Riverine–lacustrine ecosystems in river–lake continua face increasing threats, yet conventional vulnerability maps often overlook local degradation drivers. This study presents an advanced satellite-based mapping framework using Deep Attention Networks (DANets) for accurate, interpretable vulnerability assessment. In the Ebinur Lake Basin, a representative dryland river system, we first built a satellite-derived evidence map of ecosystem stress aligned with the IPCC’s vulnerability definition. We then optimized DANets via two nature-inspired algorithms: Genetic Algorithm (GA) and Grey Wolf Optimizer (GWO). The optimized models demonstrated strong predictive capacity, explaining a large share of vulnerability variance (R² = 0.78 for GA-DANets; R² = 0.76 for GWO-DANets). For high/low-vulnerability discrimination, GWO-DANets was most effective and stable, with a mean AUC = 0.960 ± 0.044. Factor importance analysis identified soil organic carbon (SOC; 0.29), precipitation seasonality (0.24), and aridity (0.22) as dominant drivers. Two distinct pathways emerged: chronic degradation in arid plains, driven by low SOC and poor water retention; and acute hydrological stress in wetlands, where carbon-rich soils are sensitive to drying. This insight shifts management from uniform to targeted approaches: soil restoration in plains and water-flow protection in wetlands. By integrating metaheuristically optimized deep learning with multi-sensor satellite data, the framework offers a scalable decision-support tool for safeguarding water-dependent ecosystems. The study confirms that vulnerability in the basin follows two predictable, process-based trajectories, which can be directly linked to measurable soil and hydrological conditions. These clear patterns allow managers to prioritize interventions where they will have the greatest effect under ongoing climate pressure.

Terminal lakes are crucial to maintaining ecological balance in arid and semi-arid regions. However, these lakes are increasingly facing severe ecological degradation due to human activities and climate change. This study aims to provide a sustainable management solution for ecological security by determining the suitable water surface area of terminal lakes. We selected eight major terminal lakes in Xinjiang, including Bosten Lake, Lake Ebinur, Lake Ailik, Manas Lake, and Lake Ayding, and analyzed the changes in water surface area over the past 30 years. This research proposes a new method based on ecosystem safety demands to assess the minimum ecological water levels required to maintain the health of these lakes. The study utilizes remote sensing images, ecological security assessments, and hydrological data to analyze the water surface changes, vegetation growth, and ecological health of these lakes. Main findings include the following: (1) The water surface area of most lakes fluctuates significantly, with Bosten Lake maintaining a water surface area between 920 and 1100 km², while Lake Ebinur decreased sharply from 884 km² in 2003 to 350 km² in 2022. (2) The proposed method calculates the suitable water surface area for each lake, ranging from 25 km² (Lake Ayding) to 1050 km² (Bosten Lake). (3) The proposed method provides precise recommendations that can maintain ecological stability while optimizing water resource usage. (4) Some of the findings have already been adopted by local government departments and are applied to water resource management in Xinjiang. This study provides a new approach for defining a suitable water surface area based on ecological security demands and offers valuable insights for the sustainable management of terminal lakes in arid regions.