The evolution of slope is the result of slope process and one of the issues of geomorphic system for drainage basin. In the present study, a complex network method was adopted to build up the spatial structure of slope system in order to explore the evolution rules of slope system. Data were high-resolution digital elevation model (DEM) of 9 stages, which were gathered by close range photogrammetry during different stages of the runoff experiment. Slope form was classified into 7 slope types, i.e. flat slope, steep slope, double-straight slope, double-concave slope, double-convex slope, convex-straight slope and concave-straight slope. Slope unit was further segmented using object-based image analysis method with the terrain factors including slope, plan curvature and profile curvature. Scale parameter of slope segmentation was set to 10 units depending on change rate of local variance on segmentation model. The classification map of slope forms was completed based on the classification rules of slope unit. Slope networks referred to objects composed of slope units and connection between these units. Each node represented slope form which was the terrain condition of sediment transfer. Edges illustrated the spatial link relations between slope forms. Nine slope networks were established in 9 stages, and average degree, degree-relation and structure entropy were taken as network indices to measure network characteristics. Results showed that the degree distribution of slope network fitted power law, so it was a kind of scale-free network. Steep slope was one key node which controlled the structure of network. New nodes preferred to connect with the node with high average degree under gravity erosion, therefore, double-straight slope was apt to transform into steep slope. Moreover, slope network is a disassortative network as minus assortative factor. Slope form with low degree was easily distributed around node with high degree. In the evolution of slope system, new slope forms constantly emerged and network structure entropy tended to grow large and then stabilize under the effect of rainfall erosion. Specifically average degree and structure entropy of network were rapidly enlarged with strong rainfall in the 2nd stage. According to the change rate of network structure entropy, the evolution of network could be divided into 3 periods, which were sprouting development and mature period. Average degree of network showed periodic change because the function of each slope form had changed in different stages. Double-straight slope as initial slope had high average degree, then its average degree continuously reduced with erosion, and the number of its node constantly increased. Average degree of steep slope reached the peak in the network of the 5th stage after it consistently increased since the 2nd stage, and then it slightly dropped. During the evolution of the 2nd-3rd and 4th-5th network, sediment discharge clearly expanded under similar rainfall condition. Steep slope was one important phase for sediment transfer, and the sediment discharge of slope system was influenced by its average degree. However, the average degree of double-convex slope, convex-straight slope and concave-straight slope had a little change during the network evolution. Due to the effect of headward erosion, the average degree of double-concave slope rose gradually after the network of the 5th stage. Some steep slopes were likely converted into double-concave slope under deposition, which decreased sediment discharge of slope system. Rainfall intensity was an important factor of network structure entropy. When average rainfall intensity was beyond 1.5 mm/min, the change rates of structure entropy of the network of the 2nd, 3rd, 5th and 8th stage were high and the network structures had changed dramatically. The reason was that splash and gravity erosion were leading driving forces for slope evolution. On the contrary, its structure entropy slightly dropped in the case of the low rainfall intensity. At the point, deposition process played an active role in slope evolution. In short, slope washing, incision, combination and deposition were slope processes for shaping the structure of slope system. It is demonstrated that spatial complex network approach is a novel framework for exploring the evolution of geomorphic system. ©, 2015, Chinese Society of Agricultural Engineering. All right reserved.
