Two significant uncertainties that are crucial for landslide susceptibility prediction modeling are attribute interval numbers (AIN) division of continuous landslide impact factors in frequency ratio analysis and various susceptibility prediction models. Five continuous landslide impact factor interval attribute classifications (4, 8, 12, 16, 20) and three data-driven models (deep belief networks (DBN), random forest (RF), and neural network (back propagation (BP)) were used for a total of fifteen different scenarios of landslide susceptibility prediction studies in order to investigate the effects of these two factors on modeling and perform a landslide susceptibility index uncertainty analysis (including precision evaluation and statistical law). The findings indicate that: (1) The results demonstrate that for the same model, as the interval attribute value rises from 4 to 8 and finally to 20, the forecast accuracy of landslide susceptibility initially increases gradually, then progressively grows until stable. (2) The DBN model, followed by the RF and BP models, provides the highest prediction accuracy for the same interval attribute value. (3) AIN = 20 and DBN models have the highest prediction accuracy under 15 combined conditions, while AIN = 4 and BP models have the lowest. The accuracy and efficiency of landslide susceptibility modeling are higher when the AIN = 8 and DBN models are combined. (4) The landslide susceptibility index uncertainty predicted by the deeper learning model and the bigger interval attribute value is comparatively low, which is more in line with the real landslide probability distribution features. The conditions that the environmental component attribute interval is divided into eight parts and DBN models are used allow for the efficient and accurate construction of the landslide susceptibility prediction model.

A linear hazard-causing factor is the environmental element of landslide susceptibility prediction, and the setting of buffer distance of a linear hazard-causing factor has an important influence on the accuracy of landslide susceptibility prediction based on machine learning algorithms. A geographic information system (GIS) has generally been accepted in the correlation analysis between linear hazard-causing factors and landslides; the most common are statistical models based on buffer zone analysis and superposition analysis for linear causative factor distances and landslide counts. However, there is a problem in the process of model building: the buffer distance that is used to build the statistical model and its statistical results can appropriately reflect the correlation between the linear disaster-causing factors and landslides. To solve this problem, a statistical model of landslide density and distance of linear disaster-causing factors under different single-loop buffer distances was established based on Pearson’s method with 12 environmental factors, such as elevation, topographic relief, and distance from the water system and road, in Ruijin City, Jiangxi Province to obtain the most relevant single-loop buffer distance linear disaster-causing factor combinations; random forest (RF) machine learning models were then used to predict landslide susceptibility. Finally, the Kappa coefficient and the distribution characteristics of the susceptibility index were used to investigate the modeling laws. The analysis results indicate that the prediction accuracy of the most correlated single-loop buffer distance combination reaches 96.65%, the error rate of non-landslide points is 4.2%, and the error of landslide points is 11.3%, which is higher than the same single-loop buffer distance combination, confirming the reasonableness of the method of using correlation to obtain the linear disaster-causing factor buffer distance.

Based on GNSS observations, the co-seismic and post-seismic slip of the 2024 Noto Peninsula earthquake and the spatio-temporal pattern of afterslip are investigated in this paper. The co-seismic slip is mainly distributed in the depth range of 2 to 15 km with the maximum value of 5.94 m. Compared with the co-seismic rupture pattern, a shallow afterslip can be observed after the earthquake, and the afterslip patch is formed northeast of the epicenter. The maximum value of afterslip during the post-seismic 180 days is 1.13 m, which is situated at the longitude of 137.53°, latitude of 37.75°, and epth of 5.43 km. The spatio-temporal evolution of afterslip indicates that the fault activity has continued throughout the post-seismic 180 days, and the coverage and magnitude of afterslip have gradually increased. As time goes on, the fault activity tends to weaken, as evidenced by a decrease in slip rate. The daily images of afterslip demonstrate that the fault activity is particularly strong in the early time period following the earthquake. The maximum value of afterslip in the first week accounts for about 18% of that in the post-seismic 180 days, and the maximum slip rate reaches 0.043 m/day. In addition, the Coulomb stress analysis indicates that afterslip and most aftershocks appear in the positive Coulomb stress region, suggesting that co-seismic Coulomb stress changes may be the driving mechanism of afterslip and aftershocks.

Based on subdaily kinematic GNSS solutions, the fault slip properties during the very early postseismic phase after the 2021 M 8.2 Chignik earthquake are investigated in this paper. The very early postseismic deformations captured by near-field GNSS sites can be well depicted by the power model. The comparison of afterslip determined by daily and subdaily GNSS solutions suggests that neglecting very early afterslip can result in the underestimation of postseismic slip. Compared with coseismic slip, the cumulative afterslip of the first 24 h is mainly focused in the southeast of the hypocenter, and the shallow updip afterslip appears after this earthquake. The spatio-temporal evolution of the afterslip reveals that the patch of afterslip is immediately generated after the earthquake, and then the postseismic slip gradually grows along the afterslip patch. The magnitude of the afterslip patch varies remarkably within the 24 h following the earthquake, especially in the first several hours. Meanwhile, the spatio-temporal patterns of aftershocks and afterslip exhibit strong similarity during the first 24 h, suggesting that very early afterslip may be a possible driving factor of aftershocks. Moreover, most of the afterslip patches and aftershocks occurring immediately after this earthquake are situated in the area covered by positive Coulomb Stress Change (CSC), which implies that the immediate afterslip and aftershock activities can be influenced by the coseismic CSC. The following afterslip process further releases coseismic CSC and then influences the spatio-temporal variations of aftershock activities. Thus, the afterslip may be a possible triggering mechanism of very early aftershocks for this earthquake, alongside the effects of the CSC generated by coseismic rupture.