Landslides have been frequently occurring in the high mountainous areas in China and poses serious threats to peoples’ lives and property, economic development, and national security. Detecting and monitoring quiescent or active landslides is important for predicting risks and mitigating losses. However, traditional ground survey methods, such as field investigation, GNSS, and total stations, are only suitable for field investigation at a specific site rather than identifying landslides over a large area, as they are expensive, time-consuming, and laborious. In this study, the feasibility of using SBAS-InSAR to detect landslides in the high mountainous areas along the Yunnan Myanmar border was tested first, with fifty-four IW mode Sentinel-1A ascending scenes from 12 January 2019 to 8 December 2020. Next, the Yolo deep-learning model with Gaofen-2 images captured on 5 December 2020 was tested. Finally, the two techniques were combined to achieve better performance, given each of them has intrinsic limitations on landslide detection. The experiment indicated that the combination could improve the match rate between detection results and references, which implied that the performance of landslide detection can be improved with the fusion of time series SAR images and optical images.

Landslide inventories could provide fundamental data for analyzing the causative factors and deformation mechanisms of landslide events. Considering that it is still hard to detect landslides automatically from remote sensing images, endeavors have been carried out to explore the potential of DCNNs on landslide detection, and obtained better performance than shallow machine learning methods. However, there is often confusion as to which structure, layer number, and sample size are better for a project. To fill this gap, this study conducted a comparative test on typical models for landside detection in the Wenchuan earthquake area, where about 200,000 secondary landslides were available. Multiple structures and layer numbers, including VGG16, VGG19, ResNet50, ResNet101, DenseNet120, DenseNet201, UNet−, UNet+, and ResUNet were investigated with different sample numbers (100, 1000, and 10,000). Results indicate that VGG models have the highest precision (about 0.9) but the lowest recall (below 0.76); ResNet models display the lowest precision (below 0.86) and a high recall (about 0.85); DenseNet models obtain moderate precision (below 0.88) and recall (about 0.8); while UNet+ also achieves moderate precision (0.8) and recall (0.84). Generally, a larger sample set can lead to better performance for VGG, ResNet, and DenseNet, and deeper layers could improve the detection results for ResNet and DenseNet. This study provides valuable clues for designing models’ type, layers, and sample set, based on tests with a large number of samples.

Forest canopy height is one of the critical parameters for carbon sink estimation. Although spaceborne lidar data can obtain relatively high precision canopy height on discrete light spots, to obtain continuous canopy height, the integration of optical remote sensing image data is required to achieve “from discrete to continuous” extrapolation based on different prediction models (parametric model and non-parametric model). This study focuses on the Shangri-La area and seeks to assess the practical applicability of two predictive models under complex mountainous conditions, using a combination of active and passive remote sensing data from ICESat-2 and Sentinel-2. The research aims to enhance our understanding of the effectiveness of these models in addressing the unique challenges presented by mountainous terrain, including rugged topography, variable vegetation cover, and extreme weather conditions. Through this work, we hope to contribute to the development of improved geospatial prediction algorithms for mountainous regions worldwide. The results show the following: (1) the fitting effect of the selected parametric model (empirical function regression) is poor in the area of Quercus acutissima and Pinus yunnanensis; (2) evaluation of the importance of each explanatory variable in the non-parametric model (random forest regression) shows that topographic and meteorological factors play a dominant role in canopy height inversion; (3) when random forest regression is applied to the inversion of canopy height, there is often a problem of error accumulation, which is of particular concern to the Quercus acutissima and Pinus yunnanensis; (4) the random forest regression with the optimal features has relatively higher precision by comparing the inversion accuracy of canopy height data of the empirical function regression, random forest regression with all features, and random forest regression with the optimal features in the study area, i.e., R² (coefficient of determination) = 0.865 and RMSE (root mean square error) = 3.184 m. In contrast, the poor estimation results reflected by the empirical function regression, mainly resulting from the lack of consideration of topographic and meteorological factors, are not applicable to the inversion of canopy height under complex topographic conditions.

The production of long-term landslide maps (LAM) holds crucial importance in estimating landslide activity, vegetation disturbance, and regional stability. However, the availability of LAMs remains limited in many regions, despite the application of various machine-learning methods, deep-learning (DL) models, and ensemble strategies in landslide detection. While transfer learning is considered an effective approach to tackle this challenge, there has been limited exploration and comparison of the temporal transferability of state-of-the-art deep-learning models in the context of LAM production, leaving a significant gap in the research. In this study, an extensive series of tests was conducted to evaluate the temporal transferability of typical semantic segmentation models, specifically U-Net, U-Net 3+, and TransU-Net, using a 10-year landslide-inventory dataset located near the epicenter of the Wenchuan earthquake. The experiment results disclose the feasibility and limitations of implementing transfer-learning methods for LAM production, particularly when leveraging the power of U-Net 3+. Furthermore, following an assessment of the effects of varying data volumes, patch sizes, and time intervals, this study recommends appropriate settings for LAM production, emphasizing the balance between efficiency and production performance. The findings from this study can serve as a valuable reference for devising an efficient and reliable strategy for large-scale LAM production in landslide-prone regions.

Shangri-La is located in the eastern part of the Qinghai-Tibet Plateau, which has a fragile ecology. The plateau grassland has suffered from irreversible degradation under the influence of human activities. To address this issue, the Sentinel-2A data obtained is used in this study to calculate the RVI and build an inversion model of grassland degradation grade with GDI data, which was used to obtain the area and proportion of grassland degradation. Landscape indexes were then calculated for different degradation grades of grassland to examine the correlation between roads and degraded grassland in spatial distance and the spatial distribution characteristics of different degradation grades of grassland. The results show that the grassland area in Shangri-La was 2207.94 km², of which the heavily degraded area reaches 824.03 km², exceeding the undegraded grassland area by 172.62 km², indicating that the grassland degradation is severe. From south to north, the proportion of heavily degraded and moderately degraded grassland in townships gradually decreased, while the proportion of lightly degraded and undegraded grassland gradually increased. The townships with high percentages of degraded grassland were predominantly located in the southern area, where there was a dense road network and well-developed transport networks, particularly along National Highway 214, which is the main road in Shangri-La. Conversely, townships with low percentages are generally located in the north with dispersed roads and sparse transport lines. The study’s outcomes are significant in providing a better understanding of the current status of grassland degradation and promoting the sustainable utilization of grassland resources in Shangri-La.