Flume experiments are often conducted to reveal the detailed behavior of debris flows. Direct sampling at the lower end of a flume has been used as a method for measuring sediment concentration in flume experiments, but measurement under non-steady or non-equilibrium condition is difficult. In contrast, image analysis methods can acquire spatial-temporally high-resolution data in a non-contact manner and may complement conventional methods. In this study, we conducted experiments that homogenous sediment concentration field of 0～20% was made in water tank with 10 cm depth using coarse sediment with grain size of 2 mm and fine sediment with grain size of 0.2 mm. We filmed the experiments and verified the relationships between sediment concentration with image features in a fixed area. As a result, the mean and mode of pixel values for coarse sediment depend on sediment concentration up to 10%, the standard deviation and frequency of mode for fine sediment depend on sediment concentration up to 20%. Then, we analyzed the controlling factors of the image features and identified three processes : blocking of the background by sediment, darkening of the background, and changes in the brightness of the sediment. Also, we analyzed the influences of sediment grain size on image features, and it was shown that both the brightness index and the brightness uniformity index of the image were affected. In conclusion, these findings indicate that image analysis is a possible tool for clarifying both sediment concentration and grain size.

Two methods were proposed for adding a driftwood trapping function to a concrete closed dam, one is to install a driftwood trap directly on the upstream face of the concrete closed dam installed in the bedload section (linear layout), and the other is to install a driftwood trap on the sedimentation face of the concrete closed dam installed in the bedload section (convex layout). In order to understand the effectiveness of the two types of facilities installed upstream of the concrete closed dam, we investigated the status of the driftwood trap in the upstream area of the weir and changes in the water depth upstream of the weir through hydraulic model experiments. As a result, it was confirmed that when driftwood was trapped, the amount of driftwood flowing out of the trap surface decreased at once due to the narrower spacing between the components. To ensure the driftwood trapping function, “the length of the facility > the width of the channel” is required, regardless of the ponding or bedload conditions. If the length of the facility (width of the capture surface) exceeds the width of the channel, the progress of the weir's depth rise is inhibited. In order to control the progression of weir raising depth, “apparent channel width > 3 times the channel width” should be adopted, regardless of ponding or bedload conditions.

The advent of high-resolution and high-accuracy DEMs from photogrammetry and LiDAR is providing an invaluable tool to measure erosion/deposition. At this resolution, it is thus now essential to take into account soils' bulk density variations. From a set of experiments in the Hiru-dani experimental watershed, the authors propose a SfM-MVS method to measure soil density. From a set of photographs prior-and post-excavation, we can calculate the volume and measure the mass of the material, deriving bulk-density and other geotechnical parameters. The method has two main strengths : (1) it allows for site-specific error analysis and (2) it works on unconsolidated heterometric material. In the Hiru-dani, the bulk density of the eroded bedrock was twice the one of the forest soil and 0.4 higher than the debris-flow deposit, showing that comparing DEMs is not sufficient to precisely account for soil loss erosion, or define the volume that was transported away from the catchment.