摘要
Lake Buchanan, a major reservoir for the City of Austin area, the Texas Hydrologic Region 12, USA.Numerical climate models are increasingly being used by climate scientists to inform water management. However, successful transitions from climate models (O(10–100 km)) to water resources studies (O(100 m–1 km)) still need improved data structures and modeling strategies to resolve spatial scale mismatch. In this study, we introduce a mechanistic lake-level modeling framework that consists of a state-of-the-art land surface model – Noah-MP, a vector-based river routing scheme – RAPID, and a lake mass-balance model. By conducting a case study for Lake Buchanan, we demonstrate the capability of the framework in predicting lake levels at seasonal lead times. The experiments take into account different runoff resolutions, model initialization months, and multiple lead times. Uncertainty analyses and sensitivity tests are also conducted to guide future research.New hydrological insightsDifferent from traditional grid-based solutions, the framework is directly coupled on the vector-based NHDPlus dataset, which defines accurate hydrologic features such as rivers, dams, lakes and reservoirs. The resulting hybrid framework therefore allows for more flexibility in resolving “scaling-issues” between large-scale climate models and fine-scale applications. The presented hindcast results also provide insight into the influences of baseline LSM resolutions, initialization months, and lead times, which would ultimately help improve lake-level forecast skills.