Abstract
To analyse the driving forces of gully erosion using a present dataset of geomorphic parameters and land use/cover involves limitations because past datasets at the time of gully incision may best explain the gully formation and evolution at that time. The recent development of photogrammetric techniques enabled to estimate temporal gully volume changes. This study conducted in semi-arid Ethiopian Rift Valley used field measurements and gully volume–length relation to analyse spatial and temporal dynamics of catchment geomorphology and topographical threshold of gully heads to explain the difference in the gully volumes and area-specific gully volumes between two study sub-areas. The topographic thresholds of the gully heads, expressed by the slope (=s ) and drainage area (= a ), (i) formed in each catchment and (ii) that had the same land use/cover items (forest, grassland, and farmland) in all the catchments of each sub-area were approximated by power functions (s = ka-b ). Analysis of covariance found that these threshold lines had clear spatial and temporal patterns: the threshold lines maintained almost the same exponent b specific to each sub-area while the threshold coefficient k significantly decreased in the order of forest, grassland, and farmland. The spatial variability and its temporal changes in relief aspect of the catchment morphology can responsible for the difference in the area-specific volumes of gullies between the sub-areas, while the continuous reduction in vegetation cover over time can be the main driving force of the similar scale and changing patterns of the gully volumes between the sub-areas.
KEYWORDS: gully evolution, area-specific gully volume, gully volume–length relation, catchment geomorphology, topographic threshold
INTRODUCTION
In East and South African countries, large-scale gullies can be seen almost everywhere (Katsurada et al., 2007; Ndomba, Mtalo, & Killingtveit, 2009; Boardman, 2014). In semi-arid Ethiopian highlands (Tigray), the area-specific gully erosion rates (gully erosion rate per unit area) since gully incision to 2001 were 6.2–17.6 Mg ha-1 y-1 (Nyssen et al., 2006; Frankl et al., 2013a). In sub-humid Ethiopian highlands (Amhara), the area-specific gully erosion rates were 8.7–155 Mg ha-1 y-1 (Tebebu et al., 2010; Zegeye et al., 2016; Yibeltal et al., 2019). In semi-arid Ethiopian Rift Valley, a lowland part of Ethiopia, the area-specific gully erosion rate was 16.2 Mg ha-1 y-1 (Mukai, 2017). Most gully volumes showed an exponential increase since gully incision in Ethiopia except the ones in the areas where gully rehabilitation or soil and water conservation programmes at watershed scale were implemented (Nyssen et al., 2006; Frankl et al., 2013a). The contribution of gullying to total soil loss from the area ranges from 28% in semi-arid highlands (Nyssen et al., 2008) to 64 to more than 90% in sub-humid highlands (Tebebu et al., 2010; Zegeye et al., 2016) of Ethiopia.
Gully formation and its evolution are regulated by various factors, such as several geomorphic properties of catchments, slope gradient, land use, vegetation, and rainfall characteristics (Poesen et al., 2003; Valentin, Poesen, & Li, 2005). Land use and vegetation cover are a major controlling factor of gully initiation (Parkner et al., 2006; Gómez Gutiérrez, Schnabel, & Lavado, 2009) and its evolution (Martinez-Casasnovas, Ramos, & Garcia-Hernandez, 2009). Many studies that used a statistical approach concluded that the present land use/cover was not a decisive factor of gully erosion (Muňoz-Robles et al., 2010; Kompani-Zare et al., 2011; Frankl et al., 2013b; Mukai, 2017). Using sequential historical aerial photographs and photogrammetric technique, Gómez Gutiérrez, Schnabel, & Lavado (2009) determined temporal changes in gullied areas over ~60 years in southwest Spain, which were explained by temporal changes in land use and vegetation cover in respective years determined by the combination of aerophoto-interpretation and field observation. Martinez-Casasnovas, Ramos, & Garcia-Hernandez (2009) used similar techniques to explain the impact of land use/cover changes on gully sidewall erosion in northeast Spain. Mukai (2017) found that the gully erosion rates of the study gully networks during the respective study periods in the present and the past had significantly strong correlations with the rates of area changes in land use/cover items during the respective study periods. These pieces of evidence show the importance of an analysis that analysed the relationship between gully erosion and past land use/cover when gullies were initiated.
It is well known that gully initiation and gully head positions are related to some critical conditions, e.g., the topographic threshold, a combination of the slope at the gully head (s ) and upslope drainage area (a ), which is expressed in equation (1):
s = ka-b (1),
where the threshold coefficient k , is a constant that varies with local climate, soils, and vegetation covers, and the exponent b , is a constant which is related with the dominant process of the flow conditions in a gully channel, e.g., surface or sub-surface flows (Torri & Poesen, 2014). Land use that reduces vegetation cover, such as an increase in cultivated area and transformation of forest to grassland, tends to reduce the topographic threshold levels and increase the risk of gully erosion on sites (Parkner et al., 2006; Gómez Gutiérrez, Schnabel, & Lavado, 2009). Using sequential historical aerial photographs and digital elevation models (DEM), Parkner et al. (2006) determined the s -a relationships (topographic threshold) of the gully heads formed in the 24 active gullies and gully complexes in New Zealand back to the time when gully heads were incised. Land use and vegetation cover in individual years was determined by mainly aerophoto-interpretation. They found the s -a relationships were related to land use/cover at the time of gully incision. When the land use of a catchment was an indigenous forest, the topographical threshold value was very high. As the land use changed to pasture, to invaded scrub, and to reforestation, the value decreased rapidly, then slowly increased, and finally returned to a similar level to that under indigenous forest. Gómez Gutiérrez, Schnabel, & Lavadzo (2009) took a similar approach and proposed a topographical threshold evolution model based on land-use changes in respective study years in southwest Spain. Thus, some studies have indicated that land use and vegetation cover in the catchments at the time of gully incision explained the spatial variability in gully topographical threshold values well. However, the studies that quantitatively assessed the relationships between land use/cover and a difference in topographical threshold values were rare.
Some gully morphological characteristics have recently been used to determine temporal gully volume changes. Several studies have explored the relationship between the gully volume (V ) and length (L ) using a power equation V = aL b (VL relation; e.g., Frankl et al., 2013b). Li et al. (2017) proposed a relation between the gully volume (V ) and gully area (Ag ) using a power equationV = aAg b. These models have advantages that the length and area of a gully can be easily determined from aerial photographs and high-resolution satellite images. These photogrammetric techniques were utilised to assess long-term changes in gully volumes (Frankl et al., 2013a; Mukai, 2017).
Concerning other controlling factors of gully erosion that show temporal variation, such as catchment geomorphic characteristics, many studies analysed the present parameters and indices (Singh, Sarangi, & Sharma, 2008; Chandrashekar et al., 2015). Using a statistical approach, Tamene et al. (2006) and Haregeweyn et al. (2008) quantitatively explained the spatial variability of sediment yield and area-specific sediment yield by various controlling factors that included the present catchment geomorphic parameters in semi-arid Ethiopian highlands. However, the relationships between the temporal changes in the catchment geomorphic parameters and gully evolution (e.g., gully volume changes) have rarely been quantitatively analysed except the VL relation. Mukai (2017) quantitatively analysed the relationship between the temporal changes in the catchment geomorphic parameters and gully erosion; however, how that relationship had implications to the spatial variability of gully evolution in agroecology had not been analysed.
The objectives of this study carried out in semi-arid Ethiopian Rift Valley were three-fold. It was, first, to analyse spatial and temporal dynamics in catchment geomorphology quantitatively and explain the difference in gully evolution (i.e., changes in the gully volumes and area-specific gully volumes) specific to each of the two study sub-areas. Secondly, it is to quantitatively assess the relationships between temporal changes in land use/cover items and topographical threshold of gully heads in each sub-area, and to explain the difference in gully evolution specific to the sub-area; and (iii) to confirm that the combination of the VL relation and field measurements is feasible to assess the interactions between environmental changes and gully erosion/gully head positions.
MATERIALS AND METHODS