A real-time correction method for dynamic system differential response with consideration of uncertainty in model structure

Structural uncertainty in hydrological models is an important factor affecting accuracy of hydrological forecasting． To quantify and reduce such effect is a hotly researched area． Dynamic system response curve (DSRC) method was used to deduce mathematical relationship between model structural errors and changes in input, and to propose a dynamic system response curve method with consideration of model structural uncertainty (UNDSRC), to distinguish source of model structural uncertainty． This method was applied to Dapoling and Fushui watersheds, for a test of comprehensive performance of UNDSRC algorithm and for comparison with DSRC algorithm． UNDSRC algorithm was found to perform better than DSRC algorithm in actual basin tests, with average correction effect evaluation coefficient of 0.82 (UNDSRC) and 0.60 (DSRC) respectively． Xinanjiang model data were effectively corrected by DSRC method in the two watersheds, with similar correction effects． UNDSRC algorithm is excellent and stable and can adapt to more complex surface conditions, the algorithm corrects flood peak better than runoff depth． With identical correction accuracy, ridge coefficient of UNDSRC method is much smaller than DSRC method．