Water level monitoring plays a crucial role in managing water resources, flood prevention, and disaster mitigation in reservoirs. The technology of Global Navigation Satellite System-Interferometric Reflectometry (GNSS-IR) has been extensively utilized for water level monitoring in various types of water bodies. However, owing to significant annual water level fluctuations and narrow reflection zones in some reservoirs, this poses a major challenge for the usage of GNSS-IR in reservoir monitoring. To address these challenges, this study proposes a method for dynamically selecting the elevation angle range (DSEAR). In this study, observation data from a GNSS monitoring network established around the Kalabeily Reservoir, comprising one reference station and five monitoring stations, is utilized for water level retrieval. It has been demonstrated that this method effectively expands the usable elevation angle range and increases the reflection zones, thereby improving the success rate and reliability of the retrieval results. This method also lays the foundation for real-time monitoring of reservoir water levels. To additionally, improve the retrieval accuracy, an algorithm for water level retrieval combining multi-station multi-signal data is proposed, based on a robust estimation strategy. The results indicate that the multi-station multi-signal combination retrieval method can achieve a maximum accuracy improvement of 44.70% compared to the single-station multi-signal retrieval method. The water level retrieval sequences from 2023 also exhibit consistency with the reservoir operation mode. This confirms the significant potential of GNSS-IR technology in complex environments.