Impact of Water Level Fluctuations on Landslide Deformation at Longyangxia Reservoir, Qinghai Province, China

The construction of Longyangxia Reservoir has altered the hydrogeological conditions of its banks. Infiltration and erosion caused by the periodic rise and fall of the water level leads to collapse of the reservoir banks and local deformation of the landslide. Due to heterogeneous topographic characteristics across the region, water level also varies between different location. Previous research on the influence of fluctuations in reservoir water level on landslide deformation has focused on single-point monitoring of specific slopes, and single-point water level monitoring data have often been used instead of water level data for the entire reservoir region. In addition, integrated remote sensing methods have seldom been used for regional analysis. In this study, the freely-available Landsat8 OLI and Sentinel-2 data were used to extract the water level of Longyangxia Reservoir using the NDWI method, and Sentinel-1A data were used to obtain landslide deformation time series using SBAS-InSAR technology. Taking the Chana, Chaxi, and Mangla River Estuary landslides (each having different reservoir water level depths) as typical examples, the influence of changes in reservoir water level on the deformation of three wading landslides was analyzed. Our main conclusions are as follows: First, the change in water level is the primary external factor controlling the deformation velocity and trend of landslides in the Longyangxia Reservoir, with falling water levels having the greatest influence. Second, the displacement of the Longyangxia Reservoir landslides lags water level changes by 0 to 62 days. Finally, this study provides a new method applicable other areas without water level monitoring data.

Threshold Definition for Monitoring Gapa Landslide under Large Variations in Reservoir Level Using GNSS

The triggering threshold is one of the most important parameters for landslide early warning systems (EWSs) at the slope scale. In the present work, a velocity threshold is recommended for an early warning system of the Gapa landslide in Southwest China, which was reactivated by the impoundment of a large reservoir behind Jinping’s first dam. Based on GNSS monitoring data over the last five years, the velocity threshold is defined by a novel method, which is implemented by the forward and reverse double moving average of time series. As the landslide deformation is strongly related to the fluctuations in reservoir water levels, a crucial water level is also defined to reduce false warnings from the velocity threshold alone. In recognition of the importance of geological evolution, the evolution process of the Gapa landslide from topping to sliding is described in this study to help to understand its behavior and predict its potential trends. Moreover, based on the improved Saito’s three-stage deformation model, the warning level is set as “attention level”, because the current deformation stage of the landslide is considered to be between the initial and constant stages. At present, the early warning system mainly consists of six surface displacement monitoring sites and one water level observation site. If the daily recorded velocity in each monitoring site exceeds 4 mm/d and, meanwhile, the water level is below 1820 m above sea level (asl), a warning of likely landslide deformation accelerations will be released by relevant monitoring sites. The thresholds are always discretely exceeded on about 3% of annual monitoring days, and they are most frequently exceeded in June (especially in mid-June). The thresholds provide an efficient and effective way for judging accelerations of this landslide and are verified by the current application. The work presented provides critical insights into the development of early warning systems for reservoir-induced large-scale landslides.

A Novel Method for Extracting Time Series Information of Deformation Area of a Single Landslide Based on Improved U-Net Neural Network

This paper proposed an improved U-Net fully convolutional neural network to automatically extract a single landslide deformation information under time series based on the physical model experiments. This method extracts time series information for three different landslide deformation ranges. Compared to U-Net and mainstream superpixel method, evaluation indicators of DSC, VOE and RVD verify the high recognition accuracy and strong robustness of our method.

Detecting Long-Term Deformation of a Loess Landslide from the Phase and Amplitude of Satellite SAR Images: A Retrospective Analysis for the Closure of a Tunnel Event

Information about the long-term spatiotemporal evolution of landslides can improve the understanding of landslides. However, since landslide deformation characteristics differ it is difficult to monitor the entire movement of a landslide using a single method. The Interferometric Synthetic Aperture Radar (InSAR) and pixel offset tracking (POT) method can complement each other when monitoring deformation at different landslide stages. Therefore, the InSAR and improved POT method were adapted to study the pre- and post-failure surface deformation characteristics of the Gaojiawan landslide to deepen understanding of the long-term spatiotemporal evolution characteristics of landslides. The results show that the deformation displacement gradient of the Gaojiawan landslide exhibited rapid movement that exceeded the measurable limit of InSAR during the first disaster. Moreover, the Gaojiawan landslide has experienced long-term creep, and while studying the post-second landslide’s failure stability, the acceleration trend was identified via time series analysis, which can be used as a precursor signal for landslide disaster warning. Our study aims to provide scientific reference for local governments to help prevent and mitigate geological disasters in this region.

Application of Multi-System Combination Precise Point Positioning in Landslide Monitoring

To verify the positioning performance and reliability of multi-system combination Precise Point Positioning in landslide monitoring, we carried out a multi-system combination Precise Point Positioning calculation experiment on the monitoring data of a single landslide disaster area in Fujian Province. The coordinates of the monitoring points obtained by a continuously operating reference station and the monitoring station for static relative positioning were used as reference values. The GPS system was used as the standard system and the combined PPP solution mode of G/R/C, G/R/E and G/R/E/C was used to obtain the surface displacement of the landslide area. The research showed that multi-system combination PPP converges to the centimeter level in about 30 min. The average value of internal accordant precision was more than 1 mm after convergence, and that of the external accordant precision was more than 5 cm, which meets the centimeter-level accuracy requirements in rapid landslide deformation monitoring.