Real-time unstable slope monitoring is essential for recognition of landslide occurrence, as well as for early warning to reduce landslide-induced damages. This study investigated an unstable slope monitoring system that consists of tilt sensors aiming to establish an advanced time prediction model (TPM) for landslide early warning. The monitoring process utilized additional support devices (e.g. pipe strain gauges, water level gauges and a rain gauge) installed on a natural cut slope site. The tilt sensors could detect movements (in tilt angle) within the slope generated by heavy rains. Analysis of the recorded data revealed that the rate of movement, or tilt, was influenced by groundwater table fluctuations and antecedent rainfalls. A relationship between the tilt rate and the horizontal displacement calculated from pipe strain value has been established. Subsequently, a new classification of slope movement was proposed according to the tilt rates obtained from the deformation process of the slope. Considering the movement characteristics, two warning levels were identified such as warning and evacuation. Further, an advanced TPM was proposed in relation to realtime slope surface tilt rates. The TPM could provide efficient results at the continuous acceleration phase of the landslide occurrence. Therefore, we suggest this technique can be applied for monitoring and early warning of rainfall-induced shallow landslides.

实时不稳定边坡监测对于识别滑坡发生以及预警以减少滑坡造成的损害至关重要。本研究调查了由倾斜传感器组成的不稳定斜坡监测系统，旨在建立用于滑坡预警的高级时间预测模型 (TPM)。监测过程使用了安装在自然切割斜坡现场的额外支持设备（例如管道应变计、水位计和雨量计）。倾斜传感器可以检测大雨产生的斜坡内的运动（倾斜角度）。对记录数据的分析表明，移动或倾斜的速度受到地下水位波动和前期降雨的影响。已经建立了倾斜率和根据管道应变值计算的水平位移之间的关系。随后，根据从边坡变形过程中获得的倾斜率，提出了一种新的边坡运动分类。考虑到运动特点，确定了预警和疏散两个预警级别。此外，还提出了一种与实时坡面倾斜率相关的高级 TPM。TPM 可以在滑坡发生的连续加速阶段提供有效的结果。因此，我们建议将该技术应用于降雨诱发的浅层滑坡的监测和预警。考虑到运动特点，确定了预警和疏散两个预警级别。此外，还提出了一种与实时坡面倾斜率相关的高级 TPM。TPM 可以在滑坡发生的连续加速阶段提供有效的结果。因此，我们建议将该技术应用于降雨诱发的浅层滑坡的监测和预警。考虑到运动特点，确定了预警和疏散两个预警级别。此外，还提出了一种与实时坡面倾斜率相关的高级 TPM。TPM 可以在滑坡发生的连续加速阶段提供有效的结果。因此，我们建议将该技术应用于降雨诱发的浅层滑坡的监测和预警。