The Study Of Equifinality Of Conceptual Rainfall-Runoff Model Parameter

The concept of Equifinality come from systems engineering, means that the same end can be achieved from different original conditions and(or) by different ways. In hydrology, Equifinality, is used here in the sense of an expectation that the same end, in the case an acceptable model prediction, might be achieved in many different ways, i.e. different model structures or parameter sets. This is not to say that every simulation is equally acceptable as a simulator of the system. Each simulation can be evaluated as part of the calibration process and ranked in terms of performance or likelihood of being an acceptable simulator, if some criterion of evaluation(either qualitative or quantitative) can be specified.After comprehensive reference of previous literatures, we apply GLUE methodology to Xinanjiang Rainfall-Runoff Model simulation of the Baohe River, which is a branch of Luanhe River:1. Using rainfall and evaporaspiratin data of Baohe River, we generate synthetic runoff data, which are used to analyze the equifinality of parameters of Xinanjiang Model, whose parameters are classfied in the following way: evaporaspiration para; runoff production para; runoff separation para: and runoff concentration para. Simulated by the class, the wide range of parameter sets giving acceptable simulations and correlation of parameter and the likelihood measure are demonstrated.2. Using rainfall and evaporaspiration data of Baohe River, synthetic runoff data with normal distribution random error are generated for simulation, and the Equifmality of parameters and the correlation of parameters to likelihood measures are analyzed, which demonstrate the wide range of parameter sets giving acceptable simulations. The result is compared with that of error-freed data, and the impacts of normal errors on Equifinality of parameters of Xinanjiang Model is not salient.3. Using live observed runoff data, the same study is made and the great difference between the result of observed data and that of synthetic data shows that the difference between the structure of the watershed model and the hydrology process affect the equifinality of parameter greatly.