The vegetation in mountainous areas is abundant, and its ecological carbon sequestration ability is of great significance to maintain the sustainable and healthy development of the ecological environment. However, when estimating the carbon sequestration of mountain vegetation, the Carnegie-Ames-Stanford Approach (CASA) model assigns a uniform value to the maximum light energy utilization (ε_max = 0.389 gC/MJ), ignoring the influence of vegetation types and topographic factors on ε_max, resulting in the low accuracy of the CASA model in estimating the carbon sequestration of mountain vegetation. In this paper, the improved CASA model was combined with Landsat 8 Operational Land Imager (OLI) remote sensing image data to improve the estimation accuracy of carbon sequestration of mountain vegetation. The first was the establishment of a linear link between the terrain characteristics (slope and aspect), vegetation types, and ε_max in mountainous locations. The second was the improvement of the CASA model’s calculation method for key parameters. The different distributions of the estimation results from the two techniques in 2015 and 2016 are then compared using Landsat 8 data as the data source, and the impact of the terrain factors in the improved CASA model on the estimation results is confirmed. Finally, the improved CASA model and the CASA model are used to estimate the Net Primary Productivity (NPP) of the study area from 2000 to 2020, and the estimated results of the two models are compared with the computation results of the MODIS data NPP product. The findings indicate that the improved CASA model’s estimation results have a higher degree of fit and a better correlation. The improved CASA model aids in precisely understanding the ecological carbon sequestration potential of mountain areas and increases the estimation accuracy of vegetation carbon sequestration in mountainous areas.

The purpose of this study was to construct a geological hazard susceptibility evaluation and analysis model using three types of machine learning models, namely, random forest (RF), support vector machine (SVM), and naive Bayes (NB), and to evaluate the susceptibility to landslides, using the Puge section of the Zemu River valley in the Liangshan Yi Autonomous Prefecture as the study area. First, 89 shallow landslide and debris flow locations were recognized through field surveys and remote sensing interpretation. A total of eight hazard-causing factors, namely, slope, aspect, rock group, land cover, distance to road, distance to river, distance to fault, and normalized difference vegetation index (NDVI), were selected to evaluate the spatial relationship with landslide occurrence. As a result of the analysis, the results of the weighting of the hazard-causing factors indicate that the two elements of rock group and distance to river contribute most to the creation of geological hazards. After comparing all the indices of the three models, the random forest model had a higher correct area under the ROC curve (AUC) value of 0.87, root mean squared error (RMSE) of 0.118, and mean absolute error (MAE) of 0.045. The SVM model had the highest sensitivity to geological hazards. The results of geological hazard prediction susceptibility analysis matched the actual situation in the study area, and the prediction effects were good. The results of the hazard susceptibility assessment of the three models are able to provide support and help for the prevention and control of geological hazards in the same type of areas.

Landslide susceptibility assessment can effectively predict the spatial distribution of potential landslides, which is of great significance in fields such as geological disaster prevention, urban planning, etc. Taking Xining City as an example, based on GF-2 remote sensing image data and combined with field survey data, this study delineated the spatial distribution range of developed landslides. Key factors controlling landslides were then extracted to establish a landslide susceptibility assessment index system. Based on this, the frequency ratio (FR), random forest (RF), support vector machine (SVM), and artificial neural network (ANN) models were applied to spatially predict landslide susceptibility with slope units as the basis. The main results are as follows: (1) The overall spatial distribution of landslide susceptibility classes in Xining City is consistent, but the differences between different landslide susceptibility classes are significant. (2) The high-susceptibility area predicted by the FR-RF model is the largest, accounting for 15.48% of the total study area. The prediction results of the FR-ANN and FR-SVM models are more similar, with high-susceptibility areas accounting for 13.96% and 12.97%, respectively. (3) The accuracy verification results show that all three coupled models have good spatial prediction capabilities in the study area. The order of landslide susceptibility prediction capabilities from high to low is FR-RF model > FR-ANN model > FR-SVM model. This indicates that in the study area, the FR-RF model is more suitable for carrying out landslide susceptibility assessment.