The Influence of Temporal and Spatial Boundary Condition Variation on The Parameter Sensitivity of Wasp

With the development of computer technology, more and more application of water quality model appears in environmental management. The WASP is currently the most widely used water quality model, which contains a large number of parameters have to be calibrated before simulation. Therefore identification of the important parameter is an important step for the application of the model, which means parameter sensitivity analysis should be done first. Sensitivity analysis is the model research hot spot problems. But research about the influence on the parameters sensitivity caused by the variation of boundary condition is not enough. For a quantitative discussion of the influence on the sensitivity of WASP parameters which caused by temporal and spatial variation, sensitivity of the parameters related to model output DO, CBOD, NH3-Nand NO3-N is analyzed based on EFAST method with Simulink. The results show that: EFAST method can efficiently filter out the most sensitive parameters of the process simulation as a global sensitivity analysis method, it has a great potential of application to localize the parameter in the complex nonlinear water quality model; the significant difference between the total sensitivity and first-order sensitivity index shows that some parameters of the WASP model is strongly coupling, and "One variable at a time" method is inappropriate to evaluate the sensitivity of these parameters; and it is also found that even for the same river, WASP parameter sensitivity index changes as boundary conditions changes, which are significantly affected by the temporal variation; and as the input variable changes over time, the uncertainty caused by parameter is also significantly different; WASP model shows an obvious "equifinality", for which the traditional methods such as the global optimization calibration, fail to reflect the conversional process among the components in water body. Since that, experiments are required as a verification from another perspective when calibrate the key parameters.