Abstract We present a novel field experimental setup that can be used for studying the characteristics of landslide seismicity. The setup consists of a concrete, filled with soil, cylinder that moves along a surficial soil corridor. The emitted seismic signals are due to soil friction. The cylinder acts as an upscaled sheer-box allowing control over a number of parameters: the magnitude of normal stress on the failure plane, the degree of saturation and the type of soil. This allows for the simulation of soil friction within, or between, different geological layers under different conditions. Results are site specific, but can be easily reproduced for any geological environment. We validate this methodology by comparing the spectral characteristics of the signals emitted by the movement of the cylinder to those induced by a controlled failure of a 2.5 m high vertical face at a nearby site with very similar geology. We find a very good agreement between the two. This methodology can be used as a site investigation tool for the optimization of the deployment geometry of seismic networks for landslide monitoring, as well as to inform machine learning algorithms on automatic detection and classification of recorded signals during seismic monitoring of landslides.

中文翻译：

---

确定软土滑坡的地震特征：一种基于现场尺度剪切箱的方法

摘要 我们提出了一种新的现场实验装置，可用于研究滑坡地震活动的特征。该装置包括一个混凝土，充满土壤，圆柱体沿着地表土壤走廊移动。发射的地震信号是由于土壤摩擦引起的。圆柱体作为一个放大的剪切盒，允许控制许多参数：破坏平面上的法向应力大小、饱和度和土壤类型。这允许在不同条件下模拟不同地质层内或之间的土壤摩擦。结果是特定于地点的，但可以很容易地在任何地质环境中重现。我们通过将圆柱体运动发出的信号的光谱特征与附近地质非常相似的 2.5 m 高垂直面的受控破坏引起的信号的光谱特征进行比较来验证这种方法。我们发现两者之间有很好的一致性。该方法可用作现场调查工具，用于优化滑坡监测地震网络的部署几何形状，以及在滑坡地震监测期间为机器学习算法自动检测和分类记录的信号提供信息。