The application of remote sensing technology in grassland monitoring and management has been ongoing for decades. Compared with traditional ground measurements, remote sensing technology has the overall advantage of convenience, efficiency, and cost effectiveness, especially over large areas. This paper provides a comprehensive review of the latest remote sensing estimation methods for some critical grassland parameters, including above-ground biomass, primary productivity, fractional vegetation cover, and leaf area index. Then, the applications of remote sensing monitoring are also reviewed from the perspective of their use of these parameters and other remote sensing data. In detail, grassland degradation and grassland use monitoring are evaluated. In addition, disaster monitoring and carbon cycle monitoring are also included. Overall, most studies have used empirical models and statistical regression models, while the number of machine learning approaches has an increasing trend. In addition, some specialized methods, such as the light use efficiency approaches for primary productivity and the mixed pixel decomposition methods for vegetation coverage, have been widely used and improved. However, all the above methods have certain limitations. For future work, it is recommended that most applications should adopt the advanced estimation methods rather than simple statistical regression models. In particular, the potential of deep learning in processing high-dimensional data and fitting non-linear relationships should be further explored. Meanwhile, it is also important to explore the potential of some new vegetation indices based on the spectral characteristics of the specific grassland under study. Finally, the fusion of multi-source images should also be considered to address the deficiencies in information and resolution of remote sensing images acquired by a single sensor or satellite.

Lake water body extraction from remote sensing images is a key technique for spatial geographic analysis. It plays an important role in the prevention of natural disasters, resource utilization, and water quality monitoring. Inspired by the recent years of research in computer vision on fully convolutional neural networks (FCN), an end-to-end trainable model named the multi-scale lake water extraction network (MSLWENet) is proposed. We use ResNet-101 with depthwise separable convolution as an encoder to obtain the high-level feature information of the input image and design a multi-scale densely connected module to expand the receptive field of feature points by different dilation rates without increasing the computation. In the decoder, the residual convolution is used to abstract the features and fuse the features at different levels, which can obtain the final lake water body extraction map. Through visual interpretation of the experimental results and the calculation of the evaluation indicators, we can see that our model extracts the water bodies of small lakes well and solves the problem of large intra-class variance and small inter-class variance in the lakes’ water bodies. The overall accuracy of our model is up to 98.53% based on the evaluation indicators. Experimental results demonstrate that the MSLWENet, which benefits from the convolutional neural network, is an excellent lake water body extraction network.

Most previous studies of the Qinghai-Tibet engineering corridor (QTEC) have focused on the impacts of climate change on thaw-induced slope failures, whereas few have considered freeze-induced slope failures. Terrestrial laser scanning was used in combination with global navigation satellite systems to monitor three-dimensional surface changes between 2014 and 2015 on the slope of permafrost in the QTEC, which experienced two thawing periods and a freezing period. Soil temperature and moisture sensors were also deployed at 11 depths to reveal the hydrological–thermal dynamics of the active layer. We analyzed scanned surface changes in the slope based on comparisons of multi-temporal point cloud data to determine how the hydrological–thermal process affected active layer deformation during freeze–thaw cycles, thereby comprehensively quantifying the surface deformation. During the two thawing periods, the major structure of the slope exhibited subsidence trends, whereas the major structure of the slope had an uplift trend in the freezing period. The seasonal subsidence trend was caused by thaw settlement and the seasonal uplift trend was probably due to frost heaving. This occurred mainly because the active layer and the upper permafrost underwent a phase transition due to heat transfer. The ground movements occurred approximately in the soil temperature conduction direction between the top of the soil and the permafrost table. The elevation deformation range was mainly −0.20 m to 0.20 m. Surface volume increases with heaving after freezing could have compensated for the loss of thawing twice and still led to the upward swelling of the slope. Thus, this type of slope in permafrost is dominated by frost heave. Deformation characteristics of the slope will support enhanced decision making regarding the implementation of remote sensing and hydrological–thermal measurement technologies to monitor changes in the slopes in permafrost adjacent to engineering corridors, thereby improving the understanding and assessment of hazards.

In recent years, deep learning methods utilizing convolutional neural networks have been extensively employed in hyperspectral image classification (HSI) applications. Nevertheless, while a substantial number of stacked 3D convolutions can indeed achieve high classification accuracy, they also introduce a significant number of parameters to the model, resulting in inefficiency. Furthermore, such intricate models often exhibit limited classification accuracy when confronted with restricted sample data, i.e., small sample problems. Therefore, we propose a spectral–spatial double-branch network (SSDBN) with an attention mechanism for HSI classification. The SSDBN is designed with two independent branches to extract spectral and spatial features, respectively, incorporating multi-scale 2D convolution modules, long short-term memory (LSTM), and an attention mechanism. The flexible use of 2D convolution, instead of 3D convolution, significantly reduces the model’s parameter count, while the effective spectral–spatial double-branch feature extraction method allows SSDBN to perform exceptionally well in handling small sample problems. When tested on 5%, 0.5%, and 5% of the Indian Pines, Pavia University, and Kennedy Space Center datasets, SSDBN achieved classification accuracies of 97.56%, 96.85%, and 98.68%, respectively. Additionally, we conducted a comparison of training and testing times, with results demonstrating the remarkable efficiency of SSDBN.