Impact of terrain attributes, parent material and soil types on gully erosion
Highlights
► The rate of gully bank retreat (GBR) is estimated and related to the environment. ► Results suggest a GBR of 4.9 ± 0.13 cm y– 1. ► Lateral seepage and piping are predominant processes. ► GBR decreased from gully heads to the downstream direction. ► Soil type and parent material had not significant impact.
Introduction
Soil erosion is a natural phenomenon. However, when the removal of the soil is faster than soil formation through bedrock weathering, it becomes problematic, often resulting in the reduced ability of the terrestrial ecosystems to perform their functions, such as those of food and biomass production, storage and filtering of water (FAO, 2008).
There are three main types of water erosion (Wischmeier and Smith, 1978). The first type is splash erosion, which is the detachment and airborne movement of small soil particles caused by the impact of raindrops on the soil surface. The second type, sheet erosion, corresponds to the detachment of soil particles by raindrop impact and their removal down-slope by sheet flow (Chaplot and Le Bissonnais, 2003, Chaplot et al., 2007), while the third type is linear erosion, where the soil material is detached and transported by overland flow (Kinnell, 2004). Rills correspond to the shallowest forms of linear erosion, while gullies are sufficiently deep not to be filled by tillage operations, thus resulting in irreversible losses of agricultural land. In light of ecosystem management, it is important to understand the spatial distribution of gullies and gullying (the process of gully formation and evolution) in agricultural landscapes (Poesen et al., 2003).
Overland flow is thought to be the main cause of gully erosion. When the flow velocity overcomes a threshold resistance of the soil material, overland flow induces scouring, which initiates macro channel formation. Because of the topographic control of flow velocity, several topographical thresholds have been used for spatially predicting gully formation (Nachtergaele et al., 2002, Poesen et al., 2003). In addition, linear erosion by subsurface water movements is often overlooked. Preferential flow paths through pipes are considered as a major process of linear erosion (Fox and Willson, 2010), as well as of landslides (Uchida et al., 2001). Gullies may build-up, when soil pipes erode underlying soil horizons to the extent that tunnel collapse occurs. When the sediment concentration in the pipe flow exceeds its transport capacity, pipe blockage may happen, causing a build-up of soil water content and water pressure, which is responsible for mass movements and landslides. Once the depression forms, the changed soil surface morphology collects and concentrates overland flow, which may increase the export of sediments. Lateral seepage is another process by which the enhanced soil pore water pressure causes undercutting, which, in turn induces wall failure and depression enlargement (Fox and Willson, 2010).
The literature on the prediction of gully head cut location, based on the role of topography, is abundant (Poesen et al., 2003). The topographic thresholds for gullying, which are based on Horton's (1945) concept of overland flow generation and concentration, may vary from site to site. However, Bocco (1991) suggested that rather than be caused by Hortonian flow, gully initiation might often to be due to saturated overland flow and piping. According to various authors, the topographic parameters, mean slope gradient (S) and upslope contributing area per unit length of contour (As in m2 m− 1), are critical to assess the location and the size of gullies. It is indeed assumed that gullying occurs as the flow velocity exceeds the soil shear stress, which is mostly a function of As, as it controls the amount of overland flow and S, which determines its level of energy (Vandaele et al., 1996, Vandekerckhove et al., 1998, Desmet et al., 1999). Based on this concept, there is an inverse relationship between S and As, the upslope contributing area for gully initiation decreasing, as the mean slope gradient increases. Several topographic thresholds for gullying were found in the literature. In the US, Moore et al. (1988) found that S × As should be over 18 or lnAs/tanS up to 6.8 to initiate gully erosion. In the Belgian loess belt, the threshold was found to be between S × As0.4 = 0.025 (Vandaele et al., 1996) and S × As0.4 = 0.72 (Desmet and Govers, 1997). Predictions based on topographic thresholds have successively allowed the spatial prediction of gully erosion headcuts in some areas of the world, such as the West European loess belt (Poesen, 1989, Desmet and Govers, 1997) and the steep slopes of South East Asia (Chaplot et al., 2005), but in many other areas, the concept of topographic control of gullying is flawed.
For this reason, several authors established thresholds of upslope contributing area to account, not only for gullying by saturation overland flow, but also pore pressure induced landsliding and piping (Montgomery and Dietrich, 1988, Montgomery and Dietrich, 1989, Montgomery and Dietrich, 1994, Dietrich et al., 1993, Montgomery, 1994, Montgomery, 1999), two frequently overlooked mechanisms, operating at low flow velocities. Montgomery and Dietrich (1992) and Dietrich et al. (1993) showed that a single S × Asa threshold was insufficient at predicting gully initiation in Tennessee, USA. They further demonstrated that the “a” exponent should be between 1 and 2, with a threshold between 25 and 200 to account for the different gullying processes.
While most of the studies have investigated the factors of control of gully head cut location, only few have been performed on the controlling factors of gully side wall retreat, an important – if not the dominant – land degradation process (Daniels, 2002, Krause et al., 2003). Based on the fact that environmental factors are much easier to obtain than measurements of gully wall retreat, especially when large areas are considered, establishing a link between wall retreat and some key environmental factors, such as relief and soil type, will allow the spatial prediction of lateral gully growth and will improve the implementation of preventive measures that will reverse the growth of existing gullies.
From the available literature, gully side walls can either retreat rapidly in width during single storm events or change little over long periods of time (5000–10,000 years) (Kirkby and Bracken, 2009). The reason for such discrepancies seems to be related to the detachment of soil material from the gully walls and its evacuation possibilities. By wetting the soil, rainfall acts mainly through the slaking and spalling of the soil at gully banks (the wetted soil part breaks off from the drier material underneath and drops), but with the detachment of single particles by splash being a minor process (Chaplot et al., 2011). Moreover, during rainfall events, overland flow can infiltrate in between soil aggregates and natural vertical cracks, fostering the collapse of gully banks, a process that may be accelerated by seepage undermining erosion. As the gully walls sap, the collapsed material produces a talus slope close to the angle of repose, that may eventually protect the gully bank from further retreat. The capacity of the gully to grow laterally thus depends on the ability of runoff in the gully channel to evacuate this material. It is the premise in this paper that bank retreat processes are correlated to the bank environment. The question of whether environment factors, such as terrain morphology (mean slope gradient, terrain curvature and upslope drainage area), parent material and soil types may allow the prediction of the retreat rate of gully walls, remains to be asked. To this end, this study investigated the link between gully evolution and selected environmental factors in an active gully system of South Africa. The study was performed in the foothills of the Drakensberg, where preliminary observations have shown that gullying is an active process (Martin, 1987, Yaalon, 1987, Botha et al., 1994, Wintle et al., 1995, Rienks et al., 2000, Morgan and Mngomezulu, 2003, Chaplot et al., 2011).
In a representative 4.4 km2 degraded rangeland, the main objective was to assess the relationship between gully side wall retreat over an entire year and selected factors of the environment, such as terrain morphology, parent material and soil types. Here the focus is on permanent gullies, not ephemeral gullies and rills.
Section snippets
Study area
The study site is a 2 × 2.2 km2 area (longitude: 29.36°; latitude: 28.82°) of the foothills of the Drakensberg mountains (KwaZulu-Natal Province, South Africa) (Fig. 1). This area under rangeland is included in the Thukela Basin (30,000 km2), a river system highly modified by humans with a high density of dams and reservoirs to fulfill the needs of the agriculture and of large cities.
The altitudes in the area range between 1237 and 1467 m (Fig. 1). Higher altitudes occur in the north-east and in the
DEM and terrain attributes
The 20 m mesh DEM from the national data base (National Geospatial Information, 2000) was used in this study. This DEM was generated, using 5 m contour lines. Data points of elevation obtained with a DGPS by Dlamini et al. (2011) showed that this DEM yields an accurate representation of the landscape shape and water pathways. Several terrain attributes were estimated from it: elevation (Z); As; S; the distance to gully head cut (D), the gully width (W) the slope length factor (SL; Wischmeier and
Soil characteristics
Luvisols is the most commonly found soil type in the study area. Deep Luvisols (~ 2 m) were mainly found at backslope and foot slope position. They exhibit compact (bulk density between 1.4 and 1.6 g cm− 3) 0.5 to 0.9 m thick clayey Bw horizons (Table 1). Acrisols that show a base saturation of lower than 50% constitute the second soil type. Found on hillslope summits and plateaux, they show a dark reddish brown (2.5YR3/4) Bw horizon on dolerite, while this horizon is yellowish brown (10YR5/4) on
Characteristics of GBR
The average GBR value estimated from the 441 pins was 2.3 ± 0.59 ton ha− 1. This is a relatively low erosion rate, compared with those found in the literature. Poesen et al. (2003) reported linear soil erosion rates, as high as 65 ton ha− 1 y− 1 in the badlands of Spain and 32 ton ha− 1 y− 1 in Niger and 15 ton ha− 1 y− 1 in the neighboring country of Lesotho. The erosion rate in our study area, however, is of the same order of 5.0 ha− 1 y− 1 reported by Flügel et al. (2003) in the very close Mkomazi River Catchment
Conclusions
Two main conclusions can be drawn from this study on gully bank retreat in the Drakensberg foothills of South Africa. First, the obtained results suggest that gullying is an active process in the area with a bank retreat rate of 4.9 ± 0.13 cm y− 1. Second, there was a tendency for the rate of gully bank retreat to decrease from gully heads to the downstream direction and for higher GBR values to be found at banks located at slightly convex slopes and draining surface areas up to 400 m2. Results on
Acknowledgments
This study was performed under the umbrella of the Water Resource Commission of South Africa Project Number K5-1904, which focuses on nutrient and organic carbon erosion at catchment level. The authors gratefully acknowledge the School of Bioresources Engineering and Environmental Hydrology, Rabie Saunders Building, University of KwaZulu-Natal for providing support and assistance, Jean-Louis Janeau from the Institut de Recherche pour le Développement (IRD, former ORSTOM) for having provided
References (57)
Effect of slope on the growth and migration of headcuts in rills
Geomorphology
(1999)- et al.
Episodic late quaternary palaeogully erosion in northern KwaZulu-Natal, South Africa
Catena
(1994) - et al.
Spatial and temporal assessment of linear erosion in catchments under sloping lands of Northern Laos
Catena
(2005) - et al.
Rainfall simulation to identify the storm-scale mechanisms of gully bank retreat
Submitted at Agriculture Water Management
(2011) - et al.
Two-dimensional modelling of the within-field variation in rill and gully geometry and location related to topography
Catena
(1997) - et al.
Importance of slope gradient and contributing area for optimal prediction of the initiation and trajectory of ephemeral gullies
Catena
(1999) - et al.
Quantifying interrill erosion in a sloping-land-agricultural catchment of KwaZulu Natal, South Africa
Submitted at Agriculture Water Management
(2011) - et al.
Multi-parameter fingerprinting of sediment deposition in a small gullied catchment in SE Australia
Catena
(2003) - et al.
Spatial prediction of soil properties, using environmental correlation
Geoderma
(1999) - et al.
Threshold conditions for initiation of valley-side gullies in the Middle Veld of Swaziland
Catena
(2003)