Landslides distribution at tributaries with different evolution stages in Jiangjia Gully,

Abstract. Jiangjia Gully (JJG) is known for its high frequency and variety of debris flows, especially the intermittent surges of various flow regimes and materials. Observation indicates that the surges come from various tributaries with different landslides activities. In this study, 81 tributaries of JJG are taken from DEM with 10 m grid cells, and the hypsometric curves are used to characterize their evolution stages; five stages are identified by the evolution index (EI, the integral of the hypsometric curves) and most tributaries are in relative youth stage with EI between 0.5 and 0.6. Then 908 landslides are interpreted from Quickbird satellite image of 0.61 m resolution, and it is found that LD (LD = landslides number in a tributary/the tributary area) increases exponentially with EI, while LAp (LAp = landslides area in a tributary/the tributary area) fluctuates with EI, meaning that landslides are inclined to occur in tributaries with EI between 0.5 and 0.6, and thus these tributaries are the main material sources supplying for debris flows.



30
Landslide susceptibility assessment over large areas is considered a preliminary step for the 31 planning or design of the most appropriate risk mitigation measures. The use of statistics and physics 32 based models is considered a useful tool for landslide susceptibility assessment (Amashi et Strahler, 1952Strahler, , 1957. Meanwhile, 44 the hypsometric curves are related to tributary form and erosional process, and are used to interpret 45 landform development stages (Schumm, 1956;Strahler, 1952Strahler, , 1957, which can represent the state of 46 material storage of a tributary. In addition, the relationship between EI and tributary characteristics 47 changes with scales. For example, the dissection index of tributaries presents various relationships to 48 EI depending on scale of the tributaries. For the 5th-order tributaries, their correlation is r = 0.41, 49 whereas for the 4th-order, it is r = 0.24, and it becomes negative correlation for the 3rd-order (Hamza et Wieczorek, 1996). Therefore, the relationship between EI and landslides distribution has special 57 significance to reveal the landslides distribution in tributaries, which, however, has been gotten little 58 attention in literatures.

59
In this paper, a case study is conducted in Jiangjia Gully (JJG), where weak and similar lithology, disparate topography, sparse vegetation, and unconsolidated deposits are widely distributed in 61 tributaries. In addition, the debris flow behavior in JJG are representative, it is known for the high 62 variety of debris flows; each debris-flow event consists of tens or hundreds surges of different flow 63 regimes, velocities, discharges, and total volumes Li et al., 2013;Arai, 2017). In 64 particular, the surges are composed of different materials, suggesting that they come from different 65 sources (Xiang et al., 2015). In other words, each debris flow in JJG comes from different tributaries 66 (Bollschweiler et al., 2007;Li et al., 2012;Li et al., 2013;Li et al., 2015). Generally, the flow surges 67 are originated from different tributaries and the material supplies are mainly from landslides (including 68 avalanches, soil failures and other slope processes) (Beguería, 2006 (Fig. 1). This region undergoes active 76 neotectonic movement, faults, and folds; and rocks are dominated by slate, dolomite, limestone, basalt 77 and breccia rocks, which are easily weathered (Gabet and Mudd, 2006). The exposed strata in this gully 78 is mainly shallow metamorphic rocks of the lower proterozoic Kunyang group, accounting for about 79 80% of the whole gully area (Wu et al., 1990). Generally, weak lithology, wide faults and sparse  and also with artificial correction to ensure the accuracy of boundaries (now it is Fig. 3). In principle, the gully can be divided further into smaller tributaries, but that makes little difference for the present 100 purpose as to distinguish tributaries. Some tributaries in field are displayed in Fig. 2

114
Hypsometric integral is the area between the hypsometric curve (y=h/H and x=a/A) and 115 coordinate axis (Strahler, 1952(Strahler, , 1957, which can be defined as the evolution index (EI).

147
The curves present various types, such as convex, concave and others between them; and these 148 can be well fitted by the following function (Strahler, 1957): 150 Where k and n are parameters, with the fitting coefficient R 2 of 0.90 and higher. It is found that higher 151 the curve is, greater the k is. Meanwhile, the curve is rising as n decreases.

165
Moreover, it is found that the EI satisfies the Weibull distribution with the scale parameter of 0.02 166 and the shape parameter of 1.69 (Fig. 5). The small value of scale parameter means that EI is much 167 concentrated and EI of most tributaries in JJG is mainly between 0.5 and 0.6. The shape parameter is 168 more than 1 and the frequency of the tributaries changes rapidly with the increasing of the EI,

332
The EI values for the landslide sources are also subject to the Weibull distribution (Fig. 16) which means the concentration of EI. This also implies that landslides occur in tributaries within a 341 relatively narrow range of EI. More important point is the difference between shape parameters, the 342 bigger shape parameter in Wenchuan region means that the curve is to the right more than in JJG, 343 implying that the earthquake is inclined to induce more landslides in tributaries of big EI. As JJG is of 344 tributaries with wide range evolution stages, we choose it as the study area to reveal the mechanism of 345 landslides distribution. 346 347 Fig. 16 The EI frequency distribution of Wenchuan in Sichuan province.