This paper aims to calculate and analyze the spatial and temporal variations in the groundwater flow quantity in Nam Co Lake based on the water balance principle. The results show that a large amount of groundwater was gradually lost and that, groundwater loss decreased from 1.9 billion m³ to 1.5 billion m³ from the period of 1980–1984 to 1995–2009. The comparative analysis in the current study indicates that the decrease in the groundwater index has a strong linear relationship with the temperature of the ground surface on the Tibetan Plateau, with a correlation coefficient as high as 0.92. Moreover, environmental variations such as large-scale engineering construction projects and increases in water storage may have played dominant roles in the sudden changes in the water quantities of plateau lakes (e.g., Nam Co Lake) during the periods of 1990–1995 and 2000–2009. The increased water levels resulted in reduced groundwater losses, which may lead to the substantial expansion or gradual shrinkage of the Qinghai–Tibet Plateau lakes over short periods of time. The results of this study provide an important reference for studying the mechanisms of lake water level changes on the Qinghai–Tibet Plateau.

Rapid acquisition of 3D reconstruction models of landslides is crucial for post-disaster emergency response and rescue operations. This study explores the application potential of Neural Radiance Fields (NeRF) technology for rapid post-disaster site modeling and performs a comparative analysis with traditional photogrammetry methods. Taking a landslide induced by heavy rainfall in Sasebo City, Japan, as a case study, this research utilizes drone-acquired video imagery data and employs two different 3D reconstruction techniques to create digital models of the landslide area. Visual realism and point cloud detail were compared. The results indicate that the high-capacity NeRF model (NeRF 24G) approaches or even surpasses traditional photogrammetry in visual realism under certain scenarios; however, the generated point clouds are inferior in terms of detail compared to those produced by traditional photogrammetry. Nevertheless, NeRF significantly reduces the modeling time. NeRF 6G can generate a point cloud of engineering-useful accuracy in only 45 min, providing a 3D overview of the disaster site to support emergency response efforts. In the future, integrating the advantages of both methods could enable rapid and precise post-disaster 3D reconstruction.