Watershed hydrological similarity assessment and watershed classification are important methods for solving the problem of hydrological prediction in ungauged basins. However, traditional watershed hydrological similarity assessment cannot fully reflect the homogeneity among watersheds, and watershed classification methods cannot capture the degree of similarity between watersheds. This study proposes multidimensional indicator clustering–similarity evaluation–classification (MSC). Based on 10 watershed similarity indices, similarity evaluation and classification were conducted using 31 watersheds as examples. Integrating hydrological similarity evaluation and watershed classification, it comprises three layers: multidimensional indicator decomposition with stepwise clustering, hydrological similarity calculation, and watershed classification. Through this structure, the most similar pairs among 31 watersheds were identified, 6 main class were divided, and watersheds with unique hydrological characteristics were isolated. Finally, we verified the rationality of the results by conducting parameter transfer on similar watersheds to simulate runoff processes. The results show that the MSC framework can quickly and efficiently identify similar watersheds, effectively analyze differences in watershed clustering structures under different similarity indices, and realize rational classification of watersheds based on hydrological similarity. It reveals multidimensional watershed clustering information and interclass relationships of watershed characteristics, providing a more refined decision-making basis for watershed management. Its three-layer structure has both correlation and independence, facilitating the adjustment of weights of similarity indices or clustering methods according to specific situations. It provides new ideas and solutions for research related to parameter transfer in ungauged watersheds and has important practical application value.
Li et al. (Mon,) studied this question.