GRAIN SIZE DISTRIBUTION ANALYSIS OF LANDSLIDE DEPOSITS WITH RELIABLE IMAGE IDENTIFICATION

High-locality landslide is one of the most catastrophic hazards because of its insidious feature. Analyzing the grain size distribution form the basis of landslide failure process inversion, which is significate for prediction of rock avalanche hazards. In this paper, we propose a novel framework to handle this issue. It contains two main parts:field investigation and PCAS software. First, the high precision images of landslide deposit is obtained using UAV. Then, damage situation of local houses is analyzed in detail. Last, we quantify the grain size of the rock avalanche deposit using image data with PCAS software, and analyze the relationship between the grain size distribution and the damage situation of local houses. A case study is conducted on Pusacun rock avalanches of August 28, 2017. It turns out the following results. (1) The PCAS software can results in reliable image identifications. (2) The proportion of small particle size increases in pace with the runout distance. (3) The proportion of large particle shows bimodal distribution, but the peak value decreases along the sliding deration. (4) The similarity of lateral distribution enhances as the runout distance increases. (5) The houses have the ability for "trapping the coarse particles and discharging the fine particles". The results demonstrate the framework proposed has superior performance in related researches.