3.2 Temporal changes in catchment geomorphic characteristics
All values of the catchment geomorphic parameters (HL , A ,Pe , and HDC ) showed increasing tendencies as the gully networks expanded over time (Mukai, 2017). Over the three periods, 1957, 1972, and 2005, no areal aspect of catchment geomorphic parameters,HL (maximum catchment length), A (catchment area), andPe (catchment perimeter) showed any significant differences (p > 0.05) between the two sub-areas (Mukai, 2017). The same was true for the catchment areal aspect indices, LR(lemniscate ratio), CC (compactness coefficient), and FF(form factor; Table 1). Besides, no notable tendencies in LR ,CC , FF , and RR were observed over the three periods. The relationships between HL and A (Figure 4 (a)) and Pe and A (Figure 4 (b)) showed the same spatial and temporal trends; (i) regardless of the period and sub-area, all the data appeared to be positioned on the linear lines that represent the relationships between each pair of the two parameters; and (ii) the Vallley bottom main gullies largely extended the HL -A andPe -A relationships over time, whereas the temporal changes were limited for the Rift margin main gullies.
In contrast, the catchment relief aspect parameter HDC (height difference for catchments) showed a significant difference (p< 0.01) between the two sub-areas (Mukai, 2017). The same was true for the catchment indices, RR (relief ratio) and HI(hypsometric integral; Table 1). Over the three periods, the Rift margin showed higher RR values than the Valley bottom, showing the catchments of the Rift margin had higher sediment loss per area. TheHI values of the Rift margin shifted from 0.37 in 1957 to 0.33 in 2005, indicating the catchments were always at the equilibrium or mature stage, whereas those of the Valley bottom shifted from 0.60 in 1957 and 1972 to 0.47 in 2005, indicating the catchments recently have developed from disequilibrium/young stage to equilibrium/mature stage. The relationship between HL and HDC (Figure 4 (c)) showed a contrasting trend between the two sub-areas; (i) regardless of the period, all the data appeared to be positioned on the linear lines representing the HL-HDC relationships that are specific to each sub-area; and (ii) the Valley bottom main gullies largely extended theHL-HDC relationship over time, whereas the temporal changes were limited for the Rift margin main gullies.