稀疏数据下复杂流域的水质模拟:以赣江为例

以赣江流域为例,对稀疏数据条件下复杂流域水质模型的建立和参数识别进行了探讨．采用了结构相对比较简单的CSTR模型,将其参数划分为水文参数和水质参数,并分别进行识别．水文参数采用回归方法进行识别,对水文参数的扰动实验表明,水质对于水文参数在可能的取值范围内的变化不敏感,因此在随后的水质参数的识别中,将其固定在回归方法求出的最优值上．水质参数的确定则采用了模型方法和资料方法相结合的手段,综合考虑了赣江流域可获得的数据信息、文献资料中对于参数的经验取值和模拟者在大量数学建模中的经验,最终根据河流的不同类型确定了赣江流域的水质参数,并对模型进行了验证。

Water quality simulation in a complicated large watershed:A case study of the Ganjiang watershed

Due to lack of data, it is difficult to develop a methmatical water quality model for a large and complicated watershed. It is thus crucial to fully utilize both quantitative and qualitative information from the studied river system as well as elsewhere. Taking the Ganjing watershed as a case study, a practical way for identifying a watershed water quality model was illustrated. The results suggested that the coefficients, describing the realationships between flow and average depth and between flow and section area, across the river, presented little change along time in most observed hydrological stations. This was consistent with the observation that the hydrological conditions of the Ganjiang River were stable over years which indirectly provided evidence for the validity of the hydrological model. For the water quality components, optimization could easily lead to the estimated parameters highly based. Combination of optimization and the a prior information, in particular from similar watersheds, could give a realiable and better identified model parameters. Validation results indicated that the proposed identification approach was successful in simulating the water quality of the complicated large watershed.