参数优选是水文模型研究的一个重点问题,与模型的模拟效果紧密相关。在对广义模式搜索(GPS)算法进行详细分析的基础上，分别以2 000 km2的中国流域和760 km2的澳大利亚流域为例，输入实测气象水文数据，探讨了该算法在SIMHYD模型参数优选中的应用。结果表明，对于中国流域（Ⅰ），率定期和验证期的Nash-Sutcliffe效率系数（NSE）均达到0.7以上，水量平衡误差（WBE）的绝对值小于6%，RMSE分别为2.24 mm/d和2.21 mm/d；对于澳大利亚流域（Ⅱ），率定期和验证期NSE均达到0.7以上，|WBE|小于5%，RMSE分别为0.11 mm/d和0.13 mm/d。可以看出，GPS算法具有较好的全局收敛性和稳定性，其计算效率和精度较高, 可以有效应用于概念性水文模型的参数优化。

Parameter optimization plays a key role in the research of hydrological model. The model performance is closely related to the determination of parameters. In this paper, the Generalized Pattern Search (GPS) algorithm was analyzed, and then its application in the parameter optimization of SIMHYD model was assessed based on the observed hydrological and meteorological data in two catchments, including one located in China with an area of 2000 km2 and the other one located in Australia with an area of 760 km2. The results indicated that (1) the Nash-Sutcliffe efficiency (NSE) coefficients are higher than 0.7 and the absolute Water Balance Errors (WBE) are lower than 6% for the catchment (Ⅰ) in both model calibration and validation, and the RMSEs are 2.24 mm/d and 2.21 mm/d in model calibration and validation, respectively; and (2) for the catchment (Ⅱ), NSEs are also higher than 0.7 and the absolute WBEs are less than 5% in both model calibration and validation, and the RMSEs are 0.11mm/d and 0.13mm/d in model calibration and validation, respectively. The GPS algorithm has a global convergence and stability and high computational efficiency and accuracy, therefore it is suitable for parameter optimization of the conceptual hydrological models.

国家科技重大专项项目（2010ZX03006-006）；中科院百人计划择优支持项目（YOBROB045）；国家自然科学基金项目（40871095）