Establishing an effective real-time monitoring and early warning system for glacier lake outburst floods (GLOFs) requires a full understanding of their occurrence mechanism. However, the harsh conditions and hard-to-reach locations of these glacial lakes limit detailed fieldwork, making satellite imagery a critical tool for monitoring. Lake Mercbacher, an ice-dammed lake in the central Tian Shan mountain range, poses a significant threat downstream due to its relatively high frequency of outbursts. In this study, we first monitored the daily changes in the lake area before the 2022 Lake Mercbacher outburst. Additionally, based on historical satellite images from 2014 to 2021, we calculated the maximum lake area (MLA) and its changes before the outburst. Furthermore, we extracted the proportion of floating ice and water area during the period. The results show that the lake area of Lake Mercbacher would first increase at a relatively low speed (0.01 km²/day) for about one month, followed by a relatively high-speed increase (0.04 km²/day) until reaching the maximum, which would last for about twenty days. Then, the lake area would decrease slowly until the outburst, which would last five days and is significant for early warning. Moreover, the floating ice and water proportion provides more information about the outburst signals. In 2022, we found that the floating ice area increased rapidly during the early warning stage, especially one day before the outburst, accounting for about 50% of the total lake area. Historical evidence indicates that the MLA shows a decreasing trend, and combining it with the outburst date and climate data, we found that the outburst date shows an obvious advance trend (6 days per decade) since 1902, caused by climate warming. Earlier melting results in an earlier outburst. This study provides essential references for monitoring Lake Mercbacher GLOFs and building an effective early warning system.

Glaciers are sensitive indicators of climate change, and investigation of their dynamics is crucial for ensuring regional ecological security as well as disaster prevention and mitigation measures. Based on Landsat Thematic Mapper (TM)/Enhanced Thematic Mapper Plus (ETM+)/Operational Land Imager (OLI) imagery, the outlines and length of glaciers in the West Kunlun Main Peak Area (WKMPA) during 2000–2020 were obtained by combining a band ratio method with manual interpretation and an automatic extraction method for the glacier centerline, respectively. There were 440 glaciers in the WKMPA in 2020, covering an area of 2964.59 ± 54.87 km², with an average length of 2916 ± 60 m. The glacier count increased due to division, while the area and length all exhibited a declining trend from 2000 to 2020, at rates of −0.04%·a⁻¹ (24.83 km²) and −0.11%·a⁻¹ (66 m), respectively. Glacier retreat was primarily observed during the early period (2000–2005). Except for glaciers located above an elevation of 6250 m, the glacier area decreased with each altitude interval from 2000 to 2020, and the rate of relative change in glacier area generally decreased with increasing altitude. Moreover, except for a slight increase in north-facing glaciers, the area of glaciers facing other orientations decreased during 2000–2020. The accuracy of the empirical formula fit for glacier length was highly dependent on glacier class, with greater precision observed for smaller glaciers and lower precision for larger valley-basin glaciers due to their complex morphological structures being neglected and only a single quantitative relationship being considered. There was a time lag of 12 years between temperature changes and glacier area response in this region. The mechanism by which glacier division affects glacier change is complex, requiring dissection of multiple factors such as area, length, and terminal elevation before and after division.

Flood disasters are frequent, sudden, and have significant chain effects, seriously damaging infrastructure. Remote sensing images provide a means for timely flood emergency monitoring. When floods occur, emergency management agencies need to respond quickly and assess the damage. However, manual evaluation takes a significant amount of time; in current, commercial applications, the post-disaster flood vector range is used to directly overlay land cover data. On the one hand, land cover data are not updated in time, resulting in the misjudgment of disaster losses; on the other hand, since buildings block floods, the above methods cannot detect flooded buildings. Automated change-detection methods can effectively alleviate the above problems. However, the ability of change-detection structures and deep learning models for flooding to characterize flooded buildings and roads is unclear. This study specifically evaluated the performance of different change-detection structures and different deep learning models for the change detection of flooded buildings and roads in very-high-resolution remote sensing images. At the same time, a plug-and-play, multi-attention-constrained, deeply supervised high-dimensional and low-dimensional multi-scale feature fusion (MSFF) module is proposed. The MSFF module was extended to different deep learning models. Experimental results showed that the embedded MSFF performs better than the baseline model, demonstrating that MSFF can be used as a general multi-scale feature fusion component. After FloodedCDNet introduced MSFF, the detection accuracy of flooded buildings and roads changed after the data augmentation reached a maximum of 69.1% MIoU. This demonstrates its effectiveness and robustness in identifying change regions and categories from very-high-resolution remote sensing images.

Against the background of rapid climate warming, frequent and severe flooding disasters significantly impact socio-economic development and human life. In 2023, due to the northward movement of Typhoon Doksuri, extreme precipitation occurred in northern China, which resulted in a massive flood event in the Haihe River basin. Seven flood detention basins (FDBs) in North China were successfully implemented to effectively alleviate the downstream flood pressure. Leveraging all available Chinese satellite data, we monitored the flooding process daily, focusing on reviewing the flooding in the Dongdian FDB. The results indicate that since the activation of Dongdian FDB on August 1, the flood reached the urban area of Tianjin in just nine days and inundated the entire detention basin. Flooding persisted in the detention basin for about a week before gradually receding. The total maximum inundated area in the whole region was 307.5 km2, including 240.5 km2 of arable land, 7.0 km2 of greenhouse land, and 9.7 km2 of built-up land, with an average inundation duration of 19 days. The total cumulative inundated arable land area was 240.5 km2, with an average inundation time of 21 days. This study shows that multi-source Chinese satellite data can provide comprehensive information and adequate references for post-disaster assessments.

Building extraction is a core task in the semantic segmentation of satellite remote sensing imagery. Conventional pixel-level segmentation methods often prioritize texture over geometric structure, resulting in suboptimal performance in complex scenes affected by illumination variations, shadows, and scale changes. In this article, an innovative object-level building extraction approach is introduced to better capture the geometric structure of buildings, which incorporates superpixel segmentation to represent images as a set of adjacent regions. The proposed model consists of a cascade multi-scale fusion module (CMSFM) that progressively integrates contextual information across different receptive fields, along with a boundary-assisted loss function designed to enhance edge delineation and improve object-level accuracy. The experimental results on the WHU building dataset and the Massachusetts Buildings Dataset show that the proposed method notably outperforms other representative semantic segmentation approaches, such as FCN, UNet, DeepLab V3, and SETR. On the WHU dataset, MRLNet achieves the largest MIoU of 90.14% and the highest F1 score of 92.47%. On the Massachusetts Buildings Dataset, MRLNet attains the best MIoU of 83.14% and the highest F1 score of 90.46%. In addition, our building extraction model achieves a substantial performance improvement after the addition of the CMSFM module and the boundary-assisted loss function, demonstrating the effectiveness of these two enhancements used in our proposed model. It is expected that this research can provide a promising tool for the accurate extraction of buildings using satellite remote sensing images, which is indispensable in urban planning, disaster assessment, and other fields.