the middle reach.


P. Billi, F. Dramis / Catena 50 (2003) 353­368

(Fig. 5) and the headcut is undermined by the tumbling down flow and migrates upslope. The sediment supplied by failure of front portions is redistributed downslope on very lowangle, washout splays. In the study areas, headcut development and migration give rise to three different ground morphologies: (1) individual fronts may advance upslope and merge together forming a badland or (2) leave back a flat, gently inclined surface, that is punctuated by mounds or pinnacles of uneroded soil, and its general morphology recalls, though at a much smaller scale, that of a peneplain; (3) during heavy rainfall, the larger concentration of flow in a cuspate headcut can break the resistant bottom soil layers resulting in a fast, deep ground scouring which leads to the formation of a new gully (Leopold et al., 1964; Bull, 1997). Oostwoud Wijdenes and Gerits (1994) rejected the hypothesis of Ireland et al. (1934) that gullies may develop from rills. In fact, they say rills are not always the dimensional equivalent of gullies, since in the latter bankfull discharges are rarely achieved. During our field surveys in the Keremt rainy season the formation of rills was observed, but none of them was found to develop into a gully. By contrast discontinuous gullies were seen to fully develop during a single, short (2­ 3 h) rainstorm (Fig. 6), in agreement with the assumption of Kirkby and Bull (2000) for the formulation of their gully simulation model. Field evidence and discussions with local people revealed that small stream and discontinuous gullies may experience bankfull discharge during their formation (like the one depicted in Fig. 6) and at least once in a year.

7. Discussion In spite of the rich literature on the subject, the main factors responsible for the formation of gullies are not yet well known (Boardman, 1998). In the study areas, discontinuous gullies are originated by any process and/or small scale, ground surface feature leading to the concentration of overland flow. Other factors such as slope gradient and soil characteristics seem to play a secondary role since discontinuous gullies were observed to occur on a wide range of slopes and soils. Once the gully is formed, its morphological development is mainly governed by the flow capacity to entrain and transport boundary particles (Kirkby and Bull, 2000) and by the soil characteristics and stratigraphy that exert some control on the rate of gully widening and deepening. In the study discontinuous gullies, field observations and the downstream pattern of shear stress (Fig. 3) indicate that in the upstream portion scouring processes prevail and the sediment flux is less than a capacity load as postulated by Howard (1994) for detachmentlimited steep channels. The flow reaches transport capacity around the gully midpoint, while, in the downstream portion, the system becomes transport-limited, as assumed by Willgoose et al. (1991) in their simulations of erosional processes and drainage network development on slopes. In fact, in the downstream reach, the supply of sediment, derived from the eroding processes upstream, exceeds transport capacity and deposition takes place (Fig. 6). Previous theoretical and field studies (e.g. Meyer-Peter and Muller, 1948; Gomez and Church, 1989; Talling and Sowter, 1998) have shown that shear stress and sediment transport are related in rivers; since this relationships seems to occur also in discontinuous gullies, they can be regarded as micro fluvial systems in which basic

P. Billi, F. Dramis / Catena 50 (2003) 353­368


hydraulic processes, similar to those of larger systems, act, though at different rate and scale. According to Knighton (1999), total stream power, defined as the rate of flow energy expenditure per unit channel length, X = cQs (where c is the specific weight of water, Q the discharge, s the energy slope, assumed parallel to bed gradient), is predicted to be maximum at an intermediate location, whose position is determined by the ratio b/b, where b and b are the downstream rates of change of discharge and slope, respectively. The same author also pointed out that b is a conservative quantity having a relatively narrow range of values, hence the position of maximum stream power depends mainly on the shape of the longitudinal profile and its degree of concavity. For different river systems, as b decreases, the position of maximum stream power moves downstream (Knighton, 1999, his Fig. 2). Discontinuous gullies are relatively short and the rate of discharge increase in a downstream direction is expected to be very little. The bed profile of the study discontinuous gullies is commonly flat or has a low degree of concavity, therefore, also b is expected to assume low values. In this situation, the maximum of total stream power should be found near the mouth of discontinuous gullies. Knighton's (1999) model also predicts that specific stream power, x = X/w = sv (where w is channel width, s average shear stress and v average flow velocity), is maximum halfway between the source of the river and the location of the total stream power maximum. Following the previous considerations, if we assume that total stream power maximum occurs at the mouth of discontinuous gullies, specific stream power is expected to peak around the gully mid length. Obviously, further studies are necessary to verify this hypothesis; however, it is worth noticing here that specific stream power and shear stress are strictly related (x = sv) and that the longitudinal variation of shear stress in discontinuous gullies follows a parabolic function whose apex abscissa (maximum shear stress) coincides with the gully mid length (L = 0.46) (Fig. 3). This implies that shear stress increases and decreases at the same rate in both the upstream and downstream portion of the study discontinuous gullies, irrespective of their size and slope gradient. The pivoting point where stream power/ resisting power ratio converges to one is not located at the intersection with the slope profile, as postulated by Bull (1997), but just around the gully midpoint. As the gully grows in size, this point may migrate upstream, following the gully headhcut in the process of upslope advancing, or keeps approximately the same position if the gully is simultaneously expanding in a downstream direction. By additional field monitoring it will be possible to ascertain the development trends of discontinuous gullies (Billi, in preparation). Stream gullies are characterised by a constant, downstream increase of shear stress. In fact, these systems are dominated by erosion processes, are detachment-limited since sediment supply is commonly less than transport capacity and deposition does not take place. Their hydraulic geometry is similar to that observed in the upstream portion of discontinuous gullies and characterised by small width/depth ratios. In stream gullies, the bed gradient is almost constant, while discharge increases noticeably in a downstream direction. Specific stream power, and by analogy shear stress, maxima are therefore expected further downstream of the study reaches as indicated by the constant increase of shear stress depicted in Fig. 4. In the study areas, stream gullies can be seen as the ultimate configuration of discontinuous gullies development. As the latter expand, they can merge into each other


P. Billi, F. Dramis / Catena 50 (2003) 353­368

forming a larger gully system. The detachment-limited to transport-limited transition progressively moves far downstream from the gully head and deposition occurs in the lower gradient river reaches (depending on the grain size of the sediment supplied, Kirkby and Bull, 2000) and on flood plains. Theoretically, any factor capable to tackle the migration of the gully midpoint, where shears stress reaches its maximum, would result in a reduced rate of gully enlargement. This aim could be achieved by introducing erosion control measures designed to increase the bed roughness in the upper portion of the gully.

8. Conclusions The field survey of 16 gullies, located in areas with different climatic, physiographic and geomorphological characteristics of Ethiopia, led to group them in two distinct categories: discontinuous and stream gullies. The former are individual slope incisions originating at a variable distance from the slope divide and showing downstream variations in top-width/depth ratio and hydraulic radius. In the upstream portion of discontinuous gullies erosion and sediment transport are the dominant processes, while in their downstream part, deposition prevails. In stream gullies, both depth and width tend to increase downstream, flow energy is in excess and able to entrain and transport all the sediment derived by bank failure, bottom scouring and supply by overland flow processes. Plots of shear stress versus distance show a different pattern for discontinuous and stream gullies. The predominance of erosion processes in the upstream part, followed by deposition in the second half of discontinuous gullies, is well expressed by the downstream variation of shear stress. Its pattern follows a parabolic function reaching a maximum value around the gully midpoint, while, in the downstream portion, it decreases again to a relative minimum value near the gully mouth. In stream gullies, instead, shear stress increases linearly in association with downstream increase in discharge and excess transport capacity. Gullies severely affect large areas of Ethiopia and reliable solutions to this problem are still far from being found. The geomorphic analysis presented in this study indicates that an increase in flow resistance in the upstream portion of active discontinuous gullies could improve their stability, preventing them from developing into the larger stream gully systems. Further studies are necessary in order to define improved models of gully expansion and to design valid countermeasures, but the hydraulic processes involved in gully formation and growth should be taken into more consideration than in the past.

Acknowledgements This research was supported by the EU STD3 Programme, contract no. STD3-TS3CT92-0076: ``Land resources inventory, environmental changes analysis and their applications to agriculture in the Abaya lake region'' and by the Italian Ministry of Foreign Affairs, Department for the Co-operation with Developing Countries. The authors

P. Billi, F. Dramis / Catena 50 (2003) 353­368


are grateful to F. Vannacci, S. Accetta and the students of the 1996 Master Course of Geomorphology of the University of Addis Ababa for their help in the field data collection. A.J. Parsons and three unknown referees are acknowledged for constructive comments and suggestion on the paper contents and language.


Billi, P., Dramis, F., 1999. Geomorphic Analysis of Small Gullies Evolution and Climate Change in Ethiopia, 2nd International Convention on Environment and Development, Havana 14 ­ 18 June 1999. Abstract. Billi, P., Dramis, F., 2000. Recent climatic trends and soil erosion in some selected areas of Ethiopia. In: Feoli, E., Pottier, D., Zerihun Woldu (Eds.), Sustainable Developments in Drylands of East Africa. European Commission, DG XII, Science Research & Development, pp. 145 ­ 162. Boardman, J., 1998. Modelling soil erosion in real landscapes: a western European perspective. In: Boardman, J., Favis-Mortlock, D. (Eds.), Modelling Soil Erosion by Water, Proceedings of the NATO Advanced Research Workshop `Global Change: Modelling Soil Erosion by Water'NATO ASI Series, Series I: Global Environmental Change, vol. 55, p. 17. Bull, W.B., 1997. Discontinuous ephemeral streams. Geomorphology 19, 227 ­ 276. De Oliveira, M.A.T., 1989. Erosion disconformities and gully morphology: a three dimensional approach. Catena 16, 413 ­ 423. De Ploey, J., 1990. Threshold conditions for thalweg gullying with special reference to loess areas. In: Bryan, R.B. (Ed.), Soil Erosion--Experiments and Models. Catena Supplement, vol. 17, pp. 147 ­ 151. Ethiopian Mapping Authority, 1988. National Atlas of Ethiopia. Ethiopian Mapping Authority, Addis Ababa. 77 pp. Gomez, B., Church, M., 1989. An assessment of bed load transport formulae for gravel bed rivers. Water Resour. Res. 25, 1161 ­ 1181. Grissinger, E.H., 1996. Rill and gully erosion. In: Agassi, M. (Ed.), Soil Erosion, Conservation and Rehabilitation. Marcel Dekker, New York, pp. 153 ­ 167. Heede, B.H., 1976. Gully development and control: the status of our knowledge. USDA Forest Service Research Paper RM-169, Fort Collins, 42 pp. Howard, A.D., 1994. A detachment-limited model of drainage basin evolution. Water Resour. Res. 30, 2261 ­ 2285. Ireland, H.A., Leach, H.R., Vanvliet, R., 1934. Principles of gully erosion in the piedmont of South Carolina. USDA Tech. Bull. 63, 1 ­ 143. Kirkby, M.J., Bull, L.J., 2000. Some factors controlling gully growth in fine-grained sediments: a model applied in southeast Spain. Catena 40, 127 ­ 146. Knighton, A.D., 1999. Downstream variation in stream power. Geomorphology 29, 293 ­ 306. Leopold, L.B., Wolman, M.G., Miller, J.P., 1964. Fluvial Processes in Geomorphology Freeman, San Francisco. 522 pp. Meyer-Peter, E., Muller, R., 1948. Formulas for bedload transport. Proceedings of the Int. Ass. For Hydr. Res., erd Annual Conference, Stockholm, 39 ­ 64. Oostwoud Wijdenes, D.J., Gerits, J., 1994. Runoff and sediment transport on intensively gullied, low-angle slopes in Baringo District. In: Bryan, R.B. (Ed.), Soil Erosion, Land Degradation, Social Transition. Advances in Geoecology, vol. 27. Catena Verlag, Reiskirchen, pp. 121 ­ 141. Parsons, A.J., Abrahams, A.D., Shiu-Hung Luk, 1990. Hydraulics of interrill overland flow on a semi-arid hillslope, southern Arizona. J. Hydrol. 117, 253 ­ 273. Poesen, J., Govers, G., 1990. Gully erosion in the loam belt of Belgium: typology and control measures. In: Boardman, J., Foster, I.D.I., Dearing, J.A. (Eds.), Soil Erosion on Agricultural Land. Wiley, New York, pp. 513 ­ 530. Sagri, M. (Ed.), 1998. Land Resources Inventory, Environmental Changes Analysis and Their Application to Agriculture in the Lakes Region (Ethiopia). European Commission Contract TS3-CT92-0076 Final Report, 183 pp.


P. Billi, F. Dramis / Catena 50 (2003) 353­368

Talling, P.J., Sowter, M.J., 1998. Erosion, deposition and basin-wide variations in stream power and bed shear stress. Basin Res. 10, 87 ­ 108. Trimble, S.W., 1974. Man-induced soil erosion on the Southern Piedmont. Soil Conserv. Soc. Am., Ankeny, Iowa, pp. 1700 ­ 1970. Willgoose, G., Bras, I., Rodriguez_iturbe, I., 1991. A coupled network growth and hillslope evolution: 2. Nondimensionalisation and applications. Water Resour. Res. 27, 1685 ­ 1696.


PII: S0341-8162(02)00131-5

16 pages

Report File (DMCA)

Our content is added by our users. We aim to remove reported files within 1 working day. Please use this link to notify us:

Report this file as copyright or inappropriate


You might also be interested in

Detailed Soil Survey of Jeff Davis County, Texas
Microsoft Word - Land07e.doc
PII: S0341-8162(02)00131-5