Landslide susceptibility mapping (LSM) could be an effective way to prevent landslide hazards and mitigate losses. The choice of conditional factors is crucial to the results of LSM, and the selection of models also plays an important role. In this study, a hybrid method including GeoDetector and machine learning cluster was developed to provide a new perspective on how to address these two issues. We defined redundant factors by quantitatively analyzing the single impact and interactive impact of the factors, which was analyzed by GeoDetector, the effect of this step was examined using mean absolute error (MAE). The machine learning cluster contains four models (artificial neural network (ANN), Bayesian network (BN), logistic regression (LR), and support vector machines (SVM)) and automatically selects the best one for generating LSM. The receiver operating characteristic (ROC) curve, prediction accuracy, and the seed cell area index (SCAI) methods were used to evaluate these methods. The results show that the SVM model had the best performance in the machine learning cluster with the area under the ROC curve of 0.928 and with an accuracy of 83.86%. Therefore, SVM was chosen as the assessment model to map the landslide susceptibility of the study area. The landslide susceptibility map demonstrated fit with landslide inventory, indicated the hybrid method is effective in screening landslide influences and assessing landslide susceptibility.

The tensile strength of roots and the friction characteristics of the root–soil interface of tree species are the indicators that play a crucial role in understanding the mechanism of soil reinforcement by roots. To calculate the effectiveness of the reinforcement of soil by tree roots based on essential influencing parameters, typical trees in the coastal region of southeastern China selected for this study were subjected to tests of the tensile mechanical properties of their roots, as well as studies on the friction characteristics of the root–soil interface and the microscopic interfaces. The results indicated that in the 1–7 diameter classes, the root tensile strength of both Pinus massoniana and Cunninghamia lanceolata was negatively correlated with the root diameter in accordance with the power function. The root tensile strength of these two trees, however, was positively correlated with the lignin content but negatively correlated with cellulose and hemicellulose contents. The shear strength at the root–soil interface and the vertical load exhibited a constitutive relationship, which followed the Mohr–Coulomb criterion. As the root diameter increased, both the cohesion and the friction coefficients at the root–soil interface gradually increased, but the growth rate stood at around 15%. The cohesion value of the root–soil interface of the two trees decreased linearly with the increase in soil moisture content within the range of 25 to 45%. At the microinterface, the root surface of C. lanceolata exhibited concave grooves and convex ridges that extended along the axial direction of roots, with their height differences increasing with the enlargement of the root diameter. The rough surface of P. massoniana roots had areas composed of polygonal meshes, with an increase observed in the mesh density with increasing root diameter.

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.